Connoisseurship: Conceptual and Epistemological Fundamentals

John H. Brown, University of Maryland

[This essay appears in Jason Kuo (ed.), Perspectives on Connoisseurship of Chinese Painting, 2008. (Washington, D.C.: New Academia Publishing). That printing contains an error in section 30 concerning a reference in Berenson, which is herein corrected. Also the numbering of footnotes from fn. 15 onward is raised by one. Otherwise the versions are identical. Readers are referred to that collection for insights into connoisseurship of Chinese painting.]

The following essay attempts to lay out the basic epistemology of connoisseurship. It approaches the subject from a general philosophical perspective. While striving to be relevant to the practice of art history, illustrating the theoretical points with examples drawn from the literature of that discipline, it makes no claim to art-historical expertise.

I. The term connoisseurship: a survey of usages and conceptions.

a. Art-historical connoisseurship. Connoisseurship in these pages refers primarily to the endeavor to identify artworks by time, culture and authorship. It is not restricted to any particular branch of this endeavor or to any particular means of identification. This contrasts with the usage that limits the term to connoisseurship of the eye as opposed to scientific connoisseurship , that is, physical testing to establish time or place of origin of the materials used. If there is a better term for the totality of the art-historical endeavor, I will be happy to embrace it. Until such a one appears, using connoisseurship in this comprehensive sense has merit. Given that the context is art history, all and only aspects of connoisseurship should be embraced by the term. The more limited modes of investigation within its purview are best designated by a qualifier, as in connoisseurship of the eye or scientific connoisseurship . 1 Insofar as iconography and social history can contribute to the authentication of works, one could speak with equal justice of iconographic connoisseurship or circumstantial connoisseurship, even though those usages do not currently exist, as far as I know. 2 The essential goal of art-historical connoisseurship is to establish so far as possible the work-identifying facts of production (who? when? where?). Other circumstances of production, including motive (why?), fall outside of, but may contribute evidence toward, the essential goal. Likewise assessment of the aesthetic or other values of works plays a merely supporting role, so far as the bottom line is concerned. Yet the support offered by appraisals of quality is often so important that it is easily mistaken for the main thing.

b. Aesthetic connoisseurship. In general usage the term connoisseurship commonly designates the cultivation of discriminating appreciation of works (and other things) in general, quite apart from the specifically art-historical authentication enterprise. Its essential target is the aesthetically good, not the art-historically authentic. The term aesthetic connoisseurship is appropriately used to refer to this appreciative enterprise. As just noted it plays a role in art-historical connoisseurship, but it may be carried on independently. Further, it need not be historical at all: one may adopt the aim of refined appreciation of styles and aesthetic effects (in art or anything else) wherever and whenever they appear. As philosophers love to point out, a fake may be aesthetically superior to its counterpart original. Anachronism need not be in itself an aesthetic demerit. Whether it is an artistic demerit arguably depends on the type of artistic project that gave birth to it.

c. Value-connoisseurship in general. In the world at large connoisseurship can refer to any refined practice of value discrimination, whether aesthetic or another value is in view. Thus there are connoisseurs of horse flesh. The practices here divide according to the specific sets of values: thoroughbreds are judged by speed and endurance, hacks by comfort of gait and manageability, hunters by strength, speed, endurance, courage and intelligence, quarter horses by speed, especially over short distances, agility, and intelligence. And so forth. No doubt aesthetic values play a part in the value sets relevant to these, but even when present they need not be dominant. The only limit on the term seems to be this, that the practice deals with things of notable value where the value in question admits of fine distinctions. Thus it would idiosyncratic to speak of a connoisseur of refrigerator stick-ons. Many collectibles are doubtful candidates for connoisseurship on grounds of insufficient (and insufficiently graduated) value. But the margins are definitely hazy. Where does philately fall? Can one be a value-connoisseur, as opposed to an authenticity connoisseur, of postage stamps?

d. Authenticity. Like many other terms for cognitive and moral value, authentic admits of a range of meaning. In some contexts the term is redundant, as are true , real , actual , and genuine . An authentic original is just an original, as a real thoroughbred is just a thoroughbred. The attributive merely highlights the substantive without adding anything of substance to it. But such qualifiers can also be used to raise the standard for the things referred to. A real man is a notable specimen, well above the average. A true friend is one whose friendship is deeply entrenched, emotionally engaged, and reliable. In art history authentic is used in both senses. (1) An authentic Raphael is just a Raphael; here the qualifier's function is rhetorical rather than substantively semantic. For there is no substantive difference for the qualifier to mark. A fake Raphael is no more a Raphael than a fake dollar bill is a dollar bill. It is this semantically empty sense that expresses the essential goal of art historical connoisseurship, since determining which works are authentically of an author, school, culture, or period is merely to determine which are of the author, school, culture, or period. (2) A work of art is also praised for being authentic when it meets high standards of creativity, personal expressiveness, or other prized artistic values. Since the prevailing wisdom is that copies by other hands, especially of different periods or cultures, never meet the standards set by originals deserving this appellation, authenticity in this sense is often taken as a reliable mark of authenticity in sense 1. On the negative side sense 2 allows one to say that an artist may occasionally copy herself rather than producing an authentic work. Some aesthetic rigorists are averse to inauthentic works counting as works of art at all.

e. Authentication. Unlike the adjective, the verb to authenticate is invariably semantically substantive. It means to establish that a work is of a certain author, culture, or period, in a context of initial uncertainty about these matters. In this context establish is not used always as an absolute; art-historical usage tolerates failed authentications, as when an art historian refers to the mistaken authentications of others. But of course the aim of the authentication enterprise is to reach conclusions that are both true and solidly grounded.

f. The scope of art-historical connoisseurship. Visual art comprises works of a considerable variety of types: unique or one-of-a-kind works (drawings, paintings, carved or modeled sculpture), multiples (prints, castings), and designs. Within each of these types one finds subtypes variously facilitating or limiting individual style or expressiveness, representational or thematic subtlety, or decorative splendor. Along a different dimension, the creative process may be solitary or collaborative, swift or protracted. Along another, the level of artistic accomplishment may vary widely. These differences must obviously be taken into consideration in the connoisseur's investigations, but none of the variations falls outside the scope of connoisseurship, which is as wide as are the who, where, when questions in the field of art broadly conceived.

Copies, forgeries, unauthorized reprintings or recastings, even reproductions, by whatever hand and under whatever pretensions of originality, fall within the connoisseur's purview, as do plagiarized or forged designs. But since authentication is labor intensive, priority is appropriately given to originals, and among them, to originals of significant artistic merit, except where the art-historical project to which authentication is relevant requires a different emphasis. 3

II. Propositions concerning art-historical connoisseurship

1. Identifying properties. In attributing an artwork to an artist, culture, or time, any type of property or property-cluster may conceivably be of use, and even definitive. Style, subject matter, brushwork, materials, facture, labels, documents, and so on, may all enter into a reasoned case for a given attribution. No category of property is excluded a priori. The reason is simple. Uniqueness with regard to time, place, or author is a highly abstract relational property. It reduces to a negative, possessing some property not possessed by anything produced at other time or place, or by any other author. In the nature of the case there is no type of property that may not satisfy the uniqueness condition. For whether a property is unique to a given object O, or to a class of objects C, depends solely on what else there is. If the property is possessed only by O, or only by members of C, it identifies an individual as O or as a member of C. That is why uniqueness is a relational rather than an intrinsic property. For a property to identify a work as of an artist, culture, or time does not require the property to be aesthetically interesting or under the control of the artist or meet any condition other than being possessed exclusively by the relevant individual or class. Needless to say, unique properties typically are highly complex rather than simple, but in principle anything could suffice.

2. Recognizable identifiers. An identifying property may not, however, be demonstrably so. What art history needs are recognizable identifiers. There is no way a priori to know which these are. There is no a priori reason to believe that they constitute a single set for all artworks. A prudent connoisseur will therefore keep an open mind and welcome additions to the set of recognizable identifiers. This is simply a matter of making use of whatever valid clues come to light. The situation is essentially the same as in historical research in general. The fact that the objects of concern for connoisseurship are aesthetic products ought to exercise no constraint whatever. In particular, adventitious identifiers, ones that identify only sporadically and in an “accidental” fashion, ought not to be spurned. Much documentation is purely adventitious.

