Before Gerard David’s Holy Family surfaced in 2003, this picture had been in one single family for more than a century. The painting was acquired by the Parisian collector Prosper- Victor Martin le Roy in 1887 in his native city. We are still unaware of the exact nature of the purchase, whether he bought the painting at a sale or from a dealer or other collector. Martin Le Roy married Marie Adèle Jeanne Lebaudy, the eldest daughter of the well-known industrialist and politician Gustave Lebaudy. After his death in 1918, the painting ended up with his daughter Jeanne, who had married Jean Joseph Marie Anatole Marquet de Vasselot, an art historian specialised in medieval applied arts, especially enamels. Marquet de Vasselot became a curator in the Louvre in 1902 and later director of the Musée Cluny in Paris. He was one of the authors as well as the editor of the massive five volume catalogue of his father-in-law’s collection, that was published between 1906 and 1909 and in which the Holy Family by Gerard David was discussed in detail by his Louvre colleague Paul Leprieur. Although both Martin le Roy and his son-in-law Marquet de Vasselot donated works of art to the Louvre and the Musée Cluny, the larger part of the collection remained with the descendants of Marquet de Vasselot, following his death in 1946.

When the Holy Family was auctioned in 2003, its appearance left something to be desired. However, as could be established from examining the picture surface with the aid of ultraviolet fluorescence, its condition was superb for an early 16th-century painting, due to the fact that it had remained in one collection for so long.i The remarkable condition became even more visible after the dirt, yellowed varnish and old retouchings were removed. Minor damages turned out to be limited to the edges of the panel, probably due to inconsistent framing. The metamorphosis of the painting was breathtaking. The beautiful colours had returned – like the stunning landscape in blues and greens – as well as the sparkling light that lends the picture its transparency – the Virgin’s veil is indeed transparent – its volume, visible in the play of light and shadow on the clothes, skin and the Virgin’s golden hair. Finally, the superb painting technique revealed itself in all its splendour, especially in the flesh colours.

The Holy Family is painted on a single board of oak and on the back written texts in red and white chalk can be made out, as well as the Christie’s stock number VE469 relating to the 2003 sale at Christie’s in New York. The mark J.M.V. in red chalk refers no doubt to its previous owner, Jean Marquet de Vasselot. The picture may have been somewhat larger originally. The right side is completely intact, showing an unpainted edge and an upstanding ridge, called a barbe, indicating that the picture was painted within the frame, a common feature at the time. The other three sides have been cut, but the loss appears to have been extremely limited. I am convinced that the painter intended to present his Holy Family in this rather cropped composition, placing the Virgin and Child very close to the forefront, therefore enforcing a much deeper intimacy on the viewer and potential buyer.

The infrared reflectogram (IRR) mosaic of the painting is difficult to read because of the pattern of craquelure in the paint surface (fig. 1).ii Nevertheless the underdrawing is visible; it has been applied with a dry material, probably black chalk or carbon, only to define contour lines. The artist did not make use of parallel hatching to prepare for shaded areas. The underdrawing is sometimes quite difficult to see, because in painting out the composition the underdrawn contours have been followed precisely and are therefore often covered by painted contours.

Nevertheless, it is visible in those areas where the painted contours do not contain any black pigment, where the painting deviates from the drawn contours and where the underdrawn form was not painted at all or only in part. The right ear of the Virgin is a good example. In the paint layer the ear was covered by the transparent veil that Mary wears over her head, and can therefore only roughly be detected. The X-Radiograph that was made of the painting does not show the ear at all, indicating that no lead white was used for its flesh colours. The IRR image shows that in the underdrawing the ear was prepared very carefully with a thin line of black chalk (fig. 2).iii In some cases the underdrawn form was changed during the painting process. The apple the Christ Child holds in his left hand was enlarged at a later stage, painted over a part of the Virgin’s right hand; the thumb of the Christ Child was corrected somewhat to adapt to the larger apple (fig. 3). In contrast, the two rotten apples that Joseph holds on a wooden plate, were reduced in scale in the paint layer.iv In several areas visible in the IRR mosaic (fig. 1), one has the impression that some contours have been strengthened, like those of the bowl in Joseph’s hands, or the outlines of his fingers. The contours of the fingers of Christ’s proper right hand, held around his mother’s neck, were most certainly enforced in the underdrawing process. As for the landscape and the castle in front of it, there does not seem to be any indication that it was prepared in the underdrawing. It is unclear whether the contour drawing was made free hand, or should be regarded as the result of the use of a cartoon for mechanical transfer. The enforced contour lines, such as the fingers of Christ’s proper right hand, might be an indication for the use of a cartoon. It may have been that in some instances the traced lines – assuming that a tracing cartoon was used, not a pouncing cartoon – lost their visibility and were therefore reinforced free hand.

