NICHOLAS EASTAUGH

Authenticity and the Scientific Method

Past Approaches, Present Problems and Future Promise

Dr. Nicholas Eastaugh is a physicist and

chemist engaged in the history and analysis
of artists’ pigments. His major publication is
Pigment Compendium – A Dictionary and Optical
Microscopy of Historical Pigments. He is currently
Academic Visitor at the Research Laboratory for
Archaeology and the History of Art, University
of Oxford, and he runs the Pigmentum Project,
Oxford/London (www.pigmentum.org).

The use of scientific techniques to unmask into what he clearly felt was his own territory: 3
forgeries and fakes of paintings is much vaunted “Mr Fry having abandoned his claims as an art critic
and has a high public profile. Since the earliest applications and based the question of authenticity on the examination of
of X-radiography and pigment analysis in such renowned the surface of the picture, is it evident that the right person to
cases as the Wacker Van Gogh forgeries of the 1930s and consult is the chemist?” [Laurie 1927]
the Van Meegeren Vermeers, science has seemingly held a Fry’s response goes unrecorded, but the Editor of the
key role in the popular imagination, the expert in analysis Burlington Magazine nonetheless added his own comment:
unmasking the master faker. There is though by now a “Does not one’s past experience of the scientist’s
degree of maturity to the field, with significant demands decisions force us to accept [this] with some degree of
for such analysis ranging from due diligence questions hesitation…the question which many of our readers will
during acquisition in the public and private sectors, through be inclined to ask is, ‘How reliable are the tests on which
resolution of legal disputes, to ‘aspirational’ owners seeking Professor Laurie is prepared to rely?’ ” [Tatlock 1927]
to validate their discoveries. Nevertheless there is no While we might initially read this as yet another, minor,
generally accepted methodological approach or universally example of the interminable ‘Two Cultures’ debate of C.P.
agreed set of benchmarks for carrying out such studies, with Snow, in fact closer examination reveals a much richer
the lack of such agreed parameters potentially threatening to context. This exchange, and the events surrounding it, is
undermine what we do. This paper explores some of the key symptomatic of the emergence of a ‘scientific’ approach to
background history, the currently prevalent methodology, questions of authenticity. If we were to look for origins of
some associated problems and possible solutions presented such scientific rationalisation, then the 1920s were the point
by the use of science in authenticity studies of paintings at which methods and approaches coalesced into a coherent
discipline.
Arthur Pilans Laurie (1861-1949) remains well known
1. Introduction within this community for his book The Painter’s Methods
and Materials, still in print some 80-odd years since it
In June 1927 the Burlington Magazine published a letter first appeared just before Laurie was writing so acerbically
from A.P. Laurie. The critical focus of this letter was a review to the Burlington. Laurie was a chemist and principal of
by Roger Fry of an exhibition that had appeared in a previous Heriot-Watt College in Edinburgh who had reputedly been
edition. Laurie was exercised by Fry’s apparently straying encouraged into the analysis of paintings by the artist William

