Artwatch UK

Posts tagged “Bernard Berenson

Connoisseurship: Examinations, Debates and Snap Visual Responses

23 November 2016

The rising tide of old master “sleepers” and “discoveries” carries great dangers and demands snap judgements. Some candidates for upgrades intrigue, some look dubious, some scream “Fake!” Last week two cases caught our eye.


Fig. 1, above. Fig. 2, below. The newly discovered “Lost” Van Gogh sketchbook (above, top) rang our and many other fake bells. Then came a report that a small “Florentine School” painting on a €3-4k estimate fetched €375k at auction as a sleeping Filippino Lippi (above, Fig. 1). A link to another small work attributed to the artist at the North Carolina Museum of Art (Fig. 2, below) showed pronounced, seemingly reassuring correspondences, but something jarred.


On connoisseurship we hold that every claimant work should be rigorously “interrogated” in three crucial respects. Technically, in its physical composition; by documentation on its known or claimed histories (provenance); and, above all, by visual analysis because, in the visual arts, every picture is its own prime historical document and its inbuilt historically-generated artistic relationships constitute the primary subject of art critical appraisal and evaluation.

Failure to excise bad attributions deceives the public and corrupts oeuvres. A good picture has nothing to fear from challenges. No amount of scrutiny constitutes a threat as good pictures outlive their doubters and can fight another day. Argument is healthy and a successfully repulsed challenge can increase understanding and enhance a good picture’s lustre.

Vincent van Gogh: The Lost Arles Sketchbook


Above, Fig. 3. The controversial lost-but-now-found Van Gogh sketch book above is by Professor Bogomila Welsh-Ovcharov, a Van Gogh authority whose claims of authenticity are supported by Ronald Pickvance, author of Van Gogh in Arles, but the cover’s supposed Van Gogh ink self-portrait announces itself as a draughtsman’s pastiche, as Mark Hudson noted in the Daily Telegraph (“A romantic story but can it really live up to its promise?”, 16 November 2016).


Above, Fig. 4. Simply by placing the supposed Van Gogh self-portrait next to an autograph portrait, immense and glaring differences become apparent: the author of the “discovered” drawing has abandoned symmetry with eyes of radically different sizes and a nose that seems product of a car crash. Throughout, the author mimics Van Gogh’s pen marks without comprehension of his form, power of design, and psychological acuity.


Above, Fig. 5. Instead of a form-camouflaging jumble of marks, the bona fide Van Gogh disports five graphically discrete component parts: a light-coloured jacket; a dark shirt and scarf; a varied but, on aggregate, mid-toned face; a light-toned hat with some mid and dark-toned form articulating shading; and, throwing all other values into relief, an agitated but tonally cohering background. Each of these spheres is allotted its own graphically purposive notations. The four images we show above for comparison are easily found online in historically successive reproductions. While these reproductions vary considerably, the force and artistic coherence of Van Gogh’s graphic intent and method shines through all.


Above, Fig. 6. If we place the bona fide Van Gogh between the clumsy mimicking newcomer and a masterpiece of the greatest graphic brilliance – van Dyck’s etched self-portrait – it is clear that the Van Gogh has more kinship with the latter than with the former.


Above, Fig. 7. And if we compare the Van Gogh with an entirely autograph van Dyck etched state of a figure we find a common use of a toned background that throws both subjects’ flickeringly brilliant lights and darks into relief.

The Van Gogh Museum’s objections to the “Lost Arles Sketchbook” and its track record on Van Gogh attributions


Above, Fig. 8. Prof. Welsh-Ovcharov (top) has responded to the Van Gogh Museum’s dismissal of the drawings with a rebuttal and a challenge to debate – thereby showing conviction and good faith. Her publisher reportedly characterises the proposed debate as “an opportunity to shed light on the conditions under which the Van Gogh Museum is claiming the de facto right to a monopoly of attribution.” This is a common plaint against authorities that block would-be, high-value attributions but our impression of the museum’s judgements is favourable.

In 2006 a Van Gogh – The “Head of a Man” owned by the National Gallery of Victoria in Australia (above left, on an easel) – was challenged by the art historian and Sunday Times art critic Frank Whitford when the portrait was loaned to an exhibition in Edinburgh. The newspaper asked our opinion and, when we demurred, sent a (revelatory) high resolution full colour compilation of all of Van Gogh’s painted portraits. We supported Whitford, saying to the newspaper (as reported in ArtWatch UK Journal, Spring 2008):

“The specific warning signs that should have alerted the buyer are:

“1] It is unique in Van Gogh’s portrait oeuvre

“2] It does not fit in the stylistic chronology that exists within that oeuvre. Compare it for example with the brushwork, colours and ‘attack’ of the Old Man with Beard, painted the year before that is in the Van Gogh Museum, and the Portrait of Camille Roulin painted a year or so later and that is also in the museum. There is an enormous but clear and logical development between those two pictures, from thick, laboured, relatively coarse brushwork to much more refined and ‘decorative’ marks – but both are entirely consistent and ‘all-over’ in their treatment.

“3] If its provenance goes back no further than Germany in the late 20s or early 30s, that is particularly unfortunate. Germany at the time was notorious for the certification by scholars (for a fee or sales cut) of dud works. The dealer René Gimpel has referred to the scandalous ‘amounts obtained by means of certificates given daily by German experts to German dealers. Just as there were paper marks, so there are paper canvases, an easy way of bringing dollars into Germany…The German title of Doktor impresses the Americans. The museums are even more intent than the collectors on defending their fakes or their mistaken attributions….’ By coincidence, in the current ArtWatch Journal [No 21], Kasia Pisarek cites the case of the great Rubens scholar Ludwig Burchard who issued so many optimistic certificates that he was unable ever to write his definitive book on the artist…She has identified over 60 Burchard attributions that have subsequently fallen. It was Burchard who first upgraded to Rubens the Samson and Delilah that is now in the National Gallery.

“I would add that the fact that it seems to be admitted that it is a cut-down canvas that was glued onto a panel compounds suspicions… Why should a (presumably) then only forty years old canvas, have needed gluing onto a secondary support? It might be worth asking the Gallery curators if any scholar has questioned the picture publicly or privately.

“It may be coincidence, but two of the pictures that ArtWatch has challenged in our own National Gallery, the Rubens Samson and Delilah and the Raphael Madonna of the Pinks, no longer retain their original backs. The former was planed down to 2 or 3mm thickness and glued onto a sheet of blockboard; with the latter, the family of restorers who sold the picture in the 19th century had (most unusually) polished the back of the panel thereby removing all historical evidence.”

As we have seen more recently, the claimed lost Leonardo drawing “La Bella Principessa” that emerged anonymously in 1998 had been glued to an old oak panel. Gluing canvases or drawings onto boards conceals half the material evidence. On 3 August 2007 Andrew Bolt reported in the Australian Herald Sun:

“The curious thing about the National Gallery of Victoria’s fake van Gogh is how easily it was spotted as phoney once it went on tour…. For more than 60 years this painting hung in the NGV without anyone screaming ‘Fake!’ True, a few experts now say they had their doubts, but it was only when the NGV proudly loaned its ‘van Gogh’ to Scotland’s Dean Gallery last year that the painting was denounced. Three British critics took one look at it and snorted… Even then, there were some back in Melbourne who couldn’t accept the evidence of their own eyes, as ABC Television’s 7.30 Report found:

“‘Two of the critics include Michael Daley from ArtWatch UK and Times art critic Frank Whitford. But Robyn Sloggett [an art authentication expert at Melbourne University] has questioned their expertise. ROBYN SLOGGETT: I don’t think either of them are Van Gogh experts, certainly not known to be such…[Director of the National Gallery of Victoria] DR GERARD VAUGHAN: It is a slightly offbeat picture. It doesn’t fit into the natural progression of Van Gogh’s work at that time because it was a moment in late ’86 and into early 1887 where he was experimenting with two or three different styles. In many ways, this is slipping back into his earlier realist style of the mid 1880s where he concentrates and uses these earthier ochre colours. It is a transitional picture.’”

“Conceived at special moments” and “sometimes repeating, sometimes anticipating themselves” are commonplace apologias for disqualifying incongruities in upgrades. In 1997 and 2000 the National Gallery claimed its Rubens Samson and Delilah did not look like any other Rubens in the gallery because it was “the only work in this collection typical of the artist when he had returned from Italy in 1608”. In truth the painting was unlike the (secure) one that immediately preceded it and unlike the (secure) one that immediately followed. If a Rubens, it would be the only one on which he employed flat brushes and painted finger tips with rectangular highlights. During ABC Television’s 7 August 2006 programme (“NGV’s Van Gogh Labelled a fake”), James Mollison, a former NGV director said: “This picture has been doubted by people very often.”

The upshot of the controversy was that the NGV director announced that such was the gallery’s confidence that the painting would be submitted (voluntarily) to full technical examination at the Van Gogh Museum. A year later the Herald Sun reported the attribution’s demise at the Van Gogh Museum:

“The Dutch team used X-radiograph, digital photographs, light microscopy and paint and thread analysis. Among conclusions were: THE work’s ground layer of white paint is not found in Van Gogh’s Antwerp and Paris works. ITS use of pure ochre is not found in other Van Gogh work. THE portrait shows just the top of the man’s shoulders. Van Gogh usually showed more of the clothes. “A COMBINATION of a fairly coarse and detailed painting style”, with more detail in the mouth, eyes, skin and beard than Van Gogh used. NO reference to the portrait or the sitter in Van Gogh’s extensive letters. The experts also noted no record of the work could be found before 1928, when it appeared at Berlin’s Galerie Goldschmidt and Co.”

The Rubens Samson and Delilah emerged in Germany the following year at Van Diemen and Benedict where it was offered as a Honthorst before being upgraded by Ludwig Burchard. Previously it had been attributed to Jan van den Hoecke, a follower of Rubens. Burchard had recently upgraded the supposed Rubens ink sketch design for the Samson and Delilah (see Art’s Toxic Assets and a Crisis of Connoisseurship ~ Part II: Paper – sometimes photographic – Fakes and the Demise of the Educated Eye ).

The Newly Upgraded Filippino Lippi


Above, Fig. 9. At first glance the awakened “Filippino Lippi” (above right) seems more plausible than the new Van Gogh drawings – especially when linked to a work attributed to the painter at the North Carolina Museum of Art (above left). In terms of palette, condition and design the two seem as peas in a pod but this closely related pair triggers no recollections of anything similar in the artist’s oeuvre. If their strikingly common format suggests original incorporation in a larger work, disjunctive variations in their parapet walls and stone inscription tablets dispels the possibility. Most inexplicably of all, the new upgrade is incongruously modernist in its emphatic planar and ‘on-the-picture-surface’ geometrical vocabulary.


Above, Fig. 10. In 1901 the painting of Saints Uraldus and Fridianus was sold (not as a Philippino Lippi but as a Masaccio) to an English aristocrat, the Earl of Ashburnham. As with the recently proposed Haddo House Raphael (Fig. 10 above), there is little on the panel’s back other than a label in English for an exhibition of “Early Italian Art” (Fig. 10 top). For the Carolina Saint Donatus, the museum offers only a date – “circa 1490” – and the identity of the picture’s donor, the Samuel H. Kress Foundation. Such lacunae are perplexing because Filippino Lippi is a well-chronicled artist whose securely attributed works might easily be brought into direct comparison with the two more recent attributions.


Above, Fig. 11. The backs of attribution upgrades often prove problematic, and none was more so than the small panel “discovered” as a Duccio Madonna and Child (above right) in 1904 after having been bought in an antiques shop in Italy. It was then rarely seen until bought with fanfare (but no technical examinations) for $50m in 2004 by the Metropolitan Museum of Art, New York, in a blind “treaty” sale conducted by an auction house among a few leading museums shortly after the picture was withdrawn from an imminent comparative exhibition of Duccio and his followers that would have introduced the painting to many scholars for the first time and in an instructive context. That withdrawal – despite the painting’s inclusion in the catalogue – might have been made out of fear of repeating the demise of the owners’ second Duccio, as described below.

The back of the tiny picture had been cradled with no fewer than eight mahogany bars and when these were later removed at the Met. a hand-written ascription to Duccio’s pupil “Segna” was found on the bare poplar wood from which painted work had been stripped. The Met Duccio contained modern wire nails, a fact not acknowledged in the museum’s post-purchase technical examination reports. When we asked after the antiquity of the nails the museum claimed they had been inserted as repairs after the panel had been cradled in the 1930s. As “proof” of that unsupported chronology it was said that one of the nails had entered one of the mahogany bars. However, as we pointed out, the head of that particular nail had been visible on the front of the frame throughout all of picture’s photographically recorded history and, while some nail heads were visible most were not and therefore had been applied before the (now heavily distressed) frame was gessoed and gilded. Thus, the panel arrived in the world at the beginning of the 20th century with modern nails intersecting a cradle that concealed an awkward ascription on a stripped-down back.


Above, Fig 12. A face painted in the 14th century by a follower of Duccio (identified by Pèleo Bacci as Segna) is shown above to the right of the face of the Met’s Duccio as it was before restorations established the blue of the Virgin’s mantle to be azurite, not the requisite ultramarine. No one ever suggested that the painting on the right was by Duccio and no one judged the Met’s picture an autograph Duccio before Bernard Berenson’s wife (Mary) in 1904 with the support of Berenson’s protégé Frederic Mason Perkins. An earlier suggestion had been that it was a work of Sano di Pietro, as Frances Vieta discovered in researches at the Frick Library, New York, that were kindly made available to us.

In 1933 Perkins attributed a second Madonna and Child to Duccio and persuaded the then owner of the Met.’s Ducio (the Belgian collector, Adolphe Stoclet) to buy it. In 1989 that Duccio was loaned to the Cleveland Museum of Art and was there identified on technical examination by Gianni Mazzoni as a fake by Icilio Federico Joni who ran a forgery factory fronted by middlemen, one of whom was Berenson’s protégé Frederic Mason Perkins.


Above, Fig. 13. A cult of Supreme Art Historical Importance was activated around the Met. Duccio and part of this mythology rested on the picture’s supposedly miraculously well-preserved, little-restored, condition. Comparison of the photographs above showing its present state (right) and an earlier state (left) discloses how extensively the work has been repainted – note the altered design of the dominant eye, and the extensive reworking of the veil. The potentially falsifying nature of restorations when determining attributions remains a conspicuously under-examined area – as does the extent and nature of repainting on stripped-down “sleepers”. (But see “A restorer’s aim – The fine line between retouching and forgery”. For a fuller account of the Met.’s Duccio difficulties, see Michael Daley: “Buyer Beware”; “Good Buy Duccio?” and “Toxic Attributions?” in the Jackdaw magazine issues of Nov/Dec. 2008, Jan/Feb. 2009, and March/April 2009.)

The Newly Upgraded Filippino Lippi (continued)


Above, Fig. 14. With the two small “Filippino Lippi” pictures at Fig. 9 and below, top, said to have been painted between 1490 to 1494, precise stylistic comparisons can be made with securely attributed works in the oeuvre. What is believed to be Filippino Lippi’s self-portrait above was executed by the artist in 1481-1482 in the Brancacci Chapel, Santa Maria del Carmine, Florence.


Above, Fig. 15. Filippino Lippi’s Apparition of the Virgin to St Bernard of 1480-86 was said by Bernard Berenson (in his 1938 revised Drawings of the Florentine Painters) to comprise “Filippino’s masterpiece, the last picture in which he is still a pure Quattrocentist, in which there is no sense of the Baroque.” Is it conceivable that some years later this artist painted the two small pictures shown here above the Apparition? Berenson reports that Filippino went on to betray excesses, not to purge and severely abstract his pictorial vocabulary: “Filippino’s Baroque, however had little in common with the qualities of the genuine [Baroque] style, and much with its worst vices. These were the sins of extravagance, of wantonness – the vulgarity of the newly enriched, who feel life is enhanced by the mere act of showy spending.”


Above, Fig. 16. At the top we see how Filippino Lippi painted books before 1486 in his Apparition of the Virgin to St Bernard and before his lapse into Baroque excesses. In this secure work we not only see great technical accomplishment but a fascination with the very means by which books were stitched together in assembled folded sections. Is it conceivable that after this tour de force celebration of the book binder’s craft skills Filippino should have been satisfied with the out-of-perspective simplifications of books in the Saint Uraldus? While the opened book has been painted in utter ignorance of book binding methods, the ochre coloured book at the bottom left of the pile has managed to anticipate to a remarkable degree the appearance of a neat modern machine-bound book.


Above, Fig. 17. Again, does the chasm of technique and sophistication in this further comparison from the Apparition and the Saint Uraldus not strain credulity at claims of a common author?


Above, Fig. 18. By 1493-94 the artist had completed his Madonna and Child with St Catherine of Alexandria and St Martin of Tours as above and we see precisely the over-elaboration Berenson castigated as Filippino’s squandering “like a nabob with a heady disorderliness all the decorative motives which the heritage of antiquity, the hard earnings of his precursors, and his own fancy had put into his hands.”