3. Generally applicable identifiers. Notwithstanding the previous caution, it is perfectly reasonable for connoisseurship to seek generally applicable identifiers. Even the ideal of an all-purpose identifier for given types of works is a rational goal. These desiderata are, after all, straightforward consequences of the basic aim, essential to any intellectual inquiry, to arrive at the simplest possible solution to the general problem at hand—that is, Occam's razor principle. Now it would be highly unrealistic to assume that such a condition can be met. For there is no reason a priori to think that any all-purpose identifier exists, comparable to the DNA identifier of individuals. But if, contrary to expectations, such a one is discovered for, say, Chinese paintings on paper or silk, then of course the connoisseurship problem for that class of objects would be considerably simplified. 4 And identifiers that are generally applicable to particular classes of works are not unreasonable to hope for. Giovanni Morelli is widely believed to have posited the existence of such identifiers for Italian Renaissance paintings. (See Sections 27-28)

4. Property-clusters. Typically, identifying properties are highly complex property-clusters. Logically these are conjunctive properties, where the property P = the conjunction (or intersection) of A, B, C…. The constituent properties taken together narrow the possibilities of origin to a given time, place, or author. They are the properties cited in an art historian's argument for a particular attribution. I will refer to the constituents of an identifying property as partial identifiers.

5. Partial identifiers. As in any inquiry into the identity of a thing, it is reasonable to begin with the partial identifiers that are suggested by such knowledge as we have of variables limiting the possible time, place, or authorship, and seek to refine our discernment of these features—always keeping our eyes open for other types of partial identifiers. Hence if we are sensitive to qualities of line, depicted form, tonality and color, texture, rendered atmosphere, spatial organization, and so on, it makes sense to draw upon these sensitivities, refining them to the nth degree and deploying them as systematically as possible. But there are compelling reasons why we should also explore other partial identifiers: physical evidence, iconographic evidence (in the broad sense that includes what Panofsky calls pre-iconographical subject matter), 5 evidence of manifest pictorial skill, documentation, evidence drawn from social history, and anything else that might conceivably be relevant. For in principle any partial identifier might complete the case for a given attribution. Equally, on the negative side, any partial identifier may demolish an otherwise plausible case for a given hypothesis. For until one has checked all the possible partial identifiers, one cannot in principle tell whether the work possesses a property incompatible with the identification seemingly supported by the others. Like all other empirical findings regarding distant states of affairs that, complex in themselves, also leave complex traces of many sorts, authentications can always, in principle, be falsified by new evidence.

6. The essential contextuality of evidence. No item is evidential of anything in the abstract, by itself, but only in the context of the total system of knowledge. It is easy to lose sight of this because the context tends to be taken for granted and attention given only to the item in question. Of course not everything in the context has much effect on the evidential force of a given item, but enough does to make it almost impossible to give a truly comprehensive account of it—that is, to construct an argument in which all contributors are given due credit. Thus a connoisseur confronting a work brings a complex set of posits from common and technical knowledge from many areas, in and beyond art and art making. 6 One result of this is that whole regions of possibilities of authorship, period, and culture are ruled out at a glance, since there will be massive incompatibilities between features of the work and large stretches of art history, both temporally and geographically. Coming to Raphael's Alba Madonna , for example, we see at a glance that its rendering of bodily form and circumambient atmosphere is too advanced for the Quattrocento, to say nothing of the Trecento. This perception narrows the possibilities within which all further investigation and argument take place. When one talks about one thing proving another, it must always be understood that the evidence has this or that degree of strength relative to the context of knowledge at that moment. Another consequence is that the soundness of an evidential argument will depend on the soundness of those parts of the background knowledge that are relevant to the kind of case under examination. Our immediate certainty about the Alba Madonna not belonging to the Trecento would collapse if our suppositions proved wrong about why Giotto painted as he did (to cite the merest tip of the iceberg).

7. The essentially eliminative force of evidence. Evidence in favor of authorship, period, or culture is always eliminative in the sense of ruling out possibilities, progressively narrowing down the unfalsified remainder, ideally until only one author, period, or culture is left. It is easy to overlook this when our belief flashes straight to the end result as if we directly perceived the goal: the authorship, time, or place. Taken in context some properties in an identifying cluster will summarily eliminate otherwise promising possibilities. Physical tests, for instance, are notorious for decisively defeating subtle intuitions of skilled connoisseurs of the eye. Stylistic properties may also strongly refute hypotheses based on documents, and vice versa. It is noteworthy that physical properties often testify to the impossibility of a work belonging to an artist, culture or time, whereas stylistic ones testify only to the improbability—for example, that any hand other than Raphael's or Su Shih's could have painted the work. Subject-matter properties sometimes raise the improbability close to the level of physical or psychological impossibility. A detailed, fully legible representation of a Stanley Steamer would be more than merely anachronistic in an otherwise apparently authentic 16 th century drawing, even one bearing all the stylistic marks of Leonardo.

8. Degrees of evidence. In certain cases the accumulation of evidence in favor of an attribution is so massive that for all practical purposes the attribution is absolutely certain. Some certainties (called “axioms” below, Section 14) are in fact essential to the credibility of the practice of connoisseurship as a whole. But however diligent, industrious, and persistent the practice is, common caution forbids us to expect more than a modest degree of warrant for many, perhaps even a majority of attributions of authorship of known works. Inevitably the force of the evidence depends on the circumstances of the case. The natural and normal thing, given past practices of art making and conservation, is for there to be a wide range of attainable degrees of rational confidence, especially regarding authorship. Thus we must expect to encounter cases in which the yield obtained from further investigation will decrease without limit toward zero. There is no way to attain rock-hard certainty as to which these are, however, even after the zero point seems to have been reached. Art historians like police investigators depend heavily on totally unexpected and unearned revelations to crack hard cases. Absent such serendipities, refractory cases usually are never solved, partly because the passage of time progressively destroys whatever hidden evidence there was earlier. On the other hand, a revelation is never strictly impossible even after all the known trails of evidence have proven unproductive. Confessions are a paradigm case of unearned discovery, e.g., van Meegeren's, which was motivated by the need to defend himself against the charge of having sold Belgium 's patrimony to the Nazis, and that of Israel Rouchemovski in the case of the Tiara of Siataphernes. 7

9. Empirical observation. The security of an attribution depends heavily, directly or indirectly, on empirical observation of an ordinary sort, even in cases where the connoisseur's claim is backed only by arcane holistic qualities supposedly accessible only to the hypersensitive connoisseur. This is because rational confidence in the style or facture being authentically of the period or of a particular author rests upon generalizations about the probability of successful replication of known stylistic or facture properties. Such generalizations must be based on observation of the performance of painters of different skill levels who are trying to replicate the properties in question. Nothing short of such observation could validly underwrite claims to perceive the hand (spirit, personality) of the master as distinct from that of a copyist, in the work under scrutiny. Of course the evidence is almost always of a rather general sort, even when garnered from long experience, and it enters diffusely into the fabric of our knowledge. The result is that attributions on the basis of connoisseurship of the eye can enjoy a high degree of rational credibility only if documented by specifying the observable features that undergird the general impression and also reinforced by evidence of other kinds—documentation, physical tests of the age of materials, cultural history, and so on.