InCoRM Journal Vol. 1 No. 1 • 2009

 Holman Hunt (1827-1910). Laurie attended the University damaged Mantegna Triumphs of Caesar at Hampton Court
of Edinburgh, then King’s College, Cambridge, from where Palace in England.
he graduated with first class honours in the science tripos That Laurie was taking Fry to task for straying from art
of 1884. Laurie’s subsequent career saw him split his time to science might consequently start to look a little peculiar,
between academic work in chemistry, roles in government with an almost dogmatic insistence on separation of the
committees and book writing. His academic work included a art historian from the scientist. In fact Laurie and Fry had
lecturer’s position in chemistry and physics at the St Mary’s had contact beforehand, both sitting as panel members of a
Hospital Medical School in London, the chair in chemistry ‘committee of experts’ in one of the most celebrated cases
at the Royal Academy of Arts and, latterly, an advisory post of authenticity of the time, that of the so-called Hahn Belle
at the Courtauld Institute of Art. The crowning achievement Ferronière.
of his career was probably his appointment as principal of The Hahn Belle Ferronière is a version of the work
Heriot-Watt College, Edinburgh, in 1900 though, of course, by Leonardo da Vinci in the Louvre and we can summarise
his most enduring legacy must be his studies of the materials the case as follows: Harry Hahn, an American, apparently
and techniques of art (Who’s Who 1938 1937; Forbes 1949- acquired the painting in France. However, in 1920, just as
1950; NAHTSE). Hahn was about to sell the painting to the Kansas City Art
Roger Fry (1866-1934) is perhaps best known today for Institute, the renowned London dealer Sir Joseph Duveen
his invention of the term ‘Post-Impressionism’, a somewhat denounced the Hahn Belle Ferronière as a fake. Hahn
salacious private life and (maybe) his own painting, but he promptly sued Duveen for half a million dollars. Duveen, in a
was also a noted art historian and critic of his time, firstly of countermove, used all his influence to set up a confrontation
the Italian Renaissance, later of contemporary movements in between the two paintings, establishing a panel of 10 experts
French art. What makes Fry an especially interesting figure to pronounce judgement. The panel, selected by Duveen,
4 in our present context however is that his first grounding included the renowned art historian Bernard Berenson,
was as a scientist – he read Natural Sciences, like Laurie various leading museum directors, Roger Fry and, as a late
at King’s College, University of Cambridge. He was drawn and self-offered addition, A.P. Laurie. On viewing the Hahn
though to art during his undergraduate years, turning to and Louvre versions together these experts largely dismissed
this as his principal field of study immediately after (Woolf the Hahn version. Nonetheless, during the following six
1940; Spalding 1980; Bruneau). Such early training must years Hahn’s lawyers systematically undermined the expert
nonetheless have remained with Fry throughout his life testimony in the trial, such that by 1929 Duveen felt it
because we find him commenting not long before his death necessary to settle out of court with Hahn.
that: As John Brewer has recently written of this case:
“If ever there was a study…needing as it does the “Harry Hahn contrasted ‘the air-spun conjectures,
co-operation of so many sciences…it is surely that of Art- subjective guessings, sixth-sense flairs, and, in certain
history, and I would make the claim that the benefits it would instances, downright dishonesty produced by members of
confer would be at least equal to those it would receive. […] the Duveen clan’ with ‘reliable historical documentation’
We have such a crying need for systematic study in which founded on the ‘objective and scientific nature of accurate
scientific methods will be followed wherever possible.” historical research.’ […] The case of Hahn versus Duveen not
[Fry 1939, 3] only raised questions about the authenticity of a particular
In fact a notable feature of Fry’s way of looking at (and potentially extraordinarily valuable) picture, it raised,
art was his focus on the object and things internal to it, in stark terms, the question of whether the 20th-century
as opposed, say, to those strands in art history in the late art world was to be governed by the aesthetic opinions
nineteenth century where documentary sources were of a selfanointed elite of connoisseurs, or by the rigorous
paramount. It is not surprising to find for example Fry strictures of modern science. Duveen set out to vindicate not
taking Giovanni Morelli’s archly analytical study of only his condemnation of the Hahn picture, but the entire
Italian painters with him on an early trip to Italy (Spalding system of attribution and connoisseurship on which his
1980, 49; Morelli 1893; Sorensen). Fry had also studied hugely profitable business depended.” [Brewer 2005]
‘Old Master’ techniques, using them in his paintings and This background makes Laurie’s attack on Fry’s
advising others on their application. He was even known to methodology all the more remarkable. Clearly they had both
dabble as a restorer, notably while he was on the staff of the been present during discussions on the Belle Ferronière
Metropolitan Museum in New York and later, on the much- case four years earlier and, moreover, the trial was ongoing,

InCoRM Journal Vol. 1 No. 1 • 2009

dragging through the American courts. That Laurie would of the field as A.M. de Wild in Holland, Kurt Wehlte in
choose this particular moment (1927) to launch a public Germany and, of course, A.P. Laurie in Britain. Edward
attack on Fry, in a journal the latter had founded, is Forbes, renowned himself in the annals of this field as well
consequently at first sight surprising. In fact Laurie’s muscle as for his pigment collection, writing Laurie’s obituary in
flexing should instead be seen as more of a reflection of the 1949, explicitly credited him with being a pioneer of “…
strength of analysis of paintings at that time – Laurie would research into the methods and material (sic) of the old
probably have felt that at no point previously was there such masters by chemical study as well as the use of the X-ray,
a promising set of scientific tools and methods with which to ultra violet and infra red rays, microscopic examination,
tackle questions of attribution and authenticity. If we look at micro photographs, and the perusal of the early documents
the range of techniques developed by and through the 1920s, which bore on these studies” (Forbes 1949-1950). While
it forms a template for much practice since. Laurie did not establish this field single-handedly, Forbes’
summary nonetheless serves as a useful reference point
for the essential methodological toolkit for the scientist
2. The Disciplinary Matrix investigating art. It can be summarized as follows:

