Latest Work

  • Conclusion Conclusion
    VI Conclusion It is time to draw back and look, if not at the whole Hyakusen, at as much of him as we have managed to illuminate in this study. Dark areas remain, and doubtless many distortions, but...
    Read More...

Latest Blog Posts

  • Bedridden Blog
    Bedridden Blog   I am now pretty much confined to bed, and have to recognize this as my future.  It is difficult even to get me out of bed, as happened this morning when they needed to...
    Read More...

This email address is being protected from spambots. You need JavaScript enabled to view it.

CLP 53: 2002 "Riverbank as a Chang Dai-chien Forgery." published in Japanese in Geijutsu Shincho, May 2002.

"Riverbank As a Chang Dai-chien Forgery"

(Written, on request of Hironobu Kohara, to be published in Japanese in Geijutsu Shinchô.)

The late Chang Dai-chien (1899-1983) was surely the most versatile forger of paintings in the whole history of art. The purported dates of his fabrications range from the late sixth century (“Vimalakirti,” with the date 590 in its inscription, Fig. ---) to recent times, and encompass an astonishing diversity of styles, spanning virtually the whole history of Chinese painting. I first became aware of his forgeries of early painting in 1954-56, when I was a Fulbright student in Japan. Chang Dai-chien had brought some of the paintings he owned (including Riverbank) to Kyoto to be reproduced by the publisher Benridô in the fourth volume of a series titled Taifûdô Meiseki devoted to his collection. I got to know Chang well--we both could speak Japanese--and spent a lot of time with him. I had heard stories of his activity as a forger from my teacher Shûjirô Shimada and others, and saw a “Dunhuang” painting of a Bodhisattva made by him, with an eighth century date in its inscription, in the hands of a Tokyo dealer. (Another of Chang’s “Dunhuang” forgeries with a related inscription was later offered to the Freer Gallery for purchase; laboratory analysis of its pigments proved it to be modern, and the “Tang” silk was pronounced by our Japanese mounter Sugiura to be Japanese.) In Hong Kong, on my way back to the U.S., I saw landscapes purportedly by the tenth century masters Dong Yuan and Juran owned by the collector Chen Rentao, and in Paris a “Han Gan” handscroll painting of horses and groom in the Musée Cernuschi (Fig. ---). Returning to the Freer Gallery of Art in Washington, D.C. (where I would become Curator of Chinese Art) I was shown a handscroll titled “Three Worthies of Wu-chung,” associated loosely with Li Gonglin, which had been acquired by the Gallery while I was away (Fig. ---.) Gradually realizing that all these were Chang Dai-chien’s work, I began making a mental list of them, and trying to analyze the physical and stylistic features that identified them as Chang’s forgeries. This was a time when examples were being acquired for high prices by major museums such the Boston Museum of Fine Arts (Fig. ---), the British Museum (Fig. ---), and the Honolulu Academy of Arts (Fig. ---), as well as by private collectors. I did not make my provisional list public, however, until much later, in 1991, when a symposium on Chang Dai-chien was organized by Fu Shen in connection with his exhibition of Chang’s paintings (Challenging the Past: The Paintings of Chang Dai-chien.) I delivered a paper then on Chang’s forgeries, ending by showing why I suspected the so-called Dong Yuan Riverbank (Fig. 1) to be one of them.

I felt uncomfortable doing this, because the painting was by then owned by my good friend and sometime teacher Wang Chi-ch’ien. I have always had the greatest respect for Wang’s judgements of paintings, and have written about him as one of the two leading practitioners today of traditional Chinese connoisseurship. But I have also come to realize, over many years of watching, that the strengths of that tradition are in judging Yuan and later painting, in which brushwork and individual style are prominent; for Sung and earlier periods, when this is not the case, the methods of traditional Chinese connoisseurs are less effective, I believe, and they can make bad mistakes. Wang Chi-ch’ien and other Chinese connoisseurs feel confident that they can detect Chang’s forgeries of later artists such as Bada Shanren and Shitao, and the best of them usually can. But Chang knew very well the weaknesses of traditional connoisseurs in the early periods, and exploited them in making his “early” forgeries, of which Riverbank is the most successful.

I was not the first to question publicly the authenticity of Riverbank: Hironobu Kohara had already done so in print, in Chûgoku Nansôga nôto 6, 1977, which accompanied a volume of the series Kohara edited, Bunjinga Suihen, China vol. 2, the volume in which Richard Barnhart included Riverbank as a great early work. Kohara, rightly uneasy that his readers would think he shared this view, pointed out the stylistic anomalies that indicated the painting could not be as old as the Song period.