The first time the picture was presented to a wider public was in 1902, during that most famous exhibition, the Exposition des Primitifs Flamands et d’art ancien in Bruges, where no less than 413 paintings, diptychs and triptychs were presented. The Holy Family was exhibited under the name of Quentin Metsys, as no. 343. The catalogue of W.H. James Weale was subject to many critical reviews, which was quite understandable, given the fact that so many different and unknown pictures were brought together and many of them were never critically studied at all. In fact, the exhibition and its catalogue proved to be an important catalyst for the rediscovery of Early Netherlandish painting and the flowering of connoisseurship. Although Weale was in fact responsible for the exhibition and catalogue, he was not a connoisseur, not even an art historian, but instead a historian and an archivist. Some distinguished art historians were members of one of the several scientific committees, among them Georges Hulin de Loo and Max J. Friedländer. The latter wrote two different reviews on the exhibition and catalogue, whereas Hulin de Loo even produced an alternative scientific catalogue, presenting his outspoken views on many attributions, which fundamentally differed from those James Weale had presented.

Hulin de Loo’s entry on the picture was and still is extremely impressive and in the main still stands today. He was very outspoken in regard to the attribution. He regarded the painting as a late work by the hand of Gerard David, painted after 1515, when the artist enrolled in the Guild of Saint Luke in Antwerp. That the concept of the composition and the intimacy between mother and child derived from Quentin Metsys was, according to Hulin de Loo, no doubt the reason for the wrongful attribution in Weale’s official catalogue.v Hulin de Loo pointed out that the painting was remarkably close to the various versions of the Virgin and Child with the milk soup, especially in regard to the Virgin’s face and to the view from the window (fig. 4). For the facial type of Joseph he referred to the Adoration of the Magi in the National Gallery, London, although he did not regard that painting as by David’s own hand (fig. 5).vi

With regard to the Holy Family, Max Friedländer fully agreed with Hulin de Loo, stating that the painting was an ‘excellent late work from David’ (Vortreffliches Werk aus der Spätzeit Davids).vii Friedländer was the first to connect the Holy Family to two copies after it, at the time both in Munich (figs. 6-7).viii Two years earlier Friedländer had attributed these two pictures to the so-called Pseudo-Mostaert, e.g. Adriaen Isenbrant.ix The attribution of the Holy Family to Gerard David was generally followed by the other reviewers (Hymans, op. cit.; Migeon, op. cit.), but Weale, who himself had written a monograph on Gerard David in 1895, did not accept the new attribution, stating that there is “neither tradition nor documentary evidence” in favour of Friedländer’s classification of the picture as a late work by Gerard David. He stated that the Holy Family was painted by an Antwerp master under Metsys’ influence. His point of view was more or less shared by Franz Dülberg, hardly a connoisseur himself, who wrote a review in the Zeitschrift für bildende Kunst.x

Eberhard Freiherr von Bodenhausen, in his 1905 monograph on Gerard David, dismissed the painting as being an autograph work, instead regarded it as a copy after a lost work by David, painted under the influence of Quentin Metsys. From that moment on, the painting remained in the vicinity of Gerard David and although it was usually regarded as a late work and therefore painted for the Antwerp market, some authors remained insistent in stating that it was not by the master himself, but by a follower, like Hans van Miegroet in his 1989 monograph on Gerard David. However, Van Miegroet’s negative judgement of the picture were in fact not based on the version now at auction, but on the copy in Cologne.

Van Miegroet, like Jean Wilson in 1983 (and again in 1998), added a fourth version to the discussion (fig. 8). This painting remained in the well-known collection of Heinz Kisters in Kreuzlingen, until it was auctioned at Christie’s in London in 1970. Its present whereabouts are unfortunately unknown.