 General examination by imaging techniques such as

It is always an instructive exercise in any field to reflect
X-radiography, infrared imaging, and Ultra Violet
upon the features that define it. Who is in the community?
fluorescence.
What are our common practices? How do we describe our
 Detailed surface observations of the painting using low-
field to others? A helpful framework in these circumstances is
power microscopy.
what is known as a ‘disciplinary matrix’, largely popularised
 Analysis of the constituent materials by chemical
through the philosopher of science Thomas S. Kuhn (Kuhn
testing. 5
1970). This is the set of features – the theories, assumptions,
 Examination of ‘technique’ – the way in which these
methods and case studies – that we would all largely agree
materials were used as well as aspects such as what may
upon as constituting the basic knowledge of the field. To
be loosely termed ‘brushwork’.
understand what Laurie must have perceived as occurring
 Comparative study of historical documents for
we must examine what features we (actually or instinctively)
information on materials stated to have been used by
recognize as being present in the field.
artists in the past.
Although we are focusing on the 1920s we should not
lose sight of the fact that the use of scientific techniques It could plausibly be argued at this point that all modern
to analyse paintings was, even then, by no means new. scientific analysis of paintings derives from this basic pattern
Nadolny, for example, has traced the use of simple methods and, while more advanced analytical instruments such as the
of investigation on historical paintings as far back as the electron microscope and the gas chromatograph have been
late eighteenth century, with relatively sophisticated studies developed, in essence the same common threads of method
certainly being employed by the early nineteenth century are still being followed.
(Nadolny 2003). Important figures such as Mrs. Merrifield At the same time we should also be aware that this
and Charles Lock Eastlake researching historical materials toolkit encompasses more than just the physical processes
and techniques in the mid-nineteenth century, and A.H. required – the sampling and direct analysis of samples – it
Church, the chemist, writer on artists’ materials and first also encapsulates a philosophical approach (which we will
professor of chemistry at the Royal Academy in London (thus discuss shortly) as well as a rôle, for example, for historical
standing as a direct precursor to Laurie) laid foundations in studies and comparative analysis. It also contains a specific
critical areas for the subsequent development of this field conception of what science ‘is’ in this context: that of the
during the later nineteenth and early twentieth centuries. analytical chemist, this being quite distinct from, say, the
Use of X-rays for the examination of paintings occurred not process of the art historian.
long after the discovery of the X-rays themselves – in fact in To complete our disciplinary matrix of scientific
1896, less than a year after Roentgen announced his findings methods for authenticity we need to add two other
(Bridgeman 1964) – while all the basic elements of the components, the socalled ‘paradigm’ cases the field cites,
approaches practiced today were certainly in place by the and a philosophy of science argument that allows us to ‘test’
1930s. A group of names stands out from this latter period for authenticity.
(including but certainly not limited to) such luminaries Kuhn’s narrower meaning of paradigm – case studies

InCoRM Journal Vol. 1 No. 1 • 2009

 presented as exemplars – allows us to highlight that, the field, its contribution is also ‘played up’ – while scientists
repeatedly, certain cases come up when people discuss participated in these particular cases, it is not so clear that
science and authenticity, features that we want to emphasise their testimony necessarily brought about the specific legal
as discipline markers. Apart from the Belle Ferronière, we conclusion. (With Van Meegeren it was instead probably his
might therefore cite the Otto Wacker Van Gogh case and, confession and recreation of a painting for the court.)
of course, Van Meegeren and his forged Vermeers and de Our other component of the disciplinary matrix was
Hoochs. a specific ‘authenticity’ test, a criterion for deciding when
The elements of the Wacker case can be briefly outlined. something is or is not genuine. When a philosophical basis
Otto Wacker, a dancer, took up dealing in art in 1925 and for the methodology of ‘analytical authenticity’ is explicitly
quickly developed a ‘sound’ reputation with Van Gogh mentioned, it is invariably that of falsification, a concept
experts. Such was his renown that the 1928 De la Faille adapted from the philosopher Sir Karl Popper that has
Catalogue Raisonné of Van Gogh listed some 33 paintings become embedded in day-to-day consciousness, especially
from Wacker. Four of these, however, were subsequently amongst the ‘hard’ scientific community to whom it was
included in an exhibition where they were recognised as fakes, addressed (Popper 1934).[1] Briefly expressed we may
thereby casting doubt on the entire group. Legal proceedings state Popper’s position as being that no number of positive
were instituted against Wacker in December 1928 and, when outcomes at the level of experimental testing can confirm a
it came to trial in 1932, there was expert testimony from two scientific theory, but that a single genuine counter-instance
technical specialists. These were A.M. de Wild and Kurt is logically decisive in rejecting it. Re-phrasing, if we
Wehlte, both names still remembered. De Wild reported the have a theory about something – the existence of the atom,
presence of ‘resin’ in the paint which was ‘not used by Van planetary motion – then however many times we check it
Gogh’, while Wehlte presented comparison X-rays to show against some observation and get a ‘positive’ result, we are
6
differences between real and fake Van Goghs. only not disproving it; on the other hand any contradictory
Second, the notorious Van Meegeren case has of result should cause us to abandon our theory. Crucial to
course been discussed extensively before, but we may again Popper’s method, any truly scientific theory is required to
summarise (Coremans 1949; Dutton). During the 1930s make testable claims, otherwise (according to Popper) the
the art historian Bredius authenticated the painting known theory is ‘non-scientific.’
as the Supper at Emmaus. In the late 1940s, however, to Translating this into our present context, then according
avoid the more serious charge of wartime collaboration, to the Popperian conception of analytical authenticity
Van Meegeren confessed to having created this and scientists take a hypothesis, in this case the claim that a
other paintings. Analysis by Paul Coremans at this time painting P is by a specific artist X. This claim is then tested
showed anomalous features were present, such as that the by taking samples and analysing them, declaring if there is
natural ultramarine Van Meegeren had used was actually any counter-evidence, such as anachronisms or features alien
contaminated with cobalt blue, a pigment unavailable to to artist X, that the hypothesis that the painting is by X, is
Vermeer and his contemporaries, while cracks sometimes false. Thus we have a very clear statement of methodology.
did not pass through upper paint layers. In this way we have a set of protocols that allow
As stated earlier, such paradigm cases are important us to use scientific methods to determine authenticity.
as exemplars used by a field to define its scope, practices Analytical tools such as X-rays and pigment analysis are
and theoretical framework. Each of these cases therefore combined with knowledge of what is and is not appropriate
illustrates the role of science in the authentication process to find historically through an unambiguous test with sound
and they are commonly cited for just such purposes. philosophical foundations.
Importantly, however, the contemporary field view of
them is not necessarily exactly what happened and there is
commonly simplification, selectivity and inaccuracy in the 3. Problems …
retelling. For example, in the Van Meegeren case it was not
until the period 1967-1973 that researchers such as Keisch The disciplinary matrix just outlined provides a
used lead 210 decay and isotope analysis to show that the lead clearly established methodological approach to questions
white was modern (Keisch 1968) and Breek and Froentjes of scientific authenticity. Since the time of A.P. Laurie
identified the paint medium as a phenol formaldehyde resin and others in the 1920s and 30s, successive generations
(Breek and Froentjes 1975). For the purposes of promoting have seemingly built on this, giving us an ever deepening