However, in spite of the serious doubts expressed about Riverbank by Kohara, myself, and others, our colleague Wen Fong included it in a group purchase by his brother-in-law Oscar Tang from Wang Chi-ch’ien, eleven paintings intended for gift to the Metropolitan Museum of Art, where Wen Fong chaired the Department of Asian Art. The acquisition was hailed in a front-page article in the New York Times (May 19, 1997), with a picture of Riverbank and the claim that it was “the earliest of the three rarest and most important early monumental landscape paintings in the world” (the other two being the well-known masterworks by Fan Kuan and Guo Xi, both in the National Palace Museum, Taipei.) Riverbank was called (quoting Wang Chi-ch’ien) the “Mona Lisa” of Chinese painting.

Some weeks later, a short article appeared in The New Yorker by freelance journalist Carl Nagin, who for years had followed Chang’s career as faker, quoting me as believing Riverbank to be a forgery by Chang Dai-chien. This trouched off an explosion of controversy, with articles on it appearing in many newspapers and magazines. An angry response appeared in Orientations magazine written by Richard Barnhart of Yale University, a former student of Wen Fong who had published the painting several times and is a passionate defender of it. Nagin in turn wrote a response to Barnhart, also for Orientations, quoting, among others, Sherman Lee, retired director of the Cleveland Museum, saying that “the Met made a big mistake” and that if he were quizzed by defenders of the work, he would reply that “any idiot can see that this is a fake, and if you can’t see it, I can’t help you.” The rhetoric on both sides was becoming more and more heated.

At last, on December 11, 1999, a one-day symposium was held at the Metropolitan Museum of Art, titled “Issues of Authenticity in Chinese Painting.” The papers were simultaneously published as a book under the same title. The auditorium was packed, with much of the Chinese art community present along with reporters, collectors, and many fascinated laymen. There were ten participants delivering papers and two discussants, Apart from Kohara, Sherman Lee, and myself, and two Chinese elder scholars, all of them (except Fong himself) were former Princeton students who had worked with Wen Fong--or, in one case, with Barnhart. Barnhart also spoke, briefly and again angrily. It was widely recognized, then, that the symposium was decidedly “stacked” in favor of supporters of Riverbank. Kohara was present but unwell, and a brief summary of his paper was read by Wen Fong. It dealt with the recent history of Riverbank and the fabrication of a spurious provenance for it, as Kohara reconstructs it, especially the part played in this by the painter Xu Beihong, who colluded with Chang in the falsifying operation. Sherman Lee, also unwell, but who had spent some hours over two days studying the painting carefully in the gallery before forming a firm opinion of it, read a brief paper in which he pointed out that representationally it is a mess. He began with the pattern on the water: “It is not the shui [of shan-shui] observed in early works; only a modern could fail to see the varying tension when observing water in nature.” As for the landscape masses: “The closer we look at the details, the vaguer and more insubstantial the forms and shapes become.” And Lee concluded: “The result is a morass of starts, false starts, and half starts that point inexorably to a modern pastiche all too familiar to many of us [he refers, of course, to the well-recognized forgeries by Chang Dai-chien] and unworthy of serious consideration by our serious colleagues.”

My own paper was titled “The Case Against Riverbank: An Indictment in Fourteen Counts,” and presented fourteen reasons that indicated--or collectively, in my view, proved--that (a) the painting could not have been produced in any early period, and that (b) it was a forgery by Chang Dai-chien. These included close comparisons with Chang’s paintings done in the late 1940s-50s, either under his own name or as forgeries, in which virtually every telling feature of Riverbank can be matched; and comparisons with accepted works from early Song and pre-Song periods, in which none of these distinctive features can be found. Chang was careful to see that Riverbank did not resemble, superficially, either his own signed works or his better-recognized forgeries--a common, naive objection to ascribing Riverbank to him is to say “But it doesn’t look like Chang’s works!” But this seeming unlikeness overlies deeper structural similarities (cf. for instance Figs. 2 and 3, in which the build-up of the mountainside out of large units and the use of tree groups can be seen to be closely paralleled in a signed work painted by Chang in 1957). There is, by contrast, no style-period in the evolution of Chinese landscape into which Riverbank fits; no reliably early painting resembles it significantly. As Kohara observed, “a no-time painting [i.e. one that cannot be fitted into any period style] means a contemporary piece,” and he adds that the only recent artist who could have painted it is Chang Dai-chien.

Riverbank, moreover, is full of representational mistakes and anomalies: although Chang Dai-chien was technically skilled for his time, he could not organize his forms convincingly in space as the old masters could, or render them as coherent, readable volumes, or paint a clump of trees or bushes that can be clearly read as separate plants, as any tenth century master could and would have done. The mountaintop in upper left (Fig. 4) simply disappears--not into mist (which would be impossible anyway in this early period) or because of damage; Chang simply did not paint it in. This odd feature is seen in others of his forgeries. The flight of geese in the distant valley is headed into a mountainside, identified as that by blurry groves of trees; the only opening into farther space is far above them to the right. And, as many viewers quickly notice, the river flowing out of distance turns below into a road with people walking on it--the visual continuity is unmistakeble. This kind of confusion is also common in Chang’s paintings, but unthinkable in a truly early work.