Above, Fig. 19. How conceivable is it that Filippino might, at the same short period, have made two so diverse treatments of a man parting drapery with an advancing left arm as in these two paintings? In the one the “Blanket-like drapery dear to Filippino” (in Berenson’s term), curves, twists and folds naturally across the body, while in the other it moves as if fabricated by a former sheet metal worker with little regard for any underlying body, or even for the means by which the glimpsed parts of the (wildly varying) decorative border of the cope might ever have been united as woven material. Why the arbitrary, asymmetrical placement of indeterminate embroidered decorations on the cope’s border? What holds the cope together? Is it a giant garnet or ruby, or a small tambourine? Where else in Filippino might we encounter such flattening abstractions and lax indeterminacy of depiction?


Above, Fig. 20. If the logic and treatment of Saint Donatus’ cope border (above, left) seems plausible and suitably understated, what might have carried the same artist to the Byzantine and conceptually irresolvable twin conundrums of the cope border as encountered on Saint Uraldus (above right)? What accounts for the very different depictions of the inscribed tablets on the parapet wall? If that of Saint Donatus is somewhat overly monumental and set uncomfortably close to the top of the parapet, at least it is sculpturally resolved and satisfactorily symmetrical along its horizontal axis with its twin decorative “butterfly wings” termini. Why, then, would the twinned tablets of Saints Uraldus and Fridianus meet in the middle with single butterfly-wing termini while leaving blank endings at the outer edges of the picture’s composition? Why are these two inscribed tablets skimped and devoid of projection when the saints above are greatly more dynamic and humanly engaged – almost as if in anticipation of Raphael’s later depicted dialogue between Aristotle and Plato?

Below, Fig. 21. What theological reading of Saint Uraldus’ life prompted the vast frilled neck lizard-like display of the cope’s pink lining below? If intentionally “Baroque” in its explosive ostentation and theatrical impact, why, then, its implausible combining with a geometric severity of draperies that are more snapped than folded?


With such bizarrely anomalous visual constructs, might it not be prudent to consider the waking “Filippino Lippi” sleeper as a possible product of the late 19th and early 20th century Italian forgeries boom that was tailor-made for British and American collectors? We know that many skilful artists were employed in that trade because when the Italian Government proposed stringent export taxes in 1903 to stem the country’s out-flow of art treasures, the Florentine art dealers association petitioned that the new laws would throttle the large and thriving trade in forged art and antiquities for foreign collectors. (Where did all those often excellent works go?)

At the December 2015 ArtWatch UK/LSE Law/NY Center for Art Law conference Art, Law and Crises of Connoisseurship (the proceedings of which will be published shortly), Professor Charles Hope pointed out how effectively 20th century scholarship had winnowed previously overblown numbers of Titians, Raphaels and such. Markets are good things and the London art market has long been a very good (much envied in Europe) force for Britain, but there are developing dangers. If perceptions were to grow that previously downgraded works are being systematically rehabilitated through “sleeper-discovery” mechanisms at a time when leading houses are fighting to the death for pole position on market share, confidence in the lots on offer might evaporate. Already, certain external structural changes are weakening the London market’s traditional and much-valued symbiotic relationship with disinterested scholarship. Increasing litigation by owners against dissenting independent scholars suppresses debate and frank expert appraisal. In a paper at our conference (“Throwing the baby out with the bath water – the Demise of Connoisseurship since the 1980s”), Brian Allen, former director of studies at the Mellon Centre, warned that recent changes of philosophy and views on connoisseurship in the academic world are greatly reducing the traditionally available body of disinterested academic expertise that counterbalances purely commercial interests:

“In my own field of British art the number of so-called ‘experts’ has now diminished alarmingly as the older generation dies off not to be replaced. It seems extraordinary to me that major artists such as Stubbs, Wright of Derby and Sir Thomas Lawrence, to name but three, don’t have an acknowledged expert to whom one can turn for a reasonably reliable, independent opinion. And this has also certainly happened in other specialist fields… Younger scholars nowadays, especially those in the universities, have almost no contact whatsoever with the art trade compared to fifty years ago. Yet for many years it was perfectly possible for the two worlds to co-exist harmoniously.”

Michael Daley 23 November 2016

Problems with “La Bella Principessa” – Part II: Authentication Crisis

In Part I we discussed the look of the so-called “La Bella Principessa” drawing and showed that while it bears no comparison with Leonardo’s female portrait type, it sits comfortably among 20th century fakes (see Fig. 1). Here, we consider the singular campaign to have this work accepted as a Leonardo.


Above, Fig. 1: “La Bella Principessa” (centre) among 20th century fakes. For the drawing’s striking mismatch with secure Leonardo works – and with other bona fide associated works of the period – see Problems with “La Bella Principessa” – Part I: The Look.


The nine years long campaign to have the drawing “La Bella Principessa” accepted as an autograph Leonardo da Vinci is faltering even before our series of examinations is completed. The leading proponent, Professor Martin Kemp, is said in the May 2016 Art Newspaper (Vincent Noce’s “La Bella Principessa: Still an Enigma”) to have his “reputation on the line” in the wake of our posts and an article “How reliable are today’s attributions in art? The case of ‘La Bella Principessa’ examined”, that was published in the Polish scholarly journal Artibus et Historiae, no. 71, 2015 (“La Bella Principessa – Arguments against the Attribution to Leonardo”) by Dr. Kasia Pisarek, an independent art historian (and ArtWatch member). In delayed response to our January 2014 suggestion that the disputed drawing’s author might have been the painter/restorer Gianinno Marchig (see Art’s Toxic Assets – Part II), Kemp now alleges on his blog that we are making “scurrilous and unsupported” attempts to “divert the argument into claiming that Jeanne Marchig lied profusely”. This is not the first such slur against us from that quarter. When Professor Kemp reviewed the James Beck/Michael Daley book Art Restoration: The Culture, the Business and the Scandal in 1994 he made similarly unfounded charges which we rebutted immediately. In recent years Kemp has cast his denunciations more widely and generally against his fellow scholars. (See below.)

Notwithstanding the “La Bella Principessa” campaigners’ belligerence towards doubters, and Professor Kemp’s own publicly bullish support for the attribution, his position on “La Bella Principessa” is untenable in terms of the work’s artistic properties and its emerging circumstances – as indeed is the methodological model for attributions that he hubristically offers to fellow scholars in connection with the Leonardo upgrades that he supports (see below).

Pace Kemp’s charge of evasion, it would be evasive not to consider Marchig’s role when appraising a drawing reportedly bought by a Panamean, Jersey-based company nine years ago at a requested discount for $19,000 but which now, as a claimed Leonardo, lives in a Swiss vault and is said to be insured for $150 million. We should all consider the circumstances and nature of this particular work and the assiduous, sometimes muscular campaign to upgrade it. The old masters market is fragile. The accelerating expanionism of recent years cannot be sustained. The market cannot afford to take in too many too-hopefully upgraded Leonardos, Michelangelos, Rubens’s, Van Dycks, Caravaggios and so forth. Aside from the resulting adulteration of scholarship, markets, as we all now appreciate, lose confidence and crash when too high a proportion of toxic assets is thought to have been bundled in among the bona fide.


What is now presented as Leonardo’s “La Bella Principessa” of c. 1496 was sold anonymously at Christie’s, New York, in 1998 for $22,850 as a work without provenance. Twelve years later, Jeanne Marchig, the widow of the artist/restorer Giannino Marchig who had worked as a restorer for Bernard Berenson (who, Kenneth Clark said, sat on a pinnacle of corruption), identified herself as the vendor. She did so not in the disinterested cause of scholarship but to claim damages after sensational but unfounded and misleading media reports that fingerprint evidence had shown the drawing to be a Leonardo.

As we reported, aside from the widow’s hearsay claims concerning the ownership of the drawing by the painter/restorer, the drawing otherwise possesses not a shred of recorded history in its supposed five centuries. On the widow’s account (as variously reported by Kemp, by the drawing’s owner, Peter Silverman, and by a journalist selected by Silverman to promote the attribution, Simon Hewitt), Marchig, an unsuccessful artist who had grown rich and acquired a collection of valuable historic works, had declined to say even to his wife when or from whom he had acquired the framed drawing which he is said to have kept in a portfolio.

Jeanne Marchig dangled the possibility that he might have acquired it from Berenson, with whom he had grown close and for whom he had hidden paintings, photographs and documents during the war. Had he done so that would hardly dispel doubts and suspicions but even that possible lifeline to a past now seems unlikely: the drawing’s present owner and Kemp vainly trawled Berenson’s archives at I Tatti for any sign or mention of the drawing. The trail of this supposed Leonardo begins and ends with the Marchigs. Marchig, if his widow is reliable, said nowt and simply had. By coincidence, the widow was born in Warsaw where a book is housed that was later said (unpersuasively for reasons given by Dr Pisarek) to have contained the vellum sheet on which the drawing was made. Professor Kemp thinks the late Jeanne Marchig “a person of great credibility”. The journalist Simon Hewitt reported in the Huffington Post that:

“Jeanne Marchig was born Janina Paszkowska in Warsaw, into a family of doctors and lawyers. She was an only child: her father died in an accident before she was born. Her mother Elzbieta Chrostowska, an amateur wood-carver, took her to Sweden in 1939, where she grew up and married, became Janina Hama. The marriage didn’t last. She met the artist Giannino Marchig on a train between Stockholm and Florence, where he worked as a picture restorer. Berenson and Wildenstein were his top clients. Although a youthful exponent of racy nudes, Giannino was no lady’s man. He lived at home with his mother on the banks of the Arno. He was over 50. Jeanne was an art student. Their age-difference ran into decades. They married. People talked. Talked, too, about Giannino’s wealth. What had he done during the war? Helped hide Berenson’s collection from the Nazis, among other things. Did Berenson give him the Bianca portrait? Jeanne Marchig didn’t know. Or wasn’t saying. They moved to Switzerland. She morphed from a flirtatious livewire into a coquettish Miss Marple of unfluffy shrewdness. Giannino died in 1983. Jeanne published a sumptuous catalogue of her husband’s career and religiously kept the box of pastels he had used to restore the Leonardo.”


Above, Fig. 2: The eye of “La Bella Principessa”, top left, and above left (with a superimposed diagram). Top right, an eye drawn by Leonardo (reversed). Above right, an eye featured on a sheet of eyes drawn and reproduced as aids to students and artists in a famous drawing course published in the late 1860s – and later used by the young Picasso.


Martin Kemp and I have recently discussed the eye in “La Bella Principessa” (top left) vis-à-vis the eye by Leonardo (top right) and I am grateful to him for this.
He believes that both eyes are drawn by Leonardo. I (a left-handed draughtsman) hold that the “La Bella Principessa” eye, with its pronounced, almost Cubist, angular and planar construction cannot conceivably have been drawn by Leonardo. There is simply nothing like it in Leonardo’s oeuvre. It is a construct of an alien, more modern kind. Kemp now admits that the unnaturally thick and angular lower lid is problematic but writes: “With the exception of the angularity of the lower lid, which is in an area of some damage, it is consistent (above all the amazingly delicate lashes) with the attached [the eye by Leonardo, top right]. Leonardo’s works of art are not anatomical demonstrations. It’s easy to find ‘anatomical erors’. I find the seizing on such things is to divert the arguments into issues of a peripheral nature in the face of evidence of a non-arbitrary kind.” This is a helpfully clarifying statement, but the suggestion that the eye might have been repaired is new.

In the 2010 Martin Kemp and Pascal Cotte book Leonardo da Vinci “La Bella Principessa” The Profile Portrait of a Milanese Woman, there is no indication given in Cotte’s map (Fig. 3) of repairs to “La bella Principessa” that the eye had been damaged, and Kemp, when comparing the eye with that of Leonardo’s Windsor Castle drawing Portrait of a Woman in Profile (Fig. 5), wrote of it “Even Boltraffio could not achieve this. The structure of the eyelids, the delicate flicks to create the lashes, and the translucent iris of the eye are extremely close on both portraits…” Pascal Cotte goes further, claiming a “distinct and identical logic” with the eye in Leonardo’s portrait of Cecilia Gallerani (The Lady with an Ermine, Fig. 4). Making no reference to injuries or repairs, Cotte specifically points in his diagram (Fig. 4) to the “The juxtaposition of the edge of the lower eyelid with the bottom of the iris”. Speaking generally of the analysed physical evidence of the drawing, Cotte adds “There have been some diplomatic retouchings over the years, but this has not affected the expression and physiognomy of the face to a significant degree or seriously affected the overall impact of the portrait.”


What Kemp sees as a peripheral issue that lacks “non-arbitrary” evidential value, I take to be of the essence in the evaluation and critical appraisal of (visual) works of art. In traditional connoisseurship – an area which Kemp frequently disparages on quasi-scientific professional and leftist political grounds – the test is not to identify similarities (which exist in abundance between authentic works, copies and forgeries) but to discern differences, to discriminate between products of the authentic autograph hand and closely related but variant artefacts.

Of course, Leonardo did not draw every eye as an anatomical demonstration, but nor did he ever draw an eye in ignorance of that crucial feature’s anatomical construction. The eyeball, being an orb, determines the shape and forms of the surrounding soft protective tissue of the lids. Leonardo’s eyes and lids are constructed with curves, not straight lines. With the four images above at Fig. 2, a connoisseur’s ‘eye’ should recognise that even when constructing an eye with straight and not curved lines (as in the lower right demonstration drawing) it is possible to render the visible part of the eyeball conceptually if not literally spherical. That schematic drawing displays greater sculptural and anatomical acuity than does the more laboured and “finished” “La Bella Principessa”. It recognises and describes with three (faint) straight lines, what Leonardo depicts with curves: the line(s) of collision between the bulging soft flesh of the lower eyelid (when the eye is open) and the more taught flesh that is stretched over the cheekbone. In the infrared image at Fig. 6 we can see with crystal clarity how Leonardo saw the structure of the lower eyelid and how he set out this structure in preliminary drawing form. While Leonardo gave fluent anatomically-informed account of eyes, Marchig was insecure in his treatment (see Art’s Toxic Assets – Part II). Had Leonardo complied – against everything else in his output – with an irresistible court demand for a strictly profile treatment of a female subject’s head and torso, as Kemp claims in defence of La Bella Principessa, he would have had the wit and the judgement to render the eye, too, in strict profile. “La Bella Principessa’s” eye – which is smaller – is not drawn in accord with that strict out-of-perspective formal convention. Rather, it strays into looking both outwards and downwards, imparting an insecure, wary, not proud air.


Above, Fig. 3: The colour coded map that is said to show areas of restoration in “La Bella Principessa”, as published on page 133 of the 2010 Kemp/Cotte book Leonardo da Vinci “La bella Principessa” The Profile Portrait of a Milanese Woman. It is striking how discretely localised are the areas of “restoration”, and how fortuitously the corresponding areas of injury had fallen in the least important parts of the image.


Above, Fig. 4: Top, the (true) right eye of The Lady with an Ermine. Above, the eye of “La Bella Principessa”. In this diagram Pascal Cotte, a brilliant engineer, sees confirmation of a common author – even at his arrowed point 2 where Leonardo’s curved demarcation between the eyeball and the lower lid is set against the form-denying straight demarcations in “La Bella Principessa”.

Above, Fig. 5: Left, the eye of Leonardo’s Windsor Castle Portrait of a Woman in Profile (reversed). Right, the smaller, more angular and more sunken eye of “La Bella Principessa”.

A common fault of copyists, pastiche-makers and forgers is to get details correct while missing the larger unified relationships which, collectively, they should constitute. The author of “La Bella Principessa” not only misses the cohering sense of the eye as a ball but even misconstrues the form and anatomical function of particular details. This error testifies to forgery rather than pastiche. We will be showing “La Bella Principessa” to be a “portmanteau work” composed from features drawn from a number of bona fide Leonardos. It can hardly seem a coincidence that the most disqualifying error of drawing in “La Bella Principessa” – the lower eyelid – occurs at the very point where damage is found in the (“prototype”) work which it is most closely said to resemble. At this point the drawing’s author has clearly been required to invent rather than copy or paraphrase. Indeed, in “La Bella Principessa’s” eye we find a progressive falling off of anatomical and artistic credibility from top to bottom: most plausible in “La Bella Principessa” is the somewhat simplifying paraphrase of the upper eyelid. Less plausible is the treatment of the more complex and elusive eyeball and iris. Least plausible of all is the fabricated lower eyelid. Martin Kemp’s claim that by drawing attention to such incompatibility we seek to divert the arguments “into issues of a peripheral nature in the face of evidence of a non-arbitrary kind” is not only unfounded – much material is in train on this attribution – it betrays a technically philistine misapprehension of sound scholarly method. Let us be clear: art, not the devil, lies in artistic detail and these details testify to authorship. We have the clearest possible understanding of how much Leonardo knew and how well he gave expression/record to what he knew/saw. It is for those who would count “La Bella Principessa” as a Leonardo to explain the disparity between its eye and that below at Figs. 7 & 8, which, on Kemp’s account, were both made at the same date.


