S A M ENES S A ND SUBSTANCE RENEW ED

In this book, which thoroughly revises and greatly expands his

classic work Sameness and Substance (), David Wiggins retrieves
and refurbishes in the light of twentieth-century logic and logical
theory certain conceptions of identity, of substance and of persis-
tence through change that philosophy inherits from its past. In this
new version, he vindicates the absoluteness, necessity, determinate-
ness and all or nothing character of identity against rival concep-
tions. He defends a form of essentialism that he calls individuative
essentialism, and then a form of realism that he calls conceptualist
realism, a position he seeks to place in relation to one surviving
insight of idealism. In a final chapter, he advocates a human being
based conception of the identity and individuation of persons,
arguing that any satisfactory account of personal memory must
enforce and follow through all the normative requirements that
flow from its logically inalienable aspiration to furnish direct knowl-
edge of the rememberer’s own past. This important book will
appeal to a wide range of readers in metaphysics, philosophical
logic, and analytic philosophy.

  is Wykeham Professor of Logic Emeritus in the

University of Oxford, Fellow of the British Academy, and Foreign
Honorary Member of the American Academy of Arts and
Sciences. His publications include Needs, Values, Truth: Essays in the
Philosophy of Value () and many journal articles.
 SAMENESS AND
SUBSTANCE RENEWED

DAVID WIGGINS
Wykeham Professor of Logic Emeritus in the University of Oxford
PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS (VIRTUAL PUBLISHING)
FOR AND ON BEHALF OF THE PRESS SYNDICATE OF THE UNIVERSITY OF
CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge CB2 IRP
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia

http://www.cambridge.org

© David Wiggins 2001

This edition © David Wiggins 2003

First published in printed format 2001

A catalogue record for the original printed book is available

from the British Library and from the Library of Congress
Original ISBN 0 521 45411 5 hardback
Original ISBN 0 521 45619 3 paperback

ISBN 0 511 01291 8 virtual (eBooks.com Edition)

 Contents

Preface page ix

Preamble, chiefly concerned with matters methodological and

terminological 
 Aims and purposes 
 Formal properties of identity 
 Notions 
 Philosophical terminology: a manifesto 
 Sortal predicates and sortal concepts: and concepts versus conceptions 
 Real and nominal 
 Necessary/contingent and a priori / a posteriori 
 Formal notations 
 The mode of combination involved in ‘is the same donkey as’ 
 Questions of priority 

 The absoluteness of sameness 

 A central question about identity; and rival answers given by defenders
of the absoluteness of identity and the relativity of identity 
 Leibniz’s Law and the difficulties of relative identity 
 Five ways for it to be false that a ⫽ b 
g
 Possible examples of type-() relativity 
 Some cases that might be alleged to be of type () 
 Discussion of type-() cases 
 Discussion of type-() cases and some attempted amendments of
Leibniz’s Law 
 A mathematical example supposedly of type () 
 Conclusion concerning R, the Relativity of Identity 
 Absoluteness and sortal dependence jointly affirmed and formalized 

 Outline of a theory of individuation 

 Propostion D and the rationale of the ‘same what?’ question 
 The charge of circularity, or of emptiness 

v
vi Contents
 The identity of indiscernibles 
 Proposition D further explicated and amplified: and D(ii) as the proper
development of D 
 Existence and sortal predications 
 Further D principles 
 Miscellaneous further principles; and a doubt about counting 

 Sortal concepts: and the characteristic activity or function or

purpose of things falling under them 
 The sortal predicates of natural kinds 
 The other sortal predicates 
 Problems of artefact identity 
 Two approaches to the problem of artefact identity 
 Summary of conclusions to date: and a methodological remark 
 Transition to Chapters Four and Five 

 Individuative essentialism 

 Independence from the explicitly modal of the foregoing theory of
individuation 
 Principles and maxims governing the derivation of a modest
essentialism 
 The necessity of identity and the necessity of difference 
 Conceivability, theory and essence 
 Conceivability continued 
 Individuative essentialism and its consequences 
 That the idea of ‘haecceitas’ is as misbegotten as the word itself is
unlovely 
 The essentialist ‘must’ and ‘can’ 
 Avoiding overspecificity, allowing vagueness 
 Other de re necessities, real or putative: a framework for further
inquiry 
 The essences of artefacts and the matter of artefacts 
 One special kind of artefact: works of art and the essences of these 

 Conceptualism and realism 

 Anti-realist conceptualism and anti-conceptualist realism 
 Four clarifications 
 A conventionalist reconstruction of our modal convictions: a
conceptualist anti-realist view of essence 
 A hypothesis concerning the sources of anti-essentialism 
 An exaggeration of conceptualism, deprecated and corrected in the
light of certain truisms; and the reply to the anti-conceptualist
realist begun 
 Contents vii
 The perfect consonance of sober realism and sober conceptualism 
 The realist requirement restated, refurbished and satisfied 
 Concluding suggestions 

 Identity: absolute, determinate, and all or nothing, like no

other relation but itself 
 Three contrasted views of singling out an object 
 Back and forth between the object and the thought of the object 
 Some putative examples of indeterminate objects 
 If object a is the same as object b, then a is determinately the same as b 
 What, if anything, follows from such formal derivations? 
 Treatment of examples (a), (b), (c); of § 
 Sense and point; and sense as the work of the mind 
 On the level of reference, things cannot be simply conceived into
being or postulated into existence – not even material things with
matter putatively ready at hand 
 Once again (one last time) the things to which simple identity
sentences make a reference 
 More about the relation of identity 
 Might it ever be true to say that a was almost b, that a was almost
numerically identical with b? 
 Conclusion 

 Personal identity 

    
 An expeditious if precipitate answer to the question of personal
identity 
 Doubts, and answers to doubts: subjects of consciousness 
 The Lockean conception; and Butler’s criticisms of such conceptions 
 A neo-Lockean identity-condition 
 Butler’s central insight 
 A neo-Lockean conception 
 Unfinished business 
 The theses to be argued in this chapter 
 Co-consciousness again, and quasi-memory 
 A second and third question about Parfit’s definition of ‘Q-remember’ 
 Digression: an alternative method of definition, revealing by its
inadequacy the semantical point of the attribution of experiential
memory 
 More about ‘dependent in the right way’ 
   
 As it now appears, the state of the whole argument to date 
 Participation in the growth of knowledge 
viii Contents
 The penultimate problem and a verdict upon it, all leading in due
course to a reassessment of the original Shoemaker case 
 Brown–Brownson reconsidered 
 One last variant – and the philosophical moral of same. Finally,
human persons as artefacts? 

Select bibliography 

Index of names of persons cited or mentioned 
Index of contents (themes, theses, examples, etc)  
 Preface

When Sameness and Substance (Blackwell, ) went out of print,

Cambridge University Press agreed to take over the book. They sug-
gested that the Longer Notes be dropped and certain smaller matters be
attended to in recognition of what has happened since . They urged
that the chapter about personal identity be superseded. In the process of
my discovering how just and sensible these proposals were, then forming
the resolve to follow the substance theory through more single-mindedly
and to a greater distance, there came into being the version I have called
Sameness and Substance Renewed.
Whether Sameness and Substance Renewed is the same book as Sameness
and Substance is not a question of importance – the matter of a joke that
will fail nobody who wants to make it, or else of an exercise for the
reader (not to be attempted before reading the new Chapter Six or
without regard to the sort of ambiguities set out in Chapter One, §§–).
The present text seeks to correct all the things in the  version that I
know to be plain wrong. Then, in the same dialect of mid twentieth-
century English, it extends that version at some of the places where more
was needed. Most conspicuously, there is a new chapter about identity,
vagueness and supervenience; and, as requested, the old chapter on per-
sonal identity is entirely replaced. Those who interest themselves closely
in the annals of disputes about these subjects will have to retrieve the old
pages – from the same dust-heap of history as harbours most of
the theses and questions once explored in the Longer Notes of the 
version.
In the text from  that survived all these decisions, there has been
rewriting and abbreviation. Neither of these processes could be carried
far enough. But the reader may be assured that the present version does
not set out to transcribe everything that still seems to me to be true from
Sameness and Substance () or from the book that Sameness and Substance
itself consolidated. That earlier book was called Identity and Spatio-Temporal
ix
x Preface
Continuity, was published in  and ran to seventy-eight pages of text.
By chance or good luck, the present preface is addressed from the same
place as was the  preface. But neither chance nor luck, nor yet an
inflexible will to abandon absolutely everything save that which is central,
could have restored the same brevity or the same simplicity of purpose
that was possible in , given the wider range of well formed questions
now wanting attention.
This is not the book that I should have written if I had been starting
afresh or I had been able to train a freer fancy or a more impartial atten-
tion upon the logical and philosophical literature of identity produced
in the twenty-one years since Sameness and Substance was given to the pub-
lisher. But I have tried to update whatever has been allowed to remain –
or else to test it off the page against alternative options made newly
visible. It has been a great help that Sabina Lovibond’s and Stephen
Williams’s collection Identity, Truth and Value: Essays for David Wiggins
(Blackwell, ), henceforth their (), recently obliged me to review
everything I was committed to. Sameness and Substance Renewed follows
through the implications of the commentary I offered in Identity, Truth
and Value upon the essays presented there by Timothy Williamson,
Harold Noonan and Paul Snowdon. Not only that. It follows through
the reactions to which I was moved or provoked by the numerous other
items that I happened upon in composing replies to these three scholars.
For everything Williamson, Noonan and Snowdon did directly and indi-
rectly to provoke these reactions I am extremely grateful. In the case of
some of the other démarches in the field, however, it has seemed that the
best reaction is to take note but remain silent. Some will disappear
without the ministrations of comment or criticism. Others will not dis-
appear, but will seem misguided to anyone I can convince of the correct-
ness of the approach to identity that is exemplified in this book. Yet
others of more recent provenance must wait their turn to be read until
this book is given to the new publisher, who has waited long enough.
Acknowledgements to Noonan and Snowdon apart, as to Williamson
(who did me a further favour by accepting the publisher’s invitation to
review the whole draft), there are newer debts of gratitude, to William
Child, Stephen Williams, Christopher Peacocke, Naci Mehmet and Ian
Rumfitt, for instance, each of whom read some version of some chapter
or section or extract. I am grateful to New College and the Sub-Faculty
of Philosophy at Oxford for substantial technical assistance and espe-
cially to Jo Cartmell. Without her the task could not have been com-
pleted. From earlier times, a variety of philosophical acknowledgements
 Preface xi
must be carried forward that are no less real for being old: first (from the
monograph of ) there are special thanks to Professors P. T. Geach,
W. A. Hodges and B. A. O. Williams. From , there are acknowledge-
ments to M. K. Davies, E. L. Hussey, D. W. Hamlyn, R. A. Wollheim,
C. A. B. Peacocke, M. L. C. Nussbaum, D. F. Cheesman, J. A. W. Kamp,
N. Tennant, J. H. McDowell. Debts were incurred on a much larger
scale in – to Sir Peter Strawson and Jennifer Hornsby. In 
I made more general acknowledgements to various papers by Hilary
Putnam, Saul Kripke and Richard Cartwright, and to Leibniz’s Philosophy
of Logic and Language by Hidé Ishiguro. In divers and different ways, each
of these authors informed or strengthened the various convictions that
I needed in order to shape the characteristic, however insufficiently qual-
ified, claims of Sameness and Substance concerning the mutual dependence
of the ideas of substance, causality, law, and de re necessity.
In Identity and Spatio-Temporal Continuity, dark claims were entered about
the relevance and importance for the theory of individuation of the phi-
losophy of biology. In Sameness and Substance it would have been good if
there had been more about these matters. After the abandonment of
Longer Notes, all that remains here are certain sketchy remarks in
Chapters Two and Three. But I shall recall from the  Preface the
keen pleasure that I felt at that time on discovering how, in response to
all the facts that confront the biological scientist, Professor J. Z. Young
had arrived, in chapters Five and Six of his Introduction to the Study of Man
(Oxford, ), at a conception of identity and persistence through time
that is strikingly similar, where living things are concerned, to the neo-
Aristotelian conception that I defend:
The essence of a living thing is that it consists of atoms of the ordinary chem-
ical elements we have listed, caught up into the living system and made part of
it for a while. The living activity takes them up and organizes them in its char-
acteristic way. The life of a man consists essentially in the activity he imposes
upon that stuff . . . it is only by virtue of this activity that the shape and organ-
ization of the whole is maintained.
Two other good things that have happened since  are the recogni-
tion in the philosophical community at large of the persisting conceptual
importance, all foolish revivalism apart, of Aristotle’s biology and philos-
ophy of life; and the development by Peter Simons (in Parts: A Study in
Ontology, Oxford, ) of a new account of the part–whole relationship
that is far less alien to the present inquiry than the works of classical
extensional mereology that I criticize so relentlessly in Chapters One,
Two and Three.
xii Preface
In  it seemed that there were two important things I had to say about
identity and individuation. One came down to this. Identity was an
absolute relation, yet, despite this, identity was not bare continuity. A for-
tiori, neither identity nor even the identity relation as restricted to
material objects could be the same relation as continuity as such. Identity
and Spatio-Temporal Continuity () had been a first engagement in the
war against this idea, which was no less dispensable, I said, than it was
incoherent. Hence came the thesis of Sortal Dependency, labelled D in
Sameness and Substance (). This said that behind every true identity
claim there stands an identity covered by the concept of some particu-
lar kind of thing (in a wide range of empirical cases, a substance-kind).
Once D was in place, the philosophical work that remained was to show
how, in all their strictness, the formal properties of the relation of iden-
tity can be sustained by our kind-based individuative practices.
Among the further consequences derived from sortal dependency
were a modest but specifically individuative (contrast referential) form of
essentialism and the second of the special things I thought I had to say
in . This was the doubtfulness of the separation, supposedly obvious
or truistic and still widely insisted upon, between ontological and con-
ceptual questions. Here, even if some of the things rehearsed in the pre-
vious paragraph have come to be accepted as commonplace, I think I
have made scarcely any impression on received ideas about the sharp
division of questions of ontology from questions of ideology. (In Quine’s
sense of “ideology”. See From a Logical Point of View, Harvard, ,
p. .) As in the  version, the case for this adjustment is expressed
in Chapter Five and that which leads up to it. In Sameness and Substance
Renewed, Chapter Five prepares the way for some fresh reflections,
pursued further in Chapter Six, about identity, vagueness, determinacy
and the singularity of the identity relation. It is here that I think the con-
ceptualist insight I try to formulate in Chapters Five and Six shines a new
light onto certain familiar questions, questions already transformed by
Timothy Williamson’s sharp critique of received theories of vagueness.
See his recent book, Vagueness (London, ). Before the insight that I
call the conceptualist insight falls into the wrong hands, however, let me
say immediately that it is all of a piece with the absolute and adaman-
tine hardness of truth. (See Chapter Six, §.) The same goes for the
increasing emphasis placed upon the irreducibly practical aspect of our
acts of individuation. This may even amount to a third important thing
that I think I have to say.
Chapter Six makes a protest against the idea that, even if identity is
 Preface xiii
strictly irreducible, it must supervene somehow upon other properties
and relations of objects, and supervene in such a way that these will con-
stitute locally sufficient grounding for a judgment of identity. By this
protest, I position the theory of identity and individuation to see the
making of identity judgments in an altogether other way. Let us see it as
an extension of our practical capacity to single out things of a given kind
and then, in the light of an understanding of the behaviour of things of
that kind, to keep track of them. The fully fledged judgment of identity
outgrows its primitive origin but, according to my account, it does not
lose touch with the original enterprise that it extends.
If I had seen all this clearly in , if I had seen the opportunity it
affords for fresh modes of philosophical exposition, I think I might have
found a way to treat the questions of identity and individuation other-
wise than in the technical-sounding language of principles of individu-
ation, persistence, identity, activity. To dispense altogether with all talk
of such principles would have been a noble endeavour. But it would have
required a completely new book, one that strained less after generality or
that only achieved it by the extended demonstration and discussion of
eminent instances. Instead, the thing I shall say here about principles of
individuation and so on is simply this: given that any serious or ontolog-
ically committal use of language of this kind can only multiply the kind
of problems that philosophy has already with entelechies, forms, poten-
tialities, actualities, etc., and given that such use may threaten an explan-
atory regress (as Penelope Mackie has properly observed, see her op. cit.
at note , Chapter Four), all talk of such things needs to be understood
as notional. What would it be to treat it so? Well, here is a start. To see
that the principle of individuation for a buzzard is not the same as the
principle for a bat, to see that the principle of individuation for a teapot
is not the same as that for a housefly – there is no more to this (and no
less) than there is to seeing what a difference there is between these things
from the practical point of view of singling them out, of keeping track of them and of
chronicling what they do.
Chapter Seven, the new chapter on personal identity, focuses on
human beinghood, and recants anything I have ever said against Bishop
Butler’s objection to Locke’s account of personal identity. The chief aim
is to treat personal identity for what it is, namely a special case with a
special power to test any emerging answer to the general question of the
identity and individuation of substances. The chapter reviews briefly the
course of controversy on these matters since the nineteen sixties, when
a thought experiment of Sydney Shoemaker’s deflected me and many
xiv Preface
other philosophers towards the neo-Lockean conception of personal
identity. My completed recantation, which perseveres in doubts Bernard
Williams, Paul Snowdon and I have expressed over a long period, com-
prises considerations inter alia of epistemology and the cognitive activity
of human beings. On this basis, I seek to show that there is no non-
vacuous sense in which one can say “the ordinary further facts of human
personality supervene upon the facts of mental and physical continuity
and connectedness”. (Pace the philosophers who say that sort of thing,
those mental and physical facts are already identity-involving.) I must add,
however, that despite the completion of this recantation of all neo-
Lockean tendencies, I cling to my admiration of Locke’s Essay ·, not
least (now) of his “forensic” conception. This last has usually been taken
to support the Lockean over the human being conception of personal
identity. In the light of the considerations of physiognomy that I try now
to insist upon, I think that the chief contribution of the forensic concep-
tion is to make us (the persons that we are) see the difficulty of conceiv-
ing of a person (conceiving of one of us) otherwise than as a being with
a human form. Rereading the old chapter in Sameness and Substance that
the new Chapter Seven replaces, I find that this anticipated and spelled
out at great length a range of practical and moral apprehensions arising
from the prospect of other, quasi-functional or quasi-artefactual, con-
ceptions of personhood gaining ground. The fact that the ensuing
twenty years have intensified these apprehensions might be ground for
intellectual satisfaction. (For no other.) But the intervening years equally
suggest the need to condense the apprehensions themselves into one or
two bare paragraphs. At this point of the argument, the thing that
matters is the intimate connexion between such apprehensions, familiar
as they will now be to almost every reader, and the range of rival con-
ceptions, some of them artefactual, some of them (like mine) anti-
artefactual, of what kind of thing it is we are concerned to individuate
when we ask what a human person is.
Readers who wish to begin by seizing the main essentials of the
theory of identity and individuation which leads into all these other
things (or so I claim) should not labour too hard over the later sections
of the Preamble, which is mainly methodological and terminological.
Terminological explanations that are essential – and some of them are
indeed essential – are given again or referred back to as and when they
are needed in the body of the book. The chief purpose of the
Preamble is to place all these explanations where they belong, namely
in a single framework within which they will show themselves to be
 Preface xv
singly and collectively defensible. Those convinced of the wrongness of
my substantive conclusions or who object to the method of reaching
them ought, in due course, to take the precaution of reading the
Preamble through to the end.
Readers who are prepared to skip should read Chapter One, sections
–, and then advance immediately to Chapter One, sections  and ,
before reading Chapters Two and Three. A summary is given at Chapter
Three, section , of this material, just as a partial summary is given in §
and § of Chapter Five, to recapitulate Chapters Four and Five.
The chief aim of Chapters One, Two and Three of the book is to
place questions of individuation, identity and persistence through time
on a firmer and broader basis of theory, but in such a way that the par-
ticular point that is at issue in particular problems of identity will be
locally determined. Once matters are put onto this basis, there can be
securer standards (or so I claim) by which to judge in situ, on the basis of
the right kinds of consideration, the relevance or irrelevance to the given
case of empirical information that is collateral with the case. The result-
ing conception of individuation is principled, logically founded, yet irre-
ducibly practical. It is universal, in so far as it always appeals to ideas that
transcend the particular case, but also dialectical. It is dialectical not only
in respect of how it envisages any particular decision’s being reached but
in respect of the individuative practices that it justifies. Room is left for
these practices and the thing-kind conceptions that incorporate them to
proceed in a given case by considerations that are highly specific to it
(scarcely general at all). That will not prevent these considerations from
being universal in import. For the distinctness of the general/specific
and the universal/singular (or universal/particular) distinctions and the
compatibility of specificity with universality, I would refer to R. M.
Hare, Freedom and Reason (Oxford, ). (See his p. .) Even at this late
stage in the specialization of philosophy, light can still be cast on logic
and metaphysics from ethics and the philosophy of law.
The explanation of how the conception thus formed of identity and
individuation coheres with the invincible strictness of the laws of iden-
tity is completed in Chapter Six, which resumes and extends some of the
arguments of Chapters Four and Five. Finally, the last part of Chapter
Seven (§ onwards) offers certain general reflections about identity and
individuation and follows them through. In these concluding reflections
I see some culmination of the efforts of all earlier chapters. I hope that
this part can be read on its own without the preceding sections of
Chapter Seven.
xvi Preface
The price of making the book skippable in this way is paid by the
reader who reads it right through. I have tried however to keep to the
barest minimum the amount of repetition that is entailed by the policy.
The purpose of the Select bibliography is to include a selection
(updated ) of certain major and minor classics of the theory of iden-
tity and individuation and to make reference to other works that the
reader may find useful or on which this book most heavily depends. Only
incidentally is it a bibliography of personal identity or of anything else
besides the theory of identity and individuation. Numerous other useful
or fascinating items not included in this selection are referred to in the
footnotes. I know that many books and articles left out of the Select bib-
liography are just as good as those I have included. Philosophers hate to
contemplate such contingency, I know, but the sole aim has been to make
this bibliography short enough for it to be useful, useful in its own right
or usefully cognizant of the particular intellectual debts it happened the
author incurred in writing or rewriting this book. The author/date
system is used for references to titles included in the Select bibliography.
The chapter footnotes are part of the final defences of the theory, but
they are meant to be theoretically dispensable to the basic understand-
ing of the argument. (One regrettable departure from this policy
remains, at footnote  of Chapter Three.) But there is no attempt to push
into the text everything which points at something important. Most
especially I have not attempted this where the matter in the note leads
not back to the argument of the text but outwards from it. An example
of that is the brief discussion in footnote  (formerly ) of Chapter One
of some of the differences between a substance and an event. Another
example is the equally old footnote  of Chapter Three (now numbered
), concerning that which I regard as the chief falsehood in the classi-
cal or original form of mereology or the calculus of individuals.
The Index is intended to secure the sense of key technical terms,
printing in bold the page reference that best indicates what acceptations
I have aimed, for the length of a book, to assign stably and definitely to
certain technical terms used here.

December  ..

New College
Oxford
 Preamble, chiefly concerned with matters methodological
and terminological

Technical terms are worse to be shunned than dog or snake.

(Leibniz, Gerhardt , .)

.   

The chief aim of this book is to elaborate a theory of the individuation
of continuants, including living substances and other substances. Such
a theory ought to comprise at least three things: an elucidation first of
the primitive concept of identity or sameness; second, some account of
what it is for something to be a substance or continuant that persists
through change; third, an account of what it amounts to, practically and
cognitively, for a thinker, to single a thing out at a time. Here, with this
last task, there is the supplementary question of what it amounts to for
the same thinker, having once singled something out, later to single out
that same thing as the same thing.
From a philosopher’s attitude towards the logical and methodological
ordering of these tasks one can tell something about his or her attitude
towards the idea that the meaning of a word is a function of its use. In
this work, it is everywhere accepted that the meanings of such words as
‘same’, ‘substance’, ‘change’, ‘persist’ and ‘recognize’ depend upon their
use. The life and semantic identity of such terms is only sustained by the
activity of singling out or individuating. But the thesis of meaning as use
is consistent with two converse or complementary theses (A)(B), which
have an equal relevance to what is to be attempted and an equal claim
upon rational acceptance.
(A) The relation between the meaning and the use of such words as
‘same’, ‘substance’, ‘change’, ‘persist’ is in fact reciprocal or two-way.
Everything that concerns meaning registers upon use; but, unless we

 Preamble
redefine use, that does not imply that meaning can be reduced to use.1
Among the concerns I began by enumerating, there is no question of
collapsing the first two into the third, for instance.
(B) An interpretation of a set of linguistic uses or conceptual practices
must speak of the subject matter to which they relate. For that reason, it
must refer to the various things themselves towards which the uses or
practices themselves are directed, together with the properties and rela-
tions of these things. The child who is learning to find for himself the
persisting substances in the world, to think the thoughts that involve
them and recognize the same ones again, grasps a skill and a subject
matter at one and the same time. A philosopher who seeks properly to
understand those thoughts must proceed accordingly. Let the philoso-
pher elucidate same, identical, substance, change, persist, etc., directly and
from within the same practices as those that an ordinary untheoretical
human being is initiated into. At the same time, let the philosopher show
by example what good elucidations can be made of such ideas as these.
To this end, let him shadow the practical commerce between things
singled out and thinkers who find their way around the world by singling
out places and objects – and singling out one another. If the meaning of
the terms ‘same’, ‘substance’, ‘change’, ‘persist’, etc., is a function of use
and use is a function of the said commerce, then one by-product of this
mode of elucidation will be that the task I began by calling the third task
is undertaken in concert with the first and the second. The first and
second tasks acknowledge the importance of the third; but, by their con-
stant appropriate acknowledgment of this importance, they will in fact
absorb the third.
When the reciprocities and mutual interdependencies of concept,
practice and thing-singled-out are acknowledged and likened to those of
some seamless web, when the primitiveness of all the relevant notions is
acknowledged too, how much genuine clarification is it reasonable to
expect a philosophical theory of individuation to be able to achieve?
Well, we have rudimentary pretheoretical ideas of identity, persistence
through change, and the singling out of changeable things. By means of
these, we may arrive at a provisional or first explication of what ‘same’
means and of the actual application of this relation-word. So soon as
that is achieved, there is a basis from which to scrutinize afresh and then
11
Or even to correct and truthful use, which would be a less striking achievement. (I hear someone
scoffing at the distinction between use and correct use. Let them note that the correct use of a
word or device might only be determinable from within a whole practice, yes, but without its fol-
lowing from this that the correct use was determinable from within practice in respect of this word
or device. Cognate questions are pursued further in my (b).
 Preamble 
consolidate our logical and participative understanding of the individu-
ative practices that a thinker’s grasp of the concepts of substance, same-
ness and persistence through change makes possible for him. At the end
of this second phase, nothing will be recognizable as the philosophical
analysis of ‘⫽’. But no special mystery need remain about how a notion
of the exigency that we ascribe to the identity relation can find applica-
tion in the changeable world of our experience. Provided we do not
despise the ordinary ideas by which we conduct the untheoretical busi-
ness of the individuation and reidentification of particulars, we can
remind ourselves well enough of what regulates the principled employ-
ment of ‘⫽’. We can remind ourselves of what it is for anyone who is
bent on singling out objects to carve off from the world, or isolate from
among the objects of his experience, various continuants or things that
persist through change.
This emphasis on the practical does not mean something that it might
seem to mean if, in the cause of the crudest version of ‘meaning as use’,
a separate priority were accorded to the third of the three tasks enumer-
ated in the second sentence of the first paragraph. It does not mean that,
for the benefit of his deluded subjects, the theorist is to find a way to see
a world that might as well be one of pure flux in which nothing really
persists through change as if that world offered us objects that persist
through change. For persistence through change is not make-believe. No
sensible inquiry could abandon a datum so fundamental or so deeply
entrenched.2 It means that, arriving at the point programmatically
described, the theorist is to understand as well as he can – discursively,
practically, in the same sort of terms as those who individuate them or
in modest extrapolation from these – what it is for an object to be a
genuine continuant; it means that, when that is done, the theorist is to
describe how the charge that something did not persist is to be consid-
ered, namely on its merits, such merits being set out in terms accessible
in principle to those who take themselves to believe in genuine continu-
ants. It is in this way that we shall try to identify the point properly at
issue in some of the most bitterly contested questions of identity.
There are two complaints about the method of elucidations that will
not go happily together: () that the method is vacuous, a mere replay of
that which needs to be ‘explained’; () that the demands which the
method derives from the congruence and other properties of identity
and translates into requirements upon the positive finding of identity are
12
Nor could flux as such, or as coherently conceived, stand in the way of singling out changeable
continuants. See my ().
 Preamble
draconian, too exigent, too severe. You cannot make both complaints.
But it is better (I hope to show) not to make either.

.     

Where there is reciprocity or mutual presupposition between concepts,
analytical philosophy is always tempted into violence or arbitrariness.
We find it hard to endure the thought that, in the substantive questions
of philosophy, there is no master thread we can pull upon to unravel
everything else. Even as I deprecate this idée fixe, however, it may appear
that the chapters which follow are victim of the same illusion. For in this
book the formal properties of identity, namely the reflexivity of identity
and Leibniz’s Law (registered in the claim that, if x is the same as y, then
whatever is true of x is true of y and whatever is true of y is true of x),
will be treated as enjoying a special status. In this way, am I not attempt-
ing to insulate from legitimate criticism my opinion that these formal
properties determine what can count as someone’s singling out or
tracing an entity? In the presence of doubt concerning formal proper-
ties, is it not simple dogmatism for me to persist in saying (in effect) that
the properties of identity regulate, by reference to a claim they make
upon reason, the interpretation of thought and action as thought and
action?
In partial answer to this charge, I can only plead that something is
done in the course of Chapter One, §, to justify the view I take of the
formal properties of ‘⫽’. I do not really think they are given simply ab
extra. It is true that I liken the status of reflexivity and congruence (along
with the symmetry and transitivity that they entail) to that of the Law of
Non-Contradiction. But, bracketing Chapter One, §, my conciliatory
view would be that the issue between the opposition and me is holistic
and dialectical. If that is right, however, then the question at issue cannot
really be resolved until some opposing account of individuation is devel-
oped to the same point as the account presented here. These questions
will not be resolved until rival descriptions of individuation (and of ref-
erence) are compared with one another against the background of all
the practices that they purport to describe. (For one small step in that
direction, see Harold Noonan’s and my exchange in Lovibond and
Williams ().) The thing I have to hope is that, in the end, the reader
will convince himself that the internal difficulties of the ontology and
ideology of a position that abandons the Leibnizian conception of iden-
tity are overwhelmingly greater than any of the difficulties attaching to
 Preamble 
the position I recommend. I trust that at that point, if not before, the
reader will come to share my conviction that the Leibnizian principle is
immanent in any linguistic or reflective practices we can recognize as ref-
erence and individuation.

 .      
Corresponding to the three tasks mentioned in the first paragraph, we
have the notions identical (same), continuant and individuate.
(i) The notion of sameness or identity that we are to elucidate is not
that of qualitative similarity but that of coincidence (as an object, thing
or substance), a notion as primitive as predication and correlative with
it in the following way: if and only if Socrates is a man, then Socrates is
identical with some man, and thus (we shall argue) shares all his proper-
ties with him. (This equivalence is offered as a manifest truth, rather than
as an analytical definition of ‘is a man’ or of anything else. It is not
offered as a part of a canonical or mandatory definitional sequence. See
below, §.) No reduction of the identity relation has ever succeeded.
(See especially Chapter Six, §.) Nor yet is it called for, once we realize
how much can be achieved in philosophy by means of elucidations that
put a concept to use without attempting to reduce it but, in using the
concept, exhibit its connexions with other concepts that are established,
genuinely coeval or collateral, and independently intelligible. (Compare
here Wittgenstein, Tractatus Logico-Philosophicus, ., ., ..) Not
only is identity irreducible. Only in a vacuous sense of ‘supervene’, or a
weak and irrelevant one, does it supervene on the totality of properties
and relations other than itself. (See Chapter Six, §.)
(ii) We have to explicate what it is to be a continuant or a substance.
This explication will not amount to a definition. Nor will it be achieved
without the ineliminably practical demonstration of the ordinary per-
ceptible individuals of common experience. The explication must go
some way beyond mere demonstration. But to set out, as so many phi-
losophers have done in emulation of Book  of Aristotle’s Metaphysics,
with the high-minded aspiration to achieve an altogether purer kind of
definition of substance, and then to abandon the concept of substance
just because the result does not satisfy, is to end up doing philosophy that
is at once ill-tempered and needlessly bad. It represents the inability to
learn from Aristotle’s experiment.3

13
For my own attempts to learn from it, see my ().
 Preamble
Kant writes at § of Prolegomenon to Any Future Metaphysic: ‘People have
long since observed that in all substances the proper subject, that which
remains after all the accidents (as predicates) are abstracted, remains
unknown.’ I protest that the substances or subjects we begin with are not
unknown but known, that the only abstraction in which we need to be
interested is utterly distinct from that which is supposed to result from the
notional (mythical) removal of properties from a substance. The interest-
ing and benign form of abstraction is that which results from the ascent
from particular kinds of substance to the determinable substance of some
further specifiable kind. (Ascent to what Wittgenstein in Tractatus Logico-
Philosophicus called a formal concept.) This form of abstraction cannot part
us from our conviction that substances are things which are known to us.
(iii) The Oxford English Dictionary defines ‘individuate’ in terms of
‘single out’ or ‘pick out’, and this definition is well suited to the purposes
of this book. That which individuates – in the one sense in which the
word will be used in this book4 – is in the first instance a thinker.
(Derivatively, but only derivatively, one may find oneself saying that a
substantive or predicate individuates.) To single x out is to isolate x in
experience; to determine or fix upon x in particular by drawing its
spatio-temporal boundaries and distinguishing it in its environment
from other things of like and unlike kinds (at this, that and the other
times during its life history); hence to articulate or segment reality in such
a way as to discover x there. To single x out though, or even to prolong
the singling out of x into the effort to keep track of x, is not yet (unless
‘in thought’) to refer to x or to designate x. And one may well refer to x,
of course, without in our primary sense singling x out at all. This is not
to say that, if there were no singling out, there could be reference.
Singling out is the sheet-anchor for information about particulars.
The verbs ‘individuate’ and ‘single out’ are not intensional. If a
thinker singles out x or individuates x, and x⫽y, then, whether or not he
knows it, he singles out or individuates y. Such verbs do, however, permit
of a complementation that is intensional. A Greek could have simply
singled out Socrates; he could have singled out Socrates as Socrates; he
could have singled out Socrates as a certain man or philosopher; or he
could have singled out Socrates as the Athenian married to Xanthippe
who was represented by Plato to have stressed (Phaedrus e) the equal
importance, in classification and in carving, of ‘dividing where the joints
are’. What then is the relation of singling out and singling out as? In due
14
Contrast books about logic or metaphysics where the verb is used to stand for the relation
between a predicate and some unique thing that satisfies the predicate.
 Preamble 
course, we shall discover reason to think that there could be no singling
out tout court unless there could also be singling out as. (This is not a pri-
ority claim.) It will be declared that not just any attempt at singling out
counts as singling something out; that that which is required in a given case
derives from what the thing itself is. It will be a consequence of the
account of these matters to be given here that, for a thinker to single out
or individuate a substance, there needs to be something about what he
does, something about his rapport with x or his relational state towards x
and his practical sensibility in relation to x, which (regardless of whether
he articulately knows this or not – for all he needs is clear indistinct
knowledge, cf. Chapter Three, note  and associated text – and regard-
less of whether it is a singling out as) sufficiently approximates to this: the
thinker’s singling x out as x and as a thing of a kind f such that member-
ship in f entails some correct answer to the question ‘what is x?’ For the
philosophical cargo carried by this Aristotelian question, see Chapter
One and the chapter mottoes prefixed to it from Aristotle’s Categories.
One further and equally Aristotelian part of that cargo makes reference
to the way in which x behaves, how it acts and reacts. It will be every-
where insisted, moreover, that the singling out at time t of the substance
x must look backwards and forwards to times before and after t. And it will be
categorically denied in Chapters Five and Six, that, where it is indeter-
minate what was singled out, we have the singling out of something
indeterminate. (Even at this distance the thing denied has the distinctive
smell of fallacy.) But at this point in summarizing what is to come, I
venture well beyond explanation of terminology and deep into the phi-
losophy of the matter. Chapters Five and Six aim to complete the
account of what singling out is. If they succeed, it will become finally
clear how and why the singling something out at t cannot help but look,
as I say, both backwards and forwards to times before and after t.
In sum, let the English language fix what will be meant by ‘single out’
and ‘pick out’. Let these verb phrases sustain the practical and episte-
mological significance of ‘individuate’, ‘individuation’ and ‘individua-
tive’. Let philosophy then seek to say what individuative acts and
thoughts amount to. At this point, a reader who has had enough of pre-
liminaries may want to advance to Chapter One.

.        :      

The explications just given are intended to leave room for me to make the
following declaration. Ideally, all technical terms should (i) be defined and
 Preamble
(ii) belong in that part of the metalanguage which does not overlap with
the object language. Where there is no alternative but to allow technical
terms to penetrate into the object language (e.g. because the object lan-
guage is poor in schematic devices or devices of generalization), one might
hope that technical terms would serve the sole purpose of abbreviation, of
summarizing, and of systematizing, in terms not essentially different from
the expressions indigenous to the object language, the matters of which the
object language already speaks. No doubt the philosophy of any particu-
lar science or art will need to use the technical terms of that science or art.
But such terms will have needed to pass muster in that art or science itself.5
The semi-technical uses in this book of ‘concept’, ‘continuant’, ‘sub-
stance’, ‘coincidence’, ‘coincidence under a concept’ will stand con-
demned unless they can conform to these requirements. Maybe they will
not always live up to the ideal stated, and will to that extent stand con-
demned. But my aspiration for them is that they should be devices for
the generalization of that which has a straightforward meaning in the
object language of English – more specifically, that they will be determin-
ables of which ordinary English provides countless determinations.6 This is
everywhere important, but it is a particularly important stipulation in
connexion with the term ‘substance’. If we misunderstand determinable
notions such as this, then it is almost inevitable that we shall unintention-
ally restore the unwanted associations of ‘substance’ with doctrines of
bare particulars and qualitiless substrate.

.               : 

           
A technical term that is associated with ‘substance’ and the what is it?
question but belongs in the metalanguage is ‘sortal predicate’. I use this
15
In stating that these are the ideals to which I regard myself as answerable, I am not venturing to
condemn all philosophy that disregards them or follows some other manifesto. It is enough for
me to say that the badness of much philosophy that is bad by almost any standard can be partly
explained as the effect, inter alia, of disregard for such maxims – or of utter nescience of them.
16
The determinable/determinate distinction was revived by W. E. Johnson (Logic .XI.; Cambridge,
) out of dissatisfaction with the traditional genus and differentia account of species when it
was applied outside its traditional scope. ‘To be ultramarine is not to be blue and something else
besides, but it is a particular way of being blue’, A. N. Prior, The Doctrine of Propositions and Terms
(London, ). Pace the traditional doctrine of genus and differentia, I should say the same of being
a cat. It is not a matter of being an animal and something else that is independent of animality.
See below on real definitions, §. Manifestly ‘substance’ stands for a fundamentum divisionis in the
traditional scheme. Or, in the language of the Tractatus, one may prefer to say it stands for a
formal concept. For further discussion, see A. N. Prior, op. cit., pp. – and ‘Determinables,
Determinates and Determinants’, Mind,  (), pp. –, –.
 Preamble 
Lockean term in roughly the manner of the second part of P. F.
Strawson’s Individuals (London: Methuen, ). See especially his pp.
–. (For a discrepancy not of philosophical purpose but of detail, see
Chapter One, section  below.) Locke’s usage, Strawson’s usage and my
own are all focused or organized by Aristotle’s distinction of predications
in the category of substance from predications in the category of quality
and the other categories. See the first five chapters of Aristotle’s treatise
Categories, especially the two passages I have prefixed to Chapter One.
For Locke’s usage, see Essay , iii, :
it being evident that things are ranked under names into sorts or species only as
they agree to certain abstract ideas, to which we have annexed those names, the
essence of each genus or sort comes to be nothing but that abstract idea which
the general, or sortal (if I may have leave so to call it so from sort, as I do general
from genus), name stands for.7
Here, as in other cases, the intuitive semantics we reach for in replace-
ment of the Lockean system of ideas are Frege’s or some adaptation of
these. (For, however unfinished Frege’s original scheme may be and
whatever reservations one may have about the further elaborations that
he offers of it in Grundgesetze, the underlying ideas are as general as they
are durable.) Like other predicables, a sortal predicate expresses a sense
and, by virtue of expressing this sense, it stands for a concept. Under this
concept individual things may fall. See the diagram in Frege’s letter to
Husserl.8 To understand a predicate and know what concept it stands for
is to grasp a rule that associates things that answer to it with the True
and things that don’t answer to it with the False. (The extension of the
concept is therefore the inverse image of the True under the function

17
At § (‘Divided reference’), Quine () notes the following variants for ‘sortal predicate’: ()
individuative predicate; () articulative predicate; () substance-name; () shared, or multiply
denotative, name; () predicate which divides its reference (extension). Another variant that has
had some currency, on which see Woods (), is () boundary drawing predicate. (Cf. Frege
(), §.) All six terms serve to illuminate the difference, partially but only imperfectly reflected
in the grammatical division of noun and adjective or verb, between Aristotle’s ontologically basic
question What is x? and less basic questions such as What is x like? Where is x? What is x doing? Note
that looking at these terms in this Aristotelian way will enforce a diachronic interpretation of ‘indi-
viduate’, ‘articulate’, etc. We shall not be in the business of describing first what it takes for syn-
chronic momentary presentations (things presented) a and b to be the same dog and then
describing what it takes for a presentation now and a presentation tomorrow to be ‘concanine’.
Identity over time is just identity. The same holds of identity at a time. Such truisms should con-
dition any account of the terms of a given identity judgment. Any secure practical grasp of what
counts now as a dog regulates present judgments in the light of future and past findings about
the same thing. And vice versa. See my ‘Reply to Noonan’ in Lovibond and Williams ().
18
The letter is dated . See Dummett (), Chapter Five. The diagram is reproduced in my
() and my ().
 Preamble
determined by this rule.) To grasp the rule is to grasp how or what a thing
must be (or what a thing must do) in order to satisfy the predicate. To
grasp this last is itself to grasp the Fregean concept. Thus ‘horse’ stands
for that which Victor is and Arkle is, for instance – just as, outside the
sortal category, the verb-phrase ‘runs swiftly’ stands for that which Arkle
does. When I declare that to grasp this rule is to come to understand
what horse is or run swiftly is, someone may insist that, in that case, the
concept so spoken of, horse or run swiftly or whatever it may be, is a prop-
erty. I shall not demur, but simply insist in my turn that the notion of a
rule of correlation to which I appeal is pretheoretical. It is not indissol-
ubly wedded to an extensional criterion of concept identity. The exten-
sional criterion is the by-product, not here needed, of the
mathematicians’ regimentation of an entirely intuitive notion.9
The concept horse is not then an abstraction such as horse-hood or
horse-ness (whatever these are). It is something general or, better, univer-
sal; and to that extent it will be philosophically contentious. But horse or
mammal or carnivore surely are things that we need to speak of or quantify
over, in metaphysics and in science.10 Objects fall under them and so on
– and, under this aspect, objects can be seen as belonging to divers
assemblages, variously denominated species, sorts, kinds.11
Seen in this way, as something with instances, the concept belongs on
the level of reference (reference in general being something of which
naming is one special case). But there is another use of the word
‘concept’ which is equally common, if not more common, and this
belongs on the level of sense. It is this rival use of the word ‘concept’ that
we find in discussions that are influenced directly or indirectly by Kant.
In those discussions, talk of things falling under a concept, or of con-
cepts having extensions, may be less felicitous. Or rather, it will not come
to the same thing. Perhaps everything will fall into place, however, and
the connexion will be visible between the two uses of the word, if we try
to reserve the word ‘concept’ for the Fregean use and we prefer the word
‘conception’ to cover the Kantian use (seeing a Fregean sense as a very
special case of a conception). The connexion that there is between the
two may then be understood as follows:
19
See B. A. W. Russell, Introduction to the Mathematical Philosophy (London, ), p. , and the
further references to Ramsey, Quine and Church given in Aaron Sloman’s neglected but valu-
able article ‘Functions and Rogators’ (). See especially pp. , , .
10
For more on these, see my (), especially the references to Elliot Sober, ‘Evolutionary Theory
and the Ontological Status of Properties’, Philosophical Studies,  (), and my (). The
quantification in question is over both sortal and non-sortal properties.
11
In ordinary English and even in ordinary philosophical English, some of these terms lead a
double life perhaps, as denoting assemblages or as denoting properties.
 Preamble 
Thinker T has an adequate conception of the concept horse (an adequate con-
ception, as one says in English, of a horse) if and only if T can subsume things
under horse, knows what it would take for a thing to count as a horse, and has
some sufficiency of information about what horses are like, etc.
In a word, the conception of horse is a conception of that which ‘horse’
stands for, namely the concept of horse, or the concept horse. (Similarly,
one may speak of an idea of horse and mean by this a conception of the
concept horse.) On these terms, the right way to understand what a phi-
losopher means when he speaks of grasp of the concept horse may be to
understand him as speaking in a telescoped way of having an adequate
conception of that which the predicate ‘horse’ stands for, namely horses.
Such phraseology as ‘grasp of the concept horse’ was common in Sameness
and Substance. Given the option of reading it as just indicated, there are
places where I have seen no danger in the telescoped terminology’s
remaining in place.
At this point, with terminological explanations more or less complete
(and ready to be repeated later, as and when necessary), the reader who
has not yet postponed reading the rest of this Preamble ought probably
to advance to Chapter One.

.      

Next, in anticipation of something laid out in Chapter Three, a word on
the terminology of ‘real’ and ‘nominal’. The use of these terms turns on
the distinction between a predicate whose elucidation makes inelimin-
able allusion (or must begin by making ineliminable allusion) to members
of its actual extension (real)12 and a predicate whose elucidation can dis-
pense with such allusion (nominal). In the case of the nominal, the elu-
cidation can often be made fully explicit or articulate. In the pure case
of the nominal, it never needs to involve the deictic or demonstrative
element that is so typical of the real definition. A typical nominal defi-
nition would be Aristotle’s formula ‘x is a house if and only if x is a
shelter against destruction by wind, rain and heat’. By contrast, a typical
real definition – I call it a definition only to follow custom, this is not
really definition – might run in the following way: ‘x is a water-vole just
if x is relevantly similar to that [demonstrated] animal (letting the criter-
ion of ‘relevantly’ depend at least in part on the nature, whatever it may
12
Pending the completion at some imaginary moment of a science, cf. Chapter Three, section ,
point (v). Compare the way in which the elucidation of the sense of a proper name is designatum-
involving. Cf. here ‘On the Sense and Reference of a Proper Name’ (), by John McDowell,
to whom this Preamble is indebted also in other ways, and cf. my ().
 Preamble
prove to be, of the animal actually demonstrated)’.13 Not only does the
real definition depend on real specimens. The question of its application
and correctness or incorrectness also depends crucially on the facts
about these specimens. All the same, real definitions lie within the prov-
ince of semantics, as well as of empirical fact. No lifelike elucidation of
ordinary discourse could be contrived without them.14 Let us forget once
and for all the very idea of some knowledge of language or meaning that
is not knowledge of the world itself. In this way, let us follow through the
capital thing which Quine contended for in ‘Two Dogmas of
Empiricism’.15

.     ⁄       A PRIORI ⁄ A POSTERIORI

It is to be expected that there should be concepts any passable concep-

tion of which requires experience. Such concepts are a posteriori or empir-
ical. The semantics of a predicate that stands for such a concept (‘elm’,
‘oak’, ‘rabbit’, ‘brazen’, ‘green’, for instance) will be rooted in the actual

13
A definition can be real with respect to genus and nominal with respect to differentia. The philoso-
phers’ favourites like ‘sibling’, ‘oculist’ and ‘bachelor’ have definitions like this.
14
There may be a measure of injustice to Locke (as well as to certain other philosophers) implicit
in this usage. By using terms that are commonly conceived of as Locke’s property and stressing
constantly the importance of the real, I may seem to be claiming that, in his adherence to what
he called the nominal, Locke failed to see things which some say he did see. Locke’s doctrine of
ektypes and archetypes, and the way he contrasts ideas of mixed modes with ideas of substance,
both suggest not only that he had a grasp of the question whether a given predicate has an exten-
sion-involving or not extension-involving sense, but also that he may have anticipated the ideas
of onus of match and direction of fit which, following Michael Woods (op. cit., in footnote ) and
David Shwayder (), p. , I have found so suggestive in J. L. Austin (‘How to Talk: Some
Simple Ways’, P.A.S. –) and have allowed to influence the exposition given here of certain
thoughts that now have currency in philosophy about substantives and kinds. Such ideas cer-
tainly go back to Leibniz. See Leibniz’s Meditations on Truth, Knowledge and Ideas, Gerhardt , espe-
cially p. . See also Discourse on Metaphysics (sections –) and New Essays (Akademie, pp.
–). Some see the germ of all this in Aristotle’s account at Posterior Analytics a21 of how we
may effect a preliminary determination of a phenomenon (e.g. a certain noise in the clouds) and
then, having picked it out, gradually refine our description of its nature and work back to the
rudiments of a real (or as Leibniz would say causal) definition. It may well be that some com-
parable credit is due to Locke, if he is correctly interpreted. Chapter Three credits Putnam and
Kripke with refurbishing such ideas for present purposes.
15
See From a Logical Point of View, pp. –. This view of that which is enduringly important in
Quine’s argument accrues to me from remarks and commentaries that Hilary Putnam has
offered since the nineteen sixties. For the consistency of this thing with the modest essentialism
that will be advocated in this book (from Chapter Four onwards), see Quine’s later judgment in
Ways of Paradox ( edition): ‘There is also in science a different and wholly respectable vestige
of essentialism or of real definition . . . It consists in picking out those minimum distinctive traits
of a chemical, or of a species, or whatever, that link it most directly to the central laws of science.
Such definition . . . is of a piece with the chemical or biological theory itself [and] conforms
strikingly to the Aristotelian quest . . . This vestige of essentialism is of course a vestige to prize.’
 Preamble 
world. (See the previous paragraph.) Nothing, however, prevents the
occurrence of such concepts in judgments that are themselves a priori.
For instance, ‘all rabbits are rabbits’ requires (of anyone who under-
stands it at all) a certain quantum of experience; but, once a thinker has
got a posteriori that which is required to understand it, he needs nothing
more to see that it is true. In this sense, the judgment, whose compre-
hension is in part a posteriori, is itself a priori. Moreover, it is true no matter
what. So it holds necessarily. Notoriously, though, not everything that
involves such concepts yet holds necessarily can be a priori. Pursuing the
deictic demonstration of gold and investigating the relevant similarities
of good specimens of the stuff that is denominated ‘gold’, physical
chemists have found ways to light upon the element with atomic number
seventy-nine. Since that is the very stuff they were directed to by the
deictic demonstration, it seems gold is necessarily the stuff with atomic
number seventy-nine. (Use Russell’s theory of descriptions and
Smullyan () to understand the grammar of this claim.) But the expe-
rience that it takes to understand the semantics of ‘gold’ was insufficient
to establish this identity. It holds necessarily but a posteriori. The same
surely holds good for ‘Lucifer is Noctifer’. The a posteriori basis on which
each of these names is understood does not suffice to establish that
Lucifer is Noctifer. Only when that identity is established empirically can
modal reasoning convert this discovery into the discovery that there is no
circumstance in which Lucifer is not Noctifer. The identity holds neces-
sarily – necessarily but a posteriori.
These matters are familiar, even if they still lie on the margin of con-
troversy. But, beyond those treated by Saul Kripke, certain adjacent
questions are less familiar. Leibniz writes in On Nature Itself (G , , fol-
lowing §)
Matter resists being moved by a certain natural inertia (as Kepler well calls it)
so that . . . it requires for its motion an active force proportional to its size . . .
Just as matter has a natural inertia which is opposed to motion, so in a body
itself, and indeed in every substance, there is a natural constancy which is
opposed to change.

The point Leibniz draws to one’s attention here is a perfectly general

condition on the determinable concept material body or substance.
Relatively few ordinary thinkers are in a position to state this condition
articulately or with precision. But a person who had absolutely no con-
ception of this mark of the concept material body or substance, or who
expected no resistance at all from things that answered to determinations
 Preamble
of it (such as apple, wagon or tree), would be hard to credit with any ade-
quate conception of these concepts (either of the determinable material
substance or of such determinations of it).16
What then is the status of the truth ‘a material substance has inertia
and requires for its motion an active force proportional to its size’? It is
a candidate to hold necessarily. Nevertheless, the truth is scarcely ana-
lytic. It does not follow from logic and definitions. (What definitions are
available here?) The truth is grounded in the empirical concepts material
substance and inertia. It is grounded there in such a way as to leave no room
(apparently) even for the possibility of a material substance without
inertia. Moreover, for anyone who thinks through what is involved in a
substance’s lacking inertia, the very idea of a material substance’s
lacking inertia may seem to be excluded. Some philosophers may even
react to this by calling the truth that material substances have inertia an
a priori truth. If I do not call it that myself, it is because I strongly suspect
that, in thinking through what would be involved in a substance’s lacking
inertia and finding the way blocked to that state of affairs, one calls upon
experience for a further contribution, over and above that which was
required to grasp the empirical concepts that are involved.

 .        
In the ensuing chapters, there are notations and locutions that may give
the impression that, despite protestations to the contrary, I am engaged
in the business of constructing a formal system. But the only aim of these
notations is to effect abbreviation where abbreviation is needed, or to
make fixed and transparent whatever is taken in the context to be the
logical and inferential character of certain antecedently familiar forms
and locutions. In the case of notations such as ‘ ⫽ ’, this disclaimer
a b
donkey
is to be borne particularly in mind. This simply abbreviates ‘a is the same
donkey as b’, where that is to be understood as having whatever meaning
the language imparts to these English words. Again, at the outset of the
inquiry in Chapter One, the question is put whether x can be the same
f as y without being the same g as y. But there is no policy here to dimin-
ish dependence on informal English. To make a question in English out
of this schema, for instance, replace the letters ‘ x ’ and ‘ y ’ by names of
things and the thing-kind letters ‘f ’ and ‘g’ by predicates denominating
16
Cf. Christopher Peacocke, ‘Intuitive Mechanics, Psychological Reality and the Idea of a
Material Object’, in Spatial Representations, ed. Eilan, McCarty and Brewer (Oxford, ).
 Preamble 
kinds of thing, e.g. ‘donkey’, ‘horse’, ‘tree’, ‘mammal’, ‘animal’, ‘beast of
burden’. In sum, the codifications proposed should be seen as an invita-
tion to agree in finding a certain meaning in natural language expres-
sions of which the notations are the counterpart, and then to treat these
notations as specifying that meaning canonically.
Notation such as this, latching as it does onto one particular under-
standing of the reusable words that make up some phrase or latching
onto one particular reading of the construction in which the words are
placed, represents, if you will, a first step in the direction of a system-
atic account of the logical grammar of English. But it is no more than
a first step. Much has been learned from experiments with an approach
that is opposite to mine, i.e. with a would-be exact or calculus-building
approach. This has advanced some matters by raising questions not
previously encountered and identifying all sorts of possible answers to
these questions. Yet all too often this top down approach has had the
effect of precipitating its proponents rather suddenly into issues for
which there had been no discursive or presystematic preparation. In
this book, by contrast, the approach is that of the underlabourer –
albeit an underlabourer not denied the distant prospect of discerning a
simple Leibnizian order implicit in the syntax and semantics of
English.

.                ‘        

  ’
In the explanation described as a first step, I said how ‘a is the same f as
b’ may be read as a sort of English. This explanation was general. But
this particular form is, for purposes of this book, special and specially
important. What is the relation between the mode of combination by
which ‘donkey’ occurs in ‘a is the same donkey as b’ and that by which
‘donkey’ occurs in ‘a is a donkey’? For the sake of completeness, we need
to address this question. Logically speaking, the matter belongs here;
even though, as so often with the questions treated in this Preamble, one
who does not care yet about the answer may wish to skip or go on now
to Chapter One.
According to Quine, ‘a is the same donkey as b’ is simply a contrac-
tion or condensation of the logical form ‘a is a donkey and a is b’.17 Thus
‘– (is) (a) donkey’ should be read as a predicate distinguishable within the

17
See Quine review of P. T. Geach, Reference and Generality (); also Perry ().
 Preamble
general class of predications only by the non-syntactic marks insisted
upon at § above. (These marks are further discussed in Chapters One,
Two and Three.) A variant that I think Quine might tolerate, and
(having scrutinized all extant alternatives) I should myself prefer it, is to
see ‘a is the same donkey as b’ as saying that a stands to b in the relation
of identity as restricted to (things that are) donkeys. Hence our notation in
a ⫽b
this book: . In Russell and Whitehead’s official Principia
donkey
Mathematica notation (see Volume , *), the approved rendering would
have been (a) ⫽[ donkey (b).
The advantage of proposals such as these is that they simultaneously
show forth the distinctive contribution of ‘donkey’ to the linguistic forms
that the word ‘donkey’ helps to make up and show it as made by means
of a single mode of combination. Suppose that for one moment we rep-
resent this mode of contribution by the invented verb ‘to donkey’. Then
‘a is a donkey’ becomes ‘a donkeys’ and ‘a is the same donkey as b’
becomes ‘a donkeys and a⫽b’ or else (essentially similarly, but in a way
that brings ‘⫽’ and ‘donkey’ better into construction together): a has to
b the relation of identity as restricted to things that donkey, or a has to b
the relation same [ donkey.
So far so good. Now, helping ourselves to similar verbs for other sub-
stantives, we can quickly make sense of ‘there is something that a is’, ‘a
is the same what as b?’, ‘a is the same something as b’, and so on. For pur-
poses of such contexts, the letters ‘f ’ and ‘g’, which have served so far as
schematic, will need to take on the role of true variables,18 ranging over
the general things which we claimed to discover that thing-kind words
stand for. Examples would be person, plant, tree, house, horse, donkey,
etc.19 On these terms, if we continue with our temporary representation
of nouns by verbs, we can say that ‘a is something’ is true if and only if
a something-s; that a something-s if and only if for some f (one of these
general items) a f-s, or (᭚f)(fa). Similarly, ‘a is the same something as b’ is

18
If we were constructing a formal system, it would be a heinous offence to let the same letters f
and g do duty sometimes as schematic letters, understood by reference to the role of holding a
place for their designated linguistic replacements, and at other times as genuine variables, under-
stood by reference to whatever items ‘all’ and ‘some’ range over. But, as I have said, our nota-
tion is not introduced for the purpose of constructing a formal system. It is only abbreviatory
and disambiguative.
19
This is second level quantification but of a very limited kind. There is no question yet, of quan-
tifying over everything true of x. We quantify over the sorts of thing which this or that given
object is, in the Aristotelian sense of this phrase. For this, see again the epigraphs for Chapter
One.
 Preamble 
true if and only if, for some f, a f-s and a ⫽b or, better, a has to b the
relation of identity as restricted to things that f; or, more formally, (᭚f )

冢 冣
a⫽b
f
. Note that in the quasi-formalized version, concatenation is
taken to do the work of the verb ending. When we revert to the ordinary
English form a is an f and look back at the quasi-formalized version, this
concatenation is seen as having taken over the work of the indefinite
article too, bracketing for us the nice question of how much or little sig-
nificance to attach to the English form ‘a is a(n) . . .’.
In sum, following the variant that I propose on Quine’s proposal, we
then have the equivalence:
a ⫽b ↔ (᭚f) ⫽
a b
f 冢 冣
Let it be as clear as anything can that, in embracing and exploiting this
equivalence, one is not agreeing with Geach’s claim that always ‘x is
identical with y’ is an abbreviation for ‘x is the same f as y’ where f ‘rep-
resents some count noun supplied from the context of utterance’ (‘or else
it is just a half-formed thought’).20 One is disagreeing deliberately and
actively with that. If vindication were really needed of the bare ‘a ⫽b’,
then the right-hand side of our equivalence would vindicate it. It would
vindicate it as definite, determinate and well formed. It is determinate
in the same way as ‘I talked with someone yesterday at noon’ is determi-
nate for a truth-value.21
In Sameness and Substance the question of how ⫽ and ‘a is an f ’ are
a b
f
related was left undecided. In so far as it can now be treated as decided,
there should be no strict need to argue separately or twice over, as in
⫽
Chapter One, about the formal properties of ‘⫽’ and ‘ f ’. There is
something to be said, though, for trying to satisfy the reader separately
under each head. Again, in Chapter Two, where the doctrine of Sortal
20
See Geach (), p. . See also Geach (), p. .
21
It follows that there are two reasons why there has never been any temptation, either in Sameness
and Substance () or in Wiggins (), in each of which ‘a ⫽b’ was treated as guaranteed the

冢 冣
same truth-value as ‘(᭚f) a ⫽ b ’, to follow Geach in claiming that ‘a ⫽b’ is in any way defective
f

冢 冣
or incomplete. First, no variable is free on the right side. Secondly, being Leibnizian ‘(᭚f ) a ⫽ b ’
f
and ‘a⫽ b’ are both determinate in the further sense of excluding the possibility (see Chapter

冢
One) that (᭚g) ga & not a ⫽ b .
g 冣
 Preamble
Dependency is formulated, it may be wondered what strict need there
now is for the third clause of Principle D to invoke f-coincidence as such.
In that third clause, is more required than that there should be no differ-
ence at all between the particular f that a is and the particular f that b
is? In the absence, however, of complete reconstructions of the lifespans
of a and of b (and that is the normal case), some idea of f-coincidence
does seem integral to the ordinary working or application of the third
clause of D. In Sameness and Substance Renewed, the third clause remains
as it was. See Chapter Two, §.

 .             
Do the determinations of the foregoing section mean that there is a
danger that we may be forced to concur in a claim that adherents of
Quine are apt to repeat sometimes, as if in evidence against positions
like D? Does it mean that ‘x⫽y’ is prior somehow to the practices,
directly and indirectly invoked by D, of the individuation of things and
the assignment of things to thing-kinds?
When we do propound identity conditions for bodies, or persons, or classes, we
are using the prior concept of identity in the special task of clarifying the term
‘body’ or ‘person’ or ‘class’; for an essential part of the clarification of a term is
clarification of the standard by which we individuate its denotata.22
Well, no. There is no need at all to concur in the asymmetry Quine has
seemed to suggest. For the truth is that there can be no question in this
province, where so much is primitive, of anything’s being absolutely prior
– either logically or philosophically or psychogenetically. To be sure,
‘Juno donkeys’ or ‘Juno is a donkey’ will seem (so soon as we look at this
predication as it appears in English) to mean or imply that Juno is the
same as one of the donkeys or is identical with one of the donkeys. But,
as we shall see in Chapter Two, there is also an opposite dependence.
The practical grasp of identity itself presupposes the capacity to
subsume things under kinds, to refer to them and to trace them (or keep
track of them). But in order to trace things, one has to trace them in the
way that is appropriate to this or that kind, and then, by dint of one’s
understanding of congruence as flowing from identity or coincidence, to
assign to an object picked out at one time and again at another time
everything that is true of either. But if that is right, then not only does
sortal predication presuppose identity. Identity presupposes sortal pred-
22
W. V. Quine (). My italics.
 Preamble 
ication.23 These things belong together – or the grasp of each requires
the grasp of the other. Identity is a notion coeval with the determinable
entity of some determinate kind, which brings with it the possibility of the par-
ticular determinations that figure in particular sortal predications.
Without thinking very hard about what one is hoping for, one forms
too easily the aspiration to discover a starting point from which one
could introduce, stage by stage, as if definitionally, the apparatus of
naming, substantial predication, identification and differentiation, then
other kinds of predication, then quantification over individuals, then
quantification over sorts, then abstraction (the prototype of lambda
abstraction), then quantification over properties . . . If the process to be
reconstructed were intended to be definitional or quasi-definitional,
however, then all the hazards of piecemeal definition would have to be
guarded against (the hazard of the left hand’s undoing or altering what,
unknown to the left hand, the right has already done). The difficulties
would quickly appear of keeping careful account of what already pre-
supposes what. Given the absolutely foundational nature of that with
which we are concerned, however, and the primitiveness of the predica-
tive and other devices that are under consideration, I should claim that
it is a false picture that bewitches one who demands such a quasi-
definitional sequence. What we have here to confront is a whole skein of
connected practices. These practices are intertwined with one another.
Their relations can indeed be set out in all sorts of true equivalences. But
it does not follow they can be set out in a developing sequence of the kind
we were meant to be looking for. It is much more likely that the basic

23
Geach seeks to penetrate to a deeper level of analysis of sortal predication, by isolating an under-
lying mode of combination effected by the attribution of equivalence relations (which are not
necessarily, according to Geach, congruence relations) such as x is the same donkey as y, x is the same
apple as y, etc. This relation can then be derelativized as in ‘x is the same donkey as x’, ‘x is the same
apple as x’, etc. Finally, he offers these latter forms as definitions of ‘x is a donkey’, ‘x is an apple’.
Geach’s proposal can be separated from Geach’s case against absolute identity and his denial
of congruence.
If the proposal were offered on the level of psychogenetic theory, then it would be implau-
sible. For it is hard to avoid thinking that calling x the same horse as y presupposes picking x out
and picking it out as something (as a horse?) simpliciter. If there were nothing more to horse-
identity than there was to being a horse, whence could come the relationality that Geach dere-
lativizes?
If Geach’s proposal is simply grammatical or definitional in its intent, well, that is more plau-
sible. But it may be less clear than it looks. For as soon as we try to bring Geach’s ‘same donkey
as’ forms into the requisite relation with ‘the same what?’ and ‘the same something as’, which
are indispensable to the whole rhetoric of sortalism, we need to discern structure within ‘is the
same donkey as’. This structure cannot, however, be, for Geach, that which is suggested by the
English predicable ‘is the same donkey as’. For that apparently reintroduces both absolute iden-
tity and simple ‘donkey’ predication.
 Preamble
forms and devices have to be learned together. Just as the keystone of an
arch and the adjoining bricks can be placed together, but only if
somehow they are placed there simultaneously or they are put into posi-
tion with the help of a temporary external support, so each primitive
device is learned simultaneously and in reciprocity with each of the others.
In the language learning case, there is the possibility of partial and
interim grasp of a device. But the full grasp of any one device will
require the full grasp of many or most of the others. If semantic devices
are well made, then, once they are fully mastered, they will operate
smoothly and in concert together. Their mutual relations will indeed
allow of their being rehearsed by logical equivalences. If a sceptical
question arises, however, about whether in their supposed reciprocity the
concepts corresponding to the expressions we are concerned with are
well made, the only answer that ought to be promised is that, several
chapters into the book, there will be indirect assurance that the locutions
‘same’, ‘horse’, ‘same horse’ collaborate securely in the modes of com-
bination of which we avail ourselves in identifying things, placing them
in kinds, distinguishing them from other things and attributing other
properties to them. For the interim, the most that should be offered in
answer to the question broached in paragraph one, § above, is a non-
definitional and non-psychogenetic account of a single mode of combi-
nation linking ‘is an f ’ and ‘is the same f ’.
    

The absoluteness of sameness

Only [species and genera, among predicables,] reveal the primary

substances [e.g. Socrates]. For if one is to say of the individual man
what he is it will be in place to give the species or the genus (though
more informative to give man than animal); but to give any other
thing would be out of place – for example to say ‘white’ or ‘runs’
or anything like that.
(Aristotle, Categories b30–7, as translated by J. L. Ackrill, Oxford, , but
with italics and square bracket additions.)

[Man or animal] does not signify simply a certain qualification, as

white does. White signifies nothing but a qualification, whereas
species and genus mark off the kind of substance – they signify what
sort of substance.
(Aristotle, Categories b17–21, translation and paraphrase based on J. L.
Ackrill, op. cit., vide p. .)

.                     ;   

                       
                  
If somebody claims of something named or unnamed that it moves, or
runs or is white, he is liable to be asked the question by which Aristotle
sought to define the category of substance: What is it that moves (or runs
or is white)? Perhaps one who makes the claim that something moves
does not need to know the answer to this question in order to enter his
claim. It is not hard to envisage circumstances in which he can know that
it moves without knowing what the thing is. Yet it seems certain (for more
on this see below, Chapter Two, §) that, for each thing that satisfies a
predicate such as ‘moves’, ‘runs’ or ‘white’, there must exist some known
or unknown, named or nameable, kind to which the item belongs and
by reference to which the ‘what is it’ question could be answered.

 The absoluteness of sameness
Everything that exists is a this such. Where someone fully understands
what he is judging when he says that this, that or the other things moves,
or runs or is white, he will know what the thing in question is.1
Secondly, if someone reports further that the thing that runs is the
same as the thing that is white, then his judgment cannot be true unless
at least two conditions are satisfied. These conditions are that the thing
that runs should be the same something or other as the thing that is white (the
same horse, say, or the same man or whatever) and that the something
in question be correlative with or associated with a principle by which
entities of a particular kind may be traced or kept track of and re-iden-
tified as one and the same.
These two requirements will represent one part of what will be
referred to in this book as D, or the Thesis of the Sortal Dependency of
Individuation. (For D as elaborated, see Chapter Two, §.) But in some
readers, neo-Aristotelian formulations of this kind will instantly arouse
grave misgivings. Such suspicions will be allayed in the end (if at all)
simply by following out the implications of the classical conception of
identity. But that is the work of later chapters. At the beginning of a
search for a theory of individuation, it will be better to focus on a non-
Aristotelian question that requires nothing more of the reader than that
he should understand English sentences that employ familiar locutions
like ‘the same horse’, ‘the same donkey’, ‘the same tree’. The universally
accessible question to which these locutions naturally give rise is whether
a can be the same f as b without being the same g as b. Better and more
precisely, the question is whether a can be the same f as b, and not the
same g as b, even where a or b is itself a g.
This is the topic of Chapter One. The answer to the question will be
short and negative. But the absoluteness thereby discovered in identity
will carry with it certain unexpected consequences, the possibility for
instance that a portion of space may be co-occupied at a time by distinct
things (provided, at least, that the things in question belong to different
basic or individuative kinds). Indeed some of these consequences may be
strange enough to prompt someone to reconsider the attractions of the
affirmative or relative answer to our question. Let him do so at least once
before confronting the larger questions of identity and individuation that
will arise later – if only because our answers will need the absoluteness
that it is the work of the present chapter to establish.
The thesis of the Relativity of Identity, which I shall denominate R
11
Most recently, I have tried to answer the doubts that such claims have provoked in Wiggins
(b).
 The absoluteness of sameness 
and shall steadfastly oppose, may be held in several distinct forms.2 It will
convey the general spirit to enumerate a few of the different and equally
correct sortal answers that may be given to the question ‘What was the
thing that ran?’ Suppose the answers are man, soldier, greybeard, Greek, phi-
losopher, official of the state, the chairman (epistates) for day d,  B.C., the subject
of a certain portrait bust in the Vatican Museum, stonemason, victim of the wit of
Aristophanes, the person whose body was the only human body that was to be discov-
ered at place p at time t in the agora of ancient Athens. Then an R-theorist might
suggest that it makes all the difference to the project of keeping track of
continuants through space and time which of the concepts in such a list
one subsumes something under. Might not material particular a coincide
with material particular b when individuated under some of these con-
cepts and not coincide with b, but be distinct from it, when individuated
under others? From this apparent possibility a defender of R may want
to conclude that the notion of identity is concept- or sortal-relative, i.e.
relative to different possible answers to the question ‘a is the same what
as b ?’. (For ‘concept’ and ‘sortal’, see Preamble, §.)
D and R, the theses of Sortal Dependency and Relativity of Identity,
may seem to be made for one another. For this reason they have some-
times been confused or assimilated; and sometimes, because R appears
to entail D, R has been taken to be the one good reason for maintaining
D.3 Occasionally, philosophers have even gone so far as to argue from
12
The label is suggested by Quine’s review (), p. ) of P. T. Geach, Reference and Generality
(Ithaca, N.Y., ), and Quine’s characterization of the doctrine given at p. , where Geach
says: ‘I could not object in principle to different As’ being one and the same B . . . as different
official personages may be one and the same man.’ Cf. also Geach’s later articles at Review of
Metaphysics, ,  (), and ,  (). See Geach () ().
Among those who have followed Geach in one way or another are Eddy M. Zemach ‘In
Defence of Relative Identity’, Philosophical Studies  (), pp. , ; D. Odegard, ‘Identity
through Time’, American Philosophical Quarterly  (), pp.  ; Nicholas Griffin, Relative Identity
(Oxford, ); Harold Noonan, ‘Objects and Identity’, Cambridge Ph.D. thesis,  (see the
extracts in Noonan () and ()); Tobias Chapman, ‘Identity and Reference’, Mind, 
(); Richard Kraut, American Philosophical Quarterly (). See also Noonan ().
13
Some have held that the reverse dependence exists between Relativity of Identity and Sortal
Dependency, others that they are equivalent doctrines. The supposition that D depends for its
rationale on its being possible for a to be the same f and not the same g as b was made by
Professors Geach and Quine in their original controversy about these matters. Professor Geach
insisted on the universal legitimacy of pressing the same what? demand (see (), §§–, and
Chapter Six); and one of his arguments for it rested on the doctrine that it was logically possible
for a to be the same f as b without being the same g as b. Professor Quine in his review rejected
out of hand the possibility of a’s being the same f as b without being the same g as b. But it was
apparently on the strength of that rejection that he strongly questioned the legitimacy of Geach’s same
what? demand and the proposals he took to be Geach’s for the treatment of quantified sentences.
The one point on which Geach and Quine seem to have been agreed at that time is that the pos-
sibility of a’s being the same f but not the same g as b is what provides the principal rationale of
D. This is a point on which I am to disagree with them both.
 The absoluteness of sameness
the falsity of R to the falsity of D. But this is a formally fallacious pro-
cedure.
R and D are utterly distinct. The falsity of R will be argued very
shortly, on the basis of the principle commonly known as Leibniz’s Law,
which says that if a is b then whatever may be true of a is true of b. It will
be argued that ‘a is b’ and ‘a is the same f as b’ are equally and excep-
tionlessly subject to this Leibnizian principle. In Chapter Two, with R
put out of the way, the Sortal Dependency of Individuation will be
amplified and refined. D will then be applied to the task of finding a
schematic or formal characterization of the notions of substance and
sortal concept. Chapter Three will relate the abstract requirements
upon sortal concepts to the sortal concepts that we actually possess.4
Everything else will follow on from that.
In addition to R and D, there will be occasion to mention a thesis that
I shall call the Counting thesis, or C. It says that to specify the something
or other under which a and b coincide is to specify a concept f which
qualifies as adequate for this purpose only if it yields a principle of count-
ing for fs. C is no more to be confused with D than R is. I shall submit
that, however closely it approximates to an important truth, C is false.
(See Chapter Two, section .)
The realistic discussion of the thesis of the Relativity of Identity
requires formal argument, as in the ensuing section. It also requires the
detailed discussion of examples. That will be the work of the rest of this
chapter.

.  ’                  

    
The fact that there are many different sortal concepts under which one
may single out some individual a does not in itself imply R. For all the
alternative procedures of individuation of a under alternative covering
concepts might, when they yielded any answer, yield the same answer to
that question.5 My contention is precisely that they must do so. The
14
For other support of D, or for related theses, see, e.g., A. Prior, Analysis,  (June ); Hampshire
(); D. Wiggins, ‘Individuation of Things and Places’, P.A.S. Supp.  (), printed in more
correct form in M. Loux, ed., Universals and Particulars, st edn (only) (New York, ).
15
I shall call an individuative or sortal concept that adequately answers the same what? question for
an identity-statement s, a covering concept for s, and, as prefigured in the Preamble §, I reserve
the letters f and g to represent such sortal concepts. ‘f(a)’ means that a belongs to the sort f – col-
loquially, that a is an f or is one of the fs. In the case of an identity ‘a ⫽b’ supplemented, as any
fair discussion of R will evidently require, with covering concept f, I write a ⫽ b. For the
f
 The absoluteness of sameness 
reflexivity and congruence of identity provide logically compelling
reasons why, if a is b, or if a is the same something or other as b (same
horse, same tree, same planet, or whatever), then all different procedures
of individuating a (provided they really do individuate a) must, if they
yield any answer at all, yield the same answer with respect to a’s coinci-
dence with b. This is to say in the timeless idiom that if, for some f, a is
the same f as b, then for any g such that a is a g, a is the same g as b, or

((᭚f) 冢a ⫽f b冣) →((᭙g)(g(a ) → 冢a ⫽g b冣));

which is to say that R is false.
To illustrate this will only require the extension to ‘is the same f as’ or
⫽
‘ ’ of the formal properties that logicians associate with ‘⫽’.6 The
f
strong reflexivity of the same (dyadic identity),
(᭙x)(x⫽x),
will have for its same f counterpart the reflexivity property

(᭙x)(f(x) → 冢x ⫽f x冣).
The congruence of sameness that is affirmed by Leibniz’s Law or the
Indiscernibility of Identicals,
(᭙x)(᭙y)((x ⫽y) → (␾x ↔ ␾y))7
has as its same f counterpart:

(᭙x)(᭙y)( 冢x ⫽f y冣 → (␾x↔ ␾y)).

interpretation of this, see again Preamble §. Replacements of the ordinary predicate letters phi
and psi (␾ and ␺), as they occur in schemata of first order logic, will embrace both sortal pred-
icates and non-sortal predicates.
16
On the account of ‘is the same f as’ that I have proposed in the Preamble §, this is not a sub-
stantial extension at all. We are simply following through the consequences of the claim that this
relation is a restriction of the relation of identity. But from now until the end of this chapter I shall
not simply assume the Preamble account. I hope it will help to confirm that account if I can
arrive at the conclusion I shall arrive at without its immediate assistance.
17
The schematic letter phi holds a place for any predicate that can be true of the things over which
the variables of quantification range. Given reflexivity, the schema in the text is derivable from
(x) (y) (x ⫽y → (␾x → ␾y)). See Quine (), pp. –. Furthermore, reflexivity and this schema
suffice to prove symmetry and transitivity.
 The absoluteness of sameness
Or, treating the letter f not as a schematic letter but as a variable, we have

(᭙x)(᭙y)((᭚f) 冢x ⫽f y冣 →(␾x ↔ ␾y))

(᭙x)(᭙y)( ᭙f)( 冢x ⫽f y冣 →(␾x ↔ ␾y)).
So much for being the same f. Now for the thesis R. R entails that for
some a, b, f and g

冢a ⫽f b冣 & not 冢a ⫽g b冣 & g(a).