10. The effect of the medium on replication. Different media are variably difficult to replicate. Those that allow for preliminary drawing, gradual building up of opaque layers, and undetectable correction of slips of the hand are in general easier than those that make these preparations and revisions impossible. Chinese paintings on silk are a paradigm case of the latter. In such cases the possibility of a “perfect fake”—a copy literally indistinguishable from an original in side-by-side comparison—is excluded from the outset, although of course distinguishability does not reveal which is the original. Perhaps it is also more difficult to imitate exactly the hand of the master in such a work than it is in the case of an oil painting, for example. It is not easy to imagine a Chinese master standing in front of a copy and discoursing on his experience in helping to paint it as Giulio Romano is said to have done before Andrea del Sarto's copy of Raphael's Pope Leo X and Two Cardinals . Giulio's error is understandable because of the extraordinary likeness Andrea could achieve due to the revisability allowed by the medium. 8 The possibility of exact replication is of course also drastically affected by the character of the image. A twentieth-century non-figurative painting of the sort produced by Ellsworth Kelly is easily replicable at a level of effective indiscernibility, whereas no such possibility exists for one of Pollock's classic drip paintings. There can be no possibility of identifying Kelly's work by subtle stylistic discernment, whereas Pollock's work is made to order for that discernment.

11. The effect of the medium on physical tests. Media also divide with respect to their amenability to precise physical tests, given the cultural imperative to preserve the work. On this basis Chinese works on paper and silk are not well adapted to the tests currently available. Some of the difficulties are nicely explained in an article by John Promfret dealing with the career of the virtuoso connoisseur, painter, and forger Chang Dai-chien (Zhang Daqian) and the well-known controversy concerning his works in general and The Metropolitan Museum's Riverbank in particular. 9

12. The effect of art-world practices. Social practices affecting the making, collecting, displaying, recording, and preserving of artworks have a large effect on the ease or difficulty of attributions. For they determine how much documentary evidence comes into the public domain regarding time, place, and authorship. Modern art-world practice has considerably expanded the available evidence, and this trend can be expected to continue, saving our descendants much labor regarding attribution. Mechanical reproduction is a major aid in this endeavor. The media frenzy of contemporary society ensures that every major artist's oeuvre is minutely documented. Still more sophisticated devices are available for use when and if the incentive exists, and technology will offer more and more unbeatable tags as time goes on. Where these safeguards are used from the time of production onward, attribution of the works thus secured will be far less of a problem. To be sure, technology also allows for more and more sophisticated ways of replicating works and of defeating safeguards—forging documentation, for example. But the overall trend will almost certainly be to diminish problems of attribution of new works.

13. Documentation. The term documentation , narrowly construed, refers to records directly pertaining to the work. Works for which no such records have been found are referred to as undocumented. In current art-historical practice, however, the term covers a somewhat larger body of records bearing directly or indirectly on the work without naming or describing it: records of artists' birth, death, marriage, family, travel, training, professional career, patrons, associates, and so forth. 10 Indeed if we rely on the general meaning of the term documentation the scope of the category could be extended over the entire body of background information relevant to a given authentication, since it can be brought to bear only if presented in some transmissible form, that is, by some form of documentation. But in the present context it seems best to adhere to the current usage, with two highly significant additions. It seems appropriate to add to documents in the usual sense, first, archeological evidence of the place and time of the work, sites offering geological documentation, so to speak, which is transmitted to art historians by documents in the normal sense; and second, evidence drawn from the physical character of the work concerning time and place, the materials documenting themselves and that record being revealed by physical tests. For these sources, like written documents concerning commissions or prices or ownership, are entirely independent of the intrinsic, artistic character of the works. Documents directly relevant to a given work are by no means restricted to those of the time or culture of origin. They need only intersect with the work at some stage of its career, including the present, as in the series of records establishing a provenance. Documents indirectly relevant to authorship, period, or culture are a rather sprawling lot without firm boundaries, but nothing is gained by trying to establish a firm threshold of relevance. Documents in the present sense include images as well as texts, for example the images obtained by X-ray examination of paintings and the now-ubiquitous photographic reproductions that vastly facilitate close comparison of works in distant places.

14. The epistemic priority of documentation; axioms; core oeuvres. The practice of authentication depends absolutely on documentation being sufficient to establish in a definitive way the authorship, period, or culture of some works, which serve as starting points, or axioms, of the entire authentication enterprise. The reason for documentation being (at this level) epistemically prior to, say, style or subject matter or facture is that it provides the only grounds anyone could conceivably have for thinking that stylistic or any other intrinsic likeness and difference can testify to time, culture, place, or authorship. At the most basic level direct observation of who makes what in what style grounds beliefs as to what similarities and differences of intrinsic properties are likely or unlikely in the work of individuals and groups in given cultures. On this documentation rest all the general hypotheses we apply to works of whose making we have no direct knowledge. This epistemic priority is greatly lessened once we have obtained credible (probabilistic) generalizations, or strong, widely based intuitions, concerning style and other intrinsic properties. Then likeness or difference of those properties may by itself testify convincingly in some cases. However, in principle the strongest sort of documentation always trumps the strongest possible stylistic and other intrinsic property clues as to authorship, period, or culture. 11 The epistemic priority of documentation should not be confused with objectivity or certainty. Some documentation is quite objective and quite certain (those two are not the same, either), and some is historiographically problematic. It is a commonplace that documents themselves call for documentation.

15. Widening the oeuvre. The process of reasoning involved in establishing an oeuvre (individual or collective) is centrifugal: one expands the oeuvre outward from the set of axiomatic works. Given a set of axioms, unbeatable documentary + physical + stylistic evidence can be assembled for a somewhat larger oeuvre . This can be called the core oeuvre of the artist, period, or culture. The core expands as more evidence is uncovered, but its integrity is compromised if the standards are lowered, since any inauthentic work accepted into a core oeuvre opens the door to others. Around the core can be grouped works less strongly attested by the total evidence but sufficiently warranted for practical historical purposes. The durability of axioms will depend on the survival of their documentation and the evidence relevant to its authenticity.

16. Oeuvres without axioms. Axioms need not obtain for all artists, since given credibly established core oeuvres, say those of Botticelli, Filippo, and Filippino Lippi, other oeuvres can conceivably be credibly projected consisting of “neighboring” works not furnished with axioms of their own. Whatever the merits of Berenson's particular argument, a case like the one he developed in favor of Amico di Sandro could conceivably have been prescient. 12 For circumstantial evidence might have accumulated sufficient to justify belief in some known artist hitherto without any securely documented works, say, Berto Linaiuolo, being the author of the works Berenson assigns to Amico—without there being “axiomatic proof” of any of these works, and therefore without the oeuvre enjoying as high a degree of rational assurance as axioms confer. 13

A vast amount of art lacks the documentation required for axioms of authorship. Of course every work once had ample documentation—in the sense of transmissible evidence of authorship, time, and place, for instance, by those who witnessed its making. In the present day the basic “documentation” for Greek vase painting is limited mostly to the archaeological evidence of time and place of production, taken in conjunction with our knowledge of the culture at large. This, like physical tests of the work itself, never suffices to establish axioms of authorship, only axioms of time and place, which are more firmly grounded when stylistic variables cohere with the archaeological record. The occasional declaration of authorship found on a work itself may justify adding a name but only in a rather tentative way, since the surviving documents do not provide enough warrant for anything more decisive. Likeness of style may sometimes give credence to the attribution of a number of works to the same unknown author. But here again sparseness of documentation restricts the level of rational confidence.

17. Temporal asymmetry of possibilities . It is a truism that not everything in art is possible at every time. Artistic possibility is bounded by technical and aesthetic competences 14 and preferences 15 , as well as by the opportunities afforded by the social context. These factors affect possibilities before and possibilities after. Thus, suppose we are confronted with the task of dating a work virtually identical in all major respects to a well-authenticated original work with a secure date of production. It is immediately apparent that the possible dates of production of the first work are unevenly distributed relative to the second. The work in question may be a copy by the same hand or by an associate of the artist. But equally it may be a forgery hundreds of years later. The possibilities before the fact are essentially nil, given the degree of similarity along all dimensions and given that the well-authenticated work is correctly authenticated.