Above, Figs. 6, 7 & 8: Top, the eye of “La Bella Principessa”; Centre, the (true) left eye of Leonardo’s La Belle Ferronnière, as seen in an infrared reflectogram by E. Lambert for C2RMF as published in Leonardo’s Technical Practice, Paris, 2014; Above, the (true) left eye of Leonardo’s La Belle Ferronnière, as seen in Pietro Marani’s Leonardo da Vinci – The Complete Paintings, New York, 2000. It is, for reasons given here, inconceivable that both of these eyes could have been produced by Leonardo at the same time – c. 1496.

There is another sense in which Kemp’s espousal of “La Bella Principessa” exposes his art historical method. By drawing attention to “La Bella Principessa’s” many points of direct correspondence with the Windsor Castle profile portrait in terms of overall effects, pentimenti, anatomical details and so forth, the scholar strains credulity. The Windsor Castle portrait is clearly of a mature woman. It was made some 15 years earlier than the supposed portrait of Bianca Sforza who died when a child of fourteen. The Windsor drawing is made in another (and single) medium – silverpoint – in which Leonardo was effortlessly, supremely fluent. “La Bella Principessa”, however, was made in an unprecedented combination of materials on a never or extremely rarely encountered-in-Leonardo support. How likely, then, is it that Leonardo would produce an elaborately finished drawing in an un-encountered mix of graphic and pictorial media and of a type nowhere else encountered in his oeuvre, of a young girl who, on Kemp’s account, was drawn either directly from life or from some other unknown record of her appearance, in commemoration after her death, some fifteen years later, that would, when reversed, produce a near perfect coincidence of proportions, features and, even, pentimenti?

It would, of course, be entirely unremarkable for a forger or pastiche-maker to engineer a similar reversed coincidence of features and traits with a bona fide Leonardo work. A canny forger who happened to be a restorer of old masters – including Leonardo, as was the case with Marchig – would well appreciate the need for evidence of one or two reassuring “campaigns of restoration” in a supposed work of five centuries of age. Marchig’s widow has reportedly claimed that he had indeed conducted restorations on the front of the drawing and on the back of the oak panel to which it was – unprecedentedly and, it is now claimed, irreversibly – glued. Kemp has not discussed, as far as we know, technical evidence that has been discussed by another Leonardo scholar – Cristina Geddo – that the back of the vellum is not blank as it would surely have been had it ever faced the the elaborately illuminated frontispiece of a major book as Kemp claims. For Dr Geddo, it is reassuring that the back of “La Bella Principessa’s” vellum support bears “superimposed numbers…like others written in pen, such as a very pale inscription visible along the upper border of the sheet and the little winged dragon – at least this is what it seems – in the lower left corner. This feature, too, counts in favour of an attribution to Leonardo, who, even though he has never to our knowledge used a parchment support in his work, was in the habit of re-using the paper on which he wrote or drew.” Of course, forgers too re-use old material.


Nothing can bring greater benefit to the art world than free, frank discussions and debate. The annual three-day Hague Congress is organised by a body that addresses the subject of authenticity in art and is titled AUTHENTICATION IN ART. This year’s AiA congress (11-13 May) specifically addresses the voguish museum world hybrid discipline known as Technical Art History, the misapplied and anti-aesthetic scientism of which we have criticised since its earliest days (see, for example, the first post on this site The New Relativisms and the Death of “Authenticity”). Attendees at this year’s AiA congress comprise “art collectors, collection managers, directors of museums and galleries, art dealers, appraisers, connoisseurs, advisors, auctioneers, insurers, investors, lawyers, authors of catalogues raisonnés, restorers, conservators, material scientists and art historians.” This particular critic of Technical Art History will not be speaking even though our proposal for a paper was encouragingly received by a congress organiser who wrote:

“Dear Michael, Thank you. Very valuable to the whole set up of AiA 2016. Get back to you in the coming weeks.”

The proposal had opened:

“Technical Art History is presented as a multi-discipline, international museum-standard professional synthesis that eliminates error and delivers enlightenment when, in truth, it testifies to little more than the ascendency in museums of technicians over curator/connoisseurs. This putsch began with the creation of in-house museum restoration departments where staff restorers could no longer be sacked. The National Gallery in London claims pioneering authorship of the new hybrid discipline and it perfectly reflects the new pecking order.”

We were subsequently “dis-invited” through a form notification to unsuccessful applicants. Among this year’s speakers will be Professor Kemp, a member of the AiA’s advisory board, and Pascal Cotte, of Lumière Technology. Cotte was, as mentioned, co-author with Kemp of the 2010 and 2012 English and Italian editions of a book of advocacy, La Bella Principessa – The Story of the New Masterpiece by Leonardo da Vinci. (Kemp has been critical of Cotte’s recent, go-it-alone book Lumière on the Mona Lisa – Hidden Portraits.) A working group was set up to organise this year’s congress. It is comprised entirely of conservators or conservation scientists. Curators and connoisseurs are not represented. One member of this advisory group was David Bomford, who is presently the Chair of Conservation and Head of European Art at The Museum of Fine Arts, Houston. Formerly, as a senior restorer, Bomford was the architect of the National Gallery’s presently stated cleaning philosophy – which we had discussed in some detail in our proposed congress paper (“…The false assurances of Clark’s aesthetics/science sleight of hand haunt and deform the National Gallery. Its official conservation guide declares restorations to rest on individual restorer’s own aesthetic inclinations…”)


In his 2014 AiA congress paper (“It Doesn’t Look Like Leonardo”) Kemp discussed two other Leonardo upgrading attributions with which he is associated. Namely, the two versions of the Madonna of the Yarnwinder (one of which was restored badly, Kemp once complained, by Marchig) and the massively restored wreck of a panel painting, the Salvator Mundi. His paper’s abstract ran:

“The state of methods and protocols used in attribution is a professional disgrace. Different kinds of evidence documentation, provenance, surrounding circumstances of contexts of varied kinds, scientific analysis, and judgement by eye are used and ignored opportunistically in ways that suit each advocate (who too frequently has undeclared interests). Scientific evidence is particularly abused in this respect. The status of different kinds of evidence is generally not acknowledged, particularly with respect to falsifiability… I will attempt to bring some systematic awareness into this area, which is a necessary first step in establishing some rational protocols. The case studies will be drawn from Leonardo.”

On reading this abstract with its scattergun slurs “opportunistically”, “disgrace” and “undeclared interests”, we laughed out loud. Partly because of the grandiose title – “The 2014 Hague Congress Authentication in Art – What happens when the painting you are buying, selling, investigating, exhibiting, insuring – Turns Out to be a Fake or a (Re)Discovery…” – chosen at a time when Kemp and others had failed to achieve a consensus of support for the drawing he had portentously dubbed “La Bella Principessa”. But also because, as mentioned, two decades earlier we had experienced Kemp’s invective and sneering distaste for traditional connoisseurs whom he sees as “a self-proclaimed (and often class-based) elite whose skills are insulated from systematic scrutiny”.

In his review of the James Beck/Michael Daley book Art Restoration – The Culture, the Business and the Scandal”, the authors – much as with Kemp’s peers today – were found professionally wanting and morally deficient: “…Their problem is that they seem unwilling to acknowledge the status of different kinds of evidence…The slanting of arguments, manipulation of quotations, and rigging of visual evidence may be effective journalism but it is poor history…” A more focussed barb was aimed at artists’ evaluations of conservation treatments: “Stalking throughout their book… is a very particular notion of ‘Art’ and its creators. ‘An Artist’ (ie What Beck and Daley understand as an artist in today’s terms) is adduced as the most important arbiter of the criteria for the treatment of our historical heritage. I am unclear about the identity of this archetypal beast.” Unclear indeed.

Kemp’s professional aversion to the views and judgements of artists is presented as a token of “higher”, more philosophically sophisticated notions of scholarly method and a scientist’s preference for non-judgemental, non-subjective “evidence” within it.


While some are cowed by Professor Kemp’s trademark abusive critical put-downs, others, like the blogger and art “sleeper” hunter, Bendor Grosvenor (21 April), openly admire them. We called Kemp’s bluff in a letter to the THES (2 May 1994):

“…he alleges… a misuse of historical and material evidence. Professor Kemp’s notions of misuse would seem to be singular: he complains, for example, that Professor James Beck and I accepted Charles Heath Wilson’s clear and detailed testimony that Michelangelo had extensively revised his frescoes with glue painting, ‘with unquestioning approval’. This is presented as proof of our ‘lack of discrimination’. But Wilson saw what he saw and said what he said… Does Kemp have any grounds for rejecting Wilson’s record?… Kemp is silent on this evidence. Why? The photographic evidence we supply of restoration-induced injuries is impugned by Kemp as ‘rigged’. It is nothing of the sort – most of it was provided by the restoration authorities themselves… Does Kemp wish to defend that restoration? Is he in possession of any photographs which tell a different story?”

No grounds for rejecting Wilson were offered in reply. No contra-testifying un-rigged photographs were ever produced.
For a fuller account of the spurious charges raised by art conservators and Kemp to Art Restoration, see “Why are picture restorers allergic to appraisal?”, Jackdaw, May/June 2016.


In this year’s AiA congress paper Kemp returns to his 2014 AiA congress attack on the shortcomings and abuses he perceives in the methodologies and behaviour of all other scholars in the field, albeit in muted form. Today his abstract reads:

“A speech on Technical Art History and the way he [Kemp] implements the research on Leonardo’s La Bella Principessa: the varieties of evidence and arguments, and how reactions to the attribution shed light on the disorderly nature of current methods”.

A clue to where he might be going can be found on his blogsite where he has published a “reworked” version of his 2014 paper that excludes his earlier linking remarks on other Leonardo attributions he supports – “Science and Judgement by Eye in the Historical Identification of Works of Art”.

The methodological schema Kemp outlines and censoriously offers to others seems little more than an overly complicated regurgitation of the tendentious, the self-evident and the true-by-definition. It leans heavily on and misapplies Karl Popper’s famously illuminating discussion of scientific knowledge and methods. Kemp seems to crave an aesthetic equivalent of the decisive Popperian test of scientific “falsifiability”. This is a vain, misdirected quest. Because of the profound differences between appraisal of works of art and the technical analysis of their constituent materials, Kemp is forever complicating and caveating his proposed model method. He compiles tables of hierarchies that are organised into polarising dualities. He is constantly extricating himself from fogs of his own making:

“In this and the following table, I am using the terms ‘scientific’ and ‘art historical’ in a conventional way without intending to signal that they can be isolated in the actual practice of attribution – and certainly not to suggest that the application of scientific analysis necessarily delivers more certainty than art historical evidence. In the actual practice of art history, its isolation from scientific analysis is all too common.”

Which? What? To help orientate us, Kemp devises a table of criteria that are more traditionally “art historical” with the “the most malleable [being] at the top.” But nothing ever comes to rest. Here, “I have added standard kinds of evidence relating to provenance and documentation that are highly constructive.” If baffled by the usage “constructive”, the reader must back-track to an earlier explanation that “The kinds of evidence and explanation that can be subject to various degrees of falsification can be grouped under two headings: constructive and permissive.” Thus, “By constructive I mean those that add positively and accumulatively to the case being made for a specific attribution. By permissive I am signalling those that present no obstacle to the attribution being made, i. e. they offer a nil obstat.”


When Picasso was asked what he thought of the philosophical discipline aesthetics, he replied: “Aesthetics is to the artist what ornithology is to birds.” In Kemp, everything is dunked in pseudo-philosophical terminology. Take the simple clear self-explanatory notion “judgement by eye” which he offers in preference to snobby, elitist, class-based etc “connoisseurship”:

“As we have seen, judgement by eye plays a key role in key scientific techniques. Although the most constructive of the kinds of art historical evidence, documentation and provenance, do not rely upon judgement by eye, it is common that this kind of evidence is not available or is less conclusive than we would wish. In many cases judgement by eye necessarily provides the actual starting point, before other kinds of investigation are undertaken. This is often the situation when a previously unknown or unrecognised work first emerges with specific claims attached to it. Let us try to formulate some propositions about judgement by eye in a somewhat Popperian manner”

Why, apart from intellectual snobbery, in a “Popperian” manner? Well, they help Kemp to conclude with the twin observations that “Judgement by eye is malleable in the light of multiple interests”, and “Judgement by eye is falsifiable only by factors outside of itself.” Kemp’s Popperian edifice is, as it were, a perniciously misleading red herring. In truth judgements by eye cannot be equated with falsifiable scientific propositions, they are a different beast – they are critical appraisals. As such they are gambits in a discussion. They can be countered by demonstrably superior, more percipient and persuasive judgements that accord better with the material circumstances and visual facts of a given work of art. Because Kemp sets the hard and “irrefutable facts” of material analysis (many of which, he acknowledges, require judgement by eye) against what he deems the “subjective”, “relative” and “malleable” act of appraisal of the (despised) connoisseur, he misrepresents what is indispensable to proper appraisals of art. Connoisseurship (the term should not frighten or repel us) cannot sensibly be treated as a natural science – properly, rigorously conducted it is an open, competitive adversarial system that is full of checks and balances wherein anything that is proposed may examined, challenged and deposed. Art world abuses certainly exist but they flourish best when legitimate criticisms and demonstrations are blocked and disqualified and critics are ruled out of court. Kemp, who despises the “class-based” connoisseur, does not respond well to criticism and prefers abuse and denigration to straightforward and healthy critical engagement. That is his loss, but also, he being a talented man, it is that of scholarship itself.


When ArtWatch UK, The Center for Art Law, and the London School of Economics Law Department, organised a conference on connoisseurship in London last December (“Art, Law and Crises of Connoisseurship”), one of the papers, “How reliable are today’s attributions in art? The case of ‘La Bella Principessa’ examined”, by Kasia Pisarek, presented a number of the interlocking art historical, aesthetic and technical criticisms she had recently published in the above mentioned Polish scholarly journal Artibus et Historiae.

Because this paper was a strong and detailed rejection of the attribution, we invited Martin Kemp to give a paper. He declined. We then invited Nicholas Turner, an independent scholar and drawings expert who had championed the “La Bella Principessa” drawing before Prof. Kemp. He, also, declined to speak. To represent the restoration-pro-active, attribution-making school of art dealers, we invited Philip Mould. Mr Mould, too, declined to speak. Shortly before the conference, Kemp, Silverman and a Polish art historian, Kasia Wozniak, all lobbied for the “balancing” inclusion in our conference proceedings of the journalist Simon Hewitt who is writing a book with the owner of La Bella Principessa, Peter Silverman – who tells us that the book is on “various aspects of the art market, sometimes highlighted by others’ and my own discoveries”. Hewitt attended the conference and, from the floor, launched an assault on Pisarek’s case – but did so, we later learned, on the borrowed authority of Kemp who had “prepped” him for the occasion on objections he had framed in response to Pisarek’s Artibus et Historiae article. Kemp later submitted his lengthy response to Artibus et Historiae for publication but it was turned down. This article has now been posted on the AiA Congress website. Perhaps, in the interests of scholarly balance, the AiA congress will now also post the article by Dr Pisarek which Professor Kemp aims to rebut? Perhaps the Congress might also consider posting our own initially valuable-to-proceedings but later rejected paper on Technical Art History?

Michael Daley, 3 May 2016

COMING SOON: The Salvator Mundi, Giannino Marchig, Left-handedness and “La Bella Principessa”

Problems with “La Bella Principessa”~ Part I: The Look

The world famous drawing that was dubbed “La Bella Principessa” by Professor Martin Kemp is insured for $150 million and lives in a “secure vault in Zurich”. It is not a portrait of Bianca Sforza by Leonardo da Vinci, as has been claimed, but a twentieth century forged or pastiche Leonardo.


In 1998 the now so-called “La Bella Principessa” appeared from nowhere at Christie’s, New York. A hybrid work made in mixed media that were never employed by Leonardo (three chalks, ink, “liquid colour”), on a support that was never used by Leonardo (vellum), and portraying a woman in a manner that is nowhere encountered in Leonardo, it was presented as “German School, early 19th century” and “the property of a lady”. It went for $22,850 to a New York dealer who sold it nine years later on a requested discount of 10 per cent for $19,000 to an art collector, Peter Silverman, who said he was buying on behalf of another (unidentified) collector whom he later described as one of “the richest men in Europe”. Thus, at that date, it was not known who owned the drawing or by whom it had been consigned to Christie’s and it remained entirely without provenance. In its nine years long life, no one – not even its new owner(s) – had taken it to be by Leonardo.

