By the early twentieth century, more than six hundred paintings bore Rembrandt’s name. By the end of it, that number had been cut by nearly half, and some of the most admired works in major collections had been reassigned to pupils, followers, and anonymous imitators. The Rembrandt Research Project, founded in 1968 by a team of Dutch art historians led by Josua Bruyn (1923–2011), set out to discipline a corpus that had swollen over three centuries of enthusiastic, and often careless, attribution. The results were severe. In the first three volumes of A Corpus of Rembrandt Paintings (1982, 1986, 1989), covering the period up to 1642, the RRP catalogued 286 paintings from the roughly 330 accepted by Abraham Bredius (1855–1946) in his 1935 catalogue raisonné. Of these, 146 were accepted, 122 rejected outright, and just twelve left in a category of acknowledged doubt. Horst Gerson (1907–1978), revising Bredius in 1969, had already reduced the total oeuvre to approximately 420 paintings; the RRP drove it below 250. The number of signed self-portraits alone was halved. Museums across Europe and America found themselves staring at paintings whose labels had become, almost overnight, institutional embarrassments.
The problem, though, was structural, and it did not begin with modern scholarship. Rembrandt ran a large and commercially productive workshop in Amsterdam from the mid-1630s onward, training pupils who were themselves already competent painters. Some fifty artists are now associated with his studio, of whom about twenty can be documented by name. Joachim von Sandrart (1606–1688), a near contemporary, recorded that Rembrandt earned between 2,000 and 2,500 guilders annually from tuition fees and the sale of student work alone, on top of income from his own paintings and prints. Pupils paid him one hundred guilders a year for instruction and were set to work copying the master’s compositions, painting from the same models, and working through shared pictorial problems, often in partitioned spaces separated by paper or linen walls. This was not a factory in the Rubens mould, where assistants executed sections of the master’s compositions under direct supervision. It was something harder to categorise: a teaching environment in which the boundaries between master and pupil were porous by design, where pupils absorbed the master’s idiom so thoroughly that their independent productions could, and did, pass for his work, sometimes within his own lifetime.
The RRP’s original methodology rested on an assumption that Rembrandt executed his paintings without workshop assistance, treating collaboration as the exception rather than the rule. Its classificatory system, dividing works into A (accepted), B (uncertain), and C (rejected), had no framework for collaborative production. That assumption, as Ernst van de Wetering (1938–2021) later acknowledged, did not accord with what documentary evidence actually suggested about workshop practice in the seventeenth century. Van de Wetering, who had joined the RRP in 1968 as a junior assistant and became its chairman in 1993 after a deep ideological dispute led to the withdrawal of the three other surviving members, quietly abandoned the A/B/C system from volume four onward. His approach was processual rather than stylistic: where Bruyn and his colleagues had described images and compared surfaces, Van de Wetering described processes, asking whether a painting’s genesis indicated that the maker was also the person who had conceived the composition. The question sounds simple. In practice, it proved enormously difficult to answer, and it led Van de Wetering in a direction his former colleagues would not have taken.
In A Corpus of Rembrandt Paintings VI: Rembrandt’s Paintings Revisited (2014), the final and most controversial volume of the project, Van de Wetering reinstated forty-four paintings rejected in the first three volumes and added approximately twenty more that had not previously been catalogued, arriving at a total oeuvre of some 336 works. This was a staggering reversal. It was also the work of a single scholar, carried out after the collective authority of the original team had collapsed. What does it mean when one person’s eye undoes the considered judgement of four?
An Old Man in an Armchair belongs squarely to this contested territory. The painting was almost certainly among the pictures at Chiswick House collected by the 3rd Earl of Burlington (d. 1753), passing by inheritance to the Dukes of Devonshire. The National Gallery acquired it in August 1957 from the 11th Duke under the Finance Act of 1956, and it was praised on arrival as a powerful example of Rembrandt’s work of the 1650s. Kenneth Clark (1903–1983), in Rembrandt and the Italian Renaissance (1966), remarked on its pronounced Venetian quality, citing the saturated colours and the use of glazes to create depth of tone, and compared it to Tintoretto, though whether Rembrandt would have known the Venetian master’s work at first hand remains uncertain. Gregory Martin, writing in Apollo in 1969 under the pointed title ‘The Death of a Myth’, was among the first to question the attribution, and by the late 1960s a more systematic comparison with securely attributed works, combined with technical analysis of materials, led scholars to recatalogue the painting as the work of a contemporary follower.