Limitations of technical competence and aesthetic preference also govern the distribution of possible dates and authors where a work bears a lesser but still very substantial overall similarity to a securely attributed and dated original work. The work in question may be an earlier work by the same author or by a predecessor, or by an able contemporary who is drawn toward the style of the artist, or by a follower or a latter-day imitator. It will be judged a stage along the way to the latter, or a variation upon it, in homage or in rivalry. Here again, however, the distribution of possibilities is heavily weighted toward time after. The rarity of anticipations in comparison with reminiscences stands in rough proportion to the degree of overall similarity.

Only when the securely attributed and dated original work is itself consciously backward-looking do the possibilities stretch out evenly in both directions or extend predominantly into the past. 19 th century Nazarene homages to Quattrocento “primitives” are a case in point in Western painting 16 , and examples are legion in traditional Chinese art.

18. Technical competence. Technical expertise affects many aspects of works, material as well as depictive. Artistic effects often depend on technical virtuosity. Different artistic cultures develop different sorts and grades of competence for various classes of work. Since expertise is hard won, it emerges only when prized and persistently sought. Skilled practice is also expensive and therefore subject to the vagaries of patronage. When artists seek to develop new production techniques, as Leonardo did, they may venture into regions of incompetence with disastrous results. 17 Some great artists have been negligent of certain technical standards. 18 For all these reasons the sort and level of competence embodied in works differs greatly over time and with all manner of special circumstances. Its bearing on authentication is accordingly complex but quite substantial.

What a particular grade of a specific technical competence in a work shows as to the author, time, or culture can be determined only by particularizing art historical research, not by recourse to universally applicable principles. Nevertheless it is worth observing that the very concept of fine (high, serious, ambitious) art entails the aspiration to extend technical competences that have an artistic payoff. Hence in cultures embodying (even if not formulating) this ‘elitist' concept of art progressive refinement and diversification of competences is normal, even though there can also be dissenting practitioners who deliberately purvey technical crudities, as many twentieth century avant-garde artists did, or who deliberately eschew the higher levels of virtuosity in some respect, as certain Chinese scholar-amateur literati artists did in regard to depictive skill. 19 (Such willful deviation from the norm rarely occurs until the artistic tradition has attained a high level of technical mastery of the relevant sort.) In cultures not embodying the concept of fine-art fineness, development and diffusion of competences necessarily figure historically in the advent of the art and are at all times potentialities awaiting a motive.

Where the aspiration to a given form of technical mastery can be assumed, an original can often be distinguished from a copy by the level of mastery displayed. Accordingly, connoisseurs rightly examine works minutely for technical defects. The telling deficiency may be graphic or depictive. “Deficiency” like “mastery” is relative to the standard of expertise applicable to the work, however. A work is properly adjudged a later copy if the level of depictive competence exceeds the highest attainment for the documented time of the original. 20

Sometimes specialized, easily recognized, graphic competences are extraordinarily difficult to replicate and therefore provide reliable marks of a masterful hand. An example of this, cited by Lo Ch'ing, is the extraordinary uniformity of wave-forms over a sizeable section of Wen Po-jen's painting Lin-wu Cave of Tai Lake , dated 1555. The difficulty resides in maintaining exact control of the forms and the flow of ink over hours of uninterrupted production. 21

19. Are authentication arguments inherently circular? While there is no inevitable vicious circularity in connoisseurship, there is obviously a danger of it. For instance, where one attributes largely on the strength of a principle of stylistic likeness (this degree of likeness of works W and X making it probable that W is by the same hand as X), and then uses the class of works thus selected as a warrant for using the same likeness-principle in fresh cases (it worked in the last case so it is even more likely to work in this new one), and then claims that the extension retroactively confirms the initial posit. If the initial attribution is not independently confirmed by external evidence—by documents discovered, for example—then the process does not really give any grounds for greater confidence in that particular likeness-principle. In practice many fallacious “bootstrap” arguments of this sort are used. On the other hand, in many cases new, independent evidence is discovered and the circle broken.

20. Style-space. Since attributions derived by connoisseurship of the eye depend on perception of likeness and difference, the reliability of conclusions based on any such determination will be jeopardized by proliferation of examples done by close associates of the master. The difficulty here is one of the style-space cultivated by the master being close pressed and even infiltrated by the work of associates, or by copyists later on. Clearly it is possible for associates to infiltrate along many dimensions, even if it remains uncertain whether they can do so in all. The most admired masters are understandably the most attractive to imitators (bona fide or other) and their oeuvres therefore the most liable to corruption.

21. Social variables re. authorship. Cultures and periods may place more or less importance on singularity of authorship. This will show itself both in (a) the process of creation, not just in overt autonomy or collaboration, but also in how publicly the artist works (there are famous cases in the Italian Renaissance of artists sealing off their workspace when engaged on an important commission); and in (b) the response of the public, particularly in the value placed by viewers on being able to recognize authorship. The same variation may occur with regard to the recognizability of the culture and time of origin of works. If a culture were entirely indifferent to all such recognizability, forgery and plagiarism would be conceptually impossible social practices, though of course copying would remain possible and would probably be an entirely acceptable, or even the normal, way to produce new works. While actual art cultures seem never to go that far, they do vary significantly with regard to stress on singularity. In those with well-developed fine-art practices singularity and recognizability tend to be highly valued and forgery a going enterprise.

22. “Autographic” works. In philosophical aesthetics the term autographic is used of symbols, 22 works counting as a variety of symbol, whose individual identity is defined by the full range of their properties, including the history of production. Hence in the field of visual art none but the unique physical object (painting, carving) can be the work. Even the perfect fake fails the strict identity test. The vast majority of artworks are in this technical sense autographic symbols. Much of the complexity and difficulty of connoisseurship is a direct consequence of this character. Proving concrete (spatiotemporal) particularity is harder than establishing visual similarity (or even indistinguishability). Autographic works need not be autonomously created. Collaborative works can be equally autographic. So can multiples conceived as a set of products uniquely resulting from the creator's activity, e.g., the authorized edition of a lithograph. 23

23. Allographic works, in contrast, are ones that allow for significant diversity, both material and within limits stylistic, among the items that realize the work, as do correct performances of composed music or plays and correct editions of poems or novels. In the visual arts the clearest case of an allographic work is a design. 24 A true copy of a design, like a true copy of a poem, counts as a full instance of it regardless of when and by whom produced. The authentication of an instance of a design requires only a showing that it is a faithful and knowing realization of the creator's specifications, not that it was produced by the designer or at the designer's direction or within the designer's culture. 25 Hence, although allographic works are not merely general types but types created at a time and place by one or more creators, they cannot be forged. However, an existing design, like a poem, can be plagiarized, that is, claimed by someone other than the creator; and a design stylistically similar to the work of a known designer can be created and fraudulently ascribed to that designer. The latter is the only possibility of allographic forgery.

24. Autograph works. In art history “autograph” 26 refers most commonly to a work or part of a work produced by the hand of the artist as opposed to the hand of assistants or restorers. The usage seems to presume that the handwork is distinctive to the artist stylistically or expressively. Here the alternative is not allographic but simply non-autograph. The term refers not to a semiotic type but merely to production by the author. There are many degrees to which a painting can be autograph in this sense. A comprehensive authentication assessment would specify the degree of the artist's participation, somewhat along the following lines:

(a) fully autograph = painted exclusively by the author;

(b) autograph in all important respects = painted with only trifling assistance, for instance in the priming coat, varnishing, or stylistically insignificant in-painting;

(c) substantially autograph = painted in most significant respects by the author, but with assistants providing enough in-painting to provide a normally sufficient test of difference of hands.

(d) Selectively autograph = the author painting certain key parts, e.g., faces, and planning the rest, while the assistants paint enough for their style to be conspicuous in the work as a whole.

(e) Authorized studio work = assistants painting substantially the whole simulating the style of the master, who retouches the painting to bring it nearer his autograph standards and perhaps signs it.

(f) Unsupervised studio work = assistants painting the whole without supervision, without or without some initial contribution by the master, in the form of sketches or even a cartoon, or some preliminary underpainting, still within the ambit of the studio.

