Amsterdam: Defining Rembrandt

Art historians are still trying to establish which paintings are by the master and which ones are not

LAST NOVEMBER West Berlin’s Staatliche Museen let it be known that The Man With the Golden Helmet was not, as long supposed, a work by Rembrandt Hermanszoon van Rijn. The painting had been one of the most popular in the oeuvre attributed to the artist. It appears on coasters, posters, and playing cards. In phosphorescent hues it has been replicated on velvet. The revelation took the public by surprise and received considerable attention in the press. As it happens, however, connoisseurs of Dutch art had suspected the painting’s misattribution for quite some time. TheMan With the Golden Helmet was but the latest casualty in a century-old battle to “purify" the corpus of Rembrandt’s w ork.

Far more Rembrandt paintings seem to have been bought and hung than the artist ever produced. No one knows howmany portraits, landscapes, history paintings, and genre scenes Rembrandt actually created (his output was prodigious), but more than a few museums still display works over his name which curators are fairly sure did not come from his hand. During the 1930s the art historian Abraham Bredius catalogued 630 paintings by Rembrandt, in public and private collections, “whose authenticity,”he wrote, “seems to me beyond all doubt.” Art historians today believe that of those 630 paintings 250 or more are very likely not by Rembrandt.

In New York, at The Frick Collection, the renowned Polish Rider, one of the most familiar paintings in the museum’s holdings, trots into an uncertain future. A few blocks away, at the Metropolitan Museum of Art, at least two works displayed as Rembrandts—The Auctioneer and Christ With a Pilgrim’s Staff—are regarded as suspect by some scholars, including one member of the museum’s staff. In the main storeroom for European paintings, behind the gallery walls at the Met, are eighteen more paintings that entered the museum as Rembrandts but were subsequently presumed to be or established as the work of others—generally of painters who have not yet been identified. Several of these are paintings that I grew to love on childhood trips into the city. A Head of Christ that once helped to clarify my image both of Jesus and of Rembrandt is now in storage.

Rembrandt has been a peculiarly difficult case, and he himself is partly to blame. In Leiden (where he was born) and Amsterdam (where he lived from about 1632 until his death, in 1669) he was a teacher of considerable repute and unusual power; everyone who came into his orbit began to paint like him. Govert Flinck, Ferdinand Bol, Carel Fabritius, Willem Drost, Gerard Dou—these and a hundred other artists learned most of what they would know about painting in the shadow of Rembrandt’s easel. In their later careers some of them continued to follow faithfully every swerve and advance in the master’s method. Although Rembrandt, unlike Rubens, never employed pupils to work on his own paintings, he sometimes worked on theirs. Hanging at the Rembrandt house, on the Jodenbreestraat, in Amsterdam, is a pen-and-wash drawing by Constantijn van Renesse that shows obvious signs of intervention by Rembrandt to strengthen the composition. (The sketch hangs expressly to illustrate Rembrandt’s role as a teacher.) In rare cases Rembrandt’s contribution to a student’s work included a signature. “Changed and overpainted by Rembrandt 1636”: these words, in Rembrandt’s hand, appear on Flinck’s Abraham Sacrificing Isaac, in Munich. On other paintings an authentic signature may appear without comment. Guild regulations in seventeenth-century Amsterdam permitted the sale of a student’s works under his master’s name.

Compounding the problem of misattribution is the existence of numerous imitations and forgeries. Though he died bankrupt, Rembrandt was a successful painter who commanded high fees. Then as now, owning a Rembrandt was a mark of distinction. Imitations of Rembrandt’s work appeared during his lifetime; their number grew as the years passed, in response to mounting demand from dealers, speculators, and collectors. The trade in forgeries waxed during the nineteenth century especially, because the hallmark qualities of Rembrandt’s paintings—the contrast, often surreal, of shadow and light, the dramatic composition, the restrained application of color—appealed powerfully to the new Romantic sensibility in European taste. If the imitations lacked a pedigree, so did all but a handful of authentic Rembrandts. Believing that only Rembrandt could be Rembrandtish, nineteenth-century connoisseurs—“expansionist” rather than “reductionist" in outlook—welcomed hundreds of dubious paintings into the fold.