There has hardly been any discussion on the exact nature of the relationship between these four versions with the Holy Family, apart from the general assumption that the two versions in Cologne and Munich should be regarded as copies after the Gerard David painting, coming from the workshop of Adriaen Isenbrant. Only Jean Wilson took a somewhat different view with regard to the version in Cologne (fig. 6).xi She regarded the Cologne Holy Family as one of three paintings from the so-called Style II Grouping within Adriaen Isenbrant’s oeuvre. According to her, this hand is visible, too, in parts of the Coronation of the Virgin in the Church of St. James in Bruges, an altarpiece that is extremely well documented as being painted by Albert Cornelisz between 1517 and 1522, with participation of one or two unknown assistants.xii Wilson states that only the figure of the Archangel Michael is by the same hand as the Cologne Holy Family and the choir of angels to a lesser degree. According to Wilson, therefore, the Cologne Holy Family was painted by an artist that assisted Albert Cornelisz in the genesis of the Coronation of the Virgin. However, Dorien Tamis has suggested on the basis of her detailed research on the altarpiece that its present surface and most certainly all the faces, betray the hand of Albert Cornelisz.xiii This may be true, but I still see some dissimilarity between the two prophets on the left and right foreground and the standardised angels. Nevertheless, Tamis is certainly right when she writes: “Other works by Cornelisz should be sought in the confusing hotchpotch of paintings that forms the so-called ‘Isenbrant’ group”.

How then, should we scale the Holy Family, in terms of its attribution, its status and its connection to the other three versions? I am convinced that the picture was painted by Gerard David himself, after 1515, when David became a member of the Antwerp Painters’ Guild and before his death in 1523. The high quality of the painting, the superb refinement of the brush strokes, visible in every single detail of the painted surface, as well as the superb nuances between light and shadow, giving form and volume to the painted textures, all point in the direction of Gerard David himself. Of course, there are quite a few parallels with other paintings by David’s hand with regard to facial types, possibly based on the repeated use of pattern drawings, but that in itself is no argument for attribution.xv Nevertheless, the flat, broad, almost square frontal face of Saint Joseph was used time and again in David’s studio, as in the Head of Christ in Philadelphiaxvi, the risen Christ on the Resurrection in New York (fig. 9)xvii or the Christ on the Transfiguration in the Bruges Our Lady’s Church.xviii The facial features themselves, including the odd strand of hair on Joseph’s forehead is another aspect that points to Gerard David as the maker, as follows from comparison with the face of Moses in the aforementioned Transfiguration or the kneeling Magus in front of the Christ Child in the London Adoration of the Magi (fig. 5). The facial type of the Virgin is Gerard David all over and the transparent veil that is draped on top of her head is a stock-in-trade model from his workshop that was taken over by other painters in Bruges, Benson and Isenbrant included. There are literally dozens of other examples of this facial type in David’s later production, such as the seated Virgin in the London Adoration of the Magi, or the Virgin in the left exterior wing of the Baptism Triptych in Brugesxix or the Virgin Mary on the Rest on the Flight into Egypt, especially the version in New York.xx

It is, however, with the Virgin and Child with the milk soup, we are able to find the most obvious connection. As can be judged from comparison with the example in the New York Aurora Trust (fig. 4), the Virgin’s face is almost identical, as is the transparent veil, that follows an identical pattern. But there is more that connects the two compositions. The two spaces are most certainly very closely connected, although the extreme cropped composition of the Holy Family hardly allows us to make that comparison. When the less cropped version of the Holy Family in Munich is compared to the Virgin and Child with the milk soup, the spatial connection is much more evident. Finally, the still life of the brownish apple, bowl of milk soup, wooden spoon and trencher in the Virgin and Child with the milk soup returns in a more compact form in Joseph’s hands in the Holy Family. And then there is the landscape. The four versions of the Virgin and Child with the milk soup Maryan Ainsworth discusses in her book on Gerard Davidxxi all share the same vista, which differs from the view in the Gerard David Holy Family. Other versions of the Virgin and Child with the milk soup, such as the version in Strasbourg (fig. 10), display a landscape that is based on the same model that was used for the view through the window of the painting under discussion.xxii

Like Hans van Miegroet, I think that paintings like these, or for that matter the various versions of the Virgin and Child with the milk soup or the Rest on the Flight into Egypt, were painted for the large-scale Antwerp market, by Gerard David, his assistants and other Bruges workshops, like those of Adriaen Isenbrant or Albert Cornelisz. It is for this reason that David may have adapted his palette to the Antwerp couleur locale, embodied by Quentin Metsys. Indeed, the soft and subdued colour pattern, with soft purples, reds and blues is very reminiscent of Metsys’ paintings in this period. The same is true in regard to the heightened intimacy between mother and child as well as the domestic atmosphere that David presented through his paintings with the Virgin and Child with the milk soup and the Holy Family. The same intimacy between mother and child, intensified by the use of subdued, but saturated colours, can be witnessed in Metsys’ wonderful Virgin and Child before a landscape in Rotterdam, from the same period, c. 1518-20 or the 1529 dated Virgin and Child in a domestic interior in the Louvre (fig. 11), painted shortly before he died.xxiii