Is that last possible? To answer that question let us suppose that the first
conjunct, ⫽ , holds and let us take the ␾ in the Leibnizian schema
a b
f

冢a ⫽f b冣 → (␾a ↔ ␾b)
as including among its replacements any predicable of the form ‘ ⫽ ’.8
a x
g
Then we have

冢a ⫽f b冣 → (冢a ⫽g a冣 ↔ 冢a ⫽g b冣).
But then, by modus ponens and the supposition that ⫽ , we have the
a b
f
consequent

冢a ⫽g a冣 ↔ 冢a ⫽g b冣.
⫽
But now, by the reflexivity of ‘ ’ and g(a), as in the third conjunct, we
g
have

冢a ⫽g a冣.
18
For what follows, see Wiggins (), pp. –. For similar derivations see Perry (), p. ;
Stevenson (), p. . Stevenson worked out a formalized framework in which this and other
claims could be formally evaluated in Stevenson ().
 The absoluteness of sameness 
Hence, by modus ponens,

冢a ⫽g b冣.
Hence we arrive at not R. So, if we suppose that the first conjunct of
R holds true, then we obtain the denial of the remainder of R. It follows
that R is not true.
The questions that now arise are whether Leibniz’s Law is true (as this
derivation presumes) and, if so, whether it is as true for ‘is the same f as’
as it is for ‘⫽’. In support of an affirmative answer, I shall urge four con-
siderations:
(i) Leibniz’s Law marks off what is peculiar to real identity and it differ-
entiates it in a way in which transitivity, symmetry and reflexivity (all
shared by exact similarity, weighing the same, having exact equality in pay, etc.)
do not.9
(ii) How, if a is b, could there be something true of the object a which
was untrue of the object b? They are the same object. People sometimes speak
of counter-examples to Leibniz’s Law. But these are scarcely more
impressive than the counter-examples to the Law of Non-Contradiction.
19
Cf. Frege’s remark in his review of Husserl, p.  in Geach and Black’s Translations from the
Philosophical Writings of Gottlob Frege (Oxford, ): ‘I agree . . . that Leibniz’s explanation eadem
sunt quorum unum potest substitui alteri salva veritate does not deserve to be called a definition; my
reasons, however, are different from Husserl’s. Since any definition is an identity, identity itself
cannot be defined. This explanation of Leibniz’s could be called an axiom that brings out the
nature of the relation of identity; as such it is fundamentally important.’ We may fault Frege’s
doctrine of definition here, and we ought to note that the eadem sunt quorum unum alteri substitui
potest principle says more than the principle that I have called Leibniz’s Law. But when construed
as a contention about Leibniz’s Law, Frege’s contention seems incontrovertible. (See also
Grundgesetze der Arithmetik, Jena, ,  Band, .)
The Indiscernibility of Identical was familiar to Aristotle (see De Sophisticis Elenchis  a37). But
I follow custom in calling it Leibniz’s Law. It is not to be confused with the distinctively
Leibnizian (and non-schematic) converse, the Identity of Indiscernibles (to be discussed in § of
the next chapter). Leibniz’s eadem sunt quorum unum alteri substitui potest principle presumably entails
both Leibniz’s Law and the Identity of Indiscernibles.
It may be asked what protects the schematic Leibniz’s Law from illegitimate replacements for
the letter phi (␾). I follow Quine in holding that intensional replacements are excluded by ‘the
incoherence of bound variables in any but referential position’. See Journal of Philosophy , 
(), p. . Quine writes ‘This version of the principle of identity is a little broader than the
version in terms of properties, since the open sentence represented by ‘␾x’ can sometimes exceed
the range of properties, for reasons unrelated to substitutivity and related rather to Russell’s
paradox.’ By ‘the version in terms of properties’, Quine means the non-schematic second level
principle that if x is y then every property of x is a property of y, which is discussed and defended
by Cartwright (), p. .
Quine’s characterization of the substitutivity schema suffices in my opinion to answer all the
questions about paradox that Geach has seen as counting against the congruence conception of
identity. See the works by Geach cited at note  above, in further reply to which, I would refer
to my reply to Harold Noonan in Lovibond and Williams (), p. , note .
 The absoluteness of sameness
Concerning modal and intensional contexts, Frege and his inheritors
have shown how ineffective such contexts are to discover to us objects
that are at once identical (or numerically the same f) and discernible.10
At their best, these Fregean arguments are given on the basis of inde-
pendently plausible accounts of what else is being conveyed in the pur-
ported counter examples.
(iii) If Leibniz’s Law is dropped, or if classical identity is dropped in
favour of some allegedly un-Leibnizian relative identity, then we need
some formal principle or other, and one of at least comparable univer-
sality, to justify the instances of the intersubstitution of identicals that
evidently are valid. The instability, indeterminacy or arbitrariness of all
extant emendations and relativizations of Leibniz’s Law only corrobo-
rate the case for a pure congruence principle such as Leibniz’s. (See
below, §, and especially footnote .)
(iv) Suppose there were terms t1 and t2 both designating z, one and the
same donkey, and suppose there were a context ␾( ) such that the result
of supplying t1 to it was true and the result of supplying t2 was false. What
ought we to say if it were suggested that the open sentence ␾(x) deter-
mined a property? Call the putative property Q. We ought to ask: How
can the donkey both have and lack the property Q? The question is
unanswerable. Let the R-theorist note that this argument can be stated,
as it is stated here, without showing any special favour between ‘⫽’ and
⫽
‘ ’. It supports Leibniz’s Law for both of these relations. In order to
f
counter it, the R-theorists will have to uncover much more complexity
than appears to be present in the innocuous locution ‘t1 designates z and
t2 designates z’. Nor is that enough. R-theorists will need to deny the very
possibility of there being such a relation as simple designation. (Chapter
Six, §, footnote  bears on these matters.)

a⫽b
.                  
g
The previous section showed why we ought not to expect to find any true
judgments whose truth requires the relativity of sameness. Nevertheless,
given that there is no shortage of apparent examples in the form ⫽
a b
f 冢 冣
& not
a⫽b
g冢 冣 & (g(a) v g(b)), it will be as well to classify them,

10
See ‘On Sense and Reference’ (Frege, ). Frege’s defence of Leibniz’s Law is fortified in a way
independent of his theory of direct and indirect sense and real and apparent reference by
Cartwright (), p. .
 The absoluteness of sameness 
however crudely. There are five types of case where a is not the same g
as b. Only three of these will provide ways for R to appear to be exem-
plified, but it will do no harm to enumerate all five.
() g may simply be the wrong covering concept for both a and b where
nevertheless a ⫽b. The evening star is the same planet but not the same star
as the morning star. For Venus is not a star. This is not a case of

( 冢a ⫽f b冣 & not 冢a ⫽g b冣) & (g(a) v g(b)).

() Venus is not the same star as Mars nor the same anything as Mars.
For in this case, for all f, not 冢 冣
a⫽b
f
. Again this is irrelevant.
() We may seem to get nearer to what is required with the case where
John Doe, the boy whom they stupidly took for a dunce, is the same human
being as Sir John Doe, the Lord Mayor of London, yet neither the same
boy (for the Lord Mayor is not a boy) nor the same mayor/ex-cabinet minis-
ter/father of five marriageable daughters. (For the boy did not attain office or
beget children when a boy.) Yet surely, it may be said, boy, dunce, mayor,
ex-cabinet minister, father of five marriageable daughters are all sortal words.
They furnish perfectly good covering concepts. One can count and iden-
tify such things, and so on. So this gives the appearance of a case where
we have

冢a ⫽f b冣 & not 冢a ⫽g b冣 & (ga v gb) & not (gb),

a case that is where a cuts out, as it were, under a sortal-concept g (e.g.
boy) but persists under another sortal-concept f (e.g. human being).
This is not the sort of case the relativist is looking for. The first thing
it shows is the necessity for care, care about tenses, care in the interpre-
tation of affirmation and denials of the formula ⫽ or ⫽ and
a b
f冢 冣 冢 冣 a b
g
care in the interpretation of Leibniz’s Law. Secondly, it shows the pos-
sibility of an interesting and highly important distinction within the class
of sortal predicates.
If John Doe is still a boy, then John Doe, the boy, will one day be a
cabinet minister and later the Lord Mayor of London, and he will beget
five children. If John Doe is no longer a boy, then John Doe the boy (or
Sir John Doe, when he was a boy) was going to be and was going to do these
 The absoluteness of sameness
things. We only thought we had a case of R because we confused the
timeless and the tensed way of speaking within one utterance. If ‘g(Sir
John Doe)’ is to be a tensed statement, it should be read as saying that
Sir John was a boy and it is true. If it is a tenseless statement, then it says
of Sir John Doe that he is, was or will be a boy or that at some time or
other he timelessly is a boy. This again is true. If you take tenses seriously,
you may refuse to say ‘Sir John Doe is the same boy as John Doe’, since
it is false that Sir John Doe is now a boy. But it is true and perfectly
unproblematic that Sir John Doe was the same boy as John Doe (was). It
is precisely for this reason that Sir John Doe is not now identified or
singled out under the temporally restricted sortal boy. From all this it

冢 冣
follows that not ⫽ , properly read, is not true. (Similarly, the author
a b
g
of Childe Harold I/II () was the same person as the baby born in 
who was christened George Gordon Byron. That author was a poet. But,
if we said the author was the same poet as the baby, that would give the
impression that the baby could write verses when he was a baby and that
is not true. The baby will, however, be the same poet as the author of
Childe Harold I/II.)
The second matter that type-() cases bring to our attention is this.
They underline the need to distinguish between sortal concepts that
present-tensedly apply to an individual x at every moment throughout
x’s existence, e.g. human being, and those that do not, e.g. boy, or cabinet
minister. It is the former (let us label them, without prejudice, sub-
stance-concepts) that give the privileged and (unless context makes it
otherwise) the most fundamental kind of answer to the question ‘what
is x?’ It is the latter, one might call them phased-sortals, which, if we
are not careful about tenses, give a false impression that a can be the
same f as b but not the same g as b. But, in fact, they do not conflict at
all with what is to be proved: namely that, for all x and all y, every
concept that adequately individuates x for any stretch of its existence
yields the same answer, either directly or via the principle of individ-
uation for the substantive it restricts, as every other genuinely individ-
uating concept for x or y to the question whether x coincides with y or
not.
The first appearance of the phased-sortal is probably the best
moment to inveigh against a misunderstanding of identity which has
culminated in attempts to show that the only true identity is the ‘identity
at a time’ of one and the same phase or thing-moment of a thing, ‘iden-
tity through time’ being held by those in the grip of this conception to
connote some different relation that holds between the different phases
 The absoluteness of sameness 
or thing-moments out of which individual continuants are supposed to
be combined or concatenated.11
Whatever the merits of this idea as a programme for linguistic reform,
those drawn to it have never appreciated the true scale of the re-orga-
nization that they are envisaging. There is no serious prospect that this
view of the identity relation will do justice to any of the actual questions
of continuity and persistence that perplex our habitual modes of
thought about identity and difference. That which these questions want
in the first instance is not replacement by other questions given in terms
of phases of things or thing-moments12 or in an alien four-dimensional
mode. The answers these questions require from philosophy ought to be
given in language that speaks as simply and directly as natural languages
speak of proper three-dimensional continuants – things with spatial
parts and no temporal parts, which are conceptualized in our experience
as occupying space but not time, and as persisting whole through time.13
Unlike sets or classes, continuants can gain and lose elements. The terms
11
Cf. Hume, Treatise of Human Nature, Bk. , §: ‘We have a distinct idea of an object that remains
invariable and uninterrupted through a supposed variation of time; and this idea we call that of
identity or sameness. We have also a distinct idea of several different objects existing in succes-
sion, and connected together by a close relation; and this to an accurate view affords as perfect
a notion of diversity as if there were no manner of relation among the objects.’
12
Against a related but distinct proposal concerning existence and persistence, see Chapter Two. See
also Chapter Six, §, where some eirenic remarks are offered on the four-dimensional mode of
speech. For a discussion of the distinction that Mark Johnston, David Lewis and other writers have
made between the endurance and the mere perdurance of things, see §. of my ‘Substance’ ().
13
It is well worth remarking that, if we respect this way of thinking, then by a transposition we
arrive at the everyday conception of event. An event takes time, and will admit the question
‘How long did it last?’ only in the sense ‘How long did it take?’. An event does not persist in the
way in which a continuant does – that is through time, gaining and losing new parts. A continu-
ant has spatial parts. To find the whole continuant you have only to explore its boundaries at a
time. An event has temporal parts. To find the whole event you must trace it through the histor-
ical beginning to the historical end. An event does not have spatial parts in any way that is to be
compared with (or understood by reference to) the way in which it has temporal parts.
At least in the light of this conception of the differences of events and continuants, there
appears to be a terrible absurdity in such claims as ‘a material object is just a long event’. For an
illuminating specimen of such claims, see C. D. Broad in Scientific Thought, p. :
We usually call a flash of lightning or a motor accident an event, and refuse to apply this
name to the history of the cliffs of Dover. Now the only relevant difference between the flash
and the cliffs is that the former lasts for a short time and the latter for a long time. And the
only relevant difference between the accident and the cliffs is that if successive slices, each of
one second long, be cut in the histories of both, the contents of a pair of adjacent slices may
be very different in the first case and will be very similar in the second case. Such merely
quantitative differences as these give no good ground for calling one bit of history an event
and refusing to call another bit of history by the same name.
This will not do. Material object and event are in some sense duals. It has recently come to seem
more and more important in philosophy to understand the notion of an event (cf. Donald
Davidson, ‘Causal Relations’, Journal of Philosophy, , ). A fair start can be made on these
questions if we take note of all the hints of analogy and disanalogy we get from the unreformed
language of things and events.
 The absoluteness of sameness
of the identity questions that we are concerned with in this book stand
for these continuants themselves, John Doe, say, not for phase-thick lam-
inations of their four-dimensional counterparts, John Doe-when-a-boy,
or for infinitesimally thin time-slices of these held fast between percep-
tual impressions, John Doe-as-caught-sight-of-there-then. And, by the
same token, the phased sortals like ‘boy’ or ‘old man’ that we encounter
in English never denote ‘phases’ of entities or (if that were different) the
entities themselves frozen at an instant. They denote the changeable
changing continuants themselves, the things themselves that are in these
phases. No faithful elucidation of identity judgments will seek to alter
this. People often speak of identity over time and distinguish this from
identity at a time. But identity is just identity. (See Chapter Six, §.)
Experience proves that here there is a choice between tedious repeti-
tion and scarcely prevailing against a deep-seated tendency of the
human mind towards confusion.14 In the equivalence x is a boy at t if and
14
It is a profound and important question, which one should long keep open, what underlies this
nisus towards disorder. That it is not provincial to any particular language or culture is attested
by the confusion, misrepresentation and misquotation to which Cratylus and others subjected
the gnomically simple insights of Heraclitus, a thinker as innocent of the confusion of ‘numer-
ical’ and ‘qualitative’ sameness as he was of any positive desire to equate opposition and contradic-
tion. Robert Coburn has given an able account of some of the sources of confusion. (See ‘The
Persistence of Bodies’, American Philosophical Quarterly, , .) For another source, scrutinize the
quotation from Hume at footnote . The capacity to make inner images is as difficult to control
as it is indispensable in most human thinking. Perhaps it will help even to explain the folly Russell
describes in ‘The Philosophy of Logical Atomism’ (Logic and Knowledge, ed. R. C. Marsh, London,
, p. ) when he writes:
Identity is a rather puzzling thing at first sight. When you say ‘Scott is the author of Waverley’,
you are half-tempted to think there are two people, one of whom is Scott and the other the
author of Waverley, and they happen to be the same. That is obviously absurd, but that is the
sort of way one is always tempted to deal with identity.
There is another confusion that is a converse of this: its aetiology will deserve equal attention from
someone who comes closer than I shall to completing the theory of individuation and identity:
The question of what we count as survival, as continuing to be the same person, is even more
difficult to answer in a precise and definite way. (Joel Kupperman, Character, New York, ,
p. .)
Nobody has any choice about going on ‘being the same person’. One is bound to do so. In
ceasing to be, one does not become somebody or something else. The difficult question is rather
this: what it takes, conceptually speaking, to survive or persist. For another case of ‘changing my
identity’, see
One of Locke’s insights was that although under the description ‘this lump of matter’ I may
be changing my identity from moment to moment, under a description ‘this person’ I am
not. (David Sedley, ‘The Stoic Criterion of Identity’, Phronesis, , p. . Cf. Hobbes cited
below at note .)
I am not a lump of matter, even if a lump of matter constitutes me. I cannot then change my
identity under the description ‘lump of matter’. I cannot change my identity at all in any rele-
vant sense of ‘my identity’. (For another sense of ‘my identity’ that may be needed sometimes,
but is not at issue here, see Chapter Seven, §.)
 The absoluteness of sameness 
only if x is a male human being who has not at t reached maturity, we have the
substance term ‘human being’, and two qualifications of it which deter-
mine proper subsets of the class of human beings. I have followed Geach
in calling these qualifications ‘restrictions’ of the concept human being. It
follows that, unlike ‘boyhood’, ‘boy’ cannot denote a phase. Still less can
‘boy’ denote a phase or stage, whatever that would be, of a human being.
A boy is a human being. No human being is a stage of a human being,
or an -year-old cross-section of a year-long space–time worm.
Suppose some boy grows up and lives seventy years. If that is how it is,
then there is only one answer to the question ‘How long did that boy
persist in being?’ – namely seventy years. He lived seventy years. He did
not live seventy years as a boy, but, when he grew up, that growing up
was not the passing away of a boy.
So much for confusions that can flow from imperfect recognition of
the presence in natural language of phased predicates and the phased
sortal concepts that they introduce. Finally, if there were need, further
refinements could be introduced at this point, e.g. between predicates
like ‘infant’, ‘adult’, ‘pupa’, ‘tadpole’, which every member of the exten-
sion of the substance term that they restrict must in due course satisfy if
only it lives so long, and predicates like ‘conscript’, ‘alcoholic’, ‘captive’,
‘fugitive’ or ‘fisherman’, of which this does not hold. This distinction,
like certain others that are there to be made, is not without interest. But
we shall not need these refinements to dispose of type-() cases repre-
sented as cases of R.

The point we have now reached is the fourth and fifth types of identity
claim where R appears to be verified. Here at last we shall find cases with
a semblance of plausibility.
() is the variant where, in the timeless idiom,

冢a ⫽f b冣 & not 冢a ⫽g b冣 & (g(a) v g(b)) & (g(a) & not g(b)).
() is the type of case where, allegedly,

冢a ⫽f b冣 & not 冢a ⫽g b冣 & (g(a) v g(b)) & (g(a) & g(b)).

We need some examples which might be said by an R-theorist to be
of type (), and then some of type (). We must understand why these
examples give the appearance of being cases of R and then understand
what reasons there are for interpreting them otherwise.
 The absoluteness of sameness

.            -(  )        

(␣) I might say to someone ‘that heap of fragments there is the jug you
saw the last time you came to this house’. They could not be the same jug
but they might be the same collection of material bits.
(␤) The visitor might be a person of tiresome ingenuity and cement
the pieces of the jug together to make not a jug but, say, a coffee pot of
a quite different shape from the original jug’s. It might then be said that
‘the jug is the coffee pot’ is true with covering concept same collection of
material bits and false with covering concept same utensil.
(␥) Perhaps the best and least strained example of type () is one of a
kind which a champion of the relativization thesis that is given in P. T.
Geach’s Reference and Generality might describe in the following way:
‘Leonard Linsky asked in his review of Reference and Generality15 why
‘Cleopatra’s Needle’ could not correspond in use to ‘the same landmark’
rather than to ‘the same (lump of) stone’. And of course it could. For all
one knows, ‘Cleopatra’s Needle’ in some people’s use does work this way.
In that case, if the stone obelisk brought from Alexandria to London in
– is eroded by the atmosphere, and is continuously repaired with
concrete, so that in the end none of the original stone is left, then we
shall have to say, ‘The same landmark, namely Cleopatra’s Needle, was
stone and is concrete.’ But now, whereas it would be true that Cleopatra’s
Needle in  is the same landmark as Cleopatra’s Needle in , it
would be false that Cleopatra’s Needle in  is the same stone as
Cleopatra’s Needle in  – or, indeed, the same stone as anything. For
it wouldn’t be a stone in . This gives a case where Cleopatra’s
Needle is in  both an f and a g, both a landmark and a stone, and
goes on being the same f but doesn’t go on being the same g.
‘To take another case, during the Festival of Britain the stone in
Meriden, inscribed to show that it marks the reputed centre of England,
was removed from Meriden to London to be put on show. Such a per-
formance is well within the limits of human folly. Well, during transport
it will have remained the same stone but not the same landmark as the
Meriden stone. Moreover, after its return to Meriden, it is questionable
whether it is any longer that same landmark. After its return, old villag-
ers alleged that it was put back some yards away from its old site.’
These cases, (␣) (␤) (␥) seem to qualify, if they qualify at all, as cases
of type (), though for one or two of them type-() analyses might also
be suggested.
15
Mind,  (Oct. ).
 The absoluteness of sameness 

.                         (  )
(␦) An argument in Geach’s Reference and Generality, p. , might prompt
the following suggestion. Whatever is a river is water. Suppose I moor
my vessel at the banks of Scamander when that river is in full torrent.
The next day, the river on which my vessel is now moored is the same river
as the river on which I moored it yesterday. But, even though rivers are
water, the river is not the same water. The water on which I moored it is
now part of the Aegean Sea.
(␧) John Doe the boy is the same human being as Sir John Doe the Lord
Mayor, but not the same collection of cells as Sir John Doe.16
(␨) ‘[I]t may be said, without breach of the propriety of language, that
such a church, which was formerly of brick, fell to ruin, and that the
parish rebuilt the same church of freestone, and according to modern
architecture. Here neither the form nor the materials are the same, nor
is there anything in common to the two [sic] objects but their relation to
the inhabitants of the parish; and yet this alone is sufficient to make us
denominate them the same.’17 So we may say of Hume’s church that the
present church is the same church as the old parish church but not the
same building or the same stonework as the old parish church.
(␩) At Paddington Railway Station I point in  to the Cornish
Riviera Express and say: ‘That is the same train as the train on which
the Directors of the Great Western Railway travelled to Plymouth in
.’ Same train, yes, it may be said, but not the same collection of coaches and
locomotive.
(␪) A petitioner asks to see the same official as she saw last time. The
man she sees is the same official but not the same man. (Cf. Geach, op. cit.,
p. .)
(␫) The Lord Mayor is not the same official as the Managing Director of
Gnome Road Engineering Ltd (indeed they often write one another
letters) but he is one and the same man.
(␬) Dr Jekyll and Mr Hyde were the same man but not the same person or
personality. (Cf. Locke, Essay, , ,  and .)
(␭) ‘There is but one living and true God . . . and in unity of this
Godhead there be three Persons of one substance, power, and eternity;
the Father, the Son, and the Holy Ghost.’ (Article  of the  Articles.)
16
Cf. Hobbes, De Corpore, Ch. , §: ‘It is one thing to ask, concerning Socrates, whether he be the
same man and another to ask whether he be the same body; for his body, when he is old, cannot
be the same as it was when he was an infant, by reason of the difference of magnitude; yet nev-
ertheless he may be the same man.’ 17
Hume, Treatise . (p.  in the Everyman edition).
 The absoluteness of sameness
This is to say that the Father, Son and Holy Ghost are the same God but
not the same person.18
Some of these examples are more convincing than others, but none
of the examples (␣)–(␬) is sufficiently secure to provide an independent
argument for the conceptual possibility of (␭), which is the most difficult
case. I shall argue that all the perspicuous cases repose on ambiguities of
reference or logical form or on special uses of ‘be’ (e.g. the use I shall call
the constitutive use). None of them shows a way for relations like ‘is the
same horse as’, ‘is the same river as’ (in the usage that ties them to ques-
tions of identity and existence) to escape from the congruence require-
ment. Any reader who is prepared to take this on trust (or can satisfy
himself of it) should skip now (or so soon as he is satisfied) to section 
of this chapter. The suggestion to skip is intended seriously.

.    -(  )     
(␣) and (␤) hang together. For if the jug is the same collection of bits as
the heap of fragments and the heap of fragments is the same collection
of bits as the coffee pot, then, by transitivity, the jug must be the same
collection of bits as the coffee pot. Either both or neither, then, is a true
identity-statement. The difficulty is that if the jug is the same collection
of material parts or bits of china clay as the coffee pot, that is if they are
one and the same collection of china-bits, then their life histories and
durations must be the same. (Call that the life-histories principle.) But
the coffee pot will be fabricated or assembled at t3 by my ingenious friend
and exist only from then on. The jug won’t then exist any more.
(␣) will only be what is required as a case of type () if the sentence
‘that heap is the jug you saw last time’ comes to something more than
‘the matter you see there is the same matter as the matter of the jug you
saw when you came here last time’. Similarly (␤) must not be reducible
to the unexceptionable claim that the jug and coffee pot are made of the
same matter. Otherwise, it is no longer obvious that we have the type-()
identity-statement the relativist required. To get that, the ‘is’ in ‘is the
same collection of china bits’ of (␣) and (␤) must mean ‘⫽’ and we must
take collection of china-bits as a straightforward covering concept. The ‘is’
in question must not be comparable to the ‘is’ in ‘The soufflé you are
eating is flour, eggs and milk’ or the ‘is’ of ‘The portico is wood and
18
A collection of the writings of the Church fathers on this matter is to be found at pp. –, ff.,
ff., ff. of Documents of the Christian Church, selected and edited by Henry Bettenson (Oxford,
). See also op. cit. in note  below.
 The absoluteness of sameness 
stucco’. I shall call the latter the ‘is’ of constitution, contrast it with the ‘is’
of identity, and shall attempt to prove that it is precisely this constitutive
‘is’ that we have in (␣) and (␤). If I am right, then this occurrence of ‘is
the same collection’ means ‘is constituted from the same collection’.
Suppose, with (␣), that the jug is straightforwardly the same collec-
tion of china-bits as the heap of fragments. Then, if this is a type-()
identity-statement, we are entitled to infer that the jug is a collection of
china-bits. (If Hesperus is the same planet as Phosphorus then Hesperus
is a planet.) But then there must be some collection of china-bits with
which the jug is identical. (If you doubt the principle linking identity
and predication of this sort, namely that if x is ␾ then there must be
some ␾-thing with which x is identical,19 then consider whether you will
deny this instance of it: if Hesperus is a planet then there is some planet
with which Hesperus is identical.) Suppose there to be some such col-
lection. Then, again, we have trouble from the principle that, if a and
b are identical, then they must have the same life history. Suppose I
destroy the jug. Do I then destroy the collection? Either I do or I don’t.
If I do, then both (␣) and (␤) fail to be true with covering concept collec-
tion of china-bits and fail as type-() examples. If I don’t thus destroy the
collection, then it cannot be true of the jug that it is a collection of
china-bits in the normal predicative sense of ‘be a . . .’. Nevertheless it
is true, under some other reading of ‘be’, that the jug is a collection of
china-bits. That is to say that it is china-clay. It is true in the sense that
the jug is made of china-clay or constituted of a collection of china-bits.20
This is what is predicatively true of the jug. So much seems obvious. But
the argument requires two supplementary remarks. One remark con-
cerns the behaviour of ‘same’ in (␣) and (␤) within the gloss same collec-
tion, the other remark concerns the sense of ‘collection’. Together, these
19
A principle no less true for having prompted false theories of predication, e.g. Antisthenes’ iden-
tity-theory. Cf. Aristotle, Metaphysics, b32; Hobbes, De Corpore, ch. , §. It implies no such
absurdity. The principle is denied by Noonan in Mind (), p. . Cf. also p. , where
Noonan claims something can be a man by constituting that man but be not identical with that man:
‘What constitutes a man is not identical with that man, but on my account it is identical with
something which is a man, namely itself.’ For further discussion of Noonan’s position, see my
reply to him in Lovibond and Williams ().
20
I am not saying that the possibility of this paraphrase by itself forces us to postulate this distinct
sense of ‘is’. l am saying that the independent plausibility of this paraphrase, plus the plausibil-
ity of Leibniz’s Law which would otherwise have to be amended or abandoned, plus the difficul-
ties of amending Leibniz’s Law or finding any f-restricted version of this principle for same f that
stands in the right connexion to what it is for an f thing to exist, forces us to postulate this dis-
tinct sense of ‘is’.
For collateral evidence supporting the hypothesis, as first formulated in Wiggins (), that
this distinct reading of ‘is’ needs to be postulated, see the article by Tyler Burge cited at footnote
 of Chapter Two.
 The absoluteness of sameness
remarks will occupy all seven paragraphs that intervene between here
and the transition to example (␥).
The argument for the non-identity of jug and collection does not rely
on a special or unfair construal of the term ‘collection’. The possible
construals seem to be three in number. ‘␾ (A)’ where A is in some sense
or other a collection can presumably mean either (i) that class A is ␾ or,
(ii) that each of the As is ␾, or (iii) that a physical aggregate A is ␾.
Sense (i) cannot be what we are really looking for, even if skilful and
opportunistic reinterpretations of ␾ might hold a set-theoretical inter-
pretation of the ‘A’ in ‘␾ (A)’ onto the rails for an indefinitely long time.
In the end, the only way in which one could explain breaking or scatter-
ing a set-theoretical entity would have to be parasitic on the way one
explained what one had to do to a physical configuration to break or
scatter it. At root, the thing we need to be interested in is a notion of col-
lection or manifold for which there can be no empty or null collection,21
and for which it holds that ‘if we take the German Army as our mani-
fold and an infantry regiment as a domain within it, it is all one whether
we choose to regard as elements within it the battalions, the companies,
or the single soldiers’.22 Notoriously this is not true of sets.
Sense (ii) of collection is not what we are looking for either. If I repair
or destroy an item, I do not repair or destroy each part of it. (Since each
part of a part is a part this would be difficult.) Nor in any non-
Anaxagorean universe do we wish ‘Jug (A)’ to mean each of the As is a
jug.
Sense (iii) suggests the definition of sum or fusion in Lesniewski’s mer-
eology.23 An individual X would be a Lesniewskian sum of [all elements of
the class of parts of the] jug J if all [elements of the class] parts of J were
parts of X and if no part of X were disjoint from all parts of J.24 This
would certainly seem to be the sort of thing that we are looking for,
because, according to this notion, all collections of parts of the jug,

21
‘If we burn down all the trees of a wood we thereby burn down the wood. Thus [in the concrete
sense of class] there can be no empty class.’ Frege’s review of Schroeder’s Algebra der Logik in
Geach and Black’s Translations, p. . 22
Ibid, p. .
23
For a description of mereology (the calculus of individuals) see Tarski (), p. ; or Nelson
Goodman’s Structure of Appearance (Cambridge, Mass., ), ch. . See also Woodger (), ch.
, §, and appendix E (by Tarski), p. .
For one capital philosophical defect of mereology as thus formulated, see footnote  of
Chapter Three. I am relieved and encouraged to find my reservations about these mereological
systems vindicated in Simons ().
24
The definition of ‘Y is disjoint from Z’ is ‘no individual W is a part both of Y and Z’. The ref-
erence to classes in the definition of sum of elements is eliminable (as is indicated by the square
bracketing). The ‘part of ’ relation is transitive in mereology.
 The absoluteness of sameness 
however specified (whether as china-clay bits or as molecules, or as
atoms), and all collections of collections of parts of the jug, etc., are
intended to define and exhaust one and the same Lesniewskian whole or
sum, X, of the jug. If ‘collection’ is defined in this way, however, and if
mereology is grafted straight onto that pre-existing scheme of three-
dimensional persisting things and their parts that we are operating (and
anybody who wanted to obtain our type-() or type-() contrasts would
have to be willing to operate), then perhaps the jug turns out not to be
the same collection, in this sense of ‘collection’, as the coffee-pot in (␣)
and (␤ ). For if X⫽J then among the parts of X is J itself. For everything
is part of itself. So if J is broken at t2 and there is no such jug as J after
t2, then it looks as if X does not survive t2 either.25
In fact the problems that would arise in adding mereology to a logical
system already possessed of a concept of identity defined for three-
dimensional continuants, whose parts are not specified in terms of a pre-
determined scheme of logical ‘atoms’, have been studied insufficiently.26
The reason for this is that the adherents of mereology have almost
always wished to operate a four-dimensional scheme which (as I have
already complained) reduces everyday continuants to temporal series of
slices, ‘thing-moments’ or spatio-temporal regions of the space–time
continuum, and have usually advocated a most implausible semantic dis-
tinction between identity or difference at a time and identity or differ-
ence through time. (More is said against this below in Chapter Six, §.)
For the former concept of identity, they say that ‘x⫽y’ can be defined
mereologically by the condition that x is a part of y and y is a part of x;
whereas for spatio-temporal continuity, or what is sometimes called gen-
identity, these definitions have to be supplanted or supplemented by
special conditions of a quite different character. The thing that chiefly
matters here, in a discussion of (␣) and (␤), is that, no matter how these
extra conditions may be stated and whatever alternatives there may be
to Lesniewski’s general method of defining ‘concrete collection’ and
however three-dimensional wholes may be accommodated, the same
fundamental dilemma remains. Either ‘concrete collection’ is defined in
25
It is pointless to try to counter this by redefining the Lesniewskian whole of J without including
J as part of J. One cannot destroy the non-proper part of J without affecting the proper parts of J
and doing something just as drastic to them as to J. For to shatter J, or even break it into two, is
to shatter the indefinite number of proper parts of it that lie across the break. This may amount
to the destruction of these parts. (That depends on what destruction is deemed to amount to in
the mereological framework.) The ordinary continuants of ordinary three-dimensional talk are
not built up in any predetermined way from logical ‘atoms’ and aggregates of these.
26
This remark dates from before the appearance of Peter Simons’s book. For his book, see above,
Preface paragraph six.
 The absoluteness of sameness
such a way that concrete collection X has the same principle of individ-
uation as the jug, or it is not so defined. If it is not, then the life-histories
principle debars X from identity with the jug and the type-() example
disappears.27 But if X does have the same principle of individuation as
the jug, then again, for a different reason, we don’t have a type-()
example. For under this option, the jug isn’t the same concrete collection
as the coffee pot. What is more, the chances are that the whole project
of equating thing and matter will then have degenerated into triviality.
If X is to be defined so as to be no more and no less tolerant of
damage or of replacement of parts, etc., than is the jug, then sponsors
of X will need to help themselves to the everyday continuant concept jug
so as to secure the right configuration and persistence-conditions for X.
But this is to ascend from the level of bits of things to the level of some-
thing whole, structured and jug-like, namely a jug. The jug is constituted
of certain matter and identical only with a certain whole or continuant
at present constituted in a certain way out of that matter. That is to say
it is identical with the jug. Unless the project is thus trivialized and con-
crete collection is so defined, the true statement that the jug or the coffee
pot is X must not be allowed by anybody who accepts the life-histories
principle to have the standard consequence of the ordinary ‘is’ that it is
identical with the mereological thing X. The ‘is’ must mean ‘is consti-
tuted of ’, and collection of parts will not function standardly as a normal
covering concept in the locution ‘is the same collection as’, as it figures
in our examples (␣) and (␤).
Since ‘the jug is the heap of fragments’ and ‘the jug was the same
china-clay as the coffee pot’ both boil down to identity of matter, the
supplementary remark about ‘the same’ that was promised seven para-
graphs ago is simply this: the words ‘the same’ can do appropriate duty
with constitutive ‘is’ just as readily as they can do their ordinary duty
with ‘is’ taken in the sense where it yields so-called numerical identity
(the sense paraphrasable as ‘is the same substance or continuant as’). So much
for (␣) and (␤).
27
Still a doubt may persist. Isn’t the life-histories principle too strong? Might not the jug be iden-
tical with a stretch of some Lesniewskian whole X for such time as no part of the jug is broken
or replaced? But quite apart from the support we have adduced for the strict life-histories prin-
ciple, this ‘temporary identity’ is surely a peculiar sort of identity. See now my reply to Noonan
in Lovibond and Williams (). We surely cannot give a sense to the supposition that Hesperus
might be the same planet as Phosphorus for a bit and then stop being Phosphorus. But then the
relation between the jug and the redefined whole X looks as if it cannot be the same sort of rela-
tion as that between Hesperus and Phosphorus. The conclusion for which I am arguing is of
course just this, that they are related by the one being composed or constituted of the other, not
by identity.
 The absoluteness of sameness 
Example (␥) also requires considerable unpacking, but I think its
power to convince is quite deceptive. We may begin by asking what is
meant by ‘Cleopatra’s Needle’. What it is that someone points to when
he points to Cleopatra’s Needle. There is here a special difficulty that has
to be faced by a consistent defender of the position Geach took up in
Reference and Generality. To keep (␥) in play at all as a type-() example, the
defender will have to claim that landmark and stone give different princi-
ples of identity. But by the theory of proper names defended in Reference
and Generality, the sense of a proper name is given by the principle of
identity built into the general term associated with it. It seems to follow
that if ‘Cleopatra’s Needle’ had two equally good but different ‘nominal
essences’ then it ought to be ambiguous. In which case (␥) should not sur-
prise or impress us any more than any other startling paradox arrived at
by equivocation.
Rather than object in general to this theory of proper names,28 let us
simply examine the different specifications one might give of the
received meaning of ‘Cleopatra’s Needle’. What again is Cleopatra’s
Needle? Is it a stone? If a stone is what it is, if ‘it is a stone’ properly
answers Aristotle’s what is it question, then, so soon as that stone is rotted
away completely, Cleopatra’s Needle is rotted away completely. For they
are one and the same stone. Cleopatra’s Needle, the stone, is not then
the same anything as anything that exists in . For if the stone,
Cleopatra’s Needle, no longer exists in  then it is not then the same
landmark as anything, even if something different may have come to
fulfil the same role as it did.
But perhaps the fact the stone has completely rotted away by 
does not imply that there is no longer any such thing as Cleopatra’s
Needle. Stone is not then the sense-giving sortal. It may be that monument
or monument suitable for use as a landmark is what Cleopatra’s Needle sub-
stantially is. And perhaps monuments can be completely refashioned
and still persist. But then ‘Cleopatra’s Needle in  is not the same
stone as Cleopatra’s Needle in ’ need only mean that Cleopatra’s
Needle is not made of the same material as it was in . The dates
surely qualify the verb in any case. Cleopatra’s Needle is not constituted
in  of the same material as Cleopatra’s Needle was constituted of in
. Once its matter was a (piece of) stone, now its matter is concrete.
In that case the words ‘the same’ are serving in (␥), with the versatility
already remarked upon in connexion with (␣) and (␤), to indicate that
28
Which could be defended against the criticisms of Linsky all the more effectively if the general
thesis of the present chapter were correct.
 The absoluteness of sameness
you can’t say about the material of Cleopatra’s Needle in  what you
could have said in . (A type-() analysis might also be attempted.)
These are not all the possibilities. One might think landmark was what
gave ‘Cleopatra’s Needle’ its sense. But there is something rather pecu-
liar about treating ‘landmark’ as an ordinary substance-concept suitable
for giving a proper name a sense. It is really more like a title conferred
on an object when it attains a certain position of conspicuousness.
Compare ‘chairman’ or ‘official’ or ‘president’ or ‘sovereign’. In that
case, in one use, landmark is a restriction of a sortal concept. It presup-
poses the availability in principle of some underlying sortal predicate
that says what sort of object. This takes us back to the alternatives
already mentioned. There does, however, exist the possibility of another
use, which one might call a titular use, of the phrase ‘same landmark’.
According to this, for x to be the same landmark as y, x has only to mark
the same spot as y did.29 But in this use, something else, something non-
identical with the obelisk and distinct under every genuine covering
concept, can succeed it as the same landmark. It must be this view of
‘landmark’ that (␥)’s defender exploits in suggesting that the Meriden
stone ceases to be the same landmark when it is transported to London.
‘Cleopatra’s Needle’ then turns out not to be an ordinary proper name
at all. It emerges as an abbreviation for the description ‘whatever suit-
able object of suitable dignity conspicuously marks such and such a spot
on the Embankment in London’.
The effectiveness of this critique of example (␥) does not depend on
there being a hard and fast or canonically correct answer to the question
‘what is Cleopatra’s Needle?’ The example may owe a specious plau-
sibility precisely to the fact that ‘Cleopatra’s Needle’ can remain ambig-
uously poised between these and perhaps yet other incompatible
interpretations.
With this, the alleged cases of () are concluded. It begins to appear
why there simply cannot be cases of type (). Where (᭚f) ⫽ and
a b
f 冢 冣
a b
g 冢 冣
allegedly (᭚g) not ⫽ & g(a) v g(b), either g is a substantial sortal

29
If this use exists and extends to ‘sovereign’ then all that is required for Queen Elizabeth II to be
the same sovereign as Queen Elizabeth I is that she should rule the same country. And perhaps
‘The same sovereign was a man, is now a woman’ need not signify that anybody has changed
sex (unless ‘anybody’ be thought of as adapted to perform precisely the same trick as ‘sovereign’).
Queen Elizabeth herself simply succeeded a man.
 The absoluteness of sameness 
concept or it is not. If it is not substantial then the champion of (␥) needs
to show that we have more than a type-() case or a case of constitutive
‘is’. If g is a substantial sortal concept, however, then either a or b has to
fall under g without the other falling under g. But that has been excluded.

.    -(  )             

         ’ 
The identity-interpretation of (␦) and Leibniz’s Law are incompatible.
Unlike the water, the river on which I moored my boat yesterday is not
a part of the Aegean. Rivers are indeed water but this means that water
goes to make them up. ‘Same water’ is not therefore a covering concept
for an identity statement identifying a river with something.
(␧) is fairly easily unmasked. If ‘collection of cells’ will do as covering
concept, and if ‘is a collection of cells’ doesn’t merely mean ‘is made up
of cells’, then John Doe must be identical with some definite collection
of cells, and this will have to share all properties of John Doe. What col-
lection of cells? Suppose we make ‘collection of cells’ mean ‘such and
such aggregate’ (with fixed constituents). But then one aggregate is suc-
ceeded by another. John Doe is not similarly succeeded. One aggregate
is dissipated. John Doe isn’t. But then ‘John Doe is such and such collec-
tion of cells’ has to have the constitutive interpretation. (Compare (␣)
and (␤).) Suppose we make ‘collection of cells’ mean ‘composite with a
succession of constituents through time’. Then we no longer have an
example of type () at all, i.e. change of truth-grounds by change of cov-
ering concept. For in this sense man and boy are the same collection of
cells.
(␨) resembles (␥) in a certain specious exploitation of ambiguities. The
example has intrinsic interest, but, because the analysis is at once intri-
cate and at least in part predictable, I relegate the discussion to a note.30
30
Hume continues in tell-tale manner: ‘But we must observe, that in these cases the first object [sic]
is in a manner annihilated before the second comes into existence; by which means, we are never
presented, in any point of time, with the idea of difference and multiplicity; and for that reason
are less scrupulous in calling them the same.’ Hume has less interest than we ought to have in
pressing the point I am about to make because, having distinguished respectable strict ‘numeri-
cal’ identity from what he variously dubs ‘imperfect’, ‘specific’, ‘fictitious’ identity (the lamen-
table ancestor, I suppose, of ‘genidentity’, and a notion which on one occasion he even calls a
perfect notion of diversity), he can then afford to let you say whatever you please about the dis-
reputable second notion. But we must insist – either we make up our minds to say the building
was annihilated or we do not make up our minds so.
If we do say the building was annihilated, then a certain building seems to be found in exis-
tence after it lapsed from existence. It is easy to avoid this absurdity. Here are two or three differ-
ent ways. First take ‘church’ in its first and third occurrences in ‘the present church is the same
 The absoluteness of sameness
In (␩) the appearance of a type-() case relies entirely on the failure to
say what is meant by ‘Cornish Riviera Express’. Once this is specified,
all semblance of support for R disappears. Manifestly, to admit the pos-
sibility of an express surviving its present coaches and locomotive is to
admit its non-identity with these. But then collection of coaches and locomo-
tive is a non-starter for straightforward covering concept, and we have a
constitutive ‘is’. (Alternatively, perhaps we have a titular use of ‘same
train’ or ‘same express’.)
(␪) is equally easily exposed. Suppose official a is succeeded by official
b. The petitioner therefore sees b on her second visit. She doesn’t see the
same office-holder but the holder of the same office, whoever he is. The
sentence ‘a is the same official as b’ doesn’t ascribe ‘numerical identity’
to a and b at all. It predicates something of them in common, namely
holding a certain office. In the same’s extensive repertoire this is one of the
better-known roles. (Cf. landmark, and example (␥).)
(␫) also exploits an ambiguity. Under one interpretation it is simply
false that the Lord Mayor is not the same official as the Managing
Director of Gnome Road Engineering. John Doe, that tireless official, is
both Lord Mayor and Managing Director. So the Lord Mayor is the
same tireless official as the Managing Director. The interpretation that
makes the first part of (␫) true concerns what it is to be Lord Mayor and what
it is to be Managing Director of Gnome Road Engineering. These Frege might
footnote  (cont.)
church as the old church’ to mean ‘building’. Take the second occurrence of ‘church’ to signify
what can persist when a congregation loses its church building. Mark this latter sense by capitals.
Thus the sentence signifies something like ‘the present church building embodies/
realizes/houses the same Church as the old church building embodied/realized/housed’. If this
really is a way of reading the sentence that we are concerned with, then, there must be some ‘is’
of ‘realization’ which extends the ‘is’ of constitution. (Here cf. (␮) later.) To add to the sentence
‘the new church building is the same Church as the old church building’ the qualification ‘but it
is not the same building’ imports a zeugma at ‘is’, which reverts the second time round to its ‘⫽’
sense, with consequential change in the grammatical role of the expression following ‘same’. A
second way is to take ‘church’ to mean ‘Church’ throughout in the affirmative sentence ‘the
present church is the same church as the old church’, and to take the addition ‘but it is not
the same building’ as importing a zeugma at the subject place: ‘… but the new building is not the
same building as the old’. Third, let ‘church’ mean building at its first and third occurrence and
behave in the titular fashion and unlike a proper covering concept at its second occurrence, as it
was supposed ‘land mark’ might behave in (␥) (meaning say ‘whatever building houses such and
such a congregation’). We lose the type-() identity-statement whichever analysis we adopt.
If we do not make up our minds to say that the church was annihilated then either we do not
know what to say about the example, or we say that it was not annihilated. It is then false that
the new church is not the same building as the old one. It is the same, and has simply been
repaired and remodelled. That is what we have decided to say. Again the rebuttal does not
depend on its being a hard and fast question how ‘church’ behaves in the example. As before,
the example cannot be permitted to survive by poising itself ambiguously between mutually
exclusive alternatives.
 The absoluteness of sameness 
have called concepts. And what (␫) then says is that to satisfy the one
concept and hold the one office, is not to satisfy the other and hold the
other office. To add ‘they often write one another letters’ is to make a
zeugma. For now ‘they’ must be understood, the second time round, as
‘the man who satisfies this concept and the man who satisfies that
concept’. And it is in this sense (in which they are the same official) that
‘they’ are the same man.
(␬) touches on large issues to which we shall return in Chapter Seven.
It is certain that neither human being nor homo sapiens is synonymous with
person. But this does not prove the point. And the difficulty is this. If, as
(␬) says, Mr Hyde and Dr Jekyll are the same man, then, if Mr Hyde
visited Tilbury Docks at . p.m. on  December , then Dr Jekyll
did so too, and did whatever Mr Hyde did. Now Dr Jekyll the man is a
person and he did these things at Tilbury Docks. But then is he not the
person who did these things at the docks? How then can he be a differ-
ent person from Mr Hyde? The only way to make (␬) come out true is
to give it a rather odd and implausible interpretation, and interpret it to
mean that to satisfy the concept person who is ␾ is not necessarily to satisfy
the concept person who is ␺. To have these personal characteristics is not
necessarily to have those personal characteristics. Contingently, though,
to satisfy the one concept was (in R. L. Stevenson’s story) to satisfy the
other. ‘Dr Jekyll’ and ‘Mr Hyde’ have then to be read twice over in (␬) to
make it come out true, first as standing each for a man (this individual is
the same man as that individual), the second time as standing for a
certain kind of character or personality. (These personalities, not these
men, are different.) But this spoils all prospects of seeing it as a case of
R; and the example really represents an implausible attempt to postu-
late philosophically defined schizophrenia without going the whole way
and postulating two men sharing one body, each taking his turn to
control it.
This brings me to (␭), and that which is becoming overdue, the re-
examination of Leibniz’s Law. No one could hope to exhaust the theo-
logical implications of what is at issue here or examine every possible
formulation, but the plain difficulty is that, if the Son who was God was
crucified and was the same God as the Father, then, according to
Leibniz’s Law unamended, the Father was crucified. I understand that
this involves one in the heresy of patripassionism. What is more, one appli-
cation of Leibniz’s Law is as good as any other. If the Fathers of the
Church had allowed the Patripassiani their way, then the three Persons,
Father, Son and Holy Ghost, would have been in danger of collapsing
 The absoluteness of sameness
into one another. For, in exactly the same way, all the predicates of Christ
that applied uniquely to him or applied to him at a time and place will
have applied to the Father and Holy Ghost; how then could Christ fail
to be the same person as these?
Short of heresy, which offers an indefinite variety of reformulations,
the determined defender of (␭) needs to withhold Leibniz’s Law. The
trouble is that Leibniz’s Law underwrites the substitutivity of identity,
and this is a principle not long dispensable in any form of reasoning. It
follows that the defender must find a way to qualify it or restrict its appli-
cation. Even as he brackets the full Leibnizian Principle, the defender of
(␭) must make positive provision for some however circuitous advance
a b
f 冢 冣
from ␾a and ⫽ , taken as true of some a, b and f that satisfy any
extra condition he requires, to the conclusion ␾b. There must be some
condition or other under which the circumstance of b’s having a genuine
property may be deduced from the circumstance of a’s having that prop-
erty.31
A method of dispensing with Leibniz’s Law for ⫽ that Wilfrid
x y
f
Hodges and I once tried to explore was to make do with this property of
⫽
:
f

LL.II: 冢a ⫽f b冣 →(᭙␾)冤(᭙x)冢x ⫽f a → ␾x冣 ↔(᭙y)冢y ⫽f b→ ␾y冣冥

31
For an attempt to restrict congruence by defining a class ⌬f of predicates such that Leibniz’s Law
assumes the qualified form:
[x ⫽ y → (᭙␾) [␾ 苸⌬[f(x) ∧ f(y) ∧ [␾ (x)→ ␾ (y)]]]],
f
see Nicholas Griffin, Relative Identity (op. cit. note ), pp. –. The conditions Griffin places on
⌬f are as follows:
(i) f (␰) 苸⌬f
(ii) If ␾ (␰ ) 苸⌬f then ¬ ␾ 苸 ⌬f
(iii) If ␾ (␰ ) 苸⌬f and ␺ (␰ ) 苸⌬f then [␾ (␰ ) ∧ ␺ (␰ )]苸⌬f; and conversely (provided [␾ (␰ ) ↔ ␺ (␰ )])
(iv) If ␾ (␰ ) 苸⌬f and ᭙x[␾ (x) → ␺ (x)] then ␺ (␰ ) 苸⌬f
(v) If ␾ (␰ ) 苸 ⌬f and ␺ (␰ ) results from ␾ (␰ ) by the replacement of every occurrence of one or
more variables in ␾ (␰ ) then ␺ (␰ ) 苸⌬f
(vi) Cf (␰ ,c)苸 ⌬f , where ‘Cf (␰ ,␩)’ is a complex predicate which expresses the identity criteria con-
veyed by ‘f ’ and ‘c’ is an arbitrary constant.
S. G. Williams has shown that, on these terms, Leibniz’s Law will come back unrestricted. His
argument is that, by (i) f (␰ ) belongs to ⌬f. So by (ii), ¬f (␰ ) belongs to ⌬f, as therefore does [f (␰ ) ∧
¬ f (␰ )], by (iii). But (᭙x)[[f (x)∧¬ f (x)]→ ␾ (x)]→ ␾ (x)]. So, by (iv), ␾ belongs to ⌬f. But ␾ here
can be any extensional predicate. So every such predicate belongs to ⌬f. As Williams remarks,
there is something instructive in the way in which Griffin’s very natural conditions render back
to us the classical Leibnizian principle.
 The absoluteness of sameness 
This says that, if a is the same f as b, then whatever properties transfer
from a to anything that is the same f as a, these properties will transfer
from b to anything that is the same f as b. This is not doubtful. Suppose
we know that
() Cicero ⫽ Tully
man
and
() Denounced Catiline (Cicero).
Then, by modus ponens and universal instantiation, it follows that

( ) (᭙x)(x ⫽ Cicero →Denounced Catiline (x))

man
if and only if

(᭙y)(y ⫽ Tully →Denounced Catiline ( y)).

man
But, to get anything more interesting than (), we need something that
cannot be regarded as guaranteed when unrestricted Leibniz’s Law is
withdrawn, namely the mass of truths such as

() (᭙x)(x ⫽ Cicero →x denounced Catiline).

man
In the presence of (), we could of course deduce
() Tully denounced Catiline.
But from whence do we obtain ()? On what, rather, does the relativist
take the question of the truth or falsity of claims such as () to depend?
And what, in the absence of a general principle of congruence for
x⫽y
, does he take to ground the truth of propositions that link
f
particulars and attributes in the way in which () links man, Tully, Cicero
and denounced? Why is there no similar true proposition leading to a
similar deduction with the subjects The Father and The Son, the sortal
predicate is God, and the predicate was crucified at such and such a time and
place? The answer must be that the connexion that holds between man,
Cicero, Tully and denounced and makes () true does not hold between
God, the Father, The Son, and crucified at such and such a time and place. It
might be said, as Hodges put it, that crucified does not ‘transfer over’ the
 The absoluteness of sameness
concept God. But why not? There must be some principle of some sort
to be found here from which one could gain an idea of what counts as a
consideration for or against claims such as ().
These questions become more immediately important for present
purposes – among which I do not here number attempting to refute (␭)
under all interpretations, for all I seek to show is that (␭) is not an iden-
tity-statement with any independent leverage in the larger logical
dispute – if we ask how the predicate God who was crucified can fail to trans-
fer over the concept God. Yet, if Christ is God and was crucified, then it
seems hard to deny that he is God who was crucified.
One way to block the inference to ‘Christ is God who was crucified’,
and this is a way that has attracted some thinkers, is to deny that it was
qua God that Christ had a body or was crucified.32 But if this escape is
used, then there is a simpler relativization of Leibniz’s Law which prom-
ises to provide both the restricted congruence and the adjudication that
we have wanted for sentences like (). It reads

LL.III: 冢a ⫽f b冣 →(␾(a) as an f↔ ␾ (b) as an f).

In assessing LL.III, let us now suppose that LL.II and LL.III are both
available. We are to suppose that all identity is reached through relative
identity. We are to suppose also that this is our only general account of
how a sortal concept f and a property must be related if the property is
a⫽b
to transfer from a to b where . The rest will be given piecemeal, we
f
are to suppose, from the semantics of the various predicates in all their
various predicative roles.
Now there are undoubtedly places where qua or as or similar devices
occur essentially and an object has to be characterized (␾a as an f) and
(not-␾a as a g). But I submit that this is a circumscribed and special phe-
nomenon, ill suited to the larger role for which it is being recruited.
32
This will lead to a division of predicates similar to one which Hodges pointed out to me that
Pope Leo made in his Tome of . ‘Deus per id quod omnia per ipsum facta sunt et sine ipso
factum est nihil; homo per id quod factus est ex muliere, factus sub lege . . . esurire, sitire, lasses-
cere atque dormire evidenter humanum est.’ See also Griffin, op. cit., pp. –.
On any view resting on the non-transferability of crucified with respect to God new questions
arise. For instance, would it then be possible for Christ the God to be the same anything as Christ
the person that was crucified? Either yes or no. If no, (␭) disappears. If yes, Christ is the same
God as the Father. Furthermore he is a person who was crucified. He is God then and he is cru-
cified. But then is he not God who was crucified? Again, what follows, if anything, from the fact
that Christ, who was qua person crucified, seems to satisfy the complex predicate ‘was, qua God
incarnated, qua person crucified’?
 The absoluteness of sameness 
() It may arise with what have been called attributive adjectives such as
big, small, tall, short, real, good, bad. Thus a ship can be big among destroy-
ers and small among cruisers, a man tall among Japanese but short
among Americans, a wooden duck a real decoy duck but not a real duck,
a witticism a good joke but not a good thing to say at that particular
moment.
() There can be reference via qua or as to a role that some person or
thing plays. As a general he was obliged to be present at the Court
Martial, as the best marksman in the regiment not. Sir John Doe sent the
letter as Managing Director (i.e. in the course of his duties for Gnome
Engineering), not as Lord Mayor.
() Oratio obliqua and straightforward referential opacity can also take
cover under as or qua. x sent the letter to Sir John Doe as Lord Mayor
not as Managing Director (i.e. x addressed or directed the letter ‘The Lord
Mayor’). Philip, who does not know that Cicero is Tully, may believe
Cicero, as Cicero but not as Tully, or believe Tully, as Cicero but not as
Tully, to have denounced Catiline. Philip may think of him under this
name rather than that.
These cases are well demarcated and fairly well understood. What the
present proposal seeks to do, however, is to generalize the phenomenon
they represent, without doing anything to show us how to determine the
relevant ‘qua f ’ for each and every property asserted to qualify a given
individual, or how to decide the applicability, meaning and satisfaction
conditions of the complex predicates thus arrived at. At greater length,
a rather more exhaustive and exact typology could have been devised for
qua and all similar devices. But it is certain that qua is not ubiquitous in
its operation.33 Not every adjective has an attributive use. Individuals are
many things (satisfy many predicates) otherwise than by virtue of playing
some role or other. Referential opacity is not to be found absolutely
everywhere. Moreover, qua does not always produce a single kind of
effect for which a unitary rule of the proposed kind could be laid down.
Consider the difference that, with certain precautions, ‘qua f ’ is detach-
able salva veritate from affirmative ‘␾ qua f ’ in types () and (), and not so
33
I owe reference to an example which falls outside these three kinds to Professors Geach and
Anscombe. See Aristotle, Physics, ., b. Consider the road from Athens to Thebes. It is the
same road as the road from Thebes to Athens. But the road Athens–Thebes is uphill and the
road Thebes–Athens is downhill. My objection to allowing this example (qua Athens–Thebes
uphill, qua Thebes–Athens downhill) is that either ‘road’ means an actual feature of the land-
scape, in which case ‘uphill’ collects a specification giving the direction of the slope, and there is
a simple relational predicate true of that road; or else ‘road’ means ‘journey by road’, in which
case there is no identity. (The fare for Thebes–Athens may be more than the fare for
Athens–Thebes.)
 The absoluteness of sameness
removable in type (). In kinds () and (), the qua is simply explicative
and serves to demarcate that which may be detached. Which kind of qua
is it that is involved in LL.III? Is the intention behind putting forward
LL.III to withhold transferability entirely for predicates and verbs that
do not need a qua? Or is it to require in every case the discovery of some
suitable qua?
If Jesus Christ could teach the doctors at the age of twelve, then he
was by that age good qua scholar, that is to say a good scholar. And that
was the capacity in which he taught the doctors. This he did qua man,
not qua God . . . So the Father did not teach the doctors at the age of
twelve – or so the orthodox will seek to rule. (But was it not qua God that
Jesus Christ was learned enough aged twelve to teach the doctors?) The
intention behind LL.III is clear enough. But nothing else is clear once
we insist on making the presence of ␾ qua f the general condition of
transfer of ␾ from a to some b that is the same f as a – because there is
no reason to believe one could say what qua would mean and how it
would need to be completed when it was adjoined to each of any or
every verb or predicate. On the other hand, it is unclear why transfer
should be refused to predicates that do not seem to need a qua.34

.        ( )

This book is not directly concerned with abstract entities but since the
arguments that have been used against examples (a)–(␭) have been
formal arguments, they ought, if they are good arguments, to work
equally well against purported R-examples of a mathematical or logical
character.
(␮) Suppose we have a relation R that holds only between a and b and
between c and d. Then the relation R in extension may be said to be the
set {〈a,b〉, 〈c,d〉 }, i.e. the set whose members are the ordered pairs 〈a,b〉
and 〈c,d〉. Now there are a number of different and equally allowable
definitions of an ordered pair. For example, 〈x,y〉 can be defined either
as {{x}, {x,y}} or as {{x), {∧,y}}. One might then say (from the extrasys-
tematic point of view) that the set S, namely {{{a},{ ∧, b}},{{c},{ ∧, d}}},
was the same relation as set S⬘, namely {{{a}, {a, b}}, {{c}, {c, d}}}. But
it is certainly not the same set as S⬘.
As usual the objection to accepting this apparent type-() example at
its face-value is Leibniz’s Law. If S ⫽S⬘ under the concept of relation
34
There is now an important essay on the Trinity and such matters by Richard Cartwright, to be
found in his Philosophical Essays (Cambridge, Mass., ).
 The absoluteness of sameness 
then whatever is true of S is true of S⬘ and vice versa. But S has {{a} { ∧,
b}} as a member and S⬘ does not.
Notice that any project of saving (␮) by some amendment or relativ-
ization of Leibniz’s Law plus some doctrine or other of categories is
manifestly hopeless. Suppose we say that we can only expect what is
true of S qua relation to be true of S⬘. Well, either relations are or they
are not sets.35 If they are sets then the plea is absurd, R is a set like S
and S⬘ and we still have the violation of Leibniz’s Law. If they are not
sets, however, then the true statement ‘set S is the same relation as S⬘’
must not be allowed to have the consequence that S or S⬘ (predicatively)
is a relation. We must only allow, when we are looking at this question
extra-systematically, that it represents a relation. But to block this conse-
quence is to withdraw (␮) as a genuine case of type (). If we want a
reduction of relations of degree n which strictly equates them with
something, then we shall have to isolate and utilize whatever it is that
all mathematically satisfactory definitions of ordered n-tuples have in
common.
If we do block the unwelcome consequences of (␮) (rather than vainly
try to settle which is ‘the right definition’ of ordered pair), then what sort
of ‘is’ do we have in (␮)? Presumably it is analogous to the ‘is’ of ‘Irving
is Hamlet’ or the ‘be’ of ‘that piece of sugar can be your queen [at chess]
while I glue the head of the queen back on’. It may be that the ‘identi-
fications’ associated with the various proposed reductions of arithmetic
to set-theory have to be similarly explained, in terms of realization or
representation.36

.      R ,             

    
It may be helpful to conclude with some general remarks about the
defects we have discovered in purported type-() and type-() instances
of R. Almost all of the readings of these that have given R the sem-
blance of support have depended on ignoring one of two things:
(i) the definite but limited ambiguity of certain uses of ‘is’ (ambiguity
35
‘Different things cannot be made to coincide by abstraction, and to regard them as the same is
simply a mistake. If abstracting from the difference between my house and my neighbour’s, I
were to regard both houses as mine, the defect of my abstraction would soon be made clear.’
(Frege, Grundgesetze der Arithmetik, §.)
36
For ‘theoretical identification’, see Putnam’s essay in Dimensions of Mind, ed. Sidney Hook (New
York, ), p. . For representation of arithmetic in set-theory, see Paul Benacerraf, ‘What
Numbers Couldn’t Be’, Philosophical Review,  (Jan. ).
 The absoluteness of sameness
for which we have given some collateral evidence, almost all of it depen-
dent, however, at closer or further remove on Leibniz’s Law), most par-
ticularly the ‘x is constituted of y’ reading of ‘x is y’;
(ii) the deceptive designation of certain referring expressions. (‘By
‘that Hermes’ do you mean ‘that statue of Hermes’ or ‘‘the matter of
that statue of Hermes’, or what exactly?’ Again, it is only in the presence
of Leibniz’s Law that such questions have any point.)
In the case of some putative examples of R, it is hard to say which is
the dominant confusion. Different people’s appreciation of certain cases
may differ. But prolonged experimentation has led me to believe that,
even if there is overlap among the R-candidates that they disqualify, no
one of these diagnoses will account for everything that is going on in the
full corpus of apparent examples. Neither diagnosis is dispensable.
The need we have found for the second diagnosis is particularly inter-
esting. At § of The Foundations of Arithmetic, Frege wrote (in J. L. Austin’s
translation, my italics):
While looking at one and the same external phenomenon [derselben äussern
Erscheinung] I can say with equal truth both ‘It is a copse’ and ‘It is five trees’,
or both ‘Here are four companies’ and ‘Here are  men’. Now what changes
here from one judgement to the other is neither any individual object, nor the
whole, the agglomeration of them, but rather my terminology. But that is itself
only a sign that one concept has been substituted for another.
That which is true and unshakeable here is the justly celebrated insight
of Frege for which he is preparing to argue in this passage – that
numbers attach to the concepts under which objects fall, and not to the
objects themselves. But even for a good thesis there will exist both good
arguments and bad arguments; and there is at least one bad argument
here if Frege really meant to suggest that, holding my gaze constant
upon one external phenomenon,37 I can subsume the very same some-
thing first under the concept copse then under the concept trees and, in
doing this, assign to the first concept the cardinal number one and to the
second concept the number five. Obviously a copse is trees. But the copse
itself is not identical with any tree or with any mereological aggregate of
trees. (The copse tolerates replacement of all its trees, for instance, but
neither the aggregate nor the class comprising the present trees in the
copse can survive such replacement.) There is no real prospect of this
‘this’ (the ‘dies’ in ‘dies ist eine Baumgruppe’ and ‘dies sind funf Bäume’)

37
Is the Erscheinung a sort of internal accusative of ansehen, or the whole visual field of the seeing,
or the object seen? – and, if it is the last, why suppose there is one and only one such object?
 The absoluteness of sameness 
having constant reference, even though the stuff (wood) that makes up
the this both makes up a copse and makes up five trees.38 What interests
Frege is that, according to which of these choices we make, we either
arrive at a concept with the number one or arrive at a concept with the
number five. What interests me, however, is that, if we are concerned
with the reference of ‘this’ or of any other designation whatever – con-
cerned that is about what object it designates – then we have to take care
to discover whether the ‘this’ stands for a copse or for a class whose
members are certain trees or an aggregate of certain trees. These are not
the same thing.

.      

  
Once controversy recedes and the dust settles, the result that we reached
by § of this chapter may be given again.
Given Leibniz’s Law
x ⫽y→ (␾x ↔ ␾y)
and given, by virtue of D,

x ⫽y if and only if (᭚f) 冢x ⫽f y冣,

we obtain the schema

(᭚f) 冢x ⫽f y冣 → (␾x ↔ ␾y).

So, if we choose any kind at all such that ⫽ , x will be ␾ just if y is ␾:
x y
f
(᭙g)
x⫽y
g 冢
→ ␾x↔ ␾y . 冣
Consider now the arbitrary individuals x and y, each falling under the
sort g, the individual x being the same g as the individual y. Then

38
See Wiggins (). When Socrates claims in the Parmenides that he is one man and many parts,
right and left parts, back and front parts, upper and lower parts, all different (enumerated at e,
cf. Philebus d), Plato seems to say that Socrates himself partakes both of one and of many. Of
course Frege could not say such a thing. And, if we credit Frege with following through consis-
tently the idea that ‘to use the symbol a to signify an object, we must have a criterion for decid-
ing in all cases whether b is the same as a’ (§), then we ought to deny that he is in any way
suggesting at § that the ‘dies’ has constant reference to objects. His sole interest is in the shift
of concept. (‘Neither any individual object’ is intended to dismiss the whole question of objects.)
 The absoluteness of sameness
consider an arbitrary individuative sort or kind h. Either x falls under h
or it doesn’t. If x does not belong to h, then by the principle T→ (F→
p), we have
x⫽y
→ (h(x) → ⫽ ).
x y
g h
On the other hand, if x does belong to h, then we have
x⫽y
& h(x).
g

But in that case ⫽ .39 So consider x’s predicate ( ⫽

x x x ...
). By Leibniz’s
h h

Law and ⫽ , the object y has this predicate too. So again we have
x y
g
x⫽y
→ (h(x) → ⫽ ).
x y
g h
Restoring quantifiers, we then have the following conclusion, entailing
the denial of R:

(᭙h)(᭙g) ( ⫽ → (h(x)→ ⫽ )).

x y x y
g h
In sum, let me say that it is on the twin claims of the Leibnizian abso-
luteness and the sortal dependency of identity that a theory of individ-
uation needs to be founded. It follows that, in order to obtain a theory
of individuation, we must try to arrive at a further and better account of
the kinds by which we individuate and sort particulars. This will be the
work of Chapters Two and Three.
39
If putative type-() cases of R appear to give trouble, see §, paragraph one, case ().
     

Outline of a theory of individuation

Upon those who step into the same rivers different and again differ-
ent waters flow. The waters scatter and gather, come together and
flow away, approach and depart.
(Heraclitus, Diels fragments , , text and translation after Kirk, Cosmic
Fragments, Cambridge, , pp. –.)

In the state of living creatures, their identity depends not on a mass

of the same particles, but on something else. For in them the vari-
ation of great parcels of matter alters not the identity: an oak
growing from a plant to a great tree and then lopped, is still the
same oak; and a colt grown up to a horse, sometimes fat, sometimes
lean, is all the while the same horse.
(John Locke, Essay Concerning Human Understanding , xxvii, .)

All constituents of living matter, whether functional or structural,

of simple or complex constitution, are in a steady state of rapid
flux.
(Rudolf Schoenheimer, The Dynamic State of Body Constituents, Cambridge,
Mass., .)

 .           D                ‘        ’
 
If identity were sortal-relative and sortal-relative just by reason of the
holding of R or the Relativity of Identity, then that would help to
support the Sortal Dependency of Individuation or D, as that was given
in the Preface and the first pages of Chapter One. For D is a thesis that
some champions of R have wanted to see as a kind of obverse of R. The
obverse they look for says that, on pain of indefiniteness, every identity
statement stands in radical need of an answer to the question same what?
But R is false, we have decided. It can lend no support to anything, least

 Outline of a theory of individuation
of all to the claim that there is some semantical indeterminacy in the
plain ‘a is b’ locution. (See Preamble, § ad fin.) Let us forget about this
obverse.
When D is clearly dissociated from R, that which remains is this:
D: a⫽ b if and only if there exists a sortal concept f such that
() a and b fall under f;
() to say that x falls under f or that x is an f is to say what x is (in the
sense Aristotle isolated);
() a is the same f as b, that is coincides with b under f in the manner of
coincidence required for members of f, hence congruently . . . [See
below §, paragraph .] [For ‘concept’, see Preamble §. See also
Preamble, § ad fin.]
Why dwell so particularly upon D if the three-place predicable ‘x is the
same f as y’ is going to be Leibnizian? Why is this attenuated version of
D worth recovering from the wreckage of R? It is worth recovering
because it registers the way in which, wherever there is a point at issue
in matters of identity, this depends in a systematic way upon what a and
b are. D unfolds and articulates the collateral and coeval ideas by the pos-
session of which we deploy the ideas of entity, identity and substance
upon that which confronts us in experience.
The import and importance of D may be brought out by considering
the position of a philosopher who proposes to dispense altogether with
it. So far as this philosopher is concerned, the point at issue in questions
of a and b’s identity or difference is indeed dependent on what a is and
b is, but only to an unremarkable extent. If a and b are to count as the
same, then a and b will have to agree in respect of all properties and rela-
tions, sortal properties themselves being among these properties.
Suppose I ask: Is a, the man sitting on the left at the back of the res-
taurant, the same person as b, the boy who won the drawing prize at the
school I was still a pupil at early in the year ? To answer this sort of
question is surprisingly straightforward in practice, however intricate a
business it would be to spell out the full justification of the method we
employ. Roughly, though, what organizes our actual method is the idea
of a particular kind of continuous path in space and time the man would
have had to have followed in order to end up here in the restaurant; and
the extraordinary unlikelihood (if the man himself were questioned and
these dispositions investigated) of certain sorts of memory-dispositions
existing in anyone or anything that had not pursued that path. Once we
have dispelled any doubt whether there is a path in space and time along
which that schoolboy might have been traced and we have concluded
 Outline of a theory of individuation 
that the human being who was that schoolboy coincides with the
person/human being at the back of the restaurant, this identity is
settled. And then we can say that, no matter what property ␾ is and no
matter whether the question of either entity’s instantiating ␾ figured in
our inquiry into the spatio-temporal paths of a and b, a has ␾ if and only
if b has ␾. The continuity or coincidence in question here is that which
is brought into consideration by what it is to be a human being.
I am not urging here that our actual method, focused upon the ques-
tion of this particular path of a human being, gives us a risky but prac-
tically indispensable short cut to establishing each of the indefinitely
many instances of the schema ␾a ↔ ␾b. Nor am I urging that it happens
there is no other way of establishing each instance (though that must be
a capital consideration). The contention is rather that to determine cor-
rectly the answer to our continuity question, the question about the
traceability of things through their life-histories, precisely is to settle it
that, no matter what property ␾ is, a has ␾ if and only if b has ␾. I am
also urging (as the epistemological reflection of a criteriological point)
that it is impossible even in theory to conceive of some way independent of
the prior discovery that a⫽b by which to establish that a and b have all
and only the same properties. For suppose one were to renounce all elu-
cidations of identity other than one given in terms of a’s and b’s com-
plete community of properties. Then how would one think about the
non-permanent properties enjoyed by an individual a identified with
respect to the past and the properties enjoyed by an individual b iden-
tified with respect to the present? One is only justified in pooling the non-
permanent properties of a and b if there is some other basis for the
identity of a and b than their having all their properties in common.
Finally, I would urge that, wherever the sortal concept under which a and
b coincide is the sortal concept for a kind of continuant and one can ask
‘what is it for an f to persist?’, it is the idea of a sequential history of a
thing’s doings and undergoings that comes into consideration. Without
this idea or the willingness to explore analogous points about items that
are not continuants, little sense will be made of very much that we actu-
ally do with the concept of identity. As Leibniz puts the point that I too
have wanted to insist upon, ‘By itself continuity no more constitutes sub-
stance than does multitude or number . . . Something is necessary to be
numbered, repeated and continued’ (Gerhardt , ). Indisputable to
our deployment of the ideas of entity, identity and substance is our
deployment of our countless ideas of their determinations.
Can some analogous point be made about entities that are not
 Outline of a theory of individuation
substances or material continuants? Suppose that, in making a tally of
the fs, Edward first uses the vocable ‘one’ then two more vocables. His
tally of the fs he gives as ‘three’. Suppose Kallias, in making a tally of
the gs deploys the vocable ‘hen’ then uses two more vocables of which
the second is ‘tria’. Then on this basis alone, we can say that whatever is
true of Edward’s tally number is true of Kallias’s tally number. (Or so
anyone will say who subscribes to the idea that natural numbers are
objects.) No more is needed. In practice, we don’t start with ␾x↔ ␾y. We
reach ␾x ↔ ␾y from identity.
Someone might accept this much and still insist that no philosophical
analysis or explanation can ever be achieved along the lines I intend, or
without dismantling the question of identity. In the case of continuants,
he will seek to dismantle it into a question of relating to some finer-
grained continuity relation holding between thing-moments.1 ‘Unless
something of this sort is attempted, the only thing that is left for the phi-
losophy of individuation to attempt will be as pointless as it is circular.’
The charge is worth considering, first as directed against the elucida-
tions of particular identities for which D provides the recipe, then as
directed against D itself. In the longer term, though, another answer will
appear, namely the peculiar exigency that the method of elucidation will
discover in the conditions for significant true application in any particu-
lar area of the sentence-form ‘a is the same as b’.