Imposing some sort of critical order on the Rembrandt oeuvre has taken time, and the tension between expansionist and reductionist tendencies is felt to this day. At the turn of the century Wilhelm von Bode (who acquired The Man With the Golden Helmet for the Berlin museum) collaborated with Cornelis Hofstede de Groot to produce a lavish, eight-volume catalogue of Rembrandt paintings. Von Bode and Hofstede de Groot ruled out a number of reputed Rembrandts and ruled in several new discoveries, but on the whole they remained “somewhat large hearted” (as Hofstede de Groot himself complained of an earlier generation of critics) about how much to accept as genuine. They validated 595 paintings. More generous still was Wilhelm Valentiner, whose various catalogues designated well over 700 paintings as authentic Rembrandts. An American, John C. van Dyke, halved the total in 1923. Van Dyke’s instincts were correct, but many of the paintings he excluded (or included) were not wisely chosen. Abraham Bredius subsequently raised van Dyke by more than two hundred while reducing Valentiner by about half that number. In the early 1960s a catalogue by Kurt Bauch admitted 562 paintings into the Rembrandt corpus. A contemporaneous list by Jakob Rosenberg accepted thirty-three of the paintings rejected by Bredius and rejected some forty of the paintings Bauch catalogued as authentic. In 1968 Horst Gerson, who as a young man had assisted Abraham Bredius, produced a revised edition of the Bredius catalogue in which the number of accepted Rembrandt paintings was reduced to about 420. Although each of the successive catalogues made use of, and benefited from, the alluvial deposits of artistic scholarship, none of them provided a critical apparatus of any great sophistication. In 1969, when the tricentennial of Rembrandt’s death was observed, there still existed no true catalogue raisonnè of the master’s paintings—no catalogue in which all of the physical, stylistic, and documentary evidence pertinent to every painting ascribed to Rembrandt was examined.

THE YEARS LEADING up to the tricentennial were ones of considerable stimulation in the small world of Rembrandt scholarship. Conferences were held, exhibits opened, books and monographs published. In 1968 a group of seven Amsterdam-based art historians and conservators convened to launch the Rembrandt Research Project and assume the responsibility for producing a catalogue raisonne of Rembrandt’s paintings. In so doing they assumed the obligation not only of deciding which of the pictures attributed to Rembrandt were actually his and which ones were not but also of explaining the difference. This is not to say that museums have been regarding their own collections uncritically; the National Gallery in London (a pioneer in the field of Dutch paintings) and the Metropolitan in New York, along with many other museums, continually scrutinize the paintings in their possession. From time to time, when the evidence warrants, they publish updated assessments of their holdings. The research project, however, was not to be an intramural affair, and its focus was not to be limited to the works in any one institution.

With funding from the Dutch government the members of the project set out to examine physically and analyze scientifically every painting that Abraham Bredius had attributed to Rembrandt (along with a handful of others that have come to light in the postwar years). That task has more or less been completed—though the whereabouts of perhaps a dozen pictures, presumed to be in private collections, are still unknown—and publication of the findings is now under way. The first volume of the research project’s Corpus of Rembrandt Paintings, covering the painter’s Leiden years (1625-1631), was published in 1982. The second volume, dealing with Rembrandt’s first few years in Amsterdam, will appear in several months. Three more volumes are planned for publication within the next five or six years. Needless to say, given the scholarly and commercial implications, the publication of Volume II is awaited with interest by collectors and curators around the world. The Rembrandt Research Project speaks with considerable—though not unquestioned—authority. Its members include Simon H. Levie, the director of Amsterdam’s Rijksmuseum; Pieter J. J. van Thiel, the Rijksmuseum’s curator of paintings; Ernst van de Wetering, an art historian on the staff of the Central Research Laboratory for Objects of Art and Science, in Amsterdam; Bob Haak, the director of the Amsterdam Historical Museum and the author of an important Rembrandt biography; and Josua Bruyn, a professor emeritus of art history at the University of Amsterdam, who is the spokesman for the project. J. G. van Gelder, the eminent Rembrandt connoisseur, and J. A. Emmens, the iconographer, were members of the group until their deaths.