The technical evidence assembled in subsequent decades appeared to support this conclusion. The National Gallery’s own research, presented in the 2010 exhibition Close Examination: Fakes, Mistakes and Discoveries, drew attention to the painter’s use of comparatively pure pigments: the bright red on the sitter’s left cuff consists of pure vermilion, the orange-yellow streaks in his robe are natural earth pigment laid unmixed, and the deep purplish glazes on the gown and hand are composed solely of red lake pigments. This handling of red lake glazes would be unusual in securely attributed Rembrandts of this period. Neil MacLaren’s National Gallery catalogue, revised and expanded by Christopher Brown in 1991, listed the painting as ‘follower of Rembrandt’, and the technical study by David Bomford, Jo Kirby, Ashok Roy, Axel Rüger, and Raymond White (2006) maintained this position. Christian Tümpel (1937–2009), in Rembrandt: All Paintings in Colour (1993), offered a slightly different formulation, cataloguing it as the work of an unknown collaborator in Rembrandt’s studio. For the better part of three decades, the painting hung under a label that politely withdrew a claim the Gallery had paid for.
Then came Corpus VI. Van de Wetering, cataloguing the painting as no. 221 and dating it to c. 1652, argued that its bold and innovative brushwork was too experimental, too searching, to be the product of a follower imitating a known style. A follower, by definition, follows: he works from an established model and adapts it to his own, usually lesser, capacities. But the handling of An Old Man in an Armchair does not follow. It anticipates. The loose, suggestive treatment of the left sleeve, achieved with half a dozen very broad strokes, and the way light is used to dissolve form rather than to define it, represent, in Van de Wetering’s reading, an early and significant step in Rembrandt’s development toward the rough manner of his later works. The question Van de Wetering posed is a fair one: is it really plausible that a pupil or follower, working within the orbit of the studio, would have pushed the painterly idiom in a direction the master himself had not yet fully explored?
The National Gallery’s response was measured. A spokesperson, quoted at the time of the Corpus VI publication, noted that ‘An Old Man in an Armchair is a picture that has generated much debate over the years, as many Rembrandts do, and we look forward to further discussions concerning its attribution.’ The Gallery has since amended its label to ‘Probably by Rembrandt’, a formulation that holds the question open without committing the institution to a definitive answer. It is worth pausing on that word, ‘probably’. It is not ‘yes’ and it is not ‘no’. It is a label designed to manage institutional uncertainty in public, and it says more about the limits of connoisseurship than it does about the painting.
Nor was the reception of Corpus VI uniformly approving. The art historian Michael Savage, reviewing the volume in detail, noted that while lengthy argument was presented for the reattribution, he remained unconvinced. The Burlington Magazine, in an editorial assessing the project’s final phase, observed that Van de Wetering was ‘never more impressive than when using every weapon in his armoury’ to rescue rejected paintings, but also noted that the volume, unlike its predecessors, almost entirely ignored rejected works, offering no consistent assessment of condition or provenance for many of the most contentious entries. The implied certainty of a catalogue that gives precisely 336 paintings to Rembrandt’s own hand sat uneasily with a methodology that claimed to embrace probability and acknowledge doubt. And there was a more uncomfortable observation: Van de Wetering had invoked Bayesian reasoning as the theoretical underpinning of his approach, but the catalogue itself was, as Savage noted, profoundly frequentist in its conclusions. Paintings either were or were not by Rembrandt. No probabilities were assigned, no grey areas acknowledged.