Since every part of a painting is the work of some hand, it follows that every part is “autograph” in relation to someone. But in practice the expression seems to be used almost wholly in reference to the master artist. Thus the portions of Raphael's Transfiguration painted by Penni are not typically referred to as autograph Penni. Exceptions are made in extraordinary cases, e.g., there is no oddity in Leonardo's angel in Verrocchio's Baptism of Christ being referred to as autograph Leonardo because his intervention is worthy of a master, revealing as it does a major emerging artistic personality different from that of his then master.

Justice requires us to acknowledge that the presumption of stylistic or expressive uniqueness of an artist's handwork must not be allowed to pass as a universal, let alone a necessary truth. Rather it functions as an unstated background supposition on the strength of normal artistic practice. Thus autograph status would not be accorded to the handwork of the artist in a moment of entirely routine or mechanical activity. The term applies rather to the artist in the midst of focused artistic endeavor—the artist operating as such. Moreover, if an artist forcibly constrains her natural artistic impulses in order to imitate an alien style, the result will not qualify as autograph. 27 The reason is that such handwork is of scant artistic interest.

There is no way to boil all the diversity of autograph/non-autograph concepts down to a few, rigorously exclusive categories—and there is no reason for consternation at this irreducible diversity. A simpler set of concepts would be unfaithful to actual artistic practice, just as a simpler practice would not require so much conceptual diversity). In any case we can always say what we mean in clear terms, even if quite long phrases are required. Where the context makes clear what is meant, a simple term may suffice. Thus in reference to frescoes and paintings in oil and tempera, autograph refers to a range from (b) to (c); it does not require as much exclusivity as (a), which might be called super-autograph . But autograph western drawings and watercolors satisfy (a), as do Chinese calligraphy and painting on silk and paper. Works falling in (d) and below would be referred to as studio rather than autograph. But full elucidation of cases requires that the underlying particulars be set forth.

25. Establishing autograph status. Obviously it will be intrinsically harder to establish autograph status in sense (a) or (b), strictly construed, than it will to reach a more generalized conclusion, with one large exception. In cases where the medium and facture curtail collaboration of the master-assistant sort the default supposition is that the master's hand determines everything or nothing. Not always, of course, for it is often possible for a later hand to add something. Still, connoisseurship, especially connoisseurship of the eye, has a clearer field in which to exercise its sense of the artist's hand where the medium limits the possibility of revisions and collaborative handwork.

26. The legitimacy of authorship. Let me underscore that nothing in the above warrants saying that authorship is not a viable concept. The lesson is merely that there are many sorts of authorial contribution (hence the need for a variety of concepts). The sensible course here is to assimilate the complications into a better articulated conceptual scheme. If one rejects authorship on such grounds, one would have to reject all the interesting and useful concepts (green, woman, particle, light, apple, individual). The aim of connoisseurship is not to determine one and only one mode of “true” authorship, but to deal with the variety of cases in as specific (= informative) a way as possible.

27. Local or “measurable” properties. Most of the “measurable” properties of depicted form traditionally stressed by connoisseurship of the eye are not strictly measurable. Thus the width of the metacarpus, length of the index finger, bulbousness of the earlobe, curl of the lip, those notorious Morellian indices, are depicted properties, not graphic ones. That is, they are properties that the lines and shapes on the picture surface represent, not properties of the lines and shapes themselves. Hence one could take measurements of them only if the pictorial convention, as embodied in the work, permitted precise determination of depicted magnitudes or ratios. But this is never the case in art; it is possible only in technical drawings that use strict parallel projection.

Thus most references to “measurable physical properties” are in fact references to more or less imprecise pictorial properties. The degree of imprecision is determined by how closely the pictorial convention (as employed by the artist) prescribes the absolute or relative magnitude. Thus if two viewers disagree about whether two depicted forms (two ears or noses) are depicted as being exactly the same or different, no procedure of exact measurement can settle the question. Rather, the determination must, in the end, be made by the (well-informed) eye, though measurements of marks on the picture plane are often helpful in refining the perceptiveness of that eye.

Typical references to these properties in fact involve another imprecision. For pictorial practices are quite imprecise, sometimes deliberately so, sometimes from incapacity to be more exact. Conventions for placing things in space, for color and lighting, and so on, are not sharp-edged enough for pictures to have a precise pictorial content. Vagueness and ambiguity abounds, as does outright inconsistency, without necessarily frustrating artistic purposes. Indeed indeterminacy and inconsistency often facilitate these purposes, most obviously where the effect desired is that of a diffuse impression, a quick glance, or a mysterious world. It also facilitates the organization of the design on the picture plane, where unity can be achieved only by softening the effect of recession and protrusion.

28. Morellian properties. The local forms, e.g., of hands and ears, traditionally associated with Morelli, 28 need not be unique to a painter to serve as potent identifiers. They function always within a previously narrowed context. For example, where we deal with originals, not copies, the presence of period or school indicators makes it necessary to compare the “Morellian” features only within those boundaries. The features need be unique to the artist only within his period or school—or for that matter within a given stage of her career. More generally, the presence of a Filippo Lippi ear in a work could deceive only where the rest of the features are tolerably consistent with the work's being a Filippo Lippi.

It is not necessary for an artist always to render a local feature in the same way, even during a period of her career, for it to help identify. For (a), even if an artist uses a variety of ear forms, the set may still be unique to her. And (b) deviations need not obscure authorship if they can be explained by special circumstances, e.g. where too much uniformity would be conspicuous, as in a row of figures like those in the Brancacci Chapel fresco cited by Wollheim. 29 (c) Occasional deviations do not keep the typical form from being unique to that artist. The typical form can still identify wherever the artist uses it. Thus there is no need to claim, as Morelli once did in a moment of exuberance, that an artist's characteristic forms never deviate at a given career stage. One must only avoid summarily rejecting a work that deviates. (d) Since in any given work there are many such local forms, deviation in one need not prevent the others from providing sufficient evidence of authorship.

29. Holistic properties. Many of the properties important in connoisseurship are “regional” or holistic rather than local. Regional properties are relational properties that arise from (are “supervenient” upon) local properties distributed in a pattern. The color or curve of an earlobe is a local property, the overall color or tone of a work a holistic one ( witness Berenson on a painting now in the Norton Simon Collection, Madonna and Child with Adoring Angel : “the golden flesh, the golden red hair, and the mauve of the Child's tunic produce an effect of light tone never paralleled in Botticelli”). 30 Some holistic properties are as “measurable” (accurately discernible) as most local ones are, for instance an overall effect of golden light. But frequently they consist of somewhat elusive impressions based on distributions of local properties that seem beyond our capacity to pin down. Still, they are crucial for artistic purposes. Artists aim to embody them, art lovers delight to perceive them. Refinement of our capacity to discriminate different degrees (strictly different rank orderings) of holistic properties is practically a definition of aesthetic refinement of our faculties. Aesthetic production and reception is largely a venture in sensitizing oneself to these, in achieving intensifications and attenuations of them in works, and in discerning and relishing them in works viewed. Connoisseurs rightly insist that many of these properties can be sensed in a very precise way. The “spirit” of a work or oeuvre or period is just an extreme case of a holistic property. Chinese connoisseurship famously proceeds largely by reference to such diffuse properties: “vital force,” and so on.

30. Consistency and inconsistency. Among the holistic properties of oeuvres or their subsets is the property of consistency (continuity, regularity). Even more interestingly, so is the property of creative inconsistency. This is not the complement of consistency—not just any violation of consistency will do. Rather it is a departure from an achieved, therefore now “easy,” consistency that conveys an effect of freshness (= good novelty), a sense of creative power in reserve showing itself in a departure that could be the basis for a new sort of valuable consistency. Such inconsistency is thus a new, looser sort of consistency (continuity, regularity). There is no real paradox here, merely a complication of just the sort intrinsic to the notion of fine-artistic fineness.