In Volume I of its catalogue the research project considered ninety-three of what Bredius had accepted as early paintings by Rembrandt. It authenticated forty-two of them, rejected fortyfour, and reserved judgment on seven. Another hundred or so paintings will be considered in Volume II. Although project members are reluctant to discuss the contents of the catalogue’s next installment, it is probably safe to assume that what has always been viewed as a particularly fruitful period in Rembrandt’s career—the years 1631 to 1634, when clients had to beg him as well as pay him, according to one contemporary source—will be shown to have been somewhat less fruitful than has been supposed. It is also likely that the number of paintings assigned to one or another of Rembrandt’s students will be modestly augmented. And several of the authentications put forward in Volume I are almost certain to be withdrawn in Volume II.

It is impossible to consider the work of the Rembrandt Research Project without thinking, at least at first, in terms of winners and losers. One is dealing with an artist whose paintings, were they all to be sold individually at auction, would probably fetch an average of $5 million apiece. (A painting by one of his better students would be lucky to go for a tenth of that sum.) But mercantile concerns are not uppermost in the minds of the members of the research project. Their interest in delimiting the Rembrandt oeuvre derives chiefly from an academic interest in the painter himself: only when the apocryphal is excised from the acknowledged body of his work can Rembrandt’s special qualities be identified and assessed with any confidence. The preface to Volume I states: “Our attempt to define and purify Rembrandt’s oeuvre amounts to an effort to find rational, communicable arguments to support our opinions.” In large measure, then, the task of the research project has been to develop criteria for recognizing what is Rembrandt-like— criteria that involve issues not only of quality and style but also of method and technique. Josua Bruyn says, “These are matters in which not everybody is much interested—an unfortunate result, I think, of people being concerned with authenticity and value. But they are a large part of what is new and interesting in what we have to offer.”

I MET WITH PROFESSOR Bruyn one day last winter at his office at the Central Research Laboratory for Objects of Art and Science. It was bitterly cold in Amsterdam, and the city was covered by a mottled glaze of ice. The Rijksmuseum’s Victorian bulk, normally framed by Bruyn’s windows, was barely visible through the morning mist. In one corner of the spacious office, on an easel, sat a black-and-white photograph of a painted eye, many times enlarged — the right eye of the ringleader in Rembrandt’s The Conspiracy of Julius Civilis, now hanging in Stockholm. (Julius Civilis, who led the Batavians in a revolt against Rome, lost his left eye.) Josua Bruyn himself seemed to epitomize one image of the Old World art historian. He is tall and lean, with silver hair brushed neatly back from his brow. His manner is at once genial, formal, and polite. Recently retired from university life, Bruyn now devotes most of his time to the Corpus of Rembrandt Paintings.

“The uniqueness of the project lies in the intensive and extensive scope of the research,” he said. “No scholar—and there had never before been a team— had been given the opportunity to examine all of the paintings attributed to Rembrandt, much less to do so over a limited period of time. We were able to do this, so that a much better comparison became possible. I am not aware of any similar project involving the work of a single artist. To some degree—we had our own variations, of course, our own requirements—we took as a model the methodology of the Corpus of FlemishPrimitives, which was an effort begun in the 1950s to catalogue Netherlandish paintings of the fifteenth and sixteenth centuries. These publications discussed the physical aspect of paintings—for the first time describing the pictures not simply as works of art but as objects. ‘They included information on the materials used and whenever possible information discovered through x-rays, infrared examination, and so on.”

Beginning in the late 1960s, Bruyn said, and continuing for half a decade, the members of the research project traveled in pairs (and in different combinations) to Melbourne, Tokyo, London, Newr York, Berlin, Los Angeles, Leningrad—anywhere a Rembrandt was reputed to exist. They went to Kenosha, Wisconsin, to Burghley House in Northamptonshire, to the parish church at Le Mas d’Agenais, in the Garonne. They visited several paintings more than once. Expenses were met through a generous subvention from the Netherlands Organization for the Advancement of Pure Research. To aid in the description of each painting—and to make sure that the same information was collected for every work—the team developed a questionnaire.