.          ,          

By the elucidation of identity advocated here, one who knows what he
is saying when he says that ‘a ⫽b’ ought to be in a position to explain,
where a and b are continuants, that a is a continuant which . . . and b is
a continuant which _ _ _, and then to expand each specification separ-
ately in a way sufficient to make it determinate (with the help of the
world) which continuant a is and which continuant b is. In the course of
this being achieved, as a necessary condition of the truth of the identity
claim, some common sort f will have to be found to which they each
belong. That is the point, we are saying, at which the primary question
of identity can come into focus. Is the f that a is any different from the f
that b is? Or do these fs coincide? What is there here to invite the charge
of vacuity or circularity?

11
For my view of attempts that have been made along these lines, see Chapter One, § and Chapter
Six, §. Against the claim that philosophy cannot dispense with such efforts, see Preamble, §.
 Outline of a theory of individuation 
The first charge might be that, if the author of the statement ‘a is b’
says in this way what each item is, then his amplification of the judg-
ment already involves the general idea of continuant-identity. Reply:
there is no harm at all in this provided two conditions are satisfied. First,
the notion of continuity that is invoked must not be the kind of conti-
nuity that we have affected to disbelieve in altogether. In so far as con-
tinuity is something general, the continuity in question must be the
determinable continuity of such and such sort-specific kind, of which man-coin-
cidence, donkey-coincidence, tree-coincidence can be the determinations.
Secondly, when the author of the statement ‘a is b’ says what the a is and
what the b is, he must be able to do this without prejudging the question
of the truth of the identity-claim. But this question can be left open. It
can be left open even if the answer is, if correct, then necessarily
correct.
The second alleged vacuity or circularity relates to the terms of D
itself. In order to trace a, one must find out what a is, or so I have said.
But what will count as knowing what a is in the Aristotelian sense? What
counts as a sortal concept for a continuant? Suppose the answer depends
in part on the idea of a principle of continuity for a. What then of the
determinable continuity (or coincidence) of such and such sort-specific kind? Has
the nature of this determinable been explained?
Here we are caught in a circle which would be vicious if we thought
we were bringing the concept of identity into being by means of other
ideas prior and better understood. The circle would be vicious if we
could not appeal to some extant a priori understanding of the identity
relation or we could not invoke a going practice which will effect the
sortal articulation of individual continuants as this f or that g or whatever.
Happily, though, we lack none of these things, unless we want to offer
up Leibniz’s Law to some numen of confusion. So, where the theoreti-
cal aim is only to enhance the understanding we already have of same-
ness, as this is ascribed or not ascribed within the class of continuants,
the doubt can be met. For D puts us in mind of that which Aristotle’s
isolation of the what is it? question also recalls to us, namely our practi-
cal mastery of the business of assigning things to kinds and our capac-
ity to get a grasp upon the nature of a thing. Compare Metaphysics
a 10–12. It is only by virtue of that capacity of ours that we can keep
track of a thing through change and distinguish between doing this cor-
rectly and making a mistake in doing it. As for ‘coincide’ (and its corre-
late ‘to be continuous with’), I hope that the meaning of this technical
term may be further secured by the following schema:
 Outline of a theory of individuation
where f is what x is (in Aristotle’s sense), x coincides with y under sortal concept
f if and only if the way in which x is f-related to y suffices for whatever is true
of x to be true of y and whatever is true of y to be true of x.

Of course, this schema, which is no more than a reminder of matters

already rehearsed, requires some collateral understanding of the what is
it question, of implication and of what it amounts to for the schema (␾x
↔ ␾y) to hold true no matter what extensional predicate replaces the
letter phi. But all this is in fact available to us; and the schema vindicates
both the fidelity and the unity of the conception of identity recapitulated
by our method of elucidation. We need not exaggerate what has been
accomplished, but now let us complete the third clause of D, as it is given
at the start of this chapter, by supplying the words ‘congruently, that is
in such a way as to entail the Leibnizian community of what is true of a
and true of b’.
This approach to identity will rightly remind the reader of divers doc-
trines of ‘criteria of identity’. Some philosophers have used ‘criterion’ to
mean ‘way of telling’ or ‘conceptually determined way of telling’.
Where they have meant that, however, there is a contrast with the pro-
posal advanced here. An answer to the what is it question does both less
and more than provide that which counts as evidence for or against an
identity. It does less because it may not suggest any immediate tests at all.
It does more because it provides that which organizes the tests or findings.
It organizes that which has been wanted by those who have asked for a
criterion of identity in the constitutive sense of ‘criterion’ in which Frege
asked for such a criterion.
My approach here is only a variant on Frege’s. If it differs in any
important way, that is only in virtue of a pessimism (corresponding to a
difference of subject matter) that I feel about the prospects for strict phil-
osophical analysis. It is true that some who have followed Frege
(Foundations of Arithmetic, §) in asking for the criterion of identity have
required from such a criterion some breakdown of identity into materi-
als that are singly independent of identity. But this was not necessary.
Nor is the Fregean demand inextricably linked with the puritanism of
those who have insistently required criteria of identity in fanatical
pursuit of ontological parsimony. Such philosophers have wanted to
refute claims that this, that, or the other kind of entity exists. That is not
my present concern, however. Nor was it Frege’s.
The real and abiding interest of the demand for the criterion of iden-
tity is surely this. Wherever we suppose that entities of kind f exist, we
 Outline of a theory of individuation 
are committed to ascribe some point to typical identity questions about
particular fs; and, in so far as identity is a puzzling or problematic rela-
tion, the first concern of the philosophy of any subject matter must be
to enhance our powers of finding the elucidation (whether or not we use
here the language of criteria) for its disputed identity questions. We need
to reinforce and enrich our powers of determining what it is we must
look to in order to answer these questions. (‘What does the matter turn
on?’, we must ask.) Such explications or glossings may be as discursive
and heterogeneous as we will from case to case, and as irreducibly prac-
tical (which implies that they will not be fully expressible in mere words),
provided that they all originate from something we can recognize as a
unitary understanding of identical and the same. Once it is understood
what organizes the tests or methods that we actually use to answer ques-
tions of identity, it will be possible to see the way backwards from any
tests we do use to the underlying point of these tests or methods; and
that is the trail back to what the tests are tests of, and why they are tests
of it. The special effectiveness of the ‘what is it’ question is that, in the
case of continuants, it refers us back to our constantly exercised idea of
the persistence and life-span of an entity; and it makes manifest the con-
nexion, which it was always evident that there must be, between the
entity’s identity through time and its persistence, between its persistence
and its existence, and between its existence and its being the kind of
thing it is.

.              

It is no confusion of epistemology with philosophical elucidation or
analysis to require, as we have, that ‘for all ␾, a has ␾ if and only if b has
␾’ should be the consequence and not the criterion of an identity judg-
ment’s being true, or so I conclude. Furthermore, as will emerge in the
course of the elaboration of the consequences of D, there really can
exist for a kind k some effective, albeit not verbally explicit, condition for
‘⫽’ whose satisfaction by any a and b belonging to k will have the amiable
property of securing that, no matter what property ␾ is, a has ␾ if and
only if b has ␾. But, in the meanwhile, it will be useful to enlarge for the
length of one short section upon a principle that has been in the back-
ground for some time now. That is the Identity of Indiscernibles. This is
the principle which results from taking the converse of the first level
schema that I called Leibniz’s Law, and transposing it to the second level
– the need for this transposition can scarcely be sufficiently emphasized
 Outline of a theory of individuation
– by prefixing to the antecedent a second-level quantifier with scope con-
fined to that antecedent.2 Thus the Identity of Indiscernibles is
[(᭙␾) (␾x ↔ ␾y)]↔ (x⫽y).
The shortcomings of this principle are intimately connected with the
failings of community of attributes and relations as an elucidation of
identity, and also with the failure of all reductive accounts of ‘⫽’.
Leibniz’s own view of the principle may be made plain by quotation:
There is no such thing as two individuals indiscernible from each other. An
ingenious gentleman of my acquaintance, discoursing with me, in the presence
of her Electoral Highness the Princess Sophia, in the garden of Herrenhausen;
thought he could find two leaves perfectly alike. The Princess defied him to do
it, and he ran all over the garden a long time to look for some; but it was to no
purpose. Two drops of water, or milk, viewed with a microscope, will appear
distinguishable from each other. (Correspondence with Clarke, p.  in the H. G.
Alexander edn, Manchester, .)
I have said that it is not possible for there to be two particulars that are similar
in all respects – for example two eggs – for it is necessary that some things can
be said about one of them that cannot be said about the other, else they could
be substituted for one another and there would be no reason why they were not
called one and the same. Moreover, if they have diverse predicates the concepts
too, in which these predicates are contained, will differ. (Pp. – in F. Schmidt,
Leibniz: Fragmente zur Logik, Berlin, ; cf. p. , L. Couturat, Opuscules et frag-
ments, Paris, .)
If we ask about the strength of the antecedent ‘for all ␾, ␾x ↔ ␾y’ in this
rendering of Leibniz’s reading of his principle, then it is clear that it is
as if he considers the principle to be protected from triviality by his
excluding from the range of this variable ‘␾’ not only predicates com-
pounded from ‘⫽’ itself but also such predicates or relations as ‘five miles
S.W. of Big Ben’. Indeed, he must think all predicates presupposing
place-, time- or thing-individuation are excluded.
It is always necessary that beside the difference of time and place there be inter-
nal principles of distinction . . . thus, although time and place [external relations,
that is] serve in distinguishing things, we do not easily distinguish them by them-
selves . . . The essence of identity and diversity consists . . . not in time and space.
(Nouveaux Essais , . and ; cf. , ., and letter to De Volder, Gerhardt , .)

12
At the same time providing for the concatenation of property variable and individual variables
to be understood as effecting predications. So, officially speaking, the Identity of Indiscernibles,
given in the maximally perspicuous form, says that x is identical with y if and only if, for all pure
properties ␾, x has ␾ if and only if y has ␾. For the question of which properties are pure, see
the ensuing paragraph.
 Outline of a theory of individuation 
Not everyone who has wanted to defend the Identity of Indiscernibles
has been prepared to follow Leibniz into his theories of relations,
space and time. But, unless one is prepared to follow argument where
argument leads and delimit the range of the property variable in
something approximating to Leibniz’s manner, the formulation ‘for all
␾, ␾x ↔ ␾y’ becomes a relatively trivial condition. Once predicables
involving ‘⫽’ or its congeners and its derivatives are included within
the range of the variable, the formula is neither an analytical explica-
tion nor even a serviceable elucidation of identity. For the formula
manifestly presupposes identity. It presupposes it in such a way that
any particular issue of identity will move round and round in a small
circle.
On the other hand, if one gives the principle of the Identity of
Indiscernibles the Leibnizian and more interesting interpretation, or if
(as I note that Leibniz probably did not) one envisages the principle
being used, not merely for the refutation of identity claims, but for their
positive or effective determination, then strange results follow.
Wittgenstein noted at Tractatus Logico-Philosophicus . that the Identity
of Indiscernibles rules out a possibility that involves no evident incoher-
ence at all, namely that of two objects’ having absolutely all their prop-
erties in common. Max Black and others have explored the difficulties
one will have in trying to deny the logical possibility of a universe con-
sisting only of two qualitatively indistinguishable spheres. (See Black
().) This matter will be resumed in Chapter Six, §.

 .           D                  :

  D ( ii)               D
At this point we must revisit a claim already entered, namely that, no
matter how disparate in respect of temporal and other identifications the
two designations are that flank the identity symbol in a true identity of
the form ‘a is the same f as b’, a phased or restricted sortal predicate can
always be supplanted salva veritate by a comprehensive unrestricted sortal
predicate, or (as I shall say) a substance predicate, to produce an equally
true affirmative identity-statement.3
13
There are many sortal predicates whose status as restricted sortal predicates is not syntactically
or even etymologically manifest, e.g. boy, foal, cub, gaffer, graus (Greek for an old woman); and
many whose precise status is not only structurally unmarked but has also to be discovered empir-
ically, e.g. tadpole, caterpillar. In the end, though, all phased sortal concepts are either latently
or manifestly restrictions of underlying more general sortal concepts. Boy is definable as human
being that is male and biologically immature, and so on. [footnote continues]
 Outline of a theory of individuation
This is the point at which it becomes necessary to distinguish between
two claims. I shall call them D(i) and D(ii). D(i) says that for all times t at
which a continuant a exists there is a g under which a falls at t; or more
perspicuously
D(i): (x)(t)[(x exists at t) →(᭚g) (g(x) at t)].
As will be apparent, all that this guarantees for each continuant is a suc-
cession of phased sortal concepts. It does not guarantee the availability of
one concept with application for all times. It does not guarantee the
thing the substance assumption envisages, namely that there will be
some preferred concept in a hierarchy of sortal concepts that a thing will
fall under throughout its existence. If the idea of continuity is to eluci-
date the truth-conditions of identity statements in the special way envis-
aged by D, then we need a stronger principle. It is this
D(ii): (x)(᭚g)(t)[(x exists at t) → (g(x) at t)].
Here, as before, ‘x’ ranges over three-dimensional continuants, ‘t’ ranges
over times and the letters ‘f ’ or ‘g’ range over individuative or articula-
tive concepts which ‘divide their reference’ and answer the question what
is x?
I hold that D(ii) is true. But let us begin with a putative counterexam-
ple to it. Consider the story of Lot and his wife, as given in Genesis,
chapter :
. . . . And it came to pass, when they had brought them forth abroad that [t]he
[Lord] said [unto Lot]: Escape for thy life; look not behind thee, neither stay
thou in all the plain . . .
. Then the Lord rained upon Sodom and upon Gomorrah brimstone and
fire from the Lord out of heaven . . .
. But his wife looked back from behind him [Lot] and she became a pillar of
salt.
footnote  (cont.)
Nothing in the argument to follow here is meant to depend upon a certain conceptual con-
servatism from which philosophical inquiry into substance and identity should be free, viz. the
supposition that simply by being a language-user one can tell for any given sortal, e.g. the sortal
tadpole or pupa, whether or not it is a substance-sortal or merely a phased sortal. Room must be
found for the empirical and surprising discovery that there is something which is first a tadpole
and then a frog – let us designate what goes through the whole cycle, what becomes this and then
turns into that, batrachos – or is first a pupa or chrysalis and then becomes a perfect insect. The
ensuing argument will leave room for that.
‘Another type of linguistic shortcoming [in biology] is illustrated by the persistence of our ten-
dency to identify organisms with adults . . . it is not just adults we classify when we classify organ-
isms . . . We can speak of the egg as the primordium of the future adult, but not of the future
organism because it already is the organism’: J. H. Woodger ‘On Biological Transformation’, in
Growth and Form: Essays for D’Arcy Thompson, ed. Le Gros Clark and Medawar (Oxford, ).
 Outline of a theory of individuation 
Encouraged by this narrative, one who doubts D(ii) will say that we can
understand the story through and through. He will urge that it corre-
sponds to a possible world, declaring that the champion of D(ii) must
make a difficult choice. Either that champion trivializes D(ii) altogether
by inventing arbitrarily a concept woman-pillar, which a thing satisfies by
being first a woman and then a pillar of salt; or else the champion must
postulate a real and substantial discontinuity (as judged by standards of
the answer given there to the ‘what is it?’ question) between Lot’s wife
and the pillar of salt.4
At the outset let it be agreed between me and the objector to D(ii) that,
if we avoid the trivializing alternative, then there is no substantial sortal
concept suitable to cover the identity between Lot’s wife at the beginning
of the story, before she looks back, and the forty-foot pillar of salt that is
still there to be encountered on the Jebel Usdum near the Dead Sea.
Indeed if we could invent sortal concepts simply at will, then the real
content of the assertion that something lasted till t and then ceased to
exist would be trivialized completely. For if one were unconstrained in
the invention of a substantial concept by which to represent that a thing
persisted, one would be equally unconstrained in the invention of a sub-
stantial concept by which to represent that it failed to persist. We could
have it either way, so to speak. We do not think of things like this,
however. As Aristotle perceived, there is an important difference that we
mark between a thing’s changing and its being replaced.5 If we took the
14
The counterexample to D(ii) might be fortified by suggesting that, in the story, the change in
question was continuous. The opponent of D(ii) may suggest that one may imagine that at one
stage in the process narrated there was something that was a candidate to count both as a human
body and as something crystallizing into salt. In defence of the claim that the story determines
a possible world, he will point out that, as told, the story does not even require violation of a
plausible principle that adds to D(i) but is weaker than D(ii). This might be called the principle of
sortal continuity:
If an entity x survives from t to t’, then there is some articulation of the time between t and
t’ into periods P1, P2, . . . Pn such that, for every period Pi, (᭚f) (f(x) throughout Pi) & (f(x)
throughout the adjacent period Pi±1).
This principle marks adequately the necessity (he will say) to distinguish between the persistence
of a thing through change and its replacement by some quite other thing. The story of Lot’s wife
respects this principle, the critic will insist.
For further discussion of the example of Lot’s wife, see Wiggins (a).
15
Cf. Aristotle: ‘Unqualified coming to be and passing away takes place when something as a whole
changes from this to that. Some philosophers hold that all such change [coming to be and passing
away] is mere alteration. But there is a difference. For in that which underlies change there are
two factors, one relating to the logos of the subject [i.e. what it is], the other relating to the matter
that is involved. When the change affects both of these, coming-to-be or passing-away will occur.
But when the change is not in the logos but only in the qualities [i.e. in respect of categories other
than substance], i.e. when the change is a change in accidents, there will be alteration’ (De
Generatione et Corruptione, a21 ff.). Aristotle’s is a stronger claim (given Leibniz’s Law and the
 Outline of a theory of individuation
Lot story seriously, if we treated it as if it were a historical narrative,
then, enamoured as we are of the change/replacement and true/false
distinctions, we should need to interrogate the narrative very closely, and
press all sorts of questions about what sort of a being Lot’s wife really
was. We should need to think of that question as one to which an answer
could be discovered ‘in spite of ourselves’. But we do not do so. It is
simply a story. It is perfectly notorious that not every story corresponds
to a possible world. (Chapter Three, §, bears on this, as does Chapter
four.)
If we do not want D(ii) to be trivially true, then we need to begin by
taking seriously the sortal predications we actually use. (That is how I
arrive so swiftly at the denial of identity between Lot’s wife and the pillar
of salt.) But even if we follow this counsel, some may ask: what forces us
to think that the mass of as yet uninvented sortal predicates which might
be truly applied to a woman such as Lot’s wife have to cohere together
in the kind of structure that is needed to verify D(ii)?
Well, remembering that all the sortal predicates that apply to a thing
at all can be used to cover identity claims concerning it, let us now deploy
our previous finding that R is false. If R is false, then all the sortal con-
cepts that apply to a thing at all must be concordant with respect to such
answers as they furnish for identity questions relating to the thing. If, as
I suppose, the best explanation and characterization of this concordance
is that all true sortal attributions to a given thing are restrictions of one
substance-concept (or are restrictions of a set of substance concepts that
are individuatively concordant), then that will make sense of other facts
such as this: in practice, something or other almost always lends a high
(even indefinitely high) degree of determinacy to questions we ask about
whether, at a given time, an object a still exists. Our capacity for massive
agreement about this is much more remarkable than our occasional dis-
agreement.6 The question I urge upon anyone reluctant to accept D(ii)
footnote  (cont.)
arguments that are shortly to be presented for D(ii)) than the principle of Sortal Continuity. But
the power of both is attested by the lengths to which poets and story tellers will go, when con-
cerned to pass off unqualified passing away as persistence-through-alteration, to create the
however mendacious impression that the conditions for alteration are scrupulously respected.
The trick is to describe the change as happening very gradually. See particularly Ovid,
Metamorphoses. One must add that the metamorphoses of Proteus –– see, e.g., Homer, Odyssey ,
– – represent a special artistic problem, corresponding to a conceptual impossibility of
really imposing dimensions. See Chapter Three, §.
16
If there are any disagreements, that will suggest that sortal predications can be contestable. For the
idea that it can be seriously contestable, in a way that can stir moral disquiet, what this or that pro-
visionally identified thing is, see Chapter Seven, especially § following. The concordance require-
ment, so far from being inconsistent with any of this, helps to bring out what may be at stake.
 Outline of a theory of individuation 
is this: what could better or more plausibly ensure that, if an individual
appears to fall under several sortal concepts, then these concepts agree
in the identity and persistence conditions that they ascribe to it?7
How could an entity escape this argument (such as it is) for D(ii)? It
could escape only by lacking a principle of identity and persistence
made fully determinate in the manner we have envisaged by a substance-
concept. But it would be no great defect of the argument for D(ii) if, at
this point, we rested it perfectly explicitly upon such an assumption. The
assumption says that in the ordinary business of individuation, questions
about the identity and persistence of the substances that we have knowl-
edge of possess definite answers, answers that are not normally undis-
coverable in principle. The same assumption says that, within the
ontology of changeable Aristotelian continuants to which so many of
these familiar things belong, we find ourselves committed to think that
many things have definitely ceased to exist.8
17
Nicholas Griffin (see Relative Identity, op. cit., p. xi and p.  note) has complained about this argu-
ment that it assumes that no two ultimate sortals can overlap (overlap, that is, without one’s being
a proper part of the other). But the argument does not assume that. It implies it. The picture the
argument in the text motivates is this. If we have a structure comprising only sortal concepts
(concepts that answer the what is it? question and autonomously assign principles of individua-
tion to objects that instantiate them) and these concepts stand in relations to one another such
as that of having members in common, one concept’s restricting the other, identity etc., and if
the structure is not closed to further or ‘missing’ concepts or individuals, then, where concepts
C() and C() have any members in common, there will be three cases:
The first case is where neither C() nor C() is a restriction of any other concept and each is
an ultimate sortal. Here, if the concepts have common members, then, because they will cover
identities relating to these common members and R is false, C() must be identical with C() or
extensionally equivalent to it.
The second case is the case where one of the two sortal concepts C() and C() is ultimate and
the other is not. In this case, if there are common members, then the non-ultimate one restricts
the ultimate one.
The third case is the one where neither C() nor C() is ultimate. Here, if C() and C() have
common members, there may be overlap without either concept subsuming the other. This is
the case of cross-classification. But then some sortal concept that is ultimate will subsume both
C() and C().
It is my conjecture and hope that this picture of things, founded in the nature of sortal con-
cepts and the absoluteness of identity, concedes everything that deserves to be conceded to the
over-stringent demand that sortal concepts should form a hierarchy. (On which see Woodger
(), p. ; Richmond Thomason, ‘Species, Determinates and Natural Kinds’, Nous,  (),
pp. –, who was criticized by Howard Kahane, ‘Thomason on Natural Kinds’, Nous, 
(), especially p. . Thomason’s proposal deserves to be dissociated from the idea that the
hierarchy must be arrived at by a process of dichotomous division. Cf. Aristotle, De Partibus
Animalium, aff and Plato, Politicus, d.) At the same time, the picture I am suggesting does
not, I think, rule out the sorts of cross-classifications that seem indispensable to ethology, linguis-
tics, etc. On cross-classification, see Chomsky, Aspects of the Theory of Syntax (Cambridge, Mass.,
), pp. –.
18
This implies that there are no principles of conservation that apply to continuants as such. But
to deny that there are such principles is not of course to deny that the matter of such continuants
 Outline of a theory of individuation
If we defend D(ii) in these terms, then a question remains whether an
entity could escape from the space where these assumptions hold good.
In Chapters Five and Six an argument will emerge for the claim that
none can escape. But that argument (and a limitation upon it that will
emerge in Chapter Seven, § following) must wait for its moment.
Rather than pursue undecidable fantasies about what it would be like
if the presuppositions of our practice were groundless and unjustifiable
or search for every conceivable pretext not to take them seriously, I urge
the reader first to observe and marvel at the near unanimity that has
been reached (and is almost effortlessly extended to a whole range of
new cases) about what constitutes the ceasing to exist of an everyday
continuant of this or that kind.

.             

The story of Lot’s wife has the effect of facing us with another question
to which sortal concepts are no less relevant than they are to identity.
What is it to cease to exist? An orthodox account says: to cease to have
any properties. But this is not necessary. Of Aristotle, the author of
Nicomachean Ethics, I believe that he was bald. So this Greek, though long
since dead and non-existent, now has the property of being believed by
someone to have been bald. Similarly Heraclitus is still misunderstood
and Frege has become famous since his death. Are these not properties?
Suppose we devised a stricter notion of property to exclude these and
we said that ceasing to exist was a matter of ceasing to have any strict
properties. Would that count as a full answer to the original question? I
do not think so. The question about tensed existence arose from the
desire to know what it was for an individual to be no longer available to have
such and such properties. Maybe it will cast light on that question,
however, if we respect the distinction between predications in the cate-
footnote  (cont.)
is subject to such principles as physics and chemistry can demonstrate. Nor is it to deny that con-
tinuants may share this matter with the other entities, if any, for which there do exist principles
of conservation.
We pass close here to the problems of the continuance of stuff. There are profound and diffi-
cult questions to be asked about this. I do not deny that these questions interact with some of the
most difficult questions in the philosophy of chemistry and physics. It would be a wonderful
thing, if life, will, and the mixed economy lasted so long, to embark upon them. But those ques-
tions are not these questions. And these ones are easier.
For a careful and searching exploration of the attempt, by what he calls the Relational Theory,
to circumvent the claim in the text to which the present note is affixed, see Burge (), see espe-
cially p. ) – a gratifying vindication also of the constitutive ‘is’ and of the constitutive approach
to things and their matter that was advocated in Chapter One.
 Outline of a theory of individuation 
gory of substance and predications in the category of quality or relation,
and we follow in an older tradition to say that an individual a’s having
ceased to exist at t is a matter of there being nothing identical with a that
belongs to the extension at t of the ultimate or substance kind that is a’s
kind.9
A little more still needs to be said about the meaning of D(ii). There
are countless predicates in English that have the appearance of sortal
predicates but are purely generic (animal, machine, artefact) or are pure
determinables for sortal determination (space-occupier, entity, substance).
Like a substance-concept for x, these apply to x at every moment of x’s
life-span. But to adduce any of these as verifying the existential claim
that we find in D(ii), ‘there is a g . . . (etc),’ would sell the principle short.
The reason why D(ii) is true has to do with the special role of the
substance-concept man, horse, willow tree . . . in marking simultaneously
what a thing is, what matters turn on with regard to its persistence and
what matters turn on with regard to identity claims relating to it. It will
be the work of subsequent D principles (and of later chapters) to
suggest the further conditions that will need to be placed on the sort-
variable f if D is to possess itself of the whole strength that it is
intended to enjoy.

 .       D      
It will be inquired in the next Chapter how any predicate at all could
measure up to the responsibilities that go with the title of standing for a
substance-concept. But, in advance of Chapter Three, let us assess the
magnitude of the office that it now appears such a concept must dis-
charge and try to reckon up in some further principles, without any
regard to economy or completeness or independence of statements, the
duties that have already been laid upon it.
19
Note that ‘a exists at t’ is not on the theory we have advanced here an incomplete assertion. It
says (᭚f) (f is a substantial concept for a and a is in the extension at t of f). This is a complete
assertion, and it well reflects the apparent completeness of actual denials and assertions of tem-
poral existence. Someone bent on rejecting D(ii) might advance a theory to be sharply distin-
guished from mine. He might claim that ‘a exists’ is an incomplete assertion and propose a
doctrine of relative existence. By this theory of existence, the distinction we have drawn between
phased and substantial sortal concepts is simply misconceived. For by this new theory, whatever
concept f is, a thing may always cease to exist as an f but continue to exist as a g. About this pro-
posal I should remark that it may well be the only consistent option for someone bent on reject-
ing D(ii); but it would constitute a daring reconstruction of the principles of the continuant
ontology and of the whole way we at present think and speak. As I have said already, D(ii) was
intended to reflect the ontology implicit in our present way of thinking and of puzzling ourselves
about everyday identity questions. According to that, many things do definitely stop existing.
 Outline of a theory of individuation
First, in summation of what has been said already, we have
D(iii): a is identical with b if and only if there is some concept f such
that () f is a substance-concept under which an object that belongs
to f can be singled out, traced and distinguished from other f enti-
ties and distinguished from other entities; () a coincides under f with
b; () ]coincides under f § stands for a congruence relation: i.e. all
pairs 〈x,y〉 that are members of the relation satisfy the Leibnizian
schema ␾x if and only if ␾y.
D(iv): f is a substance-concept only if the grasp of f determines (with
or without the help of further empirical information about the class
of fs) what can and cannot befall any x in the extension of f, and what
changes x tolerates without there ceasing to exist such a thing as x;
moreover f is only a substance-concept if the grasp of f determines
(with or without the help of further empirical information about the
class of fs) the relative importance or unimportance to the survival of
x of various classes of changes befalling specimens of f (e.g. how close
they may bring x to actual extinction).
Clause () of D(iii) answers to the requirement that community of prop-
erties should be not the basis but the consequence of the satisfaction of
an acceptable criterion of identity. It may seem either emptily unattain-
able – in the case where it is not allowed that anything at all could guar-
antee the satisfaction of the Leibnizian schema – or magic of some sort.
But it is not magic. D(iii) simply recapitulates the principle on which in
our actual practice, sustained as that is by the idea of the nature of an
object and (where applicable) its life-span, predicates are applied to a and
b or (as the case may be) withheld from a or b in respect of any instant
or period belonging to the life history of this (these) persisting continu-
ant(s). This is not to say, however, that it will always be possible to write
down an explicit f-involving condition that suffices for the identity of a
with b. The thing we can write down will only be some verbal expression
of the larger practical understanding enjoyed by one who can convert
his grasp of the concept of an f into some capacity to determine what
is at issue in any particular case where it is asked, with respect to an a or
a b which is an f, whether a is the same as b.
No doubt a sceptic may say that our practice has no business to be sus-
tained by such ideas as that of an object’s life-span or nature, etc.,
however filled out in particular cases. He will say that any practices
founded in them must be precarious in the extreme. But here, as usual,
our chief concern must be to describe from within and to sustain from
within the ideas that the sceptic attacks, and then to enumerate such
 Outline of a theory of individuation 
sources of support as they do have. It is not the purpose to render them
immune to all doubt or difficulty – still less to minimize the considerable
commitment that someone incurs when he says what something is. Such
commitment is indeed precarious, but no more precarious than anything
else that is worth the candle. It should be added that it will be rare for
any sortal concept that qualifies by D(iii) as a substance predicate (or as
determining a principle of tracing) to come up to the standards a posi-
tivist would propose for observationality or direct experiential confirmability.10
I do not think this prevents a predicate that qualifies by the standard set
by D, D(i), D(ii) and D(iii) from furnishing a principle of individuation
for members of its extension. Nor does it prevent D(iii) from determin-
ing and organizing any recognizably empirical tests that may be elab-
orated for whether a or b is an f thing, for whether a is the same f as b,
and for a or b’s continued existence.11
Connoisseurs of the literature of identity and individuation will have
noticed that little room has been left in these formulations of D princi-
ples to accommodate the distinction of which so much has sometimes
been made between synchronic principles of boundary drawing (some-
times called criteria of distinctness) and diachronic principles of reiden-
tification (criteria of ‘transtemporal’ identity). This is not an oversight.12
It is my contention that, as principles that purport to be of one or the
10
Across the full range of their received applications, not even ‘white’ or ‘flexible’ or ‘soluble’ do.
11
The same point, or almost the same point, can be seen from a completely different angle.
Consider the biologically informed description of the Protozoon Paramecium (Class Ciliata):
Paramecium reproduces by both asexual and sexual means . . . Sexual reproduction is by
means of a process called conjugation. In Paramecium not all individuals are alike: there are
two different classes which are known as mating types . . . Conjugation begins with a pairing
and then partial fusion of the anterior ends of two individuals of opposite types. The micro-
nucleus in each undergoes meiosis and then there is an interchange of nuclei between the
two individuals. In each individual a new micronucleus is formed by the fusion of one of its
own micronuclei with one from the other individual. This is essentially fertilization. The
macronuclei degenerate. ( J. A. Moore, Principles of Zoology (New York, ), p. .)
Contrast with the above the description that might be given by an observer who was attentive
but not biologically informed. Such a description would leave room for the possibility that the
individual paramecia a, b survive the process described. It is principles D(iii) and D(iv) that artic-
ulate the rationale for insisting, on the basis of Moore’s description, that they cannot survive it.
With D(iv), it should become finally clear why, in the sense of ‘coincide’ that I employ in order
to elucidate identity, a thing does not coincide with its matter. Coinciding is not simply being in
the same place at the same time. Just as there is no such thing as mere continuity so there is no
such thing as coinciding otherwise than under a concept. A house, for instance, does not coin-
cide with its bricks and mortar. For in ‘the house is bricks and mortar’ the is is constitutive. Here
we do not have identity. Cf. Chapter One, §.
12
See my symposium paper ‘The Individuation of Things and Places’, section , Proceedings of the
Aristotelian Society, Supplementary Volume  (); see M. J. Woods’s reply, ibid.; see Woods ().
See also my reply to Harold Noonan in Lovibond and Williams ().
 Outline of a theory of individuation
other sort are filled out in the direction of adequacy, the distinction dis-
appears. It is of course a severer test of someone’s grasp of the concept
of cat to ask him to determine whether this cat is the same cat as the cat
that licked the saucer clean on Monday of last week than it is to ask him
to count the cats in the parlour now at . pm, making sure that none
is counted twice. It does not follow, though, that the thing tested by the
first assignment can be seen as built up from the cat-lore needed to meet
the second assignment plus some altogether different kind of lore or
some different sort of grasp of what cats are.

In so far as the sense and semantics of substance words come into ques-
tion here, or our grasp of this sense, these things must wait for Chapter
Three, §§–, as must the justification and full explication of the D-
principle I shall place next:
D(v): f is a substance-concept only if f determines either a principle
of activity, a principle of functioning or a principle of operation for
members of its extension.
The Leibnizian echo of ‘activity’ is deliberate – as will emerge when the
activity of natural things is distinguished from the fulfilment of function that
is characteristic of organs and artefacts.13 In Chapter Three, where this
distinction will be made, a key role will be played by a principle that is a
direct consequence of D(iii):
D(vi): If f is a substance-concept for a, then coincidence under f is fully
determinate enough to exclude this situation: a is traced under f
and counts as coinciding under f with b, a is traced under f and
counts as coinciding under f with c, yet b does not coincide under f
with c.
Consider what happens when an amoeba divides exactly in half and
becomes two amoebas. We are committed by D(vi) and the other doc-
trines of this chapter to find that neither of the resultant amoebas
qualifies as coinciding under the concept amoeba with the original
amoeba. Chapter Three will show how the peculiar facts about a
natural kind, discovered a posteriori, could in any given case be enlisted
to support this ruling. But two things are immediately obvious. The
first is that, since not both amoebas can be identical with the original,
and since neither has a better claim than the other, no sensible theory
13
In advance of Chapter Three, however, let me commend Hacking (). The Leibnizian echo
made by ‘activity’ is deliberate but, outside the monadological framework, it does not have to
import anything very different from ‘way of being, acting and reacting’ – something a stone
might have.
 Outline of a theory of individuation 
will want to count either amoeba as identical with the original one.14
The second obvious point is that judgments of coincidence and iden-
tity must not be withheld ad hoc simply because transitivity is threat-
ened. There must be something that one who grasped the a posteriori
concept amoeba could see (not necessarily independently of general
knowledge of the kind) was lacking in any idea of amoeba-coinci-
dence that was ready or poised to count amoeba division as a way of
persisting for amoebas.
Let us not make the wrong denial here. The process of division is
of course the process by which Amoeba proteus, for example, is per-
petuated. Particular amoebas perpetuate it. But Amoeba proteus is
not a substance or a particular. It is something universal (a kind or a
clone of organisms or cells all produced asexually from one ancestor
or stock from which they are genetically indistinguishable). The
thought we were concerned with in relation to D(vi) concerned not
Amoeba proteus but the particulars that instantiate and perpetuate it.
So the second point that we were concerned with in the paragraph
preceding this may be beneficially expanded to run like this: if
amoebas (amoebas plural, the particulars) are to be objects of refer-
ence, then there must be something or other that one who understood
amoebas and grasped the a priori constraints upon the determinable
continuant-identity (is the same something or other as) could in practice
find to be independently wrong with the suppositions that generate
the contradiction that D(vi) proscribes. Moreover, there is something
wrong, as would be apparent to anyone who contrived to observe and
understand this general process of division.
One other thing deserves remark. Principles such as D(vi) not only
enforce our formal understanding of the identity relation. As we see
from the work that they do here, they bear witness to our grasp of the
distinction between particular and universal.

14
Contrast Prior, ‘Time, Existence and Identity’, P.A.S.,  (–). The view I have put in the
text conforms with the only thorough attempt that is known to me to work out a logic of divi-
sion and fusion, namely that of Woodger (), p. . One amoeba becomes two amoebas. But
here, in a manner that will remind the reader of a thesis frequently defended in Chapter One,
the verb ‘becomes’ receives an analysis making it correspond to the ordinary ‘becomes’ in the
same way in which constitutive ‘is’ corresponds to the ordinary ‘is’ of predication and identity.
The matter of the original amoeba – the ‘it’ – is the fusion, or the matter, of the two new ones
taken together. There is matter such that first a was constituted of it, and then b and c were con-
stituted of it.
Further light is shed on the amoeba question by the ‘Only a and b’ rule of Chapter Three, §.
We shall call this D(x), a principle that will come into its own in Chapter Seven.
 Outline of a theory of individuation
.               ;    
       
From the falsity of R we have
D(vii): there are no essentially disjunctive substance-concepts (f or g)
coincidence under which might allow a to be the same (f or g) as b
and allow a to be the same (f or g) as some c that was distinct under
every covering concept from b.
Some disjunctive sortal predicates are innocuous in this regard, e.g.
when one disjunct f is subordinate to a higher sortal. (For example,
animal or mouse is innocuous. a cannot be the same mouse as b and a differ-
ent animal from b. The sortal predicate reduces here to animal.)
Disjunctive sortal predicates will be innocuous when the corresponding
conjunctive sortal is a satisfiable concept (e.g. animal and mouse, which
reduces in the opposite direction to mouse). But for that case disjunctive
sortals seem to be as superfluous as they are innocuous.15 The argument
for D(vii) is not, of course, an argument against the essentially disjunc-
tive (f or g) being a concept. It is an argument against its being a sortal
concept.
D(viii): If f is a substance-concept for a then, however indefinitely
and unforeseeably the chain of a’s f coincidents a, a’, a’, a’’’ . . .
extends, whatever is truly or falsely applicable to one member of the
chain must be truly or falsely applicable to every member of the chain
whatever. But then all fs must belong to one category.
This follows a fortiori from Leibniz’s Law, and is trivial. But it is worth
deducing for the sake of the things it brings out against certain allegedly
possible metamorphoses, for instance that of Proteus into fire at Homer,
Odyssey , –. These difficulties will present themselves again in
Chapter Three.

I now come to the doubt. It would be all of a piece with the speculative
tradition to which these speculations belong to add
D(ix?): f is a sortal concept only if it furnishes the materials for the
question, and for the determination of a definite and finite answer to
the question: ‘How many fs are there in region r at time t?’
This is more or less the claim C of the first section of Chapter One.
Many or most of the concepts that it is the aim of this and the next

15
It is sometimes objected that we must have disjunctive sortals for certain kinds of counting oper-
ation. But ‘There are  women or shadows here’ means ‘The number of women plus the number
of shadows⫽ ’. We do not need disjunctive sortal concepts to find our way here.
 Outline of a theory of individuation 
chapter to characterize do satisfy this principle. For all I know, everything
traditionally accounted a concept of material substances satisfies it. But
it is one thing to be able to say how many fs there are in some particu-
lar context and another thing to provide the conceptual prerequisite for
the thing that D(ix) seems to require: namely a general method of enu-
merating fs. Subject to any doubts that Hobbes’s puzzle of Theseus’ ship
may arouse concerning artefact kinds (see Chapter Three, §), the
concept crown gives a satisfactory way of answering identity-questions
for crowns. But there is no universally applicable definite way of count-
ing crowns.16 The Pope’s crown is made of crowns. There is no definite
answer, when the Pope is wearing his crown, to the question ‘how many
crowns does he have on his head?’ Something similar will apply in the
other cases of homeomery.
Given that someone knows how to count and can also be relied upon
to count correctly in some particular situation, then to see if they give
the right answer there to the question ‘how many fs?’ shows whether they
can pick out fs and isolate them correctly from their background and
from one another. That is what carries C of Chapter One, § and the
putative principle D(ix) so close to the truth. But the objection to taking
these general principles as necessary for a concept’s being individuative
is that someone might be able to tell reliably whether this arbitrary
f-thing was the same or not the same as some independently identified
f-thing, even though there were far too many different ways of articulat-
ing f-things for there to be any definite answer to the question how many
fs there were to be found in this or that given place. To ask for the latter
as well is to ask for more than is needed to ensure that f is a sortal
concept.
Frege wrote in Foundations of Arithmetic, §, ‘only a concept which (a)
delimits what falls under it in a definite way [der das unter ihn Fallende
bestimmt abgrenzt], and (b) which does not permit any arbitrary division of it
into parts [und kein beliebige Zertheilung gestattet] can be a unit relative
to a finite number’. His condition (a) could be naturally developed to
16
Cf. wave, volume of fluid, worm, garden, crystal, piece of string, word-token, machine. This is a rag-bag.
These are all kinds of individual decomposable into matter, though there is a good sense of ‘sub-
stance’ in which not all are kinds of substance. So far as concerns any general doubt or the
general logic of identity and individuation, it must be worth taking note of homeomerous items
in other categories, such as colours and their shades, quantities, wanderings, wonderings, read-
ings, insults.
With the exception of this minor disagreement about countability taken as a necessary condi-
tion of a concept’s being a sortal concept, there is here no disagreement with Strawson’s account
of these matters in Individuals, or with the written and oral tradition concerning these predicates.
The intention has not been to cut loose of that tradition, but to correct and enlarge it.
 Outline of a theory of individuation
cover precisely that which we have intended by our conditions upon
being a sortal concept. But there is no reason to think that the second
condition, italicized here, is for Frege’s special purpose otiose. His par-
ticular interest in this passage concerns counting, and the second condi-
tion, condition (b), seems to represent a precise further requirement that
has to be satisfied for there to be a universally applicable distinction
between right and wrong answers to the special question ‘how many?’
Like D(ix), this requirement goes appreciably beyond any requirements
it would be sensible to place upon the generality of sortal concepts that
are enjoined upon us by our concerns with the what is it? question.
D(ix) fails. But D(ix) is not the last candidate to consider. Chapter
Three, §, will introduce another, namely D(x), the Only a and b rule. In
due course, anyone who follows through the rationale for articulating
such principles will surely want to frame others.
     

Sortal concepts: and the characteristic activity or function

or purpose of things falling under them

The barley drink disintegrates and loses its nature unless it is con-
stantly stirred.
(Heraclitus, Diels fragment .)

Nothing is permanent in a substance except the law itself which

determines the continuous succession of its states and accords
within the individual substance with the laws of nature that govern
the whole world.
(Leibniz, Gerhardt , .)

That which is a whole and has a certain shape and form is one in a
still higher degree; and especially if a thing is of this sort by nature,
and not by force like things which are unified by glue or nails or by
being tied together, i.e. if it has in itself the cause of its continuity.
(Aristotle, Metaphysics a22–5, as translated by W. D. Ross.)

 .              

Among the best candidates to play the roles of sortal and substantial
predicates, as these roles were described in Chapter Two, are natural
kind words. We now reach the question whether even these will fill the
bill. Given a kind k associated with the sortal concept c introduced by
some sortal predicate or substantive, how can c determine what is at
issue, what the matter turns on, when it is asserted or denied that x
and y, which instantiate k, are one and the same? D(iii) as glossed by
D(v) requires something to be latent or implicit within the sense of a
natural kind identification that will determine, for x, for y, for typical
other members of the kind, some further specifiable principle of
activity. It seems that here the natural kind concept c must make some
constitutive link between the relevant questions of identity and the
empirically ascertainable causal and dispositional properties of

 Sortal concepts
members of the kind, as we find them in the actual world. But how is
this link secured?
It is certain that, if we are looking to the sortal concept to play this role
in matters of identity and individuation, then there is no help to be got with
the problem from the explications of kind words offered in the tradition of
nominal essence as traditionally conceived, or from accounts that seek to
specify the sense of ‘sun’ or ‘horse’ or ‘tree’ by a description of such things
in terms of manifest properties and relations or in terms of appearances.
It is now a very familiar point that these are unsatisfactory, even when
taken on their own terms. If we read proposals of this sort literally or in
the way in which they invite us to understand them, they always fail either
of necessity or of sufficiency for membership in the intended kind. Nor is
this their only failure. By the very form in which nominal essence accounts
are given, it is impossible that they should reconstruct our actual under-
standing of natural kind substantives. They leave unexplained not only the
way in which our conceptions of a given kind of thing can evolve while still
being conceptions of the very same natural kind, but also the non-arbi-
trariness of that evolution. A more satisfying account will emphasize the
contribution that the world itself makes to those conceptions and paint
into the picture the instinctual, cultural and technological preconditions of
our catching hold of one another’s public meaning when we make refer-
ence to kinds. These last are an indispensable part of the background for
our collective interest in natural kinds and the advancement of our shared
understanding of them.1
11
‘To be sure – there has to be a great deal of stage-setting before one can read a [dictionary] def-
inition and guess how a word is used. But in debunking . . . the fact that something so simple . . .
can convey the use of a word [certain philosophers] forget to be impressed by it . . . there is a
great deal of stage-setting, but it is rarely stage-setting specifically designed to enable one to learn
the use of this word.’ See Putnam (), p. .
In subsequent writings Putnam has stressed how any good account of these matters must incor-
porate not only stage-setting but also the division of labour and the dependence of non-experts
on experts in the identification of natural and other kinds:
The [‘probability theory of meaning’] cannot as it stands incorporate either the linguistic divi-
sion of labour (the fact that confirmation procedures for being gold, or being aluminium, or being
an elm tree, or being N.N. are not the property of every speaker . . . speakers defer to experts for
the fixing of reference in a huge number of cases), or the contribution of the environment (the
fact that the extension of a term sometimes fixes its meaning and not vice versa). In my view, the
criteria used by experts to tell whether or not something is gold are not ‘part of the meaning’
of gold (e.g. the word doesn’t change its meaning in the language if the experts shift to a differ-
ent set of tests for the same metal), yet they are part of a mechanism for fixing the extension
of gold. (‘Reference and Understanding’, the Collected Papers, Vol.  (Cambridge ).)
Compare Leibniz, Nouveaux Essais, .., ... See also .., which suggestively anticipates
and answers some of the difficulties that have been found in the deictic–nomological account,
 Sortal concepts 
Hilary Putnam’s counter-proposal is now well known.2 This proposal
(or reconstruction, if you prefer to call it that) holds that, for the expli-
cation of the sense of a substantive or a sortal predicate, the thing that
needs to be laid down is that x is an f (horse, cypress tree, orange, caddis-
fly . . .) if and only if, given good exemplars of the kind that is in question,
the best theoretical description that emerged from collective inquiries
into the kind would group x alongside these exemplars. (See also
Preamble §, and my ().) An object x is an f just if x is similar in the
e.g. by D. H. Mellor in ‘Natural Kinds’, British Journal of the Philosophy of Science,  (). Take
an element and one of its isotopes. Which one is the natural kind? Leibniz would surely say that
it was a mistake to insist on any general decision. It may be implicit in our however provisional
classificatory practice that the stuff we are aiming for is the most specific kind possible. But
Leibniz also leaves room to say that it is implicit in our use that what we are aiming at is a kind
of stuff that plays a certain role in a particular kind of explanation. In the case of isotopes (one
may insist), that which is fundamental is the part played by the element in chemical reactions.
Electron shell usually matters more in this connexion than atomic number. Normally then the
relevant similitude should embrace both isotopes under one element. (But there may be grounds
for another decision in special cases, e.g. those that were marked by the name ‘deuterium’). This
does not reduce real classification to our decision. The decision relates to the identification and
refinement of the interest that is in question, not to the inclusion or exclusion as such of a given
thing or stuff.
12
See also Kripke, Lecture , ‘Naming and Necessity’ (). Compare Leibniz, N.E. ...
Related conclusions have been reached by Vernon Pratt (), pp. –. See also W. V. Quine
(); and Hidé Ishiguro ().
The terminology in which I shall expound all this will coincide with Frege’s usage wherever
Frege had a serviceable term, and will sometimes diverge from Putnam’s. A sortal predicate has
a sense, a reference or designation, and an extension; and there pertain to it both a concept (which may
have marks) and a conception. On these matters see my () and () and see Preamble § and
following. Summarizing these, here is the terminology I try to adhere to:
[] The reference of a sortal predicate such as ‘horse’, i.e. what ‘horse’ stands for, is the concept
horse, the general thing horse.
[] The extension of a sortal predicate such as ‘horse’ consists of the entities falling under the
concept it designates, i.e. horses.
[] The sense, or contribution to truth-conditions, of a sortal predicate may be elucidated by specify-
ing what concept the predicate stands for, and seeking to impart a certain conception of that.
[] A conception of horse is a set of beliefs concerning what horses are, or what it is to be a horse.
The conception is in no way the same as the concept. The conception is of the concept.
[] Finally, a mark ␾ of a concept f is any ␾ such that (᭙x)(x is an f→ ␾ x).
I remark that almost any appreciable change in people’s beliefs about horses might constitute a
shift in their conception of horses. Such a shift, taken far enough, might in the end deserve to
be accounted as a change in sense. But only in a rare or extreme case could a change in concep-
tion result in anything so extreme as a shift in the reference of ‘horse’. This is not to deny that it
may result in a correction of previous conceptions of that extension.
I have tried to present Putnam’s theory in a manner that (a) is compatible with a supplemen-
tary doctrine of his about the stereotypes which support human understanding of substantives
(stereotypes being standardized sets of beliefs or idealized beliefs associated with terms, and
belonging with conception); (b) leaves sense itself relatively insensitive to minor shifts in concep-
tion, and to minor changes in the criteria used by experts; but (c) involves no unnecessary or
dubious claims to the effect that the sense (contrast reference) of a natural kind word is invari-
ant through change in theories, beliefs or conceptions. On these matters see, mutatis mutandis, J.
H. McDowell (), and above, Preamble, § and following.
 Sortal concepts
relevant way to genuine exemplars (where the proper standard of rele-
vance is not abstract or general but internal to the nature that would be
revealed by study of f-exemplars). It should be noted that Putnam’s
account, like that of Kripke, not only bears upon the semantics of words
for natural kinds whose instances are particulars or substances. It also
bears upon the semantics of words for natural stuffs. That matter is
equally important, and the parallelism is important and illuminating. In
this book, however, the chief or central concern is with particulars/sub-
stances and their kinds.
It is fully compatible with Putnam’s suggestion (which should not be
read as a recipe for the production of formal definitions of the substan-
tives to which it applies, only for real definitions in the sense of Preamble
§) that the theoretical description that comes into question in a given
case should make reference to both the microphysical and the macro-
physical. (Some of the useful theoretical notions in this area, gene-pool for
instance, really make reference to both.) As this figures in Putnam’s
account, the act of deixis will not prescribe at what level the relevant
facts should lie. Nor is it implied by Putnam’s proposal that the people
who use a natural kind word will know what the true theory of the kind
is – only that, if they take the word seriously as the name of a natural
kind, then their linguistic practices will finally depend upon the suppo-
sition (however inexplicit, and not always borne out)3 that such a theory
is there to be found. All the doctrine implies is that the determination of
a natural kind stands or falls with the existence of lawlike principles,
known or unknown, that will collect together the extension of the kind
around two or three good representatives of the kind. In order for the
name to stand for a natural kind, everything depends on whether there
is some nomological grounding for the application of the name. If there
is, and if the predicate is worthy to survive as a natural kind term, then
the holding of the relevant principles is constitutive of its exemplifica-
tion by its instances. To be something of that kind is to exemplify the dis-
tinctive mode of activity that they determine.
In this conception we are driven back to an unmysterious but pre-
empiricist notion of substance (a notion that is here to be progressively
enriched) which it would have been both possible and advantageous for
Aristotle to distance from his concern with final causes:
13
Witness what befell ‘caloric’, ‘phlogiston’, and Leuwenhoek’s ‘hominids’. On such terms see
Kripke, op. cit. (footnote ), and Lecture One, ibid., paragraph one, and his addendum at pp.
–. See also G. Frege, Nachgelassene Schriften, ed. H. Hermes, F. Kambartel, F. Kaulbach
(Hamburg, ), p. .
 Sortal concepts 
Things which exist by nature . . . such as animals and the organs of these or
plants and the elementary stuffs . . . have in them a principle of change or rest
(in respect of place or growth and decline or alteration generally) . . . the nature
of a thing being the source or cause of non-accidental change or rest in any-
thing to which it [the nature] belongs . . . (Physics Book II, Ch. );
and which Leibniz did divorce from all final causality except that of
God’s decree:
Aristotle has well called [nature] the principle of motion and of rest, though that phi-
losopher seems to me to take the term more broadly than its accepted meaning
and to understand by it not only local motion and rest in a place, but change
and stasis or permanence in general (p. ) . . . [The] divine law once estab-
lished . . . has truly conferred upon [things] some created impression which
endures within them, or . . . an internal law from which their actions and passions
follow (p. ) . . . if the law of God does in fact leave some vestige of him
expressed in things . . . then it must be granted that there is a certain efficacy
residing in things, a form or force such as we usually designate by the name of
nature, from which the series of phenomena follows. (On Nature Itself, or on the
Inherent Force and Actions of Created Things, Gerhardt ,  ff., as translated at
pp.  ff. by Leroy E. Loemker in Leibniz, Philosophical Papers and Letters, nd
edn, Dordrecht, .)
Putnam’s account of natural kind words, as combined with the impor-
tance we ascribed to the what is it? question, provides a fresh understand-
ing of one part of the point of these doctrines. However rough and
unfinished such a suggestion may seem when compared with the variety
and complexity of all the classifications that are implicit or explicit in our
linguistic practices (consider our divers substantives for genera, species,
subspecies, crossbreeds, variants, etc), it has the signal virtue of reveal-
ing or recalling to us a possibility scarcely even imagined by those who
have preferred to theorize in abstraction from particular kinds of sub-
stantive, or preferred to erect transcendental arguments upon the foun-
dation of such supposed truisms as ‘because all our concepts are tools
for the intellectual handling of our sensory intake, we cannot make sense
of any statements about the world except ones admitting of a broadly
phenomenalistic analysis’.4
14
Jonathan Bennett in ‘The Age and Size of the World’, Synthèse,  (). Of the numerous phi-
losophers who have been party to this conception I know only one who dared to draw the proper
conclusion, viz. Nietzsche.
We speak of a ‘serpent’; the designation fits nothing but the sinuosity (the serpere), and could
therefore appertain also equally to the worm. What arbitrary demarcations! What one-sided
preferences given sometimes to this, sometimes to that, quality of a thing . . . What therefore
is truth? A mobile army of metaphors, metonymies, anthropomorphisms; in short a sum of
human relations which become rhetorically intensified, metamorphosed, adorned, and after
 Sortal concepts
Not very many things are entirely new in philosophy. The insight we
have borrowed here has been anticipated several times. But that which
earlier attempts have lacked, with the exception perhaps of Leibniz’s, is
the explicit combination of two distinctive features: () the component
of deixis, which requires of any candidate for kind f that it be relevantly
like something that is directly or indirectly identifiable and is actually an
f, and () the explicit acknowledgment of the indispensable part played
in the determination of standards of relevant likeness by the laws or
lawlike propensities of the real world. With or without this feature,
however, all views of this kind have the distinctive consequence, which
is attested by experience, that it is possible for what appear astonishingly
like fs to be not fs, whereas the most improbable seeming specimens may
turn out to be fs after all. (Hidden structure dominates apparent, as
Heraclitus puts this at fr.  Diels.) It can also be a matter of prolonged
and difficult inquiry gradually to improve currently accepted standards
or conceptions of what it is to be an f.
According to the account suggested by Putnam, the semantics of a
kind word will accommodate all this. There is room for one and the same
concept of what it is to be an f, anchored as it is to examples that are
grouped together in virtue of resemblances that are nomologically
grounded, to be unfolded gradually in a succession of different and
improving conceptions. As the primitive mentality gives way to the
protoscientific, it can be predicted that human beings will entertain the
thought that it is extraordinarily mysterious what the most ordinary
objects are. They may come to believe they are gradually awakening
from a pre-conscious state in which they never grasped properly what
footnote  (cont.)
long usage seem to a nation fixed, canonic and binding; truths are illusions of which one has
forgotten that they are illusions. (On Truth and Falsity in the Ultramoral Sense.)
Ingenuity and elaboration of the phenomenal conception can postpone this outlandish conclu-
sion. The conclusion Nietzsche draws about truth can be blurred by the deliverance of common
sense that worms do not in fact belong to the extension of ‘serpent’. But might they just as well
have? If the semantical conception that sustains Nietzsche’s inference to his conclusion is in vio-
lation of common sense – and, in the presence of an alternative conception of the working of
substantives, it becomes all the easier to suspect that the phenomenal conception enjoys no
special obviousness of its own – then the common-sense finding that worms are not serpents is
irrelevant and the ingenuity is misplaced. It is a mere postponement of the startling conclusion
that Nietzsche draws.
Semantical fact is almost always more interesting than transcendental philosophical fiction;
and for the imagination a hundred times more potent. We do not really have to choose between
() the phenomenal cum reductive way of Berkeley and Mach, which assimilates the criterion for
being an f to some set of tests for f-hood, and () the opposite way, which assimilates what would
be required by way of a real test for f to the constitutive criterion for f-hood.
 Sortal concepts 
they encountered, nor comprehended it even when they learned to rec-
ognize it.5 Moreover, the doctrine illustrates in a schematic way how a
word can enter the language on the slenderest and most provisional basis
of theory – just enough to yield what is required for the recognition of
an extension (which can itself be subsequently purified) – and draw
credit on a draft that Nature may or may not finally honour.
Once one has seen the possibility of a natural kind word’s having
such a debut, one will be tempted, in the case of natural things, to
replace all doctrines that distinguish in the customary way between the
real and nominal essences of a given kind of thing (contrast our use of
these notions at Preamble, §) by an account of the different stages in
the unfolding of what it is to be an f. Let us try to understand and
describe a process by which clear but indistinct knowledge – knowledge
sufficient for recognition but without the enumeration of distinguish-
ing marks – is replaced by clear, more distinct (better itemized) or more
adequate knowledge.6 In the light of an adequate account of this, it
should pose no difficulty to suggest that one could pick out (say) a
tadpole and identify its kind by saying ‘the members of the kind are
anything relevantly like this’, while leaving the actual standards of rel-
evance to be scientifically determined at macrolevel and microlevel;
then that one can be surprised to find a principle of individuation for
members of the kind committing one to count into the kind a frog –
not of course as a present tadpole, but as a member of the very same
kind as the Batrachian specimen recently demonstrated. Here the dis-
covery may or may not result in some modification of the received
sense. But all that it needs to be is a realization of the possibility, which
was always there, of fuller or more detailed specification both of what
it is to be an f and of which things are f.7
15
Cf. Heraclitus fragments ,  Diels; and Leibniz ‘[les hommes] sont empiriques et ne se gou-
vernent pas que par les sens et les exemples, sans examiner si la meme raison a encor lieu’
(Nouveaux Essais, Gerhardt , p. ). See also § of the Preamble, with footnote .
16
Leibniz describes this process in several places and connexions. See especially his account of
clear but indistinct knowledge at ‘Meditationes de Cognitione, Veritate et Ideis’, Gerhardt , op.
cit., especially p. . The Leibnizian distinction between clarity and distinctness is one I have
availed myself of constantly, here and in subsequent chapters, in developing the account given
here of sortal concepts and our conceptions of their determinables and determinations. See also
my ‘Reply to Fei Xu’ (a).
17
Again someone might pick out an amoeba under a microscope, say ‘that cell [or that creature],
and anything [relevantly] like it, I call an amoeba’, and commit himself in that way to the exis-
tence of some (as yet unknown to him) standard of scientific similarity. What he commits himself
to is to live with that standard and the associated principle of activity when he finds it out, and
to arbitrate persistence questions, when he graduates to these, in empirical and logical accord
with these. We have seen in Chapter Two (see D(vi)) how this minimal commitment leads into a
 Sortal concepts
Frege said that the sense of an expression determined its reference,
and not the reference the sense. If Putnam’s theory is correct, this order
of determination ought not to be expected to hold in the case of natural
kind words. Just as the sense of a proper name (contrast definite descrip-
tion) is best explained by saying what its reference is, so, in the process
of the teaching and elucidation of the sense of a natural kind-predicate,
everything depends upon providing specimen satisfiers of the predicate.
These are held as hostages for that which Frege would have called its ref-
erence – or as hostages, one might say, for the what it is to be an f. (For the
real coherence, despite verbal differences, of Putnam’s and Frege’s views
see my ().) Putnam’s suggestion does not, however, imply that refer-
ence is prior to sense. The principal effect of this supposedly salutary
exaggeration would be to raise difficult and perfectly irrelevant questions
about the possibility of prelinguistic deixis and the preverbal categoriza-
tion of experience. Putnam’s suggestion only implies that at this point
sense and reference are mutually dependent and correlative notions.
Suppose that all this is even roughly right. Then it contains most of
the answer to the problems that we have posed about the demands of
D(iii), D(iv), D(v) and all the other D principles. If there have to exist
true lawlike principles in nature to underwrite the existence of the multi-
ply instantiable thing that is the reference of a natural kind-predicate, if
lawlike principles of this kind have to exist in order for that general
thing’s extension to be assembled around the focus of actual specimens
or for a reality-invoking kind of sense to be conferred on the term stand-
ing for the concept f, then they must also determine directly or indirectly
the characteristic development, the typical history, the limits of any pos-
sible development or history, and the characteristic mode of activity of
anything that instantiates the kind.
The conclusion is not a novel one:
I also maintain that substances cannot be conceived in their essence as bare or
without activity, that activity is of the essence of substance as such . . . above all
it is necessary to consider that the modifications that can befall a subject natu-
rally or without miracle derive from limitation and variation in a real kind, or
from a constant and absolute original nature . . . and wherever one finds some
quality in a subject, one must believe that if one understood the nature of the
subject and of the quality then one would understand how the quality could
footnote  (cont.)
further commitment in this case – the commitment to count every splitting of an amoeba as the
demise of that amoeba. The commitment, though its working is logically constrained, is
empirically grounded. It is grounded in the actual activity of these unicellular creatures. (For the
interaction of logical and empirical, see further Chapter Six, §§–.)
 Sortal concepts 
result from them. So in the order of nature it is in no way arbitrary whether
substances have these qualities or those, and God will only give them those qual-
ities which naturally befit them, that is the qualities which could be derived from
the nature of the substances themselves as intelligible modifications of them.
. . . Everything that is must be such as to be intelligible distinctly, if only one
were admitted to the secret of things . . .8
The deep interest of this conception of the activity of a substance and
of Leibniz’s idea that the nomological foundation of activity is some-
thing supervenient upon fundamental laws of nature (see Gerhardt ,
, quoted at the head of this chapter) utterly transcends all the partic-
ularities of Leibniz’s finished view of substance (which in the end
dethroned Aristotelian continuants, seen as imperfect from this point of
view, in favour of monads). Its implication for the theory of individua-
tion outlined in Chapter Two may be summed up in an idiom that is offi-
cially alien to this book but, for the purposes of this particular thought,
very concise. Where fs make up a natural kind, the only possible worlds
we need consider, for these are the only possible worlds having within
them any entity that is an f, are worlds sufficiently similar in nomologi-
cal respects to the actual world to exhibit specimens relevantly similar to
actual specimens. This means that, for the special case of natural kinds,
we find that ordinary necessity virtually coincides with physical neces-
sity, and ordinary possibility with physical possibility. (Cf. Ishiguro loc. cit.)
It was difficult, until we thought about the sense of natural kind words
in Leibniz’s or Putnam’s fashion, to imagine how the requirements of
the last chapter could possibly be satisfied. But the principal difficulty
that now remains is epistemological, and is shifted to another point,
which is at once right and familiar: how to assure oneself, once one real-
izes what a considerable commitment such an identification will involve,
that any particular entity really is an f, or a g, or whatever.
It may be instructive at this point to try to illustrate how the theory of
individuation, the semantics of natural kind words and general findings
of theoretical biology are in a certain way made for one another. (The
18
Leibniz, Nouveaux Essais, preface. Compare, in connexion with the contentions of the preceding
paragraphs, Leibniz in Couturat, Opuscules et Fragments Inédits de Leibniz (Paris, ), pp. –.
I think that in this series of things there are certain propositions that are most universally true,
and which not even a miracle could violate. This is not to say that they have any necessity for
God; but rather that when he chose the particular series of things he did choose, he decided
by that very act to observe these principles as [giving] the specific properties of just this par-
ticular series of things. (Cf. Gerhardt , , quoted at the head of this Chapter Three.)
I have followed here the interpretation of Leibniz’s view that is defended in chapter , especially
§–, of Hidé Ishiguro ( and ) .
 Sortal concepts
significance of this exceeds the simple pleasure one might have in finding
such Leibnizian anticipations as Gerhardt , : ‘For there to be a
certain persisting law which involves the future states of that which we
conceive as one and the same continuant, this is what I say constitutes a
substance’s identity.’) Starting off with the almost pre-theoretical idea of
a sortal predicate whose sense is such as to depend on the sort of thing
that lies in its extension – the kind of predicate that cries out for real def-
inition – we are led to speculate what holds together the extension. So
soon as we find that, we find lawlike norms of starting to exist, existing,
and ceasing to exist by reference to which questions of the identity and
persistence of individual specimens falling under a definition can be
arbitrated. Such norms will supervene on basic laws of nature, we have
supposed; they may be understood as certain exploitations, so to say, of
these laws. But now we are led by simple conceptual considerations to
precisely the account of living substances that biologists can fill out a pos-
teriori by treating them as systems open to their surroundings, not in equi-
librium with those surroundings, but so constituted that a delicate
self-regulating balance of serially linked enzymatic degradative and syn-
thesizing chemical reactions enables them to renew themselves on the
molecular level at the expense of those surroundings, such renewal
taking place under a law-determined variety of conditions in a determi-
nate pattern of growth and development towards, and/or persistence in,
some particular form.9 (In certain cases there may be definite alternative
forms keyed to different environments, e.g. locust/grasshopper,
axolotl/salamander, but again this is something that is founded in the
solid basis of lawlike propensities.)

.            

Almost everything that has been said so far has been mainly directed at
words standing for the various species of natural substances. The
account could be extended and adapted without overwhelming difficulty
to predicates of genera, wherever these were still determinate enough to
be autonomously individuative. (See especially requirements D(iv) and
D(v).) And mutatis mutandis certain of the claims so far made would also
apply to words for the natural organs, heart, liver, foot, brain, and even to
19
Cf. Aristotle, Physics, b15: ‘For these things are natural which, by a continuous process origi-
nated from an internal principle, arrive at some end. The same end is not reached from every
principle, nor any chance end, but always the tendency in each is towards the same end, if there
is no impediment.’
 Sortal concepts 
geographical or geological terms such as river, lake, spring, sea, glacier or
volcano. In the latter case, it will not be wildly inappropriate to speak of
principles of activity. In the former, we can honour D(viii) in terms both
of principles of activity and of modes of functioning or operation. It would be
interesting to enter into detailed questions about just how well these
predicates measure up to the D requirements of the previous chapter;
then to enter all the appropriate and considerable reservations.10 But,
apart from the intrinsic interest of these kinds of entity, it would be dis-
tractive, and it would be seen as mere postponement by those who hold
that it is on artefact kinds especially that the individuative theory pro-
posed in Chapter Two must stumble.
Consider ordinary artefact-words such as clock, chisel, drinking vessel,
drill, stove, pen, spade, table. Putnam’s account is not happily extended to
these. Of course the easiest way to explain, in a given case and context,
what a tool or artefact of some sort is might be to proceed by the use
of examples. But the deictic nomological method is scarcely indispens-
able. Moreover, in the absence of the right sort of commentary, it could
be positively confusing or lead to an implausible narrowing of the
reach of the artefact kind. There are virtually no lawlike sentences to
be had about particular utensil, implement or tool kinds as such. There
scarcely could be. Clocks, for instance, may be made of a variety of
different kinds of material and may function by radically different
kinds of mechanism (cf. Locke, Essay, ..). Artefacts are things to be
encountered in nature, and subject to its laws. But they need not be col-
lected and classified together as this or that artefact by virtue of inter-
nal resemblances or scientific resemblances. Artefacts are collected up
not by reference to a theoretically hypothesized common constitution
but under functional descriptions that are precisely indifferent to spe-
cific constitution and particular mode of interaction with environment.
A clock is any time-keeping device, a pen is any rigid ink-applying
writing implement, and so on. The description gives what it is usually
impossible to specify in the other cases, an explicit nominal essence;
whereas a finite and determinate set of marks fully suitable for defini-
tional purposes is precisely not what the members of natural kinds
10
There need be no one physiological theory to cover all animal hearts of all kinds. There is no
one specific structure which a heart must have, nor any one fully specific mode of biological func-
tioning that a heart must as such exemplify. To be a heart (simply contrast a human heart, a
sheep’s heart) an organ need only be an organ of some definite physiological kind or other playing
a particular role in the circulation of the blood.
Such terms may seem then to be hybrid between functionally defined terms shortly to be char-
acterized and natural kind terms.
 Sortal concepts
endowed with a scientifically palpable real essence may be expected to
have in common.11
For the theory of individuation generally, and for the theory here
advanced especially, there is then a crucially important difference
between natural things and artefacts, both in respect of conditions of
persistence through time and in respect of the satisfaction of the require-
ments upon sortal predicates sketched in Chapter Two. Where there is
dispute concerning an object identified as a member of a natural kind,
one can readily conceive of getting more scientific facts. Consider the
nineteenth-century discovery that the elvers Leptocephali were in fact the
young of the species Conger Eel, or the humbler but in some sense
proto-scientific discovery that tadpoles become frogs and maggots
houseflies. These discoveries changed conceptions (not concepts). In the
light of the deictic/nomological theory, it will be apparent how such dis-
coveries can engage with the semantics of ‘eel’ or ‘frog’ even as they
advance the collective grasp of concepts. (For the thing they reveal is
allowed for in advance in the fixing of the designation of the kind
word).12 We can see why identity questions about members of natural
kinds may be expected to find the notion of identity at its best. Empirical
discovery playing the part it does play here, it is plain too why they are
the most unsuitable of all candidates for any conventionalist treatment.13
By reference to the natural propensities that constitute the foundation of
the natural kind concept, we can make a clear distinction between
normal activity and what happens under conditions where there is inter-
ference. (Cf. Chapter Seven, §.) It is in no way up to us what to count
as persistence through change or through replacement of matter. There
11
Concepts of artefact kinds, being verbally analysable in a way that is excluded for concepts of
natural kinds, offer more prospect of generating truths that are analytic (in the strict sense of
‘analytic’, viz. truths reducible by explicit definitions to truths of logic). Even here though, and
in the place where such truths would be less interesting, the matter is not unproblematic – if only
because of the high degree of actual vagueness inhering in the definitions of artefact kinds and
the scope there is here for innovation. In any case, neither here nor in other chapters does any-
thing whatever hang on analyticity. It would do no damage whatever to the doctrines of nature,
substance and necessity put forward in this book if the only analytic truths were the logical truths
of first order logic, as afforced by means of formally correct, purely stipulative definitions.
I have claimed that Putnam’s account of natural kind words is not happily extended to arte-
fact words. But I ought to mention Putnam does seek to apply his account to such words, e.g. to
‘pencil’. See ‘The Meaning of “Meaning”’, at pp. – of Mind, Language and Reality,
Philosophical Papers, Volume  (Cambridge, ).
12
Here, as throughout, I hope to maintain the terminology proposed in footnote  of this chapter
and the disposition set out in Preamble § and §.
13
For the original impulse to conventionalism, see Plato, Symposium d3 and following, and
Thomas Reid (), , . For the association between conventionalism and parts, see Peter
Simons (), Chapter .
 Sortal concepts 
is some sort of contrast here with the identity problem of artefacts. To
some limited extent, there may be shadowy counterparts even for arte-
facts of that which a natural kind concept furnishes straightforwardly.
But in the case of damage to a watch, say, or extensive replacement of
its parts, there is nothing theoretical or extra to discover, once the ordi-
nary narrative of events is complete. For there is no such thing as the
natural development of a watch or a lawlike norm concerning the beha-
viour of watches as such.
It will be useful then, before we grapple with the problems of arte-
fact identity, to return to the first chapter of Book Two of Aristotle’s
Physics:
animals and their organs and the elementary stuffs . . . differ from what is not
naturally constituted in that each of these things has within it a principle of
change and of staying unchanged, whether in respect of place or in respect of
quantitative change, as in growth and decay, or in respect of qualitative change.
But a bedstead or a cloak or whatever, qua receiving the designation ‘bed’ or
‘cloak’ . . ., i.e. in so far as it is the product of craft, has within itself no inher-
ent tendency to any particular sort of change. Though in so far as an artefact
happens to be composed . . . of whatever mixture of natural elements, it does
incidentally, as so considered, have within itself the principle of change which
inheres in its matter. So nature is a source or principle of change and staying unchanged
in that to which it belongs primarily i.e. in virtue of the thing itself and not in
virtue of an incidental attribute of the thing. I say ‘not in virtue of an inciden-
tal attribute’ because, for instance, a man who is a doctor might cure himself.
Nevertheless it is not in so far as he is a patient that he possessed the art of med-
icine. It was incidental that he satisfied both descriptions. And something
similar holds of everything that is an artificial product (b8–28).
There is an insight here that is so important that it deserves a formula-
tion free of the locutions ‘qua’, ‘in so far as’, ‘considered as’ at which
Aristotle clutches to give voice to his thought. (These are the locutions
by which some Aristotelians still persist in obfuscating this and many
other good thoughts: compare the discussion in Chapter One, § under
(␭).) Let us say instead that a particular continuant x belongs to a
natural kind, or is a natural thing, if and only if x has a principle of
activity founded in lawlike dispositions and propensities that form the
basis for extension-involving sortal identification(s) which will answer
truly the question ‘what is x?’ For purposes of this chapter, and for pur-
poses of the theory of individuation, it is not the question of whether
a thing was fabricated but rather the difference between satisfying and
not satisfying this condition that makes the fundamental distinction.
Loosely, and because there is no other handy term, I shall continue to
 Sortal concepts
call objects that fail this crucial condition ‘artefacts’. But this is without
prejudice to the question, which is scientific not philosophical, of the
possibility (which I have no wish to prejudge) of the artificial synthesis
of natural things.
It should go almost without saying that the distinction of natural
thing from artefact is presented here in a fashion conformable at every
point and in every particular with the plausible scientific belief that,
however we have arrived at the individuation of a given thing, the thing
will be subject to the fundamental laws of physics and chemistry. For
the purposes of that belief, there is no more difficulty in our distinction
between natural things and other things than there is in making a
further distinction, within the class of natural things, between (a) those
things which, being alive, are not in chemical and energy equilibrium
with their surroundings but suck from their environment the energy
that they need for their typical activity or their molecular self-renewal
and replacement, and (b) those natural things that maintain a typical
mode of activity without being alive and cannot help but be in equilib-
rium with their surroundings. (Take the Chesil Bank, for instance, or
the River Nile or the Caspian Sea.) It is true that live things exemplify
most perfectly and completely a category of substance that is exten-
sion-involving, imports the idea of characteristic activity, and is
unproblematic for individuation. But not all extension-involving con-
cepts are concepts of live things and I should stress that non-equilibrium
with surroundings is an additional and special distinguishing mark. In
fact, terrible confusion will result from mindless assimilation of the live
/not live distinction to the other two distinctions. In other words, ()
extension involving or not extension involving, with respect to the specification of their
concept; () not synthetically produced or synthetically produced; () live or not live
(as elucidated in terms of non-equilibrium and of self-renewal at the
molecular level) – these are three distinctions, for all that living animals
fall on the same side of each of them. Moreover, once we move in this
way towards a more fundamental basis of distinction, we need not
flinch from trying to understand some harder cases: an india rubber
ball for instance (individuated by a stuff sort that is extension-involving
and that simultaneously lends nomological import to a thing-kind, the
ball itself being something artificially produced, not live) – or even a
wasps’ nest (something individuated by reference to other things which
are themselves individuated in the extension-involving fashion, an
animal artefact not a human artefact, an artefact constrained in respect
of its material constitution, not live).
 Sortal concepts 

.             