31. Intuitive discernment. Although the perception of subtle holistic properties is “intuitive,” a point much stressed by writers on connoisseurship, 31 it is essential not to construe this discernment in a mystical way. It is solidly rooted in features of line, design and facture (chiefly brushstroke) that are individually perceptible without use of aesthetic sensitivity. The “mystery” of the quality consists in two things: first, the elusiveness of the pattern of relationships among those basic features that enables them to produce the overall effect; and second, the “metaphorical” character of the effect itself. Thus “vital force” is rooted in base features that collectively produce the aesthetic effect; and that effect is a perceived (or felt) likeness between the pattern of features and experiences of vitality of body and spirit. Further, the likeness is cross-categorial, that is, it cannot in principle be increased to the point of exact likeness or indistinguishability. The experience of a line cannot be exactly like the experience of animal or spiritual vitality. Given these sources of the effect, whenever a question arises about whether work W has more vital force than work X, the issue can be negotiated rationally by the disputants pointing out as completely as possible the features on which their perception is grounded and the likenesses that operate in that perception. There is no reason to suppose that a good faith effort to reach agreement must fail. 32 That is, potentially these perceptions are intersubjectively valid, even universally so. The possibility of well-founded connoisseurship of the eye rests absolutely upon this potentiality being realized.

32. Intersubjectivity. Two intersubjectivities need to be achieved in relation to the holistic properties just referred to. One intersubjectivity concerns the properties themselves. The other relates to their description, which often seems subjective in the sense of “poetic” or idiosyncratic to a given writer. As just indicated, the descriptions are typically metaphorical. Despite their reputation for being “subjective,” closer inspection will reveal that there is no sharp difference between them and literal descriptions in point of vagueness or variability from speaker to speaker in a close-knit linguistic community such as the community of art-historical scholarship. When Syndey Freedberg says that in Titian's St. George (ca. 1509) “The armour of the Saint is taken as occasion for a turbulence of light,” 33 the description gains specificity from the metaphor. All the controls listed below can be exercised as rigorously upon metaphorical as upon literal descriptions. There is therefore no special problem of describing artworks stemming from the indispensability of metaphor. As with literal descriptions, the problem is simply (!) one of achieving the needed level of accuracy and clarity.

33. The role of ostension. In this regard a distinction has often been noted between descriptions standing by themselves, capable of serving satisfactorily in absence of their referent, and descriptions whose full utility is dependent on the presence of the referent or an adequate surrogate. When the painting (or reproduction) is present, descriptions may use “ostension,” that is, direct reference to, that is, pointing to, the work that manifests the property. Thus descriptions such as “spatially incoherent” or “quasi-cubist” (said of a Chinese landscape) or “foreground figures plastically realized while background figures are reduced to heads with doubtful bodies plugged into the openings left by the former” have a quite specific sense. There may be no need to work up a more self-sufficient formulation, and it may be counterproductive to try. Such reliance on ostension is not essentially different from a botanist speaking of different sorts of foliage forms by terms that require the referent to be displayed before it is well understood. All this is natural and normal, by no means unique to aesthetic descriptions. Descriptions normally get their specificity from known comparison cases, or points of reference, which establish standards (of slimness or corpulence, of brightness or dullness, of heaviness or lightness, slowness or alacrity, and so on). It is necessary to have a suitable comparison case in mind, if not always before one's eyes. Where the property described is unfamiliar and subtle, however, it may be essential to surround it with relevant comparison cases; or where it is complex and metaphorical, as is frequently the case in connoisseurship.

Given that an individual connoisseur's intuitive perception is sufficiently stable and important to weigh heavily in her mind regarding an attribution, there is reason to expect that intensive, protracted, freely experimental, collegial effort by the connoisseur and others will develop a descriptive practice (aided by ostension) sufficient both to convey the property the connoisseur discerns and to reveal at least many of the local features on which it depends. The working assumption should be that no such property will forever resist all efforts at comprehension. No je ne sais quoi should be accepted as the final word.

34. Verbal formulation. Verbal descriptions of stylistic and facture properties (the latter referring to brushwork and other properties of the physical “make” of the work) cannot be either perfectly comprehensive or perfectly specific. They cannot encapsulate the full individual character of their subject, whether this be a part or the whole of a work. This limitation is not unique to descriptions of artworks or things of aesthetic interest, or cultural as opposed to natural things, or to any category as opposed to another. It is a truism that descriptions are in some measure general and selective. Indeed, their normal purpose requires that they be so—that they not give more information than is relevant. From this truism, however, it is wrong to conclude that descriptions have some particular ceiling of specificity or comprehensiveness. Descriptions can be made increasingly subtle and comprehensive without known limit. The utility of descriptions is not in general dependent on any particular level being attained in either of these dimensions, but only on the level being appropriate to the need.

Most of the hurtful imprecision in describing artworks comes not from any intrinsic limitation of description as such, but from the describer not knowing enough about the subject of description or from the art world at large not having developed a sufficient common understanding of the subject and sufficient descriptive resources to provide an adequate description for the purpose at hand; for instance, to articulate what in the work is truly distinctive to an artist, school or period or what properties of the work justify a particular assessment of its quality.

35. Seeing differently. It is often said that everyone “sees differently,” as if subjective differences excluded objectivity across the board. But while it is true that no one sees everything at all times identically to anyone else, it is quite false that no one sees anything at any time to have identically the same property as is seen by another. A large part of our cognitive career is taken up with learning how to see the same properties as others do, and we succeed for a considerable range of cases. The range has no detectable boundary, and we are constantly extending it in some directions by improved methods, just as we are constantly losing ground in other directions from falling out of practice, e.g., skill in following faint spore through the bush or in cutting complex angles in wooden beams by sight. It is clear that refining perception requires persistent effort. The lesson is obvious, and it runs directly counter to the skeptical conclusions drawn from commonplaces about the utter and irredeemable uniqueness of each individual's perception, conclusions whose power to persuade is rooted in ideology, not in research.

The correct application of those commonplaces to connoisseurship is not dogmatic skepticism but rational caution. Prudent connoisseurs freely concede that subtle recognitions of likeness and difference depend upon constant practice where that practice is subject to correction in light of (a) fixed benchmarks, for example undoubted originals and undoubted copies; (b) the judgments of colleagues; and (c) the entire body of external evidence bearing on the cases at hand. This sort of intuitive perceptiveness is to that extent “constructed” by an ongoing process, but the process is subject to painstaking intersubjective criteria. Intuitive perceptions, that is to say, are only credible to the extent that those criteria are met. The life of the conscientious connoisseur is a constant exercise in confirming and correcting her intuitive perceptions at all levels, local and global.

36. Acknowledging uncertainty. Wherever an attribution question is impossible to resolve in a reasonably decisive way, using strictly valid methods and standards, the uncertainty should be acknowledged. A fine balance must be maintained between overeagerness and excessive caution. The first is endemic among proprietors, private and institutional. The second is a subtler temptation for those whose chief claim to fame is critical discernment and who therefore hate to be wrong.

37. Aesthetic quality. The ultimate in intuitive discernment is reached in perceptions of the aesthetic quality of a work, since such intuitions necessarily depend on appreciation of the relevant local and holistic qualities, which are precisely what is aesthetically good about the work. That a work is aesthetically good is simply and solely a matter of the aesthetic goodness of its relevant qualities. Where the value intuition concerns the work as a whole, virtually all of its properties that are meant to be perceived or understood are relevant. Accordingly the connoisseur's study must be searching, unhurried, and renewed after intermissions. On the other hand any aspect of the work that is subject to artistic management, the structural design, the depicted spatiality, atmosphere, light and shadow; the figuration, the color harmonies, the emotional expressiveness, or the presentation of narrative and thematic content, may be the subject of a separate aesthetic evaluation. 34 Similarly parts can be assessed separately from the whole. Since aesthetic quality (of parts, aspects, and the whole) is a prime aspiration in art of all sorts, at all times and in all cultures—nothing is more central to the concept as well as the practice of art 35 —it is bound to feature prominently in connoisseurship. Artists are ranked by the aesthetic quality of their work, even though that is not by any means the only measure of their significance. Aesthetic quality is also the default measure of the work of masters as opposed to that of students or copyists—defeasible though that measure be. The stages of an artist's career must also be understood in part by reference to aesthetic quality (of aspect or whole) gained or lost, for that is the artist's own measure of success or failure.