In a 2012 book (Lost Princess ~ One man’s quest to authenticate an unknown portrait by Leonardo da Vinci), Silverman claimed a successful upgrading to Leonardo and described how he had gained the support of distinguished scholars including Professor Martin Kemp who had formulated an elaborate hypothetical history in which the drawing was said to be a Leonardo portrait made either from a living subject in celebration of her wedding or in commemoration after her death in 1496.

Nonetheless, the drawing failed to gain a consensus of scholarly support and is rejected in centres like New York, London and Vienna. Carmen Bambach, the Metropolitan Museum’s Renaissance drawings authority dismissed “La Bella” on the grounds that “It does not look like a Leonardo”. Thomas Hoving, a former Metropolitan Museum director, held it to look “too sweet” to be Leonardo. ARTnews reported that the Albertina Museum’s director, Klaus Albrecht Schröder, had noted “No one is convinced it is a Leonardo”. In the Burlington Magazine Professor David Ekserdjian suspected it to be “counterfeit”.


In matters of attribution the most important consideration is the look of a work. Many things can be appraised simultaneously but, conceptually, the “look” of a work might be broken down into two aspects: an initial at-a-glance response to a work’s effects and appraisal of its internal values and relationships; and, a comparison of the effects, relationships and values with those of bona fide productions of the attributed artist, or with those of the artist’s students, associates or followers. It can also be useful to compare the looks of works with those of copyists and known forgers. It might fairly be said that in connoisseurship, as in the evaluation of restorations, visual comparisons are of the essence. (In ArtWatch we take pride in the extent to which we seek out all possible comparative visual material and regret that some institutions still hinder our efforts in this regard.)

Above, Fig. 1. If we put aside questions of attribution and simply look at the group above, we find works of remarkably similar figural motifs and formats that clearly relate to and derive from a most distinctive type of 15th century Italian profile female portrait. These similar-looking works are similarly sized, being, respectively from left to right:

A Young Woman, 14 and 1/4 x 10 inches;
“La Bella Principessa”, 13 x 9 and 3/4 inches; and,
A Young Woman, 18 x 12 1/2 inches (here shown mirrored).

All show young women depicted in the strict early Renaissance profile convention made in emulation of antique relief portraits on coins and medals. Although very widely encountered (see Fig. 4), Leonardo side-stepped the type in order to intensify plastic and expressive values with sculpturally-purposive shading and axial shifts in the bodies and gazes of his portraits (see Fig. 6). The portrayals above are strikingly similar in their head/torso relationships; in their absences of background; in their highly elaborated coiffures which offset ‘sartorially’ skimped and unconvincing simplifications of costume; in their sparse or wholly absent depictions of jewellery; and, even, in their almost identically cropped motifs. Collectively they might be taken as a suite of variations on a simple theme. We take all three to be twentieth century Italian artefacts. At least two of them are linked to Bernard Berenson and the two on which reports have been published have unusual and problematic supports.

As mentioned below, the Detroit picture is painted on top of photographic paper. It is suspected that it might have been a photograph of the Frick sculpture to which the painting was initially related. The “La Bella Principessa” is drawn, exceptionally for Leonardo, on a sheet of vellum which appears to have been removed from a book and it is, most unusually, glued to an oak panel. The panel itself is a curiosity: although a number of “butterfly keys” have been inserted into its back, as if to restrain splitting, there is no evidence of splits in the panel and, if there were, the present four such keys in such a small panel might be considered restoration “over-kill”. If the panel had split while the vellum was glued to it, the drawing would have split with the panel. The fact that the vellum has been “copiously glued” to a (possibly pre-restored) oak panel makes it impossible to examine the back of the drawing which is said by one of its proponents, (Cristina Geddo, an expert in Leonardo’s students and Milanese “Leonardesques”) to bear “superimposed numbers…a written inscription…[and a] little winged dragon – at least that is what it seems.” For Geddo, this unexamined content is reassuring: “This feature, too, counts in favour of an attribution to Leonardo, who, even though he never to our knowledge used a parchment support in his work, was in the habit of re-using the paper on which he drew.”

(In reading the compendious literature on this proposed attribution, we have sometimes wondered what might be allowed by its supporters to count as evidence against the attribution.)


The portrait on the left, A Young Woman, was bought in 1936 by the Detroit Institute of Arts as by Leonardo da Vinci or Andrea del Verrocchio. The institute’s director, W. D. Valentiner, made this attribution on the strength of clear correspondences with the curls in the hair of Leonardo’s painting Ginevra de’ Benci (see Fig. 6) in the National Gallery of Art, Washington, and with those found in the above-mentioned marble sculpture in the Frick Museum, A Young Woman, given to Andrea del Verrocchio. (Valentiner had made a study of Leonardo’s work in Verrocchio’s workshop.) In 1991 Piero Adorno, specifically identified the Detroit picture as Verrocchio’s lost portrait of Lucrezia Donati. Notwithstanding seeming correspondences with secure works, this picture is now relegated to “An Imitator of Verrochio” – and this is an extremely charitable formulation. In Virtue and Beauty, 2001, David Alan Brown described it as “a probable forgery by its anachronistic materials and unorthodox construction”. “Probable” [!] because: “after a recent technical examination, the picture turns out to have been painted on photographic paper applied to a wood panel that was repaired before it was readied for painting. And at least one of the pigments employed – zinc white – is modern…” Valentiner judged one of two Leonardo studio works of the Madonna with a Yarnwinder to be “more beautiful than the Mona Lisa”.

The portrait on the right, A Young Woman, was attributed to Piero Pollaiuolo by Berenson in 1945. While this figure is perhaps the most attractive of the above three, with its nicely constructed counterbalancing of the thrusts in the neck/head and torso, and its credibly proportioned arm, the work itself has, so far as we can ascertain, sunk without trace. In truth, this female profile portrait type has been assailed by forgeries. Alison Wright notes in her 2005 book The Pollaiuolo Brothers, that “Complications for the historian lie both in the fact that the subjects of most female portraits are no longer identifiable and that, because of their exceptional decorative and historical appeal, such portraits were highly sought after by later nineteenth- and early twentieth-century collectors, encouraging a market for copies, fakes and over-ambitious attributions.”

The portrait in the centre (“La Bella Principessa”) has been precisely attributed by Kemp to Leonardo as a book illustration portrait of Bianca Sforza of 1495-96.


Above, Fig. 2. In My dear BB (an incalculably valuable new resource edited and annotated by Robert Cumming), we learn that in November 1930 Kenneth Clark’s wife, Jane, wrote to Berenson: “K has seen Lord Lee’s two new pictures…The Botticelli Madonna and Child you probably know too. K thinks the latter may be genuine about 1485 or rather part of it may be, but it is not a pretty picture…” A footnote discloses that Lee had bought The Madonna of the Veil, a tempera painting on panel in 1930 from an Italian dealer for a then huge sum of $25,000 (Fig. 3). It was widely accepted by scholars as autograph Botticelli and published by the Medici Society as a “superb composition of the greatest of all Florentine painters”. Clark, doubting the attribution on sight, objected that it had “something of the silent cinema star about it” – and he likened the Madonna to Jean Harlow (Fig. 3). Lee donated the picture to the Courtauld Institute Gallery in 1947. In June 2010 Juliet Chippendale (a National Gallery curatorial intern working in association with the Courtauld Institute MA course) disclosed that scientific examination had identified pigments not known before the 18th and 19th centuries and worm holes that had been produced by a drill. It is now designated a work of the forger Umberto Giunti (1886-1970), who taught at the Institute of Fine Art in Siena and forged fresco fragments.


Four months later Clark wrote to Berenson: “Just in case Lee has sent you a photograph of his new Botticelli may I ask you to forget anything Jane may have reported me as having said of it. It is one of those pictures about which it is best to be silent: in fact I am coming to believe it is best for me to be silent about every picture. Did I tell you that my Leonardo book was a mare’s nest. The man had sent photographs of two drawings from the middle of the Codice Atlantico. They must have been early copies done with some fraudulent motive – perhaps the book really did belong to Leonardo – he certainly had read it – & some pupil thought to enhance its value.”

Above Fig. 3. The young Kenneth Clark (then twenty-seven years old) displayed an admirable “eye” by spotting a fraud on sight some eighty years ahead of the pack. Is it better for a connoisseur to see but not speak than it is not to see at all? Undoubtedly, it is. Would Clark have enjoyed his meteoric rise had he humiliated the mighty and exposed the big-time fraudsters of his day? (That question might be taken as self-answering.) If Clark bided his time on Berenson, eventually he delivered an unforgiving former-insider’s repudiation in 1977 by chronicling how Berenson had “sat on a pinnacle of corruption [and] for almost forty years after 1900… [done] practically nothing except authenticate pictures”


Above, Fig. 4. In the middle and bottom rows we see three bona fide works of the female profile type – respectively:

Portrait of Bianca Maria Sforza, c. 1493, by Ambrogio de Predis, The National Gallery of Art, Washington;
Domenico Ghirlandaio’s 1488-1490 Giovanna degli Albizzi Tornabuoni, Museo Thyssen-Bornemisza, Madrid; and,
a portrait of Beatrice d’Este tentatively attributed by Kemp to Ambrogio da Predis.

The differences between this trio and the works in the top row are pronounced and eloquent. The secure works are highly individuated and immensely richer in their effects. Collectively, they do not constitute an inadvertent suite. Individually, they are greatly more various compositionally. Collectively, they are markedly richer in jewellery and ostentatiously sumptuous costumes. The distinctive physiognomies of their subjects derive from living persons, not from other art or photographs of other art. Flattery and loving attention are channelled more into the costume and bling than into the facial features. In every respect the opposite is the case in the top row where prettiness has been held at a premium with an eye on the modern photographically-informed market.


Above, Fig. 5: As mentioned, “La Bella Principessa” and her two companions are of a piece, and of a type never followed by Leonardo whose female portraits (see below) pioneered an unprecedentedly complex and sophisticated evocation of real, sculpturally palpable women in tangible spaces or landscapes. To include the figurally impoverished and stylistically anachronistic “La Bella Principessa” in Leonardo’s oeuvre would disjunct his revolutionary arc of insights and innovations in portraiture. Such inescapably disruptive consequences have been ceded tacitly by Kemp, “La Bella Principessa’s” principle defender – some say advocate. In “La Bella Principessa ~ The Story of the New Masterpiece by Leonardo da Vinci” (Kemp’s 2010 book jointly written with Pascal Cotte of Lumiere Technology and including chapters by the drawings scholar Nicholas Turner and the recently discredited fingerpints expert Peter Paul Biro), Kemp converts an intractable problem into an asset by begging the essential question. That is, he underwrites “La Bella’s” credibility on an assertion that “Any important new work, to establish itself, must significantly affect the totality of Leonardo’s surviving legacy over the longer term.” Without question, the de-stabilising inclusion of “La Bella Principessa” would produce knock-on effects, but arguing backwards from that predictable disturbance to some endorsement of its source is patently unsound methodologically – the inclusion of any atypical work, whether bona fide or forged, into an oeuvre would affect its “legacy”.


Above, top, Fig. 6: Left, Andrea del Verrocchio’s Lady with a Bunch of Flowers of c. 1475; and (right) Leonardo’s (hypothetically extended) Ginevra de’ Benci of c. 1474-1478.

Above, Fig. 7: Left, Leonardo’s The Lady With an Ermine of about 1489-90; centre, Leonardo’s La Belle Ferronnière of about 1495-96; right, Leonardo’s Mona Lisa (La Giaconda) of about 1503-06 onwards.

In the group above we see extraordinary development in Leonardo’s portraits of women over the last quarter of the fifteenth century as he strove to incorporate the entirety of sculptural, plastic, figural knowledge, and to surpass it by making it dance to an artistically purposive tune liberated from the happenstance, arbitrary lights of nature on which sculpture then necessarily depended. Some have attributed the Bargello sculpture, the Lady with a Bunch of Flowers, to Leonardo on the grounds that its subject was Ginevra de’ Benci, the subject of Leonardo’s painting. Others have seen Leonardo’s authorship of it in the beauty of the hands. In Leonardo da Vinci and the Art of Sculpture, 2010, Gary M. Radke holds that the two works show differences that emerged in the mid-1470s between the two artists. Against this, it has been suggested that the painting might originally have borne a closer relationship to the sculpture with a possible inclusion of hands in a fuller length treatment. A study of hands by Leonardo was incorporated in a hypothetical and digitally realised extension of the painting by David Alan Brown (Virtue and Beauty, 2001, p. 143). Frank Zollner sees the painting as marking the point (1478-1480) at which Leonardo broke away from “the profile view traditionally employed in Florence for portraits of women” in favour of the three-quarters view in order to impart “a pyschological dimension to his sitter – something that would become the hallmark of Renaissance portraiture”. Which is all to say that Leonardo had broken away from the profile convention some sixteen to eighteen years before, on Kemp’s hypothesis, he made a solitary and exceptional “return” to it.

Speaking of the reconstruction of Leonardo’s Ginevra de’ Benci painting, Brown writes:

“Ginevra’s portrait, the lower part of which was cut down after suffering some damage, may have included hands. A drawing of hands by Leonardo at Windsor Castle, assuming it is a preliminary study, aids in reconstructing the original format of the picture. As reconstructed, Leonardo’s portrait may be seen to have broken with the long-standing Florentine convention of portraying women in bust-length profile. In seeking an alternative to the static profile, Leonardo, like Botticelli, seems to have turned to Verrocchio’s Lady with a Bunch of Flowers in the Bargello, Florence. Because of the sitter’s beautiful hands which mark an advance over the earlier head-and-shoulders type of sculpted bust, the marble has even been attributed to Leonardo. But the highly innovative conception of the half-length portrait bust is surely Verrocchio’s achievement. What young Leonardo did was to was to translate this sculptural protype into a pictorial context, placing his sitter into a watery landscape shrouded in a bluish haze…”


For the owner and the art historical proponents of “La Bella Principessa”, the very chronology of Leonardo’s female portraits constitutes an obstacle. Given Leonardo’s famous eschewal of strict profile depictions of women, the onus is on those who would include “La Bella Principessa” (- albeit as a solitary and exceptional stylistic regression that was undertaken without ever attracting attention or comment) to make a double case.
First, they must show how and where “La Bella Principessa” might plausibly have fitted within the trajectory of Leonardo’s accepted works. Second, they must demonstrate by comparative visual means that “La Bella Principessa” is the artistic equal of the chronologically adjacent works within the oeuvre. Kemp has proposed the precise date of 1495-96 for the execution of “La Bella Principessa” but, conspicuously, has not presented direct, side-by-side visual comparisons with Leonardo’s paintings. Instead of comparing “La Bella Principessa” of 1495-6 directly with Leonardo’s La Belle Ferronnière of about 1495-6, Kemp writes:

“If the subject of Leonardo’s drawing is Bianca, it is likely to date from 1495-6. In style, it seems at first sight to belong with his earlier works rather than to the period of the Last Supper. However, Leonardo was a master of adapting style to subject. Just as his handwriting took on an earlier cast when he needed to adopt a formal script, so his drawing style could have reverted to a meticulous formality, appropriate for a precious set-piece portrait on vellum of a Sforza princess.”

“If”? “Could have”? “At first sight”? The pro-attribution literature is bedecked with daisy-chains of such tendentious and weasel words and terms. With which earlier works is “La Bella Pricipessa” deemed to be artistically comparable or compatible? With the Ginevra de’ Benci of c. 1474-1478? With The Lady With an Ermine of about 1489-90? Never mind the red herrings of handwriting and the giant, near-obliterated historical figures of the Last Supper, what of the relationship with Leonardo’s (supposedly) absolutely contemporaneous La Belle Ferronnière of 1495-96? (On this last we volunteer a pair of comparisons below.)

Above, Top, Fig. 8: Leonardo’s La Belle Ferronnière, left, and the “La Bella Principessa”.
Above, Fig. 9: Details of Leonardo’s La Belle Ferronnière, left, and the “La Bella Principessa”.

Kemp insists: “The Lady in profile [“La Bella Principessa”] is an important addition to Leonardo’s canon. It shows him utilizing a medium that has not previously been observed in his oeuvre…It testifies to his spectacular explosion and development of novel media, tackling each commission as a fresh technical challenge. It enriches our insights into his role at the Milanese court, most notably in his depiction of the Sforza ‘ladies’ – whether family members or mistresses. Above all, it is a work of extraordinary beauty.”

Even if we were to assume that for some reason Leonardo had opted to “revert” in 1496 to a type he had never employed, what might explain a pronounced indifference in “La Bella Principessa” to the detailed depiction of the “stuffs” of costume with which the artist was simultaneously engaged in La Belle Ferronnière? Given that Leonardo clearly appreciated and celebrated the fact that courtly costume required sleeves to be made as independent garments held decorously in place by ribbon bows so as to permit undergarments to peep through; and, given that Leonardo lovingly depicted not only the varying thicknesses of the costume materials but every individual twist in the threads of the elaborately embroidered band in La Belle Ferronnière, how could he possibly – when working for same ducal master, at the same time – have been so negligent and indifferent in the execution of “La Bella Principessa’s” costume? Kemp acknowledges and offers excuse for the distinct poverty of the costume: “It may be that the restraint of her costume and lack of celebratory jewellery indicates that the portrait was destined for a memorial rather than a matrimonial volume.” In so-saying, he jumps out of one frying pan into another.