Some of the questions were the obvious ones. What was the physical state of the picture? Was it painted on a canvas or a wooden support? Did it have a signature, and was the signature authentic? When and under what conditions was the picture viewed, and by which members of the team? Other questions concerned the ground (an application of paint to prime the support), the paint layer, the brushwork, the varnish. Still others concerned the scientific evidence. Was an x-ray photograph available or was it possible to have one made? How about ultraviolet or infrared photography? In the case of a painting on a wooden panel, what were the results of a dendrochronological investigation? (Dendrochronology is a method for dating wood by observing the pattern of rings that a specimen displays.) Another family of questions involved the circumstantial evidence. What was known of the painting’s provenance? Did this knowledge include any documentary evidence? Do any copies of the painting exist, and when were they painted? Was the picture ever reproduced in an engraving or other print? When and by whom? The team members were asked to describe each painting in minute (some would say laborious) detail—its subject matter, color composition, even mood. Finally, they were asked to render a preliminary opinion.

Working in this down-to-earth manner, the research project compiled dossiers on more than 600 paintings. Then, meeting in Amsterdam once a week, the team began to discuss its findings. One by one the pictures were assigned a letter: A for works of unquestioned authenticity, C for works by a hand other than Rembrandt’s, and B for works on which the team was unable to reach a decision. This sorting process is still going on.

“The way we come to a decision varies a great deal,”Bruyn said. “There are cases where the two members who actually went to see and describe the picture have come back with very clear-cut ideas. And these have been mostly so convincing that there was hardly any discussion needed. It was rather a matter of how to argue the case in the catalogue, how to present it. In other cases it can be very difficult. It could be that the authors of the description were hesitant themselves and presented all sorts of pros and cons. This is when the process becomes complicated and sometimes very long. Right now, while preparing Volume III, we have to deal with many landscapes, and landscapes turn out to be much more difficult than one might expect. One finds oneself considering picture A, which is absolutely convincing, and picture G, which is not, but in between there are B, C, D, E, and F, and now one has to determine at what stage of the gradation the answer begins to turn negative.

“We try to reach a consensus and almost always do. If you look at Volume I you will see that formal dissents are very, very rare. Everybody, of course, has a sort of preconceived idea of how the great master relates to his pupils and his followers or remote followers or even his falsifiers, and applies these ideas to decisions in individual cases. In the course of the years, however, several of our preconceived ideas have been subject to considerable change. For instance, we long assumed that certain paintings done on panels, which were not by Rembrandt, were imitations from a much later date. When the evidence of dendrochronology showed that this was not always the case—that some of these paintings were remarkably early—then it was obviously time fora reassessment. In other words, if the dendrochronology contradicts you once, it’s disturbing, but when it happens three or four times, you start thinking you’re mistaken in principle. In this case, pictures that we used to consider remote imitations or fakes we now think were probably done in Rembrandt’s workshop itself. We have greatly extended the idea of the workshop. One of the nice things about Volume II is a chapter on guild regulations and workshops in seventeenth-century Holland, especially with regard to paintings. We must not forget that Rembrandt was running a business. His workshop had to produce pictures. And that is what it did. The idea of art as the personal creation of an individual spirit is a fairly modern idea.”

ROUGHLY A THIRD of the works considered in Volume I by Bruyn and his colleagues were painted on wood— one of the easiest materials to place in time. The width of the rings added year after year by any tree varies with annual climatic conditions. By considering the overlapping patterns established in pieces of wood whose dates are known and that are from the same general locale (say, northern Europe), it is possible to work backward and construct a “map” of successive tree-ring patterns over many centuries. By matching a specimen of wood to this map, scholars can establish in what year the tree from which the specimen was taken was cut down. They can also tell whether or not panels have been hewn from a single trunk. The fact that the Bust of an Old Man in a Cap, in The Hague, about which the research project has reservations, was done on oak from the same tree that provided panels for Simeon in the Temple and Minerva in Her Study, which are accepted as authentic Rembrandts, is one reason why the Old Man’s status is listed as B rather than C. Of course, many painters may have ordered from a common supplier; and Rembrandt, who undoubtedly ordered in bulk, was buying for his students as well as himself.