Another of my “counts” against Riverbank was directed at the soft, “brushless” rendering of the earth masses, with the ink rubbed on instead of applied in strokes, so that the forms are blurred and merge confusingly (see Fig. 5); this is in contrast to the rendering of earth forms in genuine tenth century paintings, such as the one excavated from a Liao tomb (Fig. 6), in which the application of ink in loosely-applied streaks shapes the masses strongly. Other counts pointed out the unnaturally strong light-and-shadow treatment of the land forms, which gives the painting a striking, dramatic effect impossible for early periods; and a compositional formula in which the separate component masses that make up the mountainside are animated and muscular, with diagonal thrusts balanced by counter-thrusts for a highly dynamic assemblage, contributing also to the dramatic effect of the whole. This way of constructing a landscape was not used before the time of Dong Qichang, one of whose paintings I illustrated for comparison (Fig. 7). That painting and Riverbank have more in common than either has with any truly early Chinese landscape.

Still others of my “counts” pointed out that the collectors’ seals on Riverbank do not match those on more acceptable works; that no secure reference to it can be found in any old catalogue or other text; and that the “signature” on it is highly suspect for a number of reasons. All these lay a false trail, as do the sets of spurious seals and inscriptions and false correspondences with recorded paintings planted by Chang on others of his “early” forgeries, for later owners and researchers to follow and be duped.

My last “count” dealt with the spurious provenance that Chang constructed for the painting, with help from his friend the painter Xu Beihong. Other noted collectors and connoisseurs are known to have helped Chang in the same way for others of his forgeries, writing colophons filled with misinformation on paintings they must have known were fakes. Why they did this is a question; that they did it is beyond doubt. A crucial document for Riverbank’s supporters is a letter written by Xu Beihong in which he relates how he purchased the painting in Guangxi province in 1938 but allowed Chang to borrow it, and eventually to acquire it himself (Fig. 8). Kohara believes that this letter is itself a forgery, not written by Xu; I myself, while respecting Kohara’s view, am more inclined to see it as written by Xu at Chang’s request, to “document” events that in fact never happened. In either case, the letter is itself highly problematic. Believers claim it was addressed to a certain Mr. or Miss Sun (probably Sun Duoci, a woman) and was in a Japanese private collection; in fact it has for years been in the hands of a Taipei dealer, who acquired it from Xu Beihong’s divorced first wife Jiang Biwei. The addressee, moreover, is now believed to be Chang Dai-chien’s brother Zhang Shanzi. The letter has now become an embarrassment for the believers in the painting, who carefully avoid referring to it. Yet there is no other evidence that Xu Beihong acquired the work in 1938, or that it existed then at all.

Another very shaky episode in the fabricated account of Riverbank’s recent history is when Xu Beihong allows Chang Dai-chien to acquire the painting, which he has already had in his possession for many years (without showing to anyone else, much less publishing it.) He gives it to Chang, according to the story, in exchange for a painting owned by Chang that Xu Beihong enthusiastically admired, Jin Nong’s Returning by Boat in a Rainstorm (Fig. 9). Kohara believes that painting to be itself a forgery by Chang; I am not sure of that, although he may well be right. In any case, it is a modest work, its value only a tiny fraction of what Riverbank would be worth if it were genuine. Believers in the story see this strange exchange as exemplifying Xu Beihong’s unworldliness and disregard for monetary considerations. In 1950 Xu wrote a long inscription to accompany the Jin Nong painting (Fig. 10), relating once more (quite irrelevantly) the whole story of how he had acquired Riverbank and let Chang take it away, trying again to “document” the false history that he and Chang had constructed for the painting.

All this, and other inventions that make up the “provenance” of Riverbank, are intended to explain how the painting could appear suddenly, previously unpublished and unknown, in the mid-1950s, when Chang Dai-chien reproduced it in the Benridô volume and his close friend Xie Zhiliu published it in his Tang Wudai Song Yuan mingji (1957) with high praise and claims for its authenticity as a work by Dong Yuan. This whole account is so full of inconsistencies and obvious untruths that, far from supporting the antiquity of the painting, it adds heavily to all the other evidence against it. (Anyone wishing to read more on this matter of the fabricated provenance should read my article and Kohara’s in Issues of Authenticity, or the Chinese translation of mine published in the Taipei journal Tang-tai (Contemporary), no. 152, April 2000.) None of these problems of provenance, however, were raised in the symposium by anyone except Kohara and myself, nor were any of our objections answered, or any defense of it offered. We were, as I wrote him later, “stone-walled”--when any questions about it were raised by members of the audience, the supporters simply looked the other way.