If “La Bella Principessa” was made after Bianca Sforza’s death, from whence did the likeness derive? One reason why Kemp settled on Bianca as the preferred candidate subject for “La Bella Principessa” was that while (disqualifying) likenesses of the other Sforza princesses existed, none survives for her – she is an image-free figure. Kemp offers no indication of a possible means for Leonardo’s (hypothesised) post-death conjuring of Bianca’s supposed likeness other than to claim that “Leonardo has evoked the sitter’s living presence with an uncanny sense of vitality.” This again begs the crucial question and fails to consider any alternative explanations for the image’s qualities. (We will be showing how the profile of “La Bella Principessa” could well have been a “portmanteau” composite image assembled from one particular work of Leonardo’s and from that of another, unrelated painter.)

The most strikingly “Leonardesque” feature on the costume of “La Bella Principessa” – the knot patterning around the (implausible) triangular slash in the outer garment – is a source of further concern and constitutes evidence of forgery. First, the motif on which much effort will have been expended, is brutally cropped along the bottom edge of the sheet, as if by a careless designer laying a photograph into a book. Why would any Renaissance artist, let alone Leonardo, design a complicated feature so as to “run it off the page”? Further, the illusion of form (created by lights and shades) in the patterning is feeble in the extreme for Leonardo – as when compared with his treament of relief seen in the above embroidered passage in La Belle Ferronnière, for example. Leonardo probably better understood than any artist in history the vital connection between a thing made and a thing depicted. He took bodies and organs apart to understand their construction and he sought to create mental models that would make the otherwise terrifyingly arbitrary and capricious forces of nature graspable if not checkable. Most seriously of all, as our colleague Kasia Pisarek has noted and reported, while the patterning present on “La Bella Principessa” matches none found in any work of Leonardo’s, it more closely matches that found in a carved marble bust by Gian Cristoforo Romano in the Louvre – see La Bella Principessa – Arguments against the Attribution to Leonardo”, Kasia Pisarek, artibus et historiae, no. 71, 2015. (To receive a pdf of Dr Pisarek’s article please write to: )

Michael Daley, 24 February 2012.

In Parts II and III we examine: the provenance of “La Bella Principessa” and the work’s problematic emergence from within the circle of Bernard Berenson; the claim by the forger Shaun Greenhalgh to have produced “La Bella Principessa” in Britain in the 1970s; the spurious “left-handed-ness” of “La Bella Principessa” and the low quality of, and the means by which the drawing was made…

Art’s Toxic Assets and a Crisis of Connoisseurship

30 September 2014

“Buy land”, Mark Twain advised, “they’re not making it anymore”. This logic ought to apply to the old masters but does not. Land makes sound investment not only because of its scarcity and its potential for development but because, in law-abiding societies, it comes fixed with legally defendable boundaries. Karl Marx, plundering English classical economists, held that all value is unlocked by human labour – but all labour does not generate equal values. In given periods and places all painters work pretty much with the same materials but their artistic transformations of those materials are various and unequal in accomplishment and merit. Such differences drive reputations and hence the market value of artists’ works but they do so in ways that are intrinsically problematic.

Artists’ reputations may or may not endure. With many surviving works the identities of authors are either not securely established or entirely unknown. In such cases paintings are appraised and then attributed to particular artists or schools. Attributions, however, are neither guaranteed nor immutable. They are made on mixtures of professional judgement, artistic appraisal, art critical conjecture and, sometimes, wishful thinking or deceiving intent. They remain open to revision, challenge, manipulation or abuse. The experts who make attributions exist in professional rivalry with one another (sometimes with vehemence) and while their disagreements are signs of art critical health, a consequence is that legal guarantees for attributions are untenable and non-existent, as some buyers later discover to their costs. Buyers are advised in the small print to beware and to proceed on their own judgement. With art, as we recently pointed out (see Endnote 1) it can be safer to buy a second-hand car than an old master painting (- and few people would dream of buying a house without legal searches and a structural survey.)

“Scientific” red herrings

In recent years attempts have been made to impart quasi-legal assurances to attributions by appealing to the authority of supposedly “scientifically verifiable” technical proofs. The exercise is vain and, technically, philistine: by its very nature, art is not reducible to scientifically quantifiable component parts. The technical evidence cult reflects a collapse of confidence in powers of connoisseurship on the one hand and a grab for cultural and institutional power by technocrats and bureaucrats on the other. The new hybrid discipline “Technical Art History” in which restorers, conservation scientists and curators pool expertises in attempt to arrive at professionally impregnable positions, has proved pernicious. Art-politically, this united front seeks to neutralise all charges of art critical and methodological failure with professional mystification and displacement activities – by fostering a “closed-shop” mentality and claiming that its mysteries are beyond the reach of any outsiders [2]. The new technocrats insufficiently appreciate that paintings are no more and no less than the products of artists who, working by brain, eye and hand, fix values and the relationships between values so as to produce specific and unique artistic effects that can be comprehended by others using eyes and minds in response. In the visual arts the visual should remain paramount – what you see is what it is about. Art loving viewers and professional art experts alike might be said to have duties of appropriate response to art itself and not to its shadows and encumbrances. It is the optically perceived quality of artists’ artefacts that drives reputations and market values. Understanding art is not the same thing as poking and poring over the component parts of its fabric – let alone presuming, as “restorers” (or now, “conservators”) perpetually do, to undo and redo its features at regular intervals. What matters is what you see, not what might be said or thought to lie under the surface.

Managing lapses of connoisseurship

This is not, of course, to say that technical examinations can serve no purposes. Rather, it is to say that in matters of art attribution and appreciation technical examinations of the physical composition of works might supplement informed visual appraisals but they cannot stand in lieu of them. Nor can the supposedly disinterested and neutral character of technical examinations themselves be taken at face value. In practice, with every technical investigation and its resulting “findings”, someone, some institution, some interest group, has commissioned/conducted the exercise and controlled its dissemination. Paintings in powerful institutionally-protected locations (particularly major museum) can be afforded dispensations from otherwise injurious findings [2]. It sometimes seems that just as banks are now too big to be allowed to fail, so big museum attributions cannot be allowed to fall, whatever evidence and arguments accumulate against them [3], for fear of undermining public, political and art market confidence.

Follow the money and look at the drawings

Concerning the frequency of art world upgrades, it would seem easier to grow old master drawings than paintings. Where only 250 sheets of drawings were attributed to Michelangelo in the 1960s, today that oeuvre has been expanded to over 600 sheets. Although drawings do not command the high prices of paintings they can greatly assist their attributions. In the late 1920s a firm of antiquarian dealers in Holland, R.W.P. de Vries of Amsterdam, sold a number of old master drawings some of which have ended in museums, and two of which concern us here (Figs. 1 and 2). Neither of these had a provenance (i.e. a proven history of previous ownership). Both had simply materialised in the dealers’ hands with old master attributions. The first sold in 1927 for 26 florins (guilders), some € 235.80 at today’s values. The second sold two years later for 750 florins, some €6,801.91 today. The first was attributed to van Dyck, the second to Veronese. Neither attribution survived and the original perplexing ratio of value between them (which approached thirty to one) has reversed dramatically.

The Veronese attribution crashed in 1984 when Richard Cocke published his catalogue raisonné Veronese’s Drawings and dismissed the drawing with the single (apt) sentence: “The heavy forceful cross-hatching in the drapery and the forms of the head and hands have nothing to do with Veronese.” That drawing sold in 1991 at Christie’s for £7,000 as “attributed to Agostino Carracci”. In contrast, the former van Dyck drawing morphed into the work that sold at Christie’s on July 10th as an autograph Rubens ink sketch for a world record Rubens drawing price of £3,218,500. The former “van Dyck” has thus enjoyed a 14,000-fold increase of value since 1927.

The extraordinary success of the van Dyck that is now a Rubens was due only in part to Christie’s masterful promotion. It was very much on the strength of its current art-historical position that the drawing was drum-rolled as the starred lot in a sale of part of the prestigious I. Q. van Regteren Altena drawings collection. Most helpfully of all, the drawing was precisely characterised as Rubens’s “first thought” preparatory ink sketch for the National Gallery’s Samson and Delilah painting (Fig. 4). Notwithstanding its anomalous traits (see our previous post), its artistic shortcomings and its dubious provenance, the drawing remains bolstered by its crucial allotted role in a sequence of three Samson and Delilahs, two of which have been acquired by museums (Figs. 3 & 4). Although Christie’s July 10 sale realised more than twice its highest estimates and broke many records for individual artists, only one of the top ten works went to an art gallery or museum. Two were sold on to the trade. Seven, including the Samson and Delilah drawing, went to anonymous individuals.

Making four Rubens’s

Christie’s catalogue entry burnishes the drawing’s pedigree with upbeat optimism. It is said for example: “When I. Q. van Regteren Altena bought the drawing in 1927, he listed it in his inventory under its traditional attribution to Sir Anthony van Dyck (1599-1641). That attribution also accounts for an earlier owner’s inscription of the letters ‘V.D.’ in the lower left corner.” What traditional attribution? Which earlier owners? Christie’s account of the provenance begins: “with R.W.P. de Vries Amsterdam; from whom purchased by I.Q. van Regteren Altena on 20 December 1927 for 26 guilders (‘387.t. A. v. Dijck. Samson & Delilah’)”. And that is all. There had been no previous owners and no evidence exists of any “traditional” reception as a van Dyck – or anything. Any suppositions aside, all that can safely be said is that this drawing emerged from nowhere at a time when forgery was rife and the art world suffered from what Bernard Berenson [!] described as “the universal tendency to ascribe a given work of art to the greatest artist to whom wishful thinking and excited imagination can ascribe it.” (“Essays in Appreciation”, 1958, p. 95.)

Christie’s entry continues: “With the emergence of the finished painting and the connected oil sketch the drawing’s significance rapidly became apparent.” There was no rapidity and the claimed significance is mythic. The supposed second stage oil sketch or modello did not appear until 1966. The claim that, “The picture of Samson and Delilah was only rediscovered in 1929”, also misleads. The painting was not “rediscovered” as a Rubens. It had never been a Rubens. When it appeared in 1929 it was, just like the ink drawing three years earlier, without provenance and it was not judged a Rubens by its German dealers, Van Diemen and Benedict, who were offering it as a Honthorst. It was later upgraded to Rubens in a certificate of authenticity by Dr Ludwig Burchard and it then sold in 1930 to August Neurburg, a German tobacco magnate.

Burchard was a leading Rubens scholar, but today his attributions have a notoriously poor record [4]. Far from the ink drawing being corroborated as a first stage sketch by the arrival of the painting, Burchard had upgraded the painting on the authority of the drawing which he had himself upgraded to Rubens in 1926. In Christie’s catalogue the drawing’s “Literature” begins with Burchard’s attribution: “L. Burchard, ‘Die Skizzen des jungen Rubens’ in Sitzungsberichte der Kunstgeschichtlichen Gesellschaft, Berlin, 8 October 1926, p. 30, no. 2.” At that date no one had previously owned or discussed the work. Burchard thus upgraded a drawing that had never been exhibited and was in a dealer’s hands without any provenance. Notwithstanding his claims on behalf of the drawing, in 1927 both the dealer selling and the collector buying still held it to be a van Dyck.

When the modello eventually appeared in 1966 it had no provenance. Its history consisted of a hearsay account (from the anonymous lady vendor) of an ancestor said to have bought the work for a few shillings in an antique shop in York during the 1930s because she liked the frame. This supposed Rubens oil sketch had been painted on a support that is found in none of the artist’s oil sketches – on a soft, conifer wood, not on his customary oak panel. Its appearance was, for a Rubens oil sketch, disturbingly close in design and effects to those of both the ink drawing and the finished painting (see Figs. 2, 3 and 4). Its arrival completed an unicum in Rubens’ oeuvre: a suite of stages of work without evidence of development. Notwithstanding that problem, the modello on the wrong wood was given to Rubens by Christie’s themselves, to join the company of a panel painting whose back, it later emerged, had disappeared in an operation for which no one acknowledged responsibility, and a drawing whose back was concealed by being pasted onto a second sheet even though it bore drawing itself. The modello sold to a London gallery for £24,000, going to a private collector before passing through Agnews to the Cincinnati Art Museum in 1972. The last of the trio to emerge, this technically problematic work-without-provenance was the first to achieve museum status. At some point, pieces of wood were removed from its sides (creating a closer compositional alignment with what is now the National Gallery painting) and, at another, the Cincinnati museum claimed the panel to be oak. Presently the wood is not identified, the work being described as on “panel”.

Why? Why? Why? Delilah?

In July 1980, the supposed third stage, the Samson and Delilah painting, was sold by Neurburg’s heirs through Christie’s to Agnews, acting on behalf of the National Gallery, for a then Rubens world record price of £2.53m. In 2002, with two parts of the Samson and Delilah trio now secure in museums and the third in a respected private collection, Sotheby’s sold a painting, The Massacre of the Innocents (see Fig. 13), as an autograph Rubens on the back of its perceived shared characteristics and collections history with the National Gallery’s Samson and Delilah for £49.5m, to Lord (Kenneth) Thompson. Even though those paintings are riddled with problems (see “Is this really a Rubens?” Michael Daley, Art Review, July/August 1997, and “Is this a Rubens?” Michael Daley, Jackdaw, October 2002), and the Samson and Delilah had been challenged for over a decade [5], the price was an outright old masters’ world record. Thompson loaned the Massacre to the National Gallery and then bequeathed it to the Art Gallery of Ontario, Toronto, thereby making it publicly available and greatly enhancing its pedigree. Thus, today, three high valued well-placed but individually problematic museum Rubens’s owe their positions to a belated acceptance of Burchard’s initial attribution of what is still a privately (but now anonymously) owned ink drawing.

Who cut Samson’s toes?

The reason why all of these subsequent Rubens upgrades rest on the authority of this ink drawing is because of a glaringly anomalous feature in the National Gallery painting – the fact that the toes of Samson’s right foot are cropped by the edge of the picture. This was not because the panel had been trimmed at some point. Rather, it is because the painting simply stops disturbingly, inexplicably, at the beginning of the toes. Thus, without the drawing’s seeming testimony that Rubens had planned to crop Samson’s toes by cropping his own initial design within a precisely drawn ruled box that anticipated (even before he had executed an oil sketch) the final format of what is now the National Gallery painting, that painting could never have been attributed to him. This is so for reasons that are implicit in Burchard’s 1930 certificate of authenticity. It read:

“The photographed painting on the other page is one of Peter Paul Rubens’ major works from the time of the master’s return from Italy. It must have been painted in 1609 or 1610. With Rubens’ agreement, Jacob Matham reproduced the painting with a copper engraving around 1615. As witnessed by the inscription of the painting, the picture at that time was in the possession of Antwerp mayor Nicolas Rockox. Indeed, the inventory of Nic. Rockox’ estate, dated 19 Dec. 1640, lists the picture as “Eene schilderne…(Annales de l’Academie d’Archaeologie de Belgique, Anvers 1881, p. 437). On pp. 143-44 in vol. I of 1886, the five-volume catalogue of Rubens’ work by Max Rooses, the painting is described in detail as number 115, based on the Matham engraving and mentioning the Rockox inventory. The picture itself remained as unknown to Rooses as to all literature since. It is further notable that a picture of an interior by Frans Francken (Pinakothek Munchen No 720), which appeared to be of mayor Rockox’s living room, showing the painting in pride of place above the mantelpiece, while in an adjoining room is the picture of the “Doubting Thomas” which we know Rubens painted for Rockox. According to S. Hartveld of Antwerp, the room with the mantelpiece exists even today in the Kaiserstraat in Antwerp where Frau Gruter-Van der Linden now lives in the Rockox house. A sketch for the Samson picture (pen, varnished, 16.4 x 16.2) is in Amsterdam in the collection of Mr J.Q. Regteren, Altena. The picture is in a remarkably good state of preservation, with even the back of the panel in its original condition.” [By courtesy of the National Gallery Archives Department.]

Note, even as Burchard asserts that this is the original painting of the subject that Rubens is known to have made shortly after 1608, he acknowledges that the original painting itself had universally been understood to have been lost since 1641. (To this day, despite detailed and sustained searches, nothing connects the present version to the original painting.) Crucially, Burchard also acknowledges that the appearance of the original Samson and Delilah had been recorded in two contemporary copies, one of which had been supervised by Rubens. Both of these copies by two artists who likely worked decades apart, testify that Samson’s original right foot had not been (improbably) cropped at the toes, as in the National Gallery version, but had originally been painted intact and set comfortably inside the composition and consistently with the artist’s known manner. See, for example, the almost contemporary, probably pendant (and near mirror-image compositional group) Cimon and Pero – “Roman Charity”, at Fig. 9.