InCoRM Journal Vol. 1 No. 1 • 2009

knowledge of artists and their materials as well as new generally accepted as a legitimate and useful process. It
analytical solutions to problems of identification (such as involves the application of analytical methods to uncover
non-invasive methods, advanced organic analysis and new contradictory evidence for what something is believed to
imaging techniques) that mean that we can go further and be. Attribution on the other hand is both pro-active and
deeper than ever before. There are well-demarcated spheres controversial. It involves taking a painting without prior
of expertise – art historian, conservator and scientist – assumptions and, through analytical means, determining
working together symbiotically to solve problems. There is authorship. Few scientists currently make such claims, and
also an increasing awareness and use, in both institutional those who do tend to be those using mathematical analyses
and commercial arenas, of the contributions of scientific of images rather than chemical analyses of materials, these
analysis. being identifiable and separate sub-disciplines.
Or is this really a true picture of the discipline? A To authenticity and attribution studies a third process
counter-view might run: should in practice be added: the determination of date. While
In fact there has been little critical examination of the clearly connected to issues of authenticity and attribution it
procedures for decades. The apparent stability of methods can nonetheless be a process carried out quite independently
applied actually hides a lack of rigorous assessment of the of any specific authorial question. Through use of direct
protocols used or whether they truly tell us what we want dating techniques like radiocarbon or dendrochronology, or
to know. Accurate and reproducible results are held back by applying a detailed knowledge of changing patterns of
by a lack of knowledge, access to equipment, and reliable use of materials, it is possible to give estimates of when
sources of reference data. Interdisciplinarity is often a something was created. Such evidence can therefore be
fiction, with ultimate judgement deferred to non-scientific used to underpin a more specific belief, such as that of
dogmas, a process that technical specialists connive with. authorship. Moreover the creation of a date specific test that
7
Nor has there been any serious research into either how the can be used for our falsification process – explicitly that of
overall and detailed process functions or how we can really anachronism – has been much easier than, say, finding a
do it better. material or technique always or never used by an artist. An
The truth probably lies somewhere in between, but if assessment of date is thus the most oft-stated approach to
we want to properly disentangle the state of the discipline, scientific authenticity.
then we need to examine such questions as: At the expense of discussing some of the broad
issues mentioned above (and some that were not) we will
 How well defined and reliable actually is our
instead take Popperian falsification as an exemplar of the
methodology?
problems that exist. Falsification theory has a number
 What are the questions really being asked?
of well recognized problems, such as the so-called
 Who is defining these questions and what impact does
underdetermination problem and the Quine-Duhem problem.
this have on the processes we use and the conclusions we
There is also a failing specific to the application of analysis
reach?
to falsification of art works, which will be called here the
 What are the unstated assumptions we use and what
‘hypothesis test’ problem. From this we shall see that it is
impact do these have?
possible to critically examine what we do, identify specific
 Are we working within a system that fails to let us answer
difficulties that we may or may not be aware of, and find
legitimate questions?
solutions where necessary.
 Are there unstated methodologies that we use without
The underdetermination problem concerns the making
justification?
of unwitting assumptions by choosing one hypothesis to test
 What is the effect of pragmatic constraints, such as time,
rather than another. Simply put, for every theory imaginable,
money, and confidentiality? 
another, contradictory, one may also be conceived that would
 What is the nature of any interdisciplinary process?
explain the same facts just as well. So, we might ask, why
In partial answer we should perhaps first make some this hypothesis and not another? For example, why choose
clarification of what is done when practising ‘scientific ‘This is a painting by X painted in 1626’ over ‘This is a
authenticity’. For example, we generally grasp that there is a painting by somebody who liked X in the 19th century and
distinction between the process of determining authenticity wanted to emulate him,’ or even ‘This is a disinformation
and that of making an attribution. The application of science exercise perpetrated by a secret committee determined to
to the determination of authenticity is widely practiced and bring down the art establishment’? Who decides which