A paper in the symposium by Maxwell K. Hearn, student of Wen Fong and Curator of Asian Art at the Metropolitan, was titled “A Comparative Physical Analysis of Riverbank and Two Zhang Daqian Forgeries.” A generally recognized example of Chang’s forgery, a landscape attributed to Dong Yuan’s follower Juran in the British Museum (Fig. ---), was the principal work used for the comparison, which was aimed at “proving” that Riverbank could not be another forgery by Chang, but must be genuinely early. Hearn’s paper was, needless to say, hailed by supporters of Riverbank as a final refutation of the arguments of the non-believers. I countered this claim, in my paper as delivered (not as published), by quoting from a paper titled “The Problem of Forgeries in Chinese Painting” published in 1962 by none other than Wen Fong himself. In this he related how in the mid-1950s a work now recognized by nearly everyone as a Chang Dai-chien forgery, the “Dunhuang” painting of a Bodhisattva with an eighth century date in its inscription that had been acquired by a Tokyo art dealer, was put through technical analysis by laboratory specialists in Tokyo--and passed their inspection with great success. “From the physical standpoint,” Fong writes, “the forgery was almost perfect . . . microsopically and chemically, it was thought that everything looked as one might expect of a handsome specimen of Tang workmanship. There were, in fact, plans afoot to publish the findings, as a standard textbook on technical analysis of a Tang painting.” Scientific evidence, he concludes from this, can never suffice in itself to prove authenticity, but must be interpreted in the light of stylistic and other evidence. Nearly forty years later, he was to write that both physical and stylistic analysis prove that Riverbank “cannot be a work of the 20th century nor the creation of the renowned modern forger . . . Zhang Daqian.” I commented that I would invoke the 1962 Wen Fong against the 1999 one and continue to argue that the painting can be 20th century and by Chang Dai-chien.

The British Museum’s “Juran” painting, borrowed for the occasion, was hung beside Riverbank in the gallery, in the hope that people seeing the two together would be persuaded by their dissimilarity that they could not both be forgeries by Chang. But, as I pointed out, Chang’s forgeries are carefully painted and “aged” in such a way that they will not look like each other. It would be like hanging the Dutch forger Van Meegeren’s Disciples at Emmaus, the masterwork among his forgeries of Vermeer, painstakingly painted over a period of seven months and carefully aged, beside one of the sloppier forgeries he made later: they would not look alike, but both would be Van Meegeren forgeries.

The conclusion of all this seems inescapable: Riverbank cannot be a genuinely old painting, and should not be allowed to muddle the histories of early Chinese landscape painting that we try to construct. If a European would-be old master painting were to turn up suddenly in the hands of a known master forger, and the recent history claimed for it proved to be full of holes, there would scarcely remain any room for argument, even if the painting were stylistically convincing for that period and master (as Riverbank certainly is not.) By some curious double standard, a Chinese would-be old master painting is not held to the same criteria, and apparently can escape--for now--the judgement that its European counterpart would receive.

A number of leading connoisseurs and museum specialists from China were invited to the symposium, and as a group seemed convinced by the painting and the arguments made for it, accepting it as at worst an early work. Wang Chi-ch’ien himself was there, and spoke (through his daughter) about his faith in the painting and its importance. An unfortunate aspect of the controversy is that it seemed to pit Chinese scholars, who are uncomfortable with having the strengths of their great tradition of connoisseurship called into question by outsiders, against foreign specialists, who mostly have deep doubts about Riverbank (although many of them, for various reasons, decline to make their doubts public.) Discomfort has been expressed also over the way the controversy divided our field of study into two camps, with a degree of hostility on each side. There are those who now have adopted an “in between” position, wanting to see Riverbank as some kind of old painting without being able to make a convincing argument for any particular dating or art-historical placement. Many others recognize it as a modern pastiche, but are not familiar enough with Chang Dai-chien’s styles to see it as his work. The controversy will not be over soon. But, as has often been noted, forgeries have only a limited life span, and what is accepted by leading experts one year looks obviously wrong to any good graduate student twenty or thirty years later. It will be that way with Riverbank: we have only to wait.

James Cahill, Honolulu, February 2002

Latest Work

  • Conclusion Conclusion
    VI Conclusion It is time to draw back and look, if not at the whole Hyakusen, at as much of him as we have managed to illuminate in this study. Dark areas remain, and doubtless many distortions, but...
    Read More...

Latest Blog Posts

  • Bedridden Blog
    Bedridden Blog   I am now pretty much confined to bed, and have to recognize this as my future.  It is difficult even to get me out of bed, as happened this morning when they needed to...
    Read More...