Our concern passes now to the problems associated with singling out
artefacts and the similar difficulties we shall have with other things that
cannot furnish the theorist of identity with a principle of activity.
Summing up the argument so far in the general (if qualified) claim that
natural things are individuated by reference to a principle of activity nat-
urally embodied, but ordinary artefacts are individuated by reference to
a parcel of matter so organized as to subserve a certain function;14 and
advancing next to the consequences of the finding that there is nothing
nomological to make up for the logical indeterminacy that seems to
inhere in the individuation of artefacts, one feels the temptation looming
to exaggerate the degree to which questions of artefact identity are
matters of arbitrary decision. Even here, though, and even for the most
typical cases of our own artefacts, conventionalism is not immediately
plausible. For human purposes and decisions might enter into the inven-
tion and modification of human artefact-kinds without its following that
any particular questions of identity and difference (between this artefact
and that artefact) should qualify for decision by stipulation.
Artefact identity does, however, present some difficult problems. D(iii)
requires that the contribution that is made by an artefact-word such as
‘clock’ to the sense of ‘x coincides with y under the concept clock’ should
suffice to render this relation both an equivalence relation and a congru-
ence relation. But how is this to be secured? And how are we to get for
artefacts the effect that was got for natural things by the notion of activ-
ity? (The conceptual effect, I mean.) It would be unrealistic and absurd
to suppose that coincidence under the concept clock required a clock’s
continuous functioning or expect the functioning spoken of in D(v) to be
more than remotely analogous to the activity of natural things. A clock
may stop because it needs winding up. Such a pause does not prejudice
its persistence. A clock can stop because it needs to be repaired; and
again it persists, however long the lapse before the repair. (The nominal
essence of clock does not make ‘broken clock’ problematic for existence
or peculiar in the manner of ‘dead person’.) The nominal essence of clock
must involve a stipulation of some sort concerning the capacity to tell
the time. But surely the uninterrupted continuance for all t of the capac-
ity at t to tell the time at t will not be stipulated. This is too strong. The
14
Contrast works of art, which do not normally, qua works of art, have any function. See below,
Chapter Four, §.
Concerning the expression ‘principle of activity’, see Preface.
 Sortal concepts
only loss that could count to any appreciable degree against the persis-
tence of a clock is a radical and irretrievable loss of the time-keeping function.
(Sometimes, even under this circumstance, the clock itself may be held to
have survived, however irreclaimably damaged.) Another reason why a
clock-persistence condition has to be undemanding is that the repair of a
clock apparently permits both disassembly and replacement of parts. We
do not look back to the time when a clock was being repaired and say that
the clock’s existence was interrupted while it was in a dismantled condi-
tion. For a thing starts existing only once; and in the case of a clock its
proper beginning was at about the time when its maker finished it. In the
case of dismantling, there is not even a time limit upon reassembly.
When these several points of discontinuous functioning, disassembly,
and part replacement have been accommodated, the condition of coin-
cidence and persistence for clocks that emerges is not merely weak. It is
so undemanding that there seems to be nothing to prevent one clock,
identified by a description applying at the time when it was functioning
normally, from clock-coinciding (i.e. coinciding in a manner supposedly
sufficient for identity) with two distinct clocks, each identified under a
description applying to it after some radical repair or muddled reassem-
bly in a moderately disorderly clock maker’s workshop. But the coinci-
dence of one thing with two things breaches requirement D(iii), D(vi),
and almost every other interesting declaration made in Chapter Two.15
Nor is there one piece of clock – the spring, the regulator, the escape-
ment, the face, the case . . . which the concept clock could suggest that we
should revere as the ‘focus’ or ‘nucleus’ of a clock, and which can help
us past this difficulty. Here, and in the extremes of ingenuity and oppor-
tunism to which such problems pushed the Roman laws of accessio and
specificatio (cf., for instance, Justinian, Institutes Bk , . et passim), we see
that, however questionable it may be, Leibniz’s transition in the decade
of his writing the Discourse of Metaphysics from Aristotelian substances to
Leibnizian monads, and his demotion of things with parts to the status
of mere aggregates (for a characteristic statement see Gerhardt , ),
did not arise from a simple misconception of the logical level of the
number word ‘one’.
A good place to embark on the difficulties of artefact identity is with
a case that Thomas Hobbes put forward expressly against those who
regarded ‘unity of form’ as a principium individuationis: on this view, he
suggests,
15
It is worth noticing that it fails even the weaker requirements that relative identity theorists would
have to place upon the relation is the same clock as.
 Sortal concepts 
Two bodies existing both at once would be one and the same numerical body.
For if, for example, that ship of Theseus, concerning the difference whereof
made by continued reparation in taking out the old planks and putting in new,
the sophisters of Athens were wont to dispute, were, after all the planks were
changed, the same numerical ship it was at the beginning; and if some man had
kept the old planks as they were taken out, and by putting them afterwards
together in the same order, had again made a ship of them, this, without doubt,
had also been the same numerical ship with that which was at the beginning;
and so there would have been two ships numerically the same, which is absurd.16
The unity of form theory that Hobbes means to attack here is not dis-
similar from ours. In our version of the theory, the finding of a’s coin-
cidence with b under the concept ship needs to be the makings of a
sufficient condition of a’s being the same ship (unless the concept ship
is to be not a sortal concept). It might seem that in our theory the diffi-
culty Hobbes mentions is escaped because our theory favours the
repaired ship. For the constant availability of this ship, and its uninter-
rupted service as the sacred ship by which the annual voyage was made
from Piraeus to Delos, may seem to mark this ship out as the one that
is continuous in the manner of a ship with Theseus’ ship. But is such
a fact really enough to trump all the claims of a reconstituted ship,
which may well be much closer not only in matter but in its design, its
manoeuvrability and its handling qualities to the ship that Theseus
sailed? Can one who favours the working ship ignore all of this?
Cf. D(v).
Few of us will hurry to make a choice between the two options as so
far specified. Some will even say we do not have to choose. They will ask
us to imagine its being decided, in an age that no longer believed in
Apollo but still believed in Theseus, to erect a monument to Theseus and
to put his ship upon the monument. Then antiquarians might say that
the ship put together from discarded planks was the right one to raise up
there. But dispute might break out about this matter between them and
16
De Corpore, , H (Molesworth, p. ). Hobbes’s source for the puzzle is Plutarch’s life of Theseus:
The vessel in which Theseus sailed and returned safe with these young men went with thirty
oars. It was preserved by the Athenians up to the times of Demetrius Phalerus [floruit 
B.C.]; being so refitted and newly fashioned with strong plank, that it afforded an example
to the philosophers in their disputations concerning the identity of things that are changed
by addition, some contending that it was the same, and others that it was not. (§§–; cf.
Plato, Phaedo A; Xenophon, Memorabilia ...)
I do not know who proposed the final twist in the story that made it into Hobbe’s objection. An
addition to our modern store of problems in this area is furnished by Jonathan Barnes in ‘Bits
and Pieces’ in Matter and Metaphysics, ed. J. Barnes and M. Mignucci (Bibliopolis, ); another
by T. W. Rhys David’s The Questions of King Milinda (New York, ).
 Sortal concepts
the superannuated priests who favoured the working ship.17 The diffi-
culty (it will then be said) is a certain incomparability between the posi-
tions. It might be eirenically suggested that, if one party is looking for an
archaeological relic and the other for a functionally persistent continu-
ant, then the whole dispute ought really to be traced to a disagreement
about what it is for something to be such a sacred ship. The antiquarian
who favours the reconstituted ship has a different interest, it might be
said, from the priest who favours the continuously repaired ship. Both
are stuck with the identification ship but, having different interests, they
do not mean quite the same thing by ‘ship’. The priests and the antiquar-
ians find different entities with different principles of individuation in
one and the same place.18 Neither party can vindicate its view by refer-
ence to the ‘laws of development’ of a ship.19 There are no such laws.
17
A mereologist making reference to the mereological definition of identity
(a⫽b) ⫽ df [(a is a part of b) & (b is a part of a)]
might well side with the antiquarian and against the priests. Evidently, though, it is a precondi-
tion of the acceptability of the mereological definition of identity, supposing that this is what is
invoked on the antiquarians’ behalf, that, for any non-empty class of individuals ␣, there should
be one and only one sum of all elements of ␣. That is to say that the sum axiom of mereology
must be unrestrictedly true, true whatever entities we admit into the class ␣, including continu-
ants that persist through loss or gain of parts. (The sum axiom is the second postulate in the
tenseless presentation of mereology given by Tarski, op. cit. in footnote  of Chapter One. See
the text of Chapter One for the definition of ‘sum’.) Under the conditions stated, however, it no
longer looks obvious that this axiom is true, unless ‘part’ is given a sense specially designed to
save the plausibility of Tarski’s postulate . In that case, any mereological definition of identity
that depends upon the axiom loses any direct usefulness it might have had for the antiquarian
cause. For my own conversion to the view that Tarski’s second postulate is false, see Wiggins
(). For a general account of the doubtful status of the second postulate see now Simons
(). Note also that, on a strictly mereological view, it is neither here nor there for the sum or
fusion whether the elements are together in one place.
Relative identity theorists are apt to claim for their view of such puzzles as Theseus’ ship a
particular naturalness. Griffin in Relative Identity, p. , provides what he calls the ‘simple and
appealing’ solution that the original ship of Theseus is not the same ship as the reconstruction,
though it is the same plank collection. There is then nothing peculiar at all about the case. The
reader will have to decide for himself whether it is a virtue or a vice in the relative theory that it
so easily distances the disputants from the issue which they suppose they disagree about – and
from the ontology of absolute identity.
18
For careful discussion of some of these issues, see Brian Smart, ‘How to Reidentify the Ship of
Theseus’, Analysis, .  (April ), pp. –, and ‘The Ship of Theseus, the Parthenon, and
Disassembled Objects’, Analysis, . (Oct. ), pp. –. But see also the response by Francis
W. Dauer, ‘How not to reidentify the Parthenon’, Analysis,  (–), p. .
19
There may then be a temptation to dismiss the question as merely psychological. But we must
not confuse the fact that it is in some sense a psychological matter whether we ourselves come to
adopt the priests’ or the archaeologists’ view with its being a merely psychological matter which
one is Theseus’ ship – or an arbitrary matter. Compare Frege (), §:
The objectivity of the North Sea is not affected by the fact that it is a matter of our arbitrary
choice which part of all the water on the earth’s surface we mark off and elect to call the ‘North
Sea’. This is no reason for deciding to investigate the North Sea by psychological methods . . .
 Sortal concepts 
This third view seems as unsatisfactory as the uncritical ship-continu-
ity view and the antiquarian-cum-mereological view. Let us call it the
naïve conceptualist view. Admittedly it divorces the antiquarian view,
sensibly enough, from crude mereologism. But in the process it makes
the antiquarian view seem unstable. By its nature antiquarianism looks
back to Theseus. But Theseus looked forward to the priests who would
continue the cult he took himself to be establishing. The competing con-
ceptions of Theseus’ ship really are rival views. Philosophical quietism
is a sensible outlook upon some questions, but in the present case it does
not convince. We need at least one more idea.

 .                   

    
Two further methods suggest themselves for the resolution of such diffi-
culties. These will define a fourth and a fifth view of the example. The
fifth will emerge in the form of an improved application of the
unity/continuity of form view advocated in this book. The fourth is an
alternative approach widely adopted in the present-day literature of
identity and individuation.20 This runs contrary to the spirit of Chapter
Two and involves our seeing a certain further complexity in the notion
of f-coincidence.
The fourth view proposes that we start with the relation in which x
and y stand just in case y is a coincidence-candidate (of type f) for identity
with x. We then say that y veritably coincides under f with x if and only if y
is an f-coincidence candidate for identity with x and nothing distinct from y is as good
an f-coincidence candidate for identity with x as y is. If one wanted to proceed
like this, one could then elaborate an account of the degrees of strength
in a candidature, and arrange to distinguish best from second best, third
best . . . claimants. In application to the present case, the best candidate
theorist might say (i) that, where Theseus’ ship is constantly repaired
without any retrieval of the parts that are shed, the working ship persists
into an open future; (ii) that, where a ship is reconstituted and the

If we say ‘The North Sea is , square miles in extent’ . . . we assert something quite objec-
tive, which is independent of our ideas and everything of the sort. If we should happen to
wish, on another occasion, to draw the boundaries of the North Sea differently . . . that would
not make false the same content that was previously true. (Cf. Dummett (), p. .)
20
I am indebted to my colleague Naci Mehmet for persuading me to take this approach more seri-
ously than once I did and helping me to understand it better. He will wish me to enter the normal
disclaimers on his behalf.
 Sortal concepts
working ship is no more, the reconstituted ship is Theseus’ ship; (iii) that,
where a ship is reconstituted from discarded planks but the working ship
is not destroyed and continues working, the working ship is Theseus’ ship
and persists into an open future. (These decisions reflect the perceived
strengths of considerations that are meant to count, however defeasibly,
in favour of an identity.)
In the application to our case, the best candidate proposal, so given,
perpetuates some questionable preferences, especially concerning the
working ship. (Does it really not matter, as in (i), provided that the
working ship continues and is the sole candidate, what changes the can-
didate undergoes? Could Theseus’ ship continue in the shape of a Dutch
barge or a Chinese junk?) But the proposal might easily be hedged or
qualified. Nor is it necessary to take exception to the fact that the best
candidate theorist’s method is dialectical or (in the correct and non-
condemnatory sense) casuistic. The chief objection is rather that this
approach proves inappropriate, as so far characterized, to questions of
identity. The most general form of the objection is this. In notionally pur-
suing object a in order to ascertain its coincidence or non-coincidence with b, or in
retracing the past history of b to ascertain its identity link with a, I ought not to need
to concern myself with things that are other than a or other than b. This is not to
deny that singling a out or singling b out may involve contrasting a or b
with a background of other things. But the identity of a with a, of b with
b, and of a with b, once we are clear which things a and b are, ought to
be a matter strictly between a and b themselves. Let us call this D(x) or
the Only a and b rule and adjoin it to the other D principles of Chapter
Two.21 In its official form – see the sentence italicized above, as glossed
by the two succeeding sentences – this is not a principle that is stated in
the object language. Contrast D, D(i), D(ii), D(iii). It is akin to D(iv) or
D(v). It is a dialectical rule for the adjudication of persistence and coin-
cidence claims founded in the nature of the identity relation.
Consider the cases (ii) and (iii) mentioned in the last paragraph but
one. Here it is supposed that the reconstituted ship claims identity with
Theseus’ ship; and whether this title is accorded to it or not depends at
least in part on the presence or absence of the working ship. Does that
contravene the Only a and b rule proposed in the previous paragraph? It

21
Compare Bernard Williams (), a discussion antedating in the philosophical literature of
identity the rediscovery of Miss Barcan’s proof (), of the necessity of identity. See also his
(), pp. –. At Philosophical Explanations (Oxford, ), pp. –, Robert Nozick animad-
verts sceptically upon this insight of Williams’s. But I do not think he measures up to its full
logico-metaphysical rationale.
 Sortal concepts 
appears that it does. In the present version, the best candidate theory
sees the reconstituted ship as a weaker claimant than the working ship.22
If so, then the working ship candidate and the reconstruction candidate
must be distinct. But then (as remarked) the reconstruction’s identity
with Theseus’ ship, wherever it is identical, seems to depend on a third
thing.
I should expect that, at this point, someone will ask why one should
not simply reword the fourth view. Why not say that, where the claim
of the reconstituted ship is good, namely in case (ii), this ship, precisely
by reassembling Theseus’ ship, ipso facto reconstructs a ship that was at
one point a working ship, was constantly repaired, then broke up
(though some of its parts were recovered)? That suggestion, it will be
said, has the effect of preserving the Only a and b rule. And I agree that
it does. But to preserve the rule in this way effectively dispenses alto-
gether with the best candidate formulation. (See below, the fifth view.)
The reconstructed ship is Theseus’ ship, if that can be made out, not
because it is the best candidate for the role but because it is in this
reconstruction from the original materials that the effort to track
Theseus’ ship from its beginning onwards finds Theseus’ very ship.
One finds the ship that is Theseus’ ship and thereby one finds a ship
that was reconstituted by restoring to Theseus’ ship, after accidents
that befell it at some intermediate moment, original planks and mate-
rials.
Someone who was as sure as Hobbes was that working ship and
reconstituted ship could coexist might try to salvage the rivality that the
best candidate view assumes by asking us to consider the reassembled
ship ‘on its own’ – or as it might have been if it had been put together
before the breaking up of the working ship. But that only starts more
trouble. Do we want to say of this reconstruction that it is premature but,
if only it had appeared later, it would have been identical with Theseus’
ship? Or do we want to say that at the break-up of the working ship, the
prematurely reassembled ship becomes identical with Theseus’ ship? (Can
literal sense be made of being not identical at one moment and then

22
For that which is surprising in the very idea of weaker and stronger claimants, compare Arnauld,
letter of  May  to Leibniz:
I can as little conceive of different varieties of myself and a circle whose diameters are not
all of equal length. The reason is that these different varieties of myself would all be distinct
from one another, otherwise there would not be many of them. Thus one of these varieties
of myself would necessarily not be me: which is manifestly a contradiction. (Gerhardt , ,
translated by H. T. Mason.)
 Sortal concepts
becoming identical?23) Neither of these suggestions is at all promising.
Let us remember that the title in question is not the title to the sobriquet
‘Theseus’ ship’. It is the title to identity with Theseus’ ship, a particular
ship originating from the eighth century BC.
Similar troubles arise for the best candidate view where the finding is
that we have case (iii), namely the working ship’s defeating the candidate
by reconstitution. The best candidate theorist says that the working ship
is Theseus’ ship by virtue of its being a stronger candidate than the
reconstructed ship. That infringes the Only a and b rule. Of course, we
can respect the rule better if we can find some other basis on which to
say that the working ship is indeed Theseus’ ship. But here we must
remember that this other basis might have nothing to do with the rela-
tive ‘strength’ or ‘weakness’ of the claims of reconstituted ships against
working ships. One might say instead that, in tracking Theseus’ ship
through its journeys and various repairs, one comes upon the working
ship, not a reconstruction from original materials. The idea of candi-
dacy and strength of candidacy could, of course, be restored to the scene
by someone’s pointing insistently to the reconstructed ship, and describ-
ing its claims as not quite good enough, etc. But once that line is pre-
ferred, there is further trouble. We are laid open to the thought that this
reconstructed ship, now lying in Piraeus, this very ship – if only the
working ship, which is the so-called ‘better candidate’, were not still
extant and plying once yearly to Delos – would have been Theseus’ ship.
Someone might perhaps say that. But the idea that on these terms,
namely the simple absence of the stronger contender, the ship now
deemed the worse candidate would have been Theseus’ ship itself seems
23
I will give an argument that literal sense ought not to be made. Suppose that we can make sense
of a and b’s being identical at time t, or
a⫽b
f, t
Suppose then that it is true that
a⫽b
f, t
and suppose that, in addition to existing at t, a will also exist at t´. Then so will b exist at t´. For
a is at t the same f as b is. So
a⫽b
f, t⬘
Indeed, anything that can be truly said at t about a can be truly said about b too, whether the
thing said be past, present or future. Either that or we don’t know what it is for a and b to be iden-
tical at t. How then can temporal identity require anything less than full Leibnizian congruence.
Generalizing then, we have for any f,
᭚t冢 冣
a⫽b
f, t
→ ᭙t⬘ 冢 a⫽a a⫽b
f, t⬘
→ 冣
f, t⬘
.
 Sortal concepts 
to be absurd. There is a temptation to add as a separate modal step:
nothing might have been a different entity from the entity it actually is.
(Compare Chapter Four, §.) But it is the utter oddity of such a claim
about the weaker candidate that discovers to us the real intuitive grounds
for doubting that anything might have been a numerically different
entity from the one it actually is. The absurdity of the idea confounds
the understanding of ‘⫽’ that is implicit in everything we think and say
about identity. If the issue is one of what founds what, then the impos-
sibility of conceiving of an entity’s not being identical with that with
which it is in fact identical seems to be the real foundation for the modal
claim, rather than the other way about.
A simple conclusion suggests itself – that the best candidate theory,
lacking obvious means to talk sensibly or intelligibly about plural candi-
dates for identity, must cede place to a theory that identifies, for a given
circumstance and a given question of identity, the best way to think of a
thing and the best way to track the thing. Preferably this will be a theory
that excludes, by its operation and application, the very possibility of dis-
tinct rivals for identity with Theseus’ ship. Such a theory will surely reg-
ulate the dialectic of same and different by the Only a and b rule. An
adherent of this fifth view will say that D(x) could only be abandoned on
pain of forswearing any claim to the effect that the dialectic just spoken
is a dialectic of real sameness and difference.24
The search for the fifth view already promised and hinted at may use-
fully take off from a simple reflection: the problem with which Theseus’
ship confronts us is that of conjoining the ordinary commonsensically
strict notion of identity with the commonsensically loose requirements
that we place upon artefact persistence. Even if we are to relax cau-
tiously the demands we have been placing on artefact identity or we are
to recognize the tolerance that so many of our conceptual practices
require and in real life rarely abuse, we can scarcely grant ordinary arte-
fact concepts or the things that satisfy them an indefinite tolerance of
disassembly, repair, discontinuance of function, and part replacement.
Unless we grasp this nettle, and are prepared for some degree of reform,
either of concepts or of the philosophical account we give ourselves of
our concepts, we shall not disarm Hobbes’s ship paradox. Will this mean
that we are faced here with a stark choice between abandoning the laws
24
This is not of course to deny that a third thing may be evidentially relevant (as in detective rea-
soning) to the question whether a⫽ b. For further discussion of D(x) see Harold Noonan,
‘Wiggins, Artefact Identity and Best Candidate Theories’ and ‘The only X and Y Principle’, both
in Analysis,  (). But see also Nathan Salmon (), pp. –.
 Sortal concepts
of identity and abandoning – in the spirit of the maxim no entity without
identity – the received ontology and ideology of artefacts? That fearful
outcome is to some extent anticipated in the high metaphysical tradition
of substance that seeks, for related reasons, to demote artefacts from the
status of genuine entities.25 But in truth, the data before us suggest a
much less exciting conclusion.
Suppose that we forget analogies with living substances and, running
to the opposite extreme, we remake our conception of artefact persis-
tence on the model of the identity-condition for quantities that has been
proposed by Helen Morris Cartwright.26 The condition she proposes
does not exclude change, but it excludes all addition or subtraction of
matter whatever. As remodelled and adapted to the formulation of a
persistence condition for artefacts, this condition would need to be made
more demanding in one way, in order for it to require some however ves-
tigial continuance of the thing’s capacity to subserve the roles or ends
the artefact was designed (as that very artefact) to subserve. (Cf. D(v).)
On the other hand, the ‘no subtraction’ condition could be relaxed
somewhat without our confronting the risk that condition D(vi) will be
breached. Even on these revised terms, the possibility can still be
excluded of competition à la Hobbes between different candidates for
identity with Theseus’ ship. Some of an artefact’s matter may perhaps
be exchanged with matter from its surroundings, provided that all
replacement of material parts is referred back to the first state of the fin-
ished artefact, and provided only, if the reader will forgive the comical
precision of this first attempt, the artefact retains more than half of that
original matter (or provided that it retains, where such is definable, the
material of some individuatively paramount nucleus). Under this condi-
tion there will never be more than one claimant to the title of ship of
Theseus. Under this condition, it can remain determinate whether the
ship of Theseus itself has been traced effectively through time. (For the
requirement of determinacy, see Chapter Six, §, thesis (iv).) Finally, by
reference to some strict condition like the one we began with or some
25
Aristotle maintained that natural things are the real beings par excellence to which everything
else is secondary. Leibniz maintained that the title of real unity must be reserved to ‘animate
bodies endowed with primitive entelechies’. Cf. Leibniz, Nouveaux Essais, ..: ‘It is true that
there appear to be species which are not really unum per se (i.e. bodies endowed with a genuine
unity or with an indivisible being which makes up their whole active principle), any more than
a mill or a watch could be . . .’: ... ‘In a word, perfect unity should be reserved for animate
bodies, or bodies endowed with primary entelechies; for such entelechies bear some analogy to
souls and are as indivisible and imperishable as souls are.’ Cf. also .., on which see R. Coburn
(). For further discussion, see my (), p. .
26
See Helen Morris Cartwright (), pp. –.
 Sortal concepts 
modification of that, it becomes possible to see our linguistic practices
as underhung by a sort of safety net that justifies the confidence we feel
in our capacity for controlled opportunism. The lesson we find ourselves
learning from Hobbes is one that we can see to be reasonable in itself,
namely that the ‘unity of form’ conception of individuation needs to be
carefully constrained in its application to things that evolve in a way that
is not nomologically controlled and that submit to repair by replacement
of parts. The question of their persistence needs to be stringently regu-
lated by considerations relating to change of matter. (This is not,
however, an essentialist claim. It relates only to the limitations upon
change. See below Chapter Four, §. Nor yet is it surprising.27 See
Chapter Six, §.)
I do not want to claim too much for the fifth approach. It is doubtful
whether any admissible relaxation of the strict condition we began from
can be guaranteed to account for each and every judgment of artefact
identity by which we are rationally tempted. But in respect of the more
audacious judgments of artefact identity, where the concept will exceed
its strength and take us too far from the original function or material con-
stitution of the artefact, the time has come to point out that it is not at
all obvious that the judgments in question always do demand to be read
literally as statements of identity. When someone gives his watch to the
watchmaker to clean and repair, the thing he wants back may, on a very
sober and literal minded construal, be either that very watch (by the
unproblematic criterion) or else a watch with a certain obvious relation
to his (a watch of the same kind, in better working order, enjoying con-
siderable community of parts, etc.). If he wants more than that, if he
thinks of his watch as an antiquarian or historical relic from a better age
or as a work from the hand of a great artist, then he should take more
precautions than we normally do take. He should care about its original
constitution. The truth is though that, for some practical purposes, we
simply do not mind very much about the difference between artefact sur-
vival and artefact replacement. (A negligence that in no way undermines
the real distinction between these.)
So much for how matters seem to stand with the normal artefacts
which, in everyday life and outside the museum, we freely permit to
27
It is not surprising in itself. But one or two of its consequences may surprise (in advance at least
of legitimate qualification). Where the repairs to a working ship are reaching their constitutive
limit, you will not always be able to tell by one careful look whether the ship in front of you is the
same as the ship you saw yesterday. There is moreover one conceivable circumstance where car-
rying a plank from one place to another and joining it to a reconstruction that is in progress from
discarded original planks will count as shifting Theseus’ ship itself from one place to the other.
 Sortal concepts
be repaired, altered and (without our realizing it very well) replaced.
But could there not be artefact concepts that were less permissive in
their definition (as required by D(iv) and D(v)), were more like natural
kind terms in the specificity of what they required of their instances,
and did not raise the problems that are raised by artefacts that are
subject to disassembly, part-replacement and discontinuance of func-
tion? The answer to this question is that there surely could be. True, I
have not been able to find any normal artefact concepts which require
of their exemplifications anything remotely analogous to unremitting
obedience to some specialized principle of activity that conditions the
sense of the artefactual substantive itself. There is, however, one very
special kind of artefact whose survival-requirement is extremely
detailed and specific. This is the work of art. We shall postpone the
special features of this and cognate concepts until the last section of
Chapter Four, where we shall maintain that the conceptual need
which a natural law will supply for the identity of a natural thing but
not for an ordinary artefact kind, the artist’s conception of his activity
and its eventual product can supply for the work of art. But works of
art and their like are in a special class of their own among artefacts,
and most especially perhaps for purposes of the theory of individua-
tion.
Rather than explore this or other analogies with the principle of activ-
ity of a natural substance (e.g. the analogy furnished by such social arte-
facts as an administration, or a governing body recruited and replenished by
a formal procedure) or seek to extend further the discussions begun in
this chapter, 28 it is time to provide an outline of what is supposed to have
been achieved in the first three chapters.

.             :  

        
⫽
(i) The formal properties of identity, both ‘⫽’ and ‘ ’ and ‘the same
f
something as’, are transitivity, reflexivity, symmetry, and the congruence
defined by Leibniz’s Law, although the dyadic relational predicates ‘is
the same as’ and ‘is, for some f, the same f as’ have a content that
amounts to more than could be explained, to one who lacked the
28
Since the time when this chapter was first laid out, the most important new contribution to these
questions of criteria of identity is that of Timothy Williamson, Identity and Discrimination (Oxford,
).
 Sortal concepts 
concept of identity altogether, by a bare rehearsal of these principles.29
A fortiori, the principles themselves are not dispensable.
(ii) If (i) above is true, then R is false.
(iii) There are two distinct standpoints from which D as stated at the
beginning of Chapter Two can be maintained. It may be held that a
could be the same f as b without being the same g. That is the philosoph-
ical contention R. Or it may be held that to say what a is is to subsume
a under a predicate which either gives a’s principle of continuity and
individuation – its principle of activity or its mode of functioning – or
else restricts some other predicate that gives this principle. This princi-
ple furnishes the wherewith to vindicate the absoluteness of identity.
29
See Chapter Two, §, and Chapter Six, §. Neither there nor in the sentence to which this foot-
note is affixed do I seek to controvert Quine’s claim that, for a given language, ‘we can settle
objectively and absolutely what predicate to count as the identity predicate, if any, once we have
settled what notations to count as quantifiers, variables and the truth functions . . . The require-
ments [on treating] an open sentence ‘␾ x y’ as ‘x ⫽y’ are strong reflexivity and substitutivity . . .
If ‘␾’ and ‘␺’ both meet [these] requirements . . ., then they are coextensive’. See ‘A Reply to
Professor Marcus’ in Ways of Paradox (New York, ), p. .
Quine’s claim itself is not to be gainsaid. But the two words I have italicized, namely ‘if any’,
are not unimportant. They need to be there, as Quine points out, because it may be that in a
given language no open sentence in ‘x’ and ‘y’ is strongly reflexive and substitutive. There is also
another reason why they have to be there. This too can be put in Quine’s words: ‘If the universe
is taken as that of persons, and the predicates are interpreted in ways depending on nothing but
people’s incomes then the proposed manner of defining ‘x⫽ y’ will equate any persons who have
equal incomes; so here indeed is an unfavourable case where ‘x⫽ y’ [defined in the manner indi-
cated here and in ‘Reply to Professor Marcus’] does not come out with the sense of genuine iden-
tity.’ (W. V. Quine (), §, p. .)
The first conclusion that I draw from the difficulty that Quine acknowledges is that one cannot
say perfectly generally that the sense of ‘⫽’ in a given language L is exhausted by the fact that
it is a reflexive relation and a congruence relation with respect to L-predicates. (Compare
Chapter Six, §.) The second thing I think I see here is support for something I have already con-
tended: namely that questions about identity and questions about kinds of thing are not clearly
extricable from one another. The indefinable primitive forms ‘x ⫽y’ and ‘x is an f ’ need one
another. Each must be present for the other to operate. See Preamble (ad fin.).
Whatever Quine might say about these conclusions, they are oblique to Quine’s interests in
the passage I have cited. He goes on to consider there the suggestion that matters might be
straightened out by reconstruing the members of the universe as whole income groups. (Cf.
(), p. .) He says that, whether or not we do that, ‘no discrepancies between [the surrogate
for identity] and genuine identity can be registered in terms of the theory itself ’. I don’t think
this last would furnish everything he needed to one who wanted to dismantle the identity rela-
tion into reflexiveness and congruence. But in the background is something else that is well worth
questioning: namely whether absolutely everything that grounds interpretation for a given lan-
guage must be such as to register in the explicit predicative repertoire of the language itself. It
would not suffice for these purposes to say (plausibly enough and in line with everything we have
said in Chapters Two and Three) that there is a process by which an interpreter’s implicit or
practical understanding of the role of language in the life and conduct of its speakers can be
made more and more explicit. For, first, the process may not terminate. Secondly, in so far as the
process does not terminate and the predicative resources and repertoire of the language are con-
stantly extended towards semantic saturation, the intended definition of surrogate identity given
by means of exhaustion of predicates will never quite be available. See further Chapter Six, §.
 Sortal concepts
(iv) The absoluteness just spoken of is vindicated as follows. A sub-
stance-predicate’s standing by virtue of its sense for a kind f as in (iii)
ensures this: that a’s coincidence with b under the concept f will imply
the community of all of a’s and b’s predicates. In so far as this require-
ment on the sense of substance predicates is satisfied, it will follow that
f-coincidence is not only an equivalence relation, but also a congruence
relation. Indeed, nothing less will suffice for a theory of individuation of
the continuants of our accepted ontology.
(v) The absoluteness requirement stated in (iv) is satisfied by concepts
introduced by sortal predicates whose definition is extension-involving
and deictic-nomological in the manner described by Leibniz, Putnam,
Kripke and others. These are natural kind terms. The requirement is not
so straightforwardly satisfied by ordinary artefact concepts, which are
not extension-involving or nomologically founded. Artefact kinds con-
front us with serious problems. Here some revision of what might appear
to be our current individuative practices or intuitions may seem to be
required. But one suspects that tacitly these practices respect already
some corresponding restriction. It may be that the most that is required
is some revision of distorted or ill-considered conceptions that accord
too little significance to the original matter of an artefact.
(vi) All this has been explained by talk of the coinciding of entities
under a concept. The locution ‘coincides’, rare though it is by compar-
ison with the other words and forms we are using it to elucidate, and
technical though my use of it may be, is not an invented one. Whenever
we have written ‘x coincides with y under the concept f ’, our sentence
could be rewritten (without loss of anything but the strictest synonymy
and occasional convenience) in the form ‘x coincides with y as an f – in
the manner, that is, of an f ’. The most this new version can do perhaps
is to prompt a reader who thinks he needs something more to remind
himself of what is there already, latent in his own irreducibly practical
understanding of ‘the same f ’. For all I know, ‘coincides’ could drop out
of the object language altogether if a better observed and more com-
plete account were achieved of what it is to say what a thing is. But,
pending that’s being achieved, it is by talk of coinciding that one may
hold substance, identity and individuation far enough apart to give a
view of the many strands by which they are connected. Note especially
here the relative perspicuity of the right-hand side of the schema by
which ‘coincides’ is explicated in § of Chapter Two. (See especially
there paragraph four.)
(vii) Neither in D as further elaborated nor in any other place has it
 Sortal concepts 
been supposed that, in the normal business of making identity judg-
ments, we deduce the judgment in question from other claims about
other properties and relations. Judgments of identity are sui generis judg-
ments. (They are refutable, no doubt, by reference to other properties
and relations but they are not deductively based on them.) In their own
way, they represent the prolongation and diversification of our powers
of carrying out the practical business of singling out and keeping track
of a thing. With this prolongation and diversification there comes into
being a dialectic of same and other. It is the regulation of this dialectic
of same and other by D principles, which themselves arise from the formal
properties of identity, that qualifies the dialectic as centrally concerned
at the levels of action and thought with the strict relations of identity and
difference. D principles give partial expression to practical and logical
norms for judging matters of persistence and non-persistence, identity
and difference. These norms articulate the Leibnizian conception of
identity that is stated in (i) (ii) (iii) (iv) and developed further in Chapter
Six. Not least among the D principles (and central to that which marks
identity off from other notions) is D(x), the Only a and b rule.

.               

It will be instructive to try to apply all this to persons, to works of art and
to other special cases of natural and artificial things. But, in advance of
all that, the view of individuation that has begun to emerge must be
brought into some relation with the larger questions of Realism,
Idealism and Conceptualism to which a theory of individuation is finally
answerable.
Call the philosopher who denies R and denies D the Bare Absolutist.
Call the philosopher who accepts R the Relativist or R-theorist. Call an
upholder of my position a D-Absolutist. Chapter Four draws out certain
of the individuative and essentialist consequences of D-Absolutism.
Chapter Five seeks to communicate my conviction that it is typical of the
R-theorist to exaggerate the autonomy of thought in the singling out of
objects of reference. In Chapter Five and then in Chapter Six, I try to
create the conviction that, by contrast with the R-theorists, the Bare
Absolutists, if they genuinely believe in continuants at all, are disposed
by their denial of D to take too lightly the conceptual preconditions of
singling out. I think of it as the role of the D-Absolutist to see the artic-
ulation of reality in a way that corrects both faults at once. The mind
(you may feel moved to say) conceptualizes objects. Yet objects impinge upon
 Sortal concepts
the mind.30 On the basis of this double claim, a new response will be
mustered to the problems of identity and vagueness. When that has been
done, the ground will have been cleared for an application of all these
thoughts to the case of persons and their identity.
30
One who defends such a thesis can scarcely fend off too soon certain misreadings of what he
means to be committed to. Perhaps ‘the mind conceptualizes objects’ will appear to carry the
suggestion that the very same objects could have been conceptualized in different ways, had the mind
been differently constituted. But, if so, then the suggestion that appears to be carried is hereby
cancelled. That is not what the sentence in the text means or implies. On these matters see
further Chapters Five and Six.
If the sentence to which this footnote is annexed still seems to conflict with the sentence pre-
ceding, then it is worth reflecting that for the mind to ‘conceptualize an object x’ is not for it to
construct x, or even for it to construe x (whatever that would mean). Rather (the person with) the
mind, making sense (so far as possible) of divers lookings and explorings and perceivings, being
ready to deploy clear indistinct ideas of sorts or kinds of thing (see note ), finds x – which is not
yet to say that x was the only thing there to be found. Almost always there will be other things
there to be found.
    

Individuative essentialism

Nature loves to hide.

(Heraclitus, Diels fragment .)

We despise obvious things, but unobvious things often follow from

obvious things.
(Leibniz to de Volder, Gerhardt , .)

 .                         

               
Little or no use has been made up to this point of the notion of neces-
sity. We have resisted the idea that a theory of individuation must be
a set of judgments about all possible worlds, or occupy itself with prob-
lems that are special to the making of statements of explicit necessity
de dicto or de re. In the case of the things that belong to natural kinds,
we went to some lengths to show that, for all the purposes of identity
and individuation that have concerned us so far, it is enough to have
regard for the lawlike propensities of members of the kind. It is
enough to look for true generalizations, open and counterfactual-
sustaining, about the past, present and future entities of this world.
(Enough and more than enough. Our practical understanding of
identity and individuation makes do with less.) It is true that, in con-
nexion with artefacts and at the point when the argument for the Only
a and b Rule was introduced, something modal appeared to be
involved. But in so far as modality obtruded itself there, it was conse-
quential upon the pretheoretical perception of an absurdity that
would be marked as such by almost anyone who reflected on the impli-
cations of the business of individuation.
Marking the end of modal abstinence, I now follow matters through
to the ‘essentialist’ consequences of the theory of individuation

 Individuative essentialism
expounded in the preceding chapters. Someone who finds these conse-
quences incredible will have the choice between criticizing the theory for
entailing them and criticizing the derivations now to be attempted. But
the final product will be an essentialism that is at once modest and (in
case that is thought to matter) austerely extensional. In the argument to
be offered here, which will not depend on the idea of reference to this
or that particular thing but on the necessary condition of that very
thing’s being singled out from the rest of reality or individuated (not,
however, on the suspect idea that these necessary conditions will
combine to form a distinguishing or particularizing essence of the thing),
any gaps that the reader finds should be criticized under the head of
argumentative insufficiency. Deficiencies should not be supplied from
principles or premisses offered by philosophies that aim for something
stronger. To the adherents of possibilia, substantive possible world theo-
ries or individual (particularized) essences, as to champions of essential-
isms that deliver easily or immediately the conceptual indispensability to
anything of its actual origin, the essentialism of particular substances
offered here may seem laughably modest or timid. But I am uncertain
of the rationale for maintaining anything stronger.

.                      

            
First an enumeration of the principles and assumptions ((A) – (H)) that
will govern the derivation of essentialism to be attempted in later sec-
tions. By a derivation nothing very formal will be intended. (A) – (H)
simply prepare the way for a dialectical exploration of that to which we
are committed by our simplest thoughts about continuants and our sim-
plest individuative practices:
(A) The first maxim is to dispense, in everything relating to the deri-
vation, with possibilia. Compare Arnauld
I have no conception of these purely possible [individual] substances, that is to
say the ones that God will never create . . . One can conceive of possibilities in
the natures which he has created . . . but I am much mistaken if there is anyone
who dares to say that he can conceive of a purely possible substance, for I am
convinced in my own mind that, although one talks so much of these purely
possible substances, nonetheless one never conceives of any of them except
according to the notion of those which God has created.1
11
Arnauld to Leibniz,  May , Gerhardt , –, as translated by H. T. Mason. Cf. W. V.
Quine, ‘On What There Is’ in his ().
 Individuative essentialism 
(B) The second principle is the requirement that the argument should
be independent of (and go deeper than) the model theory of possible
worlds.2 In so far as possible worlds are mentioned, they are intended
only as an expository device.
(⌫) The third principle or assumption I shall need is given succinctly
by A. J. Ayer:
We can significantly ask what properties it is necessary for something to possess
in order to be of such and such kind . . . the answer may be to some extent arbi-
trary, but at least some answer can be looked for. On the other hand, there is no
such definition of an individual.3
It is the consequence of this that, to make clear which thing thing x is, it
is not enough, pace the friends of individual or particularized essences
(by which I mean essences that it is supposed can be specified without
the use of identity yet determine an identity) to say however lengthily
that x is such and such or so and so. We have to say that x is this such and
such, or that so and so. This is perfectly obvious when we think of trying
to specify one particular entity by mentioning short or simple predicates
(other than identical with x or the like). Is there any reason to believe that
by making ordinary predicates ever longer and more complicated one
will overcome the obvious non-sufficiency or non-necessity for identity
with just x that infects all the relatively simple predicates true of x? Let
us be realistic then and forget about individual or particularized
essences.
(⌬) The fourth assumption or principle can again be given in alien
words, again those of Arnauld4 and Ayer (ibid.):
I can think that I shall or shall not take a particular journey, while remaining
very much assured that neither one nor the other will prevent my being myself.
So I remain very much assured that neither one nor the other is included in the
individual concept of myself.
[But] we have to maintain some anchorage in reality if our references are to
be successful . . . There appear to be no general rules for deciding what this
anchorage may be . . . One could imagine that the Pyramids were built at a
different time, or perhaps even in a different country. If one anchors Dickens to
12
See Williamson in Lovibond and Williams (), p.  and following.
13
The Central Questions of Philosophy (London, ), p. .
14
The source of the quotation is the same letter from Arnauld to Leibniz, Gerhardt , , trans-
lated by H. T. Mason. My citation of Arnauld should not be taken as tantamount to agreement
with him about the interpretation of Leibniz. Leibniz was not an essentialist. See my (a). It
would have clarified his purposes to distinguish always, as he sometimes did, between the
Leibnizian concept of an individual and the essence of that individual. Always, though, Leibniz
insists on the contingency of my going or not going on a journey (say).
 Individuative essentialism
other items in his biography, one can conceive of his not having been a writer:
if one identifies him by his writings one can perhaps conceive of his having lived
in a different century. But could we consistently place him in the distant future,
or in prehistoric times? . . . It is a rather arbitrary question in this sort of case.
Appearances perhaps to the contrary, one positive thing Ayer is claim-
ing here does not depend upon the theory of reference as such. Nor
does it depend (even if Ayer expresses the point in a way that invites
this charge of confusion) upon a denial that one can counterfactually
conceive at t of some individual x not having the properties by which,
for the purpose of conceiving at t of x’s being otherwise, one identifies
this x in the actual world. Rather the point is that, if serious conceiv-
ing is to have to do with real metaphysical possibility and it is excluded
that a given thing can be just anything, then we shall need to arrive at
an understanding of conceiving that yields the following conclusion–:
For any entity, there are properties or complexes of properties such
that nothing could count as conceiving of that entity as possessing
them; likewise properties or complexes of properties such that nothing
could count as conceiving of the entity as lacking them.5 It is among
the things brought to light by the Chapter Three discussion of D(x),
the Only a and b Rule, that one of the excluded conceivings is the con-
ceiving of a thing as not the very individual that it is. That point will
now be argued further, as will the exclusion of the conceiving of an
individual’s having a different principle of individuation from its actual
principle.
The positive point of agreement between Ayer and me may be set out
as follows then. Where a thinker A conceives of an individual x as ␾ (and
by moving from ‘refer’ to ‘conceive of ’ we can only make Ayer’s claim
more foundational), ␾ cannot assume just any value. There are restric-
tions on ␾ and they depend on which entity in particular the entity x is.
What we have in (⌬) is not a reduction or elimination of necessity and
possibility (that is not our purpose), but the following elucidation of pos-
sibility and necessity de re:
15
The idea of metaphysical or conceptual possibility invoked here is the dual of metaphysical or
conceptual necessity. (Which is not to say that the latter is our sole route of access to it.) The con-
ceptual necessity of things being thus or so must not be confused with or assimilated to the anal-
yticity of the sentence that says that things are thus or so. Nor must concepts be assimilated to
meanings understood a priori. The concept of a horse or tree is an a posteriori concept. The truths
involving horses or trees that are truths one has to be aware of in order to grasp fully the concept
horse or tree (i.e. in order to get an in any way adequate conception of the concept horse or tree) are
a posteriori truths. Those truths are semantically enabling for that kind of discourse. Here, in
saying this, it may be that we exceed any claim Ayer could even have contemplated approving.
(For more on these questions, see Preamble, §.)
 Individuative essentialism 
(i) x could be ␾ if and only if it is possible to conceive of x’s being ␾,
where the intended acceptation of the verb ‘conceive’ (⬇ ‘think
right through and in such a way as to count oneself as impeded by
any incoherences or contradictions’) is to be worked out in tandem
with the sober essentialism that is here and now to be excogitated
(see below, footnote );
(ii) x must be ␾ if and only if it is not possible to conceive of x being
not ␾;
(iii) The position of the boundary between what one can conceive of x
and what one cannot conceive of x depends on x, i.e. depends on
which thing or what thing the thing x actually is; the position of this
boundary may depend on a matter of degree, or appear to be some-
what arbitrary; in which case, it must be unsurprising that at least
some of the properties that it is definitely impossible (i.e. well
beyond the last point of possibility) to conceive of x’s not having are
rather unspecific.
(E) The next guiding principle for the construction to be attempted is
that the relational claim A conceives of x’s being ␾ or A conceives (␭z) (z is ␾) of
the entity x may be cleared of all extensionalist and nominalist suspicions
by viewing its connexion with A conceives that on the model of Quine’s
theory of the relation holding between believing of and believing that.6 The
relation we are concerned with is between the thinker A, the entity x
(however described), and an attribute ẑ (fz) or (␭z)(fz). The relation is
triadic. (Quine has shown how, at a certain price, the attribute which is its
third term can be nominalistically reconstrued, but nominalism is not the
issue here.) Similarly, ‘x is such that it is possible to conceive (␭z) (z is ␾) of
x’ represents a complex dyadic relation between x and this attribute.
(Z) Ordinary intuitive grammar suggests that ‘necessarily’ and ‘possi-
bly’ can either govern a complete sentence (as in sentences that submit
naturally to treatment within what Quine calls the second grade of
modal involvement, namely complete sentences lying wholly within the
scope of ‘necessarily’) or govern a simple or complex predicate. (Or
might they somehow govern the predicative tie by which one who makes
an essentialist attribution seeks to link the subject of his attribution and
some property? I shall not enter into the question here of whether,
strictly speaking, the ‘necessarily’ qualifies copula or predicate.) If it is
asked what is meant by the latter usage, which lies outside all grades of
16
See ‘Quantifiers and Propositional Attitudes’, Journal of Philosophy,  (), reprinted in Ways of
Paradox (New York, ), at pp. –. For an important amplification of the theory, see Brian
Loar, ‘Reference and Propositional Attitudes’, Philosophical Review (), p. .
 Individuative essentialism
modal involvement approved by Quine,7 then I reply first that this is
something the reader is already familiar with, under the verbal forms ‘x
can be ␾’, ‘x could be ␾’ and ‘it is possible for x to be ␾’; secondly, that
(⌬) above was already a partial elucidation of some such usage; thirdly,
that anyone who speaks English is already deeply committed to a whole
range of locutions that resemble this one in lying outside Quine’s
canon.8 For, if we supplant de re ‘possibly ␾’ and ‘necessarily ␾’ by
‘can/could (be) ␾’ and ‘must (be) ␾’ (‘cannot / could not help but (be)
␾’) we shall notice that, so far as logical form goes, they seem to have
formal affinities with the de re predicates of ability, obligation, capacity
or disposition, which people who have heard nothing of possible worlds
apply effortlessly to the actual things of the actual world when they say
that A can do otherwise than he is doing, that D could not help but strike
C, or that X, Y and Z are soluble or fragile or fusible.9 None of these
claims has a natural de dicto translation.
(H) I represent with the sign  the undifferentiated de re ‘must’ or
‘necessarily’ that I presume to underlie as a genus the various species of
de re concept. For the corresponding de re ‘can’ and ‘possibly’ I use the
sign POSS. NEC and POSS are interpreted here as working with predicates
abstracted in a manner now to be described.10 To form the expressions

17
See W. V. Quine, ‘Three Grades of Modal Involvement’, Proceedings of the XIth International Congress
of Philosophy. Brussels , Vol.  (Amsterdam: North Holland Publishing Co., ) reprinted
in Ways of Paradox (see footnote ), p. . See also ‘Reference and Modality’ in Quine (), st
edn (this essay was rewritten for the  edition).
18
Compare Ian Hacking () for an analysis which is amenable to the same semantical treatment
and delivers the same results as the present chapter seeks to establish.
The truth predicate is defined for a language containing this ‘necessarily’ in my ‘The De Re
“Must”: A Note on the Logical Form of Essentialist Claims’, and for a richer language containing
it by C. A. B. Peacocke in ‘Appendix to David Wiggins’ ‘The De Re “Must”’’, both in Truth and
Meaning: Essays in Semantics, ed. Gareth Evans and John McDowell (Oxford, ). See also my ().
For the idea of treating ‘necessarily’ as a modifier of predicates, or (which is not quite the
same) of the copula, see ‘The Identity of Propositions’, in Geach (), p. ; R. Cartwright,
‘Some Remarks on Essentialism’, Journal of Philosophy,  (); Wiggins (), p. ; R.
Stalnaker and R. Thomason, ‘Abstraction in First Order Modal Logic’, Theoria,  (); G. E.
Hughes and M. J. Cresswell, An Introduction to Modal Logic (London, ), note ; John Woods
‘Essentialism, Self-Identity and Quantifying In’, in Identity and Individuation, ed. Milton K. Munitz
(New York, ); Wiggins (). See also Christopher Kirwan, ‘How Strong are the Objections
to Essence?’, Proceedings of the Aristotelian Society,  (–); Tyler Burge, ‘Belief De Re’, Journal of
Philosophy (). For the historic sources of this view of the matter, see for instance Aristotle, Prior
Analytics a15–23, De Interpretatione Ib26ff; De Sophisticis Elenchis a23–3a.
19
See also pp. – of Wiggins ().
10
The decision to signal these occurrences specially by this particular notation begins without prej-
udice either for or against the appropriateness of ⵧ and 䉫 to this role. The decision leaves it
open whether the pairs NEC and ⵧ and POSS and 䉫 will prove to be related as an ordinary truth-
definition for open and closed sentences involving the logical constants relates (say) predicate-
negation and sentence-negation.
 Individuative essentialism 
to which NEC and POSS are to apply we take any open sentence, say ‘x is
a man’ or ‘if x is a man then x has the property G’ or ‘x is identical with
y’, and bind the free variable or free variables in the open sentence with
an abstraction operator ␭. Thus: (␭x)[Man x] or (␭x)[Man x →Gx] or
(␭x)(␭y) [x⫽y]. These abstracts may be read ‘the property that any x has
just if x is a man’, ‘the property that any x has just if, if x is a man, then
x has property G’, ‘the relation in which any x and y stand just if x is y’
respectively.
We may now express the judgment that an entity or a sequence of
entities 〈. . .〉 falls in the extension of some property or relation so desig-
nated as follows:
Caesar has (␭x) [Man x]
Everything has the property (␭x) [M(x) →G(x)]
〈the evening star, the morning star〉 have (␭x) (␭y) [x⫽y],
or, more handily and conventionally, we may exercise the option to
employ a simple juxtaposition of the ␭-term and a subject term, to be
marked by angle brackets, thus:
(␭x) [Man x], 〈Caesar〉
(᭙z) ((␭x) [M(x)→G(x)], 〈z〉)
(␭x) (␭y) [x⫽y], 〈the evening star, the morning star〉.
So far these are mere ␭-equivalents of simpler sentences. But putting
   or     to work on these abstracts, and leaving the subject term
incontrovertibly outside the scope of the modality, we have
[ (␭x) [Mx]], 〈Caesar〉
(᭙z) ([ (␭x) [M(x)→G(x)]], 〈z〉)
[ (␭x) (␭y) [x⫽y]], 〈the evening star, the morning star〉.
The last says that the evening star and the morning star necessarily have
the relation that any x and y have if and only if they are identical. It is a
second level sentence introduced by means of ␭-abstraction. More col-
loquially and idiomatically, it says that it is necessary for the evening star
and morning star to be identical. Informally, there is much use of such
versions in this book. Indeed I would emphasize that the modal enrich-
ment of the first order logical framework by modality and second level
notions is explained in the English vernacular. Once the second level
framework is in place, however, the official explication runs in the lan-
guage of properties and the having of properties – and predicate letters
take on the official role of variables. In this way things are managed in
such a way that the de re ‘necessarily’ starts life with credentials no worse
 Individuative essentialism
than those of Quine’s relational senses of saying and believing (already
referred to in (E) above). Just as Ralph believes of the man in the brown
hat (or of the man seen at the beach; it makes no difference, for they are
identical) that he has the property or attribute ␭x (x is a spy), or just as
the man in the brown hat (alias the man at the beach) is universally
believed to have the attribute spyhood – so, in this prima facie innocuous
sense of ‘necessarily’, Caesar (or the conqueror of Gaul, or the consul
of  , it makes no difference how you refer to him and it makes no
difference what philosophical account is given of definite descriptions)
is necessarily such that he has the attribute man. Following the parallel,
we can say that the relational ‘[NEC (␭x) [Mx]], 〈Caesar〉’ stands to
notional ‘ⵧ (Man (Caesar))’ as the relational ‘Of Caesar is universally
believed (␭x) [Man x]’ (⬇ ‘It is universally believed of Caesar that he is
a man’) stands to the notional ‘It is universally believed that (Man
(Caesar))’. Compare (E) above.
At the point where the ␭-abstraction operator first appears, let one
example and five equivalents illustrate the two-way rule under which it
operates:
(a) Caesar is a man
(b) Caesar has the property of being a man
(a⬘) Man (Caesar)
(b⬘) (␭x) (Man x), <Caesar>
(c) the value for the argument Caesar of the function from objects to truth-
values (␭x) (Man x) is the truth-value true.
English permits the two-way transition between (a) and (b). The ␭-
abstraction rule permits a similar two-way transition between (a⬘) and (b⬘).

.                        

   
Taking Leibniz’s Law in a second-level form
For all F such that F is a genuine property of x or y,
(x⫽y)→(x has F⬅y has F)
and making the assumption (against which no argument from opacity or
unintelligibility now presses) that NEC properties are genuine derived
properties, let us begin by obtaining a proof of the complete coincidence
of the necessary satisfaction and the actual satisfaction of the predicate
of identity. In other words, let us try to construct the NEC counterpart of
the quantified modal theorem, sometimes controverted:
 Individuative essentialism 
(᭙x) (᭙y) ((x ⫽y) →ⵧ (x ⫽y)).
Consider any individuals H and P (Hesperus and Phosphorus say) such
that
H ⫽P.
Now H is necessarily H, so the following is a truth about H:
[NEC ␭x ␭y (x ⫽y)], 〈H,H〉;
and, the scope of this  being confined to the abstract, there then
exists the abstractable property of being necessarily identical with H or
(␭z) [[NEC ␭x ␭y (x⫽y)], 〈z,H〉].
But H itself has this property, and whatever property H has P has. So
(␭z) [NEC [␭x ␭y (x ⫽y)], 〈H,z〉], 〈P〉.
Then by the two-way rule of ␭-abstraction, we have
[NEC [␭x ␭y (x⫽ y)]], 〈H,P〉.
But now, since nothing depended on the particular choice of the H and
P which were such that H⫽P, we have the necessity of identity:
(᭙x) (᭙y) ((x ⫽y)→ [NEC [␭x ␭y (x ⫽y)]], 〈x,y〉)).
This proof adapts a famous proof of the necessity of identity which
was given by Ruth Barcan Marcus in .11 Its merit when given in this
form is that it makes evident that all substitutions within the Barcan
proof can be made on manifestly extensional positions, lying outside the
scope of ‘necessarily’.12 (To show this is the chief point of deploying ␭-
abstraction here.)
Miss Barcan’s proof was long received with incredulity by those
committed to the mutual assimilation (much criticized in more recent
times by Kripke and others) of the categories of necessity and a priority,
and rejected on the grounds that the identity of evening and morning
star was an a posteriori discovery. But even if statement ascertainable a priori

11
See Barcan (), Wiggins (). For an exemplary restatement of the Barcan proof, with a rea-
soned review of all the formal and philosophical issues, see Williamson ().
12
This holds of both the deduction and the truth definition required for that minimal fragment of
first order logic which adjoins  to abstracts of the form [␭x ␭y (x⫽ y)] in the object language.
In the metalanguage the additional requirement is to adjoin  to abstracts of the form
[␭x ␭y (x satisfies y)]
 Individuative essentialism
to be true and necessarily true statement coincided perfectly in their exten-
sions, Miss Barcan’s theorem could still stand in our version. For the
conclusion is not put forward here as a necessarily true statement. (On
this we remain mute.) It is put forward as a true statement of de re neces-
sity. The thing that the proof comes down to is simply this: Hesperus
is necessarily Hesperus, so, if Phosphorus is Hesperus, Phosphorus is
necessarily Hesperus. The only conceivable point left to argue is
whether there is a de re use of ‘must’ in English. But the onus is on the
contingency theorist and the anti-essentialist at last. The latter has to
dispel as illusion what seems to be fact – that in English there exist
many such de re uses.
That is not all. By showing that all identicals are in fact  identicals,
the proof will furnish one other thing to anyone who is party to a natural
intuition: the intuition that designations of things that are necessarily
identical ought to be intersubstitutable within the scope of ‘necessarily’.
One who is party to that intuition no longer needs any act of faith to
believe that the ordinary modal logic of ⵧ and 䉫, permitting quantifi-
cation into the ‘necessarily’ and ‘possibly’ contexts represented as ⵧ and
䉫 contexts, can give blameless expression to de re necessities. The 
identity of all actual identicals revindicates this as an entirely reasonable
belief. And, on this basis, we need not demur at the necessity of identi-
cals given in the familiar form
H ⫽P→ⵧ (H ⫽P).
Advancing from this point, one may argue for something more con-
troversial, namely the necessary distinctness of things that are actually
distinct. Improving on earlier derivations that invoked stronger princi-
ples, Kripke gave a proof of this on the sole basis of the necessity of
identity and the so-called Brouwersche principle
not A→ ⵧ (not ⵧ A),
which is equivalent to
A →ⵧ 䉫 A.13
Once one is given the Brouwersche principle and its / counter-
part in the lambda notation, the  difference of actually different
13
A model theoretic vindication of this formula would depend on the claim that accessibility
between worlds is symmetrical. For discussion of this see my p. , note , in Lovibond and
Williams () and the paper by Williamson in the same volume. Williamson offers a more
general discussion, of the Brouwersche axiom and of whether it is indispensable to the proof of
the necessity of difference.
 Individuative essentialism 
things is readily available. But, in the light of the revindication of the ⵧ
and 䉫 notation now afforded by the  version of the proof of the
necessity of identity, the reader may be spared the vexation of following
Kripke’s proof in the  notation.
When stated in the ⵧ 䉫 notation, Kripke’s proof may be given as
follows (see Williamson, op. cit.). Begin with the necessity of identity as in
Barcan and Kripke,
(H ⫽P) →ⵧ (H ⫽P).
Contraposing this, we have
Not ⵧ (H ⫽P)→Not (H ⫽P).
Being proven from a theorem, this is a theorem. So we have the neces-
sity of it:
ⵧ (Not ⵧ (H⫽P) → Not (H ⫽P)).
And then, by the modal principle ⵧ (A →B) →(ⵧA → ⵧB), we have
ⵧ (Not ⵧ (H ⫽P))→ⵧ (Not (H ⫽P)).
But by the Brouwersche principle, we have
Not (H ⫽P)→ ⵧ (Not ⵧ (H⫽P)).
In which case, by virtue of the last two formulae and the transitivity of
→, we have
Not (H ⫽P)→ⵧ (Not (H ⫽P)).
In other words, if Hesperus is different from Phosphorus, then it is nec-
essary that Hesperus be different from Phosphorus. And the same for all
other cases. If Venus is distinct from Mars (and it is), then Venus is nec-
essarily distinct from Mars.
Such formal results as this help to confirm and consolidate a meta-
physical conception which the necessity of identity alone might almost
have sufficed to vindicate. According to this conception, facts about the
identity and difference of individuals are part of the necessary structure
of reality and are completely invariant across possible worlds. (See
Williamson ().) It is simply not an option for us even to describe or
envisage possible worlds in a way that would conflict with the identities
and distinctnesses that actually hold. Rather, actual identities and differ-
ences have to condition the specification of any world whatever that
counts as a candidate to contain actual things – things of the real world.
 Individuative essentialism
Moreover (again as Kripke has emphasized), the identities of the things
within a possible world are specified along with their other properties or
relations.

.       ,        

So much for the first step towards the modest essentialism that was
promised. It remains to be seen whether, within the apparently immo-
bile framework that these theorems discover to us, there is anything
interestingly essential to establish about this or that given individual. In
this search let us begin gently, however, mindful of all the doubts that are
still provoked by the very idea of essence. And let us consider first a
thought that is drawn from an area where extensionality reigns abso-
lutely, namely set theory. This thought is most aptly expressed in our
NEC notation, because that is better suited than is the ⵧ 䉫 notation to
mark a certain asymmetry that appears immediately here between sets
and the objects that are members of them:
(᭙x) (᭙␣) (x 苸 ␣ → NEC [(␭z) [x苸z]], 〈␣〉).
This claims that, even if an object might have belonged to different sets,
any set that contains an object contains it necessarily.14 In this way,
expression is given to an idea (not in practice completely incontrovert-
ible, but our chief interest in it is exempli gratia) which lurks unacknowl-
edged in the explanations that certain friends of extensionality who are
also set theorists give of the identification and individuation of classes,
as these are conceived in mathematics.
‘What sets attributes apart from classes is merely that, whereas classes are iden-
tical when they have the same members, attributes may be distinct even though
present in all and only the same things.’15

14
On asymmetrical formulae such as these, see my () and () cited at note ; also Kit Fine,
‘Essence and Modality’, Philosophical Perspectives, Vol. , ed. J. Tomberlin (Atascadero, Calif.,
Ridgeview Publishing Co., ); also (a work in which my grasp slipped of the point stressed in
footnote  of Chapter Three) ‘Mereological Essentialism, Asymmetrical Essential Dependence,
and the Nature of Continuants’, Grazer Philosophische Studien /: Festschrift for Roderick Chisholm
(), –.
15
Quine (), p. . (Cf. Quine (), p. : ‘Classes are like attributes except for their iden-
tity conditions.’) The reader is urged to focus for a moment on the ‘may’ in the sentence I have
cited in the main text. For another use of the same ‘may’, see p.  of Goodman’s ‘A World of
Individuals’ (Problems and Projects ()), ‘While a class of individuals is uniquely a class of just
those members, a whole made up of individuals may also be made up of quite other parts.’
Neither of these is an epistemological ‘may’. Nor yet is either of them de dicto. Do they exem-
plify any grade of modal involvement officially approved by Quine?
 Individuative essentialism 
‘We use the word ‘set’ in such a way that a set is completely determined when
its members are given.’16
Suppose that we try to apply these criteria, and we are invited to think
of a thing ␣ simply identified as the entity (whether class or attribute we
do not yet know) to which there belong the items x and y and only these.
Then it seems that, if we are to envisage ␣ for what it is, the question we
have to ask is whether ␣, the very thing ␣, could have dispensed with the
particular entities x and y. If it could – if ␣ could lack x or could lack y
– then ␣ is not a set or a class.17 That is the thought that is suggested by
Quine’s distinction of classes and attributes, and suggested equally by
received justifications of the axiom of extensionality given in terms of
membership determining set-identity.
The idea that motivates this way of talking is that, whereas there is no
constitutive connexion between being this or that property and any
actual extension, or between being this or that man and some actual par-
ticular position in space–time, there is such a connexion between the
membership and the identity of a set. The thought that is needed – and
I should hope by now to have cleared it of the suspicions expressed by
Quine in ‘Reference and Modality’18 – is the de re modal thought that, if
␣ is a class containing x and y, then ␣ could not have lacked x.19 Set
16
Patrick Suppes, Introduction to Logic (Princeton, N.J., ), p. .
17
Compare Kit Fine, ‘First Order Modal Theories I – Sets’ Nous,  (), pp. –.
18
In pursuit of a different purpose, one critical in effect of the ‘may’ we seem to have found (note
) even in texts of Goodman and Quine, Cartwright has expressed a reservation that is cognate
with Quine’s:
(x) (y) (z) ((z苸x ↔ z苸 y) →x ⫽y ) . . . will be a theorem of any pure theory of classes but pre-
sumably of no pure theory of attributes . . . but it is a difference between theories, and I know
of no coherent way in which it can be supposed to carry over to the objects dealt with in the
theories. It does, of course, reflect a difference between the concepts class and attribute.
Richard Cartwright: ‘Class and Attribute’, Nous,  (). (See also R. Sharvey, ‘Why a Class
Can’t Change Its Members’, Nous,  ().) My claim is that the formal proposals that have been
sketched in § save from Quine’s doubts and Cartwright’s doubts not only the essentialist ‘must’,
‘may’, ‘can’, ‘could’, ‘able/unable’ and not only the usage into which Quine and Goodman
themselves seem sometimes to lapse, but also the kind of language cited in note .
19
Suppose someone doubted the necessity of the membership relation. Then the question would
be whether he could combine his doubt with any argued or principled affirmation of extension-
ality, or advance on behalf of extensionality such claims as ‘a set is nothing more than a unity
constituted by its members’ (Richmond Thomason, Symbolic Logic (London, ), p. ). If
there is no other way of identifying such a unity than via its constituents, and such a unity can
be given simply by giving all its members, then is not its identity derivative from these, and deriv-
ative in a way in which the identity of a perceptibly demonstrable horse or tree is not derivative
from that of any particular cells or derivative from the identity of any particular sequence of
spatio-temporal positions or particular sequence of paired space–time positions and material
components? Against those not convinced, here is a final or Parthian shot: try conceiving the
unit set of x, namely {x}, as the unit set of y where y苷x. Now try conceiving the pair set {x,y}
as lacking either x or y. . . .
 Individuative essentialism
theorists who say that it is a peculiarity of sets to be determined by their
members, or who distinguish sets from attributes in Quine’s way, are
surely saying at least this. And it need occasion no surprise that, in order
to delimit an area within which extensionality will reign, in order to
justify intuitively the axioms of extensionality for the mathematicians’
world of sets and exclude attributes from the intended interpretation of
Greek letter variables, one has to trespass for a moment outside the
delimited area and talk in a language of richer expressive resources.
Suppose so much can be accepted. Then, given (⌬) of section , one
might say that the pair set {Eiffel Tower, Crystal Palace} is essentially a
set, and essentially a set with just these members, because nothing could
count as envisaging that very set in a way that implied that it was not a
set, or that it lacked these members.
At risk of recycling the obvious but warned by experience that there
is often no alternative, I stress that the claim under discussion is not that
it is impossible to envisage such sets as {Eiffel Tower, Crystal Palace}
under any other description than ‘pair set whose sole members are Eiffel
Tower and Crystal Palace’. One might conceive of this set under the
description ‘the pair set whose members are in Z’s opinion the most
remarkable works of nineteenth-century engineering in the capital cities
London and Paris’, or in indefinitely many other ways. To show that the
set itself, the very one we are concerned with, could lack Eiffel Tower,
one must envisage that set as lacking Eiffel Tower. Once we tried to do
this, the question would be what it was we could succeed in envisaging as
lacking the Eiffel Tower.
It may seem for a moment that one can envisage anything of any-
thing, even ‘lacking Eiffel Tower’ of {Eiffel Tower, Crystal Palace}, pro-
vided that the identification the envisager starts off with is as vague as, say,
‘entity mentioned or had in mind at t by F. Hausdorff’. But that is the
same mistake in another guise. Starting off with so vague a description
of the thing mentioned by Hausdorff should not make it easier but
harder for the would-be envisager to be sure that he has, in the serious
sense, conceived of that entity under the description ‘lacking the Eiffel
Tower’. The sort of possibility we are interested in should not be such as
to be augmented by ignorance.20 The same goes for the necessity and pos-
sibility that Ayer was discussing in his own enunciation of principle (⌬).
The conceiving must be with regard to some determinate thing that
answers to some specification that identifies it, however contingently, as
20
Cf. Arnauld’s Objection ‘De Natura Mentis Humanae’, against Descartes, at p. , vol.  of
Oeuvres de Descartes, ed. Adam and Tannery (Paris, –).
 Individuative essentialism 
that particular thing. Then the question must be how this thing can be
counterfactually envisaged; and what it is possible to envisage of it must
depend on what the particular entity is, whether Dickens, the Pyramids,
or whatever. The more you know about the thing, the more authorita-
tive your findings of possibility and impossibility ought to be.

.             

A handy way of summing up (⌬) (E) and (Z) as they combine with the
conception of individuation just illustrated is this: x could have the prop-
erty ␾, or it is possible for x to have ␾,21 if and only if it is genuinely pos-
sible to conceive of x’s having ␾; and a thinker genuinely conceives of x
as having ␾ only if there is some sortal concept f such that: (i) f adequately
answers the Aristotelian question what x is, and commits anyone who
singles a thing out as an instantiation of f to an identity-cum-persistence
condition for x; (ii) f and ␾ are cosatisfiable by x, and if x had the prop-
erty ␾ that would not preclude x’s being singled out as this very instan-
tiation of f.22 Let us call this the anchor constraint. It may be protested
that it does not render it an entirely clear or non-disputable matter
whether x could or could not have had the property ␾. But this is no
ground for complaint. For often it really isn’t an open or shut question.
(The contestability in question echoes that of questions to be explored
in Chapter Seven.) The role of philosophy here is not to render hard
questions easier, but to cast light upon what is at issue. This anchor con-
straint will be further elaborated just once, at the beginning of §. But
it will not be amended or modified.
21
See Hacking (), for this variant. For the distinction of could and can, which could be made
explicit here by the identification of a place for a relative to what supposition parameter (‘could’ rep-
resenting here the bare supposition that x exists), see e.g. S. A. Handford, The Latin Subjunctive,
pp. f. (cited by J. L. Austin at Collected Papers (Oxford, ), p. ) and Alan R. White, Modal
Thinking (Oxford, ), Ch. Two. See also my op. cit. in Lovibond and Williams (), p. . In
the language of that piece let me say that I am focusing here on the flat sense of ‘could’ and the
correspondingly exigent dual sense of ‘must’.
22
What would refute this contention and its expansion at § below? A weaker requirement that
arose equally naturally from a theory of particular identification, respected the absoluteness of
identity, and represented the substantial requirement that (⌬) puts upon the de re relation of con-
ceiver, object and attribute. The case for this requirement is that it makes sense of our more
reflective practice when we make suppositions. (Our practice, I mean, where it does not idle or
substitute fantasy for the careful conceiving, the careful thinking right through, of the alterna-
tives to the actual.)
It follows that the proper response to the proposal is dialectical: to propose a weaker or differ-
ent principle that promises the same sort of friction or impedance upon claims of possibility that
are idle or baseless. This is one part of the reply to Penelope Mackie (). For another part,
see Preface, paragraph . See also note .
 Individuative essentialism
The interest of this elucidation lies not in a claim to be an analysis or
reduction of de re possibility – it is multiply disqualified from those tasks
by its re-use of the modal concepts and the importing of a notional
thinker to do the conceiving – but in the stringency of the necessary con-
dition that it proposes for de re possibility. It illustrates how there might
be such a thing as a de re necessity to which a concrete particular was
subject. And it sees the issue of de re necessity as all of a piece with the
practical business of individuation.
According to this elucidation, it is possible, for all that has been
shown so far, to conceive of Caesar’s having a different career; it is pos-
sible to conceive of this very man’s not being consul in  , or his
not conquering Gaul, or his not crossing the Rubicon, or even of his
not in fact being male or not living through adolescence. Maybe it is
even possible to conceive of Caesar’s not having his actual parents. If
there is a difficulty about that, then we shall come to it in due course.23
What then do the constraints so far framed suggest that it is properly
impossible to envisage of Caesar? The this such conception of individ-
uation and the paradigm of the previous section, combining in the elu-
cidation now proposed, suggest that among the things it is hardest to
conceive of Caesar’s not being is human being. For if anything plays
here the part that we found the concept set or class to play in our spe-
cimen argument, then the concept human being plays it. Or so it would
appear.
I allow that there is one doubt even about this restriction. Someone
might concede to me that there must be some sortal concept f such that,
whatever else Caesar is envisaged as being, no genuine envisaging of
Caesar excludes his being this instantiator of f. But, despite conceding
that, the questioner might wonder whether there was a less specific
sortal concept than human being that was capable of fixing the persistence
conditions of this entity and answering the question ‘what is Caesar?’
Could not the generic property animal serve in the role of f ? (See §; and
see Chapter Seven for the other candidate, namely person.) It may be said
that all that Caesar needs is to be such and such an instantiator of
animal.
This is by no means the only obscurity or perplexity to be encoun-
tered in adapting the {Eiffel Tower, Crystal Palace} argument to the
23
For Kripke’s argument for the opposite conclusion see below, §. A word is needed here on the
status in my treatment of claims that something is de re possible. They are by their nature provi-
sional, and they are defeasible by further arguments why the property in question cannot after
all be conceived of some object. Chapter Six, §7, bears on the reasons for this.
 Individuative essentialism 
things to be encountered in nature. But before we prune back the doc-
trine it must be allowed to put on more leaf.24

.                  