Tree-ring dating is one of many relatively new tools that historians and conservators are employing in their critical analyses of old paintings. Another is microscopy. When focused on the surface of a picture, a microscope aids in the study of brushwork; when focused on a cross section, it furthers an understanding of how the paint layer was built up. Chemical analysis can identify the constituents of the paint, allowing the detection of anachronistic pigment. Ultraviolet photography is sometimes helpful in establishing where paintings have been retouched, and infrared photography may show how the paint was applied. X-ray photographs reveal the presence of particularly dense matter, such as leadwhite, which was used predominantly in the earliest stages of a painting. The revelations of an x-ray examination can be especially useful. Among other things, x-rays frequently expose “pentimenti” —the changes made in a painting’s composition. The latest advance is neutron-activation autoradiography, which involves irradiating a painting and then placing photographic film in contact with its surface for differing periods of time. Autoradiography can expose the chemical composition of a picture, allowing glimpses of the picture’s evolution.

Josua Bruyn is quick to point out that whereas scientific evidence rarely may suffice to condemn a painting, it can never, in and of itself, authenticate one. What it can do is expand the base of information that scholars have to work with—and, on occasion, resolve some vexing questions. For example, when xray photographs demonstrate the existence of additions or subtractions in the underlying composition of a painting, they are showing changes of mind. For obvious reasons, copies of existing paintings tend to display few pentimenti. Thus x-rays may suggest that a painting—regardless of who it is by—is either an original piece of work conceived of and composed by the artist himself or a copy made from another picture. Largely on the basis of x-ray photographs the team in Amsterdam concluded that The Artist in Oriental Costume— despite its “somewhat pedantic execution”—was by Rembrandt rather than a poor copy of some lost original.

The ability to probe the substructure of oil paintings has also made possible a reconstruction of the operating procedure that the young Rembrandt followed. Rarely, it seems, did he work out a design in advance on paper. (Almost no sketches for paintings survive.) Rather, he did his sketching right on the support, in dark paint. The ground, consisting of some combination of chalk, leadwhite, ocher, and quartz, would have been previously applied by a dealer, outside the workshop. On top of the sketch Rembrandt began his underpainting, or “dead-coloring,” in a brownish monochrome—probably mixed from leftover paint on his palette. Sometimes the underpainting was highlighted with leadwhite. (Several studies for etchings, such as Joseph Telling His Dreams, in the Rijksmuseum, and Ecce Homo, in London’s National Gallery, provide a good approximation of what the monochrome layer must have looked like.) To judge from the overlapping of various painted areas, Rembrandt worked in planes and in a fixed sequence, starting with the background. He seems to have employed “recipes” for the depiction of materials like fur, gold, and silk. In one picture after another they are handled in the same way. The effect of gold thread or jewelry is invariably suggested bythick ridges of bright paint. To bring a moustache into relief Rembrandt would often sculpt it by lightly scratching the surface with the butt of his brush.

The conclusions advanced by the research project, like those advanced byRembrandt scholars in decades gone by, ultimately come down to a matter of opinion. They come down, in other words, to connoisseurship: to what the editors of the Corpus characterize as “an inner measure of what a painter thinks or feels to be effective, permissible or beautiful.” This presupposes a special knowledge born of empathy and a special confidence born of knowledge—a mode of thinking “impossible to describe,” Bruyn told me, which compels the scholar to recognize in one painting and not another the hand of Rembrandt. It involves, in effect, the ability to walk in one’s own mind on the pathways created by another’s. Connoisseurship is a matter, finally, of discernment: of sensing when to discount an alarming piece of evidence and when to discount a reassuring one.

Despite the uncertainties involved, Bruyn is hopeful that the Corpus of Rembrandt Paintings will withstand both time and scholarly scrutiny. Although the Amsterdam team has accepted as authentic several paintings doubted by others (Bellona, in New York, and An Old Man Asleep, in Turin, are cases in point), Bruyn believes that few, if any, of the paintings rejected by the research project will enjoy a similar reversal of fortune. Rather, if the project has erred, it is most likely, in his opinion, to have been on the side of generosity. “That our judgments w ill remain firm one can reasonably expect,” he said. “I would never say—I would never pretend—that they should be considered definite. But more or less firm. Of course,” he conceded with a tiny smile, “the chances that in some cases we have been too optimistic are certainly not negligible.”

