Synthese (2012) 186:15

DOI 10.1007/s11229-011-9988-3
INTRODUCTION
Diagrams in mathematics: history and philosophy
John Mumma Marco Panza
Published online: 7 September 2011
Springer Science+Business Media B.V. 2011
1 Introduction
Diagrams are ubiquitous in mathematics. From the most elementary class to the most
advanced seminar, in both introductory textbooks and professional journals, diagrams
are present, to introduce concepts, increase understanding, and prove results. They
thus fulll a variety of important roles in mathematical practice. Long overlooked by
philosophers focused on foundational and ontological issues, these roles have come
to receive attention in the past two decades, a trend in line with the growing philo-
sophical interest in actual mathematical practice. Seminal contributions include the
historical/philosophical analysis of diagrams in Euclids geometry offered by Manders
in his 1995 paper (2008),
1
The date of publication of Manders paper is misleading.
Though it was not published until 2008, it was written in 1995 and widely circulated,
and became highly inuential as a manuscript. the logical studies of diagrammatic
reasoning contained in Gerard Allwein and Barwises compilation (1996), and Netzs
historical study on the place of diagrams in Greek geometry (1999).
These works exhibit a broad range of intellectual perspectives. To bring people
from these separate perspectives together, two workshops devoted to the history and
philosophy of diagrams in mathematics were jointly organized by Departments of Phi-
losophy and Classics at the University of Stanford, and REHSEIS (a research center
1
The date of publication of Manders paper is misleading. Though it was not published until 2008, it
was written in 1995 and widely circulated, and became highly inuential as a manuscript.
J. Mumma
Division of Logic, Methodology, and Philosophy of Science at the Suppes Center for History and
Philosophy of Science, Stanford University, Palo Alto, CA, USA
e-mail: [email protected]
M. Panza (B)
CNRS, IHPST (UMR 8590 of the CNRS, the Univ. of Paris 1, and the ENS - Paris), Paris, France
e-mail: [email protected]
1 3
2 Synthese (2012) 186:15
of the CNRS and the University of Paris 7). The rst workshop was held at Stanford
in the fall of 2007, the second in Paris in the fall of 2008. The present special issue is
the direct result of these workshops.
Because of the central position of Euclids geometry, in both the history and philos-
ophy of mathematics, the diagrams of Euclids geometry have been a major topic in
the recent research on diagrams in mathematics. This is reected in the make-up of the
issue. Part I, which contains over half of the contributions to the issue, concerns Euclid-
ean diagrams, addressing either their place and nature in the manuscript tradition, their
role in Euclids geometry, or their relevance to early modern philosophy and mathe-
matics. The contributions of part II examine mathematical diagrams in more modern
and/or advanced settings. In the remainder of this introduction, we briey describe the
content of these papers.
The two articles included in section I.a are devoted to geometric diagrams in the
manuscript tradition of ancient Greek texts. As Saito points out in the beginning of
his article, the diagrams we nd in modern versions of ancient Greek mathematical
texts are far from representative of those appearing in the manuscripts from preceding
centuries. These manuscripts result from various scribal traditions that exhibit various
practices in the production of diagrams. The general aim of both Saitos article and
Gregg De Youngs is to illuminate significant features of such practices.
Saitos concern is with how the diagrams of the manuscript tradition relate to the
geometrical content of the arguments they accompany. They often satisfy conditions
not stipulated to hold (e.g. by instantiating a parallelogram as a square) or alterna-
tively misrepresent conditions stipulated or proven to hold (e.g. by depicting angles
stipulated as equal as clearly unequal). There are, further, notable features with the
diagrams representing reductio ad absurdam arguments, and the diagrams represent-
ing arguments that range over many cases. To shed light on the different manuscript
traditions in these respects, Saito presents a case study of the manuscript diagrams
that accompany two propositions in book III of Euclids Elements.
De Young focuses on a specic manuscript tradition. He explores the basic archi-
tecture (i.e. the relation of diagrams with white space and text) of medieval manu-
scripts and early printed editions of Euclidean geometry in the Arabic transmission.
Many features of Euclidean diagrams remain constant through a long transmission his-
tory, although differences in scribal abilities exist. Yet there also seems to be a degree
of freedom to adapt diagrams to the architectural context. Adaptation to print brings
subtle changes, such as title pages, page numbers, and fully pointed text. The demands
of a long-held calligraphic aesthetic ideal and the necessity to compete against tradi-
tional manuscripts within the educational marketplace combined to favor the use of
lithography over typography in Arabic printed geometry.