InCoRM Journal Vol. 1 No. 1 • 2009

 ‘reasonable’ hypotheses (whatever ‘reasonable’ means) statement of cut-off points, but which, in practice, are not.
should be considered? Significant argument often seems to rest A good example is patent dates, which might at first seem
on conflicting ideas of what people think the (sole) hypothesis to present strong termini – dates before which they had not
should be rather than, say, determining an appropriate set of yet been invented but after which they were a product on
hypotheses and then testing these side by side. the open market. Whereas one could expect the flaw to be
Another difficulty with Popper, known as the Quine- that pigments were available to be used at an unknown date
Duhem Problem, illustrates the extent of care needed in applying earlier than the patent date, in fact the converse will be argued
scientific methods to questions of authenticity. The Quine- here (Eastaugh 2006). A good illustration is the case of the
Duhem problem essentially states that there are inevitably phthalocyanine pigments. Phthalocyanines form a highly
multiple components to any hypothesis. For example, when important class of organic pigments of twentieth century
testing the hypothesis that ‘This painting is by X,’ a range of introduction that provide stable and strongly chromatic blues
other aspects are also being tested, including the fundamentals and greens used for everything from paint to the coating on
of analytical methods (‘elements are identifiable on the basis CD-ROMs. An accidental discovery apparently took place
of their characteristic X-ray emissions’), the reliability of in 1928 at the Grangemouth works of Scottish Dyes Ltd.,
comparative information (from ‘this is how X painted’ to which directly led to the recognition and development of
‘these were the pigments available to X’), investigative skills phthalocyanines; the first patent, British Patent 322,169, was
(‘this paint sample comes from an original area of paint and to Dandridge, Drescher and Thomas of Scottish Dyes in 1929.
not later restoration’) and so forth. If the test ‘fails,’ it could be Detailed studies of the chemistry and commercialization of the
due to any one of these parts. Philosophy of science has several manufacturing processes took place in the earlier 1930s, with
strong responses to this problem, mainly aimed at determining formal public announcement in the press in 1935. According
the analyst’s confidence in each element of these component to most histories of the discovery of these compounds,
8 hypotheses, but these have not commonly been appealed to in however, there were plausibly two prior reported syntheses
our field.[2] Usually (in this author’s experience at least) the of phthalocyanines, first Braun and Tcherniac in 1907 at the
communities involved in these decisions accept the reliability South Metropolitan Gas Co. in London, and then Diesbach
of analytical science and tend to blame either the analyst or the and von der Weid during 1927 at the University of Fribourg.
presence of restoration. These are of course actually informal, But neither group recognized their discoveries as the potential
tacit, specifications of component hypothesis likelihoods. The commercial success phthalocyanines they would become.
true challenge however is to ascertain what these component (Eastaugh et al. 2004)
hypotheses really are and then establish objective measures So, what is the terminal date? A hard line approach would
of reliability for each, something that requires a profound surely argue for the earliest demonstrable identification as the
knowledge of the domain(s). Moreover, the use of ad hominem terminus, which would place the date for phthalocyanines at
arguments for example – those that seek to dismiss an argument 1907. Others might argue that it has to be the first explicit
on the basis of who is making the argument rather than the merits discovery and characterisation, in 1928. Then again, why not
of the argument itself, such as referring to what institution they argue for the full-scale commercialization, sometime in the
do or don’t belong to – is simply unacceptable. mid-1930s?
Lastly, as we saw earlier, a fundamental aspect of In fact if one examines the history of painting materials
Popperian falsification is the provision of a ‘test,’ the one finds a strikingly similar situation in every case. Even
experiment that is applied to a hypothesis in an attempt to if it is believed that there must be a ‘magic moment’ of
falsify a prediction of it. Our specific exemplar is anachronism, discovery (such as supposedly happened with Prussian blue)
since it is the most easily defined and widely used of these there is often not only uncertainty as to precisely when this
tests. Commonly the ‘anachronism test’ is expressed through ‘magic moment’ took place, but also questions about how
the concept of terminal dates, the dates before and after long it took for the discovery to have any practical impact.
which a pigment or other material was not available. Such This phenomenon is well understood, though not apparently
data have been collated in a number of forms, such as the so in painting analysis circles, even if the better review
important pioneering paper by Kühn (Kühn 1973), though articles implicitly demonstrate it (Laver 1997; Keijzer
this author has also not infrequently seen use of supposed lists 2002). So-called ‘diffusion of innovation’ studies, a field
of ‘terminal dates’ that are actually nothing of a sort. with a surprisingly long history of which the chief modern
In practice these might better be called ‘pseudo- exponent has been Everett Rogers in his classic Diffusion
termini,’ as they use what initially appears to be a clear of Innovation (Rogers 1962), provide a fuller understanding

