Did Rembrandt paint the Mona Lisa?

With a sinking heart I read in The Independent of 19 February a story about a new research project on Leonardo da Vinci. “Some of the world’s most eminent art historians, led by the Oxford academic Martin Kemp, are about to conduct the first comprehensive scientific study of the great man’s [painted] oeuvre, putting such iconic images as the Mona Lisa under the cold, impartial gaze of science…”

That much was fine. I am all in favor of the cold, impartial gaze of science. I am even more in favor of the study of groups of works of art rather than the incidental examination of individual works. If the Universal Leonardo Project is going to do this, and do it systematically, applying the same tests to all the objects concerned, it will be a great boon. At a reported £1.5 million, part of it supplied by Bill Gates, the project would be a bargain.

My dismay began with the choice of objects to be examined. To set out to study “each of Leonardo’s works,” as Kemp is quoted as saying, is to rob the project at the outset of much of its scientific value. Of the paintings presently attributed to Leonardo, only one or two is documented solidly from his lifetime (1452-1519); conversely, a dozen paintings that are documented as Leonardos do not correspond to an existing work. Present-day expert opinion puts the number of extant Leonardo paintings at a range between 15 and 25 – a margin that reveals that “the great man’s oeuvre” is a matter not of fact but of opinion.

To be really useful, a project of this kind would begin with the systematic examination of a large random sample of Italian paintings from a long period of time, say from 1450 to 1550. A database of the physical characteristics of such a sample would enable one to say what is typical and what is different about those 15-25 paintings. If one isolates them from the start, nothing one learns can possibly distinguish them from other work of the period.

My unhappiness deepened when I read on: “The project has an objective that will have every museum with a Leonardo getting nervous – the question of attribution.” This was truly terrible. Scientific measurement is sometimes able to prove that a work is not what it appears to be. Forgers who are careless enough to use synthetic pigments that were not invented until after the period of the originals they are faking can be expected to be caught out in the laboratory. However, laboratory science is fairly useless when it comes to proving positive attributions – that a certain work of art was made by a certain individual. Artistic techniques and materials were too generally used for that.

True, the Universal Leonardo Project adds one interesting new tool to the kitbox of the paintings scientist: fingerprinting. (In this it was preceded by the excellent work of Nancy Lloyd at the Fogg Art Museum, who found fingerprints in clay models from the Bernini studio.) “Although without a guaranteed Leonardo fingerprint to authenticate the works, a common print found on all the paintings must surely be that of Leonardo, the experts believe.” Do these experts really think that they will find identifiable prints in all the paintings under study? And if the prints do not match, what then? Will they disattribute a painting if they find a divergent fingerprint? If not, then they have no basis on which to advance a matching fingerprint as positive evidence.

Then came the part that got me groaning. “The scenario” of the Leonardo project, according to The Independent, “is similar to another recent attempt to definitively analyse an Old Master’s corpus, the Rembrandt Research Project. This group of experts is credited with in effect rewriting the Dutch artist’s oeuvre after it found many of his paintings misattributed. The experience hurt at least one eminent museum. The Mauritshuis in The Hague, which had thought of itself as the proud possessor of three Rembrandt self-portraits, suddenly found itself with just one. Its star work, Self-portrait with a gorget, was ruled a 17th-century copy and the real thing was found to be a battered work, unrecognized and gathering dust in a Nuremberg museum.”

If I were Bill Gates, after reading that I would take my money back. In 1983, in vol. 1 of its Corpus of Rembrandt paintings, the RRP published the Hague painting as an original and that in Nürnberg as a copy; in 1991 the Mauritshuis version was included in the RRP-controlled exhibition Rembrandt: the master and his workshop, with an entry in the catalogue by an RRP member. It was in reaction to that exhibition that the German art historian Claus Grimm argued for the reversal that was accepted by the Mauritshuis before the RRP came around. (In November 2002 I tested the case in a lecture to an international congress of 1000 medical pathologists. I put the two versions on the screen and asked these non-art historians cold which one they preferred. They voted in overwhelming majority for the Nürnberg version, which the RRP called an inferior copy.)

If anything, the Rembrandt Research Project is a convincing demonstration that scientific examination is useless for establishing the individual authorship of Old Master paintings. That may not have been the intention, but to date it is the most abiding contribution of the RRP to the use of science in the service of connoisseurship. Leonardists who ignore that lesson do so at their peril.

© Gary Schwartz 2003. To be published in Loekie Schwartz’s Dutch translation in Het Financieele Dagblad, Amsterdam, 15 March 2003.

This item is installment 179 in a bi-weekly column on art-historical matters (mainly Dutch) published by Gary Schwartz in Dutch in the Saturday supplement of an Amsterdam daily, and in English on the Internet discussion list ffdys.

To subscribe to the list, mail Gary.D.Schwartz@let.uu.nl