A perplexing silence

It was in defiance of such hard historical testimony that Burchard claimed his own upgraded ink drawing to be not only by Rubens but, specifically, to be his preliminary sketch for the former Honthorst painting that is now in the National Gallery. When attributing that painting to Rubens Burchard executed a sleight of hand by implying but not stating that the ink drawing (which had only recently been sold as a van Dyck) was by Rubens. The truth is this ink drawing-from-nowhere and without-history had needed to exist if the Berlin Honthorst were to be presented remotely credibly as a Rubens. Had Burchard sincerely believed that the cropped-foot drawing was Rubens’ original ink sketch, he would have felt himself the agent of a remarkable double art historical coup: first, for having identified a famous masterpiece that had been lost for 289 years; second, for having further established that both of the contemporary copies of that original Rubens’ painting (through which it had been known for centuries), had been compositionally misleading in identical manners.

Conspicuously, Burchard trumpeted neither of these “discoveries” [6]. His diffidence contrasts markedly with the reaction of the day’s leading Vermeer scholar, Dr. Abraham Bredius, who believed in 1937 that he had found an unknown Vermeer (in what was the first of a stream of Han van Meegeren fakes). Firstly, Bredius’ certificate of authenticity was ecstatically and unreservedly fulsome: “…I found it hard to contain my emotions when this masterpiece was first shown to me and many will feel the same who have the privilege of beholding it. Composition, expression, colour – all combine to an unity of the highest art, the highest beauty”. Secondly, he rushed news of his discovery onto the scholarly record via the Burlington Magazine (“A New Vermeer”, November 1937).

If Bredius betrayed credulousness as an eighty-two year old scholar, what of Burchard’s manoeuvres as a forty-four year old at the peak of his powers? It can only be said that suspicions are in order. When, shortly after the First World War, the great German scholar, Wilhelm von Bode, was reproached for having certificated an implausible Petrus Christus, he replied, “You don’t understand the intricacies of the German language. After a brief description of the subject I say ‘I have never seen a Petrus Christus like this!'” (- “The Partnership”, Colin Simpson, 1987, p. 240). One must suspect that Burchard’s twinned and circular Rubens attributions were made sotto voce out of fear that his “attributional” heist might be exposed by anyone with an alert eye who appreciated that it is surprisingly common for later copies of original works to be cruder compositionally cut-down and abridged versions – and who would, therefore, recognise the “Honthorst” as a prime member of that type.

We have found that not only are such insensitively truncated pictures frequently encountered (in Rubens twice-over with the Samson and Delilah and the Ontario Massacre, and in artists like Leonardo, Raphael, Caravaggio and Annibale Carracci – see opposite) but, also, that with a little effort they can in almost every instance be shown to post-date the superior models and prototypes from which they derive. As shown opposite, in copyists’ hands, no part of an original composition can be considered sacrosanct. As well as toes, dogs’ noses and cupids’ wings, even portions of dead infants have been cropped to fit pre-existing images to new supports and formats. Mistaking a copy for an absent original is one thing. Disregarding clear and contrary historical evidence, as Burchard would seem to have done, is another altogether. Knowingly elevating adulterated versions to a master’s oeuvre pollutes the well of scholarship and ultimately threatens the credibility of the field.

Such lapses of critical judgement are as common in appraisals of restorations as they are in the making of attributions. How much or little of an original surface has survived the vicissitudes of time and “conservators” attentions might seem a lesser matter but it is not. Professional art critical failures to spot the tell-tale differences between autograph and studio works are the twins of failures to recognise restoration-induced injuries. The differences of states within individual works can be as pronounced as the differences between autograph and studio works (see Figs. 28a, 28b, 29 and 30). Failures of judgement in both areas are frequently found in even the most high-ranking individual scholars.

Making two Caravaggios in one decade

Within little more than a decade the late Sir Denis Mahon upgraded two pictures to autograph Caravaggio status. This might seem unremarkable given that Mahon was a prolific finder/maker of old masters. What is remarkable is that he did so with two versions (of more than a dozen) of the same painting – Caravaggio’s The Taking of Christ. This Caravaggio survives in two formats, one being a truncated version of the other. Mahon managed to endorse one version of each type, doing so in the wake of two “investigative” restorations in which each team claimed revealed authenticity on the basis of its own “discoveries”. (Mahon had serious form in the double attributions stakes – we discuss opposite a painting of Annibale Carracci where he authenticated one version and later suavely switched to another, less abridged, picture. See Figs. 25-30.)

During the first restoration in 1993 in Dublin, a long-attributed Honthorst copy was found to have been made largely without revisions and it was declared the original autograph Caravaggio by Mahon precisely by virtue of its revisions-light painterly fluency. This version was of the truncated type. In Rome in 2004 Mahon conferred autograph Caravaggio status on a work from Florence (where acquired from the Sannini family) that was found to have been made with many and major revisions taken to be “serious afterthoughts as was Caravaggio’s wont”. This version was composed in the larger format and Mahon reportedly said he had “no doubt that this was now the original work”. Dublin was not best pleased and Mahon promptly rowed his position back and claimed that both versions were now original but that one was rather more so than the other. (See “New twist in the tale of two Caravaggios”, Daily Telegraph, 17 February 2004; “A dangerous business”, Michael Daley, letter, Daily Telegraph, 19 February 2004; and, “The real Caravaggio is . . . both of them” Daily Telegraph, 20 February 2004.)

Like the two R.W.P. de Vries of Amsterdam drawings, the two “autograph” Mahon Caravaggios have enjoyed unequal fortunes. In 1993 the (revisions-light) Dublin Caravaggio was loaned to the National Gallery in London and then, permanently, to the National Gallery in Dublin. The later 2004 Florence/Rome Caravaggio with numerous major revisions and other “cast iron” technical proofs enjoyed no institutional protection, being still in private hands. Its cause seems to have fallen into abeyance following legal disputes over ownership. In 2005 the initial 1993 “discovery” of the now institutionally protected Dublin Caravaggio (Mahon enjoyed a long-standing relationship with the National Gallery in London, as a trustee and as a generous benefactor-in-waiting) became the subject of an illuminating, if somewhat parti pris book, “The Lost Painting”, by Jonathan Harr.

In an epilogue, Harr has described a falling-out over the ownership of the Florence/Rome version. Technical examinations of the painting were ordered by court prosecutors without the knowledge of the owners. They were carried out by Maurizio Seracini, a leading private technical diagnostician who has examined something like half of Caravaggio’s output. The pigment Naples Yellow, which contains the metal antinomy, was found. Because that pigment is presently said not to have been used on paintings before 1630 (or “from around 1620”, according to Wikipedia), and therefore twenty years after Caravaggio’s death in 1610, Seracini held the painting inauthentic. Harr accepts the force of this technical testimony and, concluding that Mahon had demonstrably blundered in his support for the Rome/Florence painting, imagines that that old scholar’s long-time adversary, Roberto Longhi, might now be enjoying “a mirthless laugh” over Mahon’s discomfiture. The conclusion was hasty and perhaps too trusting of technical testimony.

It is certainly the case that the presence of a modern, manufactured pigment within the fabric of a supposedly old painting can safely be considered fatal to an attribution. However, Naples Yellow is not a product of a known and precisely dated modern manufacture – such as Prussian Blue of 1704 – it is ancient and greatly pre-dates Christ. Harr acknowledges that the pigment is found on a painting of 1615 by Orazio Gentileschi – just five years after Caravaggio’s death. Harr further reports that traces of this pigment had been found on another Caravaggio, his Martydom of St Ursula, which is owned by Banca Intesta in the Palazzo Zevallos, Naples. He reports a suggestion that the offending material might have come from an 18th century restoration that had subsequently been removed. Such hypothetical exculpation would only be necessary if claims that Naples Yellow could not have been used by anyone before 1630 were Gospel and if the painting’s attribution was insecure. Neither is the case. The Martyrdom is one of Caravaggio’s most reliably and completely documented works so there can be no question about its authenticity. Further, it was almost certainly his last work. It was recorded as still being wet in May 1610. If this painting contains antimony, and unless evidence exists to support the former existence of a now entirely disappeared 18th century restoration, we should accept that this material has now been found in two Caravaggio paintings and adjust the technical literature chronologies accordingly.

In this episode, we see that negative hard “scientific evidence” can be discounted on the basis of assumptions, hunches, and suspicions. We also see that the claimed chronologies of materials within the literature of technical analysis are moveable and, only ever, provisional feasts. (For such chronologies to be considered reliable it would be necessary for every painting in the world to be analysed at the same time by the most advanced technologies – and even then, subsequent technical advances would require further examinations: it is common for old formerly “advanced” tests to be re-run in conservation departments when new and improved apparatus become available.) We have asked Seracini, in the light of Harr’s comments, if “it is still the case that the presence of antimony is considered an absolute technical disqualification in paintings made before 1630?” Meanwhile, Jacques Franck, the Consulting Expert to The Armand Hammer Center for Leonardo Studies at The University of California, Los Angeles, advises that:

“The best scientific bibliographic reference concerning the history and chemistry of pigments over here is: J. Petit, J. Roire, H. Valot, “Des liants et des couleurs pour servir aux artistes peintres et aux restaurateurs”, EREC éditeur, Puteaux, 1995. Regarding Naples yellow, it says: ‘(Lead antimonate yellow) was rediscovered in Europe at the end of the Middle-Ages and was later mentioned in a document dating from 1540, “Pirotechnia”. The oldest recipes, written in 1556-1559, were supplied by Cipriano Piccolpaso…who was a painter of ceramics”

Although those recipes were indeed written primarily in connection with ceramics, given that they existed before Caravaggio’s birth (1571) it should never have been insisted that knowledge of them could not have been obtained by contemporary painters. As it happens, a study on Lorenzo Lotto’s pigments was made in connection with the exhibition “Lorenzo Lotto” (Venezia, 1480 – Loreto, 1556-57) at the Scuderie del Quirinale in Rome in spring 2011. On that occasion, more than fifty Lotto paintings spanning from 1505 to around 1556 were studied using non-invasive techniques by Maria Letizia Amadori, Pietro Baraldi, Sara Barcelli and Gianluca Poldi. The authors’ report (pages 2 and 19):

“About yellows, he uses both lead-tin and lead-antimony (Naples yellow) pigments, the latter found by XRF, in works starting from 1530 to the last years: it can be related to the ‘zalolin da vasarj’ cited by Lotto in 1541 in his account book (Libro di spese diverse)”, and, “As XRF analyses show, in some works, starting from 1530 to the last years of the century, also lead-antimony (Naples yellow) pigments, can be found, together with the previous yellow or almost alone: they can be related to the “zalolin da vasarj” cited by Lotto in 1541 in his account book (Libro di spese diverse).”

Thus, the presence of antimony would seem not to have given grounds for dismissing the Florence/Rome version of the Taking in the courts. Perhaps we can see that it might have been more to the point for the courts to require the production of the best possible photographs of as many of the versions as possible to permit visual comparisons of the two rival versions. There are many indications of the limitations of modern conservation practices to be had in Harr’s fascinating account. On page 169 he describes an encounter between the Dublin National Gallery of Art’s two picture restorers, Andrew O’Connor and Sergio Benedetti (who had re-attributed the Hontorst Taking to Caravaggio, and who had experienced “a fleeting moment of doubt” about his attribution while cutting ever larger ‘windows’ through the painting’s varnish):

“One day, about three weeks after the painting’s arrival, O’Connor and Benedetti crossed paths in the studio. Benedetti was staring at the painting. He stood with his arms crossed, his eyes narrowed in concentration, his mouth compressed into a frown. ‘Look at the arm of Judas’, Benedetti said to O’Connor. ‘What do you think?’ O’Connor studied the painting. ‘What are you getting at?’ he asked. ‘It seems too short, doesn’t it?’ said Benedetti. It did…O’Connor realised that Benedetti was wrestling with his doubts. ‘Well’, said Benedetti finally, ‘he wasn’t a perfect anatomist. He made other errors like this. In the Supper at Emmaus, the apostle’s hand is too large.’”

In this recollection we might be witness to a double failure of art critical methodology. Given his doubts, Benedetti might have assembled all available photographs of the many versions of this painting to determine whether or not the short-coming that concerned him was unique or common to (some or all) other versions. A greater lapse may be evident in the fact that while Benedetti expressed anxiety over the arm of Judas, he seems not to have done so over the compositionally and emotionally more important advancing left arm of the fleeing St John who is seen behind Christ and Judas. In the Dublin version, the arm of St John is cropped above the elbow and not above the wrist as it is in the Florence/Rome version. (On the compositional function of the arm in the Florence/Rome version, see comments at Figs. 21 and 22.)

To repeat what should be self-evident: pictures are made to be looked at. When, as with this Caravaggio, multiple versions exist we should make hard detailed visual comparisons of each against the others, if necessary (and it could hardly be otherwise when so many versions exist) by photographic means. When later copies or engravings exist we should make careful comparative estimations of their relationships to the various contenders. Whenever there are cut-down versions of more expansive compositions, we should always consider which state is likelier to have been the primary and which the secondary one. Visual comparisons in attributions, as in restorations, are of the essence. They should never be neglected, let alone discounted, on the authority of some technical evidence that may or may not be soundly framed; that may or may not be selective or loaded in its presentation; and, that will, in any event, soon be rendered obsolete by more up-to-date equipment. The informed human eye is our best “diagnostic tool” in the study of art and will remain so no matter how much money and resources might be thrown into technical studies. It remains the greatest tragedy that Bernard Berenson so badly debased his own critical currency with his shady Duveen dealings. On the primacy of the visual in visual art forms he was peerless:

“I am here concerned with names in painting. When I pronounce the words Giotto, Michelangelo, Leonardo, Giorgione, Durer, Velazquez, Vermeer, Ingres, Manet, Degas and hundreds of others, each stands for certain qualities which I expect to find in a painting ascribed to them. If the expectation fails, then no argument, no documentary evidence, be it biographical, historical, psycho-analytical, or radiological and chemical will persuade me.”

That was and is how it should be.

Michael Daley


1 The Times, letter, 13 August 2014:

“Sir, Gerald Fitzgerald (letter, Aug 12), misses an important point when calling for a tiny levy on art sales to fund an independent centre for provenance research. Although such a levy might cost only .05 per cent of annual art sales, currently standing at some $60 billion, if effective, such a centre would reduce the supply of works on the market by something like 40 per cent – at least in the view of the late Thomas Hoving, a former director of the Metropolitan Museum of Art, New York. The art world is very quick on its feet: when calls were made in the 1930s for an independent centre of art restoration research, then director of the National Gallery in London, Kenneth Clark, promptly established a department of conservation science in order, as he later confessed, to ‘have in the background what purported to be scientific evidence to “prove” that every precaution had been taken’. Although self-policing may be an unrealistic ambition, governments could help considerably and at little cost by making it a statutory requirement that vendors should disclose all that is known and recorded about the provenance and the restoration treatments of works of art. As things stand, it can be safer to buy a second-hand car than an old master painting.”

Michael Daley, Director, ArtWatch UK, London

2 The Massacre of the Innocents which came up at Sotheby’s on 10 July 2002 as a very recent Rubens upgrade is a case in point of misleading assurances and over-ridden technical evidence. In a long sale catalogue entry it was said that technical analyses and condition reports had been commissioned and that these were available on request. The implication was clear: we have exercised all possible due diligence and this painting has emerged with flying colours. That implicit reassurance evaporated on a close reading of the material – as we reported in the October 2002 Jackdaw (“Is this £49.5 million painting by Rubens?”). The reports were, by their nature dense and couched in technical language. Nonetheless they clearly contained information that was highly injurious to the attribution and to the picture’s claimed early dating of c. 1609-11. One technical fact alone should have sunk the attribution. It was found in the last paragraph of the last report. As we put it: “The author of a report on the tree-ring dating…concludes that a date of execution for the picture only becomes ‘plausible from 1615 upwards’.” In other words, the panel on which this picture was painted could not have been manufactured at the time the picture is said to have been painted – and this dating could not be amended because, like the Samson and Delilah, the picture was only remotely credible on stylistic grounds if seen as the product of a (fancifully claimed) brief stylistic abberation in Rubens’ oeuvre said to have occurred on his immediate return from Italy in 1608. As well as being on wood that was too recent, the picture contained the wrong materials: “A pigment, orpiment, that is found in no Rubens is present here. A second pigment, smalt, said to have been in use ‘mainly in the mid-seventeenth century’ and which seems only to be found in Rubens’ later works is also present. The orpiment yellow is anomalous not only in its presence but in its manner of application – it is mixed with lead-tin yellow. Such a combination is said to be ‘unusual since it was considered unstable’ and, even, to be a practice ‘not encountered in 17th century works’”. This was not just a twice-over dead attribution: “Speaking of Rubens’ debt to classical sources, the anonymous author of the catalogue entry correctly concedes, ‘one of the background figures appears to derive from the Borghese Gladiator’. There follows immediate self-disavowal: ‘it cannot’ so derive, he/she contends, because ‘though famous in subsequent centuries, the Borghese Gladiator was not excavated until late in 1611”. This painting on the wrong (too recent) wood, with what would normally be considered disqualifying (out of period)materials, and which contained a miraculous allusion to a future event, was presented to the world as a major art historical discovery. That “discovery” had taken place very shortly before the sale. The upgrading of this centuries old studio work had been made by just five experts only three of whom were identified. We put the question: “Can it be right that we are all being asked to share this leap of faith when the experts, displaying a seeming ignorance of – or disregard for – so much germane material evidence, have yet to declare their hands or publish accounts of their vital endorsements?”