InCoRM Journal Vol. 1 No. 1 • 2009

of this idea. Inventions and their impact on society follow of introduction of a new pigment, more and more paintings
a distinct and almost invariable pattern. A small group will be found to include that pigment. Typically the same
known as ‘innovators’ are the primary discoverers of things, diffusion of innovation pattern will be manifest in all new
followed by a larger group of ‘early adopters,’ followed by introductions, including not just different materials such as
the main ‘early majority,’ the similarly large ‘late majority’ supports and paint media, but also techniques – comparable
and then the ‘laggards.’ The theory provides a useful model patterns can be found in such features as red/grey double
of how things actually come to be used, few at the beginning, grounds popular in the seventeenth century for example.
with a cumulative uptake over time, and all those who are When applied to paintings, the core implication of this
going to use it by the end. is that it is far more likely that we will encounter a material
Unfortunately this view fundamentally undermines in a painting when that material is available in abundance.
our use of Popper and his falsification method since we no Conversely, it is extremely unlikely that the pigment will
longer have a clear test; there is no obvious cut-off point be encountered when it had just been discovered, or even
with which to define any ‘before’ and ‘after’, leading us into for some period of time after it is initially introduced
a falsificationist nightmare. commercially. The probability is that one has to wait until
well after, say, a patent, to see the invention widely employed.
Such probabilistic judgements are used widely and reliably in
4. … And Solutions life in general and can also be used in matters of authenticity.
I would be far more sceptical of a painting for example if it
It would be unfortunate if we were forced to abandon required me to believe in the earliest known use of a material
the use of scientific methods in studies of authenticity, or technique than if it was just a run-of-the-mill occurrence.
especially as (apart from putting us out of a job!) it seems Exclusive, yes-or-no judgements respond poorly to real-
intuitively clear that there is a role for such processes. life situations whereas formalised ‘is-this-reasonable?’ type 9
How, therefore, can we resolve this situation? In practice statements based on sound knowledge and methods of
there seem to be a number of strategies, of which we will analysis are both informative and useful.[3] A ‘likelihoodist’
look at two. The first to be described here is based on a solution therefore provides us with a means of specifying,
revision of how we apply and use chronologies; the second quite precisely, whether a proposition such as ‘Liubov Popova
is a brief exploration of how we might approach formalising used titanium dioxide white in the 1910s’ is reasonable or not
comparative analysis. (it is not, the likelihood being close to zero).
As a concept, diffusion of innovation helps us In fact the approach is richly productive, as it opens up
towards our first solution. We find that we can deconstruct a range of other practical steps. In particular it encourages
the terminal date problem such that instead of giving a a far more complex view of the history of materials. If we
dichotomous solution it provides us with a measure of can escape from the notion that we have to find clearly
‘reasonableness’ across the period of introduction. In the unambiguous situations, like Rembrandt and synthetic
case of phthalocyanines discussed earlier, the ‘discoveries’ ultramarine say, and use marginal probabilities, then we can
that took place prior to the grant of patent in 1929 were make use of information on many aspects of pigment history.
made by the group we call ‘innovators’; then, once the A simple (and simplified) case might be Prussian blue: in its
managers at Scottish Dyes Ltd. were persuaded of the early life it was made from crude starting materials such as
significance of the discovery and they moved towards blood, but was later synthesised from purer chemicals. If we
commercialisation we see the ‘early adopters’ arrive, and can differentiate these types analytically, then we have not
so forth. More importantly we can see that there is a) no one dating pigment but two. And so forth.
sharp cut-off representing a terminal date and b) a pattern We are currently exploring ways of taking these kinds
of increasing use over time that represents the gross uptake of data and combining multiple results to give overall
of the innovation. This can be usefully re-expressed here as date estimates, with interesting and promising results.
the likelihood of finding a product in use at any particular Such approaches are also opening up the possibility of
chronological period. Imagine that something is invented incorporating different types of data, notably those that
and someone wants to monitor how widely it is being used. might be considered as coming from the ‘art historical’
As time passes, surveys should indicate that increasingly domain, such as the availability of a source image or the
more people are using the invention. Similarly, the analysis structure of a workshop. An important future development
of paintings will confirm that as time elapses from the date is likely to be sophisticated model building, such as already