THE REMBRANDT Research Project has attracted its share of criticism, sometimes for reasons of its own making and sometimes not. There was, to begin with, a rather understandable reverence among many scholars for the body of work that had come to be associated with Rembrandt, and a corresponding reluctance to reopen the files on many pictures that had by the late 1960s acquired an almost mythical Rembrandtesque quality. The research project also encountered some rather fierce scholarly resistance to the very idea of a team approach. To be sure, a team effort was not without precedent. Von Bode and Hofstede de Groot, for example, had collaborated on the eight-volume Rembrandt catalogue. But many art historians were skeptical of the notion that half a dozen specialists, with differing fields of expertise, could rival in coherence and intuitive grasp the achievement of a lone, profoundly erudite connoisseur.

Some critics have voiced methodological concerns—noting, for instance, the highly variable conditions under which the pictures were examined, or the rather lengthy interlude between the viewing of the pictures (in the early 1970s) and the writing of the critical summaries (in the early 1980s). The most commonly voiced fear has been that the research project, in its zeal to rid the Rembrandt oeuvre of impurities, would prove to be overly reductionist. Seymour Slive, a professor of art history at Harvard and a former director of Harvard’s Fogg Museum, worried openly in 1973 that the team might end up “throwing out the baby along with the bath water.”

The members of the research project at times provoked anger for behavior that was considered high-handed. For one thing, they tended to be secretive. Although most museums and private collectors cooperated with the Dutch researchers, the team members generally kept their observations to themselves— perhaps a justifiable procedure, given the scope of the undertaking and the preliminary nature of the on-site inspections, but one that did not always sit well with professional colleagues. A visiting delegation from Amsterdam came to be known among curators as “the hit team.”In preparing the first volume of its catalogue for publication, the project neglected to give the owners of the pictures involved any notice of its conclusions or its reasoning. Josua Bruyn concedes that this was a needless breach of courtesy. A comment I have heard repeatedly about the members of the team is “Well, they’re Dutch.”

Despite these early tensions the work of the group seems to have earned a degree of acceptance. Seymour Slive, w ith whom I spoke one morning at his Cambridge, Massachusetts, home, was still somewhat concerned about the pessimism of the project with regard to certain pictures. “The conception that onesenses is perhaps a little too rigid for Rembrandt,” he said. “Even Rembrandt had his Monday mornings, and even his pupils had days when they were better than the master.” Slive believes that several of the pictures that the project listed in its B (“can’t decide”) section in Volume I—including the artist’s self-portrait in Stockholm and portraits at the Maritshuis, in The Hague, and in the Getty Museum, in Los Angeles—should be promoted to A (“authentic”) status. And he asks, what is the status of The Good Samaritan, in London? The picture is signed, and dated 1630, but is not mentioned in Volume I, devoted to the Leiden years. Will it be included in Volume II and assigned a post-Leiden date?

However, on the whole, Slive said, the work of the Rembrandt Research Project ought to be applauded. “You have to remember where we were with regard to Rembrandt’s work,” he said. “Rembrandt was an unusually productive artist. You must jump forward to Picasso before you find someone comparable. We have almost three hundred etchings—Rembrandt is rightly regarded as the greatest etcher of all time— and about fourteen hundred drawings, and between three hundred and six hundred paintings. The first catalogue raisonné of Rembrandt’s etchings was done in 1751, and has been worked over since by ten or twelve scholars. A catalogue of the drawings was not done until two centuries later. It was published in 1954 by Otto Benesch. But until the Rembrandt Research Project, this kind of careful analysis had never been done on all of the paintings—an analysis that included the scientific and technical evidence, iconography, a list of all the known copies, the provenance, the ground, the support, and so on. There had never been a full-dress catalogue raisonné. I think they’re too wordy in their descriptions, and I’d like to see more visual evidence, more photographs.”

With regard to the previous generations of art historians who assigned a surfeit of pictures to Rembrandt’s hand, Slive is inclined to be charitable. “They did the spadework that made the Rembrandt Research Project and a lot of other work possible,” he says. “In my view, it’s not proper to spit on the head of the man whose shoulders you’re standing on.” Slive guesses that perhaps half of the paintings from the period 1631—1634 which Bredius attributed to Rembrandt cannot withstand critical scrutiny. Two thirds of that number, he adds, have already been condemned in the court of scholarly opinion.