3 Jonathan Harr reports in his 2005 account of the upgrading of a Honthorst to Caravaggio (“The Lost Painting” p. 222) that when the picture, The Taking of Christ, was examined at the National Gallery in London it was found that its ground (priming layer) was anomalous: Ashok Roy, the head of science, observed, as Harr reports, that “the composition of this particular ground was strange – ‘bizarre’ was the word used. It contained reds and yellows and large grains of green earth, a pigment composed of iron and magnesium. Grounds usually contained lead-based pigments and calcium, which dry quickly. Green earth dries slowly. This primer looked to Roy like a ‘palette-scraping’ ground – the painter had simply recycled leftover paints from his palette board to make the priming layer.” Well, yes, someone evidently had – but what in Roy’s detailed technical analysis of the ground might have suggested that on this occasion Caravaggio had departed from his own habits in order to do so? When the painting was exhibited in a special exhibition (“Caravaggio ~ The Master Revealed”) at the National Gallery of Ireland in 1993, the catalogue gave a different spin to Roy’s research: “Analyses have shown that the ground is composed of a brown pigment, heterogeneous and unevenly applied. Several pigments were mixed with it: lead white, red and yellow ochre, umber and large granuli of green earth.” On a casual reading: impressive and reassuring technical detail and expertise. No mention of bizarreness. No acknowledgement of what was for Dr. Roy, a perplexing departure from Caravaggio’s known practices. On page 160 Harr reports that Sergio Benedetti (the Dublin National Gallery of Art restorer who first made the attribution)“saw immediately that the painting had been relined at least once before” and judged the present lining canvas to be at least a hundred years old. In the National Gallery catalogue Benedetti reported that “the picture has undergone at least three interventions, probably accompanied each time by a relining of the canvas. One of these linings caused a shrinking of the surface in some limited areas.” What is not said is that Benedetti two of the three-plus hypothecated linings had been made by Benedetti himself the first having caused cracking. Harr reports that after the first lining “There is much dispute about what happened next. For Benedetti, restoring the Taking of Christ was the greatest moment in his professional career, and to this day he adamantly denies that he had any problem relining the painting. O’Connor and others at the gallery, however, tell a very different story. According to them, he came close to ruining the painting.” Andrew O’Connor, the Gallery’s chief restorer, said that Benedetti had elected to use a densely-woven Irish canvas rather than wait for an appropriately matching loose-weave canvas to arrive from Italy. When Michael O’Olohan, the gallery’s photographer, who had made detailed photographic records of every inch of the picture’s surface, saw the painting immediately after its first relining, he could not believe his eyes and recalled “There were areas that had hairline cracks, like a sheet of ice that has started to melt, a flash of cracks all over it. I was shocked. I couldn’t believe it.” O’Connor explained that because the Irish canvas was densely woven, “it did not absorb the [water-based] glue at the same rate as the old Italian canvas. It had not dried properly and had contracted, pulling with it the Italian canvas and raising ridges, small corrugations, in the paint surface. Along these corrugations, the paint layer had cracked and lifted.”

4 In the ArtWatch UK Journal No. 21, (“The ‘Samson and Delilah’ ~ a question of attribution”), Kasia Pisarek wrote: “Dr. Ludwig Burchard was an active Rubens attributionist in Berlin before the Second World War and in London afterwards. Several paintings formerly attributed to Rubens’s school or studio or even to another artist (such as Sampson and Delilah), were reinstated by Burchard as by the master. I traced many of his attributions – he was not infallible in his judgement and changed his mind. Surprisingly, over 60 pictures attributed by Burchard to Rubens were later down-graded (in Corpus Rubenianum) to studio works, copies or imitations.”

5 The principal challenges to the attribution came from two artist/scholars, Euphrosyne Doxiadis, author of the award-winning 1995 book “The Mysterious Fayum Portraits: Faces from Ancient Egypt”, and Kasia Pisarek whose 2009 doctorate dissertation was entitled “Rubens and Connoisseurship ~ On the problems of attribution and rediscovery in the British and American collections (late XIX – XX c.)”. In 1986 Euphrosyne Doxiadis began researching the painting’s credentials with fellow art students Steven Harvey and Siân Hopkinson. Their findings were compiled in a report submitted to the National Gallery in 1992 and which is now held in the painting’s dossiers. (It is also available online at this site: Their challenges to the attribution were covered in reports in the Times (“Artists raise fresh doubts on gallery’s Rubens masterpiece”, 22 September 1996, and “Expert denounces National Gallery’s Rubens”, 25 November 1996), and in The Independent on Sunday (“Tell-tale sign that £40m Rubens could be a copy”, 21 May 2000). Researches begun in 1990 by Kasia Pisarek prompted two articles on 5 October 1997 by the Sunday Times’ art critic, Waldemar Januszczak (“A Rubens or a costly copy?” and “National’s £40m Rubens could be fake”). In the latter article, the then director of the National Gallery, Neil MacGregor, conceded that “the scholar raises some serious questions that I cannot easily answer”.

6 As Dr. Pisarek put it in the ArtWatch UK Journal 21 (“The ‘Samson and Delilah’ ~ a question of attribution”): “Both the rediscovery and the sale of this early Rubens masterpiece should have been well publicised in the press, yet there are no records of it in any art magazine (I checked most art journals published in 1929-30). However, other, even minor, Rubens discoveries could easily be traced (‘Forgotten Rubens found in Austria’ – Art News, 1930; ‘Van Diemen sells notable Rubens’ – Art News, 1931 etc.) Strangely, the Samson and Delilah was not even included in Valentiner’s ‘Unknown Masterpieces’, co-edited with Burchard, and published in 1930, which presented important little-known and rediscovered paintings. Dr. Burchard only wrote about it briefly in 1933, and only in a short note.”

Comments may be left at:

Printer-friendly PDF version of this article

Above, Fig. 1: A chalk drawing that originated with the firm R.W.P. de Vries of Amsterdam in 1929 and sold as a Veronese for 750 florins (guilders) or some €6,801.91 at today’s exchanges.
Below, Fig. 2: An ink and wash drawing that originated with the firm R.W.P. de Vries of Amsterdam in 1926 and sold the following year as a van Dyck for 26 florins (guilders), or some €235.80 at today’exchanges
Above, top, Fig. 2: The ink and wash drawing sold on 10 July 2014 as a preliminary ink sketch for Rubens’ Samson and Delilah painting.
Above, middle, Fig. 3: An oil painting on panel that sold at Christie’s for £24,000 in 1966 as Rubens’ oil sketch (or modello) for what is now the National Gallery’s Samson and Delilah painting.
Above, Fig. 4: The oil painting on panel sold for £2.53m at Christie’s in 1980 to the National Gallery as Rubens’ original Samson and Delilah.
The three works above are claimed to comprise an entirely autograph suite of successive stages of Rubens’ treatment of Samson and Delilah.
Above, top, Fig. 5: An engraved copy (here as a mirror image) made in c. 1611-14 of Rubens’ (now lost) original Samson and Delilah painting.
Above, Fig. 6: A detail of a painting (made before 1640) by Frans Francken of the original Rubens Samson and Delilah as it was displayed in the home of his friend and patron Nicolaas Rockox. This painting and the engraving above both show that Samson’s right foot was originally intact and set comfortably away from the edge of the painting.
Above, top, Fig. 7: A larger detail of Frans Francken’s c. 1630-35 oil painting A Feast in the House of Nicolaas Rockox, showing the original Rubens Samson and Delilah in pride of place in Rockox’s home.
Above, Fig. 8: The National Gallery Rubens’ Samson and Delilah when on loan in 2007 to what is now the Rockoxhuis museum, Antwerp.
Above, top, Fig. 9: Rubens’ painting Cimon and Pero – “Roman Charity” of 1611-13 (here as a mirror image) in the Hermitage, St. Petersburg.
Above, Fig. 10: The National Gallery Samson and Delilah painting.
Comparison of the two works shows in the former, the exceptional grace, composure of design and warmth of colouring for which the artist is revered, while the latter asserts an uncharacteristic stridency that required the National Gallery to posit a “special-but-brief” stylistic Rubens interlude.
Above (left) Fig. 11a: Cimon’s feet, as painted by Rubens. Above (right) Fig. 11b: The right-hand edge of the National Gallery Samson and Delilah.
It is not credible to suggest than an artist so brilliantly attentive to feet and hands might have painted the foot encountered in the National Gallery.
Above, top, Fig. 12: The version of Rubens The Massacre of the Innocents that is owned by the Musée des Beaux-arts in Brussels.
Above, Fig. 13: The version of Rubens The Massacre of the Innocents loaned to the Art Gallery of Ontario, Toronto.
Just as the National Gallery’s The Virgin of the Rocks (below) is a cut-down replica version of the Louvre’s Leonardo original, so the Ontario Massacre of the Innocents is a cut-down version of the larger canvas at the Musée des Beaux-arts in Brussels. Although now said to be a “studio replica” the latter was judged original by such eminent Rubens authorities, as Gluck, Held, Van Puyvelde and Michael Jaffé.
The cropping of motifs in the Ontario version seems particularly insensitive as it includes the two murdered infants who, in the Brussels version, were depicted whole and set (like Samson’s original toes) comfortably inside the edge of the painting. How likely is it that Rubens would have cropped his figures in this manner or, if by chance he had, that a copyist would presume to extend and make whole his composition ?
Above, Figs. 14a and 14b. The regretably unequal photographic quality of this comparison does not mitigate the disturbing cropping of the infants in the Ontario version (left) which, like the National Gallery Samson and Delilah, spent many years as studio copy in the Liechtenstein Collection.
Above, top, Figs. 15a and 15b: Left, the Louvre’s original Leonardo da Vinci The Virgin of the Rocks; right, the National Gallery’s later version of the painting.
Above, Figs. 16a and 16b: The infant St. John in Leonardo’s The Virgin of the Rocks (left) and (right) the infant in the National Gallery’s later version of the painting.
In the latter we encounter an uncharacteristic indifference to design, sloppiness of treatment and iconographic brutality in the depiction of an infant saint. While the securely autograph Louvre painting has never been in question, considerable argument has arisen over the extent to which Leonardo’s hand is present in the National Gallery version.
In the catalogue to the National Gallery’s 2011-12 exhibition “Leonardo da Vinci ~ Painter to the Court of Milan”, the gallery’s head of restoration, Larry Keith, (who had restored the Virgin of the Rocks prior to the exhibition), was in no doubt that the London version was entirely autograph. He wrote of “discoveries” made in the course of restoration:
“…What we discover is a painter firmly grounded in traditional practice who was able to stretch his methods and materials to express unprecedented intellectual and artistic concerns. However, these painterly interests were only a part of a larger pursuit; he believed that careful observation of all manner of natural phenomena was essential for both new knowledge and a deeper understanding….The National Gallery Virgin of the Rocks is a painting that is at once unique and highly representative of how Leonardo worked. Produced in fits and starts over the last 15 or so years of a commission that took 25 years to complete, it is a composition of the most artful complexity and an image where local colour was sublimated to the newer demands of tonal unity…The National Gallery Virgin of the Rocks…is manifestly uneven in finish and execution but, perhaps, paradoxically, this quality allows us to explore key issues in his painterly practice – methods, materials, collaboration, delegation and finish – and thereby understand better the larger question of the relationship between his painting techniques and his artistic intent…”
Needless to say, this conviction that the picture is an entirely autograph, unique-but-representative Leonardo is not universally accepted. Even at the National Gallery, Leonardo’s authorship has not always been accepted. In 1947 the curator Martin Davies took issue with the picture’s very many doubters (who included the recently former director of the gallery, Kenneth Clark):
“It has to be admitted at the outset that the identification of Leonardo da Vinci’s pictures is by no means the sure and simple thing one might think. It is a fact that there exists no picture of his Milanese period that has not at one time been rejected by famous critics; except for the Cenacolo, which is ruined, and hardly suitable for stylistic criticism at all! The whole subject of Leonardo’s style is therefore somewhat doubtful; but in the particular case of the Virgin of the Rocks in the National Gallery, there has been a good deal of agreement that Leonardo himself painted little or none of it…”
Davies believed the critics to be wrong, but in making his case he conceded many things germane to our concerns here. He acknowledged that this painting was a replica and that it was “quite likely under these circumstance that he [Leonardo] had no great interest in the work”. Although a replica in the sense that Leonardo had been obliged to paint a second version of a commission, Davies draws an ingenious distinction: “the picture is not simply a replica” because so much time had passed that Leonardo had left one artistic era and entered another, making “the picture […] the replica of a work in an older and different style”. Leonardo’s new style “was perhaps expressed rather imperfectly, because the picture is a replica.”
The National Gallery’s suggestion that its “Rubens” Samson and Delilah does not look like any of its twenty-odd secure Rubens’s because he had worked for a brief period in a style like none of his others was a desperate denial of the fact that its “out-of-style” traits stem from its true status as a replica. A more frank acceptance of the Virgin of the Rocks’ acknowledged replica status might might have spared decades of convoluted apologias. Where Larry Keith sees in the Virgin of the Rocks material evidence throughout that “careful observation of all manner of natural phenomena was essential for both new knowledge and a deeper understanding”, another student of Leonardo and Nature, Ann Pizzorusso (who trained as a geologist before becoming an art historian) took an entirely contrary view. For Pizzorusso, the gallery’s claims of some radical shift of style as a means of accounting for the London picture’s problems were entirely and demonstrably without foundation. She was clear on this site that no shift of style could account the picture’s problems because none had occurred:
“Using a date of 1510 for the Virgin and St. Anne and a date of 1483-86 for the Virgin of the Rocks, both in the Louvre, we have proof that Leonardo did not change his style, and that, if anything, he became more fanatical in his quest for geologic accuracy, developing new paints and techniques for natural depiction and driving his students to deliver the most accurate depiction of nature in their own works. So we must ask the question ‘How and why could Leonardo have changed his style to produce a work so lacking in geological and botanical accuracy as the Virgin of the Rocks in the National Gallery in London?’ There is no evidence Leonardo changed his style and now, with the recently cleaned Virgin and St. Anne, we have that proof. We also know that his students were inculcated with his passion for accurate depiction of natural objects so we must also exclude his students as authors of the National Gallery work.”
Writing nearly a decade earlier than Davies, Kenneth Clark, discussed the head of the angel in the London Virgin of the Rocks in his 1938 book of (marvellous black and white comparative photographs) “One Hundred Details from Pictures in the National Gallery”. Of the angel’s head, he wrote “This is the one part of our Virgin of the Rocks where the evidence of Leonardo’s hand seems undeniable…” For Clark, changes in Leonardo’s work over the years were evident, but unlike Davies later and Keith much later, he seems not to have seen evidence of the Later Leonardo equally and everywhere across the painting. For Clark, this curate’s egg of a picture was, in only select parts, very, very good indeed. Of the angel’s head:
“Beautiful as it is, this angel lacks the enchantment of the lighter more Gothic angel in the Paris version. It embodies the result of Leonardo’s later researches in which ideal beauty and and classic regularity of chiarascuro were combined, with a certain loss in freshness, but with an expressive power which almost hypnotized his contemporaries.”
Clark was onto something interesting when speaking of Leonardo’s “hand” – the characteristic touch and surface of his paintwork. It so happens that there was a tool to hand that could have been the greatest boon to those charged with making attributions: high quality micro-photography. Clark, as his own two books of National Gallery details show, was certainly alert to the potency of high quality photographs but he used his comparisons of details to flag up differences between artists in their treatments of similar subjects. That was a perfectly interesting and instructive application. He overlooked, however, the possibility (and the great profitability) of taking, assembling and collating many thousands of details from the most secure, “Gold Standard” paintings, so as to create visual benchmark indicators of artists’ distinctive methods. (Just imagine Morelli and His Ears in an era of digital photography and computers.) If the failure to pursue such programmes in the immediate impoverished years after the Second World War might be excusable, what excuse exists in today’s digital era? The pioneering photographs (shown here at Figs. 18 and 19) by Professor A. P. Laurie in his 1949 book “The Technique of the Great Painters” constituted a perfect template for a means of more accurate visual appraisals – we surely have fewer excuses today than any generation in history for stumbling as if half-blind through the minefield of attribution?
Below, Fig. 17: Martin Davies’ 1947 large format essay on the gallery’s Virgin of the Rocks carried 16 highly informative plates (including this one below of the infant St. John which appears to suggest multiple but vain attempts to keep the toes within the picture?
Above: an unexplained cropped foot
Below: an almost never-used photographic method of comparing brush strokes
Above, Fig. 18: Professor A. P. Laurie explained the significance of this pair of spliced photographs in his 1949 book “The Technique of the Great Painters”:
“This illustration is a photomicrograph of the highlight on the shoulder of [Rembrandt’s] Woman Bathing, National Gallery, No.54. The patch pasted on is from a photomicrograph of a picture whose attribution had to be tested. It will be seen that the brushwork is identical in both cases. It is possible for a skilled forger to imitate a signature, but it is quite impossible to combine the quality of the paint, the nature of the brush, and the handling of the painter, so as to reproduce this complete identity.”
Below, Fig. 19: Prof. Laurie explained the significance of the brushwork below in these terms:
“There is a very interesting portrait of Verdonck [in the National Gallery of Scotland] holding in his hand the jawbone of an ass. It was known from an engraving that such a picture must have existed, but it had apparently disappeared. The Edinburgh gallery possessed a picture by Frans Hals of a man holding a wine glass in his hand. An X-ray revealed that underneath the the wine glass was a painting of the jawbone of an ass which had been painted out by some restorer and replaced by the wine glass. On careful cleaning, the restorer’s work was removed…[this photomicrograph reveals] the rapidity with which Frans Hals laid in stroke after stroke with absolute certainty. In fact the painting seems to be alive, and one can almost see the brush moving over the surface. it would be impossible to mistake this work for the brushwork of Rembrandt…”
Above, Fig. 20:“From Duccio to Raphael ~ Connoisseurship in Crisis”, James H. Beck, Florence, Italy, 2006
In this his last book, the late Professor James Beck of Columbia University, and the founder of ArtWatch International in 1992, wrote:
“Two paintings, a mini aspiring Raphael da Urbino Madonna and an equally tiny aspiring Duccio di Buoninsegna Madonna were sold for record prices in 2004. The first was bought by London’s National Gallery and the second by New York’s Metropolitan Museum of Art. These objects and the mode in which their attributions to their famous presumed authors were achieved document a breakdown in modern connoisseurship. The two objects represent a total expenditure of public money exceeding 100 million dollars for pictures the size of a sheet of paper. These remarkable sales could not have transpired without the participation of art experts whose role was indispensable in offering authentifications of the pictures. This book will seek to define the system of attributing works of art, examine the methodology, treat in depth case studies of recent connoisseurship including the two pictures just mentioned. In addition to what is regarded as a monumental failure on the part of the experts, the use and misuse of public funds is an issue that lies just beneath the surface.”
Above, top, Fig. 21: The version of Caravaggio’s The Taking of Christ in the National Museum of Art, Odessa.
Above, top, Fig. 22: The version of Caravaggio’s The Taking of Christ that was formerly in the Ladis Sannini collection in Florence; was then restored in Rome and authenticated by Sir Denis Mahon; and, is presently being held during legal proceedings.
This small pair of photographs from 1967 is sufficient to show the profound compositional consequences of an extension of one work or a truncation of another. Regardless of the photographs’ poor quality and regardless of the paintings’ relative merits, (both of these, incidentally, have been supported as autograph), the question can be posed in the abstract: Which of the two compositional formats is likelier to be the prime version? Further, if Caravaggio had painted in the truncated format, would he or a copyist then likely have added an extension to the arm of the fleeing disciple in another version? Our feeling is that the Florence format has to be considered to be superior compositionally; more dynamic dramatically; less like a stiff and claustrophobic tableau; and, altogether more expressive of the magnitude of the pandemonium and horror that attended Judas’ fateful act. Whether the Florence picture is the original autograph version has to be established but reports of its pronounced revisions weigh in its favour. Desperately needed is a collation of high quality photographs of all the versions of the paintings, along with detailed photographs of the same, or greater, quality of those published by Prof. Laurie.
Above, Figs. 23 and 24: The Dublin and Rome/Florence versions of Caravaggio’s The Taking of Christ, as reproduced in the Daily Telegraph. Sir Denis Mahon deemed both of these works – at the same time – to be the Caravaggio original.
Above, Figs. 25 and 26: The Prado’s Annibale Carracci’s Venus, Adonis and Cupid, of c. 1588-90, top, as photographed in 1965 (by Hauser y Menet) and before restoration; and, above, as seen after a restoration funded by The Fundación Reale.
Of the two versions (see a detail of the rival Vienna picture below at Fig. 28b) Mahon has supported both as the authentic original work – but this time did so consecutively, not simultaneously, as with the Caravaggio Taking. He championed the Vienna picture until the Prado one emerged. Unabashed, he saw merit in his own mistake, saying (in the 2005 exhibition catalogue) of his critical re-positioning :
“When I first wrote about this composition, some fifty years ago, my observations on style and chronology were based not on the Prado painting, since this was as yet unknown, but on the excellent early copy in the Kunsthistorisches Museum in Vienna and on the preparatory drawings for the figure of Adonis in the Uffizi. When the Prado painting was first published in 1965, by Pérez Sánchez, it was gratifying to realize that, although all those of us who concerned ourselves with Emilian painting had mistakenly assumed that the Vienna picture was Annibale’s original, one’s intutitions about the importance of the work and where it fitted in the artist’s evolution were confirmed.”
This was dissimulation: had Mahon been alert to what might be called The Problem of Arbitrary and (otherwise) Bizarre or Inexplicable Croppings, he would have spotted the tell-tale warning in the cropped nose of the hound on the right of the Vienna version. This would have been the more likely had he consulted, as well as figure studies in the Uffizi, the etched copy of the original made in of 1655 by Luigi Scaramuccia (see Fig. 27, below). This delightful record shows not only that the hound’s head (like Samson’s toes elsewhere) had been set comfortably inside the picture, but, also, that the landscape at the top right was more extensive and contained an architectural feature (doubtless of some iconographic significance). Curators and restorers too often disregard the testimony of graphic artists, when, within their limits and styles, they are essentially respectful of the works they were paid to copy. (A copyist inclined to go his own way would likely get less not more employment.)
Below, Fig. 27: Luigi Scaramuccia, Venus, Adonis and Cupid, 1655, second state, The British Museum (here mirrored).
Above, Figs. 28a and 28b: Details of the Prado’s Carracci Venus, Adonis and Cupid (left), and the Vienna Kunsthistorisches Museum version (right).
If Mahon corrected one error with this painting, he perpetuated others. The catalogue to the exhibition that celebrated the Prado’s restoration, produced the customary self-congratulatory sponsor’s waffle (here The Fundación Reale). Less forgivable was Mahon’s claim that the restoration helped establish the date of the original work. Mahon had been a belligerent champion of National Gallery restorations when at their worst in the post-war years, mocking, in tandem with the gallery’s head of science, the objections of scholars like Sir Ernst Gombrich (who had to wait a third of a century for a full technical vindication of his objections – see How the National Gallery belatedly vindicated the restoration criticisms of Sir Ernst Gombrich and 24 November 2011)
What is unsaid in the hype of big business-sponsored restorations, is that a restorer can never recover what has been lost and that by cosmetically dressing up degraded works, imparts a spurious simulation of health and historical veracity. No restoration exhibition should ever take place without the inclusion of all extant visual records of the work(s) in question. If we disregard the testimony that exists in this area, we enter a world of “art conservation” make-believe. In doing so, we leave ourselves ill-quipped to address the most urgent questions of attribution and condition. Sadly, with this Carracci painting, the two versions have experienced what restorers euphemistically call “different conservation histories”. Which means is that they have suffered to varying and unequal degees, physical assaults on their fabrics and their pictorial skins. We are all obliged to acknowledge and address these terrible truths. Not least because all the inherent difficulties of making attributions are exacerbated by these various histories of “treatments”. On the testimony of the etching, it would seem that the Vienna hound lost considerable shading to the side of his head, while his elaborately jewelled collar survived much better than that seen in the Prado version. This tells us that neither work remains a true witness to its own original self and that, therefore, theories and judgements made on the basis of the pictures’ present selves should come with careful qualifications and health warnings, and not with some facile celebration of glorious recoveries.
The differences that restorations make to individual pictures can be as great or greater than the differences that might originally have existed between an authentic original work and an extremely high quality copy of it. It should be accepted that one of the consequences of past restorations is that making sound appraisals of the merits of once closely related versions of paintings is made the more difficult. Some indication of how dramatically transforming restoration treatments can be can be might be gauged by the pair of details below (Figs. 29 and 30) from the Prado’s records of the same painting. Properly read, their inclusion, and that of the two states of the Scaramuccia etching in the Prado exhibition catalogue might constitute a most useful contribution to knowledge and understanding in this arena.
Click on the images above for larger versions. NOTE: zooming requires the Adobe Flash Plug-in.