InCoRM Journal Vol. 1 No. 1 • 2009

 takes place in archaeological applications. of painting technical data that currently exist are also largely
Refining dating methods, however, is not the only inadequate as a result of poor access, standardization and
solution. In practice there is commonly informal use of interoperability issues. This is a problem not only for scientific
comparative methods based on what may broadly be called authenticity studies but also its sister discipline of technical
‘similarity’ where statements about ‘alikeness’ are used. art history, one that needs to be seriously addressed if we are
As with the falsification issue, little systematic research to progress in these areas.
has been made into the formal methods both required and The nature of the comparative process is perhaps even
available. more open at the present time. Objectively, we are looking
We can unpack the area to some extent nonetheless. for a means capable of taking diverse information such as
For a start there is a basic justification of process in that number and types of pigments, features of technique, perhaps
when an artist is expressing his ideas through the medium of even aspects of pictorial composition, and converting it into
paint he is making a series of choices that are significantly a robust measure of similarity or difference. Many of these
constrained by the time and place in which he works, as in fact exist, so much so that the choice can be confusing. For
well as his own situation and predilections within that. Thus example, so-called ‘discriminant analysis’ sounds appealing,
the set of materials chosen, how these are combined and since it is designed to predict membership assignment into
then used, represents the physical manifestation of both the dichotomous groups (such as ‘Democrat voter/Republican
artist’s creativity and the broader socio-economic context. voter’, or ‘real/fake’) on the basis of potentially diagnostic
Typically we thus get questions such as: What pigments, features of the individuals to be assigned. Unfortunately in
paint media and artists’ techniques does one expect to find practice there are significant difficulties for our application,
at a given time and place, or used by a specific artist? Are including the major issue that you need to have a set of data
the pigments, paint media, artists’ techniques found in the where the outcomes (group memberships) are clearly and
10
painting examined sufficiently similar to what would be unambiguously known, something not necessarily easy to
expected for that time, place or artist to justify a claim to define with authenticity questions. Instead, on the basis of
authorship? What is of actual consequence here is the specific studies we have made, neutral data exploration approaches
‘uniqueness’ of an artist and his situation. Can we discover where there is no such knowledge prerequisite seem to
physical attributes of paintings that are characteristic perform better. Clustering techniques for example allow
and, therefore, potentially diagnostic at different levels of differentiation and grouping of cases from which a critical
resolution, from broad time and place down to (ideally) an evaluation of the meaning of the groups formed, as well as
individual practitioner? Broadly speaking the answer should the extent of their distinctness, can take over.
be yes, though for such approaches to work two elements are We have also been exploring a technique known as
essentially required: first, sufficient data of high quality that Case Based Reasoning (‘CBR’). In CBR reasoning is based
is relevant to the problem and, second, a well-characterised on remembering: ‘remindings’ facilitate human reasoning in
method of comparison between cases. many contexts and for many tasks, ranging from children’s
A common practice for any field is the presentation simple reasoning to expert decision-making (Leake 1996).
of case studies and surveys of differing extents. For us As a technique, however, CBR compares the decisions taken
this ranges from conservation-related examinations of a in past cases to guide choice in a current, undecided, case.
single painting to entire oeuvre-busting exercises such as Such cases may be closely related to the original cases, or
the Rembrandt Research Project. Apart from their intrinsic else a novel one. A key benefit for us is that as an approach it
interest regarding the works concerned, they also act as is tolerant of smaller case groups than, say, clustering, while
important benchmarking exercises providing raw data for at the same time offering the possibility of ‘learning’ – the
comparative evaluation. Unfortunately such studies also tend simple addition of newly decided cases add to the overall
to be arbitrary and difficult to use as reference points for our discriminatory power of the system. We expect to be able to
purposes since there is a tendency to use what is available report on these results in the near future.
rather than what is ideal, have poor coverage, inconsistent
methodology, restricted analysis, provide interpretation rather
than data, and so forth. Up until the present time there has been 5. Conclusion
no satisfactory system of data consolidation for such studies,
with relevant information often published piecemeal or not Day-to-day experience often raises questions about
at all. Moreover, such intra- and inter institutional databases one’s basic ideas and approaches. How do you deal with

InCoRM Journal Vol. 1 No. 1 • 2009

issues that seem intuitively important but are irresolvable At the present time it is unclear where these studies
with the techniques available? Or quantify the residual will lead us. The growth of technical art history, the study of
uncertainty that one may feel after a judgment has been physical aspects of paintings, for example is bound to feed
made? Many of the areas explored in this paper have been new and important information into the field as a whole. The
framed in direct response to questions such as these that the need for reliable data for broader scientific studies makes
author has faced in his own practice. Systematic reflection revisiting questions of how to determine ‘authorship’ and
for this paper led to some interesting lines of thought. For date essential. The way these things are achieved in practice
example, it was quite surprising to realise that A.P. Laurie though will surely depend on us as a discipline continuing to
would probably recognise so much of the way science is ask pragmatic, self-reflective questions about our common
applied to questions of authenticity today, some 80 years assumptions. We must not take things for granted.
after he was writing to the Burlington Magazine about the
respective roles of the chemist and the art historian. That
the methodology has been so constant would most likely 6. References
have pleased Laurie. He was operating in an environment
Breek, R. and W. Froentjes. 1975. Application of pyrolysis
where Bernard Berenson could in all seriousness during the gas chromatography on some of van Meegeren’s faked
Belle Ferronière trial repeatedly pour scorn on technical Vermeers and Pieter de Hooghs. Studies in Conservation.
knowledge of pigments, X-rays, and chemical analysis as 20(4): 183-189.
Brewer, J. 2005. Art and Science. A Da Vinci Detective Story.
‘matters beneath a gentleman connoisseur’ (Brewer 2005,
Engineering Science 1/2: 32-41.
38). To Laurie, we might imagine, the widespread acceptance Bridgeman, C.F. 1964. The Amazing Patent on the Radiography
of his analytical approach and its apparent integration today of Paintings. Studies in Conservation. 9: 135-139.
into decision-making about paintings would count as a Bruneau, A.P. Fry, Roger Eliot (1866-1934). Oxford
Dictionary of National Biography. 11
vindication. At the same time we must also recognise from www.oxforddnb.com/view/article/33285 (accessed 14/01/08).
our own perspective that, while study of the physical does Coremans, P.B. 1949. Van Meegeren’s Faked Vermeers and de
not, as Berenson believed, put us into the lower classes, we Hooghs. Trans. A Hardy and C. Hutt. London: Cassel.
Curd, M. and J.A. Cover. 1998. Philosophy of Science: The
should nonetheless continue to critically and systematically Central Issues. W.W. Norton & Co.
examine the methods that we use. Stasis does not necessarily Dutton, D. www.denisdutton.com/van_meegeren.htm
equal reliability. (accessed 14/01/08).
Eastaugh, N. 2006. Scientific Dating of Paintings. inFocus
The central part of our discussion here additionally
Magazine. 1: 30-39.
concerned some subtle but important points about how we Eastaugh, N., V. Walsh, T. Chaplin and R. Siddall. 2004.
take the data we derive from paintings analysis and then use it Phthalocyanine group. The Pigment Compendium CDROM.
to arrive at decisions. This focus was deliberate in that maybe Oxford: Elsevier.
Forbes, E.W. 1949-1950. Arthur Pillans Laurie. College Art
we can at times become over-enamoured of the technology we Journal. 9: 206-207.
employ, at the expense of due consideration of interpretation Fry, R. 1939. Last Lectures. Ed. K. Clark. Cambridge:
and its methods. There are evident flaws in oft-cited rationales Cambridge University Press.
Keijzer, M. de. 2002. The history of modern synthetic inorganic
for determining authenticity and so we must seek solutions.
and organic artists’ pigments. Contributions to Conservation.
One of the most important points however was that viable Research in Conservation at the Netherlands Institute for
approaches that can resolve such problems do already exist. Cultural Heritage (ICN). 42-54.
Keisch, B. 1968. Dating works of art through their natural
These approaches (notably statistical data analysis, diffusion
radioactivity: improvements and applications. Science
of innovation, and case-based reasoning) have been developed 160(3826): 413-415.
and tested widely in numerous disciplines, their behaviour Kühn, H. 1973. Terminal dates for paintings from pigment
and robustness well understood through study and use. While analysis. Application of Science in Examination of Works of
Art. Boston: Museum of Fine Arts. 204 et. seq.
we need to find the most appropriate of these techniques for Kuhn, T.S. 1970. The Structure of Scientific Revolutions. 2nd
our situation, the formalism that they bring can actually help ed.
us identify strengths and weaknesses in our own approaches. Ladyman, J. 2001. Understanding Philosophy of Science.
Routledge.
To be able to assess the reliability of the judgements we make
Laurie A.P. 1927. The Burlington Magazine for Connoisseurs.
in these ways is of fundamental importance. In the process 50(291): 342-344.
we also find that such methods can also radically open out the Laver, M. 1997. Titanium Dioxide Whites. In Artists’ Pigments.
way in which we view systematic approaches to authenticity A Handbook of Their History and Characteristics, ed. E.W.
Fitz Hugh. 3: 295-355.
questions.