WHAT HAPPENS TO rejected paintings? To begin with, they are revalued. This may have a devastating effect on private collectors. (To paraphrase Dr. Johnson, there are many people w ho have but one Rembrandt, and it’s a wrong one.) Museums, though, generally manage to take their losses in stride. For one thing, their insurance premiums should decrease. A few of the better seventeenth - century Rembrandtesque paintings have found new, though perhaps temporary, places in the work of the master’s most gifted students—“the dustbins into which unaccepted Rembrandts are so often thrown,” as Josua Bruyn says. (Bruyn ventures, carefully, that “in considering the expansion of the work of Willem Drost, one shouldn’t forget to look at The Polish Rider.”) Major museums with significant Rembrandt collections have tended to hold on to their non-Rembrandt Rembrandts; these pictures represent an important scholarly resource for understanding the master’s environment. Many of the very finest paintings remain on display, with appropriate emendations of their nameplates. The popular Woman Paring Her Nails, at the Metropolitan in New York, now bears the attribution “Style of Rembrandt.” So does Pilate Washing His Hands, in the same museum.

A good many pictures, of course, have been removed from the galleries. One day last winter Walter Liedtke, the curator of Dutch and Flemish paintings at the Metropolitan, showed me into a large storeroom on the second floor where much of the museum’s collection of European paintings resides unseen. The paintings hang on floor-to-ceiling racks that can be rolled out for viewing. A row of forty racks, each holding as many as ten paintings, lies on either side of a wide walkway—Italian, Spanish, and French pictures on the left, English, Dutch, and Flemish on the right. Two of the racks hold nothing but what Liedtke calls “our not-Rembrandts.”

He rolled them out. It was oddly moving to watch the paintings slide into view, like seeing bodies pulled from drawers in a morgue—unclaimed, unidentified, yet somehow familiar. Here were The Admiral and His Wife (a pair of companion paintings), a painting titled Portrait of a Woman in a Yellow Dress, A Sybil, the Head of Christ that I remembered, and more than a dozen other pictures, including two that once passed as portraits of Rembrandt’s wife, Saskia, and his son, Titus.

“Most of these,”Liedtke said, embarking on a short tour of this potter’s field, “were acquired by the Metropolitan before the Second World War—all but two, actually. The portraits of the admiral and his wife have been assigned to Ferdinand Bol, but I don’t think that’s likely. They could be English, from around 1720. The Sybil has been assigned to Willem Drost. I think that’s right. I think it’s Drost. The portrait of Saskia is an imitation. Look at the signature. It’s lying in a pool of old paint. The artist kept doing it over to get it right. The portrait of Titus is an imitation too—probably nineteenth-century. See how flat the picture is? The 1800s produced a lot of Rembrandts, but the imitators didn’t have the same way of seeing things as the seventeenth-century painters. Even a minor painter in the seventeenth century had to produce something like a replica of the real world. A figure had to have solid form, and there had to be depth in the space around it. In the nineteenth century that concept of painting was relaxed. A picture came to be seen as a decorated surface. It was meant to be an analogy to reality, not a replica. As a result, even when they were copying, the nineteenth-century painters rarely got Rembrandt right.”

Why, I asked Liedtke, if their authenticity was in dispute, were two paintings still hanging in the Met over the name Rembrandt? “Well,” he said, “there are too many ‘Rembrandts’ hanging on the walls of a lot of museums. It’s a delicate problem. Some of the paintings are gifts, and they’ve been given on the condition that they hang. For all we know, some of the pictures, including ours, may actually be Rembrandts. We don’t move more quickly because it’s embarrassing to say ‘This isn’t by Rembrandt’ and then a few years later have to announce, ‘Oh, well, it really is.’ Also, with some paintings, which are perhaps not by Rembrandt, there is a certain obligation to preserve the attribution until you can say who they’re by. This is the case, I think, with The Polish Rider. And finally, why take down a picture that you thought was good enough to hang in the first place? The picture hasn’t changed.”

I was pleased to discover that museum curators think this way. Next year a major exhibition of Rembrandt paintings will be mounted in the United States. It w ill consist of ten pictures each from the Metropolitan, the Rijksmuseum, and the Staatliche Museen. And The Man With the Golden Helmet, whomever it’s by, will be among them.

—Cullen Murphy