Visual Arts: Rembrandt’s Imagination

I envision Rembrandt with chalk or pen always at hand, sketching from life and imagination constantly. This is also how he taught his pupils, who like him also produced numerous drawings related and unrelated to paintings or prints. Why do so many experts disagree?

By Gary Schwartz


Rembrandt, Three syndics of the clothmakers' guild, ca. 1661-62 Berlin, Kupferstichkabinett References: Benesch 1178. Bevers 2006, nr. 55

Rembrandt, Three syndics of the clothmakers' guild, ca. 1661-62 Berlin, Kupferstichkabinett References: Benesch 1178. Bevers 2006, nr. 55

In an earlier column I illustrated a large number of Rembrandt drawings that are currently accepted by the leading experts—or at least never doubted by them in print—that are closely related to etchings and paintings by the master. They cover every category in which he worked—histories, portraits, genre scenes, landscapes—and they relate to full compositions as well as details.

Very few are the exact equivalent of the etching or painting concerned. Rather, they show Rembrandt at work the way he was described by his pupil’s pupil Arnold Houbraken: trying out in successive drawings variation after variation on a projected theme.

When he finally decided what he wanted, the finished painting or etching seldom resembled the preliminary drawings “like twins,” a phrase that Houbraken uses in deprecation of artists other than Rembrandt who did produce mechanical preparatory drawings. To Houbraken, Rembrandt’s preparatory drawings were part of a dynamic process.

The passage I quoted there goes on: “He was better at this than anyone. No one else I know made so many changes in sketches of one and the same object“ (vol. 1, 1718, pp. 257-58). A few pages on (p. 261) we read: “Various pupils of his have told me that he sometimes drew a figure in 10 different ways before putting it on the panel.”

This I find a very satisfying and convincing image of Rembrandt at work. Looking at the drawings related to etchings and paintings, this is exactly what we see. I envision him with chalk or pen always at hand, sketching from life and from imagination constantly. This is also how he taught his pupils, who, like him, also produced numerous drawings related and unrelated to paintings or prints.

Why then has a consensus taken hold among the leading experts that the opposite is the case? Why do they think that Rembrandt did not work from drawings on his compositions and portraits? A modest statement of that case was made by Otto Benesch, in his essay, published in book form, Rembrandt as a draughtsman (1960; p. 6): “Drawings by Rembrandt which were projects for compositions, single studies for particular details of compositions painted or etched afterwards, or gifts for particular occasions, are comparatively few in number.” This is a defensible statement insofar as it concerns the number of existing drawings. There are more drawings by Rembrandt that are not noticeably related to panels, canvases, and plates than drawings that are.

Later specialists pushed the issue much further. In volume one of A Corpus of Rembrandt Paintings (1982), Ernst van de Wetering states a proposition that has since guided his philosophy of Rembrandt’s creative process. In a chapter on “Painting materials and working methods” (p. 22), van de Wetering writes:

The number of drawings by Rembrandt showing the whole of a composition is . . . remarkably small . . . . There is every reason to assume that Rembrandt did not work out his compositions on paper first, but sketched them direct on the actual support. He must thus have been in the category of painters of whom Karel van Mander wrote: “that some, well-practised, experienced and working with a firm hand . . . are used to drawing fluently by hand on their panels what they have seen already painted in their mind’s eye.’

This opinion (repeated verbatim in volume four of the Corpus, 2005, p. 145) apparently satisfied the Rembrandt Research Project (RRP) so completely that it makes no further point of the issue. Nowhere in the Corpus is it noted that there is a major discrepancy between Houbraken’s picture of Rembrandt as the most obsessive draftsman he knew, specifically with regard to preparatory drawings, and that of van de Wetering, who thinks that in principle Rembrandt did without preparatory drawings for compositions. In its extensive entry on “The Night Watch” in volume three of the Corpus (1989), the RRP publishes the single drawing by Rembrandt that has been brought into connection with that monumental painting, but it does not remark on the fact that only one such drawing has survived.

The idea that Rembrandt could have created that painting, with its 16 portraits, 16 auxiliary figures, countless appurtenances, elaborate architecture, and complex composition, on which he worked for some three years, without having made numerous preparatory drawings seems to me not only unlikely but, no matter how vivid the images in his mind’s eye, inconceivable.

Young Girl Leaning Out of Window Young girl looking out of window

Rembrandt, Young woman looking out of a window, ca. 1645
Related to painting in Dulwich Picture Gallery, London (Bredius 368)
London, Courtauld Gallery
References: Benesch 700. Royalton-Kisch 1992, sub nr. 50

That question alone—“did Rembrandt really paint The night watch without recourse to drawings?”—should have warned his fellow Rembrandt specialists that something was awry. However, far from challenging van de Wetering’s proposition, they embraced it willingly. In 1985 in an entry to his indispensable catalogue of the Rembrandt drawings in the Rijksmuseum, Peter Schatborn says flatly, “Rembrandt seldom drew studies for paintings . . .” (p. 13). Schatborn does not limit this reductive attitude to preparatory drawings alone. In 1991 in the exhibition catalogue Rembrandt: the Master and His Workshop: Drawings and Etchings (p. 13), he wrote, “When research into the drawn oeuvre has advanced further, it will probably emerge that Rembrandt drew much less than has always been assumed.” Concerning the etchings, Martin Royalton-Kisch wrote in 2000 in Rembrandt the Printmaker (p. 64), elaborating on a paper that was published in 1993: “Rembrandt made no less than 300 etchings and most were created directly on the copper plates without recourse to preliminary drawings . . .”