Suppose that every natural thing x satisfies throughout its actual exis-
tence some sortal concept that any thinker who singles out x has to treat
as invariant (cf. D(ii)), and has to treat as invariant even when engaged
in conditional and counterfactual envisagings of states and histories
alternative to x’s actual states and history; and suppose that this concept
is a natural kind concept which not only says what x is but also leads a
posteriori to a passable criterion of identity and persistence for members
of its extension.25 Then not only will such sortal concepts be owed the
kind of honour once upon a time accorded by Aristotle and by Leibniz
to the infima species (cf. Aristotle, Metaphysics Z; Leibniz, Nouveaux Essais ,
.; , .), provided only that they really do represent the least spe-
cific account of their bearers that will suffice to articulate these very
bearers from the rest of reality. It will also be possible to derive some
potentially surprising a posteriori consequences.
Suppose that ‘human being’ (and every other natural kind term,
whether species or genus) has its sense fixed by reference to some
hypothesized generic constitution, a constitution which users of the
term are committed, despite their ignorance concerning it, to think of
as exemplified (under the physical laws of the actual world) by the actual
instances that they encounter and group together as men. (Suppose the
simple fact of their use of the term commits them, as claimed in
Chapter Three.) And suppose G is some predicate standing for an
abstract property, e.g. one pertaining to the gene pool perhaps (albeit
not only to that, see Chapter Three, §), and scientifically partially
definitive of that constitution and/or determined by that constitution.
Now consider anything that is a human being. Surely he or she is nec-
essarily if-a-human-then-G. It may be objected that the conditional gov-
erned by this ‘necessarily’ is no genuine necessity, because it is
24
There are anti-essentialists of a moderate persuasion who might let pass the set-theoretic argu-
ment of Chapter Four, section , on the grounds that sets are abstract things and not identifiable
independently of a mathematical theory, but resist strenuously the extension I have attempted
to material substances. It seems dubious, however, that conceptual-cum-individuative depen-
dence on a mathematical theory is the real principle of the distinction between abstract and con-
crete. Witness the entities of modern physics.
25
I find my arguments about this anticipated in part by A. MacIntyre’s article ‘Essence and
Existence’ ().
 Individuative essentialism
contingent upon the laws of nature and lawlike propensities that actu-
ally obtain and other laws and propensities might have obtained. But
already Chapter Three has confronted that. If the concepts of man and
human being behave in a manner consonant with the deictic explications
of ‘man’ and ‘human being’ envisaged in that chapter, then the very
existence of such things as human beings is contingent on that same
thing. For it is on the same actual laws that the delimitation and signif-
icance of the human being-definitive constitution depends.26 Since the
G-property and human being depend for their instantiation on the very
same circumstance, and nothing counts as human entirely indepen-
dently (as it has now appeared) of possessing the G-property, it follows
that a human being is necessarily if human then G. Moreover, once that
is established, we can advance beyond this merely conditional property.
We already have the beginnings of an argument to show that whatever
is a human being is necessarily a human being. If that is right (cf.
Preamble, §, on ‘individuate’; Chapter One, mottoes and §; Chapter
Four, § and in due course §), then we can necessitate the consequent
of the conditional. It will be true a posteriori that anything that is a human
being is necessarily G.27
26
See again Chapter  of Ishiguro (). Our concern is with the general form of this kind of
argument. Once that was accepted, the time would come to enter into the differences between
(a) Aristotelian infimae species; (b) the kinds furnished by zoology or other sciences, which need not
perhaps be depended upon to do from their own resources the right sort of justice in the right
sort of terms to the sort of differences we might want there to be between a St Bernard dog and
a Pekingese; and (c) the individuative kinds (not necessarily articulated in a zoological way or
another scientific way) that are promoted by our construction in this chapter. On these matters
see P. F. Strawson’s Critical Notice of Sameness & Substance (), Mind,  (), pp. –.
27
The argument relies of course on the  counterpart of ⵧ (p→ q)→(ⵧ p→ ⵧ q). See p. 
of Wiggins ().
It is my hope that in the account here employed of the de re ‘must’ and ‘can’ there resides some
justification for modes of expression that are very natural to some scientists, and possibly indis-
pensable to the attempt to describe how scientific theory explains what certain things are. For a
passage susceptible of this interpretation and best explained, not by an implausible transposi-
tion into de dicto terminology, but by de re necessity backed with a notion of conceiving not open
to the charge that the less you know the more you can conceive, consider Erwin Schrödinger’s
explanation, at p.  of What is Life? The Physical Aspect of the Living Cell (Cambridge, ), of the
Heitler–London quantum theory of the chemical bond:
a number of atomic nuclei, including their bodyguards of electrons, when they find them-
selves close to each other, forming ‘a system’, are unable by their very nature to adopt any
arbitrary configuration we might think of. Their very nature leaves them only a very numer-
ous but discrete series of ‘states’ to choose from . . . The transition from one of these config-
urations to another is a quantum jump. If the second has the greater energy (‘is a higher
level’), the system must be supplied from outside with at least the difference of the two ener-
gies to make the transition possible . . . the molecule will of necessity have a certain stability,
the configuration cannot change unless at least the energy difference, necessary to ‘lift’ it to
the next higher level, is supplied from outside.
 Individuative essentialism 
It is sometimes complained that essences ‘explain nothing’. But if the
foregoing argument is correct then, apart from the least interesting and
least controversial necessary properties (e.g. identity with Caesar, which is
too special, and self-identity, which is too general), they do have a philo-
sophical function. For, first, there are predicates that stand for essential
properties of a thing and register the condition whose satisfaction is a
direct or indirect prerequisite of the very thing’s being articulated at all
from the rest of reality. These predicates are not in the business of
explaining anything much, because they are presupposed to there being
anything to explain. But then, in the second place, there are more inter-
esting predicates, like G above. These correspond to a posteriori or scien-
tific accounts given by some (at the appropriate level) relatively
fundamental theory of the natural kind or articulative sort to which a
thing belongs, a theory whose non-availability even in principle would subvert
the possibility of a sound deictic demonstration of the particular kind or
articulative sort presupposed to the thing’s being what it substantially is.
It is to the satisfaction of this complex requirement that such properties
as the G-property owe both their place in the theory of individuation
and their necessary application to human beings.
We may add to the maxims of § a principle long since acknowledged
in every interesting or thoughtful essentialism of natural substances.
(⌰) The practical scientist does the business; but the philosopher keeps the
books. (Nelson Goodman, Problems and Projects (Indianapolis, ), p. .)
I agree with those who deny that forms are to be used in seeking to explain
specific and special causes. (Leibniz, Specimen Dynamicum.)28

.      ‘  ’            

          
Particular individuals have essences without which they would not ‘be
what they are’ – would not exist. But apart from logically particularized

28
Leibnizens Mathematische Schriften, ed. C. I. Gerhardt (Berlin and Halle, –), Vol. , p. .
Cf. Discourse of Metaphysics, Ch.  (Gerhardt, Philosophischen Schriften, Vol. ). ‘I agree that consid-
eration of these forms is of no service in the detail of natural philosophy, and must not be used
for explaining phenomena in particular. And it was in this that our scholastics failed, and the
Physicians of past times following their example, believing that they could account for the prop-
erties of bodies by mentioning forms and qualities without going to the pains of examining the
manner of operation; as if one were willing to content oneself with saying that a clock has the
horodictic quality resulting from its form, without considering in what all this consists . . . But
this failure and misuse of forms must not make us reject something knowledge of which is so
necessary in Metaphysics.’ (Translation by Lucas and Grint.) Cf. also Chapter  ibid.
 Individuative essentialism
essential properties such as necessarily identical with Caesar, which are spec-
ified through a designation of the thing in which they inhere, we must
expect the de re essences of individuals to be shared or shareable in prin-
ciple. The requirement that essences determined otherwise than
through identity itself should be unique to particulars (like almost any
other attempt either to say anything or to deny anything by means of the
idea of haecceity or this-ness) is the product of confusion. We get from the
‘this’ of ‘this f ’ (where f represents some sortal specification) all the par-
ticularity that is required for anchorage to the actual entity of which we
are to conceive in various counterfactual ways. Whether or not Strawson
has demonstrated the necessity for thought as we know it of individuals,
he would appear to have shown at least this: that, wherever thought does
recognize individuals, the functions of ‘this’ and ‘such’ are in the end
(regardless of any limited interchangeability of role) mutually irredu-
cible. (Cf. Chapter Six, §.) It is true that Russell’s Theory of
Descriptions may be seen as a proposal for redrawing one apparent fron-
tier between predicating and designating. But whatever we make of that
theory, naming or designating (which is a linguistic function that is entity-
involving in a way in which predicating need not be) and predicating are
two functions. So far, we lack any reason to believe in the existence of
any haecceitas that is defined or manufactured otherwise than via desig-
nation of the owner of the property in question. It is hard to think of
anything true and significant that could be said using the idea of this-
ness (which-ness?) not better said while respecting the distinctions desig-
nation/predication and particular/universal. It is harder still to imagine an
approach to the identity of a thing that makes its explanatory beginning
with haecceitas.
Conspicuous among the non-particularized properties of Caesar that
pass the test of being invariant under all counterfactual speculations that
can count as successful de re conceivings of Gaius Julius Caesar, we must
expect to find properties with negations not coconceivable with his sortal
identification human being (or, more weakly, not coconceivable with
animal, see below, §). Our provisional expectation is that each of the
properties (save those defined through identity itself) that Caesar must
have will in principle be multiply satisfiable.

.           ‘   ’  ‘  ’

Of course the locutions ‘can’ and ‘must’ are normally put to other, more
specific, work. The range of available meanings of these modalities is
 Individuative essentialism 
highly various. De re modal claims are usually based upon people’s abil-
ities or inabilities, their capacities or incapacities as of some time and in
relation to the circumstances of that time, or upon their obligations or
debts or compulsions of that time.29 If someone must repay £ then the
source of the necessity is perhaps a borrowing transaction and a conse-
quential state of indebtedness from which they have not been released.
Or if someone cannot help but slide down a slope then that which neces-
sitates their falling may be a trip, the gradient of the slope, and their lack
of the strength that would be needed to regain control. This is more
unalterable than a state of being obliged to pay money (except perhaps
in Sicily). But the proposal that is put forward in this chapter is that we
should see the de re necessity of essence as the limiting case of the other
de re necessities with which their form appears to group them. The essen-
tial necessity of a trait arises at that point of unalterability where the very
existence of the bearer is conditional upon the trait in question. Here, at this
point, a property is fixed to its bearer by virtue of being inherent in the
individuation of it – inherent in the very possibility of the drawing of a
spatio-temporal boundary around the object in the light of some prin-
ciple of activity of the thing. The closer the source of the attribute to the
singling out of the thing itself – the more it is bound up with the whole
mode of articulating reality to discover such an object there – the more
exigent, obviously, is the necessity that, if there is to be any such thing as the
bearer, it should have the feature in question. The de re ‘must’ of causal
inflexibility here passes over at a certain threshold into the inflexibility
of the metaphysically necessary or the (in one defensible sense) concep-
tual (the concept being the a posteriori concept of being identical with that
very thing). This metaphysical necessity is not of course analytical or (in
any but the loosest sense) logical (see Preamble §). There is no reason,
I believe, why this should make the essentialistic de re attribute any less of
a real attribute of the thing itself.30

29
There are constant temptations to confuse differences in the grounds for modal attributions with
differences in their meaning. It is not necessary to settle here the question whether we have here
an array of distinct senses disposed around a focal idea, or a univocal idea which can conjoin
with different parameters to yield a variety of different kinds of semantic output. See also my op.
cit. in Lovibond and Williams (), pp. –.
30
Pace Wittgenstein who, for the same sort of reasons as I am rehearsing in this chapter, isolated
such attributions but dubbed them non-descriptive.
There is a sense in which an object may not be described. This is that a description may
ascribe to it no property whose absence would reduce the existence of the object itself to
nothing. Description may not express what would be essential for the existence of the object.
(Philosophische Bemerkungen §§–. Cf. also Tractatus Logico-Philosophicus ..)
 Individuative essentialism

.          ,      

Individuative essentialism has put on more leaf. Ought it now to be
pruned back? We need to consider the charge of overspecificity in the
choice of human being as the highest individuative sortal for Caesar.
Thereafter, we can explore yet other directions in which this essential-
ism might grow.
The objection we postponed was this: why could not the necessary
anchor between the actual Caesar and some thinker’s envisaging some-
thing putatively of Caesar be secured by virtue of the subject’s being
individuated under the description this mammal or even this animal?
Suppose first that this objection were correct, and that one of these
unspecific sortal specifications were a good enough sticking point. Even
then there would be de re necessities to be discovered. Animal excludes
stone, insect, paddle steamer, number (and even mereological fusion of animal parts).
The objector might say that the resulting essentialism lacked much of
the interest of the position first stated. But our doctrine of natural kinds
would still have a real point even at the level mammal or animal; and, even
at such a distance from the specific, there might still be something to be
made of scientific discoveries, however much more abstract the choice
for the G property of § would then have to be. The damage the objec-
tion can inflict is limited then. But the real reply to the objection ought
to be that ‘this animal’ is not a good enough identification of Caesar to
sustain the envisaging of him as not a man. Or (better) so soon as we sup-
plement ‘this animal’ in the way in which it needs to be supplemented
in any context in order to be properly individuative there, or in order for
it to furnish any answer to the question ‘which thing is being referred
to?’, that which results is more specific than ‘this animal’.
It is not to be denied that the words ‘this animal’ suffice to express a
rough and ready identification in ordinary contexts of what things are.
But this is because ‘animal’ so easily takes on an individuative force from
a context and/or some other sortal predicate that is ready to hand.31 But
31
Or because it does not matter very much to know more than roughly what the thing in question
is. ‘What’s under that blanket?’ ‘It’s an animal – I should sit on the other blanket if I were you.’
In the context, it is probably perfectly obvious that it is not a leopard or a gila-monster under the
blanket. This is ordinary indoor life. (One can sit next to the blanket, for instance.) It doesn’t
matter whether it’s a cat or a dog or a nanny goat under the blanket. It doesn’t matter exactly
what is under the blanket.
At New Essays .. Leibniz represents Locke as saying, ‘les mots deviennent généraux
lorsqu’ils sont signes d’idées générales, et les idées deviennent générales lorsque par abstraction
on en sépare le temps, le lieu, ou telle autre circonstance, qui peut les déterminer à telle ou telle
existence particulière’. In his objection to this doctrine, Leibniz exploits the distinction that I am
 Individuative essentialism 
the designation ‘this animal’ is complemented in all sorts of different ways.
In itself it determines no single principle of individuation. (Cf. D(iv),
D(v).) How then would it be complemented in the context of the claim
‘N.N. conceived of this particular animal’s not being a man’, where ex
hypothesi there was nothing like ‘man’ to supplement ‘animal’ in the
anchorage of the object conceived to the conceiving? The chief defi-
ciency of animal construed as an autonomously individuative concept is
that in taking something for an animal simpliciter we are not committed
to conceiving of it as a thing with any particular way of coming into
being, of behaving, of interacting with other things, of developing or of
ceasing to be. But if anything at all is to be singled out determinately as
‘this particular animal’, then there is no alternative but for the answer to
the what is it? question to rest on some sufficiently specific causal gener-
alization to provide conditions of identity and persistence for members
of its extension. (These are generalizations to which, whether or not he
knows or wishes it, anyone who subsumes a thing under the concept will
be committed.) This requirement will be breached in the attempted
envisaging of Caesar as not a man, unless there is a sortal term that is
lower than ‘animal’ and either coextensive with or higher than ‘man’ or
‘human being’, and this term is individuative and it suffices to fix Caesar
with Caesar’s very principle of individuation. These conditions are not
empty. Their purport is brought out in conditions such as D(ii), D(iii),
D(iv) and D(v) of Chapter Two. If someone can find such an ultimate
sortal concept for Caesar, if it is higher than man or human being and it is
sustained by generalizations sufficiently specific to fix the identity and
persistence conditions that are presupposed to individuation, then they
are at liberty to substitute it for man or human being throughout the argu-
ment of previous sections, and weaken all my claims accordingly. They
will not have been reduced to nothing.
At this point, however, it may seem that an element of the vagueness
that Ayer predicted (and we anticipated at (⌬) of § above) has crept in.

pressing here. The principles we have to use in keeping individuals under continuous observa-
tion can be made to yield genus terms. There is abstraction from species to genus. (Cf. Ishiguro
()). But there is no abstraction from individuals to species. The principle we use in observ-
ing and tracing a particular does not come with a species concept which we get by abstraction from
individuals. That would be impossible, for in so far as we have a species concept for an individ-
ual, the possession of this concept and the ability to pick out the thing under it cannot be separ-
ated. For genera terms, however, which presuppose individuation, there is no such difficulty. ‘Je ne
disconviens point de cet usage des abstractions, mais c’est plutôt en montant des espèces aux
genres que des individus aux espèces. Car (quelque paradoxe que cela paraisse) il est impossible
à nous d’avoir la connaissance des individus et de trouver le moyen de déterminer exactement
l’individualité d’aucune chose, à moins que de la garder elle même.’
 Individuative essentialism
But it is no part of the essentialist’s purpose to deny that in exploring the
grounds of de re modal attributions we encounter matters of degree.
Vagueness in the delimitation of kinds is not a new admission.32 Nor
does the need to decide whether or not a certain threshold has been
reached entail that the choice to make or not make a given modal asser-
tion is unprincipled or arbitrary, or a matter of no moment. For the
admission does not undermine any of the point or interest or determi-
nacy we have shown how to attach piecemeal to these questions, or
degrade the complexity of what is involved in seeking to establish, well
below the level of how things superficially appear to be, whether one can
or cannot envisage some property ␾ of Caesar. In the end, it may or may
not prove, in given cases, that there is indeterminacy. If (over and above
vagueness) there is indeterminacy, then it may be expected that this will
be an echo of the contestability that can attach to the practicalities of indi-
viduation itself – a matter resumed at some length in Chapters Six and
Seven.

  .     D E R E      ,        : 

             
The derivations of § and the consequential reflections of § and §
placed us in the following position. The identities and distinctnesses of
individuals make up a metaphysical framework for counterfactual spec-
ulation. That framework constrains us in how we can envisage any given
individual’s being or acting. Translating this into the language of pos-
sible worlds, one might say (with Kripke) that, when we construct worlds
that are alternatives to the actual world, the specification of a world
needs to specify, in so far as anything hangs on this, which individuals they
are that figure in that world.33 Moreover, to place a particular individual
32
Leibniz made it willingly in Nouveaux Essais: ‘Les passages d’espèce en espèce peuvent etre insen-
sibles, et pour les discerner ce seroit quelquefois a peu pres comme si on ne sçauroit decider
combien il faut laisser de poils à un homme pour qu’il ne soit point chauve. Cette indetermina-
tion seroit vraye quand meme nous connoistrions parfaitement l’intérieur des creatures dont il
s’agit. Mais je ne vois point qu’elle puisse empêcher les choses d’avoir des essences réelles inde-
pendamment de l’entendement, et nous de les connoistre: il est vray que les noms et les bornes
des espèces seroient quelque fois comme les noms des mesures et des poids, où il faut choisir pour
avoir des bornes fixes’ (..). Compare also .., .. and ...
33
Just as we have to be ready, if it matters, to write into the construction of alternative worlds the
colours or weights of objects, unless this information be deductively entailed by other specifica-
tions already given. If you find it incredible that this specifying should be needed, you are not
taking into account what identity and difference amount to or the proper content of these
notions. It is likely that you are holding onto the illusion that pure properties and relations can
fix identities. For a further attempt to show that this really is an illusion, see Chapter Six, §.
 Individuative essentialism 
within a construction/supposition/world limits that construction/sup-
position/world. For x cannot be represented there in a manner that is at
variance with what x itself substantially is – or in a manner at variance
with anything that flows from that. Here, as so often in this book, the
sense of ‘what x is’ is the one that Aristotle marked out for philosophy.
Apart from being human and having the property G, are there any other
limits on how Caesar can be suppositionally represented? What else
flows from the requirement that, in our envisaging any ␾ at all of Caesar,
␾ should be cosatisfiable with some individuating sortal concept f for
Caesar (human being, say)?
The anchor constraint given in § suggests at least one further
essentialist thought. Wherever ␾ is genuinely conceivable of x and x
could have or could have had the property ␾, the property and some
sortal identification f of x co-satisfy the following requirement: that if
x had ␾ then that would not stand in the way of x’s being singled out
as this particular instantiation of f; x’s having ␾ would not be in con-
flict with this f-identification of x. It is only possible for x to have ␾ if
x’s having ␾ would not obstruct x’s being identified somehow, for the
purposes of conceiving of x’s having ␾, as the particular f it actually
is.34
Let the property of being not-␺ be the negation of a long, long dis-
junction of interesting properties of x, properties that would normally
be invoked in the support of a demonstrative identification or (as for us
now) an indirect identification of Gaius Julius Caesar. Suppose it is asked
whether Caesar could have the property not-␺. Consistently with the
supposition of Caesar’s having not-␺, could he be singled out or individ-
uated as this or that identical-with-Caesar-instantiation of the concept
human being? Or would the putatively total disruption of Caesar’s actual
properties that is postulated by the putative envisaging of Caesar as
having not-␺ subvert the very possibility of finding a collateral descrip-
tion in the form ‘this or that instantiation of human being’ that both
homed upon Caesar and was co-conceivable with not-␺?
Let it be clear that we do not ask this question because we expect the
satisfaction of the predicate ␺ to establish the title of suppositions that
34
At p.  of ‘Replies’ in Lovibond and Williams (), I promised that on some other occasion
I would attempt to expand the Aristotelian account of what an individual is, in order to enrich
the Aristotelian constraint on the construction of suppositions. That is what I am attempting
here. The outcome will be methodologically consistent with that earlier effort, though in detail
different. The aim is to look for more limitations and necessities than were discernible from any-
where on the route chosen on that other occasion. I have not tried to incorporate here all the
points about essentialism argued in that earlier discussion.
 Individuative essentialism
project it onto Caesar to be suppositions relating to Caesar. The suppo-
sitional exercise itself confers that title. (It confers it as of right, which is
not to say that that title cannot be put into jeopardy.) The point at issue
is only that, if the answer to the question posed is that the supposition of
Caesar’s having not-␺ does indeed rule out the kind of identifiability just
spoken of (for provided that ␺ is carefully enough and laboriously
enough constructed, that does seem to be the right answer), then ␺
stands for a property that Caesar must have. Moreover the necessity of
Caesar’s having ␺ is something over and above the necessity of his being
a human being. The thing shown is the strong de re necessity of a
complex property, the one ␺ stands for, that is neither in the category of
substance nor yet imported by something in that category.
What if anything does all this suggest about certain familiar de re
necessities philosophers claim to be able to discover about individuals?
Saul Kripke once asked ‘by what right’ one would call a person sprung
from different sperm and egg from the actual ones ‘this very man’.35
Kripke’s question is surely not intended to imply, in a manner at var-
iance with Kripke’s own insights in this area, that every counterfactual
speculation about Julius Caesar involves the thinker who undertakes it
in establishing his title to identify the individual concerned as Julius
Caesar. Perhaps Kripke’s argument in the passage cited for the neces-
sity to a thing of its actual origin is that the speculator has to be able
to rebut the charge that he has lost his grasp of his subject of discourse
if he conceives of this subject with changed parents or origin. But now
one must ask: can one not rebut the charge of losing the subject of the
supposition by claiming to speculate about how the man whom Brutus
murdered in  B.C. would have fared if (secretly, say) Marius had
been his father? To rebut the charge of losing the individual, there
must always be available to the speculator, consistently with his speculation
and in a way not destructive of the framework in which the question
is posed, some way of specifying the man the speculator means by
Julius Caesar. That seems right. Could a convincing argument be
made along these lines for the thesis of the necessity of origin? It may
well be that a determined attempt to follow through the implications
of the anchor restraint will reveal an important asymmetry between
earlier and later identifying attributes of an individual.36 That will not
35
See p.  of Kripke ().
36
In her admirable new paper ‘Identity, Time and Necessity’, Proceedings of the Aristotelian Society, ,
 (), pp. –, Penelope Mackie explains the plausibility that so many have found in the
thesis of the necessity of origin by showing how it can be based in a branching model of pos-
 Individuative essentialism 
carry one all the way to Kripke’s own arresting contention however –
or so I hazard.

 .                     

   
Suppose someone responds to the hardness of problems of artefact per-
sistence rehearsed in Chapter Three not with proposals about how to
dispense altogether with claims of identity in the more extreme cases of
subtraction, addition and repair, but with a more demanding condition
for a’s counting as the same artefact as b – a condition arrived at by the
incorporation for each artefact kind of its nominal essence and then by
some further requirement about addition and subtraction of matter.
Might this sort of enrichment of the survival conditions of a given arte-
fact kind (already touched upon in Chapter Three, §) be the basis of
certain de re necessities that are frequently mooted for artefacts?
Before we try to reach any conclusion about artefacts and their matter,
we have to be on our guard. We have tried everywhere to distinguish the
question of survival through change – how the object must be at stage n
⫹ given that it was thus and so at stage n – from the question of what can
be envisaged of an entity only conditionally upon the existence of the
entity. Let us be sure to maintain that distinction in what follows and let
us focus on the second question. Conditionally upon its existence, and
only upon that, one surely can envisage a given table’s not being so pol-
ished, not being quite so tall, not being chamfered at the corners. That
is not very hard. But can one envisage in this way its being not a table,
or its not even being a piece of furniture fabricated for a cognate

sibilities concerning actual individuals. According to this model, the ways in which an individ-
ual might have been different are restricted to divergences from actual history: ‘It follows that
any complete alternative possible history for Julius Caesar must have some segment that is the same
as some portion of his actual history’ (p. ). Mackie calls this the ‘overlap requirement’. She calls
it this without enthusiasm – for she deliberately confines her efforts to explaining why the thesis
of the necessity of origin is found attractive and refrains from advocating the thesis herself – and
she bases the need for some such requirement as that of overlap on a version of the ‘anchoring
requirement’ that has motivated the present chapter. See Ayer’s wording of our (⌬) in § above.
I note, however, that, for obvious reasons, she moderates the literal force of the overlap require-
ment to avoid the ‘absurdly strong principle [as it applies to x within an alternative world] that
requires complete coincidence of properties at some time. Nor . . . need it be taken to require that
the overlap include sameness of temporal or spatial location’ (footnote ). But then the question
arises why overlap must be interpreted to require either sameness of material constitution or
coming from the same sperm and egg. Perhaps this doubt about whether there is here any argu-
ment for the necessity of origin as such converges upon the unspoken reservation that discou-
rages Mackie from coming out in any positive advocacy of the necessity of origin.
 Individuative essentialism
purpose, or conceive of its enjoying the negation of the disjunction of
all predicates that relate to its having been fashioned for cognate pur-
poses? That is too much. One cannot envisage any of these things. These
suppositions fail requirements analogous to those already argued for. Let
us search then for de re necessities lying between these extremes of ease
and difficulty in envisaging the table as otherwise than it is.
Let us ask first: can one genuinely envisage my table’s being made of
different wood?37 Here there is little or nothing we can postpone to the
day of an empirical discovery. For all the physical facts upon which the
problem is based may be available already. My provisional answer is this:
that difference of material is something one can envisage without jeop-
ardizing the title of the supposition to relate to this very table.38 The
essence of this or that table does not demand any particular matter, only
the right kind of matter for that artefact.
Here, in front of me, is a table, one much used and leant upon in
writing, a table damaged ten years ago by an inundation of water
through the roof, but carefully repaired, revarnished and repolished. Its
whole simple history is known and available back to the occasion of its
being bought in Chalk Farm, London, NW, in . Let us suppose
that its previous owner and user was also its maker. Let us suppose the
table came into being in , after the intending artificer came upon
several pieces of some Indian hardwood. Could the artificer not, when
he made the table, have taken other pieces? Surely he could. The table
came into being when he did whatever he did to the pieces that he
managed to find. That which constitutes the table he then constructed
and distinguishes it from anything else at all, is something that resulted
from the acts he visited upon the pieces he took at the moment of
making it and the kind of wood those pieces were to which he did what-
ever he did in order to make the table. Why look for more? The table
can in principle be traced back to the time when it was made. But, con-
sistently with everything that we can show in this way, surely this table
37
I venture to think that the argument given at footnote  of Kripke () for the necessity of
constitution depends on treating constitution as if it were identity. For the distinction, see
Chapter One, §.
38
Let us contrast a table, taken together with its wood, and an objet trouvé (on which the artist might
stamp his initials, suppose, and the date and place of the find), taken together with the object
itself before the artist confers upon it that status. In the second case, <objet trouvé, the object
itself>, simply by virtue of the genre, the object found is not transformed into the work of art. Or
rather it is not transformed in the way in which the hardwood is transformed into a table. The
objet trouvé is not transformed thus, even if the artist crops, trims or polishes the thing found. For
in the case of objet trouvé, the stuff of the work of art is the object itself. Contrast the situation (as
I claim it appears) with <table, hardwood pieces>.
 Individuative essentialism 
could (metaphysically could) have been made from other pieces of wood.
This is not to say that it easily or non-disruptively could. If we knew
more about the table’s actual history, then we might find that the pieces
actually employed were the only suitable pieces of wood that were in any
way available to its maker for a table anything like this. We might find
that amazingly many things would have had to be different for the table
to have been made in  from different pieces of hardwood – espe-
cially when we realize that, as a matter of fact, placed as he was, the pre-
vious owner was the only craftsman who was in a position to make such
a table and we remember that, at that time, suitable wood for making
furniture was very scarce. In that case, not only is there no categorical
‘could have been differently constituted’ to be affirmed here (in the sense
of potuit or poterat). There rarely is in such cases. Maybe all the hypothet-
ical ‘coulds’ that are true (all the ‘coulds’ that have the sense of potuisset)
are surprisingly remote from reality. That would be interesting, in its
way. But not even this would show there is no world where the table has
different matter. If there is just one such world, then that suffices for the
flat metaphysical ‘could have’. And what on earth could make the sup-
position of its having other matter incoherent?
Would I go so far then as to say that we are free to distinguish between
two worlds as qualitatively similar as you like, each containing one table
very like my table, and free to do so by identifying the table in the one
world but not the table in the other world as my table? I am committed
to reply that we can do this.
Suppose though that the only relevant difference between the two
worlds is this: that in world one the table has the matter of my table in
the actual world, and in world two it doesn’t. Even here, in the face of
this description of the two worlds, I should say that I can complete the
supposition corresponding to world two. I think I can do this by stipulat-
ing that in world two, and not in world one, it is my table that the sup-
position relates to. What could prevent that? Where such a supposition
is entertained, all that has happened is that our supposition about where
the table is has been separated from our supposition about where such
and such a parcel of matter is, taking advantage of their independent
specifiability and of the non-identity between table and wood.
‘But how can you, where two worlds are qualitatively indistinguish-
able, make a ruling of identity in defiance of the one thing that remains
to distinguish the one table from the other, namely the identity of the
matter?’ Answer: How can you, who ask me this question regarding two
worlds that are qualitatively indistinguishable, insist that the table in one
 Individuative essentialism
but not in the other world is made of such and such matter? Only surely
by allowing supposition to settle which pieces of wood in which world are
the actual pieces making up my actual table. So ‘supposition’ is my
answer too, subject of course to the Aristotelian constraint. Worlds are
the shadows of our suppositions and they take on their identity from
these. Suppositions themselves take on their identity from (inter alia) the
objects they relate to. If they sever themselves from these objects, then
they collapse (or cease to relate to that to which they purport to relate).
But the supposition that involves the table’s being constituted from
different matter does not sever itself from the table that stands in front
of me.

.      :     

        
Where does this leave us? It seems wrong, or so I conclude, to expect that
concepts of particular sorts of artefacts will be more prolific than con-
cepts of particular sorts of natural things in the de re necessities that they
confer upon their instantiators. And, apart from properties such as that
imported by the predicate constructed in paragraph  of §, which will
be as readily available for artefacts as for natural things, the necessary
properties of particular artefacts will come nowhere near to being logi-
cally peculiar to them.
Nevertheless it is natural to ask: could there not be some special sort
of artefact-kind for which, as a very special case, something closer to a
Scotist theory of individual natures held? And could not such an indi-
vidual nature also constrain the identity of the matter making up a given
member of this artefact-kind?
To obtain even an approximation to a case where a Scotist theory of
individual nature was applicable we should need a thing-kind which was
special in one particular way. The instantiators of the thing-kind would
have to be like this: to understand what this one or that one is, to under-
stand what it is and which one it is, would be to understand countless
particular details as so central to the instantiator’s identity that its integ-
rity depended on the presence of all of them, and its existence depended
on virtually all of them. Particular objects of this thing-kind, being
essentially conceived so particularly, could scarcely be envisaged at all as
lacking very many of the particular properties thus designated as theirs.
It follows that any variation in the central set of properties which were
specific to one such special object’s being the particular that it is would
 Individuative essentialism 
then – for this is the consequence of the requirement – have to count as
a definite step towards the destruction of the thing or the replacement
of it. At the same time, though, if we are to find here any approxima-
tion to particularized natures, then anything belonging to this special
kind of artefact kinds would have to be an individual, a material particu-
lar. It must not be a mere type, or a recipe instantiable by different
material particulars.
Are there any sorts of sorts like this? I have argued in another place
that easel pictures, carved sculptures and frescoes come as close as any-
thing can come to meeting such requirements.39 In the culture of the
West, as in certain other cultures, we think of the artist as making some-
thing which is calculated by him to have an effect which cannot be char-
acterized in instrumental or non-aesthetic terms, and cannot be
identified independently of some totality of relevant features that are the
individuating features of this or that individual work. It follows that,
once the work is finished, to visit any interference upon the features com-
prising this set is to threaten the work with obliteration, or with destruc-
tion by replacement, however gradual.40 At the same time, works of art
seem to satisfy the other condition we specified. For we do not conceive
of an easel picture, fresco or carved sculpture as a template for the repro-
duction or multiplication of some particular image. Finally, we attach no
less importance to the artist’s effect being brought about by the same
material means as the artist envisaged and contrived than we attach to
the effect’s itself remaining agelessly utterly indiscernible from one age
to the next.41 What a principle of activity does very completely for a
natural thing, and the function of an ordinary artefact does imperfectly
for the artefact (see above, Chapter Three, §), the artist’s conception of
his own making of the work, his conception of its outcome and his cal-
culation of the effect of the work on an ideal spectator must do for a
painting. We do not look for law-like principles of activity for paintings
or particular paintings; and it is equally true that, being no ordinary arte-
facts, paintings and carved sculptures do not have a function. But any
particular work has an aim, a highly particular aim, that is internal to its
39
See Wiggins (), pp. –.
40
Note that there is vagueness here, because partial destruction, hostile to integrity but insufficient
to bring the work to the point of destruction or replacement tout court, is an idea that makes sense
in this sort of case. That, after all, is the sad condition, all too easily aggravated by mindlessly
scientistic restoration, of many fine works.
41
We may not require the second of these things at all – if only because the artist himself cannot
have expected it. He may have expected change, patina, fading or whatever. He may even have
counted upon these alterations.
 Individuative essentialism
being that very work. The hypothesis I advance for consideration is that
this at least can generate very particular de re necessities.
The work of art has necessarily – is such as to be de re inconceivable
or unenvisageable as definitely lacking – any sufficiently rich complex of
features that has essential occurrence in the artist’s own implicit or
explicit practically realized account (placed where it is in whatever
context of artistic understanding) of this very piece of his work. This is
the claim. It entails that there was an important residue of truth in
certain idealist accounts of works of art. At the same time, I submit that
this way of conceiving these special things is precisely not the way in
which to conceive of an ordinary substance or of an ordinary artefact.
The contrast that this brings to light serves well to mark something note-
worthy in the understanding we have so far accumulated of ordinary
substances.
    

Conceptualism and realism

Quand nous ferons naître la pensée, elle naîtra ainsi dans un

univers déjà rangé.

When we come to exhibit the birth of thought we shall find that it

is born into a universe that is already ordered.
Simone Weil, Leçons de Philosophie, ed. A. Reynaud (Paris ), p. ,
translated by Hugh Price, Lectures on Philosophy (Cambridge, ).

The relation between my consciousness and a world is not a mere

matter of contingency imposed on me by a God who happened to
decide the matter this way rather than that, or imposed on me by a
world accidentally preexisting and a mere causal regularity belong-
ing to it. It is the a priori of the judging subject which has precedence
over the being of God and the world and each and everything in
the world. Even God is for me what he is in consequence of my own
productivity of consciousness. Fear of blasphemy must not distract
us here from the problem. [But] here too, as we found in the case
of other minds, the productivity of consciousness does not itself
signify that I invent or fabricate this transcendency, let alone this
highest of transcendencies.
Edmund Husserl, Formale und Transzendentale Logik (Halle (Saale), ),
pp. –.

.  -             -      

Chapter Four was an exploration of the conceptual limits of our think-
ing about the things that belong to this or that particular ultimate indi-
viduative kind. It pointed to the riches that will lie hidden within any
individuative kind which is a natural or real kind. In the examination of
how we are constrained in our thought about individuals belonging to
kinds that need to be deictically-cum-nomologically specified, we were


 Conceptualism and realism
discovering that which plays in the determination of the essences of such
particulars the part that verbally explicit definitions can play in determi-
nation of the nominal essences of particulars such as artefacts. In both
sorts of case, the constraints on thought derive directly or indirectly from
that which is meant by what we think or say. But in the deictically spec-
ified case, where the sortal predicate is like ‘frog’ and unlike ‘house’ and
has its sense fixed in a manner that is reality-involving, meaning is con-
ferred (or so we claimed) in a distinctive way. And this in its turn gener-
ates conclusions that appear to go well beyond those conventionally
accounted as semantical. To anyone who thinks he knows how to make
such distinctions, it will seem to be a metaphysical thesis, not a semanti-
cal thesis, that being a human being or having the G property (see
Chapter Four, §) is an absolute individuative prerequisite for anything’s
being Caesar. For this reason, the arguments of Chapter Four will attract
the hostility of at least two schools of thought. Designating my own posi-
tion (simply for brevity and convenience) conceptualist realism, I suggest
that one of its likeliest critics may be called the anti-realist conceptualist, and
another the anti-conceptualist realist.
The anti-realist conceptualist may or may not be an R theorist, and
may or may not have found intelligible the defence of principle D
attempted in Chapter Two. What he is certain to find questionable in
the extreme is the purported absoluteness (non-relativity to topic,
context, or level), of the necessity that I have alleged (exempli gratia) to
attach to the particular entity Caesar’s being a human being.1 The anti-
conceptualist realist, on the other hand, whose position is addressed at
§ following, will scarcely object to the categorical conceptions of neces-
sity, substance, truth or whatever that lurk in earlier chapters; but he will
doubt (as Michael Ayers does, for instance)2 that anyone ought to count
as a realist who is prepared to embrace the sort of conceptualism that
informed the argument that I gave for D and that motivated the deriva-
tion of de re necessity. There is a third form of opposition that might come
from latter-day Humeans who are convinced of the theoretical impos-
sibility of a de re necessity that is not reducible to the analyticity of some
sentence. But I have said my piece against this school, and shall now
direct attention upon the other two.
‘It is all very well’, my new opponents are likely to object, ‘talking as
you did in the previous chapter of the constraints on envisaging a thing.

11
See Bas C. van Fraassen (–).
12
See Michael R. Ayers (). See also Parts One and Three, Vol.  of Ayers ().
 Conceptualism and realism 
Suppose that the place marked by the z-variable in “it is possible to envis-
age z’s being ␾” is, as you claim, an extensional position and somehow
confers extensionality upon “z” in “z can be ␾”. We shall not object. We
are also ready to suppose that, when Quine claimed that to be necessar-
ily greater than seven was not a trait of a number but depended only on
the manner of referring to the number,3 he overlooked the kind of neces-
sity that you have found outside the range of Quine’s three grades of
modal involvement and claimed to find traces of in Quine’s and
Goodman’s own writings. We shall not even demur when you claim that,
if a thinker genuinely conceives of z as ␾, there must, for purposes of
the kind of possibility you were concerned with, be some sortal concept
f such that z can be identified as this or that instance of f, f and ␾ being
coconceivable of z. This last requirement may perhaps justify you in
adopting, for purposes of your brand of possibility, what Quine called
an “invidious attitude towards certain ways of specifying a thing and
favouring [some over] other ways [of specifying x] as somehow reveal-
ing the essence of the object”. But these concessions, in the particular
form in which you are trying to extract them, are perfectly unimportant.
For the most you can ever obtain by these means is a conceptualism that
is shaped and conditioned by human powers of envisaging. These
powers are the product, both time-bound and culture-bound, of a con-
ceptual scheme which is determined by interests that are very special.
However deep seated they may seem to us, such interests in the world
that surround us are scarcely implanted in us by nature in order that our
beliefs or theories should mirror nature itself. You are deceiving yourself
if you think you have found some philosophical engine by which to invest
with real essences the concrete entities of the world. For us, there is no
doubt a certain apparent indispensability in the everyday conceptual
scheme that we, from our limited and cosmically insignificant point of
view, bring to bear upon the world. But the real entities of the real world
are whatever they are, independently of whether human beings exist or
not. They are perfectly indifferent, so to speak, in this matter of how
human beings conceive of them. That is not our only complaint. A
further charge that we should urge against you is that the conceptual
scheme for which you are working to get this privilege is scarcely very
different, even now, from the scheme that systematically delayed the
progress of natural science beyond the jejune approximations of
Aristotle.’

13
See Quine (), pp.  and .
 Conceptualism and realism
The moral that the anti-realist conceptualist seeks to draw from all
this is that the aspiration to contemplate, however distantly, the real
essences of concrete things is best forgotten about. For in nature itself,
he will say, there are no modalities.
The moral drawn by the anti-conceptualist realist, on the other hand,
is not that essentialism as such is mistaken. The conclusion he will come
to is that the realist aspiration is inimical to any position like the concep-
tualism that I have been arguing for. He will persist in the idea that the
realist ideal is to be valued above any supposed conceptualist insight. But
he will hold fast to his conviction that, in the long run, there must be
some better approach than that which is supposed to lead through (A) –
(H) of Chapter Four to the real essences of concrete things.
In answer to these critics, I shall be content for the most part to
clarify the claims of the particular conceptualism defended here to be
one form of realism. For the objection I have just rehearsed compre-
hends in each of its two forms a number of misapprehensions whose
removal will be far more instructive and far more liberating of thought
than exhaustive responses to the divers accusations and counter-
suggestions that each critic could muster on his side. Once conceptu-
alist realism is recognized for what it is, its metaphysical plainness will
be manifest.

.          
(a) The objection given in the previous section depends partly on a mis-
understanding of one point that Chapter Four will have made familiar.
The conditions that have been characterized as peculiarly central to the
articulation of this or that particular thing and its division from the rest
of reality concerned, not the world’s need to contain Caesar, but only
the kind of thing Caesar had to be if there was to be any such thing as
Caesar, namely (say) a human being. The necessity in question is at once
crucial with respect to the question of the individual’s being there to
articulate, and peculiarly innocuous. The necessity is categorical with
respect to Caesar, but from another point of view it is a conditioned
necessity. The singular thing about this necessity resides in the character
of its condition. The condition comprehends all states of affairs in which
there exists . . . well, the very entity in question.4
14
For the reasons for putting the matter in this way, see § of my article ‘The Kant – Frege –
Russell View of Existence: a rehabilitation of the second level view’, in Modality, Morality & Belief,
ed. Sinnott, Armstrong, Raffman and Asher (Cambridge, ).
 Conceptualism and realism 
(b) To put a sentence forward as a true statement of de re necessity (or
as a de dicto necessity for that matter, if the de dicto were in question) is not
to put the sentence forward as in some special way proof from revision,
correction, or the boredom of our descendants. Those who object in
such terms to essentialism as an expression of the stick-in-the-mud men-
tality or of ill-advised reaction have insufficiently distinguished these
several statuses.
(c) It is necessary to reiterate the sincerity of the essentialist adher-
ence to principle ⌰ of § of Chapter Four. Essences of natural things,
as we have them here, are not fancified vacuities parading themselves
in the shadow of familiar things as the ultimate explanation of every-
thing that happens in the world.5 They are natures whose possession by
their owners is the precondition of their owners being divided from the
rest of the reality as anything at all. These natures are delimited by ref-
erence to causal or explanatory principles and purposes that are low
level perhaps; but they are fully demanding enough for something to
count as their being disappointed or frustrated. (Witness the longish list
of would-be sortal concepts that have definitely failed.) It is true that,
whatever these principles may once have seemed to be, the principles
we have called the laws of development of individual natural sub-
stances (that is the laws associated with individuative kinds of natural
things) are not themselves the scientifically basic laws of the physical
world. (Or rather, they need not be.) Nor in general are they reducible
(in the strict and proper sense) to such basic laws. Nevertheless, the
kind-bound laws of coming to be, of distinctive activity, and of passing
away are nomologically grounded. They are supervenient upon, or better
(as Leibniz might put it) consentient with, the more basic laws that are
immanent in all things.6
15
We do not look for scientifically ultimate explanations at this level of conceptualization. By the
standards of , years ago it was not necessarily unscientific to look for them there, as Aristotle
justifiably and non-vacuously did. (See the interpretation offered by Alan Gotthelf in ‘Aristotle’s
Conception of Final Causality’, Review of Metaphysics,  ().) Of course, it has been discov-
ered since Aristotle that the search for fundamental explanations and principles, and for mech-
anisms underlying the phenomena, requires not only a new vocabulary but also a new ontology.
On this see below, §. But let us take a sane view of all this. The failure of Aristotelian science is
final. Its failure does not entail, however, that science has shown that every explanation in any
way worth having of anything worth knowing must eventually find expression at the level of the
science that has displaced the human world view.
16
Cf. Leibniz, Gerhardt , , op. cit. at head of Chapter Three. Not only are these ideas separa-
ble from the teleological view of nature (see Chapter Three, Section ). They are separable also
from the teleological conceptions of certain theorists of Natural Law about the levels of excel-
lence that natural things may realize or attain to. For an amusing and summary account of these
ideas see H. L. A. Hart, The Concept of Law (Oxford, ), p. .
 Conceptualism and realism
(d) The realist conceptualist may cheerfully admit that the sortal con-
cepts of which we are possessed, and which he argues to be presupposed
to our articulation of reality, are (in a certain fashion and sense – see §
below) the creatures of our interests (provided that these interests are
generously enough conceived to include our curiosity). He may also
allow that (often at least) there may be no one way in which we must artic-
ulate reality, nor any one level at which we must. The thing that does not
follow from any of this, he will say, is that, relative to some determinate
epistemological-cum-practical interest, there is arbitrariness in the dis-
criminations we make, or in the existence conditions that we ascribe to
the members of the natural and artefact kinds under which our interests
(as properly and generously conceived) oblige us to make sense of expe-
rience. For the supposed arbitrariness of whether or not this or that
concept is brought or not brought to bear upon reality – better (see
Preamble §) the supposed arbitrariness of whether we deploy the con-
ception corresponding to this concept and look for fs in reality, or deploy
the conception corresponding to that concept and look for gs in reality –
cannot be translated into any corresponding arbitrariness in the articu-
lation of the things themselves that fall under the concept.7 Arbitrariness
or vagueness that attaches to the determination of the extension of the
sortal concept has a different source (see Chapter Four, §).
Nobody will find any of these confusions or misapprehensions very
inviting – at least when they are anatomized in this way. But that does
not mean that they are dead or inert or have no indirect influence.
Moreover, confusion apart, it must also be allowed that the thinking the
anti-essentialist opposes to the essentialist conception is sustained by the
enchantment of an image that is vivid in the extreme.

.                      

   :           -       
  
‘At bottom, everything that can be said about the world, can be said in purely
general statements, without modalities. There is no thisness beyond suchness,
but every actual individual is individuated already by the properties it has in this
world; hence can be denoted in principle by a definite description in which the
quantifier ranges over actual existents alone. At this bottom level the only neces-
sity we can countenance is purely logical or verbal necessity which, like God, is
17
Cf. p.  and other remarks I make in § of ‘The De Re ‘Must’: a Note on the Logical Form of
Essentialist Claims’, in Truth and Meaning: Essays in Semantics, ed. Evans and McDowell (Oxford,
).
 Conceptualism and realism 
no respecter of persons. In this modality whatever Peter can do Paul can do also.
A semantic representation of this will use a conventional identification of indi-
viduals in different worlds, but since every individual plays each possible role in
some possible world, every choice of conventional identifications which does not
violate the identity principle that no two existents in world ␣ have all the same
properties in ␣ yields the same result.
‘To make sense of our world in a convenient fashion, however, we raise certain
regularities to the status of laws and (not independently) certain attributes to the
status of natures. In the formal mode, this means that some statements assume
the office of assumptions which may be tacitly used in all reasoning, and certain
predicates are chosen to form a classificatory scheme. Once this is done, we
produce relative or tacitly conditional modal qualifiers . . . A proposition P is
peculiarly about individual e if we can change the truth value of P by permuting
e with another individual e⬘ (while leaving all other individuals fixed), but cannot
change that truth value by any permutation which leaves e fixed . . . In a full
model, no proposition peculiarly about a particular entity can be necessary; for
[in a full model] the necessary proposition is closed under all permutations.’8
What Van Fraassen describes here is a sort of reconstruction in two
stages of how it is possible for us to have our present ideas of substance
and de re modality. The reconstruction is meant to tell us what we should
make of these ideas and of the fact that we have them. Not too much,
he suggests. The reconstruction is not put forward by Van Fraassen as
an ordinary historical hypothesis or even as a piece of psychogenetic
theory. Taking it rather as it is intended, we shall find that one thing is
crucial to the reconstruction, as to the attitude that it recommends
towards substance and modality: the two stages of which it speaks must
be intelligible to us and describable by us, the theorists, in just the rela-
tionship that the account postulates. How otherwise can we credit Van
Fraassen’s own words with even the semblance of intelligibility?
The difficulty in this respect attaches to the first stage. Here, whatever
Peter can do Paul can do also (Van Fraassen says) – indeed any individ-
ual whatever can do. For anything, it seems, can do or be anything at this
level, the only constraints being those imposed by logical necessity in the

18
Van Fraassen, part  of ‘Essence and Existence’ (–). Cf. part :
The nominalist’s first and basic move in this game is to say that all natural necessities are
elliptic for conditional verbal necessities. This sheet on which I write must burn if heated,
because it is paper – yes. But the only necessity that is really there is that all paper must burn
when heated. This is so, but means only that we would not call something paper if it behaved
differently . . . When sufficiently refined, the position that all non-verbal necessities are ellip-
ses for conditional necessities ex vi terminorum can be held.
Cf. also the same author’s ‘The Only Necessity is Verbal Necessity’, Journal of Philosophy (Feb.
).
 Conceptualism and realism
narrowest sense which, whatever dues it pays to ‘indiscernibility at a
world’, is ‘no respecter’ of identity in the proper sense. The question
then is: can we really understand this first stage of the reconstruction?
Only, I submit, if we can make sense of an entity that is nothing in par-
ticular. (Cf. the justification and refinement of ⌬ of Chapter Four,
section , undertaken in § and §.)
Van Fraassen’s adherence to the Identity of Indiscernibles is impor-
tant here. Contrast with his explicit reliance upon the principle every-
thing that we have claimed in Chapter Two against the Identity of
Indiscernibles and have elsewhere implied (in agreement with P. F.
Strawson, Individuals, the chapter on monads and passim) about the indis-
pensability and irreducibility of the demonstrative function. To turn this
into this-ness and, having found no way to understand this-ness as a sort of
such-ness, to seek then to patch matters up by espousing the Identity of
Indiscernibles – this is a counsel of confusion, confusion that is needless.
But there is scarcely anything unconfused that can be said or denied by
the use of the words haecceitas and this-ness. What these words purport to
help one to say or deny is no less mixed up than the noises that they make
are unlovely. (Cf. Chapter Four, §.)
Finally, I would claim that the whole charm of Van Fraassen’s picture
depends on one’s allowing that one can first have an ontology of partic-
ulars conceived as bare, and then at a second stage introduce, as an
instrument of understanding, explanation and discovery, what Quine
and Geach have called an ideology. But is it not a mysterious suggestion
that a whole range of attributions, including sortal attributions which say
what various everyday things are, could be determined at this second
stage (in radical dependence, Van Fraassen suggests, on some human
viewpoint), by being superimposed upon a completed first stage ontol-
ogy which is at once bare yet existentially determinate? For anyone
versed in the notion that an ontology is by no means an empty or uncom-
mittal thing, this is as strange as being asked to believe that one could
distinguish between (first) a neutral or concern-uncontaminated ontol-
ogy of jokes (what it is for x to be a joke, or for x and y to be the same
joke, or for there to be such a joke as z), and (second), a quite separate
and interest-contaminated range of predicative attributions to jokes
(‘funny’, ‘comic’, ‘witty’, ‘off-colour’, ‘whimsical’) or things in jokes
(‘incongruous’, ‘ludicrous’, ‘absurd’). Is it compatible with a good theory
of existence to suppose that the first stage (ontology) could have been
provided for while leaving absolutely everything else to be determined
at the second (ideology)? Let anyone still charmed by Van Fraassen’s
 Conceptualism and realism 
picture rehearse these questions again, mutatis mutandis, for natural
numbers, quantities, shades, . . .
Obviously, if ontology were completely insensitive to all changes in
theory, in ideology or in how things are conceived, and if one and the
same fixed set of entities could be descried by conceptions of the world
that were different to just any degree that you like, then that would
subvert not only Chapter Four but also the account offered in Chapter
Two, §, of the relationship between existence and belonging to a sort.
I should hope that the positive account of existence offered there had
the attraction of simplicity. There is a range of basic sortal attributions
that we apply to various everyday things – ‘this is a horse’, ‘this is a tree’,
‘that is a man’. These belong to the level of ontology and, at least to
this extent, ontology and ideology must contaminate one another
immediately. What is strange is that the anti-essentialists whom I am
attacking9 accept all these attributions in their unmodalized form, and
then (one stage too late, in my opinion, for they have already consented
to pick out the thing and to involve themselves, however minimally, in
some proto-theoretical conception of entities of the relevant kind)
adduce as a reason to deprecate the suggestion that any of these things
had to be a horse, or a tree or a man, the anthropocentricity of the view-
point that underlies and conditions the attributions. But here as every-
where the question is: of what could one ever be speaking if one allowed
that it might equally well have been a prime number or a fire shovel,
though it was in fact neither?

.                  -

       
I advance for the reader’s consideration a hypothesis: the anti-realist case
against essential properties has rested on a sequence of three ideas.
() It is conceivable that, in the absence of creatures capable of thought
as we know it, the particular kinds of individual substance that we
ourselves recognize in this world would never have been discovered.
(Indeed nothing forces anyone to discover in a place even that which
is there to be discerned in terms of his own conceptions.) In that
place, in the room of what we ourselves find there, or (so to say) out
of its matter, another race of creatures might have articulated and
19
Russell was an opponent of essentialism. But these paragraphs are not directed at his position.
For he pursued the logic of his position (and the logic of my paragraphs here) to the point of
rejecting the ontology which is presupposed to saying ‘this is a man’. See op. cit. note .
 Conceptualism and realism
individuated quite other kinds of things with very different principles
of individuation.
From this questionable if not altogether implausible premiss the
conceptualist anti-essentialist goes on to conclude:
() It is perfectly possible for what is there to be discovered in a given
place, e.g. for a substance which is there, to be conceptualized in at
least two quite different ways. The progress of science would pre-
cisely consist in conceptualizing it in better, and radically different,
ways. So, he argues:
() The constraints arising from the conceptualization and individua-
tion of the particulars that we recognize in the world, and can draw
spatio-temporal boundaries around, must not be represented as
necessities on what is actually there independently of mind or
thought. The most objectionable feature of all the objectionable fea-
tures of essential attributions is that they purport to record such
necessities.
It is at stage () that everything goes wrong here. What is this substance
out there that can be conceptualized in radically different ways, which
can be seized upon in thought by the anti-essentialist, but can have rad-
ically different principles of existence and persistence ascribed to it?
This is surely an entity with inconsistent properties. If the charge of
having a ridiculous conception of substance sticks anywhere, it sticks on
anyone who argues via () from () to (). But it is only by means of some
such conception that an anti-realist philosopher could crowd out alto-
gether the possibility, which is important but always neglected, that, even
if a conceptual scheme capable of exploring and understanding the
physical world might dispense altogether with the articulation and isola-
tion of x from its environment (because it could afford to ignore x alto-
gether), still the essential attributions upon x record or reflect conditions
at once fundamental and invariant upon x in particular’s existing and
being singled out in reality.

.             ,    

                    ;    
       -             
If the preceding section succeeded to any significant extent, it will have
gained favour for some sort of conceptualist view of individuation. But
conceptualism may seem to have strange consequences.
Leszek Kolakowski writes:
 Conceptualism and realism 
‘The picture of reality sketched by everyday perception and by scientific think-
ing is a kind of human creation (not imitation) since both the linguistic and the
scientific division of the world into particular objects arise from man’s practical
needs. In this sense the world’s products must be considered artificial. In this
world the sun and stars exist because man is able to make them his objects, differ-
entiated in material and conceived as ‘corporeal individuals’. In abstract,
nothing prevents us from dissecting surrounding material into fragments con-
structed in a manner completely different from what we are used to. Thus,
speaking more simply, we could build a world where there would be no such
objects as ‘horse’, ‘leaf ’, ‘star’, and others allegedly devised by nature. Instead,
there might be, for example, such objects as ‘half a horse and a piece of river’,
‘my ear and the moon’, and other similar products of a surrealist imagination.’10
I am bound to agree with any anti-conceptualist realist who asserts that
what Kolakowski says here is impossible to believe. I should hold that any
conceptualism which is bent on picking its way past the confusions that
have sustained for so long the conflict between realism and idealism must
indeed strive to do justice to the insight for which Kolakowski is seeking
expression. But it will see the whole interest of these problems as resid-
ing in the problem of finding the elusive, correct and unexciting formu-
lation of that insight. A conceptualism that aspires to the title of a modest
and sober realism must keep at arm’s length any such ‘sense’ as the sense
in which horses, leaves, sun and stars could be supposed to be artefacts.
It must hold on to the distinctions so laboriously worked out in Chapter
Three, §. It must leave itself room to express all the proper reservations
(concerning the lack of appropriate causal or explanatory principles,
etc.) about the prospects of finding a genuinely sortal notion with the
same extension as half a horse plus a piece of river. And it must hold a nice
balance, adjusted to what is in fact a subtly reciprocal relation, between
() the extent to which the conceptions that we bring to bear in order to
distinguish, articulate and individuate things in nature are our inventions
and () the extent to which these conceptions make reference to concepts
that we find or permit nature itself to discover to us, thereby informing and
regulating our conceptions of them. (For the distinction of conceptions
and concepts, see Preamble §.) Conceptualism properly conceived must
not entail that, before we grasped these concepts, their extensions did not
exist autonomously. It must insist that natural things and their concepts
existed independently of whether our conceptions of them were des-
tined to be fashioned or the things themselves were destined to be dis-
10
Leszek Kolakowski, Towards a Marxist Humanism (New York, ), pp. –. My attention was
first drawn to this passage by Michael Ayers, who has expressed doubts whether my own sort of
conceptualism leaves me in a position to disagree with it.
 Conceptualism and realism
covered falling under these concepts.11 Its most distinctive contention is
that, even though horses, leaves, sun and stars are not inventions or arte-
facts, still, if such things as horses, leaves, sun and stars were to be singled
out in experience at all so as to become the objects of thought, then some
scheme had to be fashioned or formed, in the back and forth process
between recurrent traits in nature and would-be cognitive conceptions
of these traits, that made it possible for them to be picked out.
For someone to single out a leaf or a horse or a sun or a star, or what-
ever it is, that which he singles out must have the right principle of indi-
viduation for a leaf or a horse or a sun or a star . . . For to single out one
of these things he must single it out. Such truisms would scarcely be
worth writing down if philosophy were not driven from side to side here
of the almost unnegotiable strait that divides the realist myth of the self-
differentiating object (the object which announces itself as the very object it
is to any mind, however passive or of whatever orientation) from the sub-
stratum myth that is the recurrent temptation of bad conceptualism.12 It
is easy to scoff at substratum. It is less easy to escape the insidious idea that
there can be the singling out in a place of a merely determinable space-
occupier awaiting incongruent or discordant substantial determinations
(individuatively inconsistent answers to the question of what it is).13 But
no substance has been singled out at all until something makes it deter-
11
Cf. Hobbes’s definition: ‘a body is that, which having no dependence upon our thought, is coin-
cident or coextended with some part of space’. De Corpore ,  (Molesworth, p. ).
12
Rebounding from all this, there is a temptation to add nonsense to confusion and say that a thing
is nothing but a sum (in some inexplicable sense of ‘sum’) of properties and/or appearances, or
a set of universals each sitting in the next one’s lap. This is in fact the notion of thing favoured
by some anti-essentialists and sometimes also attributed to Leibniz (entirely wrongly, see my
()). How it is to be combined with the idea that objects so conceived have no essences finally
passes comprehension. Consider, for a start, the negation of the conjunction of these properties.
And if the reply to that is that these collocations of properties have shifting membership, then
inquire after the principle determining membership. The answer to this will smuggle back the
notion of a substance or a continuant.
13
There will exist too indefinitely many items with too many distinct principles of identity and per-
sistence which you might find in that place – the thing, the parcel of stuff that makes up the thing,
and the mereological sum of all the components of that parcel, to name but three. See also here
A. M. Quinton, The Nature of Things, p. . For the claim that the same portion of space can be
co-occupied by different things, see my ().
Against all such arguments, Michael Ayers has written (, p. ; and compare M. C.
Bradley’s critique at Australasian Journal of Philosophy,  () concerning the principles of tracing
that have been defended in Chapter Two):
it might be thought that in unravelling the sweater we remove or destroy one principle of
unity, and [that] by cutting the [woollen] thread [of the ball of wool from which the sweater
was knitted] we remove a second. Each operation is possible without the other, and so each
principle of unity seems independent of the other [giving in one and the same place two
different things of different kinds]. The latter and more realistic view, however, is that, if a
sweater consists of a single thread, then this means only that the different parts of the wool
 Conceptualism and realism 
minate which entity has been singled out; and for this to be determinate,
there must be something in the singling out that makes it determinate
which principle is the principle of individuation for the entity and under
what family of individuatively concordant sortal concepts it is to be sub-
sumed (see Chapter Two, §).14

.        


An anti-conceptualist realist who takes exception to the essentialism
arrived at by means of principle ⌬ of Chapter Four will view with sus-
picion any suggestion that truisms are available from which, in the pres-
ence of a strict notion of identity, anything like this will follow: that the
individuative scheme we bring to bear upon experience or the fashion in
which (as a race, or as a culture, or as a culture at a time) we articulate
reality can determine the kinds of thing we shall single out (fs, gs or what-
ever).15 Such claims may seem to the realist to be tantamount to saying
that the active inquiring mind does not merely construe reality but con-
structs it – as if under certain conditions the mind could fashion not only
concepts or conceptions but truth itself. Is this not a particularly unap-
pealing form of idealism, he will ask?
The literal purport of the conceptualist claim which the realist has put
under discussion is not that mind constructs reality. If it comes close to
seeming to say that, this is only because the truth has here to rub shoul-
ders with two opposing doctrines that are so similar in their pretensions
hang together in more than one way, and so have a unity that is more difficult to destroy than
would otherwise be the case. Roughly, the unity or structure of a sweater knitted from a
hundred separate threads is destroyed by unravelling, and the unity of a single thread by
cutting; while the unity of a sweater knitted from a single thread survives either operation
but not both. Thus no serious doubt is cast on that apparently essential ingredient of a realist
theory of the identity of physical objects, the principle that the same matter cannot be sub-
jected to more than one principle of unity at the same time: i.e. cannot compose more than
one individual simultaneously.
If this says that in some cases (composition from one thread) unravelling a sweater is not a way
of destroying it, but that in other cases (composition from several threads of wool) things are so
different that unravelling is enough for destruction, then I am happy to let the reader assess the
price of Ayers’s ‘realist principle’. I hope it will be clear why I believe that the principle is dis-
pensable to realism, to objectivity, and to the nomologically grounded conception of substance.
(Compare here certain difficulties that I claim to see in a similar principle of Eli Hirsch’s: see
footnote  of my ().) Singling out that sweater is not singling out that wool.
14
A fallacy for the unwary: It is indeterminate what thing A singled out; therefore A singled out something inde-
terminate.
15
Michael Ayers (), Eli Hirsch (‘Physical Identity’, Philosophical Review (), especially pp.
f.) and Fei Xu, among others, have taken exception to these claims. See now Wiggins (a).
 Conceptualism and realism
and underlying absurdity as to leave only a small space between for truth
to stand. Our claim was only that what sortal concepts we bring to bear
upon experience determines what we can find there – just as the size and
mesh of a net determine, not what fish are in the sea, but which ones we
shall catch. It is true that the individuative conceptions that are brought
to bear at any point will come with notions about the ways in which
things of a given kind behave. These notions will bear on persistence
conditions. But this does not imply that, once things of a given kind, fs
or gs, are lighted upon, the individuative scheme we bring to bear will
itself determine something further – a principle of activity or a persis-
tence condition. What that principle is for fs or gs depends only on fs or
gs. We discover it in tandem with grasping the concept of an f or a g and
applying our conception of it to the fs and gs we light upon. Moreover,
our conception of the things that answer to the concept is given to us in
a reciprocal process of the kind that Chapter Three was meant to illus-
trate, a process comprising not only deixis, not only allusion to the nature
whatever it may be of that which is demonstrated, but practical and pro-
bative interaction with the things themselves that are in question.
Much hangs here on identifying correctly the point, if any, where
there is choice or freedom in the articulation of reality. The near cer-
tainty of incomprehension of the tenets of conceptualist realism by a
different kind of realist is so great that even the reader who has seized
what I would say here may forgive me for attempting one more state-
ment of the point at issue (and for returning to the matter once again
in the course of Chapter Six). As soon as a set of individuative con-
cerns is determinate, and an interest and a set of sortal concepts is fixed
upon, nothing that can be done by one who grasps the concept will
determine whether or not there is anything at a given place and time
that actually satisfies the given concept. Any freedom we enjoy must
come at an earlier point. Whatever may be the truth to be found in the
idea that what concepts (better, what conceptions of what concepts) a
thinker applies to the world depends on his inventiveness in the fash-
ioning of conceptions and the devising of hypotheses, whatever the
truth of the idea that the accident or caprice of his interests and/or the
thinker’s conceptual daring and enterprise will carry him at his pleas-
ure to places that nothing else would have carried him to – this exhausts
his freedom. For, once arrived at a place, what he finds there is not at
his pleasure. Whether or not there exist things that answer to the
concept f, and what they are like, cannot depend on his invention or
his concerns.
 Conceptualism and realism 
.            ,       
      
It is not enough to remove confusions. The anti-conceptualist realist is
not necessarily someone confused by anything at all, except the differ-
ence between good and bad conceptualism. And he is most unlikely to
think it relevant to what he has to say that one can count as objectively
as you like, and answer the question what is it? as accurately as you like
under sortal concepts of utterly provincial or frivolous provenance. His
real preoccupation is surely not with objectivity, for it presupposes that.
He will probably allow that we have well-tried canons for recognizing
the correctness or incorrectness of subsuming this or that object under
such and such a concept. Nor is he bent on denying that the truth with
which such predications are concerned can transcend our means of rec-
ognizing truth. Evidently, the thing that chiefly concerns the anti-
conceptualist realist is to preserve our thought’s prospects of passing
beyond the narrowly anthropocentric.16 But here it seems to me that an
essentialist of my persuasion can make common cause with him. For, as
we have conceived it, an individuative scheme for the articulation of
reality and the singling out of natural things can be wide open to reality.
There is nothing to prevent the scheme from being regulated and cor-
rected constantly by the things in reality themselves and/or our evolving
conceptions of things in reality. Here I would hark back, of course, to
the distinction proposed in Chapter Three and Preamble § between
nominal or stipulative definitions, e.g. those founded in the function of
artefacts (whose nature depends on us, or has depended on us, collec-
tively), and the real or causal definitions (strictly not ‘definitions’) that we
hunt after in nature, identifying their targets by ‘tacit reference to the
real essence of this or that species of body’ (Locke, Essay ..), first
annexing these essences by marks that can stand proxy for them (marks
that ‘tiennent lieu’, Leibniz, Nouveaux Essais, ..). With the second
kind of definition, deictically directed and provisional in content, we
cannot of course avoid the risk in which our explanatory aspirations
inevitably involve us, namely that, where we are unlucky, we may have
been searching for something that was not there to be found. But here
16
For the distinctness of these concerns, and for the Peircean conception of the advance of science
beyond and away from the anthropocentric, see ‘Truth, Invention, and the Meaning of Life’, in
my Needs, Values, Truth (, ; ), pp. –.
I find that my conception of realism here is in some respect anticipated, albeit in terms more
optimistic than I should think necessary or desirable, by Milton Fisk in ‘The World Regained’,
Journal of Philosophy (), see especially pp. –.
 Conceptualism and realism
too, when all goes well, things themselves can shape our conception of
things.
‘You assert that the notion of substance is formed from concepts, and not from
things. But are not concepts themselves formed from things? (Nonne ipsi con-
ceptus formantur ex rebus?) You say that the notion of substance is a concept
of the mind, or an entity of reason as they say. But the same can be said of any
concept.’ (Leibniz to de Volder.)17
In so far as Leibniz is making a completely general claim in this passage,
it is no doubt falsified in part by the case that he always ignores (or, where
he does not ignore it, underestimates), viz. artefact kinds and cultural
objects where things are fashioned to answer to a conception and there
is no question of shaping the conception of the concept to the things
that prove to fall under the concept. But in the other case, there need be
nothing to prevent human understanding from coming ever closer to
finding features of the world whose articulation is not sensitive to the
more narrowly practical or provincial explanatory interests that first
moved us to inquire into the nature of things. It may be true that the
normal thing-kind concept (even where the understanding sees that not
as the creature of our understanding but as something in nature that the
understanding has prepared itself to be able to discover or light upon) is
rarely an absolutely basic concept conceived as basic by physical science
– still less a concept so fundamental that no real understanding of nature
could ever look past the extension of this concept. It may also be true
that, however deep we go, an intelligence of a superior kind might latch
on to a more fundamental and explanatorily more basic class of things
whose identifying attributes pulled their weight in a far better and more
universal theory than do any of the kinds that we, in our lowly condi-
tion, shall ever discriminate. Let us not argue here about that. To what-
ever extent this is true, the causal-cum-explanatory interests that give us
the most sophisticated and scientific sortal conceptions we possess may
indeed be somewhat provincial. But if that is so, then a proper answer
to the realist critic ought not to insist that the concepts these conceptions
are conceptions of are never provincial to any degree. One part of the
better answer is that the critic has not shown that all genuinely explan-
atory insights must be framed at the deepest level of fundamental theory.
Moreover, even for purposes of the purest theory of science, the critic
may be in error. His doctrine would certainly be unmanageable in the
daily practice of giving expositions of the relatively fundamental theo-
17
Gerhardt , p. . Cf. Nouveaux Essais, ., §.
 Conceptualism and realism 
ries we do possess. The less exigent condition of acceptance that is
imposed on the sortal notions that we continue in practice to employ is
not in any case an empty one. And equipped as it is with such notions,
our present scheme of ideas surely does represent a respectable, however
unimpressive, stage in some however gradual process of the revelation
of reality. Only sneering discontent with all would-be factual thought or
speech as such, or a gloating dissatisfaction with it as any sort of a record
of reality (and these are not the vices of the anti-conceptualist realist),
could prompt someone to deny this.

.           
By way of partial summary of this and the previous chapter, I shall close
with some suggestions or speculations concerning the relation of the sci-
entific world view and the less theoretical view of reality which lags
behind the scientific world view but is practically indispensable to the
sustaining of any meaning that might be attributed to our thoughts.
(i) Many of the scientifically significant shifts in our conceptions of the
world represent not shifts in our conceptions of a settled ontology, or
shifts in the individuative conceptions proper to one and the same set of
entities, but the abandonment for certain explanatory purposes of the old
set of entities and the old ontology. The ‘ingredients’ that composed the
things of the old ontology may then be inventoried, if that needs to be
the question at issue, in other counts of the stuff of the world. In so far as
questions about the old set of entities are answered rather than simply
abandoned, the answer may be given under the new order by bridge
principles (whose ad hoc character it will do no harm to acknowledge) or
even by recourse to considerations (again ad hoc) of a mereological kind.
(ii) Where a shift in conception is a genuine revision of the old con-
ception of the old entities and the same entities are recognizable, i.e.
where the shift is an improvement or development of the individuative
conceptions which were constitutive of the old ontology, the old ontol-
ogy can happily take over the improvement. (Cf. the G property of § of
Chapter Four.) If the deictic-cum-Leibnizian account of concept-for-
mation is correct, then conceptual schemes are designed from the outset
to allow for this kind of improvement.
(iii) Scientific progress towards an explanatorily more fruitful concep-
tion of the world that abandons an old ontology for certain explanatory
purposes may or may not need to discredit the ontology of the older, less
theoretical conception of the world. (Cf. Chapter Six, §.) There will be
 Conceptualism and realism
cases where that older ontology still counts as adequate in its own terms.
The older ontology may yet be cotenable with the more theoretical con-
ception. Contrasting the actual discrediting of entities of some kind, pal-
pable or impalpable, with the discovering of new entities at the atomic
or subatomic level, let us not conceive the latter as determining the level
to which everything else must be reduced (in the serious sense of
‘reduce’18), even if this is the level at which macroscopic events are prom-
ised certain sorts of explanation. Let us note too that these promised
explanations may or may not be forthcoming and may or may not be
completely formulable at that level; and may, even if forthcoming, leave
high and dry but perfectly unharmed the familiar macroscopic entities
in which we cannot abandon our interest. Indeed there are some prac-
tical interests we cannot become blind to, and some entities in which it is
impossible for us to lose our interest (most notably perhaps the entities
to be treated in Chapter Seven).
(iv) A new conception of the world and its accompanying ontology
will not come bare of individuative concepts or of de re necessities of its
own. As always, the thing that generates these de re necessities will be the
requirement that everything conceivable of the new entities be cocon-
ceivable in the new ontology with that which is their most fundamental
sortal identification. Principle ⌬ of Chapter Four concerned a necessity
that is not in the narrow sense logical necessity. But its strength (contrast
certainty) is equal to that of logical necessity. For once the theoretically
fundamental sortal property f is fixed upon and its extension comes to
light, it is not for thought to renege, even hypothetically (lest thought lose
its grasp of its object), upon the determination of how a thing falling
within that extension has to be in order to be an f (belong to f).
18
For the serious sense of ‘reduce’ deployed in philosophy, let me commend the following:
Most people are natural metaphysicians, and it is an easy passage for them from the unassailable
methodological doctrine that physics and chemistry are applicable to biological objects, to the
metaphysical doctrine, that living organisms are ‘nothing but’ physical systems. This . . . retards
the search for explanatory hypotheses on the biological level, just as a purely behaviouristic
approach to psychology retards and discourages the search for other hypotheses in that science . . .
There is one more point to be mentioned in connexion with the doctrine of the reducibility
of biology to physics and chemistry: people who hold the doctrine do not in fact believe it. If you
want to reduce biology to physics and chemistry, you must construct bi-conditionals which are
in effect definitions of biological functors with the help of those belonging only to physics and
chemistry; you must then add these to the postulates of physics and chemistry and work out their
consequences. Then and only then will it be time to go into your laboratories to discover whether
these consequences are upheld there. From the fact that people do not do this, I venture the guess
that they confuse reducibility of biology to physics and chemistry, with applicability of physics and
chemistry to biological objects. ( J. H. Woodger, Biology and Language, Cambridge, , pp. –.
Compare also Alfred Tarski’s Appendix E for Woodger’s earlier work, The Axiomatic Method in
Biology, Cambridge, .)
   

Identity: absolute, determinate, and all or nothing, like no

other relation but itself 1

Wherever identity is real, it admits of no degrees. Thomas Reid ()

Identity is the vanishing point of resemblance. Wallace Stevens

 .                            
When something is singled out, an object of some sort impinges on a
conscious subject, and the subject, in having the de re thought that he has
when he is so impinged upon (a thought that may issue in a claim such
as ‘That bald man has been standing out there in the snow for four
hours’), takes the object for something that it is (a bald man). The subject
apprehends the object in at least one way correctly (even if, in all sorts of
ways that are neither cognitively nor practically crucial, he misappre-
hends it). Let us label this claim (). () is not drafted in order to exclude
animals from the role of subject.
We can also say () that, when the subject singles out an object, his
thought is answerable for its success to the nature and condition of the
object singled out and his thought counts as the kind of thought that it
11
For those sensible enough to have reached this point by skipping the many pages that stretch back
from here to the summary of the first three chapters (see Chapter Three, §), or for those who
simply prefer to begin the book here, I resume at this point one key theme, repeating certain
essentialist and conceptualist claims in order to push them further. (For some of these, see the
partial résumé furnished at Chapter Five, §.) Bearing in mind the interests of those stalwart
enough to have arrived hither by the longer route, the résumé is made in a fresh, albeit highly
partisan fashion. At some points the text derives from my contribution to T. Pettit and J.
McDowell (eds.), Subject, Thought and Context (Oxford, ). See also my ().
A signpost for the reader: first, this chapter seeks to support Reid’s claim and kindred claims
about identity. The chosen route is through a further development of the thesis of conceptual-
ist realism. Then the chapter seeks to confute a general conception of identity apparently visible
in Wallace Stevens’s dictum (torn disobligingly from its context) and in related formulations. (I
cite Stevens from Anne Righter in Shakespeare and the Idea of the Play (London, ), p. .)
Identity will emerge in this chapter as determinate, irreducible, autonomous and not interest-
ingly supervenient on anything else whatever – like unto no other relation.


 Identity: absolute, determinate, and all or nothing
is by virtue of being answerable in that particular way to that particular
object.
At this point, the contentions of Chapter Five will suggest a further
claim (): that, when an object is singled out by a subject in the manner
described in () and (), it will not be determinable without reference to
the content of that sort of thought on the subject’s part what object it
was that impinged upon his mind. For the question of what impinged on
his mind is not a simple question of what material something bumped
into his sensory-cum-neural apparatus. At any given time all sorts of
things will have done that. (Was his thought of a substance, of a stuff, of
an event . . ., and, if of a substance then a substance of what kind, size,
or typical life-span?)
Among philosophers favourably disposed to the conjunction of () and
(), there is one who will deprecate any suggestions which the verb
‘impinge’ may seem to carry of a material-or-causal transaction yet take
keen pleasure in the important (however inadequate) concession to his
viewpoint that he sees in () as strengthened by (). For the sake of
having a name, let us call this philosopher, among the various inheritors
of R, an idealist. (Why an idealist? Compare the however reluctant use
Kant makes of the appellation in the Critique of Practical Reason, Preface,
ad fin.: ‘Names that designate the followers of a sect have always been
accompanied with much injustice: just as if one said that N is an ideal-
ist. For although N not only admits but even insists that our ideas of
external objects have actual objects or external things corresponding to
them, yet N holds that the form of the intuition does not depend on
them, but on the human mind’ (trans. Abbott).)
Another kind of philosopher may be content enough with () but reject
() and be markedly unwilling to allow that it is only by reference to a de
re thought conceived as irreducibly de re that it can be determined what object
impinged on the consciousness of the subject. Let us call this philoso-
pher, who is among the inheritors of not-R, a materialist.2
There is a third stand someone may take. Let us give the name of con-
ceptualist or a conceptualist-realist to anyone who is eager to deny R
and accept () and () together (as literally understood); happy to say that
any full or proper development of () and () taken together would issue
in (); and happy to deny that there is any reductive level (of retinal
12
He is to be numbered among the bare absolutists of Chapter Three, §. In practice, not all bare
absolutists will want to call themselves materialists. Nor will all dissociate themselves vehemently
from every distinctive claim that is comprised under () or (). Here, as in Chapter Three, §,
philosophical stereotypes are useful in exposition, but will not carry any weight in argument.
 Identity: absolute, determinate, and all or nothing 
stimulation or whatever) at which some theory could keep track together
of objects singled out and of thoughts of such objects. (He is likely to
say that, if such objects and thoughts are to be conceived together in the
kind of reciprocity already adumbrated in the conjunction () () (),
then a theory of this kind is impossible.) Such a philosopher is to be
numbered among the D-absolutists of Chapter Three, §. That is my
crowd. His position grows out of the claim developed in Chapter Five
that our cognitive access to objects in reality is made by this such concep-
tions, where this such conceptions were said to be present in advance of
the recognition of any particular object as this or that kind of a thing,
but were seen as wide open to correction in the light of empirical dis-
coveries relating to entities of the general kind to which some particular
thing belongs.