The Perpetual Restoration of Leonardo’s ‘Last Supper’ – Part 1: The Law of Diminishing Returns

8th February 2012

The restoration of Leonardo’s “Last Supper” (1977 -1999) was defended last month at the National Gallery symposium on the artist’s methods and influence, by Pietro Marani, the Leonardo scholar who co-directed the restoration from 1993 onwards. When we pointed out that while the previous restoration had been praised for recovering all of Leonardo’s authentic paintwork, the last restoration had in fact introduced vast areas of new painting, Marani held that restorers and curators may impose their own generation’s values and interpretations. Appeals to the authority of one generation cuts little ice given how happy each proves to be to undo and redo its predecessor’s work – and, besides, given that on past form we are already half way along to the next generation’s “Last Supper” restoration, the one presently being defended is already approaching its own “sell-by date”.

In its early days, the last restoration was presented as a near miraculous liberation of the original and entirely unadulterated handiwork of Leonardo da Vinci, when that claim, far from being a novel one of its day, was, as mentioned, a re-run. In 1958, John Canaday, the art editor and art critic of the New York Times, reported how, following the war-time bombing of the refectory housing the mural in Milan:

When the sandbags were removed…the ‘Last Supper’ was in such pitiful condition that historians were ready to list it as a war casualty. But since then the picture has been gone over as definitively as the combination of exhaustive scholarship and high technical skill make it possible to preserve once and for all whatever is discoverable of the original work. The results are better than anyone expected, although the painting still suggests a ghost at best or, at worst, an embalmed relic…”

Taking their cues from restorers, commentators often dismiss past treatments and celebrate more recent scientifically under-pinned “definitive”, “once and for all” and “miraculous” ones. (Just this week, a Rubens – his “Cain Slaying Abel” – that has been restored at the Courtauld Gallery with funding by the Bank of America, is said in a press release to have stabilized the painting for nothing less than “the next one hundred years”.) The “Last Supper” rescue operation took place in two stages in 1947-49 and 1952-54. In the first, the restorer Mauro Pelliccioli (see Fig. 1) supervised by a former Superintendent of Fine Arts in Milan, Fernanda Wittgens, anchored the disintegrating paint with shellac. As Wittgens’ wrote in the Christmas 1954 issue of Art News, “This was not ordinary shellac, of course, but an absolutely colourless one recently produced in England by a chemical process that removes all wax.” With this shellac, Wittgens went on, Pelliccioli performed his wonders:

…it produced the greatest miracle of the entire restoration. The reattachment of the painted surface was achieved to perfection, and the colours acquired a new consistency as though they had been actually rejuvenated by the shellac that made them adhere to the wall…Mauro Pelliccioli, by doing away with all glues soluble in water, had permanently eliminated the danger of mould: he had at last hit upon a tremendously strong, transparent adhesive impervious to humidity.”

When the shellac was settled, Pelliccioli began scraping off the repaints of earlier restorers. Bernard Berenson visited the scaffold and later reported how, with no more than a penknife, a razor and a drop of turpentine, Pelliccioli had “touched bottom” by removing the “multiple restorations of centuries” and allowing the paint of Leonardo, “deteriorated by the centuries but no longer deturpated by incompetent hands” to be seen once more. In his diary note on October 21st 1953 Berenson said that Pellicciolli had known precisely “where to scrape” – a point echoed and amplified in H. H. Pars’ 1957 “Pictures in Peril”:

Pelliccioli was able to distinguish those parts of the painting where nothing of the original painting was left, and those parts where overpainting and restoration concealed Leonardo’s own brilliant colours. Step by step these were revealed in the now firmly-fixed painting until we are now able to see Leonardo’s ‘Last Supper’ in better preservation than for many generations, deteriorated through the centuries it is true, but no longer marked and deformed by incompetent hands.

Thus, to everyone’s satisfaction, the work had been physically rescued and what authentic Leonardo paintwork could be exposed to view had been liberated. Earlier restorers’ repainting was left in place only where it covered bare wall. Just twenty-one years later in 1975, Pinin Brambilla Barcilon, a former student of Pelliccioli’s, reported that fragments of paint were falling off the mural. Two years later, following tests, she began re-securing those parts of the paintwork that were becoming detached. Her minimal and straightforwardly necessary conservation measure was to mushroom, in parallel with the restorations of Michelangelo’s Sistine Chapel ceiling frescoes, which began in 1980, and Masaccio’s Brancacci Chapel frescoes, which began in 1981, into one the most protracted radically ambitious, corporately-funded and controversial restorations of modern times.

It can now be seen that 1977 marked the high water mark of confidence in the recuperative powers of restorers and in the legitimacy of their increasingly bold and experimental methods. In the previous decade, heroic actions in wake of the floods in Florence had carried restorers to unprecedentedly high levels of public respect (even when they opted to strip frescoes from walls). In Britain, following a spuriously engineered triumph with the restoration of Titian’s “Bacchus and Ariadne” at the National Gallery, criticism had been declared dead by the victors. The National Gallery launched its Technical Bulletin in 1977 and in it, the Gallery’s head of conservation science, Joyce Plesters, mused complacently: “one or two readers may recall the furore when the cleaning of discoloured varnishes from paintings…began to find critics”. In the same year, Kenneth (Lord) Clark published his two volumes memoir “Another Part of the Wood” and “The Other Half” in which he pronounced picture cleaning “a battle won” and claimed credit for having taken the first steps, as director of the National Gallery (1934-1945), by installing a “scientific department with all the latest apparatus”. He had done so not because he believed in the “application of science to the problems of cleaning”, but because “until quite recently the cleaning of pictures used to arouse extraordinary public indignation and it was therefore advisable to have in the background what purported to be scientific evidence to ‘prove’ that every precaution had been taken.”

Duping the public in such manner occurred in Italy. In 1981, a year after the start of the Sistine Chapel ceiling restoration in Rome, the restoration of the Brancacci Chapel commenced. Speaking of the prior tests made in connection with that restoration, the author Ken Shulman cites an Italian art historian (“Anatomy of a Restoration”, p. 156), who had said:

Let’s be honest and admit what all restoration directors will say in private. At the beginning of any restoration, you order as many tests as you can imagine, fully aware that only about five per cent of them will be of any use during the restoration. The rest of the analyses are merely window dressing.”

With Leonardo’s “Last Supper”, amidst all the preliminary testing, no-one seemed concerned by the fly that was present in the ointment of Pelliccioli’s celebrated reductive, purist restoration: his liberation of Leonardo’s paintwork had come at a cost in terms of artistic legibility. A law of diminishing returns had been set in motion that would (as we will see in Part 2) produce panic and confusion among the restorers and their supervisors.

Michael Daley

Comments may be left at:

Printer-friendly PDF version of this article

Above, Fig. 1: Pinin Brambilla Barcilon (left) and Mauro Pelliccioli (right) at work on Leonardo’s “Last Supper”.
Above, Fig. 2: The “Last Supper” in the refectory of S. Maria delle Grazie, Milan, in 1900.
Above, Fig. 3: The “Last Supper” before 1943.
Above, Fig. 4 and below, Fig. 5: The refectory of S. Maria delle Grazie after the bombing of 1943. The wall bearing the “Last Supper” is present behind the tarpaulin seen below.
Above, Fig. 6: The rebuilt refectory with the “last Supper” after Mauro Pelliccioli’s interventions of 1947-49 and 1952-54.
Above, Fig. 7: The figure of Christ after Mauro Pelliccioli’s restoration.
Above, Fig. 6: The cover of the third part of the catalogue for the 1983 exhibition “Studies for the Last Supper” at the National Gallery of Art, Washington. Below, Fig. 7, a de Stijl-esque work by Cecil Touchon which uncannily echoes the seeming “aesthetic” of the restorer’s bewildering method of cutting multiple ‘windows’ through the skin of the mural during the cleaning.
Click on the images above for larger versions. NOTE: zooming requires the Adobe Flash Plug-in.