The articles included in section I.b all share the general goal of deepening our philo-
sophical understanding of the diagrams of Euclids geometry. In his paper, Marco
Panza advances an account where geometric diagrams are fundamental to Euclids
plane geometry in xing what the objects of the theory are. Specifically for Panza
geometric diagrams perform two indispensable theoretical roles. First, they provide
the identity conditions for the objects referred to in Euclids propositions; second,
they provide the basis via their concrete properties and relations for the attribution
of certain properties and relations to these objects. After explicating this conception
1 3
Synthese (2012) 186:15 3
in general terms, Panza demonstrates what it amounts to vis--vis the Elements with
a detailed and thorough analysis of the definitions and the rst 12 propositions of
book I.
Aconsequence of Panzas account is that Euclids geometry is in a sense a construc-
tive theory. It does not concern a pre-existing domain of objects, but instead objects
brought into existence by the constructions of geometers. The issue of John Mummas
paper is the extent to which such an interpretation of Euclid can be given in formal
terms. At the center of his discussion is a proof system he developed where Euclids
diagrams are formalized as part of the systems syntax. Mumma argues that his for-
malization is superior to others in accounting for the spatial character of the theorys
constructions. It is not immediate, however, that the resulting picture of the theory
qualies as a constructive one. Mumma closes the article by pointing to the philo-
sophical work that is needed for it to be understood as such.
A prima facie feature of the diagram based reasoning of Euclids geometry is that
perception is involved in a distinctive and essential way. But howis it so involved? The
purpose of Annalisa Colivas paper is to develop the philosophical concepts necessary
to address the question. The key concept for Coliva, as the title of her paper indicates,
is that of seeing-as. A segment joining opposite vertices of a square in a diagram can,
for instance, be seen as a diagonal of the square, or the side of a triangle composing
the square. Coliva articulates seeing-as as philosophically sharp notion, and argues
that the ability is required to engage in diagram based geometric reasoning. She also
offers some initial thoughts on how her analysis bears on the status of geometrical
knowledge with respect to the a priori versus a posteriori and analytic versus synthetic
distinctions.
In her paper, Sun-Joo Shin considers Euclids diagrammatic arguments in order to
examine commonly held views on the strengths and weaknesses of diagrams in com-
parison with symbolic/linguistic representationsnamely the views that the latter are
superior for giving proofs and the former are superior for brainstorming and discovery.
Shin relates the rst viewto Locke and Berkeleys opposing accounts of the generality
of Euclids proofs, and argues that Berkeleys account offers a way for understanding
diagrams as a legitimate means for general proofs. She proposes that a major reason
behind the second view, aptly illustrated by Euclids proofs, is that diagrammatic rep-
resentations of individual mathematical objects are effective in generating the right
associations and connections with what is already known.
Section I.c contains papers that explore how the diagrammatic method of Euclids
geometry inuenced and/or was interpreted by thinkers in the early modern period.
Though mathematics was pushing far beyond classical Greek geometry in this period,
Euclids Elements nevertheless maintained its position as a foundational text and thus
formed part of the intellectual context in which philosophical or methodological reec-
tions on mathematics were carried out.
Such reections in Humes Treatise formthe subject of Graciela De Pierriss paper.
In part II of Book I of the Treatise Hume asserts that perceivable yet indivisible min-
ima compose geometric continua, but do so in a confused manner. Consequently, for
Hume, geometry does not possess the exactness of arithmetic, which concerns clearly
distinct discrete quantities. De Pierris argues that these doctrines follow in a princi-
pled way from an epistemological model in which phenomenologically given sensory
1 3
4 Synthese (2012) 186:15
images are fundamental. She also explains how Euclids geometry for Hume, despite
its relative inexactitude, qualies as a demonstrative science in being based on simple,
easily surveyable diagrams.
In her paper, Katherine Dunlop explores the role of diagrams in Newtons Prin-
cipia by considering Newtons claim that the reasoning of the book is distinctively
geometrical. She notes that Newton conceives geometry as the science of measure-
ment, and argues that his characterization applies to the propositions of Section II,
Book I. As other scholars have noted, these constitute a foundation in the sense that
they show certain spatial quantities to be related as measures to nonspatial quantities,
in particular, force and time. Dunlop takes Newtons point to be that the measures
he supplies can be manipulated in the same way as elements of diagrams in classical
geometry. In the case she examines in detail, Newtons proof of Keplers area law, the
principal measure of time is just the diagrammed trace of an inertial motion. Dunlop
contends that Newton aims to share in geometrys certainty by resting determinations
of equality and relative size (between nonspatial magnitudes) on relationships that are
open to sensory inspection.