The present consensus concerning Rembrandt’s drawings was put into words compactly by Bob van den Boogert of the Rembrandt House Museum in an entry in the catalogue The Mystery of the Young Rembrandt (2001), nr. 34: “There are very few surviving drawings by Rembrandt that can be identified as direct preliminary studies or composition sketches for paintings. This scarcity is explained by Rembrandt’s working method: he usually sketched out the initial layout for a composition on the painting itself, in the first phase of the painting process.” (Van den Boogert then goes on to disqualify entire categories of drawings as “authentic preliminary studies” for paintings or etchings, assigning various other functions to drawings that are related to Rembrandt compositions.)

Over and against this view, I propose that the scarcity of direct preliminary studies is explained not by Rembrandt’s working method but by the loss over time of the overwhelming majority of all the drawings he ever made. In 2006 (The Rembrandt Book, pp. 105-06), I wrote:

Rembrandt’s inventory included twenty-three albums, one basket and one package of drawings by the master, plus “various packets of sketches, by Rembrandt as well as by others. Most of the albums are said to be “full of” drawings or sketches. How full was full? [Evidence is presented to back up an estimate of 375 sheets per lot, which] would yield more than nine thousand drawings in the sale.

Numbers of this kind look absurd at first sight, but in fact 9,000 drawings, divided over the years 1625 to 1656, amount to fewer than one drawing a day . . . . Consider too that those were the drawings Rembrandt owned in 1656. They do not include those that he sold or gave away in the preceding 30 years, or those he threw out. (They also do not include those he made in the remaining 13 years of his life.) Since no more than about 600 or 700 drawings are currently accepted as original drawings by Rembrandt, we must face up to the likelihood that more than 90 percent of his drawings have been lost. At that rate, all suppositions concerning what Rembrandt did not draw are off . . . Rembrandt was not a less but a far more prolific draftsman than we now know.

To repeat my question: Why has a consensus taken hold among the leading experts that the opposite is the case? For this I can offer two explanations. One is that connoisseurs of Rembrandt drawings have been devoting most of their energies to eliminating from his oeuvre sheets that were assigned to him too optimistically by their predecessors. Their minds have been set on reducing the number of authentic Rembrandt drawings.

A more important factor is that they have failed to acknowledge that the great extent of loss precludes the kind of conclusions that they draw from the existing drawings. They have fallen prey to what in logic is called the fallacy of ignorance or fallacy of lack of imagination: “Because there appears to be a lack of evidence for one hypothesis, another chosen hypothesis is therefore considered proven.” Because there appears to be a lack of evidence that Rembrandt employed drawings for his paintings and etchings, the hypothesis that he did not do so is considered proven.

In fact, there is evidence enough that Rembrandt did use drawings in this way. The lists I published in my earlier column contain numerous examples of this kind, touching on 47 etchings (about 15 percent of the number accepted) and 37 paintings (10-15 percent of surviving paintings—there too losses have been suffered). One can quibble about this or that attribution or connection or dating or function, but in most of these cases we see Rembrandt behaving exactly the way that is described by Houbraken and denied by van de Wetering, Schatborn, Royalton Kisch, van den Boogert and others. The lack of such evidence concerning the remaining 85 percent of the etchings and 85-90 percent of the drawings fits in better with my hypothesis that the drawings were lost in those numbers, than with their assumption that absence of proof is proof of absence. I propose that we think again about Rembrandt’s creative process and look again at the drawings.

© Gary Schwartz 2010. First published on 22 January 2010 on the Schwartzlist and TheArtsFuse.


Gary Schwartz was born in Brooklyn, New York in 1940. In 1965 he came to the Netherlands with a graduate fellowship in art history and stayed. He has been active as a translator, editor and publisher; teacher, lecturer and writer; and as the founder of CODART, an international network organization for curators of Dutch and Flemish art.

As an art historian, he is best known for his books on Rembrandt: Rembrandt: all the etchings in true size (1977), Rembrandt, his life, his paintings: a new biography (1984) and The Rembrandt Book (2006). His Internet column, now called the Schwartzlist, appeared every other week from September 1996 to April 2007 and has been appearing since then irregularly. His most recent book on Rembrandt is one of the six titles nominated for the Banister Fletcher Award for the most deserving book on art or architecture of that year.

In November 2009 Schwartz was awarded the coveted tri-annual Prize for the Humanities by the Prince Bernhard Cultural Foundation of Amsterdam.

Please address reactions to

Leave a Comment

Recent Posts