The broader problem remains unsettled. Signing practices in the seventeenth century offer no secure ground. Rembrandt signed his paintings far more frequently than Rubens (1577–1640), who signed only about five works in his entire career, but this frequency has not made authentication easier. Contemporary sources cannot confirm whether pupil works sold from the studio also carried the master’s signature. Bruyn himself conceded, in volume three of the Corpus, that while it was ‘conceivable that Rembrandt signatures were appended in the workshop by his studio assistants, as a kind of trademark’, there was no direct evidence for the reverse situation either: the master putting his own name on a pupil’s work. The theory of studio signing was, in effect, developed to explain away paintings that bore signatures but failed stylistic tests. Once those stylistic tests themselves were called into question, the theory lost its anchor.
Paintings dismissed as workshop productions for decades have, in several recent cases, been confirmed as autograph after renewed examination, sometimes overturning judgements that had relied largely on photographic reproductions rather than direct study of the work itself. The pattern has repeated often enough to constitute its own kind of evidence: the same apparatus that once expelled paintings from Rembrandt’s oeuvre is now being used to readmit them. On what, then, does authority finally rest? On the eye, which the RRP was founded to discipline? On scientific analysis, which can identify materials but cannot say who held the brush? On the institutional weight of whoever speaks last?
An Old Man in an Armchair sits at the centre of these questions, and it is unlikely to leave them any time soon. The painting’s ambition is visible in its handling: the rough, suggestive strokes of the left sleeve, the dissolved contours where the figure meets the chair, the saturated warmth of the red lake glazes. These are the marks of a painter testing the limits of an idiom, pressing paint toward something it has not yet been asked to do. Whether that painter was Rembrandt, working his way toward a new pictorial language, or someone close enough to him to anticipate where that language might go, is a question the evidence cannot conclusively settle. Perhaps the more useful question is what it tells us about the nature of the line itself: the line between master and pupil, between invention and imitation, between the hand that conceives and the hand that executes. The Rembrandt Research Project spent nearly half a century trying to draw that line with precision, and the result, as often as not, was to demonstrate that precision is the one thing the evidence will not support.
References
Bomford, D., Kirby, J., Roy, A., Rüger, A. and White, R. (2006) Rembrandt. London: National Gallery Company
Brown, M. (2014) ‘Rembrandt expert says National Gallery utilisation was “a vast mistake”‘, The Guardian, 23 May. Available at: https://www.theguardian.com/artanddesign/2014/may/23/rembrandt-expert-national-gallery-painting-old-man-armchair (Accessed 26 Decemebr 2024)
Clark, K. (1966) Rembrandt and the Italian Renaissance. London: John Murray
MacLaren, N., revised by Brown, C. (1991) National Gallery Catalogues: The Dutch School, 1600–1900. 2nd edn. 2 vols. London: National Gallery Publications
Martin, G. (1969) ‘The Death of a Myth’, Apollo, XC, pp. 266–267
RKD – Netherlands Institute for Art History (n.d.), Follower of Rembrandt van Rijn (1606–1669) ?, An Old Man in an Armchair, m RKDimages database entry no. 286285 . Available at: https://rkd.nl/images/286285 (Accessed 28 December 2025)
Schwartz, G. (1985) Rembrandt: His Life, His Paintings. Harmondsworth: Penguin
Tümpel, C. (1993) Rembrandt: All Paintings in Colour. Antwerp: Fonds Mercator
Van de Wetering, E. (2014) A Corpus of Rembrandt Paintings VI: Rembrandt’s Paintings Revisited, A Complete Survey. Dordrecht: Springer
Van de Wetering, E. (2016) Rembrandt: The Painter Thinking. Amsterdam: Amsterdam University Press
Van de Wetering, E. (2014) ‘A reattribution to Rembrandt of Old man in an armchair (1652) in the National Gallery, London’, The Burlington Magazine, 156, no. 1335 (June 2024), pp. 394–404.Available at:https://www.burlington.org.uk/archive/back-issues/201406 (Accessed 26 December 2025)