In the Critique of Pure Reason Kant often illustrates his claims with geometric
proofs from Euclids Elements. One thus might look to recent work on the diagram-
matic character of these proofs to gain a better understanding of Kants philosophy of
geometry. In his article, Michael Friedman argues that such a project would be misdi-
rected. He argues specifically that Kants notion of a schema for geometrical concepts
and his distinction between pure and empirical intuition precludes a diagrammatic
interpretation of his theory of geometrical intuition. Such an interpretation would fail,
for Friedman, to explain why Kant takes the constructions of Euclids geometry to pro-
vide an a priori framework for empirical space. Friedman relates these points to Kants
theory of space, and the role of geometry and spatial intuition in the transcendental
deduction of the categories.
Dominique Tournss paper, the rst of section II.a, serves nicely as bridge between
the papers of part I and those of part II. In the article Tourns provides a survey of
the changing role of diagrams in the history of differential equations from the sev-
enteenth centurywhen there were still vestiges of the diagrammatic approach of
Euclids geometryup until the time of Poincarwhen the inuence of Euclid on
mathematics was entirely absent. The article is a useful resource for those interested in
understanding diagrams in mathematical subjects that are not ostensibly or exclusively
geometrical in content.
The other contributions of section II.a examine diagrams in the thought and work
of two giant gures in the history and philosophy of modern mathematics.
Danniele Macbeths proposes a new way of reading Freges logical notation, as
a sort of diagrammatic language within which to exhibit the contents of concepts.
Taking Frege s proof of theorem 133 in Part III of his 1879 treatise Begriffsschrift as
an illustrative example, Macbeth aims to show just how a proof in Frege s language
(so read) can extend our knowledge despite being strictly deductive.
In his paper, Ivahn Smadja seeks to clarify how Hilbert understood the relative
positions of diagrams and axioms in mathematics. An easy, and overly simplistic,
characterization of Hilbert on the issue would be just to attribute to him the view the
view Shin examines in her contributioni.e. diagrams belong to the realm of heuris-
1 3
Synthese (2012) 186:15 5
tics and discovery while axiomatics belong to the separate realm of justication. That
it would be overly simplistic is revealed by Smadjas study of the connections between
the diagrams found in Minkowskis work in number theory and Hilberts work in the
foundations of geometry. Smadja specifically describes the links the former have to
Hilberts axiomatic investigations of the notion of a straight line, and Hilberts concern
with the conceptual compatibility of geometry and arithmetic.
The papers included in section II.c, the last group of papers, concern contempo-
rary mathematics and linguistics. At the heart of Solomon Fefermans discussion are
innite diagrams for the proofs of central theorems in set theory, model theory and
homological algebra. The diagrams are innite in the sense that they depict innitely
iterated constructions. Feferman contends that they are more or less essential for
understanding and accepting the proofs, and considers whether this serves as evidence
against the so-called formalizability thesis: the idea that all mathematical proofs admit
of formalization. He argues for the conclusion that it does not in the light of recent
discussions of the formalizability thesis.
Sketch theory, a mathematical theory that treats the diagrams of category theory
as mathematical objects, is the topic of Brice Halimis paper. Halimi presents a tech-
nically sophisticated discussion of the theory in order to bring out its relevance to a
variety of philosophical and foundational issues. These are: the capacity of a mathe-
matical diagram (in category theory and elsewhere) to represent not just a static object
but also the dynamics of a proof; the role and nature of diagrams within category
theory; and the alternative ways category theory and set theory provide a semantics
for mathematics.
A general point of many of the papers discussed so far is that diagrams can have
important theoretical functions within mathematics. Arancha San Giness paper pro-
vides evidence that the same is true in linguistics, in the most direct way possible.
In it she presents a diagrammatic account of the role played by anaphoric pronouns
in natural discourse. The account, specifically, is a diagrammatic system of represen-
tation for sentences containing such pronouns. After motivating and describing the
system, she shows how it accounts for a variety of linguistic phenomena that have so
far resisted uniform treatment.
John Mumma & Marco Panza
Stanford, Paris, May 2011
References
Allwein, G., & Barwise, J. (Eds.). (1996). Logical reasoning with diagrams. New York: Oxford University
Press.
Manders, K. (2008). The Euclidean diagram (1995). In P. Mancosu (Ed.), The philosophy of mathematical
practice (pp. 80133). New York: Oxford University Press.
Netz, R. (1999). The shaping of deduction in Greek mathematics. A study in cognitive history. Cambridge,
USA: Cambridge U. P.
1 3