Naturally skeptics suspect that intuitions of aesthetic quality are incorrigibly subjective. But in fact there is wide agreement within artistic traditions and among those connoisseurs who make the most finely differentiated distinctions within given ranges of works. Few philosophical skeptics make the grade—or even try to. Connoisseurs when challenged sometimes find it convenient to plead nolo contendere and go right on with their useful, indeed indispensable, evaluative practice. In my view the success of their collective practice is the most reliable indication of how much intersubjective warrant aesthetic value-intuitions can acquire. Merely theoretical doubts, or doubts based on unseasoned perceptions, are too circumstantial to carry much weight. 36

38. Autograph worship. Connoisseurship is haunted by the glamor of the true original, the artist's hand, and so on. Consignment of a work to workshop or school status casts it out of the stellar into the lunar sphere or worse, even when in point of artistic quality the work may be as good as or even better than some members of the select circle. While there are good reasons for favoring autograph works in certain respects, the distorting effects of autograph worship should be recognized and resisted. Among these distortions are the tendencies (a) to assign more artistic value to autographs just because they are autographs and (b) to give works autograph status just because they are artistically good enough to be such.

39. The aesthetic rationale of connoisseurship of the eye. Part of the legitimate appeal of connoisseurship of the eye consists in the enhanced cultivation of our aesthetic powers that it promotes. This should be taken as a consolation where the use of these powers for identification in a given case is refuted by scientific evidence. Equally it should be a sufficient reason for continuing to try to discern the artist's hand and personality even in the (extremely) unlikely event that scientific methods entirely supersede connoisseurship of the eye for purposes of attribution. The aesthetic experiences of intuitive connoisseurs, so eloquently and appealingly expressed in the literature, play a central part in the refined enjoyment of art. They deserve respect independently of their objective reliability since they rank among the highest and most essential consummations of the entire artistic enterprise. It seems likely that this high value, taken together with the concentration of energies needed to attain it, explains the hyperbolic claims made by traditional connoisseurs for the authority of their intuitions of authenticity.

40. Interplay of connoisseurship and other art-historical specialisms. As already indicated, connoisseurship in the sense used here is foundational for art history, in that art-historical accounts of style or content or fabrication or training or institutions or social relations or any other art-historical project depend absolutely on knowing what works are authentically of the relevant time, culture, or author. Until one has a fairly firmly established period or cultural oeuvre to deal with, no iconographical or social or any other kind of history of that art can be written. At the same time it is equally evident that connoisseurship draws evidence from the findings of these other specialisms. All contribute in one way or another to authentication projects. The interdependence of all parts of art history should by right override and expunge the spirit of rivalry that has often existed within the field of art history as a whole, connoisseurs in the West pitted against iconographers and, more recently, social historians and those who cultivate “visual studies” against all their forebears; and in China advocates of the textual approach resisting the intrusion of the visual approach. Every specialism has its role to play, none can stand by itself. Reflection on the epistemology of connoisseurship may help to bring home the obvious lesson: collegiality is as axiomatic to method as truth is to aim. 37


1 Silbergeld, Jerome, “Chinese painting studies in the West,” Journal of Asian Studies XLI (November 1987), speaks of “brush-oriented connoisseurship” and “structural connoisseurship” in Chinese art studies. Both are varieties within what in the West is called connoisseurship of the eye .

2 A reason against thinking of any of these specialized practices as sorts or schools of connoisseurship is that none of them deals with all the evidence manifestly relevant to a given authentication. It seems better to regard them as lines of inquiry within connoisseurship.

3 The importance of the roles of various copying practices in the Chinese artistic tradition demands that copies be carefully studied. Thus exhibits such as Seeing double: copies and copying in the arts of China , The Art Museum, Princeton University, 2001, are not just appropriate but necessary. Liu, Cary Y., “Seeing double: copies and copying in the arts of China ,” Orientations 32: 3 (March 2001).

4 However, a form of connoisseurship of the eye would still be highly relevant to art history. See infra, section 38.

5 Panofsky, Erwin, Meaning in the visual arts , Garden City, N.Y 1955, 28ff.

6 An indication of the breadth of the evidence presupposed by connoisseurship of the eye can be gleaned from what is said about the education and training needed by the aspiring connoisseur: general cultural knowledge, extensive knowledge of art history, protracted training in discriminating shades of stylistic difference, observation of and hands-on experience in the making of art, and so forth. For connoisseurship as a whole, to this can be added the knowledge and skills serving the connoisseur's purposes coming from other fields: paleography, epigraphy, archaeology, physics, chemistry, radiology, anthropology, ethnology, history of costume, classics, theology, and on and on. The global reach of the evidence is now widely acknowledged. (e.g., Fu, Marilyn and Shen, Studies in conniosseurship: Chinese paintings from the Arthur M. Sackler Collection in New York and Princeton , Princeton : Princeton University Press 1973, pp. 18f.). Of course it is not supposed that any single person can be the master of all types of evidence.

7 Johnson, Kathryn C., Fakes and forgeries , Minneapolis Museum of Arts, 1973, entry 1.

8 Vasari, Giorgio,. Lives of the most eminent painters, sculptors, and architects , trans. by Mrs. Jonathan Foster (London: George Bell, 1878) iii, 216ff.; also in Lives…, trans. by Gaston DuC. De Vere, (New York: Harry Abrams, 1979) ii, 1031-2 (text), 898 (illustration). A mark had to be made on the back of the copy to indicate which was which for the benefit of those who arranged the deception. Vasari convinced Giulio of his error only by showing him this device. Sarto's copy is discussed in Shearman, John, Andrea del Sarto (Oxford: Clarendon Press, 1965), ii, 265-6 and illustrated in black and white, i, 136a; another discussion and a full color illustration appears in Natali, Antonio, Andrea del Sarto: maestro del “maniera moderna” (Milan, Leonardo Arte srl, 1998), 154-6.

9 Pomfret, John, “The master forger,” The Washington Post, magazine, 1/19/99, pp. 14ff, esp. p. 18. The scholarly treatment of the Riverbank , from which many lessons can be learned, is found in Issues in authenticity in Chinese painting, Judith G. Smith and Wen C. Fong, eds. (New York: Metropolitan Museum of Art, 1999).

10 Documentation so conceived will include what Berenson calls “tradition.” Berenson, Bernard, Study and criticism of Italian art [Second Series] (London: G. Bell and sons, 1931 [1902]), 116-119.

11 Particularly telling in support of the power of documentation is the documentary evidence of ghost painters employed by many famous Chinese masters as well as high court amateurs. See Cahill, James, The painter's practice: How artists lived and worked in traditional China (New York: Columbia University Press, 1994), 136ff. Equally instructive is the corrupting effect of false documentation, which can be corrected only by research uncovering more reliable documents. An outstanding example in recent Chinese art studies concerns “the degree to which standard accounts of Chinese artists are idealized and untrue to their realities,” faults which can be corrected by collecting “letters, diaries, jottings, untypically revealing inscriptions.” ( Ibid. , 5)

12 Berenson's original argument is set forth in Berenson, Bernard, The study and criticism of Italian art [First Series]( London : G. Bell, 1903 [1901]), 46-69 . His confidence in the reality of Amico had waned by 1932 when he reattributed some of Amico's works to Filippino Lippi (Italian pictures of the Renaissance ( London : Phaidon, 1952 [1932]). In 1938 he disowned Amico (“this delightful, if mythical personality”) altogether. He had, he said, “returned to the subject with a better eye, a better method, and greater knowledge…” He proceeds to discuss in detail the reasons relating to the drawings in the hypothetical oeuvre. (Berenson, Bernard, The drawings of the Florentine painters, amplified edition (Chicago: University of Chicago Press, 1938), Appendix VI.) In the meantime Amico di Sandro had been adopted by at least two art historians of note, Van Marle, Raimond, The development of the Italian schools of painting (New York: Hacker Art Books, 1971 [1923-38] ), vol. xii, 245-266; and Neilson, Catherine Bishop, Filippino Lippi (Westport, Conn.: Greenwood Press, 1972 [1938]), 20-37.