.                     

            
So much for the three contrasted views we have conventionally labelled
idealism, materialism and conceptualism. My hopes rest, of course, with
conceptualism, as developed and defended in the preceding chapters yet
still held open to further refinement. My belief is that, as we explore
further among the logical properties of the sameness relation, this third
doctrine will make it intelligible, in a way that no other doctrine can, how
a relation of such logical exigency as identity can have actual applica-
tion to the world of experience.
The thought of the object is the kind of thought it is (we said) by virtue
of being answerable to that object and the condition of that object. That
is one dependence. But, according to (), as poised to assimilate (),
there is an opposite (though unequal) dependence. The object is what it
is, whether or not it is singled out. But the object does not single itself
out.3 Nor need the simple commitment to engage with nature and that
which is there independently of us require us to single it out. What thing-
kind conceptions we have or deploy is not determined under this kind of
13
Here (I am aware) I yield to the temptation to try to convey something by issuing the denial of
something that is really nonsense. What would it be for an object to single itself out? The claim
may best perhaps be read as a grammatical remark (made in the material mode) following from
the explanation of ‘single out’ given at Preamble §. One more remark about this misgiving.
Even when (on other occasions) I have yielded to allied temptations and tried to convey the point
by denying that there are ‘lines in Nature’, I have never taken this to be a denial that there are
edges in nature, or that edges can be causally efficacious. Indeed they can inflict wounds or abra-
sions. Rather, I have wanted to insist that edges of that kind mark out imperfectly or scarcely at
all the boundaries that are drawn by the singling out of continuants or substances.
 Identity: absolute, determinate, and all or nothing
compulsion. Nature does not reach down and lodge them in a pigeon
hole or letter box for us. But nor yet is the matter determined quite arbi-
trarily. Rather, our store of thing-kind conceptions comes into being
under the influence of our experience, our constitution, our ways of
dwelling in the world and in reciprocity with our active concerns, prac-
tical and intellectual alike. The thing-kind conceptions that we arrive at
in this way play an indispensable role, the conceptualist says, in the busi-
ness of individuation. As always, however, the claim this imports, namely
that ‘which sortal concepts we apply to experience determines what we
shall find there’, needs to be understood in the unexciting way in which
one understands ‘the size and shape of the net determines the catch’.
The size and mesh of a net determine not what fish are in the sea but
which ones we shall catch . . . (for the rest, see Chapter Five, §) – even
as our expectations of what was there to be caught conditioned our
choice of that net and affected at an earlier point some netmaker’s
design for that net.
Conceptualism thus stated will be drawn inexorably to the idea,
already apparent in Chapter Five, that x is only a genuine continuant if,
in a sense of ‘single out’ to be gradually refined, x could in principle be
singled out as this such-and-such or that so-and-so, where the singling
out of x is answerable to multiple constraints that are at once empirical
and logical. It is to that idea that I should have recourse if I were called
upon to justify the assumption I have made throughout this book that,
in a case where it is indeterminate what has been singled out, we should
not say that the thing singled out is something indeterminate. (Would
that not be the merest fallacy?) Rather (I have said) no substance at all is
singled out until something makes it determinate which entity has been
singled out; for this to be determinate, there needs to be something in the
singling out, something from the sides of the object and the thinker,
which would yield a principle of individuation for the entity; so the sin-
gling out of a substance s at time t must reach backward and forward to
any or all of the times, before and after t, at which s exists. What after all
is an indeterminate object? What other idea can we have of such a thing
than that of an entity that is individuatively indeterminate? And what
other idea can that be than the idea of an entity with regard to which it
is indeterminate which thing it is – the idea of a thing that is the crea-
ture of time t, is ‘synchronically’ perfectly identical with itself, but is
‘diachronically’ possibly less than perfectly identical with all sorts of
more or less ‘other’ things? These notions are so strange in themselves,
so apparently ludicrous, that I think one’s first duty must be to show how
 Identity: absolute, determinate, and all or nothing 
readily and easily they can be dispensed with and how unnecessary it
ever was to suppose that the relation of identity could suffer from inde-
terminacy. On the other hand, I know that in philosophy the strangeness
of notions such as indeterminate identity may even lend them some pos-
itive allure. So there is no alternative but to seek to draw out their logical
implications. For the purpose of showing that these are unfavourable to
indeterminate identity, I shall follow a line of argument first pursued in
another form by Gareth Evans.4 (See below §.) Evans’s derivation
points, I believe, in the direction in which the conceptualist is headed
already. Meanwhile though, in advance of the attempt that I shall make
to show that the conceptualist notion of an object is inescapable, it is
time to lower the level of abstraction and have in front of us two or three
would-be examples of an indeterminate object.

.                     

(a) Imagine a monstrous birth apparently possessed of two heads but
having only one trunk, heart, liver . . . ‘It is indeterminate’, it may be
suggested, ‘whether this is one animal or two animals. So, by suitable
pointings supported by suitable explanations, it will be possible to have
a case where it is indeterminate whether this ostended thing (that which
was born, let us say) is the same as that ostended thing (the animal whose
head this is, for instance).’5
(b) ‘Circumstances can arise that make it unclear whether or not a club
started at some date is the same as a club that exists at some later date
. . . The question of identity has no answer; the facts do not determine
one.’6
(c) As we saw in Chapter Three, Hobbes adapts Plutarch and offers us
a story in which Theseus’ ship is gradually repaired with new spars,
while the old planks and spars are simultaneously collected and reassem-
bled in a reconstruction of the ship in its original condition. Suppose
someone wishes to maintain, as I myself have in partial response to such
puzzles,7 that in such cases an entity can persist through some measure
14
There have been disputes about the interpretation of Evans’s article, ‘Can there be Vague
Objects?’ (), p. . And there is no firm consensus about the best way of resolving the incon-
sistency between the two stipulations that Evans enters about the relationship between his ⌬ and
his ⵜ. Leaving these disputes behind, I follow Williamson () and the line of argument given
in my ().
15
Michael Ayers suggested that I must consider such examples and furnished this one.
16
Cf. John Broome, ‘Indefiniteness in Identity’, Analysis,  (), referring to Derek Parfit.
17
For the rest of the response and further discussion, see Chapter Three, §.
 Identity: absolute, determinate, and all or nothing
of disassemblage, loss of parts, reassemblage of these parts elsewhere
and replacement of present parts, but that there are further extremes of
simultaneous disassemblage of the thing, loss of parts to other structures,
or reparation with new parts, that are simply too extreme relative to the
proper reference point, which is the original condition and matter of the
entity – too extreme, that is, for the repaired entity to count as having
persisted. If that is right, it may be claimed (though I am not sure that I
myself should claim this), then there are bound to be relatively late stages
in the story of something’s enduring changes of this sort at which it is
not yet fully determinate whether the original entity has survived or
ceased to be.

.   a              b ,    a 
            b
When we are faced with such examples, it may be asked what there is to
be said in favour of a theory of individuation that does not hasten to
make room for objects that are indeterminate in respect of identity
(objects such that it is indefinite which things they are).
Suppose that a is b. Then, given Leibniz’s Law, whatever is true of a
is true of b. Now a is determinately a. But if so, then b is determinately
a. So, by conditional proof and the symmetry of identity, if a is b, a is
determinately b. In which case there is no future in the supposition that
one could say that a was b but refuse to affirm that it was determinately b.
Putting the matter more formally, and letting ‘⌬’ mean ‘determi-
nately’,8 we have
(i) a⫽b (Hypothesis)
(ii) ⌬ (a ⫽a) (Truism)
(iii) ⌬ (a ⫽b) ( (i), (ii), Leibniz’s Law)
(iv) (a ⫽b) →⌬ (a ⫽b) (conditional proof)9

18
Note that this is not Evans’s use of ⌬. The resemblances between Evans’s argument and the one
now to be presented would need to be taken point by point, and on their merits. Note also that,
if the sign ⵜ were introduced here, it would mean ‘not determinately not’ and, being duals, ⌬
and ⵜ would be consistent. The relation of the two signs would be similar to that of 䊐 and 䉫.
Never mind that there is no one-word translation of ⵜ into English.
19
There may also be doubt about the use of the rule of conditional proof in the derivations of
a ⫽ b → ∆a ⫽ b and a ⬆ b → ∆a ⬆ b. For recent discussion, see Richard G. Heck, Jr, ‘That There
Might Be Vague Objects (So Far as Concerns Logic)’, Monist,  (), pp. –. See espe-
cially p.  following. Heck offers a scrupulous commentary on Evans’s argument with ⌬ under-
stood not as here (see note ). But Heck’s general conclusions, not less his constructivist
inclination, transcend that framework and they will cohere in substance with those defended in
this chapter.
 Identity: absolute, determinate, and all or nothing 
The reader will perceive instantly the parallelism between this deriva-
tion and the Barcan Marcus proof of the necessity of identity given in
Chapter Four, §§–. The parallelism will be equally apparent between
the interpretive issues that arise in each case. If there is doubt about the
kind of context that is created by ‘determinately’, for instance, then it
may be that the doubt is to be circumvented by recognizing a predicate
adverb and forming complex predicables by ␭-abstraction. (See Evans,
and see again Chapter Three, §.)
When he has explored these parallels, the reader will be able to come
to his own view of the prospects of following the paradigm of Chapter
Four and extending the proof of a ⫽b→ ⌬ (a ⫽b) into a proof that a 苷b
→⌬ (a 苷b). Just how hard it would be to deny even the second and
stronger of these claims has been shown by Timothy Williamson.10
(Compare the proof in Chapter Four of the necessary distinctness of
non-identicals.)
Again, it will be equally obvious that the derivation (i) (ii) (iii) ⵫ (iv), is
available for any interpretation at all of ⌬ that verifies premiss (ii) and
preserves the referential transparency of the context it creates. For
instance, if we read (ii) as ‘a is absolutely identical with a (i.e. not as a
matter of degree)’, then we can deduce Reid’s conclusion that if a is b
then a is absolutely b (not as a matter of degree).

. ,      ,             

         
Let it be clear that (iv) should not be understood as entailing that every
true identity sentence will remain true when prefixed with ‘determi-
nately’. That runs counter to the intention of those who advance (iv) as
a truth. For identity sentences may contain descriptions, and indetermi-
nacy of an irrelevant kind may result from vagueness in these descrip-
tions. Consider the not altogether implausible claim ‘The greatest ruler
was the wisest ruler’. There is indeterminacy in this claim. At the outset,
it may be insufficiently clear what wisdom or greatness are, as they figure
here. And, even when that is made clearer, it may not be open or shut
who is the greatest or who is the wisest. The one thing to be insisted upon
and read as following from (iv) is only this: once we prevail sufficiently
over indeterminacy by elaboration of the signification and point of the
10
See Williamson () in Lovibond and Williams (). In ‘Replies’ (in Lovibond and Williams
()), where Williamson seeks to persuade me of the necessity of difference, I abandon previ-
ous reservations and accept the proof that he offers. See also Chapter Four, footnote .
 Identity: absolute, determinate, and all or nothing
original judgment, and once we identify the candidates to satisfy this
description and that description, we shall find that, if a is b at all, then a
is determinately b. That matter is open or shut.11
At this point, and allowing (they will say) for the explanation just
given, opponents of (iv) may yet argue that, if a is a vague or indeter-
minate object, then we cannot claim that a is determinately a. It is
here they will say (that is, at (ii)) that all derivations of principles like
(iv) are bound to fail. But to this I should reply that, even if identity
were a matter of degree and a were an indeterminate object, we still
ought to be able to obtain a (so to speak) perfect case of identity, pro-
vided we were careful to mate a with exactly the right object. And
surely a is exactly the right object to mate with a. Their indeterminacy
or vagueness matches exactly.12 (Here I aim to speak in terms these
opponents will find it more difficult to reject as unintelligible than I
need to find it.)
A more sophisticated philosopher will intervene here with a different
reservation: ‘I believe that the world leaves many things vague or inde-
terminate, even about substances. But that does not mean that I shall be
found rejecting the principle (iv) that you claim to have proved. The
thing I say is that the world can leave it vague or indeterminate whether
a is the same as b. I do not say that a can be the same as b but indeter-
minately so. Who on earth would want to say that? Rather I say the
world may leave it indeterminate whether a is the same as b at all. This is just
one of the many consequences of the claim to which I really do sub-
11
We may say, if we wish, that the ‘determinately’ in ‘If a is b then a is determinately b’ has a smaller
scope than any description that stands in the places marked by ‘a’ or ‘b’. Thus the central ques-
tion about the determinacy of the relation of identity will evade the approach of, for example,
Richmond Thomason, ‘Identity and Vagueness’, Philosophical Studies,  (), pp.  ff. In
making this proposal about scope, I follow Arthur F. Smullyan ().
12
I am going along with the thought (which is the sceptic’s and not mine) that a vague object needs
to be an indeterminate object. As will appear, I am not an enthusiast for that, either as a thought
about ‘vague object’ or as a thought about ‘indeterminate object’. See text below and note .
A related idea, which may exert some adverse influence in this dispute and may make (ii) seem
more dubious than it ought to seem, is that ‘determinately’ cannot be prefixed at all, without
destroying full truth, to a sentence in which vague expressions occur. But consider familiar cases
innocent of all suggestion of indeterminate objects and well outside the disputed area, where
indeterminacies manifestly trade off:
Determinately, it is raining if and only if it is raining.
Determinately, the evening star is as famous as the morning star is.
And a similar response may be directed against the complaint that Leibniz’s Law is suspect
within this controversy, because (as someone has said to me) it was never formulated to manage
vague predicates. Any vagueness in the open sentence for which ‘␾ ’ holds a place in the equiv-
alence ‘(␾a ↔ ␾b)’ within the Leibnizian scheme a⫽ b→ (␾a ↔ ␾b) will surely trade off against a
similar vagueness in the same open sentence on the other side of the equivalence.
 Identity: absolute, determinate, and all or nothing 
scribe, namely that we live in a world of real and considerable vagueness
and indeterminacy.’
Such a protest may seem to invite cross-examination under the triple
auspices of the principle (iv), of the principle of contraposition trans-
forming (iv) into the thesis that, if a isn’t determinately b, then a isn’t b
at all, and the principle (as secured by Williamson) of the determinate
difference of things that are not identical, a ≠ b → ∆ (a ≠ b). Even without
the benefit of the principle of classical logic (a ⫽ b) v (a ≠ b), the issue is
not in doubt. Indeed, if the contestants simply limited themselves to (iv)
and the weak form of contraposition that is permitted in intuitionist
logic, the issue still would not be in doubt. Surely then the space that the
sophisticated philosopher tries to point to does not exist. Or so says one
in my theoretical position. But the thing that has become apparent in the
years since the publication of Evans’s article is that, given the endless
possibilities for variant logics of indeterminacy, the cross-examination
just described may not be the right way for a conceptualist to create con-
viction. In order to undo the present deadlock, it may be better to focus
upon three other questions.
The first question relates to the status of the presumption that the
vagueness whose existence nobody doubts entails indeterminacy. There
should be no such presumption.13
Second comes the question of what account the sophisticated philos-
opher is assuming of what ‘the world’ is. Of itself, in the absence of a
13
For the dissociation, see Williamson ( and ). For Williamson, the question of vagueness
and identity is one battle in a larger and longer campaign against a host of semantical proposals
that engage with vagueness and borderline cases by modifying the scheme of valuations that clas-
sical logic furnishes for reference, predication and judgment. Questioning the coherence of these
alternatives, Williamson has defended the determinacies of the classical scheme and interpreted
vagueness as a phenomenon not calling for new kinds of valuation but as a product of the pre-
dictable unknowability within certain subject matters of cut-off-points. An expression will be
vague, according to Williamson, if it generates borderline cases where such ignorance is to be
expected.
On this kind of basis, one might want to say that being such as to generate borderline cases
of this or that kind was one feature among others of the sense of certain sorts of predicate; that
from this feature of sense nothing automatically follows about these predicates’ reference or
semantic value, or about the valuation of sentences involving them. One might also be tempted
to think that, once this sense/reference distinction is enforced, there is nothing in the phenom-
enon of vagueness to support a presumption to the effect that reference or truth themselves
might have graduated or non-classical valuations.
Since the question chiefly at issue in this book is not the sustainability in general of
Williamson’s position but the specific question of identity and the interaction between identity
and vagueness, there is no evident need to enter into Williamson’s general thesis. The thing I
shall aim to supply for this special case, however, is that which others may ask Williamson to
supply much more generally, namely an account of what will ensure determinacy or supply
determinacy positively to a subject matter that trades actively in considerations importing vague-
ness.
 Identity: absolute, determinate, and all or nothing
significant question with a point and a clarifiable sense, the world deter-
mines nothing. In the presence of such a question, however, and such a
sense – and that is a part of the point of introducing the contribution of
the individuating subject – there is no predeterminable limit to what
might be objectively determined.
Third there is a more specific question, which is the chief question
here. If one of the main points of the objector’s intervention is to defend
the picture of a world of indeterminate objects, what sort of things will
such a world contain?
It may be said that one candidate is some mountain the ordinary indi-
viduation of which leaves over numerous questions of the form ‘Is this
foothill a part of x? Is that foothill a part of x?’ But this is a strikingly
poor illustration of what would be needed. For it can be perfectly determi-
nate which mountain x is without x’s extent’s being determinate. A moun-
tain is not, after all, something essentially demarcated by its extent or
boundary.14 It is not as if there were just as many mountains to be found
with x’s peak as there were rival determinations of x’s boundary. An idea
like that could not even occur to one with the good fortune to be inno-
cent of classical extensional mereology.15
If a mountain of indeterminate extent is a bad model for an indeter-
minate object, then how is the idea of an indeterminate object to be lent
significance? We are thrown back on the bare and abstract specification
which says that an indeterminate object is an object that is such that it is
indeterminate which object it is. But if the opponent’s plea is that, in this
sense, there are indeterminate objects, then my response is as follows: let
x be such an object. Surely it is not indeterminate that x is the same
object as x. Whatever else may be indeterminate about x, x is determi-
nately x. Suppose then the question arises whether a certain object y is
identical with x. The candidate y can only be x if y has every property x
has. One of the object x’s properties is that of being determinately the
same as x.16 Unless y has that property, y isn’t x. So it seems that, what-
14
Contrast perhaps a sea, and all the special difficulties that flow from the plausible idea that a sea
is simply some roughly designated area of water. How serious though (one is moved to ask) is our
existential commitment to seas as such? Contrast our undoubted commitment to quantities or
expanses (e.g. of sea water). These do admit of arbitrarily exact determination. Nb. also the
falsity of D(ix), explained in Chapter Two, §. On these matters, see also R. M. Sainsbury,
Concepts without Boundaries (Kings College London, Dept. of Philosophy, ISBN     X,
).
15
See Chapter Three, note . ‘Classical extensional mereology’ is Peter Simons’s term. See
Simons ().
16
Among other points, surely this was Evans’s point, or one of them. See also the discussion by
Nathan Salmon in Reference and Essence (Princeton University Press, ) at pp. –.
 Identity: absolute, determinate, and all or nothing 
ever candidate is presented for identity with x, it can in principle be
resolved whether the candidate is identical with x. But now – what can
remain of the idea of an object such that it is indeterminate (rather than
merely unknown) which object it is? And what remains of the idea of a
world of indeterminate objects? In the end, I think we must set more
store by questions such as these than on the question whether, in strict
canonical logic as such, the principles of the determinacy of identity and
determinacy of difference can or cannot be vindicated as theorems in a
manner that no party to these disputes will have any right at all to regard
as question-begging. (By this time, there are too many such parties.) How
then are these inquiries to be pushed forward?
I could scarcely prove that there is only one way forward. But now is
surely the moment for the conceptualist to offer an alternative concep-
tion. The world is not an object that is determinate or indeterminate. Nor
is it an assemblage of indeterminate objects. If something simply has to
be said in the form ‘the world is . . .’ (and how could it be shown that that
is a legitimate demand?), it may be better to say ‘the world is simply the
possibility of singling out and referring to actual things which have prop-
erties and relations’. Furthermore, if that will do for the present, then this
is the moment for the conceptualist to deploy the plausible idea that every
judgment can be filled out ad libitum from the context of its cognitive for-
mation; then to explain positively and constructively (quasi-construc-
tivistically perhaps) how it can be that every well-made singular term with
a reference has a determinate reference; then to explain how it can be
that every question that is genuinely an identity question17 is determinate
for an answer known or unknown; and how it can be that (understood in
the manner described in the first part of this section) every identity that
holds at all holds fully determinately and indeed absolutely.

.         ( a ) (b) ( c )   § 

Before any such explanation is attempted, there is illumination to be had
from the consideration of the three examples (a), (b), (c) already pro-
vided. I hope that examination of each example will lend plausibility to
the claim that the contextual supplementation spoken of in §, para-
graph one, is already one part of our standard practice. That practice
17
A question may be none the worse for not being in my exigent sense an identity question. The
continuum question has the appearance of an identity question. But the appearance is inessen-
tial. Ask instead whether a totality with a larger cardinality than the totality of natural numbers
can have a cardinality smaller than the totality of the real numbers.
 Identity: absolute, determinate, and all or nothing
deserves an emphasis independent of the theorem status I would claim
for our determinacy principles.
Example (a) is the birth comprising two heads, one trunk, one heart,
one liver. . . . In a case such as this, further inquiry may suggest that the
right thing to say is that we have here one non-viable animal and it has
two heads. Or it may suggest that there are here two animals, each as
capable as unseparated Siamese twins can be of separate life and inde-
pendent conation. In that case the totality of that which was born com-
prises two animals.18 So far, there is no relevant sense in which we have
an indeterminate object. All we have is an event of giving birth that
either issued in one animal or else issued in two animals. To say that
much is not yet to say that we have a thing singled out and it is indefinite
which thing it is. But now someone may ask this: what if there is simply
no point in insisting that one of these options is the better one? And what
will one in my theoretical position say if someone exploits the supposed
equal acceptability of those options to declare that it is indefinite
whether that which was born is or is not the same as that object (that two-
headed animal) and indefinite again whether that which was born is the
same as the animal to whom that [indicated] head belongs? My reply is that, if
my opponent says this sort of thing, then the time has come to protest
that, even if (ex hypothesi) you can take each of two options, that does not
entail you can take both at once. If there is a two-headed animal in the
offing, then what was born is or was a two-headed animal. Relative to
that decision the identity is definite. Or if that seems wrong because the
associated identity is not definite (e.g., because a mere decision is not
enough to make it definite), then there is no identity. But that entails that,
so far from the options being equally good, the other specified option is
better. If so, we have established that that which was born comprised two
animals. (Compare again the case of inseparable Siamese twins that
survive into maturity.) More generally, it is just an illusion that there is a
this we can avail ourselves of here which will make as good a reference
as you like, as well-backed as you like, and make this reference regard-
less of whether the identity question still remains indeterminate.
Example (b) was the club. Suppose it is alleged that the association of
persons meeting in  is indeed the same as a club that appeared to its

18
Perhaps then we have an animal that comprises two animals. This is simply a case of homeomery,
not of indeterminate objecthood. For individuation and number concepts, see Chapter Two, ad
fin.
For more general issues that arise here about ostension, see Chapter One, especially § and
§.
 Identity: absolute, determinate, and all or nothing 
members to be lapsing (or to have lapsed) in . The first question is
whether there is any good reason to allege this identity. It cannot be
made to hold by a simple decision on the part of those meeting in 
(even if decisions on the part of the founders are relevant to the ques-
tion). On what terms, if any, was the club dissolved in ? What had
been the articles of association? The thing that really needs to be made
out is the point of the claim of identity, and the case for the identity. If
the case is not good enough, then the thing that we are prompted by (iv)
to say is that the clubs are not the same. Why object to that?
Example (c), Theseus’ ship, is a more complex matter (as we have seen
already in Chapter Three, §). But what does the question of identity
turn on here? Without using the idea of identity, one cannot say. But
using it freely and imagining a set-up where we are not unduly hampered
by problems of unknowability, perhaps one can contrive to say some-
thing at least. First, we have a reference to Theseus’ ship; and then (let
us suppose) we have a reference to the ship whose late return in  
delayed Socrates’ drinking of the hemlock. Behind each reference stands
the singling out of the object of that reference. What exactly did each of
these singlings out catch hold of ? To determine that, we have to find
some common way of further amplifying, developing and specifying each
singling out,19 reaching backwards and forwards along the life span of
each object, working towards a more and more complete answer to this
question. (Advice: it will save work and serve clarity to do what we nor-
mally do without thinking, namely begin work on the object that is spec-
ified by reference to the earlier time.) In the end, either we shall find that
we have rendered it completely manifest that, as fully spelt out, these are
the singlings out of one and the same ship, or else we shall find that we
have collected an x and y such that, for some genuine predicable ␾ that
is true of one or other of x and y, it holds that ␾x and it does not hold
that ␾y.
On what principle, then, must we determine the temporal extent of
the life span of Theseus’ working ship? Well, it depends in the first
instance on what ships are and what that ship was. We have said that
one can allow for the occasional disassembly of the ship; that one can
allow for a measure of replacement of parts; that one can allow for
modifications of the ritual or religious functions of the ship, with or
without structural alteration. The thing one cannot allow (or so it was
19
This stipulation, which ensures the compatibility and commensurability of the singlings out as
further amplified, corresponds to the ‘there exists a sortal concept f such that . . .’ clause in the
statement of principle D given in Chapter Two above.
 Identity: absolute, determinate, and all or nothing
declared in Chapter Three), if Theseus’ ship is to be that which is being
singled out, is arbitrarily much of all three kinds of change. (If we do
allow arbitrarily much of all three, we may find we have let in
Hobbes’s competitor ship.) Perhaps each such change is pro tanto insig-
nificant. But that which matters at each point is the overall distance of
the new condition at that point from the original condition of the ship
that Theseus dedicated to Apollo. (If we are careful about this, if we
recognize that any real candidate to be the ship whose late arrival
delayed Socrates’ death in   must inherit the answerability of
that   ship to the condition of Theseus’ own ship, if we insist also
that not arbitrarily much of the matter of Theseus’ ship can be dis-
carded, then nothing can colourably emerge as a rival ship during the
lifespan of the ship.) And in the end we must say no (or refuse to say
yes) when too many such changes are too variously combined with one
another or too much of the original matter is lost from the ship. The
better our understanding of what is at issue, the more exact our sense
will be of when to say no.
Needless to say, this way of regulating Sorites reasonings respects the
vagueness of predicates that are vague. Identity is not one of them. It
respects also the graduability of predicates that can qualify a thing to
some degree. Identity is not one of these either. Note also that that which
the determinacy of identity enforces is that each and every positive deci-
sion of identity must be judged by the same strict standard as we ought
to require for determinate (and absolute) identity.
That ruling is tailored to identity. But what is so unfair or so unman-
ageable about it? Suppose that, in reaching forward from the origin to
later times, we get to a point t after which, using that backward-looking
standard, we cease to be satisfied that the very same thing is being singled
out. Then we shall say that what we have at t is indeed Theseus’ ship.
But, when confronted with something picked out by reference to times
later than t, we shall refuse to affirm that Theseus’ ship is this ship.
Contraposing (iv), it will then be open to us to find that Theseus’ ship is
not identical with any item we are presented with that is found after t.
Before we thought, we might have been in doubt. But having thought
about what identity involves, it will be apparent to us that there is no
alternative but to permit the logic of identity to regulate our individua-
tive thinking. By this logic, a ship found after t has to be a new and differ-
ent ship that emerged from the process of successive remodellings of a
ship that was older and different. Provided we see this decision as itself
arising from the exigencies of singling out a continuant that has a history
 Identity: absolute, determinate, and all or nothing 
and an original condition that cannot be read off it at a glance, the deci-
sion will appear reasonable and inevitable.

.       ;                  

Evidently, as one engages with putative examples of objects that are
indeterminate and one presses certain questions about them, the posi-
tion one finds oneself moving toward is this. If that which presents itself
as a simple identity question has a real point and this point is the point
attaching to a veritable identity question that is open and in good
standing,20 then the question must relate to objects each of which could
be identified in its own way. In that case, the sense of each of the terms
that figure in the question will allow of some sufficient completion for
the term to engage with some conception of an object that is singled
out and marked off from all other things. My caricatural idealist is
wrong if he supposes that a conception of this sort is in any sense the
free creation of the thinker. It is empirically conditioned and a posteri-
ori at every point. But the idealist is not wrong to call the conception of
each object – as opposed to the object that answers to the conception
– the work of the mind. Moreover, that mind-work is not only empiri-
cally constrained. It is constrained by the determinable idea of an
object, by Leibniz’s Law, and by anything else at all that Leibniz’s Law
can deliver to us.
According to this picture, the sophisticated philosopher who said he
was prepared to accept (iv) but went on to declare that it was still open
to him to maintain that the world might leave it indeterminate whether
object a was object b, invokes an idea of object-hood which is as difficult
as it is dispensable. If it does not make sense to speak of such objects, if
every putative example disappears on examination, then one ought not
to allude, even indirectly, to the possibility of such objects. Not only that.
According to the conceptualist, the kinds of entity invoked by the sophis-
ticated philosopher are entirely superfluous. In describing how a thinker
who singles out objects finds these things in the world, we do not need
to say that he conceives or fashions them out of preconceptual objects.
As theorists, we do not need to speak of indeterminate things at all in
order to do justice to the thinker’s conceptual activity. Nor yet do we
need to abandon realism. For even if, pace the materialist, individuation
20
Contrast the question about the wisest and best ruler when it is reconstrued as a question about
the relation (of subsumption or whatever) between certain virtues and excellences. See also note
.
 Identity: absolute, determinate, and all or nothing
is (as the idealist says) mind’s work, this work, in so far as it is construc-
tive, lies entirely at the level of sense. It is work of construal, of making
sense of our experience, of seeking things and finding them or of
making oneself ready to happen upon objects and recognize them for
what they are.
No doubt, as we have said already, different practical or cognitive pur-
poses will make different sorts of construal necessary, and these may be
expected to issue in different ‘versions of reality’. But (pace the Nelson
Goodman of Ways of Worldmaking, to whom, however, not everything in
the present chapter ought to be alien or unwelcome), these cannot be
allowed to be conflicting versions. Emancipation from the correspondence
model of truth is one thing. Emancipation from truth or logic is another.
If two versions conflict, not both are right. But in order to conflict, they
must have the same subject matter. In so far as I am allowing for differ-
ent versions with different a posteriori conceptions applied for different
practical or cognitive purposes, it is sensible for me to expect that differ-
ent things will be singled out and different subject matters will be pro-
posed. Finally, I point out that, if there are different versions of reality,
that does not mean that anything and everything is a correct version (or
a version correct by its own lights).21
Every significant sentence, by being the sentence it is and having the
contextually completable sense it has, sets itself a goal that it either
attains or fails to attain. But, in doing this, it does not itself make the
verdict on whether that goal has been attained. In the same way, every
singling out of a thing, being the singling out of this f or that g, construes
reality in a certain way. But it does not construct the f or the g from its
own a priori materials, as if to validate itself. It no more validates itself
than the focusing of a camera fixed deliberately at a certain spot creates
that which the camera records when set there at that focus. A de re
thought that is a singling out thought uses tested a posteriori materials to
set a standard for itself that the world can disappoint or gratify, in accor-
dance with whether there is or is not to be discovered the very object the
thought purports to be a thought of. Even if it takes a particular sort of
empirically and logically constrained search to light up the object, the
object is not posited into being. The mind recognizes by the exercise of
conceptions the object that is there to be recognized, even as the object
21
None of this would be worth saying at such length if so many would-be modern persons had
not got excited at the prospect of coming to conclusions I am warning against or they did not
take the conclusions they threaten us with for a liberation or a triumph. As things are, I shall go
on remorselessly, for two more whole paragraphs.
 Identity: absolute, determinate, and all or nothing 
impinges upon the mind that has the right sort of thought to single out
that object.22 (See Chapter Three, at the end.) Finally, then, the world
can disappoint or gratify the claim that object a, singled out under one
mind-originated conception, is the same as object b, an object singled
out under another mind-originated conception.
In sum, once we differentiate that which a judgment contrives by
mind’s work to say from reality’s verdict on what is said, once we distin-
guish sense from reference or truth-value, sober conceptualism is indis-
tinguishable from a sensible version of realism. (Compare the
conclusions of Chapter Five.) Indeed, once we distinguish sense from
reference, individuative constructivism emerges as the one form of
realism that explains how reference can be determinate, explains how
veritable identity questions can be determinate for an answer, and
explains how whatever identities actually hold can hold both absolutely
and determinately.

.       ,             

                                      ‒
                      
   
So much for the role of conceptualism as it conditions the concept of an
object seen as the correlate of an act of singling something out. And so
much for the way in which the relation of identity between objects
admits neither degrees nor approximations. (For more on the latter, see
below, §.) If it is a pre-condition of such a relation as this coming to
seem exemplifiable and possible that there be some freedom that one
enjoys at the level of conception, then now is the moment to fill out the
account (already begun) of that which limits that supposed freedom and
effortlessly contains it.
In the third and last edition of Reference and Generality (Cornell
University Press, Ithaca and London, , see p. ), P. T. Geach
refined a puzzle he had first proposed for discussion many years earlier.23
22
Even in the case where the thing singled out simply confronts the thinker, rather than being the
object of his search, there has to be some however inchoate conception of things of the kind that
is instantiated by the object. There needs to be some sort of readiness. In the absence of that,
think how easily one can look directly at something without seeing it. Of course, one can also
see a thing without knowing what it is. For the sort of account I should offer of that, see ‘Reply
to Fei Xu’ (a).
23
Professor Geach communicated the puzzle to me in  or  and no doubt to other philos-
ophers. For my then response, see ‘On Being in the Same Place’ (), p. . Geach traced the
idea for the puzzle to something he had found in William of Sherwood. For earlier versions of
 Identity: absolute, determinate, and all or nothing
The fat cat sat on the mat. There was just one cat on the mat. The cat’s name
was ‘Tibbles’: ‘Tibbles’ is moreover a name for a cat.24 – This simple story leads
us into difficulties if we assume that Tibbles is a normal cat. For a normal cat
has at least , hairs. Like many empirical concepts, the concept (single) hair is
fuzzy at the edges; but it is reasonable to assume that we can identify in Tibbles
at least , of his parts each of which definitely is a single hair. I shall refer to
these hairs as h1, h2, h 3, . . . up to h1,000.
Now let c be the largest continuous mass of feline tissue on the mat. Then for
any of our , cat-hairs, say hn, there is a proper part cn of c which contains pre-
cisely all of c except the hair hn; and every such part cn differs in a describable way
both from any other such part, say cm, and from c as a whole. Moreover fuzzy as the
concept cat may be, it is clear that not only is c a cat, but also any part cn is a cat: cn
would clearly be a cat were the hair hn plucked out, and we cannot reasonably
suppose that plucking out a hair generates a cat, so cn must already have been a cat.
So, contrary to our story, there was not just one cat called ‘Tibbles’ sitting on the
mat; there were at least , sitting there! Of course this would involve a great deal
of overlap and sharing of organs among these , cats, but logic has nothing to
say against that; after all, it happens on a small scale between Siamese twins.
All the same, this result is absurd. We simply do not speak of cats, or use
names of cats, in this way; nor is our ordinary practice open to logical censure.
The verdict of absurdity is if anything understated. By this account of
matters, Tibbles, who is an ordinary cat, must have among his proper
parts at least , cats. By similar arguments, will not just any Tib that
Tibbles contains prove to contain cats among its proper parts?
Geach’s response to the questions raised by Tibbles is as follows-:
Everything falls into place if we realize that the number of cats on the mat is
the number of different cats on the mat; and c13, c279, and c are not three differ-
ent cats, they are one and the same cat. Though none of these , lumps of
feline tissue is the same lump of feline tissue as another, each is the same cat as
any other: each of them, then, is a cat, but there is only one cat on the mat, and
our original story stands.

footnote  (cont.)
the puzzle that have become visible as a result of Geach’s discussions, see Stoicorum Veterum
Fragmenta, ., and the commentary thereon given by David Sedley at pp. – of ‘The Stoic
Criterion of Identity’, Phronesis ().
24
Geach’s explanation of this usage may be reproduced as follows: a name for an A can be
explained as a name associated with the criterion of identity: Same A. Whereas a name of an A
names something which is an A, but need not be associated with the criterion of identity.
Chapter One (see also my reply to Noonan in ‘Replies’ (Lovibond and Williams, )) will
have made it clear that, on my view, one who uses a name n for an A object x is committed to
the possibility of making it determinate which of the As he has picked out by the name n and is
equally committed to the possibility of making it determinate that it is to x rather than the matter
or the parts of x that he is referring. (Cf. Chapter One § on the ‘is’ of constitution.) But to make
that determinate is to commit oneself to the principle of individuation for x. In practice there is
simply no need for a name of an A that is not a name for an A.
 Identity: absolute, determinate, and all or nothing 
Thus each one of the names ‘c1, c2, . . . c1,001’, or again the name ‘c’, is a name
of a cat; but none of these , names is a name for a cat, as ‘Tibbles’ is. By
virtue of its sense ‘Tibbles’ is a name, not for one and the same thing (in fact, to
say that would really be to say nothing at all), but for one and the same cat. This
name for a cat has reference, and it names the one and only cat on the mat; but
just on that account ‘Tibbles’ names, as a shared name, both c itself and any of
the smaller masses of feline tissue like c12, and c279. For all of these are one and
the same cat, though not one and the same mass of feline tissue. ‘Tibbles’ is not
a name for a mass of feline tissue.
So we recover the truth of the simple story we began with. The price to pay
is that we must regard ‘ _______ is the same cat as _______ ’ as expressing only a
certain equivalence relation, not an absolute identity restricted to cats; but this
price, I have elsewhere argued, must be paid anyhow, for there is no such abso-
lute identity as logicians have assumed.
That, no doubt, is one way to answer. But for those who reject R and
affirm Leibniz’s Law,25 there must be another response. I think the best
response would be one that adds nothing to the description of the puzzle
situation, but shows instead how something needs to be taken away. My
suggestion is that one should scrutinize carefully the purported defini-
tion of cn. Our freedom to single out this rather than that was not the
freedom either to remake the concept cat or to reconfigure its instantia-
tions. Suppose you take the concept cat as you find it, you look for its
instantiation in object c901 and you count c901 as a cat. Then you are com-
mitted to track down all of c901.26 But that (if it is anything at all) is
nothing lesser than Tibbles. If Tibbles has such a hair as h901 (or has a
tail), then you cannot define c901 into existence as the cat that lacks the
hair (or lacks that tail). For there is no such cat.
At this point, the reader will remember the analogous claim that was
advanced in Chapter One, §, namely that, if John Doe was a boy and
later a man, there is no possibility of picking out or fixing upon an
object, a boy, John-Doe-as-a-boy, that is distinct from the substance John
Doe, who grew old and was by  an old man. (This is not to say that
one could not in  have picked out John Doe as a boy. But here the
25
My response to Geach’s charges against the coherence of Leibniz’s Law and absolute identity is
given at ‘Replies’ (Lovibond and Williams, ), p. .
26
There is here a faint but interesting analogy with something that Robert Nozick has urged. See
Philosophical Explanations (Oxford, ), p. . The point that I think I see there in Nozick is in no
way proprietary to the Closest Continuer theory, defended by Nozick. The point I think I see in
Nozick’s discussion (see especially his citation from Tversky) relates to the level of conceptions
or senses. It is a misunderstanding of the concept cat (and a failure to grasp the point of the concept)
not, in applying it to the set-up Geach describes (namely, Tibbles on a mat), to deploy the whole
reach of the concept. It is worse still, having denied the concept its full application, to institution-
alize this misdemeanour in the manner of Geach’s definition of cn.
 Identity: absolute, determinate, and all or nothing
‘as a boy’ is to be read in construction with the verb ‘pick out’, not with
‘John Doe’.)
The technical sophistication of those who tinker with the proposals I
am opposing here is completely out of scale, I suggest, with the small
heed they are prepared to pay to the most invaluable commonplaces of
the logical theory of definition, not least its admonitions concerning
creative definitions.27

.     (     )             

           
At this point, at least one sceptical response is to be expected against the
claims entered in the course of the preceding sections: ‘In this chapter
you purport to be showing, against the current of much present day
opinion, that identity is determinate, all or nothing, and much else
besides. You also purport, by constrained redeployment of an idealist
insight, to explain how identity and difference can be all these things. But
your whole effort depends upon something you will scarcely permit to
be scrutinized – the assumption that the items to be singled out in asso-
ciation with the terms of an identity sentence must themselves be con-
tinuants to whose earliest and latest moments all true reference must
reach out. You pay no attention to the conflicting opinion that the real
purport of an ordinary judgment of identity over time is to say that the
references of the two terms (thing-moments, phases or whatever one
chooses to call them) stand in a “unity” relation, the relation of making
up some given whole. This option has been described many times in the
literature of identity in a manner that you will say is alien to your con-
cerns.28 But it can also be described in terms congenial to your sortal
approach:
‘The sortal [predicate] helps identify the referents of a statement by giving the
relation that holds between occurrences – parts, instances, stages, phases, etc. –
of the appropriate sort when they are occurrences of a single object of the indi-
cated kind: specifying this relation and designating an occurrence identifies the
referent. When an occurrence is identified but no sortal provided, indefiniteness
occurs because an occurrence can be an occurrence of different kinds.’29
27
Here, as elsewhere, I would refer to Chapter  of Patrick Suppes, Introduction to Logic (Princeton
N.J., ).
28
See, for instance, Rudolf Carnap, Introduction to Symbolic Logic and its Applications (New York, ),
pp. ff., , ff.).
29
John Perry, review, Journal of Symbolic Logic,  (), p. . Compare his article ‘The
Importance of Being Identical’, in The Identities of Persons, ed. A. O. Rorty (Los Angeles, ).
 Identity: absolute, determinate, and all or nothing 
In the light of this thought (the objector continues) why should it not be
suggested that a statement of diachronic identity such as “a is the same
horse as b” qualifies as true if and only if the stage or occurrence a and
the stage or occurrence b make up not just any old concatenation but a
suitable concatenation of horse-stages? Understanding the substantive
“horse” involves understanding what sort of a concatenation this has to
be. Can you not recognize here a point that you yourself want to urge –
albeit in different words? (Indeed I would refer you here to your own
().)’
Anyone who makes such an objection will probably go on to point out
how cannily such an approach evades the claim, of which so much has
been made in this chapter and elsewhere, to the effect that reference and
singling out must stretch all the way back to the first moments of a thing
and arbitrarily far forward, however proleptically, towards its last.
My response is this: the view cited by the objector seems to be all of a
piece with the widespread idea that first, and supposedly unproblemat-
ically (however singular terms are to be construed), there is synchronic
identity, as in ‘ ⫹ ⫽’, ‘The Dome of the Rock is the Dome of the
Rock’ or (falsely) ‘Paris is Rome’; and then, differently and/or more
problematically, there is diachronic identity, as in ‘the man in front of me
is the same person as the boy I shared a locker with at school’ or (?)
‘Hesperus is Phosphorus’. It would be good, though, to know the ratio-
nale for this differentiation. Is it founded in the claim that, insofar as we
refer and refer deictically at t, nothing can be involved in the deixis that
is not actually there and present at t?30 If that is the idea, then (rather
than object, as in fact I do, by protesting that this is an untenable doc-
trine of reference, that bare words at a time cannot of themselves fix
their intended interpretation, that the formulation itself of the doctrine
is radically unstable, etc), I shall say it leads into three doubts or difficul-
ties.
‘My candidate for the analysis of personal identity . . . is the unity relation for persons, that rela-
tion which obtains between two stages if and only if there is a person of which both are stages’
(p. ).
I note for the record that this is not the account of these matters given nowadays by the oppo-
nents of the substantialist position that I occupy. For this, see David Lewis, op. cit. below at note
. In Lewis’s version, it is far from clear that the position now called perdurantism would be at
all exercised to deny my claim that singling out reaches all the way back to the first moments of
a thing or arbitrarily far forward towards its last. For that is what a sum of horse-stages precisely
does. As regards Lewis, however, I do question the reason that Lewis gives for importing talk of
stages (see note  already cited). Moreover, the second part of the third objection against Perry
raises serious questions about the coherence of the very idea that horse-stages could play the role
Lewis supposes that they play in predication or individuation.
30
The Doctrine (Specious) of the Deictic Indispensability of the Deictically Present?
 Identity: absolute, determinate, and all or nothing
First, one wonders about the status of a horse-stage concatenation as
itself a potential object of reference. Is this not a continuant?31
Secondly, one wants to protest that, in English, for which sameness is
just sameness, where time-specifications cling to verbs or qualifications
not to nouns, there is no trace whatever of any bifurcation among the
senses of ‘is’ or ‘is the same as’.
Thirdly, one objects that these occurrences/parts/instances/stages/
phases, etc., seem multiply problematic. For instance, not everything that
looks for a moment or behaves for a moment as if it were a horse is a
horse. See Chapter Three, §. Can anyone truly imagine the business of
recasting into stage-talk the sorts of explication invoked there in connex-
ion with natural kind substantives? The criterion for being a horse is
essentially dispositional and diachronic. If horse-stages made sense at
all, it would be easier to go from horses to horse-stages than to go, as the
objector wishes, from horse-stages to horses. Nor is there any single
recipe one can imagine for transposing the normal range of attributions
to horses into attributions to horse-stages or horse-occurrences. (There
is no recipe the mastery of which could be plausibly attributed to a
thinker versed in the language of substances and their states, disposi-
tions, etc.) Indeed, the whole resolve to postulate horse-stages and
deploy them in complication of the truth-conditions of identity-
sentences can only remind one of the conceptualist excesses criticized in
the preceding section.
I should say that, in this province of inquiry, reasoned persuasion is
relatively rare. But let me set out an argument once put forward against
substantialist positions like mine by D. Gabbay and J. M. Moravcsik.32
31
Two possible Answers: () The concatenation belongs to a different system of ideas from that of
continuants. (See discussion later.) () One could indeed refer to a concatenation of horse-stages,
but such a reference would not be properly deictic. As regards (), I would ask: why can one not
refer, whether ‘deictically’ or not, to the person who as a boy shared a locker with me and is now
at such and such a spot in December ? If one can do this, then what objection can the theory
of reference provide against an ontology of continuants that makes no provision for stages?
32
See ‘Sameness and Individuation’, Journal of Philosophy, . (). A quite different kind of
argument has now been put forward by David Lewis in On the Plurality of Worlds (Oxford, ),
pp. –, and in ‘Is there a problem about persistence?’ Proceedings of the Aristotelian Society, Supp.
Vol.  (). Lewis’s argument is answered by Mark Johnston (ibid.) – also by Aristotle (in effect).
Aristotle’s reaction to it is given in my ‘Substance’ (), §:. (See also, for a different overall
account, Peter Simons in ‘On Being Spread out in Time: temporal parts and the problem of
change’, in Wolfgang, Spohn et al. (eds.), Existence and Explanation (Kluwer, ).) The Aristotelian
answer should not however be seen as a response to Lewis that accepts his problem and then
resolves it. Rather, it defuses the problem. It puts us into a position to say: if an object is straight
now but will soon be bent, what is the problem about that? Only a misunderstanding of the
ontology of times that temporal adverbs like ‘now’ and ‘soon’ make possible can turn familiar
facts like this one into some problem about ‘straight’ and ‘bent’ turning into relation words.
 Identity: absolute, determinate, and all or nothing 
Their discussion has the further merit, over and above that of present-
ing a recognizable argument, of avoiding the classic confusions that
haunt the subjects of identity and individuation.33
Gabbay and Moravcsik begin by inviting one to consider the sentence:
() The young woman I met eight years ago turned into the senior lec-
turer whom I met last year.
and its putative ‘restatement’
() The young woman I met eight years ago is the same as the senior lec-
turer I met last year.
Then Gabbay and Moravcsik say:
‘[If] they are equivalent, a substitution in one should have the same result as the
same substitution in the other. If we substitute in () on the right side the expres-
sion we have on the left [‘The young woman I met eight years ago’], and if we
construe the expressions as simply referring to the woman, then we get
() The young woman I met eight years ago turned into the young woman I
met eight years ago,
which is false. . . . When we perform the same substitution in () to obtain:
() The young woman I met eight years ago is the same as the young woman
I met eight years ago,
the result fails to be equivalent to (  ) . . . ( ) taken naturally is a truism, but (
 ) is not. This shows that the ‘is the same as’ in () has a sense distinct from
identity; it also shows the need for interpreting the referring expressions in ()
and () as denoting stages of entities instead of the entities themselves. Of
course, once we construe the referents as stages, the sameness involved cannot
be identity.’
Taken just as it stands the argument arouses the suspicion of something
analogous to a cancelling out fallacy.34 But we can run it through again
like this: ‘Suppose () and () are both true. Then substituting identicals
within () in the way licensed by () we obtain (). But () is false.
Therefore ‘is the same as’ in () must have a sense distinct from that of
identity, and the referring expressions in () and () must denote stages
of entities, not the entities themselves.’
Change is more primitive than any ontology of times. Let us not visit upon ‘I shall soon be bald’
the problems that come to be perceived in ‘For some t, I am [timeless] bald at t and t is not too
long after now.’
33
For a classic dissolution of several classic confusions, see Vere Chappell ().
34
‘We cannot infer that if two propositions verbally differ precisely in that one contains the expres-
sion E1, and the other expression E2, then if the total force of the two propositions is the same,
we may cancel out the identical parts and say that E1 here means the same as E2’ (Geach (),
p. ).
 Identity: absolute, determinate, and all or nothing
Everything in this argument depends in the first instance upon (). Is
() given in ordinary English? How then in ordinary English – or, if ()
is not in ordinary English, then how in philosophical English – are we to
understand the mode of combination it exemplifies between the verb
and its subject(s)?
The answer to the first question is that no dictionary that comes to
hand, nor any current catalogue of phrasal verbs and their uses, cites
any sentence that is like () in respect of the second position that ()
furnishes for a singular term (‘the senior lecturer whom I met last
year’). More typical of the sentences we find in English are these: ‘cat-
erpillars turn into butterflies’, ‘the experience has turned him into a
sad and bitter man’ (thus ‘by this experience he has been turned into
a sad and bitter man’), ‘The witch turned the prince into a frog’. In
itself, this is of no great consequence. (Though the finding can be
strengthened further by consideration of examples with the verb
‘become’ and the oddities of ‘become identical with’.) But it points
towards the right answer to the second of the two questions just posed.
Consider the sentence:
Following a not altogether credible tradition, the chroniclers represent that, in
his youth, Prince Hal was a dissipated roysterer who was suddenly changed into
a model king.35
The first thing to remark here is that we shall have no difficulty with the
substitution that is made possible by the truth that Prince Hal is the
same man as Henry V of England. If Prince Hal was in his youth a dis-
sipated roysterer who was suddenly changed into a model king, then
Henry V was in his youth a dissipated roysterer who was suddenly
changed into a model king. But then the example suggests, as do all the
others, that, in order to give a sense to (), we need to identify some new
state or form or condition that was assumed by the young woman whom
Gabbay or Moravcsik met in , or a property she came to exemplify.
Let this be the property of being a senior lecturer. Why not see () like
this then? It says that the young woman whom Gabbay met in 
came to have the property (␭x)(x is a senior lecturer), the very property
exemplified by the woman he met in , or else it says that the young
woman Gabbay met in  came to have the property (␭x)(x is a senior
lecturer and Gabbay meets x in ). It does not matter very much
which you say.
35
Cf. Frederick W. Moorman, The First Part of Henry the Fourth (London and Glasgow, undated), p.
xxx.
 Identity: absolute, determinate, and all or nothing 
Back to (). All that () and the intersubstitutivity of identicals now
gives us is that the senior lecturer Gabbay met in  came at some time
(when?) to have the property exemplified by the senior lecturer he met
in  (or the property simply of being a senior lecturer he met in ).
The senior lecturer in question wasn’t, in other words, born that way.
The new sentence is not false – just trivial.
There are other possible construals of () – too many in fact. If ()
were an ordinary sentence of ordinary English, I should be greatly
embarrassed by the arbitrariness attaching to the choice. But, as things
are, () needs to have a sense conferred upon it. Still less can the truth of
() make mandatory for us any business of transposing () or () into the
language of occurrences and stages. The positive case for such transpo-
sition still needs to be made. (See further below.)
In the here and now, pending fresh argument, let us cease to resist the
appearances – the appearance, I mean, that identity is just identity and
need not be divided up between synchronic and diachronic, and the
appearance that, with ‘Hesperus is Phosphorus’, ‘Mary Ann Evans is
George Eliot’ or ‘the man before me is the same person as the boy I
shared a locker with at school’, matters are just as they seem. Each of
these statements is true if and only if the persisting thing mentioned on
the left is one and the same as the persisting thing mentioned on the
right. Let us make sense of such identities without invoking either object-
stages or the so-called unity relation. My plea is made in full awareness
of the widespread assumption that, in the end, all questions of identity
and persistence should be transposed to a four-dimensional framework.
Let me now demonstrate that awareness by breaking for one moment
my long silence about that assumption.
The assumption comes in two forms, which it is helpful to distinguish
carefully. In its first form, the contention is that questions about contin-
uants are transposable into problems about scientifically palpable items
that are there to be discovered at (more or less) the same addresses in
space and time. In this form, the assumption is usually inarticulate. It is
part of a world view that has a powerful appeal. But no principles have
ever been provided by reference to which the fidelity of such transposi-
tions could be judged.36 I think it has been assumed that such principles
are not needed because everything that lies at a given location in space
and time must be one and the same. This assumption is unwise, however,
as will appear to anyone who pays attention to the full variety of
36
Compare the complaint made by Woodger in the citation given in the last footnote of Chapter
Five.
 Identity: absolute, determinate, and all or nothing
adjectival or substantial predications applicable to things discoverable at
a given location.37
The second form in which we find the assumption in favour of the
four-dimensional framework goes like this: the time has come for us to
abandon completely the attempt to operate the old discredited Aristotelian
ontology of everyday thinking. The time has come to stop trying to apply
modern physics to ordinary continuants by means of ad hoc mereologi-
cal improvisations. Instead, let us apply the language of modern physics
directly to that which we encounter in experience.
On this second view, the theory of identity is simply one case among
hundreds where we should make the effort to see the world as a scien-
tist officially does – and a case where the scientist himself, in the labor-
atory that he so frequently describes in lamentably unofficial or
Aristotelian terms, should try to follow through his own discoveries. If
only we will all do this, it may be said, then we shall usually find, entirely
satisfyingly, that the question we wanted to answer disappears of its own
accord.
One cannot help but admire the honesty of so fresh and engaging a
proposal. If I do not concur with it, that is only for two reasons: first
because, in practice, it seems almost impossible to find again in the new
framework either the things or the properties that our philosophical and
everyday concerns and questions involve us with. In this new framework,
one might well find events – for the category of event is common to the
four-dimensional and the everyday frameworks – but scarcely the events
articulated by our own ways of individuating or characterizing events.
As for the counterparts of the things, properties and substances that are
close to human life or that force us into the philosophy of identity, these
will not be discoverable at all. Solving problems by making them disap-
pear can be an excellent strategy in philosophy. But it cannot minister to
all needs or the great generality of questions. Indeed it is tempting to
predict that any sincere and prolonged attempt to arrive at a four-
dimensional reformulation of every legitimate question or concern that
exercises us and every predicate that is indispensable to us would only
fortify the conclusion towards which common sense already beckons,
namely that the everyday scheme and the four-dimensional scheme are
complementary not competing schemes of thought, each with its own
strengths and limitations.38
37
See my note ‘On Being in the Same Place’ (); and see Frederick C. Doepke ().
38
On all these matters, see Peter Simons (), p. , and, more generally, pp. –; also D. H.
Mellor, Real Time (Cambridge, ).
 Identity: absolute, determinate, and all or nothing 

  .                    

Identity is an equivalence relation, a congruence relation, a determinate
relation, an absolute relation, or so we have concluded. But can it be
reduced somehow to other properties and relations? Philosophers rarely
or never avow the explicit belief that it has to be so reducible. But I
venture to think that some such belief must have been at work when, a
long time ago, Jaakko Hintikka was moved to rule39 that
Each possible world contains a number of individuals with certain properties
and with certain relations to each other. We have to use these properties and
relations to decide which member (if any) of a given possible world is identical
with a given member of another possible world.
If Hintikka is right, then strictly one cannot specify world one by saying
that in world one Alcibiades is blue-eyed and specify world two by
saying, without more ado, that in world two Alcibiades, the same man,
is green-eyed. If nobody is moved to say something analogous to this
about other basic properties or relations than the identity relation – if no
problem is held to exist about saying that in world one all swans are white
and in world two all swans are black – then why is the claim so perenni-
ally attractive where identity is concerned?
Perhaps the relevant thought is that the special thing about identity
is that it is unobservable, and unobservable because it’s somehow con-
sequential on something else. But can one not see directly (albeit fal-
libly) that it is one’s brother, not one’s uncle, who has come to call on
one?
Perhaps the relevant thought is that, even allowing for the point just
registered, identity is inscrutable except as resemblance (its ‘vanishing
point’, etc.) and that resemblance can only be a matter of properties and
relations. But it hardly takes deep philosophy to reflect that resemblance
is neither sufficient for identity nor necessary. It is only necessary in so
far as, at any given time, x must exactly resemble x.
But maybe brisk philosophical exchanges of this sort will only deepen
the apparent mystery about identity and strengthen the resolve of a
certain sort of empiricist to account for our mastery of it in terms of
whatever is cognitively less taxing. Since we work so happily with iden-
tity (he whispers), how can it help but ride on the back of something that
is more manifestly manageable?
39
See ‘The Semantics of Modal Notions’, Synthèse,  (), pp. –. For further comment on
these questions, see Chapter Four, § ad fin., §, §.
 Identity: absolute, determinate, and all or nothing
Within technical philosophy, some have seen hope of such a quasi-
reduction in the ‘elimination’ of identity for which Hilbert and Bernays
first gave the recipe and which Quine describes as follows:
‘Still nothing has been said [in Word and Object] as to the make-up of the admis-
sible vocabulary of unanalysed general terms. But of this particular we may be
sure: [the predicate] ‘⫽’ will in effect be present, whether as an unanalysed
general term or in complex paraphrase, at least provided that the vocabulary of
unanalysed general terms is finite. For, write ‘if Fx then Fy’ and vice-versa with
each of the absolute general terms of the vocabulary in place of ‘F’; also write
‘(z) (if Fxz then Fyz)’ and ‘(z) (if Fzx then Fzy)’ and vice versa, with each of the
dyadic relative terms in place of F; and so on to ‘(z) (w) (if Fxzw then Fyzw)’, etc.
The conjunction of all these formulae is coextensive with ‘x⫽ y’ if any formula
constructible from the given vocabulary is; and otherwise we can without con-
flict adopt that conjunction as our version of identity. In so doing we impose a
certain identification of indiscernibles, but only in a mild way.’40
Wherever such eliminations are possible (I suppose the thought runs),
there is a clear sense in which the predicate ‘is the same as’ is indeed
reducible. Thus ‘x is the same as y’ is reducible by being replaceable by
another and longer open sentence in x and y, an open sentence that is
bound to have the same extension. How better or more satisfyingly could
the whole mystery be dissipated of identity and our grasp of identity?
Chapter Two, §, has committed me to be suspicious of this claim.
Congruence should indeed flow from identity, but congruence as such
will scarcely encapsulate the whole of identity. (Unless identity is already
incorporated within congruence. See also Chapter Three, note .)
The first thing to remark is that the recipe Quine has given for the
elimination of identity from a first level quantificational theory is indeed
one recipe. But the results of the application of this recipe do scant
justice to the evident univocity of ‘x is the same as y’. Suppose that, for
each of several languages unlike English in having a clearly delimited set
of predicates and different from one another in respect of primitive
vocabulary, we frame by exhaustion its surrogate identity predicate.
Suppose we then replace the extant ‘⫽’ predicate of the language by the
surrogate predicate. What do the surrogates framed for these various
languages have in common? Each expresses a certain congruence with
respect to the predicates of the particular language into which it is intro-
duced. But, so long as we restrict ourselves carefully to that which can
be expressed in first level terms and we are confined to that which can
register within Quine’s first level definitions of ‘x⫽y’, we cannot express
40
Quine (), p. .
 Identity: absolute, determinate, and all or nothing 
what it is that constitutes each surrogate an identity or congruence pred-
icate. For Quine’s method makes whatever sense it makes for us by its
surreptitious allusion to and dependence upon an idea which is essen-
tially second level. The second level idea is that identity is the relation
whose holding between a and b ensures that every property of a is a prop-
erty of b and vice versa. The trouble with a second level account of iden-
tity, however, is that it involves quantifying over all first level properties,
including reconstructed identity and countless other properties that
latently involve identity. It might be replied that this does not matter so
long as there is for each language some surrogate identity predicate. Such
a defence would raise difficult questions about how it is that speakers
with different active vocabularies, determining different surrogates,
understand one another – unless on the basis of a second level concep-
tion of identity. But let us not dwell on this here. For there is a further
and more specific reservation.
John Wallace points out that Quine’s method of reducing ‘⫽’ yields
rather stranger results than any that Quine himself has countenanced
explicitly.41 Applied to a theory or language with just three unanalysed
predicates ‘x is a forest’, ‘x is a tree’ and ‘x grows in y’, Quine’s proposal
forces the truth-value false upon the sentence ‘In every forest there grows
more than one tree’. Of course the defender of the method may claim
that such troubles will disappear so soon as we consider first-order the-
ories with an expressive power more closely approximating to that of
English. Wallace suggests that the claim of eliminability would need to
be rephrased as follows: if a theory has a finite number of unanalysed
predicates, then a finite number of predicates can be added to the
theory, so that in the resulting expanded theory what we usually intend
by identity is eliminable by the prescribed method. But now – at least for
the purposes of the evaluation of the idea that identity can be reduced
or explained in this way – everything is seen to depend on what predi-
cates are added in order to force the desired truth-value upon the sen-
tences involving the constructed predicate that is to be introduced in lieu
of ‘⫽’. Suppose the appearances can only be saved if monadic and poly-
adic predicates presupposing identity or place–time individuation or
thing-individuation are supplied. If so, it is upon the presence of these
that the success of the elimination recipe will depend.
41
See John Wallace, Philosophical Grammar, Stanford University Ph.D., , published by
University Microfilms Ltd, Ann Arbor, Michigan, . See pp. ff. In order to clear identity
from Wallace’s sentence, Williamson proposes that his argument be recast as relating to the sen-
tence ‘In every forest there grow many trees’, deploying a ‘many’ quantifier taken as primitive.
 Identity: absolute, determinate, and all or nothing
Here we are left with at least one question, namely the question of
which predicates are really and truly identity-free. And pending that’s
being settled, I suppose that the hope will still persist that I am trying to
dissipate. I can imagine my notional opponent’s saying this: ‘Maybe it
was a mistake for me to get involved in questions about reduction and
quasi-reduction. The thing that I want to hold out for may amount to
less than that. My thought is that, if identity is to be empirically tract-
able, then it must be that in any given situation and any given context,
all the identities of things involved are fixed or fastened down from some
arbitrarily large sufficiency of information about all the various other
predicates and relations that are instantiated there. Is not the fixation of
identity by other properties and relations the condition of significance and
of workability for the idea of identity? Whatever questions remain open
about the reducibility of identity or the elimination of Quine’s elimina-
tion recipe, identity must supervene on other properties and relations.’
That is what my opponent says. Let it be clear, however, that in doubt-
ing, as I do, whether there is any interesting sense in which identity
supervenes on other properties and relations, I do not take myself to
doubt that, in a given context with given object or objects, the identity
relation must rub shoulders with and have relations of compossibility or
incompossibility with other properties and relations. That is the normal
condition of properties from different ranges converging upon a given
object or objects. The controversial claim to which the reduction-
ist/supervenientist is now committed is rather this: even if there is no
one way in which identities are to be read off the not identity-involving
properties and relations of the entities that are involved in a given set-
up, there is, for each and every set-up, some way of determining, from
a sufficiency of information about the not identity-involving properties
and relations of the entities participating in the set-up, the identity of
each thing involved.
How is this idea to be evaluated? My opponent speaks of other prop-
erties and relations fixing identities, fastening them down, or being the
basis for ascertaining identities. But these may not be exactly the same.
Think of a case where in a detective story (say), an identity is determined
in a striking and unexpected way from some narrative about a whole
group of entities. Is it this sort of thing that the empiricist has in mind?
If it is, then the obvious objection seems to be that all that is in question
there is a deductive reconstruction from an incomplete narrative of a
fuller narrative. This is the recovery of a narrative in which things have
an identity already determined. Epistemically speaking, there is here a
 Identity: absolute, determinate, and all or nothing 
triumph. But there is no model here for the constitutive determination of
identity out of properties and relations that do not include identity.
Let this obvious finding be the occasion to reflect that the idea of
supervenience is often said by those who set store by it to be this: that
properties from a certain range of higher level properties should super-
vene on some other range of more explanatorily basic properties. But,
if supervenience is some relationship between classes of properties, then
is not the identity of the subjects of such properties treated as something
taken already for granted? The supervenientists themselves seem to
suggest this when they say (e.g.) that ‘it is not possible that two things be
indiscernible in respect of their lower level properties without also being
indiscernible in respect of their upper level properties’. On this account
of the matter, identity itself lies well outside the intended area of super-
venience.
To defuse this objection, the empiricist might want to explore the fol-
lowing suggestion: facts about the identity of objects supervene on facts
about the exemplification of properties and relations other than the iden-
tity of objects. When the claim is put in this way, it is left open what the
exemplification of the other properties and relations comes to. Maybe
these exemplifications are given in feature placing or quasi-pointillist
style without any particular commitment to any particular ontology.
Maybe matters of ontology and identity supervene simultaneously on
matters about the distribution of features.
There are things here that are difficult to understand. But this may
not matter. For my opponent has now committed himself to something
very strong and highly questionable, something equivalent to the
Identity of Indiscernibles, first mooted in Chapter Two. In that chapter
it was pointed out how strong a principle Leibniz’s Principle eadem sunt
quorum unum alteri substitui potest salva veritate becomes as soon as possession
of all the same properties is interpreted as sufficient for identity and the
properties in question are strictly confined to the pure properties that do
not presuppose the individuation of particular times, things or places.
Wittgenstein noted that any such restriction would exclude a universe
consisting only of two exactly similar spheres. Later, Strawson pointed
out, in opposing the Identity of Indiscernibles, that
‘It is necessary only to imagine the universe in question being repetitive or sym-
metrical in certain ways in order to see that there might be numerically differ-
ent points of view from which scenes presented [to an individual monad] would
be qualitatively indistinguishable, even though they comprehended the entire
universe. . . . A very simple illustration will serve.
 Identity: absolute, determinate, and all or nothing
‘Think of a chess-board . . . The problem is to provide individuating descrip-
tions of each square, and to do so in terms of the view of the rest of the board
obtainable from each square . . . The problem cannot be solved even if the view
from each square is allowed to comprehend the entire board.’ ((), pp. –)
When faced with such examples, some people will not flinch from the
thought that it is no less than a presupposition of our individuative prac-
tices that we do not live in such worlds as Strawson and Wittgenstein
have described. Because the presupposition is global, they believe it can
be lived with – pragmatically, say, and/or as a ‘framework principle’. But
matters are even worse than they suppose.
Is it also a presupposition of our individuative practices, or a frame-
work principle for it, that there are no fully symmetrical objects? (A sym-
metrical object with exactly matching top half and bottom half has parts
that are indiscernible. It is reduced effectively to half of itself by the
Identity of Indiscernibles as strictly construed. Moreover, an object that
is symmetrical about all planes which bisect it seems to be precluded
altogether. For not only would the top half of such an object be indis-
cernible from its bottom half. The left side of the bottom half would be
indiscernible from the right side of the bottom half. And as regards the
residual eighth . . . We are eventually left with nothing but a line and
then, at the limit, a geometrical point.) Are we to believe that it is a
‘framework principle’ that no object is symmetrical about all planes
which bisect it?
In the face of these difficulties, no interesting thesis of supervenience
seems defensible. Of course, it is a crucially important fact about iden-
tity, it is of the essence of identity, that it rubs shoulders with other prop-
erties and relations. But that is not enough for supervenience. Identity is
just identity. It cannot be confected from other properties and relations.

 .               a       b ,

  a                    b 
Maybe identity is, as we have claimed, autonomous, determinate, irre-
ducible, non-supervenient, primitive, not a matter of degree, all or
nothing. But, despite all that, it may appear that a question remains: is
it possible for a to be almost (numerically) the same as b? Again, where
a is not b, someone would ask: might a almost have been b? In matters of
identity, can sense be made of what is still called in post-World War par-
lance ‘a near miss’? This question is still open. The fact that identity is
all or nothing does not close it. Consider. Being tall enough to be a
 Identity: absolute, determinate, and all or nothing 
guardsman is all or nothing. Let six foot be the first acceptable height for
a guardsman and let there be no last unacceptable height. By this ruling,
one could say that being tall enough for a guardsman is all or nothing.
Nevertheless the height of ft ⁄in is a near miss. (Would it be better
to call this a near hit?) If you really want to be a guardsman and you
stand ft ⁄in, you must grow. But you do not have very far to grow. If
you fail, then this will have been a near miss.
For identity could there be anything similar? There is a case that inter-
ested the Roman jurists, who subsumed it under their doctrine of acces-
sio. Let us state it and then adapt it:
Pictura. A paints a picture upon B’s tablet. The picture is here considered the
principal thing, the test . . . being its value, and the tablet is considered the acces-
sory. So the result belongs to A. But if B is in possession, A must pay compen-
sation for the tablet, or be defeated by the exceptio doli. If A is in possession, B
may bring an actio utilis for the tablet, but he must be prepared to pay for the
picture, or himself be defeated by the exceptio; that is, if A got the tablet hon-
estly. If A stole it, B has the actio furti.42
Here is the case to consider: on Monday, A comes upon B’s tablet and
discovers that there is a sketch painted on it, a landscape with farm build-
ings, trees and mountains indicated in the distance. Because it appears
that, in one place, the picture must have been smudged before the paint
dried, A gets out his paint box and retouches the passage. On Tuesday,
A looks again at the tablet, approves his own restorations, but then itches
to improve things further. He lightly retouches several other passages.
On Wednesday, he looks again. Suddenly, he is moved to introduce into
the scene the figures of human beings and animals, at the same time
painting into the sky the sullen indications of an impending storm. On
Thursday, B discovers the loss of the sketch painted on the tablet and
makes inquiries about the whereabouts of his property. On Friday, he
brings an action for the recovery of the tablet on which A’s picture is
painted. Meanwhile A enters the claim that the picture itself is his.
Persisting in this claim, he has to offer compensation to B for the tablet
on which he, A, is ruled to have painted a picture that is not B’s prop-
erty.
We may leave to the jurists most of the questions that arise here. Our
sole concern is with one preparatory question, and one supplementary
question arising from the answer to the preparatory question. Suppose

42
Roman Private Law founded on the Institutes of Gaius and Justinian, by R. W. Leage, nd edn by C. H.
Ziegler (London, ).
 Identity: absolute, determinate, and all or nothing
the magistrate determined in favour of A’s plea to retain the picture,
on the grounds that Wednesday evening’s picture is not the same work
as the sketch belonging to B that A discovered on Monday morning.
Suppose everyone agrees that too much has happened on the tablet
between Monday afternoon and Wednesday evening for any other
finding. Suppose it has been ruled that Monday’s work was restoration
and Tuesday’s mere meddling. The meddling has changed the original’s
appearance a little, there has been significant qualitative change, but on
Tuesday the original still exists (albeit ‘improved’, A says). Obviously
then, the moment when Monday’s picture ceased to exist occurred on
Wednesday. Suppose that the crucial changes that A made to the picture,
though begun earlier in the day, were not fully in progress until . pm.
Now consider this identity claim: ‘The original sketch, the sketch that
was visible on Monday, on Tuesday, and on Wednesday morning, is iden-
tical with Wednesday evening’s picture.’ That claim is certainly false (or
so we are supposing). Yet is not the following remark, made by a witness
Caecilia, true? – ‘The picture that was on the tablet at . pm on
Wednesday was almost the same picture as the picture belonging to B that
A discovered on B’s tablet on Monday. Indeed, if A had been attempt-
ing less, if A had not taken it into his head to impose his own concep-
tion of the scene depicted, then the picture to be found on the tablet at
. pm on Wednesday would indeed have been the same picture as that
which A discovered on the tablet on Monday.’ (Cf. Chapter Four, §.)
Let it be conceded immediately that, given the finding of non-identity
between the original sketch belonging to B and A’s finished work, this
could well be a sensible, non-pointless and true thing to say. The ques-
tion is whether the proper interpretation of the claim involves identity
itself ’s holding either almost or not quite. What I should say is that, when
she offers the commentary last quoted, Caecilia has no need at all to
commit herself to that idea. For when she stated that the Wednesday
. pm picture was almost the same as the original discovered on
Monday, she can be interpreted faithfully as saying that the former
picture has a strong qualitative resemblance to the latter. Caecilia can
also be interpreted as expressing the thought that it was almost the case that
there was on the tablet at . pm, Wednesday, a picture that was one
and the same picture as the picture discovered on the tablet on Monday.
If such and such hadn’t happened, or if someone had dissuaded A from
following his impulse, then the picture on the tablet at . pm would
have been the same. But here the definite description ‘the picture on the
tablet at . pm on Wednesday’ must have the sense of ‘whatever
 Identity: absolute, determinate, and all or nothing 
picture it was/would have been’. (That is an entirely normal use, on
some views a standard use, of a definite description.)
Nothing then forces this issue. One could accept Caecilia’s remark
entirely verbatim without having any use for the idea that a can be
almost identical with b (in the numerical sense of identity) or could
almost have been b. But why should one resist the idea?
If I resist, it is on the strength of the following thought. The aspirant
who reaches a height of ft ⁄in either goes on to ft or he doesn’t. But
whichever he does, whether he succeeds or he fails, he is there and he
makes it or he doesn’t make it to ft. It is he who lives through his good
or bad luck. It is not so similarly easy to make sense of there being through-
out something that would or would not count, on Wednesday at . pm,
as numerically identical with the Monday picture. We could of course
say, for the case of a variant on the actual history in which someone
restrains A on Wednesday morning, that the picture of Monday was
almost destroyed on that Wednesday, but escaped. But in any scenario that
Monday picture can’t have been almost identical with the actual
Wednesday . pm picture. In the actual scenario, the Monday picture
was obliterated or effaced some time on Wednesday before . pm. In
the variant where A is restrained, the Monday picture is still on B’s tablet
on Wednesday at . pm. It is there, not almost there, even if it almost
suffered the fate of being painted out.
In our putative example, there is indeed, in both variants, something
that is there all along, namely the tablet that belonged to B. But B’s tablet
is not identical with either the original sketch or the new painting that A
made out of that sketch. (Here Roman Law and the doctrine of individ-
uation agree.) The tablet with B’s picture on it can indeed become by
degrees a tablet that supports the new picture that is A’s property (even
while it, the tablet, remains B’s). It is not of course to be denied that the
picture B owned can become more like the picture that will be ascribed to
A. Under A’s earlier restorative attentions, the picture that B owned can
indeed, in this sense, creep up on the status of A’s picture. But it cannot
creep up on being the same picture as A’s picture and either fail, or just
fail, or just succeed or easily succeed in being that very picture. It is no
easier for it to be almost the very same picture as A’s picture than it is for
it to be in part numerically identical with that picture.
Conclusion: When Caecilia comments, plausibly enough, that the
. pm Wednesday picture is almost the original Monday sketch, or
vice versa, she is not charitably or sensibly interpreted as saying anything
at all that would entail that this very picture, even though it is just coming
 Identity: absolute, determinate, and all or nothing
into being under the care of A, is almost numerically the same as the
picture which, on Monday, A discovered on a tablet belonging to B. (Nor
is the converse claim at all promising.) If this is right, then it supports the
thought that identity itself is not only all or nothing. It actively excludes
(or so it now appears) the ideas of just making it, almost making it, or a
near miss.
Those who want always to be sure there is some formal demonstra-
tion in the offing may see the doubts I have just rehearsed about the idea
of an entity a’s being almost identical with an entity b (and my reading
of Caecilia’s claim as altogether innocent of that idea) as furnishing an
advance commentary on the following derivation:
(i) a ⫽b (Hypothesis)
(ii) a is not almost identical with a and is not almost distinct from a
(iii) a is not almost identical with b and is not almost distinct from b
(iv) (a⫽b)→(a is not almost identical with b and is not almost distinct from b)
(v) (a is almost identical with b or a is almost distinct from b)→a苷b.
Commentary. (v) is the contraposition of (iv). (iv) is arrived at by condi-
tional proof from (iii) and (i), discharging (i). (iii) comes from (ii) and
Leibniz’s Law. (ii) is a substantive claim about a’s identity with a. In (ii),
the ‘not almost’ is read twice over in one natural way. This is the same
as the way in which it is read in (iv). (ii) itself reflects one special thing
about identity. It reflects the same thing that organizes the persuasive
efforts of all the preceding paragraphs of this section. Moreover, (ii) is
true.