InCoRM Journal Vol. 1 No. 1 • 2009

 Leake, D.B. 1996.CBR in Context: The Present and Future.
In: Case-Based Reasoning: Experiences, Lessons, and Future
Directions, ed. D. Leake. Menlo Park: AAAI Press/MIT
Press.
www.cs.indiana.edu/~leake/papers/p-96-01.pdf
(accessed 14/01/08).
Mayo, D.G. 1997. Duhem’s Problem, the Bayesian Way,
and Error Statistics, or ‘What’s Belief Got to Do with It?’.
Philosophy of Science. 64: 222-244.
Morelli, G. 1893. Italian Painters: Critical Studies of Their
Works. Translated by Constance Jocelyn Ffoulkes. 2 vols.
London: J. Murray.
Nadolny, J. 2003. The first century of published scientific
analyses of the materials in historical painting and
polychromy, circa 1780-1880. Reviews in Conservation. 4: 39-
51.
NAHSTE.
www.nahste.ac.uk/isaar/GB_0237_NAHSTE_P1161.html
(accessed 14/01/08).
Popper, Karl. 1934. Logik der Forschung. Wien.
Rogers, E. 1962. Diffusion of Innovation. New York: Free
Press. 
Sorensen, L. “Morelli, Giovanni” Dictionary of Art
Historians.
www.dictionaryofarthistorians.org/morellig.htm
(accessed 14/01/08).
Spalding, F. 1980. Roger Fry: art and life. London: Granada
12 Publishing.
Tatlock, R.R. 1927. The Burlington Magazine for
Connoisseurs. 50(291): 344. Who’s Who 1938. 1937. London:
A&C Black Ltd.
Woolf, V. 1940. Roger Fry: a biography. Richmond: Hogarth
Press. Year Book of the Royal Society of Edinburgh 1950
(Session 1948-1949). 1951. Edinburgh: Oliver and Boyd.

This article is based on a keynote lecture presented by the author at the American Institute for
Conservation Annual Meeting held in Richmond, Virginia, April 16 - 20, 2007.

[1] There are various introductions to Popper and the philosophy of science; two good recent overviews
placing him in context are Curd and Cover (1998) and Ladyman (2001).
[2] For instance when we look at the various steps in a piece of analysis we might be much more confident
about, say, that the basic technique works (since many people have developed and used it) than the
interpretation of the results. The two main solutions are the use of so-called ‘Bayesian’ approaches and ‘error
statistics,’ both of which assign probabilities to each component. See, for example Mayo (1997).
[3] It is not the place here to describe the methodologies in detail, though they are based on established
approaches to defining likelihood and manipulating such values mathematically. In essence we can apply
techniques such as the Bayesian approaches mentioned earlier.

InCoRM Journal Vol. 1 No. 1 • 2009