13 Such has occasionally happened, as in the identification of Berenson's “Alunno di Domenico” with Bartolomeo di Giovanni. See Brown, David Allen, Berenson and the connoisseurship of Italian painting , (Washington, D.C., National Gallery of Art, 1979), 42,59n123.

14 By “aesthetic competence” I mean the judgment involved in attaining a convincing unity of effect as opposed to a pastiche whose elements do not fully go together.

15 By “aesthetic preferences” I mean to include both deliberate and unwitting aesthetic choices, the latter covering those that arise from the prevailing taste of the painter's period in defiance of his intention to remain true to the period he intends to replicate. These are often cited in exposes of forgeries.

16 Examples may be found in Keith Andrews, The Nazarenes: a brotherhood of German painters in Rome , (Oxford: Clarendon Press, 1964).

17 Most notably in the utter failure of his large wall painting, The Battle of Anghiari , to adhere to the wall in the Palazzo Vecchio, but also in the early deterioration of the Last Supper .

18 The materials of Delacroix's paintings were notoriously unstable because he often disregarded good advice for the sake of immediate gratification: “It was above all a question of whether he liked the grain of the canvas, the nuances of color. If so, any objections you could make to his using them in a work were useless.” (Frederic Villot, quoted by Jobert, Barthélémy, Delacroix , trans. Terry Grabar and Alexandra Bonfante-Warren (Princeton: Princeton University Press, 1998; original edition, Paris: Gallimard, 1997), p. 9.) Joachim Gasquet reports Cézanne saying of the Entry of the Crusaders into Constantinople that he saw it “die, fade, go away.” ( Ibid .)

19 Cahill 1994, op. cit., 124ff. Part of the motivation was to distinguish oneself from artisan painters. Another was to concentrate on certain expressive qualities of brushwork rather than on pictorial verisimilitude

20 Similarly, a level of competence exceeding what is likely for the purported author may testify to production by an anonymous “ghost painter” in the Chinese tradition. Cahill 1994, op. cit., 136.

21 Lo Ch'ing expatiated upon this in the 1991 NEH Seminar on Chinese connoisseurship conducted at the University of Maryland by Jason Kuo. The image is published by Hsung shih mei-shu in Lo Ch'ing, Chuieh-miao hao-hua . Such distinct and continuous wave-patterns seem comparatively rare. An accessible Song example in the “fishnet” pattern is cited by Sherman Lee in his contribution to Issues in authenticity in Chinese painting, Judith G. Smith and Wen C. Fong, eds. (New York: Metropolitan Museum of Art, 1999), p. 82. A Yuan example is illustrated in the same volume, p. 84. In Wen Fong's Images of the mind (Princeton, N.J.: The Art Museum, Princeton University, 1984) one finds on p. 358, a detail from a scroll by Chü Chieh (ca. 1531-1585) with a looser patterning. Xu Yang's The Qianlong emperor's southern inspection tour, scroll 4, The confluence of the Huai and Yellow Rivers , 1770 (Dillon Collection 1984.6 in the Metropolitan Museum of Art in New York ), provides a heroically extended example of wave forms (again, not in the fishnet pattern).

22 Goodman, Nelson, Languages of Art (Indianapolis: Hackett 1968/1976), section III.

23 There are daunting complications in working out precisely what in the history of production defines the autographic status of multiples, e.g., what distinguishes a full-fledged autograph print from one drawn from the original plate without the artist's consent.

24 Some philosophers have argued that one-of-a-kind works are at a deep level merely designs, with the result that an exactly similar copy, like a correct performance of a musical composition, counts as the same work. That is, they have contended that all artworks are ontologically allographic. The “original,” that is, the artist's physical work, serves merely as an illustrative instantiation of the properties of that design. (Currie, Gregory, An ontology of art (London: Macmillan 1989).) Needless to say this view is not likely to appeal to connoisseurs of art.

25 By this standard, an exactly similar design created entirely independently of an existing design would be a different work.

26 “Autographic” is also sometimes used in this sense. In the present context the adjectival usage of “autograph” has the advantage of distinguishing itself from “autographic” in the semiotic sense.

27 Such a view is congenial to a philosophy of art such as that of Collingwood, R. G., Principles of Art ( Oxford: Oxford University Press 1938/1958), in which the uniqueness of expression in genuine art is carried to the limit. For him all so-called art that fails this test is mere “craft.”

28 Morelli (Morelli, Giovanni, Italian painters: critical studies of their works , trans. Constance J. Ffoulkes (London: John Murray, 1892)) has been poorly treated by most commentators, who exaggerate the importance he accords to a narrow class of local features. A careful reading of his texts shows that (a) all local features, whether of the body or anything else (he specifically mentions landscape backgrounds), are to be minutely studied; and (b) this aspect of the work forms only part of the connoisseur's evidence of authorship or school. The holistic qualities discussed below are equally evidential. In my opinion Morelli's appreciation of the total relevant evidence is virtually faultless, given that the scientific tests now available did not exist in his day. The true Morellian method—the one he advocated as well as the one he practiced—is admirably comprehensive.

29 Wollheim, Richard, “Giovanni Morelli and the origins of scientific connoisseurship,” On art and the mind (Cambridge, Mass.: Harvard University Press, 1974), 196.

30 Berenson, Bernard, The study and criticism of Italian art [First Series] (London: G. Bell, 1901/03), 49. The painting may be viewed on the Norton Simon Museum website: . Ironically for Berenson's claim, the work is now attributed to Botticelli and assigned the date of 1468.

31 E.g., by Marilyn and Shen Fu, op. cit, p. 17.

32 An interesting problem arises from a long-established tradition of putative recognition of properties such as “spirit resonance” in a class of especially admired works. (Chang, H. C., 1959. “Inscriptions, stylistic analysis, and traditional judgment in Yuan, Ming, and Ch'ing painting,” Asia Major vii. (1959)). That tradition, conveyed by declarations contained in the surviving literature, does seem sufficient to attest to the works having some property to which the “airy language” ( Ibid., 21) is appropriate. The key questions are, first, is it one and the same property that all the traditional voices are reporting? and second, if so, what property is it? The concurrence of many sensitive connoisseurs suffices to establish artistic quality of eminent degree in the works cited. But that does not identify the holistic property or properties on which the evaluative agreement is grounded. Only a concerted effort to gain a better articulated understanding of the property or properties can create the basis needed for “the traditional judgment contained in such written records… [to] provide the clue to the style of certain masters.” ( Ibid., 2)

33 Freedberg, Sydney, Painting in Italy , 1500 to 1600 (Baltimore: Penguin, 1971), 136.

34 This obviously presupposes that aesthetic quality is not to be construed narrowly, e.g., à la Bell 's significant form ( Bell , Clive, Art ( London : Chatto and Windus, 1914; New York : Frederick A. Stokes Co, 1920).

35 Whether the art be high, low, or somewhere in between.

36 Philosophical skepticism is based more on thought experiments concerning possible worlds than on the realities of actual practice. Connoisseurship need only concern the latter. Empirical skeptics, as we might call them, cite actual disagreements in evaluation, but without due attention to the manifestly imperfect conditions under which they arise. The practical realist about aesthetic value need not blanch at the spectacle of aesthetic disagreement. First, it does not affect gross differences of value, where there is virtual unanimity; second, the residual disagreement concerns subtler distinctions. Here the question is whether more intensive study and negotiation can produce non-coerced agreement among those who are maximally (and consistently) discriminating as to the aesthetic value of parts, aspects, and wholes.

37 Let me here disclaim any intention of advocating connoisseurship (or any other special area) as the main or most essential art-historical field. This canard keeps surfacing when anyone defends the legitimate role of connoisseurship in art history, though a stake should long since have been driven through its heart.