.     
In this chapter there is just one rash claim about identity (and singling
out) that I have held back from entering. I have not said that findings of
identity are uncontentious. All or nothing, no near misses, determinate,
and all the rest – that is one thing. Uncontentious is another thing alto-
gether. Identity can be all or nothing, determinate, exclude degree,
exclude near misses . . . and be contentious. But the theory of identity
that allows findings of identity to be contentious – this I have aspired to
render less contentious. Once sober conceptualism comes to the aid of
ordinary common sense, philosophy has no need (I declare) to supplant
the common sense conception according to which identity is utterly
special and unique.
    

Personal identity

You would not find out the bounds of the soul, though you traversed every path:
so deep is its logos.
Heraclitus

The human body is the best picture of the human soul.

Wittgenstein, Philosophical Investigations, p.  (trans. Anscombe).

PART ONE
 .                 
              
If the thesis of the Sortal Dependency of Identity (D) is correct, then a
simple expectation arises: namely that, in the case where A and B are
persons, the content of the sentence ‘A is the same as B’ may be eluci-
dated, through the equivalence of ‘A⫽B’ and ‘A is the same something
as B’, as requiring for its truth that A should coincide with B under the
concept person. This can scarcely be wrong. But, since the question that
it pushes into immediate prominence, of what sort of thing persons are,
leads into a thicket of philosophy, anyone who has been once through
the thicket may wonder whether it can be skirted.
One way round seems to be to reflect that, if A and B are the kind
of creature that we take ourselves to be – the kind of creatures writing
or reading this book, I mean – then an equally good answer to the same
what? question ought to be ‘human beings’. For if A and B are persons
and human beings, then A could never be the same person as B and
not the same human being as B; or the same human being as B and
not the same person. (Cf. Chapter One, § following.) So, given the
human beinghood of A and B, this furnishes a perfectly good cover-
ing concept for the identity ‘A is the same as B’. ‘Person’ and ‘human
being’ differ in sense. They may even differ in their extension. But that

 Personal identity
is immaterial.1 What matters is that here, in so far as they assign any,
the concepts person and human being assign the same underlying princi-
ple of individuation to A and to B, and that that principle, the human
being principle, is the one that we have to consult in order to move
towards the determination of the truth or falsehood of the judgment
that A is B.
Human beings, endowed as they are with a distinctive mode of activ-
ity of their own, are substances in the rich sense of ‘substance’ that has
been built up in the preceding chapters.2 Their mode of activity is
nothing unknown to us. So one might well wonder how questions of
identity covered by the human being concept could fail to find the notion
of identity at its most straightforward and unproblematical best. By
admitting of answers that are principled, such questions ought to
exclude the risk (and even the risk of the risk) of indeterminacy. See
Chapter Six. Ought it not to be safe then, in any case where it is not
doubtful that A and B are determinately identified, to see the apparent
non-determinacy of truth-value residing in a claim ‘A⫽B’ as entailing
that A is actually not B? On this view, the difficulty of undecided ques-
tions of personal identity ought to derive either from faulty formulation
or from ignorance. Or else, if human being is metaphysically troublesome
(which need not be very surprising), then we must attend again, at need,
to the question of what human beings are (cf. §) in order that we be
better placed to apply to the problem of their individuation the princi-
pled ingenuity and vigilance that ought to be promoted by any adequate
dialectic of sameness and otherness.

.   ,          :       

      
If, as some declare, these expectations (of determinacy and the rest)
are not in practice borne out, why is this? It may be said that the
reason is that the proposal just made will be deeply controversial so
soon as it is applied to sentences of the form ‘I am . . .’, as in ‘I am
11
That they are not synonymous makes easier the expression of certain truths that it would be a
pity for us to dispense with. Or so I aver at pp. – of Wiggins (b) and again at pp. –
of Lovibond and Williams ().
How may sortal predicates differ in extension but impose the same principle of individuation?
Here is one model: ‘tinker’, ‘tailor’, ‘soldier’, ‘sailor’ all impose the principle man/human being.
This is not the only model.
12
More precisely – and imprecision is dangerous here – ‘substance’ has been shown to stand for a
determinable whose particular determinations, human being, ilex, mouse, . . . must be expected to
be rich, copious and exigent in what they require of things that answer to them.
 Personal identity 
Lohengrin’, or ‘I am thy father’s spirit’. ‘Such expectations were
bound to fail’ it may be said, ‘because they have so little regard for the
fact that, first and foremost, human beings experience themselves as
subjects of consciousness.’
As a diagnosis of why personal identity is an issue that will not easily
go away, this is plausible. But it is unclear, I should say, how much further
regard, beyond that already accorded to it, a philosophical account of
personal identity needs to have for the fact of our experiencing ourselves
as subjects of consciousness. In advance of philosophy’s magnifying or
distorting things, what else is there here but this: that we experience our-
selves as embodied and changeable substances among other human or
non-human substances; that, even as we perceive things or do things or
undergo things done to us, we experience ourselves as human beings
interacting with other animate or inanimate beings (and can experience
ourselves as experiencing ourselves so)? But, if this is what things come
down to, it carries us back to the human being conception, to the phil-
osophically familiar idea of a person as possessor of P-predicates and
M-predicates,3 and then to the question of how human beings do in fact
get the idea of the distinction between that which is themselves and that
which is other than themselves or grasp therewith the moral significance
of that distinction.
At this point, various objectors will complain that too much that is
bluntly empirical has been built too soon into the claim that we experi-
ence ourselves as subjects of consciousness; that the whole tenor of the
explication just given falsifies, as if deliberately, the real depth and diffi-
culty of the original finding that we conceive ourselves as subjects of
consciousness or as conscious selves.
In the face of such a complaint, there is a temptation for a human
being theorist such as myself to find fault with the way in which the
determinable ‘conscious subject’ has now been transformed into a fully
fledged philosophical substantive of a non-empirical kind. Such a
13
See P. F. Strawson (), Chapter Three. In the argument of that chapter, I should like to recon-
strue P-predicates as predicates standing for properties of a person that are directly or indirectly
consciousness-involving and M-predicates as predicates standing for properties of a person that
are directly or indirectly body-involving. Physicalists can say then that, so far from being disjoint,
all P-predicates are M-predicates (but not of course vice versa).
Other works that have left their mark on the contentions of this chapter are Chapter One of
Stuart Hampshire, Thought and Action (London, ); Lev Vygotsky, Thought and Language
(Cambridge, Mass., ), revised translation edited by A. Kozulin – see especially the last
chapter; Lev Vygotsky, Mind in Society: The Development of Higher Psychological Processes, ed. and trans.
M. Cole, V. John-Steiner, S. Scribner, E. Souberman (Cambridge, Mass., ); David Bakhurst,
Consciousness and Revolution (Cambridge, ), see Chapter Three.
 Personal identity
theorist will say that he is reminded of the way in which the determin-
able ‘substance’ is misunderstood (and then welcomed or turned away,
on all the wrong sorts of ground) as purporting to stand for a special
and specific kind of thing. He will protest bitterly at the way in which,
rather than see the dummy sortal or determinable predicate, ‘subject’,
as the first fix by which one fumbles to subsume X (in this case some
conscious thing) under a more specific and informative sortal specifi-
cation, philosophers have come to think of a conscious subject as a
peculiar kind of thing reached by a special kind of abstraction. One
can only be amazed (he will say) that a mode of abstraction so danger-
ously similar to the procedures of creative definition4 should come to
be seen as legitimating, without further ado, a whole new branch of
philosophical thinking.5
The matter cannot be so simply wound up, however, by warnings
about the provenance and credentials of the term ‘subject’. For the
human being theory will still appear to do insufficient justice to a line of
reflection still prompted by John Locke’s account of these things: what I
am in the present (‘my present self ’) always lies under the cognitive and
affective influence of what I remember having been or having done or
undergone in the past, no less than of that which I intend or am striving
to make real in the present or the future. But if it is the nature of persons
to be remembering beings whose conception of what they themselves
are is all of a piece with their experiential memory, then some constitu-
tive connexion ought to be expected (it will be said) between their expe-
riential memory and their identity. But, in that case (it will be asked), how
can one be content in the simple expectation announced at the begin-
ning, namely that personal identity for creatures such as A and B should
amount to no more than the identity of the human beings A and B? So
soon as the Lockean insight is developed in the direction of the theory
of personal identity that it suggests, namely that person A is person B if
and only if A’s consciousness (or the subject self that A is) corresponds
to B’s consciousness (or the subject self that B is), it will prove that our
judgments of personal identity can diverge in indefinitely many cases
from those enforced by the human being account. Or so it will be
objected.
14
On creative definition, see still Patrick Suppes, Introduction to Logic (Princeton, N.J., ), Chapter
Eight.
15
The defensibility of the stand I take here is much enhanced by Quassim Cassam’s painstaking
and maybe exhaustive enumeration and critique of subject theories of the kind I am opposing.
See Self and World (Oxford, ). See my () and “Replies” in Lovibond and Williams (),
cited at note  above.
 Personal identity 
.          ;      ’          
      
Familiar though it is, the reflection just rehearsed needs to be set out
more historically and in a little more detail, if its force is to be measured
and the response to it is to be fully clear.
Locke thought that it was among the defining marks of a person to be
able to consider oneself the same thinking being in different times and
places. (See Essay Concerning Human Understanding .xxvii.: ‘A person is a
thinking, intelligent being, that has reason and reflection, and can consider
itself as itself, the same thing, in different times and places.’) In Locke’s
account, identity of person is distinguished from identity of man. The
apparent divergence of Locke’s conception of personal identity – which is
avowedly both subjective (inner) and (he says) ‘forensic’ – from the human
being conception of personal identity, which I advocate, comes to light
most clearly in cases where the human being theorists would say that B had
completely forgotten what A had done. (The example that follows is drawn
from Reid’s Essays on the Intellectual Powers of Man.) The general has forgot-
ten the exploits of the boy. He has forgotten stealing apples on some distant
occasion that was at the time noteworthy in his life. For a human being
theorist, as for one who believes in the unitary persistence of the soul, such
forgetting is not constitutively relevant to the question of personal identity.
For the Lockean, on the other hand, it is a challenge to reflection that ought
not to be impatiently foreclosed.
Among the earliest critics of Locke’s account of personal identity was
Bishop Butler, the guardian at once of common sense and the substan-
tial soul. In his polemic against the peculiarities of the divided account
that Locke had been drawn into offering, Butler continued with a logical
objection:
‘though consciousness of what is past does thus ascertain our personal identity
to ourselves, yet to say it makes personal identity, or is necessary to our being
the same persons, is to say that a person has not existed a single moment nor
done one action but what he can remember; indeed none but what he reflects
upon. And one should really think it self-evident, that consciousness of personal
identity presupposes, and therefore cannot constitute, personal identity, any
more than knowledge, in any other case, can constitute truth which it presup-
poses.’ (First Dissertation to the Analogy of Religion. (Compare Samuel Clarke,
Works, , p.  – a reference I owe to Mr Dewey Ducharme.))

One might wonder why this passage of Butler has not been permitted
to settle the whole matter once and for all in favour of a substance
 Personal identity
account, either a human being account or else a soul account. Not only
does it seem obvious that the general is the same person and the same
subject as the boy. To one content to work within the framework which
we owe to miscellaneous insights of P. F. Strawson (see especially
Chapter Three of his Individuals) and of Donald Davidson, it will appear
plainly that the most interesting point in the case sustaining Locke’s
insight could have been readily accommodated within the human being
view. Let us see how.
Par excellence, a person is not simply one of us or a proper object of
our reciprocity,6 but a subject also of interpretation, a being that both
interprets and is interpreted. In making sense of others and of ourselves
in the business of interpretation, we cannot, however, make do with the
bare idea of ‘a subject of interpretation’. We need some further notion,
however rough and ready, of what sort of thing is to count, empirically
or operationally speaking, as a subject of interpretation. The grammat-
ical substantive ‘interpreter’ lies too close to being a mere determinable
to serve the purposes of individuation. If ‘subject’ or ‘interpreter’ is to
play any properly sortal rôle at all here, if sense is to be made of how
subjects of interpretation find other subjects of interpretation, there is
need for some further determination. As we find our feet in the world
and learn to adjust to others’ expectations, as we begin to understand
what we mean by the words we have learned already to utter, as we
extend our capacity to interact with others and participate or co-operate
with them, what stereotype of personhood do we have to catch onto,
clearly if not distinctly, and learn to elaborate?7 In its individuative and
recognitional non-effectiveness, the conception furnished by John Locke
is not very much better than the bare idea of ‘interpreter’. Even here,
however, we may be able to make a new beginning from the materials
that Locke provided. People are animate beings (we may revise his declar-
ation to say) – or palpably substantial souls, if you prefer – bearers of P-pred-
icates and M-predicates, who not only think things, but do things and undergo
things, beings who reason, reflect, perceive, feel, imagine, desire, intend
and have experiential memory . . . are happy or miserable . . . conceive
16
For the thought that ‘us’ has an essential occurrence in this claim, see op. cit. at note  above.
17
Pace Peter Winch’s presidential address, ‘Eine Einstellung zur Seele’, Proceedings of the Aristotelian
Society (–), the question does not presuppose that we find our feet in the world by procedures
analogous to those that the philosophers of semantic interpretation have described. It only pre-
supposes that, beginning and proceeding however we do, participatively and interactively no
doubt, we reach some point where we know roughly what a person is. In catching on, we grasp
the stereotype, clearly though indistinctly. For my use of the word ‘stereotype’ here, see Chapter
Three, note , last paragraph. For clear and indistinct knowledge, see Chapter Three, footnote .
 Personal identity 
of themselves thus (as above) . . . who have and conceive of themselves
as having a past accessible in experiential memory and a future access-
ible in intention. . . . Is it not by virtue of our grasping practically and
following through some such stereotype, is it not by virtue of our per-
fecting our application of it in the perceptible world, that the marks of
personhood come home to us, and are available to be elaborated in phi-
losophy? More soberly, let us say that in this way the marks are assem-
bled of persons as we know them from the only case we shall ever become
familiar with, namely that of persons who are human beings. Finally, at
the last step, the position of experiential memory comes to the fore. If
experiential memory is a mark of personhood, and persons need to con-
ceive of this faculty as essential to their own way of being, then that
secures its full significance to our expectation that, when people do or
suffer something, this will impress itself on their mind, extend their infor-
mation, colour their experience and influence their future responses. In
the absence of that expectation, few if any of our finer-grained inter-
pretations of people or the practices these expectations support could
ever be sustained.
If so much is correct, then, because of its central role in interpreta-
tion, experiential memory plays a special part in the full picture of per-
sonhood. But that finding (I insist) does not imply that experiential
memory will impinge on the necessary (or sufficient) conditions of the
survival of persons or play any role in the statement of sufficient condi-
tions of survival. (Still less is memory put at the service of a construction
of some self that is distinct from the human being.) Locke can be right
about remembering as central among the marks of personhood without
Locke’s or the neo-Lockeans’ being right about personal identity.
Neo-Lockeans have ignored such partial accommodations and dis-
tracted attention from them. Human being theorists may justly com-
plain of that. But there are at least two other reasons why Butler’s
criticism of Locke’s theory of personal identity has been found incon-
clusive. The first is that so many of Butler’s admirers have appeared to
overlook the possibility (the bare theoretical possibility, I mean, but it is
one much prized by neo-Lockeans) that the act of experientially remem-
bering (remembering ‘from the inside’) planting a fig tree (say) might be
modelled (logico-grammatically speaking) on such acts as conceiving or
imagining or visualizing planting a fig tree.8 For the notable point about
18
These are acts where the mental content imagined, visualized or conceived does not call for the
subject of the act, the planter, to be specified. A fortiori, the subject need not be specified as the
same as the conceiver, imaginer or visualizer. On these and related matters, see Bernard
 Personal identity
acts such as these is that they do not make the kind of reference back to
the self which invited Butler’s charge against Locke. Taken by them-
selves, the would-be truisms adduced by the Butlerians are not enough
to vindicate Butler’s criticisms. And so much must be admitted by
human being theorists.
In the second place, friends of Butler and human being theorists have
sometimes overlooked how easily the boy whose exploits the general has
forgotten can be reunited by the neo-Lockean view with the general
himself. Lockeans were well within their rights to be irritated when
Butlerians represented that there was an easy objection against Locke to
be found in this quarter. The Lockean’s point is worth making more
explicit – not so much because it is decisive in Locke’s or anyone else’s
favour as for the sake of a fairer recapitulation of some of the saner or
more durable preoccupations of the philosophy of personal identity. In
the absence of such a recapitulation, Butler’s central insight will con-
tinue to be obscured, or its real significance will not be appreciated.

.  -      -     

Consider the relation that holds between person Q and person P when,
for some sufficiency of things actually done, witnessed, experienced . . .
by person P, person Q has at a time after the doing, witnessing, experi-
encing, some sufficient real9 recollection of doing, witnessing, experienc-
ing . . . them.10 Let us call this relation (suitably tidied up with more
about ‘sufficiency’, and glossed somehow to allow for sleep and other
forms of inattention) the relation of strong co-consciousness. Then anyone
bent on grasping the nerve of Locke’s conception of person will see that
the identity-condition he really has to consider is one that begins with
footnote  (cont.)
Williams, ‘Imagination and the Self ’, in Williams (), especially p.  and thereabouts; David
Wiggins, ‘Locke, Butler and the Stream of Consciousness: and Men as a Natural Kind’,
Philosophy,  (), especially the second section (to which the present chapter is, of course, a
necessary correction and supplement); David Velleman, ‘Self to Self ’, Philosophical Review, , 
(January ).
19
The question will soon arise whether the words ‘or apparent’ should be inserted between ‘real’
and ‘recollection’. My view is that it must not. See the end of § below.
10
The notion of sufficiency figures twice in the definition. There is a great deal to be supplied here
from the basis of ordinary norms and expectations relating to memory. At any given time one
expects a person to remember fairly well his relatively recent past; and remembering one’s very
recent past involves being able to muster a rather detailed and complete narrative. Where tk is
much later than ti, slightly less may be expected. The objections we shall urge against the
Lockean account of personal identity are not very sensitive to how these details are supplied. But
‘sufficiently’ is to be glossed realistically, not in such a way that anything whatever qualifies as
remembering provided the event is distant enough.
 Personal identity 
the idea that, so long as one persists, one normally remembers at least
from one time to the next what one was doing or witnessing or suffering
at the earlier time. Making use of that relation, one who wants to devise
a Lockean identity-condition could first define a new relation C by dis-
joining co-consciousness with its converse (that is with the relation P has
to Q if P is strongly co-conscious with Q) and then introduce the ances-
tral *C of that relation. (Compare Locke’s own use of the word ‘extend’
at Essay , xxvii, .)
With a view to stating the Lockean identity-condition explicitly, let us
say then that, if Q remembers at t⬘ doing, witnessing or experiencing a
sufficiency of what P did or witnessed or experienced at t, then the
ordered pair 〈Q , t⬘〉 stands in the strong co-consciousness relation to 〈P,
t〉. Letting C stand for the disjunction of that relation with its converse,
we have the relation
〈P, t〉 C 〈Q , t⬘〉.
Then, as promised, we define a new relation *C that an ordered pair
of person and time may have to another ordered pair. Thus
〈X, ti〉 *C 〈Y, tm〉 is true just in case
either 〈X, ti〉 C 〈Y, tm〉
or else, for some 〈W, tj〉,
〈X, ti〉 C 〈W, tj〉 and 〈W, tj〉 C 〈Y, tm〉
or else, for some 〈W, tj〉 and some 〈Z, tk〉,
〈X, ti〉 C 〈W, tj〉 and 〈W, tj〉 C 〈Z, tk〉 and 〈Z, tk〉 C 〈Y, tm〉
or else for some 〈W, tj〉, 〈Z, tk〉, . . .
Note that, in the absence of doctrine about personal identity or the
semantics of ‘recollect’ or ‘remember’, it is simply not stipulated whether
C itself is a transitive relation. But *C, introducing a chain of pairs con-
sisting of persons and times at which they remember, and requiring for
its satisfaction only membership of such a chain, gives a simple co-
consciousness relation that is transitive by definition. It makes possible
the following amended neo-Lockean equivalence:
Ip: P is the same person as Q
iff (᭚t, t⬘) (t ⬍t⬘) and 〈P, t〉 *C 〈Q , t⬘〉.
A Lockean who is not too discontent with this formulation of his inten-
tions may be expected to insist that the ‘P’ and ‘Q’ here hold places for
designations of persisting things or continuants that ‘consider them-
selves as themselves at different times and places’. I could only applaud
 Personal identity
that insistence (however unclear I find the phrase just quoted). The strat-
egy with the boy, his stealing the apples at t, and the general is that, if the
general doesn’t in , when he’s a general, remember stealing the
apples in , because he doesn’t have sufficient recollection of any
sufficiency of what he did in , then we must look for someone (don’t
say someone else, say someone), a lieutenant in the Hussars, say, and an
intermediate time, , such that (a) that lieutenant’s doings, witness-
ings and experiences as a Hussar are ones the general still remembers in
 and (b) the lieutenant remembers well enough in  stealing the
apples in .
Let it be clear that there is nothing in these explanations of C and *C
that Locke or his followers will want to see as committing them to
abandon the ontology of mental continuants.11 Let it be clear too that,
if the definition of C is read just as we stated it and that of C* is under-
stood conformably with that, then the full strength of the right-hand side
of Ip may be brought out by drawing at need upon the causal conditions
of remembering and/or real recollecting that are transferred more or
less unexamined into the definition of C. For that reason, Ip is better than
it at first appears. But this is no guarantee that it amounts to a faithful
elucidation of personal identity or even to a formally satisfactory
proposal.
Ip is a theory that the Butlerians will want to criticize. But it will be
noted that, once the ancestral of C is deployed and C’s full import is
noted and enforced, the number of apparent divergences between the
neo-Lockean account of personal identity and the human being account
appears to be immensely reduced. For the general and the boy do figure
together within the extent of a *C chain of paired persons and times.
On the other hand, no deluded empathizer who has read my diary will
become a candidate to stand in such a *C chain with me just by virtue
of being in the state of seeming vividly and completely to remember
most of what I did when I came home to find the kitchen in flames. The
chief disagreements that remain will seem to turn on cases that many
philosophers will find problematic in their own right. These are cases
such as that of a person knocked down in a road accident, who never
recovers consciousness at all even though (as a human being theorist
would say) he lies in hospital for many months afterwards, still alive; or
the person who recovers consciousness but has entirely forgotten (as I
11
Fixing on the particular times t, t⬘, of an experience and the recollection of it, nothing prevents
the Lockeans from isolating the two-place relation holding for t, t⬘ between the experiencer and
the rememberer as follows: ␭x y ( 〈x, t〉 *C 〈y, t⬘〉).
 Personal identity 
should say) who he is and everything else about himself. Such cases
remain in dispute. But they are a poor locus for a decisive engagement
over personal identity.
Here then is an interim report on the state of the argument. There is
the human being theory (in purely logical alliance with Butler and with
Reid) and there is the neo-Lockean theory. The fundamental disagree-
ment is over whether experiential memory is in some way constitutive of
personal identity over time. The human being theorist denies that it is
so constitutive, even though he ranks the capacity for memory as one
part of personhood. The Lockean persists in maintaining that experien-
tial memory is constitutive of personal identity over time. This disagree-
ment is important. Yet it is remarkable how closely, once we adjust the
Lockean position in all the obvious ways that have issued in the propo-
sal Ip, the two positions can apparently be aligned with regard to normal
cases. The scene is now clear for a more decisive attempt to resolve the
disagreement.

 .     ’        

It is not hard to reconstruct Butler’s comment on Ip. Consider the case
where we do not know yet if B is A, but we think we may be able to estab-
lish whether now B is co-conscious with A. Suppose that we know that,
probably inadvertently, A once (in August ) caused a fire in the
library-stack in the library of the Chigwell College of Commerce. It may
seem we should simply ask: does B remember anything of causing such
a fire? It seems a fair test for whether B is A. (It is not the sort of thing
people readily forget having done. Nor are many others likely to have
done the very same act.) So we arrange for the question to arise
somehow, but in a manner that does not make B jump to the conclusion
that he is about to be handed over to the police (or yet that he is due to
receive a medal). Suppose that it appears in due course that B does
remember this, so that (for practical purposes) we seem very close to
establishing the identity. What then?
In this case, Ip may appear for one moment to be an effective and
plausible principle of identity. But how does it work? Well, B’s seeming
to remember setting fire to the book stack suggests to the inquirer, in the
context of his knowledge that there was this fire in which A was involved,
that B can indeed remember causing that fire in the book stack. Does the
inquirer’s new finding not come close enough to establishing that B is
strongly co-conscious with A?
 Personal identity
One might agree that the finding narrows the gap. But one can still
ask how exactly the investigator argues from B’s appearing to remember
to B’s actually remembering this incendiary activity. After all, B is only
the same person as A if his seeming to remember is his really and truly
remembering setting fire to the book stack. The answer will probably be
that B can rehearse the act of setting fire to the book stack as if from the
point of the view of the perpetrator and it seems to be the best explana-
tion of B’s ability to rehearse these things/events as if from the point of
view of the perpetrator that B was there and did do the act.
All right. But now, on behalf of Butler and his doubt against Locke,
it must be inquired what the foregoing argument takes for granted. The
answer is that the argument takes for granted some understanding of
what it is for someone to do such an act at a place and a time, survive,
move on to other places and times, and remember later the earlier act,
place and time. So it appears to take for granted the possibility of an
account of identity other than Ip. Or (as I should say, deploying the ter-
minology of this book) it takes for granted a criterion constitutively
determining that for which all sorts of things can count as tests (tests the
favourable outcome of which can count as evidence). What is this other
account of identity? Even if it is consistent with Ip (which it may be or
may not be) or even if it subsumes Ip, this other account can scarcely be
Ip itself. For Ip presupposes the concept of real experiential memory, and,
once we see that that presupposes the idea of someone’s doing some-
thing and then correctly remembering doing it, we are reminded how
exactly experiential memory presupposes identity. Where someone appears
to remember starting that fire, they can’t be right unless they were indeed there at the
fire.12 Pending the production of some other account of the identities of
subjects, one that is anterior to Ip, the only visible candidate is the sort
of account that is offered by the human being theorist.
12
This way of explicating Butler is indebted to Bernard Williams (–) and to A. J. Ayer’s The
Problem of Knowledge (Harmondsworth, Middx., ), p. . For an earlier critique of Locke, see
A. G. N. Flew () – to which I have come to see that the  chapter, which the present
chapter replaces, was culpably unjust.
Here is another fix on the same underlying point that I want to urge against Locke. Locke
writes (.xxvii.), ‘That with which the consciousness of this present thinking thing can join itself,
makes the same Person, and is one self with it and with nothing else; and so attributes to it self and
owns all the Actions of that thing as its own as far as that consciousness reaches.’ The thinking
thing that owns Actions as its own (acknowledges them, etc) needs to have got itself somehow
some conception of itself as the thing that thinks etc. So soon as it follows through its concep-
tion of that thing, however, it must know that it is neither necessary nor sufficient for an Action’s
being that thing’s that the thing should be prepared to ‘join itself ’ to that Action. Apart from the
case of forgetfulness, the said conception needs to take account of the cases of shame, delusion,
confusion. . . .
 Personal identity 
Here the verdict must go to Bishop Butler, or so I conclude, or to his
latter-day followers. It goes to him, not for the sake of his positive
account of personal identity,13 but for the sake of the thing that is vali-
dated as telling in his objection to John Locke’s account.
At this point, the friends of John Locke may be tempted to try to
escape Butler’s point by experimenting with the definition of C and *C.
They may weaken the right-hand side, for instance, so that it shall only
require that one who is a candidate to be in some relation C to P should
have sufficient real or apparent recollection of doing, witnessing, experi-
encing whatever P did witness/experience.14 Their mistake hitherto was
to be insufficiently bold in adjusting the idea of experiential memory, the
Lockeans might say; perhaps they have been too timid in facing up to
the potentially radical ontological consequences of their original insight.
There is no formal difficulty in the weakening manoeuvre, however
questionable its philosophical motivation. But it should be noted that the
price exacted will be a further diminution in the plausibility of Ip, and
an altogether new kind of mismatch, a serious mismatch, with the
common sense conception’s verdicts on cases – as with the verdicts of the
human being conception. The neo-Lockean must do much less or much
more than tinker with the definitions of C and *C. (See below, §.)
Let us call the relation to be aimed at in all future attempts by the neo-
Lockeans the relation R.15 As for *C itself, defined as we first defined it in
the most ordinary terms of ordinary language, let us not modify it, but leave
it just where we first had it, namely as part of one possible explication of
the idea (almost certainly identity-involving, in fact, but that is a part of the
philosophy of the matter, and nothing to be settled straight off by definition
or stipulation) of the philosophical notion of co-consciousness. If C and *C
have any interest, it will transcend that of the formulation Ip.

.    -      
To the friends of Butler, that might seem to be the end of the matter. But
it is notorious that the neo-Lockeans have not wanted to give up. No
13
Butler’s own account of identity is imperfect and at some points confused. In fact, it comes as a
relief that he unreasonably refuses to elucidate a notion so simple, so final and so spiritual as that
of identity of person. For the best statement of something that is closer to what Butler, given his
beliefs and preoccupations, really ought to have said, see Thomas Reid (), , §.
14
Of course, some strength might be restored to the right-hand side by reimporting (somehow) any
causal components that are lost by inserting ‘or apparent’. See below, § following. But the con-
siderations urged in the next paragraph will still apply.
15
I call it R because Derek Parfit calls it R. Let not the relation R be confused with the thesis R of
the relativity of identity, denied in Chapter One; or C with C, denied at p. .
 Personal identity
doubt there have been bad reasons for this – blind perseverance allied
with the inertia of misplaced ingenuity, phenomenological (would-be
phenomenological) overdrive, a hypostatization of the conscious subject
(which prepares some of us to reject the conscious subject as so charac-
terized) or the misunderstanding of the traditional idea of a substance
(allied, no doubt, with an unacknowledged residual attachment to it or
dependence upon it). But perhaps there have been good reasons, as well.
In a book published thirty-seven years ago and devoted in part to the
diagnosis of a mistake that was far commoner then than it is now (in
/) – namely that of supposing that ordinary claims of experi-
ential memory have to represent themselves as claims of identity (rather
than as importing the presupposition of it) – Sydney Shoemaker pro-
posed a striking thought experiment:
‘Suppose that medical science has developed a technique whereby a surgeon
can completely remove a person’s brain from his head, examine or operate on
it, and then put it back in his skull (regrafting the nerves, blood-vessels, and so
forth) without causing death or permanent injury. . . . One day a surgeon dis-
covers that an assistant has made a horrible mistake. Two men, a Mr Brown and
a Mr Robinson, had been operated on for brain tumours, and brain extractions
had been performed on both of them. At the end of the operations, however,
the assistant inadvertently put Brown’s brain in Robinson’s head, and
Robinson’s brain in Brown’s head. One of these men immediately dies, but the
other, the one with Robinson’s body and Brown’s brain, eventually regains con-
sciousness. Let us call the latter ‘Brownson’ . . . He recognizes Brown’s wife and
family (whom Robinson had never met), and is able to describe in detail events
in Brown’s life, always describing them as events in his own life. Of Robinson’s
past life he evidences no knowledge at all. Over a period of time he is observed
to display all of the personality traits, mannerisms, interests, likes and dislikes,
and so on that had previously characterized Brown, and to act and talk in ways
completely alien to the old Robinson.
What would we say if such a thing happened? There is little question that
many of us would be inclined, and rather strongly inclined, to say that while
Brownson has Robinson’s body he is actually Brown. But if we did say this we
certainly would not be using bodily identity as our criterion of personal iden-
tity. To be sure, we are supposing Brownson to have part of Brown’s body,
namely his brain. But it would be absurd to suggest that brain identity is our cri-
terion of personal identity.’16
At the time when Shoemaker’s book appeared, and along with almost
everyone else, I was extremely impressed by this example. Yet I see now
that we were not all impressed for the same reason. Some were

16
(), pp. –. (See also pp. , , –, , –.)
 Personal identity 
impressed by the simple thought that, if the brain transfer were per-
formed with preternatural dexterity, then Brownson’s experience would
be a subjectively seamless continuation of Brown’s. Others, a smaller
group perhaps, but the group to which I belonged and would have
avowed loyalty (this is not only hindsight), would have preferred to say
that the special thing about Brownson was that he was the functional
inheritor and continuator of all of Brown’s vital faculties. This was the
reason why Brownson counted as the unique inheritor of the title to be
Brown, the reason why Brownson was Brown, that very substance.
Neither Brown nor Robinson nor Brownson was a brain. But the brain,
being the seat of memory and consciousness, was not just any old part
of the body among others.17 It was the essential nucleus of a person (of
a human being) – or so we were wont to maintain. Moreover, experien-
tial memory was more than a mere potentiality. It was a shaped and
developed capacity, conspicuous among all the vital functions of person-
hood and coeval (in persons as they are known to us) with our developed
faculties for sentience, locomotion, desire, cognition, . . . etc.18 Why then
should not the embodied faculty for experiential memory enter into the
whole principle of activity of persons conceived as we conceive them? (See
D(v), Chapter Two, §.) Why should it not enter into some plausible
emendation of the serviceable principle suggested by a misappropria-
tion of Locke’s own words ‘the identity of the same man consists in
nothing but a participation of the same continued life, by constantly
fleeting particles of matter, in succession vitally united to the same orga-
nized body’? Why not let the emendation that is needed for the sake of
the Brown–Brownson case, considered as a case of personal identity,
require for the persistence of a person the operation of the same prin-
ciple of activity and/or the participation of the same continued life
‘vitally united to the same vehicle’ or ‘vitally united to the same seat of
vital functions’?
17
See my (), p. , para. . (See also p.  of the second item cited at note .) There was
nothing to prevent one from being impressed by both kinds of consideration. I must have had
some regard to the other kind when, a little later, I wrote ‘we should scarcely allow a criminal to
escape the penalty due to the unique doer of a criminal act by contriving his own fission’. See
‘Locke, Butler and the Stream of Consciousness: and Men as a Natural Kind’, op. cit. note 
above, the precursor of the chapter that the present chapter replaces. But this kind of consider-
ation was secondary.
18
Even if Brownson had taken some time to settle down after the operation to get used to his body,
to exercise its neural connexions and resume Brown’s life – and philosophers are still apt to under-
estimate the preternatural dexterity and knowledge that the imaginary surgeon and his equally
imaginary team of anaesthetists, suturists, radiographers, laser-technicians, physiotherapists,
psychotherapists, counsellors and the rest, would have to bring to bear – , the need for such a
convalescence might not have made any essential difference to that which then impressed one.
 Personal identity

.          

In –, when I first came across Shoemaker’s example and was
drawn into this response, I ought to have been much more troubled by
the point that, even if Brownson could talk like Brown (which would not
in itself have been easy – if it was required off-stage as well as on-stage,
so to speak), he could scarcely have stood and walked and run and
jumped and smiled and sulked and earnestly entreated and frowned and
laughed like Brown. But the chief thing that then concerned me was the
realization that, because of the doubleness of the brain itself, my
response to Shoemaker’s example threatened to reintroduce the same
kind of reduplication problem that had already given me so much
trouble with the case, which I had recently disinterred from Hobbes’s De
Corpore, of Theseus’ ship.19 What if Brown’s brain was itself divided
before multiple transplantation? It was with reduplication in mind that
I drafted the personal identity condition that I defended in  to
require the vital functions of a person who persisted through time and
change to be ‘subserved by a single organized parcel of matter’.
For several reasons, this was not a stable resting point. About the con-
nexion between accepting the Shoemaker example and being a Lockean
in matters of personal identity, the reader was left with the vague (and I
should now say wrong and confused) impression that either one must
reconsider one’s positive response to Shoemaker or else, persisting in the
positive response, one must subscribe to the neo-Lockean idea that
memory is constitutive of identity. In the case where Brown’s brain was
divided and each half was supplied to the body of one or other of two
twin Robinsons, the question was whether Brownson () and Brownson
() were the same person as Brown. Admittedly, given the transitivity of
identity, the answer was no. They could not be the same because
Brownson () was not identical with Brownson (). But could nothing
more be said? More was certainly needed. But three years later, the ‘one
parcel’ stipulation and the universal–particular distinction20 accounted
for most of what I had in store to block the very different conclusion that
Derek Parfit was shortly going to arrive at in his ‘Personal Identity’
(Philosophical Review,  ()).21 Parfit’s conclusion was that, since
19
See Wiggins (), pp. , , –. Cf. Derek Parfit (), p. .
20
See Williams (–), (), (), and Mrinal Miri ().
21
I place on record here one not insignificant reminder that I mentioned in  but did not
emphasize. Not only does the full performance of all sorts of ordinary perceptual tasks require
both halves of a brain. The two halves are not in fact equipollent. Moreover, one should hesi-
tate before pronouncing it to be an unremarkable contingency that they are not. The halves are
 Personal identity 
Brownson () and Brownson () were not the same yet each had ‘the rela-
tion to Brown that matters in normal cases of personal identity’, iden-
tity itself was not what mattered in survival. He concluded that that
which mattered in survival was the relation *C, or rather the new rela-
tion, still to be elucidated, that we are calling R.
From a defender of the  account that I defended, which Parfit
was going to upstage, the chief thing that was really needed at that time
was for the defender to show how naturally and inevitably the ‘one
parcel’ condition arose from the exigencies of understanding better
what kind of thing a person or human being is. Further philosophical
reflection was badly needed to explain, and explain on the basis of the
sort of thing a person is or a human being is, what distinguished the
relation Brownson bore to Brown, if that case was to be allowed as an
identity and a true survival, from the relation that the splinters
Brownson () and Brownson () bore to Brown. That is what this chapter
must provide.
First, though, before entering into any of that, it will be good to make
as plain as possible on what terms an adherent of Bishop Butler’s criti-
cism of John Locke might reasonably prepare to accept Sydney
Shoemaker’s case. It needs to be clear that, on the neo-Butlerian view,
the intuitive (however provisional) willingness to allow the Shoemaker
case as a case of identity creates no presumption at all in favour of Ip or
any criterion like it. In concurring with Shoemaker about the case, we
are not undertaking to wait for the idea of co-consciousness to be clar-
ified or replaced by some better notion (of quasi-memory, for instance)
which can figure in a condition like Ip. Nor yet are we preparing to dis-
pense with the ontology of persons. Well or badly, we are simply think-
ing through our existing conception of the principle of activity for a
human being. It is not to be denied that, if Brownson is Brown, his life
story will be in one respect strange. But, on the still provisional terms
unequal because of the specialization of certain functions and most notably perhaps because
one side of the brain has been colonized by structures that subserve the singular and extraordi-
nary capacity of human beings to produce and understand syntactically ordered speech. If this
capacity is as intimately connected as it appears to be with our capacity to keep track of the past,
to direct our attention to our own perceptual and motor activity, to turn from the task in hand
in order to think discursively, or in order to reason about what we are doing with a prospect of
lighting on something else that needs to be adduced or compared, then the very thing that
renders brains asymmetrical is not accidentally connected with that which renders the idea of a
conscious self so interesting to us – and so productive of illusion. Perhaps it is almost as plau-
sible now as it was when Aristotle wrote the first sentence of Politics that Nature (however much
she does in vain) does nothing with no purpose. (The metaphor, as it has to be, will need to be
unpacked, of course. But I shall exercise the right to be as lazy about disintentionalizing it as
numerous present-day evolutionists are about the teleological language they themselves employ.)
 Personal identity
now contemplated, it need not count as ontologically or metaphysically
or epistemologically strange. For, as interpreted within the scope of our
decision about Shoemaker’s original case, Brownson’s cognitive states
need to be fully answerable in the ordinary way to all the norms by which
we standardly regulate the ideas of memory, experiential memory, testi-
mony and the rest. If Brownson claims to have planted a fig tree or
scaled Vesuvius or seen the Aurora Borealis from a fishing boat off
Orkney, he can only count as remembering such acts if he (Brown and
Brownson) really did these things. Moreover, if he thinks he remembers,
then that must create for him the normal if defeasible presumption that
he did indeed do this planting or scaling or was indeed there in the
fishing boat; and he must take responsibility for his account of what
these experiences were and were like, etc.
This picture may usefully be filled out a little more. Suppose that, in
a variation of Shoemaker’s example, we are keeping track, eagerly and
anxiously, of Brown’s survival, Brown being a valued friend and col-
league, mortally sick. Like every other substance, Brown is a this such.
(This is Aristotelian jargon, but usefully resonant here.) In our watching
over him tenderly and vigilantly, the such is taken care of by virtue of
our remembering who and what it is, with what principle of activity and
what particular determination of that principle, that we are to keep
watch over. The this is subsumed by the focused specificity of the such
that trains our concern onto the particular patient who is Brown. Here
the mode of activity that is Brown’s is one that is marked out, as I have
said already, not merely by his possession of the standard human capac-
ities but by his particular adaptation, development and realization of
these capacities. The special and recognizable shapes and forms that he
will have given at any point to the standard capacities will not of course
be indelible or permanent. But, if we seek to cherish Brown and prolong
his life, then in our concern for his particular realization of such poten-
tialities we cannot help but feel a very special concern for the physiolog-
ically vital foundation of these things, namely the brain and/or the
nervous system. Suppose then that, in some dreadful turn of events, it
is planned that this foundation, the brain, should be somehow trans-
posed to a new body. Then, in so far as we can still make sense of what
is happening and keep track of it determinately (these are not condi-
tions to take lightly) and in so far as we can regard the attempt as in any
way appropriate or in keeping (a condition to be linked with the first,
and not taken lightly, see below, §§–), our continuing concern must
be for the transfer of the plurality of specialized and refined faculties
 Personal identity 
and capacities22 that are supported by that brain/nervous system. But,
in the Shoemaker case, that was attended to – or so it seemed to those
as impressed by the case as I was. Nor is that all. It is remarkable, once
one thinks through the implications of the decision in favour of
Brownson and we see the Shoemaker case in the way I am recommend-
ing, that absolutely everything that Bishop Butler could want logically
speaking is already conceded to his objection (leaving on one side, I am
saying, Butler’s metaphysical conception of the soul itself). In order for
the case to be seen like this, Ip is not required, nor is anything else that
is very distinctive of neo-Lockeanism.23

.                  

The conclusion to be defended in the next phase of this chapter is that
Shoemaker’s case marks a turning point. If Shoemaker’s case preserves
identity (and the interim answer is that it does), then it must seem to be
the ne plus ultra of identity and survival. In so far as further thought-
experiments are to be taken seriously as conceivable thought-
experiments (viz. the transplanting of split brains, their carbon copying,
their pantographic reproduction, the putative teletransportation of
human beings . . .), I claim that the creatures or artefacts that these
further developments offer to our speculations will need to be seen as
analogous to clones, copies, reproductions, models – as all of a piece
with things that approximate to their originals, or simply behave as if
they were their originals. No doubt clones or models are more impres-
sive in their own way than were any pretenders to a royal throne who
were drilled and counselled in the role that they were to feign, then let
loose, so prepared, upon the world. But, however that may be, there
ought to be no temptation to think that clones, copies, reproductions,
etc., are their originals – even if those charged with managing the rele-
vant part of reality are bold enough to invite a clone to take the place of
the original (just as a bereaved spouse may, without confusion of any
kind, be urged in some cultures to marry the brother or sister of their
22
These are ‘part of his identity’, as ordinary people like to say. Or, as I might say in the language
of this book, these things figure in the specific determination of the determinable principle of
activity to be invoked in making judgments of identity through time between our colleague and
the person the doctors or surgeons will finally confront us with.
23
In disclaiming this affinity between the neo-Lockean view and the considerations that incline
one towards the positive view of the Brown–Brownson case, I want to emphasize that I read the
real John Locke’s chapter as containing certain insights that every sane view of personhood and
every sane view of identity will accept as such. See below, for instance, §§–. See also
McDowell ().
 Personal identity
deceased spouse).24 They not only fail the ‘same human being’ test, by
the passing of which it seems Brownson counts as Brown. And they not
only fail it for Butler’s reason. They also fall short of attaining to that
which ‘matters’ in survival – or so, for reasons that will complement and
extend Butler’s objection, I shall aver. Finally, though, if so much can be
established with any sufficiency of generality, the moment will come at
last to return to a lingering doubt and the main topic of Part Two of the
chapter, the doubt whether the way in which Shoemaker has Brown pass
into Brownson really does preserve Brown himself or constitute
Brownson the full and proper continuator of Brown. See, in due course,
§.

. -         ,    -   

So much for the plan and the question whether there are possibilities
that exceed the Brown–Brownson case. Let us go back now to where we
were before we registered the prima facie persuasive force of Sydney
Shoemaker’s example. Let us return to the version of Butler’s objection
that was set out in §.
Having found nothing helpful in C or C* as they were originally
defined for purposes of the formulation of Ip, and seeing no future in a
version of C that made do with ‘sufficient real or apparent recollection’,
the defender of Parfit’s view of the splinters Brownsons () and () (con-
trast the view stated in § above) now needs to invoke some mental state
or capacity that closely resembles plenary or identity-involving experi-
ential memory, yet is neutral with respect to identity. It will be this that
has to make survival conceivable without identity. At the same time,
though, the defender’s account of this state or capacity will have to
resemble the account of ordinary memory (or so it was claimed at §,
penultimate paragraph) in disallowing all sorts of specious or apparent
memories as not really memories (of any sort).
This task was first begun by Sydney Shoemaker (however unnecessar-
ily, according to my opinion, so far as the Brownson case is concerned)
24
Note in passing the ambiguity of ‘clone’: it may mean one of a group of almost exactly similar
things, but it may mean the whole group or type – a highly unpromising model for those who
want to count some unique clone of Brown as Brown himself. And under the first head, one may
mean a genetic clone or (as here) a more or less exact copy (of the same maturity and all the rest
as the originating being).
I have wondered who first imported the word ‘clone’ into present-day philosophy. My 
use of it was prompted, I think, by that of G. C. Stead. See ‘⌷⌴⌷⌷⌼⌺⌱⌷⌺ ⌻⍀⌱ ⌸〈⌻⌹⌱’,
Studia Patristica, ,  ().
 Personal identity 
in continuation of the work I have cited.25 But many stages beyond that,
the best version of R now on offer would appear to be that of Derek
Parfit. His latest version appears in a passage of his Reasons and Persons
where he is replying to an objection against his kind of account of the
phenomena that we normally describe in terms of personal memory
and personal identity.
Parfit writes as follows:
The objection would be this: ‘It is part of our concept of memory that we
remember only our own experiences. The continuity of memory therefore pre-
supposes personal identity. The same is therefore true of your Relation R [of
weak co-consciousness, which is the relation that our *C first set out to be]. You
claim that personal identity just consists in the holding of Relation R. This must
be false if Relation R itself presupposes personal identity.’
To answer this objection, we can define a wider concept, quasi-memory. I have
an accurate quasi-memory of a past experience if () I seem to remember
having an experience, () someone did have this experience and () my apparent
memory is causally dependent, in the right way, on the past experience.
On this definition, ordinary memories are a sub-class of quasi-memories.
They are quasi-memories of our own past experiences. . . . [p. ]
We do not quasi-remember other people’s past experiences. But we might
begin to do so. . . . Jane has agreed to have created in her brain copies of some
of Paul’s memory-traces. . . . One apparent memory is very clear. She seems to
remember looking across the water to an island where a white Palladian church
stood out brilliantly against a dark thundercloud.
Even if [Jane’s] apparent memories are presented in the first-person mode, Jane
need not assume that, if they are not delusions, they must be memories of her
own experiences. Even if she seems to remember herself seeing forked lightning,
she could justifiably conclude that she is quasi-remembering one of Paul’s expe-
riences. . . . Jane might have to work out whether it was she or Paul who had
some past experience. And this might sometimes be impossible. She might have
to say, ‘I do vividly seem to remember hearing that tune. But I do not know
whether it was I or Paul who heard that tune.’ [p. ]
Confronting the definition furnished at the beginning of this citation,26
reading Parfit’s condition () as saying that there is an experience (an
25
See Shoemaker (), pp. –.
26
I call it a definition not only because Parfit does but because a definition is needed. ‘Quasi-
memory’ is a technical term. Technical terms in philosophy are rarely scrutinized with the same
care that is devoted to such questions in mathematics or logic. Yet, as logicians will point out,
apparently small faults in definition can give rise in that area – and why only there? – to large
and unexpected, or even paradoxical, consequences. Consider the paradoxes of Russell and
Grelling – or indeed those of Richard and Berry relating to the idea of definition itself. If unex-
amined definitions can issue in antinomy within the sphere of logic, is there not a comparable
danger that faulty definitions will sever connections in metaphysics or issue in disabled concep-
tions of our active and cognitive faculties in the theory of knowledge?
 Personal identity
action or event consciously participated in) that I seem to remember, and
reading his condition () as saying that that experience was indeed
someone’s or happened to someone (for other interpretations seem obvi-
ously incorrect), we ought to begin, I think, by entering onto the record
the first question it invites. What is it for me, the rememberer, to remem-
ber or seem to remember that experience which was indeed someone’s?
Close similarity between how things were with the experience itself, sup-
posing that to be independently identifiable, and how the rememberer
rehearses it to himself as having been is obviously not sufficient. Countless
similar but distinct events are similarly remembered. Nor yet is it neces-
sary. Some remembering of an experience (perhaps most remembering,
really) is impressionistic and inaccurate in the extreme, without being
inaccurate enough to be disqualified from counting as really remember-
ing the experience. What is it then for someone to recall that very experi-
ence? Can the question be answered otherwise than by a condition of the
following kind: the rememberer recalls experience x if and only if the rec-
ollection putatively of x recovers to the rememberer some sufficiency of
the actual details of x and enables the rememberer to place x as an event
in his, the rememberer’s, life and . . . ?27 This sort of suggestion cannot be
welcome, however, because it reimports a requirement of identity and
depends on our using the rememberer’s own life to identify x.

 .                     ’ 

         ‘  -    ’
The second question we must enter onto the record is this: ordinary
memories are stipulated by Parfit to count as quasi-memories, but need
quasi-memories as such, defined as Parfit defines them, partake at all of
the nature of memory? Either that matters or it doesn’t. If it doesn’t
27
Well, that is how it needs to be normally (and there needs to be a ‘normally’). Abnormally (and
necessarily marginally), no doubt there is recollection of events the rememberer cannot place at all.
The ‘and . . .’ holds a place for any further condition of a causal or explanatory kind.
For fear of seeming to disagree with absolutely everything, I have acquiesced in the grammat-
ical forms deployed in Parfit’s proposal. But I should prefer to see them as condensations of con-
structions more along these lines: I quasi-remember climbing the spiral stairs of Big Ben on my
sixteenth birthday just if () I seem to remember climbing Big Ben on my sixteenth birthday; ()
someone did climb Big Ben on their sixteenth birthday; () my seeming to remember climbing
Big Ben on my sixteenth birthday is dependent in the right way on the event that consisted in
someone’s climbing Big Ben on their sixteenth birthday. Adapting the first question to this refor-
mulation, one obtains the following: What is it for my remembering, as in (), to relate to an inde-
pendently identifiable event e in virtue of whose occurrence () holds true? What is it for my
remembering, as in (), to rehearse the details of some particular event whose occurrence makes
() true?
 Personal identity 
matter, then Parfit loses hold of any non-accidental link between the
neo-Lockean conception and the ordinary memory that is conceived in
an ordinary factive manner (and whose presence in John Locke’s
account of these things made it persuasive). If it does matter, though,
then we must worry about the following. Surely I can seem to remem-
ber some experience that happened to someone, but without actually
remembering it at all. For instance, George IV was sincerely convinced,
having internalized and embraced imaginatively countless well-founded
reports of the battle, that he had been at Waterloo.28 If so, then he could
have rehearsed many of the details of the battle accurately enough and
as if from a first person or combatant point of view. But George IV had
better not count, simply on this basis, as quasi-remembering Waterloo.
Such questions as this are bound to collect the reply that all issues of
this kind are taken care of by Parfit’s requirement () that the apparent
memory be dependent in the right way on the experience apparently
remembered. That reply gives rise to the next doubt, however.
The next doubt is whether quasi-memory, having been vindicated as
memory by virtue of being causally explainable ‘in the right way’, i.e.
the way proper to ordinary experiential memory, can then fall short of
implying the identity of experiencer and rememberer. If the memory
state has to stand to the experience that it is a recollection of as the right
sort of effect, then the next question is how to give an explication of ‘the
right sort of effect’, ‘the right sort of cause’ or dependence in ‘the right
way’ that will place quasi-memory in the required proximity to memory
as ordinarily conceived and permit that non-identity. That is the third
question to be put on the record. (It generates a supplementary question,
but that is the subject of §.)
One way in which a follower of Parfit might try to answer that ques-
tion and manage all these doubts is to invite us to grasp the right sort of
way by examples, examples that he can exhibit. But here there is a chal-
lenge. Either the examples are normal cases, in which case there is iden-
tity, and we have already seen in § how intimately identity is involved in
the ordinary point of making a memory-claim; or else they are abnor-
mal. If they are abnormal and there is no identity of experiencer and
rememberer, then it will be inherently controversial how much they have
in common with normal cases so far as remembering is concerned.29
28
As Anthony Flew points out. See the article cited at note .
29
In ‘The Causal Theory of Perception’, Proceedings of the Aristotelian Society,  (), H. P. Grice
gave a new authority to ‘by-examples’ methods of explication; but this was without any
guarantee of the effectiveness of the method in cases where favour was to be gained at the same
 Personal identity
A defender of Parfit may say that in the normal case we see an event
befall someone and we take note of their capacity to recall the event
afterwards. Well, in the abnormal case, this defender will say, it’s just like
that again except that, after the initial transaction that impresses itself on
someone’s memory, the ability to recall it is transferred. For instance, it
may be transferred to two splinters of the original experiencer. In what
manner then (we ask) is it transferred? Transferred in the right sort of
way (the reply must go), and that is a way sufficiently like the way in which
the capacity to remember is carried forward by the person who is
involved in the normal way. Is it not obvious (the reply goes on) that, at
least in the Brownson () and Brownson () case, what happens is suffi-
ciently like? ‘Surely, for one prima facie disposed to allow the
Brown–Brownson case, that is not very controversial.’
Here and in response to the last claim, however, rather than dispute
whether the Brownsons () and () case is controversial (for that is itself
controversial, or so any follower of Butler will say) or insist again upon
the apparent normality (for theory of knowledge purposes) of the
Brown–Brownson case, I think one must make the defender stick to the
point. A stipulation de novo was being offered of ‘quasi-remember’. That
stipulation brought with it the commitment to say something general
about what ‘the right way’ was. The answer proposed was ‘a way suffi-
ciently like the way that is manifest in the normal case’. But, so far, this
has only been explicated or justified on the basis of one or two cases. Our
difficulty was that the normal case is the identity-involving case. What
was needed then were general reasons, not reasons relating to one or two
(or even three) particular cases, why identity of experiencer and remem-
berer was irrelevant to normality. Either that was needed or else (much
better and more correctly) we needed a general explanation of ‘right
way’ that was both plausible and identity free.
How then is one to move forward? No progress seems to be possible
here until theorists of quasi-memory and the R-relation that it is meant
footnote  (cont.)
time for a reconstrual of the phenomenon being explicated. Look towards the end of the cita-
tion from Parfit given in §. Ask whether one could pick up from putatively paradigmatic cases
of Jane’s sort (where memories of an event may seem rather like pieces of shrapnel left over from
an explosion) the mode of dependence on its associated event or experience that is required of
a mental state that is to count as partaking of the nature of memory. What ideas will be at our
disposal for us to focus on the relevant and requisite aspects of such examples if they are included
among the paradigms intended to fix the extension and intention of ‘causally depends in the
right way’? Well, evidently, the ideas of experience, having an experience and seeming to remember. But
in availing ourselves of the last we must of course prescind from all elements of remember itself
that the definition of Q-memory will seek to dispense with. ‘By examples’ is not the straightfor-
ward business it is often taken for. It is certainly not straightforward here.
 Personal identity 
to define – I shall sometimes call them the Q-theorists – measure up to
the requirements of full generality in definition-giving. Equally, though,
and by parity, let it be conceded that no progress seems possible until
those who are as sceptical as I am about these definitional efforts furnish
a better account of our own reading of the semantical and epistemolog-
ical point of attributions of experiential memory. We must clarify the con-
nexion between this point and questions of identity.

 .   :                    ,

                                       
      
On the terms that I propose, there are various ways in which the Q-
theorists might advance. I should need to shadow all of them. But at this
stage in the argument, a digression may prove to be a short-cut. We shall
understand better the utility and role of Parfit’s definition and, as a
bonus, approach closer to the semantical point of the attribution of
experiential memory if we ask why Parfit did not offer a simpler propo-
sal. The simpler proposal is that X quasi-remembers walking from St
Paul’s to Trafalgar Square (say) just if it is for X exactly as if he remem-
bered walking so except that it may not have been X who did the walking
in question. Or (better) X quasi-remembers walking from St Paul’s to
Trafalgar Square just if there holds, under the headings of causality and
whatever else, everything that is required for X to remember walking
from St Paul’s to Trafalgar Square except possibly this: X’s being identical
with the person who did the walking.
Under this proposal, the idea would be that quasi-remembering ␾-ing
is X’s remembering ␾-ing less X’s being identical with one who ␾-ed.
Quasi-remembering is remembering with or without identity of exper-
iencer and rememberer. Of course, if (as most people suppose) one’s in
the ordinary way remembering walking from St Paul’s to Trafalgar
Square entails that one did walk so oneself, then the remainder that
results from the subtraction from ordinary memory must not be reck-
oned under the scheme of X’s remembering ␾-ing and X’s not being
identical with the one who did this ␾-ing. But we have already seen an
obvious alternative, namely to say that everything is for X as if he
remembered walking from St Paul’s to Trafalgar Square except that, in so
far as that remembering is held to require X’s identity with the walker,
X may fall short of what is required.
To show that this will not do, a critic on my side of the argument will
 Personal identity
begin by protesting how easily this kind of ‘except in so far as’ can have the
effect of producing a description that is indeterminate for sense. Let the
Q-theorist consider the near vacuity of ‘It’s as if there were a thunder-
storm in progress except, maybe, in so far as that requires something to be
happening in the sky’, offered as a definition of ‘quasi-thunderstorm’.
Here a Q-theorist who saw some future in the simple proposal that
Parfit didn’t offer would surely retort with the challenge to prove that this
was a fair parody of his definition of ‘quasi-remember’.30 What then?
Before rising to the challenge to prove anything, I think the critic should
point out that one could never demonstrate in the way by which Parfit’s
defender insists it be demonstrated that ‘quasi-remember’ is like ‘quasi-
thunderstorm’ or that it is unlike. For suppose, as is usually the case with
real words for non-artificial kinds or acts, there exists no analytic defini-
tion at all of ‘remember’.31 Then any faultiness (vacuity, indeterminacy
etc) in ‘quasi-remember’ might not be demonstrable by a strictly analyt-
ical procedure – just as its soundness would not either. But not only that,
the critic ought to say. If there is no definition of ‘remember’, then there
is no future in the project of defining ‘quasi-remember’ by subtraction
from ‘remember’. For, in the absence of an analytical definition, there is
no question of arriving at the thing needed by a procedure of deletion.32
30
Another thing he might do is to claim that, as defined, a quasi-thunderstorm might be blamed
for turning the milk sour. But this scarcely restores ‘quasi-thunderstorm’ to any parity with ‘thun-
derstorm’, e.g. as the name of a causal-cum-explanatory kind.
31
In Tractatus Logico-Philosophicus (see especially .), Wittgenstein sees it as a precondition of the
definiteness of the sense of an expression that it should have some unique exhaustive decompo-
sition into semantically independent indefinables. Wittgenstein assumes that among senses there
operates something analogous to the Axiom of Regularity (Fundierungsaxiom), which excludes in
the world of sets the existence of an endless downward sequence from a given set A in the form:
. . . (D⑀C), (C⑀B), (B⑀A). But against that one will want to object, with the later Wittgenstein, that
grasping the sense of an expression cannot, in the normal or basic case, be the learning of a def-
inition – or the as-if learning of one.
32
This is the only serious thing that could be meant by the project of subtracting the concept of
identity from that of experiential memory. If someone starts on the process of definition, someone
else can, it is true, start on the process of deletion/subtraction. But how can the person deleting
be sure he has finished? How can one ever say the new definition is completed?
Someone might say that ‘swift’ (the bird) has no definition but, all the same, one could define a
new term ‘martlet’ by saying that a martlet was a swift without any feet. What would be wrong?
Well as a recipe for painting heraldic devices, that would be all right. But as the definition of a pos-
sibly non-imaginary bird? Even in the case of a bird like the swift, indefinitely many real things
about it may presuppose feet. Even if the martlet mated on the wing, could it hatch or feed its young
on the wing? If not, how does the martlet perpetuate itself ? Is the martlet in the heraldic image a
well-defined kind of bird at all then? The grammatical correctness of ‘X is a kind of swift that lacks
feet’ or ‘X remembers Y’s experience except that X may not be identical with Y’ proves nothing.
In his inaugural lecture ‘Meaning and Truth’, Logico-Linguistic Papers (London, ), P. F.
Strawson spoke of the ‘bogus arithmetic of concepts’. For a development of that thought, see §
of Jennifer Hornsby, ‘Feminism in Philosophy of Language’, in The Cambridge Companion to
Feminism in Philosophy, ed. M. Fricker and J. Hornbsy (Cambridge, ).
 Personal identity 
Defining a new term by the deleting of all uses of the identity concept
from the definition of an old term would be an honest and straightfor-
ward enough method of introducing the new term. But there is no ques-
tion of making such deletions from a definition whose actual text can
never be submitted to the blue pencil.
This response to his defensive challenge will encourage the Q-theorist
to issue another. If there is no lexical definition of ‘remember’, then how
does ‘remember’ have its semantic identity and how are we to be assured
of its soundness? Will the critic please explain himself ?
The critic’s reply must be that ‘remember’ has meaning in the way in
which most non-technical words do – not by virtue of introducing a
concept built up from simpler or more transparent concepts, but by
virtue of use. ‘Remember’ has its meaning (he may say) by standing for
a concept with a distinctive place in a whole skein of coeval concepts
that came of age together, developed together and now work together to
some present purpose. The work of these concepts in the case of
memory is to make it possible for us to undertake the familiar but multi-
ple operations and transactions by which we sustain and extend the
economy of knowledge and awareness. What is more, that work pro-
gresses. Given this picture and given that there is no question of laying
out the definition of ‘remember’ and scoring out every component that
imports ‘⫽’, the whole question between definer and critic must shift
(according to the critic) to a different and more interesting place: could a
species of memory that was identity free play the role that is played in
the epistemic economy by ordinary memory?
One major role of experiential memory, the critic will now say, is that
it supplies unmediated (albeit fallible) information that one can take oneself
to have got on the basis of one’s own experience in the past. (Is not the role of expe-
rience analogous to that which was assigned to gold or silver in older pic-
tures of the real economy?) If no way is manifest for quasi-memory to
do that, then the attempt to define ‘quasi-remember’ by ‘everything is as
if . . . except possibly in so far as . . .’ will inevitably subvert the very thing
that sustains the signification and point of a concept such as that of
memory. Any serviceable or intelligible notion of quasi-memory that
depends on the given meaning of ‘remember’ needs to respect the links
that connect experience with knowledge. And these involve identity.33
Parfit’s own proposal is still pending, still held in reserve, remember.
33
‘Des expressions et formes qui ont un sens dans l’action vraie, sont excitées et employées hors de
leur domaine – et perdent pied. Mais le philosophe ne s’en aperçoit pas’: Paul Valéry, Cahiers I,
ed. J. Robinson (Paris, ), p. .
 Personal identity
But, since a real difficulty appears to lurk here for all accounts of quasi-
memory, it will be best for the critic’s line of argument to be filled out
before we return to Parfit’s proposal.
Suppose that our use of ‘remember’ depends on our participation in
the living use of the verb phrase ‘X remembers (–)ing’. If it is among the
expectations conditioning this use of the word that the epistemological
role of experiential remembering should be to help provide us with a
starting place for the attempt (by further inquiry, in partial reliance on
the testimony of others, etc) to extend our knowledge about how things
are in the world,34 then the starting place need not be something philo-
sophically indubitable or infallible. The real charm of epistemological
foundationalism, once we forget considerations about immunity from
error, was rather this: that it saw that, if there is to be knowledge, there
have to be some unmediated starting points. Among these there must
surely be states of perceiving and remembering, including personal or
experiential remembering. But if that is right, then the first thing that
matters for the theory of knowledge is that perceptions and memories
should be accorded by that theory their full presumptive title to count as
the place from whence, however fallibly or defeasibly, we begin. For pur-
poses of the pretensions of quasi-memory to play a role analogous to
memory, it matters that quasi-memory should be a state of mind or
faculty that not merely permits an exercise of it to count as a direct
source of knowledge but constitutes it as a place to begin. But how is that
possible (one wonders) if there are doubts about how Q-memory places
the experiences it relates to, or if quasi-memory does not make the
rememberer himself responsible either for the correctness of the recol-
lection of the experience that is to be a starting point or for the reliabil-
ity of the perceptions made at the time of the experience? (Q-theorists
may appeal at this point to testimony. But our handling of testimony
depends on our proper deployment of our own experience. Nor would

34
The same goes for the role of direct perception and one’s memory of what one believes. In the
argument of these pages, I follow the argument of my article, ‘Remembering Directly’ ().
That argument, I should record, is an attempt to develop in one particular direction one of the
points made by Evans, The Varieties of Reference (Oxford, ), pp. ff. (I cannot understand
Parfit’s answer to Evans, given in his (), p. .) How dare I, having (in effect) criticized Q-
memory theorists for not demonstrating the soundness of their definitions, offer non-conclusive
arguments, all founded in the suspicion that the sense of the word ‘remember’ is at once inde-
finable and sound? I only dare because the dialectical positions are not exactly symmetrical. First,
we scarcely know of any correct analytical definitions. So the most we have is the prospect of a
correct definition of Q-memory. Secondly, ‘Q-remember’ is the neologism. We ought to refuse
to use it till it is given a determinate use. In the case of ‘remember’ it has a determinate use. The
conviction is at once defeasible and indispensable to us.
 Personal identity 
testimony itself be exactly what it now is once quasi-memory was
counted among the sources of the testimony of others.)
Suppose it has started to rain and I need to be sure that, if I am going
to go out later, then I shall not get wet. Then I need to be able to take it
as a perceptual datum that I see that it is raining, and as a memory
datum that I have an umbrella. If the question arises in my mind ‘Am I
sure I still have an umbrella?’, then I need to be able to summon another
memory, such as the recollection of having seen it very recently hanging
on a peg. I need to be able to say from personal memory not only ‘there
are pegs in the hall and by the back door’ but also ‘I remember seeing
the umbrella there yesterday night.’ From perception and memory and
memory of coming in last night, etc., I need to be able to say: ‘I’m at
home. This is the back door through which I entered. So the umbrella
must now be hanging in the corridor at the back.’
The trouble with such a train of thoughts in the form in which it
would need to be redeployed under the Q-modified conception of
memory is that the idea of an umbrella-check effortlessly organized by
someone’s idea of his own single-tracked unitary life becomes proble-
matic in a quite new way – precisely as we see it doing in the long cita-
tion from Parfit given in §. We can dispense with infallibility. What we
cannot dispense with is the organization of such thoughts by starting
points consisting of the unmediated presentations of memory and per-
ception and the accreditation of such presentations by their provenance
as unmediated. That is not all. To know how to start upon any future col-
lation of evidences I shall need to know which findings had that direct
provenance. If I am remembering in the normal fashion and I can
expect to place that which I remember, then I shall know, as and when I
need, what explanation to seek of the relevant perceptual and memory
findings, both the correct ones and the mistaken. I can weigh how com-
pelling each such finding is and try to put a body of them together into
a coherent narrative that accommodates together and explains together
their several verisimilitudes or falsisimilitudes. But how else can they be
placed so except in a partial narrative of my life? For thoughts such as these,
identity is not eliminable.
The quasi-memory theorist may react to these difficulties by suggest-
ing that the normal truth-condition for remembering be modified in
such a way that all I need to be able to claim is ‘someone in an R-rela-
tion to me saw the umbrella hanging on a peg’. If the Q-theorist is stick-
ing to his brief, however, then he has no title to that explanation. Even if
he can whitewash the ‘me’, he cannot appeal to the relation R in the
 Personal identity
explanation of ‘quasi-remember’. For quasi-memory is one of the things
he needs to use in order to construct the definition of R.

.       ‘    ’

That is the end of the digression. I hope it helps to discharge my obliga-
tions under the accord proposed at the end of §. From the considera-
tions rehearsed one has a lively sense of how wise it was for Parfit to
define ‘quasi-remember’ positively and outright rather than by an
attempt at subtraction. One sees how wise it was for him to invoke ordi-
nary ‘remember’ itself (in the existing sense) only obliquely (as in his con-
dition ()). But one has some sense too, augmenting one’s awareness of
the open questions still left on the record (§§–), of the difficulty to be
surmounted in any definition of quasi-memory that will respect the
linkage we need to assume between experience, memory and empirical
knowledge. At this point then, in advance of drawing any further con-
clusion, let us draw breath, and recapitulate the state of the argument
about Parfit’s definition of ‘quasi-remember’.
In §§–, after it seemed that an identity-free concept of ‘remember’
was needed and it was furnished by Parfit, two questions were raised
about the definition of quasi-memory quoted from Reasons and Persons; a
third then arose concerning the phrase ‘in the right way’ as that figures
in Parfit’s condition (). Condition () reads ‘my apparent memory is cau-
sally dependent, in the right way, on that past experience’. In §, the sus-
picion was voiced that identity of rememberer and experiencer, so far
from being cancellable, is in fact integral to experiential remembering –
as integral as it is to knowing what you’re remembering. This last is some-
thing you normally do by exercising your presumptive right to place the
experience apparently remembered within your own life.35 It seems to
presuppose the very thing that is still in question.
The time has come to advance to something new, if only to a new sort
of question about Parfit’s positive proposal. Here is the new question.
35
A Parfitian might respond to these difficulties by allowing that I have given a correct account of
how things must be subjectively speaking with the one who remembers. The Parfitian might
allow that the rememberer’s thoughts must indeed be as I have said. But that does not mean that
experiential memory as such requires it. My reply to this would be multiple and would ramify,
no doubt, in reply to the Q-theorist’s replies to my replies. But the first point is that there is
nothing as yet in the definition of quasi-memory that answers to the need for one who is remem-
bering and seeking knowledge to organize his thoughts in the sequence suggested at the end of
§. For it is not obvious that anything in the Q-memorist’s conception of memory can furnish
the requisite materials for starting points that can be fortified as starting points by being
addressed back to the course of the rememberer’s life.
 Personal identity 
Does Parfit’s condition () seek to discharge a responsibility to say how
memory is to be distinguished (as indeed it needs to be distinguished)
from the other mental states such as conceiving or fancy or imagination
or hope or intention . . . which present objects or events to us? If so, does
‘right way’ require the operation of a certain distinctive mode of mental
functioning? Or does () address a normative question, one also needing
attention if quasi-memory is to subserve the growth of knowledge in the
way in which memory does, a question about the goodness or badness
of the fit between how the experience really was and how the recollec-
tion presents the experience as having been?
Interpreted exclusively in either of these ways, as concerned either
with mode of functioning or with norm, condition () appears to be
insufficient to constitute quasi-memory with the very same nature, epis-
temologically speaking, as normal memory. It seems likely then that con-
dition () is intended to address both issues at once. Perhaps the thought
behind condition () is that what quasi-remembering requires is that
quasi-rehearsing a thing depend causally on some mode of mental func-
tioning or a mechanism, but a mode or mechanism which itself satisfies
some sort of normative requirement. Perhaps the normative require-
ment is this: that, for a wide variety of ␾, the mode or mechanism should
bring it about that the experience is presented afterwards as having been
␾ if and only if it really was ␾. For instance, the mechanism should bring
it about that my memory of climbing the spiral stairs of Big Ben is of a
climbing, is of a climbing in the heat, is of a climbing something vertig-
inous, if and only if the experience itself was one of climbing, in the
heat, something vertiginous . . . etc.
Read along these lines (let us not pause to make the choices needed in
order to complete the task), Parfit’s condition () certainly comes to life.
But how well does the condition then provide for anything analogous to
the distinction between better and worse (normal, identity-involving)
experiential memory? There is no escape from the worry that the nor-
mative question of how well or badly someone’s memory mode or mech-
anism has functioned or functions needs to be distinguished from the
question of how satisfactorily the mode or mechanism involved approx-
imates to a memory mechanism. Can condition () admit incomplete or
imperfect or partially wrong or oddly produced memories as quasi-memories?
If quasi-memory and memory are to be generically related, it needs to
accommodate all of these. If the putative relation is to have any of the
properties that first motivated the search for something like quasi-
memory, then the question is important.
 Personal identity
At this point we will turn again to Parfit’s definition as it lies on the
page, and then at last (with a start perhaps) we must wake up. The thing
we see that Parfit presents there, once we attend properly to the passage
I have quoted in §, is not a definition of ‘quasi-remember’ or ‘quasi-
memory’ at all. It’s a definition (he himself announces that it is a defini-
tion) of ‘have an accurate quasi-memory’. Inaccurate quasi-memory is
not provided for.
This is not a tiny oversight, but the trace of a major philosophical diffi-
culty. Nor is it hard to guess how the definition in front of us came into
being. Putting together the two distinct issues of mode of mental func-
tioning or mechanism and of norms of correctness, but being aware of
the extent to which his condition () screwed things down too tightly,
Parfit must have seen that he had to ‘balance’ his proposed equation
(equivalence). Then he found (or so I surmise) that the only way of doing
that was to add something to the left-hand side, rather than to remove
an unwanted surplus on the right. So the definiendum was changed in
the way we see in front of us.
Here one cannot help but sympathize. For it is simply not obvious how
to lighten the right-hand side, or what to remove. The trouble is, though,
that, as a technical term newly defined, ‘accurate-quasi-memory’ has
now to be seen as a (so to speak) fused term. The components appear to
have independent meanings, but, when they are taken as combining to
mint a new technical term, no separate meaning has been given to them.
Under this mode of definition, nothing Q-analogous to inaccurate ordi-
nary memory could be provided for.
To dwell, as we do so often, on examples of highly accurate and oth-
erwise perfectly ordinary remembering tempts us (I conclude) into
elision. Such examples distract our attention from the difference
between questions of mental mode/mechanism and questions of cor-
rectness.36 So soon as these issues are allowed to coalesce, as they seem
to have done in Parfit’s definition, quasi-memory leaves the epistemolog-
ical economy with a marked deficit.
In the first place, if quasi-memory excludes inaccurate memory
and/or excludes variants on the actual mental modes or mechanisms of
memory, there is already a serious problem. For the growth of knowl-
edge by criticism, refinement and correction really cannot dispense with
36
It may seem that I believe I have shown anti-naturalistically that these questions are utterly inde-
pendent. No. The mechanism is no doubt selected as one for which there can be a norm. The
operation of the norm is conditioned by the mechanism for which it is the norm. None of this
is denied by one who simply insists as I am doing on the distinctness of distinct questions.
 Personal identity 
either of these possibilities. The relation R that Q-theorists want to
define will not then be built on memory at all.
In the second place, if quasi-memory is to be akin to memory at all or
to count as a direct source of knowledge about that to which it pertains,
then there needs to be room for the whole purport of a given act of
quasi-recollection – room for its pretensions qua memory state as well as
room for that which it purports to present and room for how it presents
that thing as having been – to be determinable independently of ques-
tions of the accuracy of the recollection and the goodness of its fit with
the experience. Condition (), taken as it stands, makes this separation
problematical for Q-epistemology.
It may seem that the difficulties that arise from the conflation I allege
of questions of mode or mechanism and questions of correctness are
imported by the contingent particularities of Parfit’s definition. But that
style of definition arose out of his need to purify the memory concept of
identity. It was indispensable to the campaign to show that identity was
irrelevant to the thing which mattered constitutively in Brown’s survival
in the shape of Brownson.37

PART TWO
.        ,                
         
Here is a summation of what has been claimed so far in this chapter: for
personhood as we know it, the identity of persons coincides (I began by
37
There is a special difficulty in disentangling memory from identity. Unlike a perception, whose
occurrence ties it to what was there to be perceived and whose correctness can in principle be
regulated from other investigations of what was there to be perceived at the time and place of
the perceiving, and unlike a portrait whose title and original provenance ties it to some sitter to
whose appearance it is answerable, the act of recollecting an experience, the event e to which the
recollection is answerable for its correctness, does not permit the identification of its content or
referent, namely e, to be made on the simple basis of the place or time of the occurrence of the
act of recollection. The place and time of doing the act of recollection afford no indication what-
ever.
On the inner view of memory, the importance of this comes out in the need (already men-
tioned) for the rememberer to place the thing remembered, the experience, and thereby signal
the correctness condition of the memory-state in question, by reference to something in the
sequence of his own life. He can only do this sort of placing or fixing of e by engaging in the
kind of thought that the definition of quasi-memory does nothing to make sense of and that is
problematic for Brownsons () and ().
On the outer view of memory, the same point comes out first in the need to ascertain from
the subject what memory presents to him (see the inner view) and secondly in the need to iden-
tify by reference to the outer life of the rememberer the very experience – wherever there is one – to which
the recollective act or memory state is answerable for its correctness.
 Personal identity
suggesting) with the identity of human beings. Human beings are sub-
stances possessed of a specific principle of activity to which, in the
course of a life, each one of us gives his own yet more specific, more and
more distinctive, determination. Prominent among the specifically
human activities is our exercise of the cognitive faculties. Faced with
Sydney Shoemaker’s Brown–Brownson case, our provisional first
finding was in favour of the identity of Brown and Brownson, because
Brownson appeared to be the determinately traceable functional contin-
uator of Brown (we said) and Brownson seemed to inherit (in the manner
in which any ordinary person who has suffered no such adventures is
constantly inheriting from himself) the perfected epistemic and other
capacities of Brown. He carries these forward through time, together
(we assumed) with Brown’s other skills and abilities. Moreover, we found
on further examination that the judgment that Brownson is Brown lies
well outside the reach of Butler’s objection to Locke. If there are diffi-
culties with the Brown–Brownson case, they do not reside here. Neither
for purposes of Brownson’s thoughts, nor yet for the purposes of describ-
ing in third person mode (or in philosophical mode) the set-up that
includes those Brownson thoughts, is there the slightest need to try to
construct the identity free notion of experiential memory. Nothing
appears to prevent us from thinking of Brownson as having full cogni-
tive responsibility for the claims that he makes from direct personal
memory.
That summarizes only what happened up to §. Now for the rest.
Consider the case where we have not Shoemaker’s Brownson but two
splinters, Brownson () and Brownson (), resulting from the transplant-
ing of the two halves of Brownson’s brain into bodies of twin Robinsons
(each found available, we are to suppose, in debrained condition). Here,
we allowed that we may want to say that it is as if Brownson () and
Brownson () remember. But we declared that, on the terrain lying
beyond the case Shoemaker introduced, no stronger claim ought to be
allowed than it’s as if. For neither Brownson () nor Brownson () is the
same as Brown, and there is no newly minted, properly defined remem-
bering-of-experiences relation (or so it was argued) in which Brownsons ()
and () can stand to Brown. (That was the conclusion of §§–.) No new
memory concept that was modelled on the notion of memory that once
commended Locke’s discussion to us ought to try to make room for per-
sonal or experiential remembering in the case of Brownsons () and ().
Friends of quasi-remembering seek to improve on the finding that in
various ways it is for Brownsons () and () as if each of them were
 Personal identity 
Brown. But in so far as quasi-remembering gets us beyond that anodyne
judgment into something that is stateable without the use of the ‘as if ’, it
brings nothing but conceptual disruption.38 ‘Quasi-remember’ is ill-
defined; and in its application to Brownsons () and (), a confluence
appears of two things that scarcely mix, the idea of a person as a singu-
lar thing with an individual biography and the idea of a person as a
quasi-universal, susceptible of multiple instantiation.

.                  

The force of the considerations mustered so far would be blunted if
there were some way to contend that Brownsons () and () might after
all participate in the ordinary processes by which empirical knowledge
is expanded. Before we go further it needs to be enquired (even at risk of
doing again work done already) what prospects there are of its being
shown that, despite the fact that their ‘earlier lives’ are shared,
Brownsons () and () could in their own way keep track of things they live
through, mark ‘memories’ by their provenance, and place the experi-
ences that ‘memories’ record by the position these experiences once
occupied in their lives, etc.
I begin on this by saying that the whole case for examining this pos-
sibility may seem to depend upon the prospective availability of some-
thing like the relation R. If so, let us note how easily and unthinkingly we
can relapse into the supposition that R is available already. Perhaps this
is because we are bewitched (as I have said) by the putatively sufficient
condition that is given such prominence by the hackneyed, familiar but
special examples where the norm and mode/mechanism constraints are
perfectly and simultaneously satisfied. These happy cases furnish no
understanding of the necessary conditions of the presence of the R rela-
tion. All we really know about R we learn from purported cases of it and
the role that R is meant to play – in philosophy. But that measure of
understanding cannot (I have claimed) assure us that there really is a rela-
tion that will both perform that office and be independent of identity.
Indeed, the rationale began to appear some sections ago for insisting
upon some residue of the ‘one parcel’ condition mentioned in §.
38
At this point in the argument it may be desirable to point out that, for one who agrees to discuss
these thought experiments but is drawn to the line I am still taking at this point in the argument,
it is not at all necessary to deny that the splinters Brownson () and Brownson () have mental states.
It is not even necessary to deny that any of the respect owed to Brown should be extended, where
they actually exist, to Brownsons () and (), however pathetic they seem. They are after all repos-
itories of a kind. See §.
 Personal identity
In the second place, it is worth trying to think further about the epis-
temological condition of splinters such as Brownsons () and (). In so
far as they proceed as if they were direct rememberers, and in so far as
they take their experiential memories as direct presentations to them of
their own past, they must become party (we have said) to indefinite
quantities of error. Each of them says that he remembers a momentous
one-to-one dialogue with Isabel (say) or a confidential man-to-man
transaction with Stephen. But there are two Brownsons and neither of
them is the same as the substance that was there in converse with Isabel
or Stephen. Each survivor thinks of himself as a single person with cog-
nitive faculties, and a life altogether of his own, but such thoughts are
just wrong. Again, each of the Brownsons thinks that he is a person who
lives a life, a substance. He cannot be what he thinks he is. Moreover, if
experiential memory is to work as it must work for purposes of cogni-
tion, there is no thinkable alternative but for it to induce in Brownsons
() and () the sorts of beliefs that unsplintered people have.
Nevertheless, the beliefs that Brownsons () and () need to have are
beliefs that cannot all be true. For the sake of arranging for all this to seem
possible, why on earth should anyone want to pretend that R is well
defined when it is not?
Paul Snowdon has helped me to answer this question. The point is
that, even after issuing all these denials about Brownsons () and () and
their supposed experiential memories of Brown’s life, the human being
theorist is still faced with some impressive apparent memories. Whatever
else one says, it is hard to deny that each Brownson represents a reposi-
tory of memory traces or traces of memory traces relating to events that
took place before either of them existed. So be it. We can and maybe we
must treat them as repositories. It is as if they had each been there. There
is no more one should say, however. In Parfit’s example, Jane might be
the sole remaining repository of information about the appearance of
the portico of some remote island church destroyed, since Paul’s death,
by flood. Nevertheless, despite any gratitude we may feel for the exis-
tence of this vestige, let us be clear that Jane’s state of being such a repos-
itory falls far short of the cognitive condition of one who displays her
memories in cognitive engagement, takes responsibility for them, places
them, and exercises an option to accord them the status of ordinary
direct memory. For the same reason, the attainment of repository status
by Brownsons () and () falls short of that which would have mattered
in the survival of Brown.
A third remark. By practice, or even without practice, one can fall
 Personal identity 
readily into the frame of mind where, having been impressed by the
Brown–Brownson case, one finds it impossible to think that the
Brownsons () and () case represents the passing away of Brown. The
explanation is not far to seek. At this point in the supposed history of
Brown, one very easily falls into thinking of Brown as a thing that per-
sists in Brownson () and Brownson (). One conceives of Brown as a
thing that persists in its/his instantiations, a thing that is wherever they
are – just as the sail that lies over you, over me, and over a friend of ours,
is where I am, where you are, and where he is. (See Plato, Parmenides,
b.) In short, one thinks of Brown as a concrete universal. Nothing
need be wrong with that. Indeed it is very likely that, with a little care,
one could invent a contradiction-free way of speaking of such a corpo-
rate being, not merely as a thing with a past but as one with a future.
One could index the thoughts that are ascribed to the corporate being
by reference to the members or constituents that give local hospitality
to the thought, even as other members or constituents lack it or have
another, possibly conflicting, thought. Under the new convention, the
corporate being Brown will be large, but within him he may safely
embrace contradictions. Moreover, if we are careful, we shall be able to
say this as theorists without ourselves contradicting ourselves.
Nevertheless, Brown reconceived as such a concrete universal is not the
sort of thing whose survival was to have been described in the case of
Brownsons () and (). Nor is this how we conceive of subjects of expe-
rience when posing the question of personal identity. The problem
relates to subjects who have lives of their own, to potential authors of
autobiographies, who can say ‘I think. . . .’, rather than ‘In respect of
member m, I think. . . .’
Let it be clear that the difficulty here is not that metaphysics needs to
look askance at corporate beings. Perhaps departments of physics or
psychology or philosophy could conceive of themselves as such. By an
effort of imagination or of memory, one might find even now some way
of thinking of such departments as participating not merely in power
or glory (high ratings) but in the growth of knowledge or understand-
ing. But there is an important difference here from the sort of thing
which Parfitians might think of as surviving or quasi-surviving. For, as
we have said, Brown thinks of himself not as corporate and not as com-
posed of splinters. If a concrete universal Brown thought of himself in
(Shoemaker’s) Brown’s way, he would be wrong. Departments of
physics, on the other hand, considered as cognitive beings that do not
need to conceive of themselves as not having members or as not multiple,
 Personal identity
rather thinking of themselves as pursuing the truth about physics along
the multiple paths pursued by different individuals or teams, must see
it as their business to push physics forward by promoting individuals’
and teams’ constant effort, in seminars, in the laboratory or the
canteen, to explain anomaly and transcend inconsistency. A depart-
ment of physics has no need to suffer from the delusion that it is itself
an individual person seeking after truth. Its corporate thought had
better be that it is what it is, namely a corporate thing in search of the
truth about physics by dint of the efforts of its members. Not only that.
A corporate being such as this can offer a perfectly ordinary account
of what sort of thing its members are. Suppose a correspondingly cor-
porate notion of Brown were confected, what then? Then corporate
Brown would be conceived as a person with members which could see
itself as a person with members – and could say what sort of thing any
one of its members was. Corporate Brown is not then the sort of thing
to figure in an autobiography, a story of itself as one individual person.
Parfit’s account of survival is not, however, designed as an account of
the survival of a corporate sort of thing. It is an account of the survi-
val of things on the ontological level not of corporate Brown but of the
members of corporate Brown.39
Epistemology has long needed to engage with the collective nature
of our cognitive labours, with knowledge by hearsay, with the role of
testimony, with the conditions for the accumulation and collective crit-
icism and emendation of knowledge. Yes. But no sane enthusiasts for
these tasks will seek to dispense with the hard-won insights of the
empiricist epistemology of experience, eyewitness and recollection by
individuals with lives and histories of their own. Not even an under-
labourer can be a splinter like Brownson () or (). The charm of neo-
Lockeanism was that it seemed to engage with our given notion of
memory – of personal memory, that is. When defended by all the nec-
essary philosophical refinements, it gradually disengages with that
notion and puts into question the only account we know how to give of
the evidential fabric of our knowledge. Not only does its revised con-
ception of memory denature memory. It has now put itself in danger
of disengaging altogether with the idea of an individual ‘self ’ which
was its starting point.

39
If a Parfitian person were corporate, what account would it offer of its own members? Does each
member itself have members? In case there is a reader who wants to pursue this question and
in case that may be useful, he will find an informal statement of the Fundierungsaxiom in note .
 Personal identity 
.                  ,  
                    
          40
In the philosophical transition that will carry us slowly back to our start-
ing point, viz. the Shoemaker case in its original state, there remains one
other case I have to comment upon, namely the case where the brain of
Brown is split but only one of the two portions is successfully trans-
planted into a Robinson twin. The other dies, we may suppose. Let us
call the survivor of these events Brownson Sole. In the circumstances (it
will be asked), how can we refuse to treat Brownson Sole as Brown
himself ? If Brownson Sole claims to have seen the Aurora Borealis from
a fishing boat off Orkney on which he was the only US national, if he
gets the numerous details of Brown’s adventure right and this is no fluke
or accident, then how can we help but treat Brownson Sole as one who
was indeed there? It may seem that the only thing that is left to ask is
how well he remembers seeing the Aurora Borealis.
We can go as far as we like with the as if (I reply) and we can take the
as if as lightly or as seriously as we wish to take it. Indeed we can take the
as if fully seriously enough to experiment with questions put to a being
that we count only as a ‘repository’ or the carrier of traces of some event
they were not present at. When someone devises a whole philosophy of
als ob, that can be taken seriously too. But consider the judgment that
Brownson Sole is the same as Brown. If it is true, it ought to depend only
upon Brown and Brownson Sole. It ought not to depend on an assurance
about the existence or non-existence of another thing altogether besides
Brown and Brownson Sole. It cannot depend ad hoc on the accomplished
death or destruction or non-viability of the other half-brain that was
poised to animate some rival to Brownson Sole. Any verdict about
Brown and Brownson Sole issued in such dependence would infringe the
Only a and b rule, D(x) of Chapter Three, §, which is founded on D(vi)
of Chapter Two, § (ad fin.) and the sortalist cum conceptualist argument
for D(vi). The argument there was all of a piece, moreover, with our
commitment to the distinction between a particular and a universal. (See
above, §, the paragraph beginning with the words, ‘A third remark’.)
At this point, it is easy to imagine that someone may say that the Only
a and b rule looks plausible enough as an abstract logical requirement,
40
Readers following the suggestion in the Preface that § following should stand as a conclusion
for the book will probably be familiar enough with present philosophical trends in thinking about
personal identity to guess immediately what has been in progress here. If not, § and § of this
chapter, with or without §, will supplement §.
 Personal identity
but ask how the normal theory and practice of individuation as we have
described it could ever underwrite its satisfaction: ‘In practice, in the
actual business of making identity judgments, the Only a and b rule is
unworkable.’ My answer to that runs as follows: if Brownson Sole were
the same as Brown, where Brown is a human being and thereby a cog-
nitive being, then Brownson would have to be the same human being
and the same cognitive being as Brown. But Brownson Sole came into
being from Brown not in a manner constitutively sufficient to preserve
the transitivity of identity or to perpetuate the activity of a human being
as a cognitive being. (See § following.) On these terms, there is no ques-
tion of Brownson Sole’s being the same as Brown. The Only a and b rule
is not unworkable at all. It asks no more of us than that, when we address
questions of identity, we should persist in a way of thinking that pre-
serves the distinctiveness of identity and marks the difference between
singular and universal. If identity itself is a matter of indifference to you,
of course you can ignore the rule. (Cf. Chapter Three, §, paragraph .)
In that case, your judgments will not differentiate identity from resem-
bling, succeeding, going proxy for, being a replica of . . ., and you don’t
care. If you do care, don’t ignore the rule.

.  ‒        

There is a point here which, if it deserves anything at all, deserves to be
followed through completely. The principle we have just invoked is that,
since our judgments of identity need to be informed by consideration of
the kind of things that are in question and thus by the principle of activ-
ity of these, our positive judgments of sameness are answerable (by their
nature) to a norm that requires anyone who makes a judgment of iden-
tity through change to assure himself whether the change is one that pre-
serves this principle of activity – and equally requires him to withhold the
verdict of identity in any case where the change that is in question
would, in other cases that exemplified the same process, fail to preserve
that principle of activity.
Once such a principle is announced, it will seem that in our applica-
tion of it the question ought constantly and repeatedly to arise: ‘what
counts here and in this or that connexion as the same process?’ One will
expect that a constant need to engage with such questions will make us
party to an elaborate system of casuistry by which the specifics of each
ongoing process can be interrogated in the light of the formal proper-
ties of the identity relation and of the character of the thing-kind whose
 Personal identity 
members are subject to the process. One will expect then that anyone
making judgments of identity will have to put himself constantly into the
frame of mind of a common law barrister, as if testing his every thought
against previous rulings or precedents. But as an expectation relating to
the world that we know, doesn’t this seem extraordinarily implausible?
If there is a mystery here, I think it will disappear with the reflection
that the individuative norms by which we normally proceed so effort-
lessly (but which it seems we have such difficulty in rendering articulate
to ourselves) are incorporated already, en masse, within the grasp of
thing-kind conceptions by which we find our way about the world. These
conceptions are effective – in Leibniz’s sense clear. They are not neces-
sarily explicit or distinct. By reflection and practice we can make them
more distinct (cf. Chapter Three, footnote ). In so far as we have mas-
tered these conceptions at all, however, they not only enforce for us the
Only a and b rule that is disregarded in the proposal that Brownson Sole
be ruled identical with Brown. They mark for us the distinction between
singulars/particulars, which are not things instantiated, and universals
which are. (Cf. Chapter Two, D(vi), Chapter Three, §.) Normally we
do not need to think about what we do. When we do need to think, we
may have to struggle. (Philosophy is no sure ally here. Rather the reverse
perhaps.) But none of this need count against the principle recently
identified, the principle stated at the end of § and resumed in the par-
agraph before last.
In the light of that principle, the time has come to look once more at
Shoemaker’s version of the Brown–Brownson case. In pursuit of the
enthusiasms of yesteryear, we allowed that case. Then we disallowed all
the cases that followed on after it. But as soon as I announced (§) that I
would mark such a frontier, many philosophers will have wondered how
much difference I could insist that there is between the process by which
a whole brain is transplanted into a body ready and suitable to receive
it, that by which one half of a brain is transplanted into a body ready
and suitable to receive it and the process by which the second act is done
twice over. If the second and third cannot yield identity, can the first?
Taken neat, this question is hard to answer, or even to see straight. But
fending off the thought that questions of this sort ought to be reduced
without residue to questions about the individuation and differentiation
of surgical processes (questions for which I have just allowed we are ill
prepared), why do we not remind ourselves yet again of our method and
say, ‘Well, here as everywhere it all depends on what sort of thing we take
ourselves to be’? If we take that question more fully seriously, maybe we
 Personal identity
can recover a wiser norm of judgment than any that we exercised in
§§–. So let us go back to that point and everything that led up to it.
At the beginning we organized our inquiry around the sortal concept
human being and bracketed the concept person. But the aim was not to let
go of the idea of personhood, still less to place the question of what we
are under the alien direction of physiologists, biologists, evolutionists or
others who are expert in matters relating to organisms. The aim was only
to fight free of the philosophical conceptions criticized in § following,
then to force upon ourselves and other persons the question of what, in
thinking of ourselves as human beings and human persons, we have
undertaken to think of ourselves as being. No doubt, the answer must be
discursive and draw on our inexplicit knowledge. No doubt it will be
essentially contestable too. But what should I say?
The answer I give begins from yet another misappropriation of
Locke’s thoughts. It proposes that we apply equally to the concepts
human person and human being that which Locke said of self and of person:
Wherever a man finds what he calls himself, there I think another may say is the
same person. It is a Forensick Term appropriating Actions and their Merit and
so belongs only to intelligent Agents capable of a Law, and Happiness and
Misery. This personality extends it self beyond present Existence to what is past
only by consciousness, whereby it becomes concerned and accountable, owns
and imputes to it self past Actions, just upon the same ground, and for the same
reason, that it does the present. (Essay , xxvii, §, punctuation modernized.)
If a term is truly forensic, it surely needs a palpable and public use – a
use that is intelligible in the forum. But then the term’s use in connexion
with questions of praise and blame needs to be all of a piece with its
other uses in the interpersonal sphere – and not least with all the uses to
which P. F. Strawson drew attention when he showed how intimately our
ideas of agency and responsibility depend on human beings’ reactive
and participative attitudes towards other human beings.41 Because such
attitudes depend on the reality of human presence, that directs us to find
out the connexion between these questions and Strawson’s earlier insis-
tence on the indispensability to the metaphysics and epistemology of
personhood of the idea that a person is the bearer of both P- and M-
predicates. How a human being stands or walks or frowns or smiles or
laughs or sulks or earnestly entreats, or how he fries an egg, this is one
part of what he is – no less than is the sort of thing (‘more mentally’) he
chooses to say to this or that conversant, how musically he plays the
41
See ‘Freedom and Resentment’, Proceedings of the British Academy,  (), pp. –.
 Personal identity 
violin (if he does), or how calmly or wildly, how sensibly or recklessly or
obsessively, how magnanimously or ignobly (in a manner he is held
responsible for) he responds day by day to the predicaments of ordinary
life. Human beings’ dispositions and capacities are gradually but con-
stantly shaped and reshaped. In the process, the particular ways in which
they come in due course to do whatever they do will become distinctive
of what they have then become and can be responded to as having
become.42 If so, then P-predicates, properly understood, are a subset of
M-predicates. Unless they are, unless physiognomy and the particular-
ities of a person’s physical presence and being find their way into the
story, Strawson’s account of reactive and participative attitudes is hard
to follow through. But is enough room left for all these things by the
acceptance of Shoemaker’s thought experiment? Surely the character of
a person is not independent of his or her physiognomy, and this physiog-
nomy can scarcely be independent of the body.
Consider the same matter from another point of view, which is com-
plementary to the foregoing. In the Brown–Brownson case, after all the
events that Shoemaker describes, it seems that Robinson still walks. But
Robinson is no more (or so we were prompted to decide). We must get used
to this. It seemed that Brown was dead, but that isn’t right either (or so
it has been decided). Where Robinson seems to be, there is Brown. We
must inhibit then the affective responses we once trained upon
Robinson, and redirect our affective responses for Brown onto the
person who looks like Robinson but is Brownson. We must look into the
face of Robinson and try to see there, try to find there, Brown.
Meanwhile Brown himself, wearing Robinson’s face, must come to terms
with the difficulty that we have in finding him there. These are among
the things to be made sense of in the Brown–Brownson experiment. Can
they really be made sense of ?
In considering the characteristically human modes of activity that
appeared to be preserved in the Shoemaker case and not preserved in
the cases that go beyond it, there was much to learn about the cognitive
42
As one speaks of these things, the phrase that may come to one’s lips is ‘his doing such and such
an act in this or that particular way is a part of his identity’. When we speak at t, each of us, of
‘our identity’, what are we concerned with? Surely we are concerned with the fullest determina-
tion as of t of that which we are or have by the time t become. But this involves us with the so far
shaped principle of activity which permits certain judgments of identity over certain imaginable
transformations but disallows other judgments of identity over all sorts of other apparently imag-
inable possibilities. These other possibilities are thought of, however contentiously and by what-
ever admixture of bluff, as bringing with them the end or virtual end of the subject who
undergoes them (if not necessarily of the subject’s memory traces or the traces of those memory
traces).
 Personal identity
faculties and the conditions of their operation. But on the properly
forensic view of human being-hood, why did not considerations of
physiognomy (etc) that are familiar to us prompt, at the outset, all sorts
of other doubts or hesitations about the Brown–Brownson case? How
can the present author have been persuaded by the contingencies of
current controversy to devote six long sections of this chapter to C, *C
and R, but have allowed so fundamental an objection against most or all
of these thought-experiments to lie fallow? It was strange, just because
Brownson seemed in some ways to perpetuate Brown, for us to ignore
the developed faculties of Brown which Brownson does not inherit and
pay so little attention to all the implications of brain transfers of the kind
that were described by Shoemaker. The high quality of the actors and
mimics one sees on the stage should not lead one to think that the ques-
tion of the fit of the brain to the physiognomy of the new body which is
to receive it is as relatively simple as the transposition of music from one
instrument to another.43 But now that the Brown–Brownson Sole case
and the potentially multiple output of the process by which Brownson
Sole emerges as Brown alert us at last to further doubts, let us take them
more seriously.

.      ‒                   

.   ,            
At this point, the position must be considered of a philosopher who was
not at all convinced by that which I said about the case of Brown and
Brownson Sole (§) but is suddenly persuaded that there is a serious case
to answer concerning physiognomy and the Brown–Brownson case. The
difficulty posed by physiognomy impresses him. Nevertheless, after
further reflection, we may imagine that this philosopher insists that even
the point that he has allowed casts no doubt on a case where Brown’s
brain is transferred to the body of a Robinson who is an identical twin of Brown.
We may suppose that the philosopher also insists, with a fervour charac-
teristic of present-day controversy (if not yet of science fiction New
Comedy), that the point he accepts about physiognomy casts no doubt
at all on a complicated situation where Brownson Sole emerges from the
transplant of half of Brown’s brain to the body of a Robinson who was
one of four quadruplets. Let the first be Brown himself, a long lost son
with a name given by foster parents; let the second be the Robinson

43
Cf. Williams (–).
 Personal identity 
whose body is animated by half of Brown’s brain; let the third be a
Robinson in whom it happens that the other half of Brown’s brain dies;
let the fourth be long since dead.
In the first of these two cases, as newly conceived, the body of
Brownson Sole is for all relevant purposes indistinguishable (my adver-
sary will say) from the body of Brown. In the second case too, anything
Brown can do with his body Brownson can do with his, etc. ‘As for the
process by which Brownson Sole comes to be from Brown [this philoso-
pher may say], let that be characterized as the process by which a person
x is perpetuated in matter qualitatively indistinguishable from x’s and as the sole inher-
itor of x’s particular faculties and physiognomy. If such a thing were demanded
[he will add], then a guarantee could certainly accompany the process
just characterized. For, as thus characterized, this process can be guaranteed
not to create multiple candidates for identity with x. It can also be guar-
anteed not to collide with any of the cognitive considerations adduced
in the endless chapter on this subject by David Wiggins. It is all very well
for the purist of identity to claim that he is coming to terms with the
forensic marks of our concept of human being. It is time for the purist
to come to terms with the forensic character of our disputations about
identity. Provided that we slightly redescribe the Brown–Brownson case
and all its successors, these cases can still show everything they were
intended to show.’
My preliminary reply to this philosopher will be to voice my doubt
whether making Brownson’s face very like Brown’s face can fully over-
come the disquiet that attaches to the very idea of ‘wearing’ a face. Off
stage, one does not wear a face, only an expression of the face.
My second point of reply is that the guarantee my opponent proffers
for Brownson Sole is entirely empty if the guarantee now given for the
said process’s not delivering multiple candidates only consists in the fact
that, if it did do so, then we shouldn’t call the process in question ‘the
process by which a person x is perpetuated in matter qualitatively indis-
tinguishable from x’s own and as the sole inheritor of x’s . . . particular
faculties and physiognomy’. Within our given ontology of processes,
with going standards (however local, inscrutable or challenging to
theory) of identity and difference, a separation of the kind that my oppo-
nent is attempting simply cannot be achieved by putting a special label
upon one part of one variety of one of them. (Compare Aristotle,
Metaphysics, , a  ff.) New nomenclature cannot undo the obvious
truth that, by the processes he is deploying and the means he is using to
perpetuate Brown in Brownson, a surgeon skilled in such techniques
 Personal identity
(namely transplanting brains or half-brains into bodies matching exactly
the bodies they come from or else, for this involves the same things, into
the bodies within which they will be viable) could produce multiples, and
might indeed produce all sorts of further outcomes for which we are
even worse prepared.44
The third point of reply relates to the idea of a guarantee. This is an
idea my opponent wants to cut down to size, to mock and belittle. Let us
strive harder to make sense of it. A genuine guarantee relating to this or
that process must relate to the nature of the process itself rather than a
mere description of it. Moreover, genuine guarantees exist. However
you describe it, the process of jam-making can be guaranteed not to
produce heavy water out of ordinary water. The now standard process
by which a stone is removed from the gall-bladder or is broken up there
can be guaranteed not to remove the appendix. Another process that
comes with a certain guarantee is the natural process, sustained by the
operation of numerous laws of biochemistry, physiology and the rest, by
which a human being comes into existence, matures and eventually
ceases to be, by ‘natural death’. That process is not of course guaranteed
to save a human being from murder or from premature death by asbes-
tosis, say, or irradiation. But it is certainly guaranteed not to produce
multiples, not to transplant brains or half-brains, and not (if that were
the better way to think of Brownson) to furnish new bodies to living, con-
tinuing brains. That is what makes this familiar process and the princi-
ple of activity associated with it one part of the basis for the making of
judgments of identity. It is the lawful dependability of this process that
entitles one whose judgments are shaped by that principle of activity to
claim that his practice is answerable to the Only a and b rule. If the prac-
tice is answerable to that rule and it sees itself as reliant on dependable
processes of this sort, then the practice will properly differentiate judg-
ments of identity from judgments to the effect that the object b is the
44
‘When Parties in a State are violent, he offered a wonderful Contrivance to reconcile them. The
Method is this. You take a Hundred Leaders of each Party; you dispose them into Couples of
such whose Heads are nearest of a Size; then let two nice Operators saw off the Occiput of each
Couple at the same Time, in such a Manner that the Brain may be equally divided. Let the
Occiputs thus cut off be interchanged, applying each to the Head of his opposite Party-man. It
seems indeed to be a Work that requireth some Exactness; but the Professor assured us, that if it
were dextrously performed, the Cure would be infallible. For he argued thus; that the two half
Brains being left to debate the Matter between themselves within the Space of one Scull, would
soon come to a good Understanding, and produce that Moderation as well as Regularity of
Thinking, so much to be wished for in the Heads of those, who imagine they came into the World
only to watch and govern its Motion: And as to the Difference of Brains in Quantity or Quality,
among those who are Directors in Faction; the Doctor assured us from his own Knowledge, that
it was a perfect Trifle’ (Jonathan Swift, Gulliver’s Travels: Book , A Voyage to Laputa).
 Personal identity 
proper replacement/surrogate/proxy for object a, or of a (‘is one of the
a’s’). No doubt, as I have already allowed, the distinctiveness of identity
may in some connections be unimportant. But those cases are a poor
basis on which to engage in rhetoric about the ‘forensic character of our
actual disputations about identity’.
This threefold reply is likely to stir up at least two counter-objections.
The first will consist of a challenge to apply the positive account that I
have given of the human principle of activity to the case of the human
zygote. Normally, the zygote becomes the embryo, but sometimes it
divides and becomes twin embryos. What do I say about that? I am com-
mitted to react to this fact with a general ruling to the effect that the
human being dates from a time after the zygote finally splits or settles
down to develop in unitary fashion. The human being, the human
person, dates from the point whenceforth it is nomologically excluded
that its zygote will divide. I am happy, however, to be committed to say
that.45 An advantage of this decision, which entails that the fertilization
of the egg is one part of the assemblage and further preparation of the
materials from which the fetus develops, is that it has the effect of block-
ing the unwelcome suggestion that the fetus is the particular egg that the
sperm activated. (If we concurred in that unwelcome suggestion, the
fetus might even be accounted as just as old as the egg itself. What will
explain better than the ruling that I have proposed why nobody wants
to say that the fetus is as old as the egg itself ?)
The second objection to what I have claimed concerns the guarantees
on which the ordinary business of individuation normally relies. It may
be expected to arise from the fact that such guarantees fall far short of
any logical exclusion of the sort of events that figure in cases that reach
beyond Shoemaker’s example: ‘These guarantees are logically insuffi-
cient. But it would be foolish [I shall be told], on the part of those who
are so enamoured of their theory that they will put their shirt on guar-
antees of the kind that you mention, to seek to strengthen these guaran-
tees. For then these enthusiasts will find they have excluded the
45
I find that, years ago, intuitively and in its own way, by a route not wholly dissimilar from that
followed in the text, the Committee of Enquiry into Human Fertilization and Embryology ()
chaired by Baroness Warnock came to substantially the same conclusion. (See Lords Hansard,
 December .) On ‘become’ see p. , note .
If one is moved by moral arguments to adopt a certain conception of the person and one cor-
rectly deploys this conception upon some question of identity and difference, must the decision
about identity and difference then qualify automatically as a ‘moral decision’? I am uncertain,
if only because of uncertainty concerning the terms of the question. Leaving that uncertainty
on one side, let us not forget that seeing a moral reality can be the reminder of a metaphysical
reality.
 Personal identity
interventions of orthodox medicine and dentistry from ‘proper’ or
‘admissible’ application to human beings as such.’
To this I should reply that it is not my concern, either here or any-
where else, to represent judgments of strict identity as resting on justific-
atory premisses that culminate in the deduction of an identity or that
bring among their logical consequences the satisfaction of the D princi-
ples given in Chapters Two and Three. That is not how the strictness of
the relation of identity is to be marked or vindicated. We have long since
rejected the deductive conception of identity-judgment. See Chapter
Three, §. (See also Chapter Six, §, where I found myself reminding
the reader of the familiar fact that one can see directly, however fallibly,
that it is one’s brother not one’s uncle who has come to call on one.) The
thing which really matters, where someone makes a responsible (however
fallible) judgment of identity between the individual objects a and b, is
that the various things the person judging has regard for in arriving at it
should be the right things for a judgment of identity – that they be things
appropriate for a type of judgment which, by its nature, goes well beyond
the weaker claim that b instantiates a, that b inherits the role of a, or that
b is interchangeable or interfungible with a. One who makes the judge-
ment that a is b deploys an understanding which is all of a piece with the
business of subsuming object a under some kind that is associated with
a principle of activity (or way of behaving) that things of a’s kind exem-
plify. His understanding must be all of a piece with knowing how to keep
track of such things, with grasping things’ natures well enough to chron-
icle what they do or undergo, etc. (See Chapter Two and the various D
principles that are assembled and paraded there. See also the Preface,
last paragraph but six, for the description I see as consequential upon
conceiving thus of identity and individuation.) For any genuine object,
there is some right way of keeping track of it and this must track it as a
particular. Cf. Chapter Two, §. It is from this simple truism that the whole
business of making of judgments of identity extrapolates.
What remains then of the difficulty about splitting (etc)? The real
difficulty that the theory of individuation must have from cases of split-
ting, medical intervention, transplantation and the rest (or so I conclude)
is not that such occurrences breach the limited and specific guarantees
at which the objector is sneering (and which are all that are needed to
sustain the relevant way of thinking of something’s being securely
singled out), but that they add complications to the picture we have held out
of a substance coasting along (as it were) autonomously or under its own
steam. (Which does not mean that, even in this picture, it interacts with
 Personal identity 
nothing else.) In coming to terms with these complications, quasi-casu-
istically so to speak, without much generality but in ready appreciation
of the universality of the commitments that one creates by deciding in
one way rather than another, we need constantly to refer the issues that
arise back to our conception of the sort of thing that we are dealing with.
Minor interventions (medical, dental, orthopaedic, osteopathic . . .) in a
human being’s pattern of being and acting confront us with no concep-
tual difficulties at all. In acquiescing in these, we do not prejudice our
understanding of such a pattern. The substance’s organic independence
can still be conceived as undiminished. There is no problem of persis-
tence. (We do not commit ourselves to allow anything on the scale of the
Brown–Brownson case.) At the other extreme, however, in the so far
imaginary cases where it seems a human being is simply treated as a tem-
plate for the production of copies, it is manifest (it ought not to be con-
troversial) that ideas are changed almost out of recognition. That is what
I should say about Parfit’s teletransportation, a fictional process which
can as readily carbon-copy me twice (or thrice, or the number of times
it takes to make a regiment) as once. In so far as these cases are taken as
amounting to the perpetuation of the person Brown, we have lost hold
altogether of the notions we began with of what Brown is. The judg-
ment that the singular being Brown persists thereby is unsustainable.
Everything depends then on that which lies between the unproble-
matic cases and the cases that are out of the question. In these interme-
diate cases, where massive transplanting of organs or constant
interchange of parts is contemplated, as well as constant fine-tuning of
a human being by such expedients as gene therapy, the newly emerging
conception under which we subsume a human substance will still be the
conception of an individual thing or substance with a destiny of its own.
Nevertheless, as we proceed along the road indicated, the conception of
a human person will diverge further and further from that of a self-
moving, animate living being exercising its capacity to determine, within
a framework not of its own choosing and replete with meanings that are
larger than it is, its own direct and indirect ends. The conception will
converge more and more closely upon the conception of something like
an artefact – of something not so much to be encountered in the world
as putatively made or produced by us, something that it is really up to us
(individually or collectively) not merely to heal or care for or protect but
also to repair, to reshape, to reconstruct . . . even to reconceive.
This new or emerging conception of what a person is will perplex us
not only with philosophical variants on the problem of artefact identity
 Personal identity
but with practical questions. At the beginning, it may seem these ques-
tions will be easy enough. Later on, when less and less seems to be
excluded by the then prevalent conception of human beinghood, they
may bewilder us. I speculate that this bewilderment can only grow as
that conception is progressively adjusted to the thought that little or
nothing that a human being might more than idly wish needs to be out
of the question or excluded by our human limitations. Will not bewil-
derment then turn to total aporia as our conception of human person-
hood is adjusted to the further thought that it is not merely our destiny
that is (in large part) up to us, or our ethical identity, but even the kind
of thing it is that we are? In the here and now, at a point well short of
this limit, there has, of course, been a huge increment in the sum of
human well-being. Why deny that? But before extrapolating this gain
mindlessly into the open future, or simply rejoicing in the technological
freedom that geneticists and medical scientists have been encouraged to
create for us, there are questions to answer.
Here is one of them. If we cannot recognize our own given natures
and the natural world as setting any limit at all upon the desires that we
contemplate taking seriously; if we will not listen to the anticipations and
suspicions of the artefactual conception of human beings that sound in
half-forgotten moral denunciations of the impulse to see people or
human beings as things, as tools, as bearers of military numerals, as
cannon-fodder, or as fungibles; if we are not ready to scrutinize with any
hesitation or perplexity at all the conviction (as passionate as it is ground-
less, surely, for no larger conception is available that could validate it)
that everything in the world is in principle ours or there for the taking;
then what will befall us? Will a new disquiet assail our desires themselves,
in a world no less denuded of meaning by our sense of our own omni-
potence than ravaged by our self-righteous insatiability?46
I frame the question and, having formed it, I grave it here. But a book
such as this is not the place in which to enlarge upon it, to answer it, or
to speculate about the mental consequences of our available energies
being diverted from the gradual discovery and enhancement, within the
limits set by our animate nature, of more sustaining human ends – and
the further consequences of these energies being progressively redi-
rected to the endless elaboration of the means to ends that are less and
less often explored in thought or feeling. This book is not the place nor

46
As the book goes to press, I find a perceptive account of some related misgivings in Harry G.
Frankfurt, Necessity, Volition, and Love (Cambridge University Press, ), pp. ff.
 Personal identity 
is its author the philosopher or moralist to hold up to human beings the
image of the insatiability by which they relentlessly simplify the given
ecology of the earth and change out of all recognition the given frame-
work of human life. In a treatise on identity and individuation, it is
enough and more than enough to point to the conceptual tie that links
together these issues of life’s framework and meaning, of individuation,
and of the way in which we conceive of a human being.
In this work, the chief thing that needs to be made clear is that no
freedom that the theory of individuation leaves it open to us to exercise
in how we think of ourselves could ever liberate us from the constraints
laid down in Chapter One. Whatever it may amount to, our freedom to
remake the idea of personhood (to remake it by our pursuits and our
choices) could not entail that it would be up to us, once it was determi-
nate what a and b were, whether a certain identity judgment a ⫽b was
true or false. For it is impossible on the level of reference and truth-value
for there to be such freedom. As soon as it is determinate what a or b is,
the result of Chapter One, §, is inescapable. Moreover, once an iden-
tity question is construed in a definite way, it must be determinate what
the question turns on. Nothing is left to stipulate. Only at the level of
sense is there any room for conceptual invention to enter. At that level,
I do not deny, there is freedom to reformulate bit by bit the answer that
we commit ourselves to give to the what are we? question. It is a freedom
we cannot really escape. It is cognate with the conceptualist (residually
idealist) insight we expounded in Chapter Six and linked there with our
miscellaneous capacities to come to terms with the exigency of the iden-
tity relation. But we should think much harder how we exercise it. We
should think harder then about the choices that commit us to think of
personhood in one way rather than another. These questions are essen-
tially contestable. But from that it does not follow that they permit of
more than one answer.
Apart from the logic of individuation, what ought to constrain us in
how we think of human beinghood? Only the power of discursive reflec-
tion. How wise is it then to allow the goal of attaining the perfect state
of material and bodily well-being to fly constantly before us like the
rainbow or a will o’ the wisp? How wise is it for those who have almost
everything to project limitlessly into an open future their opportunities
to reconstruct and reconceive themselves in order to have yet more?
Where once angels feared to tread, how far will we go? If Schopenhauer
had conceived how elusive and difficult the question was going to be
found of what conception we should make for ourselves of ourselves; if
 Personal identity
he had conceived how difficult it would be found by the very substances
who exemplify the thing-kind whose nature has come into question; if
he had known what kind of commotion the question would occasion or
how sensitive the answer would be to the stability or instability of the
expectations that still prevail about the world in which we still hope to
find meaning and have our engagement; then I wonder whether he
would have wanted to persist in his optimistic declaration that ‘Just as
the boatman sits in his little boat . . . so in the midst of a world of suffer-
ing the individual man calmly sits, supported by and trusting the princip-
ium individuationis’ (World as Will and Idea, §).
 Select bibliography

For the basis of selection, see Preface, ad fin.

Anscombe, G. E. M., . ‘The Principle of Individuation’, Proceedings of the

Aristotelian Society, Supp. Vol. , –.
Aristotle, Categories, especially Chs. –.
Physics, Bk , Chs. –; Bk , Chs. –.
De Anima, Bk , Chs. –.
Metaphysics, e.g. Bk , Chs. , ; Bks , , .
Topics, aff. and a–.
De Generatione et Corruptione, bff.
Politics,  a–b.
Ayer, A. J., . Philosophical Essays (London: Macmillan), Ch. , ‘The Identity
of Indiscernibles’.
Ayers, Michael R., . ‘Individuals Without Sortals’, Canadian Journal of
Philosophy, , –.
. John Locke (London: Routledge), Vol. , Parts One and Three.
Barcan Marcus, Ruth, . ‘The Identity of Individuals in a Strict Functional
Calculus of First Order’, Journal of Symbolic Logic, , –.
Black, Max, . ‘The Identity of Indiscernibles’, Mind, , –.
Burge, Tyler, . ‘Mass Terms, Count Nouns and Change’, Synthèse, .
. ‘A Theory of Aggregates’, Nous, .
Butler, Joseph, . First Dissertation to the Analogy of Religion Natural and Revealed
to the Constitution of Nature.
Cartwright, Helen M., . ‘Heraclitus and the Bath Water’, Philosophical
Review, , –.
. ‘Quantities’, Philosophical Review, , –.
Cartwright, Richard, . ‘Identity and Substitutivity’, in Milton K. Munitz
(ed.), Identity and Individuation (New York University Press), –.
. ‘Indiscernibility Principles’, Midwest Studies in Philosophy, , –.
. ‘On the Logical Problem of the Trinity’, in Philosophical Essays
(Cambridge, Mass.: MIT).
Chappell, V. C., . ‘Sameness and Change’, Philosophical Review, ,
–.
Chisholm, R. M., . ‘Identity Through Time’, in Howard E. Kiefer and
Milton K. Munitz (eds.), Language, Belief, and Metaphysics (Albany: State
University of New York Press), –.

 Select bibliography
. Person and Object (London: Allen and Unwin), especially Ch. , Apps. A
and B.
Coburn, R., . ‘The Persistence of Bodies’, American Philosophical Quarterly, .
Doepke, Frederick, . ‘Spatially Coinciding Objects’, Ratio, , –.
Dummett, M. A. E., . Frege: Philosophy of Language (London: Duckworth),
especially pp. – and Ch. .
. ‘Does Quantification involve Identity?’, in Harry Lewis (ed.), Peter
Geach: A Profile (Dordrecht: Reidel).
Evans, Gareth, . ‘Identity and Predication’, Journal of Philosophy, .
. ‘Can there be Vague Objects’, Analysis, .
Flew, Anthony, . ‘Locke and the Problem of Personal Identity’, Philosophy, .
Fraassen, Bas C. van, –. ‘Essence and Existence’, American Philosophical
Quarterly Monograph.
Frede, Michael, . ‘Individuals in Aristotle’, in Essays in Ancient Philosophy
(Oxford University Press), –.
Frege, Gottlob, . The Foundations of Arithmetic, trans. J. L. Austin (Oxford:
Blackwell), sect. .
. ‘On Sense and Reference’ and ‘Concept and Object’, in P. T. Geach
and M. Black (ed. and trans.), Translations from the Philosophical Writings of
Gottlob Frege (Oxford: Blackwell).
. ‘Comments on Sense and Reference’, in Hermes, Kambartel and
Kaulbach (eds.), Posthumous Writings, trans. Peter Long and Roger White
(Oxford: Blackwell).
. ‘Letter to Husserl’, dated  May , in Gottfried, Hermes,
Kambartel, Thiel and Veraart (eds.), Philosophical and Mathematical
Correspondence, abridged for the English edition by Brian McGuiness, trans.
Hans Kaal (Oxford: Blackwell).
Geach, P. T., . Reference and Generality (Ithaca: Cornell University Press), espe-
cially sections –. In the third edition (), see especially p. .
. ‘Some Problems About Time’, Proceedings of the British Academy, .
. Logic Matters (Oxford: Blackwell), pp. –, –.
. ‘Ontological Relativity and Relative Identity’, in Milton K. Munitz
(ed.), Logic and Ontology (New York University Press).
Goodman, Nelson, . Problems and Projects (Indianapolis: Bobbs Merrill), Pt
. See also under Leonard, Henry S.
Griffin, Nicholas and Routley, R., . ‘Towards a logic of relative identity’,
Logique et Analyse.
Hacker, P. M. S., . ‘Substance: The Constitution of Reality’, Midwest Studies
in Philosophy, , –.
Hacking, Ian, . ‘Individual Substances’, in Harry G. Frankfurt (ed.), Leibniz:
A Collection of Critical Essays (New York: Doubleday).
. ‘All Kinds of Possibility’, Philosophical Review, .
Hampshire, Stuart, . ‘Identification and Existence’, in H. D. Lewis (ed.),
Contemporary British Philosophy (London: Allen and Unwin), rd series, pp.
–.
 Select bibliography 
Heraclitus, . Diels fragments B, B, B, B, B, in H. Diels and W.
Kranz, Die Fragmente der Vorsokratiker (Berlin).
Hobbes, Thomas, De Corpore, Pt , Ch. , in W. Molesworth (ed.), The English
Works of Thomas Hobbes (London: John Bohn, –), Vol. .
Ishiguro, H., a. ‘Leibniz and the Idea of Sensible Qualities’, in S. Brown
and G. Vesey (eds.), Reason and Reality (London: Macmillan).
b. Leibniz’s Philosophy of Logic and Language (London: Duckworth, and nd
edn Cambridge University Press, ), especially Chs. , .
Johnston, Mark, . ‘Human Beings’, Journal of Philosophy, .
Kripke, S., . ‘Identity and Necessity’, in Milton K. Munitz (ed.), Identity and
Individuation (New York University Press).
. ‘Naming and Necessity’, in G. Harman and D. Davidson (eds.),
Semantics of Natural Languages (Dordrecht: Reidel).
. Naming and Necessity, revised text (Oxford: Blackwell).
Leibniz, G. W., Nouveaux Essais sur l’Entendement Humain, especially Bk , Chs. ,
; Bk , Chs. , , , translated as New Essays on Human Understanding by
Peter Remnant and Jonathan Bennett (Cambridge University Press, ).
Discourse of Metaphysics, Gerhardt , pp. –. See next entry.
Die Philosophischen Schriften von G. W. Leibniz, ed. C. I. Gerhardt, Berlin,
–.
Correspondence with Clarke, Gerhardt II, Leibniz’s fifth paper, sections –.
‘Meditationes de Cognitione, Veritate et Ideis’, Gerhardt , pp. –.
Correspondence with De Volder, Gerhardt II. See L. E. Loemker (ed.), Philosophical
Papers and Letters (Dordrecht: Reidel, ), pp. –.
Leonard, Henry S., and Goodman, Nelson, . ‘The Calculus of Individuals
and its Uses’, Journal of Symbolic Logic, , –.
Lewis, David, . ‘Survival and Identity’, in A. Rorty (ed.), The Identities of
Persons (Berkeley: University of California Press).
Locke, John, Essay Concerning Human Understanding. , xxvii.
Reply to the Bishop of Worcester’s Answer to his letter, especially ad fin.
Lovibond, Sabina and Williams, S. G., . Identity, Truth and Value: Essays for
David Wiggins, Aristotelian Society Series, Vol.  (Oxford: Blackwell).
Lowe, E. J., . ‘Objects and Criteria of Identity’, in Bob Hale and Crispin
Wright (eds.), A Companion to the Philosophy of Language (Oxford: Blackwell).
MacIntyre, Alasdair, . ‘Essence and Existence’, in Paul Edwards (ed.),
Encyclopaedia of Philosophy (New York: Macmillan), Vol. , pp. –.
Mackie, Penelope, . ‘Sortal Concepts and Essential Properties’, Philosophical
Quarterly, , –.
. ‘Identity, Time and Necessity’, Proceedings of the Aristotelian Society, .
McDowell, John, . ‘On the Sense and Reference of a Proper Name’, Mind.
. ‘Reductionism and the First Person’, in J. Dancy (ed.), Reading Parfit
(Oxford: Blackwell).
Miri, Mrinal, . ‘Memory and Personal Identity’, Mind, .
Noonan, H., . ‘Wiggins on Identity’, Mind, .
. ‘Sortal Concepts and Identity’, Mind, .
 Select bibliography
. ‘The Only X and Y Principle’, Analysis, .
. ‘Absolute and Relative Identity’, in Lovibond and Williams (eds.).
Parfit, Derek, . ‘Personal Identity’, Philosophical Review, , –.
. Reasons and Persons (Oxford University Press), Pt Three.
Pellettier, J. F., . ‘A Bibliography of Recent Work on Mass Terms’, Synthèse,
.
Perry, John R., . ‘The Same F’, Philosophical Review, , –.
Plutarch, ‘On Common Conceptions’ A–A, quoted with other mate-
rials in Long and Sedley (eds.), The Hellenistic Philosophers (Cambridge,
).
Popper, K. R., . ‘The Principle of Individuation’, Proceedings of the Aristotelian
Society, Suppl. Vol. .
Pratt, Vernon, . ‘Biological Classification’, British Journal for the Philosophy of
Science, , ‒.
Prior, A., . ‘Identifiable Individuals’, Review of Metaphysics, .
. ‘Time, Existence and Identity’, in Papers on Time and Tense (Oxford:
Clarendon Press), pp. –.
. ‘Things and Stuff’, in Papers in Logic and Ethics (London: Duckworth), pp.
–.
Putnam, Hilary, . ‘Is Semantics Possible?’, Metaphilosophy, ; also in Mind,
Language and Reality, Philosophical Papers, Vol.  (Cambridge University
Press, ).
Quine, W. V. O., . From a Logical Point of View (Cambridge, Mass.: Harvard
University Press), especially pp. – (‘Identity, Ostension and
Hypostasis’) and pp. – (‘Reference and Modality’).
. Word and Object (Cambridge, Mass.: MIT Press).
. Set Theory and its Logic (Cambridge, Mass.: Harvard University Press),
Chapter One, §.
. Review of Geach, P. T., Reference and Generality, in Philosophical Review, ,
–.
. ‘Three Grades of Modal Involvement’, in The Ways of Paradox and Other
Essays (Cambridge, Mass.: Harvard University Press).
. ‘Natural Kinds’, in N. Rescher (ed.), Essays in Honour of C. G. Hempel
(Dordrecht: Reidel).
. Review of Milton K. Munitz (ed.), Identity and Individuation, in Journal of
Philosophy, , –.
. ‘Worlds Away’, Journal of Philosophy, .
. ‘Intensions Revisited’, Midwest Studies in Philosophy, II.
Quinton, A., . The Nature of Things (London: Routledge and Kegan Paul),
Pt .
Rea, Michael . (ed.), Material Constitution: A Reader (Lowham, Md.: Rowman
and Littlefield).
Reid, Thomas, Essays on the Intellectual Powers of Man (Edinburgh: John Bell, ),
Essay , especially Chs.  and .
Salmon, N., . Reference and Essence (Princeton University Press), pp. –.
 Select bibliography 
Shoemaker, Sydney, . Self-Knowledge and Self-Identity (Ithaca: Cornell
University Press).
. ‘Persons and Their Pasts’, American Philosophical Quarterly, , –.
. ‘Wiggins on Identity’, Philosophical Review, .
. ‘Identity, Properties and Causality’, in Identity, Cause and Mind
(Cambridge University Press), –.
Shwayder, David, . ‘Man and Mechanism’, Australasian Journal of Philosophy,
.
Simons, Peter, . Parts: A Study in Ontology (Oxford University Press).
Sloman, Aaron, . ‘Functions and Rogators’, in J. N. Crossley and M. A. E.
Dummett (eds.), Formal Systems and Recursive Functions (Amsterdam: North
Holland Publishing Co.), pp. –.
Smullyan, A. F., . ‘Modality and Description’, Journal of Symbolic Logic, ,
–.
Snowdon, Paul, . ‘Persons and Personal Identity’, in Lovibond and
Williams (eds.).
Stevenson, Leslie, . ‘Relative Identity and Leibniz’s Law’, Philosophical
Quarterly, , –.
. ‘A Formal Theory of Sortal Quantification’, Notre Dame Journal of Formal
Logic, , –.
Strawson, P. F., . Individuals (London: Methuen), especially Chs. Three and
Four.
. ‘Chisholm on Identity Through Time’, in H. E. Kiefer and Milton K.
Munitz (eds.), Language, Belief, and Metaphysics (Albany: State University of
New York Press), –.
. ‘Entity and Identity’, in H. D. Lewis (ed.), British Contemporary Philosophy
(London: Allen and Unwin), Vol. .
. ‘May Bes and Might Have Beens’, in A. Margalit (ed.), Meaning and Use
(Dordrecht: Reidel).
Tarski, Alfred, . ‘Foundation of the Geometry of Solids’, in Logic, Semantics,
and Metamathematics (Oxford University Press), pp. –. See also under
Woodger, J. H.
Wallace, John R., . Philosophical Grammar, Stanford University Ph.D., pp.
–. Published by University Microfilms Ltd, Ann Arbor, Michigan,
.
Wiggins, David, . Identity and Spatio-Temporal Continuity (Oxford:
Blackwell).
. ‘On Being in the Same Place at the Same Time’, Philosophical Review,
, –.
. ‘Identity, Continuity and Essentialism’, Synthèse, .
. ‘Identity, Designation, Essentialism and Physicalism’, Philosophia, ,
–.
. ‘Are the Criteria of Identity that hold for a Work of Art in the Different
Arts Aesthetically Relevant? Reply to Richard Wollheim’, Ratio, , –.
. ‘The Concern to Survive’, Midwest Studies in Philosophy, , –.
 Select bibliography
. ‘Heraclitus’ Conceptions of Flux, Fire, and Material Persistence’, in M.
Nussbaum and M. Schofield (eds.) Language and Logos: Essay for G. E. L. Owen
(Cambridge University Press).
. ‘The Sense and Reference of Predicates: A Running Repair to Frege’s
Doctrine and Plea for the Copula’, Philosophical Quarterly, , –.
a. ‘“The Concept of the Subject contains the Concept of the Predicate”:
Leibniz on Reason, Truth and Contingency’, in J. J. Thompson (ed.), On
Being and Saying: Essays for Richard Cartwright (Cambridge, Mass.: MIT Press);
reprinted in Roger Woolhouse (ed.), Leibniz: Critical Assessments (London:
Routledge, ), Vol. , pp. –.
b. ‘The Person as Object of Science, as Subject of Experience, and as
Locus of Value’, in A. Peacocke and G. Gillett (eds.), Persons and Personality
(Oxford: Blackwell).
. ‘Remembering Directly’, in J. Hopkins and A. Savile (eds.),
Psychoanalysis, Mind and Art: Essays for Richard Wollheim (Oxford: Blackwell).
. ‘Putnam’s Doctrine of Natural Kind Words and Frege’s Doctrines of
Sense, Reference and Extension. Can they Cohere?’, in A. W. Moore (ed.),
Meaning and Reference (Oxford University Press)
. ‘Substance’, in A. C. Grayling (ed.), Philosophy: A Guide through the Subject
(Oxford University Press).
a. ‘Sortal Concepts: a Reply to Fei Xu’, Mind and Language, , –.
b. ‘Natural Languages as Social Objects’, Philosophy, , –.
Williams, B. A. O., –. ‘Personal Identity and Individuation’, Proceedings of
the Aristotelian Society, .
. ‘Bodily Continuity and Personal Identity’, Analysis, .
. ‘The Self and the Future’, Philosophical Review, , –.
. Problems of the Self (Cambridge University Press), papers –.
Williamson, Timothy, . Vagueness (London: Routledge).
. ‘The Necessity and Determinacy of Distinctness’, in Lovibond and
Williams (eds.).
Woodger, J. H., . The Axiomatic Method in Biology (Cambridge University
Press), Ch. , p. , and Appendix E (by Tarski), p. .
Woods, M. J., . ‘Substance and Quality’ (B.Phil., Oxford).
. ‘The Individuation of Things and Places’, Proceedings of the Aristotelian
Society, Supp. Vol. , pp. –.
. ‘Identity and Individuation’, in R. J. Butler (ed.), Analytical Philosophy
(Oxford: Blackwell), nd series, pp. –.
 Index of names of persons cited or mentioned
See also the Select bibliography

Anaxagoras, 32 Ducharme, Dewey, 197

Anscombe, G. E. M., 49 Dummett, Michael, 9, 95
Antithenes, 36
Aristotle, 5, 9, 11, 12, 21, 27, 37, 49, 59, 65, Evans, Gareth, 115, 161, 162, 220
67, 77, 80–1, 86, 89, 112, 123, 131, 141,
209, 210, 237 Fine, Kit, 118-19
Arnauld, Antoine, 97, 108, 109, 120 Fisk, M., 153
Austin, J. L., 12 Flew, Anthony, 204, 214
Ayer, A. J., 109, 110, 111, 120, 204 Frankfurt, Harry, 242
Ayers, Michael, 140, 149, 150–1, 161 Frassen, Bas Van, 140, 144–7
Frege, G., 9, 27, 28, 38, 51, 52, 53, 60, 75,
Bakhurst, David, 195 79–80, 84, 94
Barcan Marcus, Ruth, 96
Barnes, Jonathan, 93 Gabbay, D., 178–80
Benacerraf, Paul, 51 Geach, P. T., xi, 15, 16, 23, 32, 34, 35, 112,
Bennett, Jonathan, 81 173–4, 179
Bernays, P., 184 Goodman, Nelson, 38, 118, 119, 125, 141,
Black, Max, 63 172
Bolton, R., 12 Gotthelf, Alan, 143
Bradley, M. C., 150 Grelling, Kurt, 213
Broad, C. D., 31 Grice, H. P., 215
Broome, John, 161 Griffin, Nicholas, 23, 46, 67
Burge, Tyler, 37, 112
Hacking, Ian, 72, 112, 121
Carnap, Rudolf, 176 Hamlyn, D. W., xi
Cartmell, Jo, x–xi Hampshire, Stuart, 195
Cartwright, Helen Morris, 100 Hare, R. M., xv
Cartwright, Richard, xi, 27, 50, 112, Hart, H. L. A., 143
119 Heck, Richard G., 162
Cassam, Quassim, 196 Heraclitus, 32, 55, 77, 82, 107, 193
Chapman, T., 23 Hilbert, D., 184
Chappell, Vere, 179 Hintikka, Jaako, 183
Cheesman, D. F., xi Hirsch, Eli, 151
Child, William, x Hobbes, Thomas, 35, 37, 92–102, 150, 161,
Chomsky, Noam, 67 170, 208
Coburn, Robert, 32, 100 Hodges, W. A., xi, 46
Cratylus, 32 Hornsby, Jennifer, xi, 218
Cresswell, Max, 112 Hughes, G. E., 112
Hume, D., 31
Dauer, Francis, 94 Husserl, E., 9, 27, 139
Davidson, Donald, 31, 198 Hussey, E. L., xi
Davies, M. K., xi
Doepke, F. C., 182 Ishiguro, Hidé, xi, 79, 85, 128–9


 Index of names of persons cited or mentioned
Johnson, W. E., 8 Rumfitt, Ian, x
Johnston, Mark, 31, 178 Russel, B. A. W., 10, 16, 32, 126, 147, 213
Justinian, 92, 189ff.
Sainsbury, R. M., 166
Kamp, J. A. W., xi Salmon, Nathan, 99, 166
Kant, I., 5, 10, 158 Schoenheimer, Rudolf, 55
Kirwan, C., 112 Shopenhauer, Arthur, 243–4
Kolakowski, Leszek, 148–50 Schroedinger, E., 124
Kraut, R., 23 Sedley, David, 32, 174
Kripke, Saul, xi, 12–14, 79, 80, 116, 118, 122, Sharvey, R., 119
130–3, 134 Shoemaker, Sydney, 206–7, 208, 212
Kupperman, Joel, 32 Shwayder, David, 12
Simons, Peter, xi, 38–9, 88, 166, 178
Leibniz, G. W., 1, 12, 13, 57, 61–3, 72, 77, 78, Sloman, Aaron, 10
81, 84–6, 92, 97, 100, 107, 109, 123, Smart, Brian, 94
128–9, 143, 153, 154, 187 Smullyan, A., 13, 164
Leo, Pope, 48 Snowdon, Paul, x, xiv, 228
Lesniewski, S., 38–40 Sober, E., 10
Lewis, David, 31, 177, 178 Stalnaker, R., 112
Linsky, Leonard, 34, 41 Stead, G. C., 212
Loar, Brian, 111 Stevens, Wallace, 157
Locke, John, xiii, xiv, 9, 12, 35, 55, 87, Stevenson, Leslie, 26
197–203, 204, 205, 207, 234 Strawson, P. F., xi, 9, 124, 187–8, 195, 234–5
Lovibond, Sabina, x Suppes, Patrick, 119, 176, 196
Swift, Jonathan, 238
McDowell, John, xi, 11, 79, 211
MacIntyre, A., 123 Tarski, A., 94, 156
Mackie, Penelope, xiii, 121, 132–3 Tennant, N., xi
Mehmet, Naci, x, 95ff. Thomason, Richmond, 67, 112, 119, 164
Mellor, D. H., 79, 182
Valéry, Paul, 219
Miri, Mrinal, 208
Velleman, David, 200
Moravcsik, J., 178–80
Vygotsky, Lev, 195
Nietzsche, F., 81 Wallace, John, 185
Noonan, Harold, x, 4, 23, 27, 37, 40, 71, 99, Warnock, Mary, 239
174 Weil, Simone, 139
Nozick, Robert, 96, 175 White, A. R., 121
Nussbaum, Martha, xi William of Sherwood, 173
Williams, Bernard, xi, xiv, 96ff., 199, 204, 236
Odegard, D., 23 Williams, S. G., x, 46
Ovid, 66 Williamson, Timothy, x, xii, 102, 114ff., 161,
163, 165, 185
Parfit, Derek, 161, 205, 208ff., 222ff., 241 Winch, Peter, 198
Peacocke, Christopher, x, xi, 14, 112 Wittgenstein, L., 5, 6, 8, 63, 127, 187, 193,
Peirce, C. S., 153 218
Plato, 6, 53, 67, 88, 93, 229 Wollheim, R. A., xi
Pratt, Vernon, 79 Woodger, J. H., 38, 73, 156, 181
Prior, A. N., 8, 24, 73 Woods, John, 112
Putman, Hilary, xi, 51, 78–9, 80, 88, 198 Woods, M. J., 9, 12, 71
Quine, W. V., 9, 12, 15–18, 25, 27, 79, 103, Xenophon, 93
111, 112, 118, 141, 184–6 Xu, Fei, 151
Quinton, A. M., 150
Young, J. Z., xi
Reid, Thomas, 88, 157, 197, 203, 205
Righter, Anne, 157 Zemach, E., 23
 Index of content (themes, theses, examples, etc)
See also Table of contents and Preface pp. xii–xvii

absoluteness of identity, see D, determinacy, brain, doubleness of, 208–9, 238

exigency of identity, incompleteness see also Brownson
(supposed) of bare identity Brownson, Brownsons, 206, 207ff., 237ff.
accessio, see Roman Law Brownson Sole, 231ff.
activity, principle or mode of, xi, 72, 77, 80,
80–3, 100, 194, 207, 209, 211, 232, C, the counting thesis, 24, 74–76; see D (ix?)
235ff., 240 C, C*, see coconsciousness relation
aggregate, see mereology camera (metaphor), 172
almost, 192 causal theory of perception and certain
amoeba (example), 72–3, 83–4 analogies with, 215–16, 222–5
analyticity, 14, 88, 110, 127, 140, 218 change versus replacement, 65–6
ancestral, 201 Chesil Bank (example), 90
anchor constraint, 109, 121, 121–3, 131 Childe Harold I–II, the author of, namely
anti-realist conceptualism and anti- Byron (example), 30, 54
conceptualist realism: clear non-distinct, ideas (Leibnizian), 12, 80–5,
the positions explained 140, 142 233
the positions anticipated, 105–6 Cleopatra’s Needle (example), 34, 41
the positions criticised and rejected in clocks and other timepieces (examples), 87, 89,
favour of conceptualist realism, Chapter 91, 101
5 clone, 211, 212
a priori / a posteriori, 12–14, 110 club (example), 161, 169
Aristotles’ sense of “is”; see what is it? coconsciousness relation, 200ff., 212
question/category connection with R, 205
artefact-identity, 91ff., 104, 169ff. coincide under a sortal concept, 5, 56, 60, 71,
artefacts, 86ff., 149, 154 104
distinguished from natural things by a conceiving, see imagining, necessary/
fundamental distinction, 89–90 contingent (in relation to conceivability)
artefacts and their matter, 100, 133ff. concept, 9–10, 44, 79
artefactual conception of personhood, xiv, conceptualism and conceptualist realism, xii,
241ff. 105–6, 159, 173, Chapter Six, 192; see D
attributive adjectives, 49–50 absolutism
axolotl (example), 86 ‘conceptualize’, 105, 106; see construe versus
construct, single out
back and forth, see reciprocity concept versus conception, 10–11, 79
bare absolutist, 105 concordant principles, 151
become, see constitutive ‘is’ congruence and community of properties, 18,
Berry’s paradox, 213 25, 56–61, 70–1, 103–4, 184
‘best candidate’ (or ‘closest continuer’) theories see exigency of identity, judgments of
of identity, 95ff. identity
biology, philosophy of, xi–xii, 64, 71–3, 85–6 connectedness, mental, xiv
boundary drawing, see single out, sortal see coconsciousness, R
concepts conservation principles, 67–8


 Index of content
constitutive ‘is’ 32, 36, 37, 39–43, 51–2, 71, edges compared and contrasted with lines in
73, 180 nature, 6 (quoting Plato, Phaedrus), 159
construe versus construct, 151, 171–6 eels and elvers (example), 88
see conceptualism, idealism elucidations (contrast analyses), 2, 5, 59–60
contestability, 66, 85, 130, 192, 243 equilibrium of a thing that is not alive with its
continuants, three dimensional, persisting surroundings, 86, 90
whole through time, 2, 3, 5, 31–3, 59ff.; essentialism, individuative, xii, 12, 13, Chapter
see synchronic/diachronic contrasts 4, 127–9, 140–4, 147–51
continuity, xii, 57 events versus continuants, 31
continuum question, 167 exigency of identity, xv, 4, 5, 24–8, 53–4, 102,
conventionalism, 88 159, 170–1
copse (example), 52–3 existence, 67, 68–9
Cornish Riviera Express (example), 35, 44 extension–involving, see deixis
corporate beings, 229ff.
covering concept, 24 face (of a person), 235–6, 237
criterion/principle of identity/individuation, fission, see splitting
xiii, 41, 53, 60–1, 69, 72-3, 77ff., 244 forensic conception, xiv, 197, 234ff., 237
see also activity see also human being conception, persons
cross-classification, 67 formal concept, 6
foundationalism in theory of knowledge,
D, xii, 18, 22, 23, 24, 55–61, 96, 103, 104, 220–1
140–1, 159, 193 flux, 3
D(i) principle, 64, 96 functioning, principle of, 72, 86ff., 91ff.
D(ii) principle, 64–9, 96, 123 Fundierungsaxiom, 218, 230
D(iii) principle, 70, 77, 84, 91, 92, 96 fusion 38, 94
D(iv) principle, 70, 84, 96 see also mereology
D(v), 72, 77, 84, 91, 96
D(vi), 72 gene pool, 80, 123
D(vii), 74 general/specific, the distinction contrasted
D(viii), 74 with the universal/singular distinction,
D(ix?), 74–6 xv
see also C geographical and/or geological things, 87, 90
D(x)=the Only a and b rule, 73, 76, 96ff., 107, grades of modal involvement, 111–12, 118
231ff., 238ff. grasshopper (example), 86
the principle has no expression in object Grelling’s paradox, 213
language, 96 guarantees, 237–9; see laws and propensities
see also universal versus singular
D-absolutist, 105, 159 haecceitas, 125ff.
definition, 7, 8, 11–12, 19, 80, 88, 175, 196, Holy Trinity (example), 35–6, 45–50
213, 216ff., 220–4 homeomerous things, 74–6, 168, 230
see also terminology see also Fundierungsaxiom, Tibbles
deixis, 5, 8, 11–12, 13, 78–80, 82, 124, 140, human being conception, xiv, 193–6, 203–4,
153, 155, 177 234ff., 242ff.; see also face, forensic
delta or ⌬ principle, 109, 121, 151, 156 conception
determinables and determinates, 6, 8 (see Hume’s church (example), 35, 43–4
note 6), 58–9, 69, 73, 194, 198
determinacy, xii, 151, 157, 160, 162, 163–7, idealism, 139, 151, 157, 158–9, 165, 171–3
170, 194–5 identity, absolute not relative, Chapter 1
dialect of sameness and difference, xv; see see also D, exigency of identity, R
judgements of identity, D and D holds, where it holds, necessarily, see
principles necessity of identity
difference, see determinacy, necessity of all or nothing, determinate, etc, Chapter 6
identity “identity”, a person’s own, 211, 235
divides its reference, see sortal concepts identity, coextensive surrogates for, 103,
division, see amoeba, Brownson, splitting 184
 Index of content 
identity ‘at a time’, 31–2, 40, 98, 177 materialism, 158–9
identity judgments, see judgments of identity memory, experiential or personal, xiii–xiv,
identity of indiscernibles, principle of, 61–3, 198ff. to the end of Chapter 7
146, 187–8 its factive character, 204, 209–30
ideology and ontology, xii, 146–7, 155–6 see also causal theory of perception,
imagining, 110–11, 118ff., 120 note, 121–3, normativity, repository
130–1, 140ff., 199–200 mereology, xi, xvi, 37–40, 43, 94, 166–7
incompleteness (supposed) of bare identity, the mereological version of identity, 39, 94
thesis dismissed, 17 Meriden Stone (example), 34, 41
india–rubber ball (example), 90 monads, 85, 92
individuate, 6–7, 18, 159–61 monstrous birth (example), 161, 168
see also single out, sortal concepts, what is it? mountain (examples), 87, 166–7
question/category
individuation, principles of, natural kind terms, 77ff.
see criterion/principles of ‘near miss’, 188–92
identity/individuation necessary/contingent: relation of the
inertia, 13–14 distinction to that of a priori / a posteriori
intensionality, intensional replacements 12–14
prevented from replacing schematic in relation to individuative essentialism,
letters, opacity and transparency, etc, 27, Chapter 4 and 5
28, 49, 115, 116, 119ff. de re construals, 111–12, 124–7
interpretation, 2, 103, 198–203 in syntactic combination with lambda
Ip, a putative criterion for personal identity, abstraction, 112–14
201ff. in relation to conceivability, 110, 111,
isotopes, 79 118–23
“asymmetrical” cases, 118
Jekyll and Hyde (example), 35, 45 consistency of necessity with vagueness,
John Doe (examples), 29–32, 35, 43, 44, contestability, etc, 111, 129–30, 192
175–6 ordinary necessity coinciding with physical
judgments of identity, seen as an extension of necessity, 85, 127
the capacity to single out, not inferred necessity of class membership, 95–102,
from information that is identity-free, 118–33
xiii, xv, 61, 105, 167, 171, 176–83, 233, necessity of identity, and of difference, 19,
240 113ff.
jug (example), 34, 36–40 necessity of origin principle, questioned,
121ff., 132
lambda abstraction, 19, 113ff. net (metaphor), 152, 160
landmark (example) 41–2 Nile (example), 90
see also Cleopatra’s Needle, Meriden Stone normativity, 223ff., 228ff.
laws and propensities relating to a kind of nucleus, individuative, 92, 207
thing, xi, 70–2, 77, 81, 82, 84, 86, 88, numbers as objects, 51, 58
107, 139, 153, 237–9
see also activity, guarantees, life histories objets trouvés, 134
principle, natural kind terms official (“same official”), 35, 44–5
Leibniz’s Law, 4, 5, 27, 28, 37, 46ff., 53–4, 59, Only a and b principle, see D (x)
114, 164, 170 ordered pairs representing relations in
see also exigency of identity, intensionality extension, 50–1
life histories principle, 37, 40, 57 organs, such as heart, liver, 86
live/not live distinction, 90
locust (example), 86 paradigms, 5, 215–16
Lot’s wife (example), 64–7 see deixis
paramecium (example), 71
M-and P-predictates (in P. F. Strawson’s sense), persons, xii–xiv, 156, Chapter 7
195, 198, 234 phased sortal concept, 29–30, 42, 64
‘martlet’, 218 ‘phases’, etc, 30–3, 175–82, 201
 Index of content
physiognomy, xiv, 208–9, 234–6 single out, xiii, 1, 2, 3, 6–7, 52–3, 128–9,
possibilia, 108 150–1, 157–61, 166–7, 169–75
possible worlds, 109, 116ff., 130, 134–6 as conditioning the idea of an object, 173
practical capacities (as presupposed to see also what is it? question
individuation), xiii, 2–3, 5,7, 9, 18–20, slices, see phases, synchronic/diachronic
105, 198 contrasts
see judgments of identity sortal concepts, and sortal predication, 8–11,
predication, 5, 10, 19, 37, 111 18–19, 21–2, 42–3, Chapter 2 passim,
Proteus, 66 63–5, 69, 77ff., 125, 128–30
sortal concordance of certain concepts, 33
qua, 48–50, 89 sortal continuity principle, 65
quantities, 100 sovereign (example), 42
quasi-remember, 213–14; definition of same species, 77–80, 83–6, 123, 124
cross-examined, 214–30, Chapter 7 splitting, 72–3, 91–102, 161–2, 168–71, 208,
passim 228, 233, 238, 240
quasi-thunderstorm, 218 see also amoeba, Brownson
stages, see phases
R or relativity of indentity thesis, denied, stereotype, 79, 198
23ff., 29, 34ff., 53–4, 55–6, 66, 105, substance, 1, 3, 5, 6, 8, 13, 14, 72, 77ff., 100,
140–1, 158 150, 178, 194, 196, 228, 241
R, a relation of psychological connectedness see also activity, laws and propensities
that the neo-Lockeans aim to define, 205 substitutivity, see congruence, exigency of
in fact presupposing person identity, xiv, identity, Leibniz’s Law
212ff., 226 subtraction, among concepts, 217, 218
see also quasi-remember supervenience, xiii, xiv, 5, 143, 186ff.
real and nominal definition, 11–12 sweater and wool (example), 150–1
realism, 105 symmetrical object, 188
see also anti-realist conceptualism, synchronic/diachronic contrasts, 9, 32, 71–2,
conceptualism and conceptualist realism 160, 176–82
reciprocity, two-way flow, back and forth, 2, 3, see also phases
18–20, 149–50, 153, 158–60, 172–3 synthetically produced, 90
reduce, 5, 156, 186
reference, 6, 41, 52–3, 176–82 teletransportation, 241
reflexivity of identity, 25 temporal parts, 31
relativity of identity, thesis of, see R see also phases, synchronic/diachronic
repository (of memories), 228 contrasts
see also quasi-remember tense, 29–30, 180
restriction of a relation, 33, 63 terminology, xv, xvi, 1, 5, 7–8
Richard’s paradox, 213 see also definition
rivers (examples), 35 Thebes to Athens, road (example), 49
Roman Law, 92, 189–92 Theseus’ ship (example), 92ff., 95ff., 161–2,
169
salamander (example), 86 Tibbles, Tib (examples), 173–6
same donkey, and its relation to same simpliciter, transplant, 206, 207, 210, 238, 241
14–20, 25, 33 see also Brownson
same place at same time, 22, 94, 106, 147, twins (Siamese), 168
148, 150, 151, 181, 243 see also splitting
same what? question, see what is it? question
seas (various examples), 90, 94–5, 166 ‘unity’ relation, 176–7
sense, definiteness of, 218 universal versus singular or particular: the
sense and point, 3, 171–3, 217–22 distinction contrasted with the
sense and reference, 9, 79–86, 165, 171–3, general/specific distinction, xv, 73, 228ff.,
243 233ff., 239, 240ff.
sets: and essentialism, 95–102, 118–23 see also D(x)
see also Fundierungsaxiom, ordered pairs use, meaning as, 1, 2, 3
 Index of content 
vagueness, xii, 111, 128–30, 137, 160, 165 works of art, 91, 101, 102, 136–8, 189–92
see also determinacy see also objets trouvés
‘world’, 165–7
wasps’ nests (example), 90
what is it? question/category, 9, 21, 24, 59, zeugma, 44–5
60, 77ff., 109, 121, 129, 131, 144, zygote (example), 239
160