Both David and Isenbrant remained in Bruges all their life and they did not produce their pictures in Antwerp. Therefore, the various versions of the Holy Family were in all likelihood painted in Bruges, and not in Antwerp. Dendrochronological analysis by Peter Klein points to an earliest possible date of 1511 for David’s Holy Family, which fits quite well with the post-1515 date most authors agree upon.xxiv One of the other panels, the version that used to be in the Kisters collection in Kreuzlingen (fig. 8), was analysed by Klein as well and may have been painted as early as 1506.xxv It is, however, more than likely that the David Holy Family was painted earlier. Based on a limited number of photographs and illustrations there is no doubt whatsoever that the Kisters version is the inferior painting of the two. In addition, detailed study of the IRR mosaics of the version under investigation has shown that several pentimenti can be distinguished, notably in the enlarged apple in the Christ Child’s hand and the adapted flexible thumb. These corrections in the painting process were taken over in the Kisters version, which comes nowhere near the refinement and play of volume the present Holy Family so evidently demonstrates. However, the Kisters version is the only direct copy after that picture and may derive from the David workshop.xxvi

The two paintings in Munich and Cologne (figs. 6-7) are evidently based upon the Holy Family by Gerard David, but cannot be regarded as slavish copies, at least not to the level of the Kisters version. First of all, the spatial arrangements of those two versions differ fundamentally. In Gerard David’s painting, the figures are extremely close to the picture plane and although only the right side has remained completely intact, the extreme cropping must have been intentional. The proportions of the Virgin and Child in the foreground have been blown up to such a degree that they more or less occupy seventy percent of the painted surface, Joseph adding another twenty, leaving just enough space for a view through the window, which is placed diagonally behind the Holy Family. In the Cologne, and to an even stronger degree in the Munich Holy Family, the proportions of Mother and Child have shrunk considerably, the painter created much more space around the figures and the window runs parallel to the picture plane, creating a completely different atmosphere and a less coherent image.

Apart from the diverging spatial definition, the paintings differ stylistically. Unlike the Kisters Holy Family (fig. 8), the two pictures in Cologne and Munich were in all probability not produced in Gerard David’s workshop. It is entirely possible that they were painted in Adriaen Isenbrant’s studio, but not necessarily. As was discussed previously, the Cologne Holy Family was linked to Albert Cornelisz’ Coronation of the Virgin in St. James’s Church in Bruges. And although the Munich painting is apparently a slightly better picture, the stylistic bonds between the two versions are strong enough to point in one and the same direction. Whether this is Adriaen Isenbrant, whose immense, but ill-defined oeuvre still awaits comprehensive study, or Albert Cornelisz, a painter with only one documented altarpiece, that in its turn has a very complicated history of production, still needs to be resolved.

Particularly fascinating are the views that are offered through the windows of the Munich and Cologne versions. Unlike David’s Holy Family where we encounter a hilly landscape in subdued colours in the typical late 15th-century Bruges tradition, the Munich and Cologne views have a much more outspoken Antwerp character. The obsessive fantasy architecture we encounter in the Cologne painting is no doubt based on the then very modern and decorative designs of Antwerp Mannerism, as spectacular drawings in Berlin, Vienna, Göttingen and London demonstrate.xxvii The typical run-of-the-mill landscape in bird’s eye-view with city walls and farm houses on the Munich picture was overly popular in the 1520s in Antwerp as well, as is testified by countless other examples.xxviii

Drawings, cartoons and patterns played an important role in the standardization and recycling of popular motifs, they were the carriers of information and every artist assembled as many drawings for his portfolio as possible. Drawings were essential in the practice of the painter and they were interchangeable, as is easily demonstrated with the famous case of Ambrosius Benson versus Gerard David, on 11 February 1519, exactly during the period at hand. xxix Benson wanted David to return his personal belongings which were locked in two chests, including several drawings, most of them from Benson himself, others from Albert Cornelisz and drawings David had taken from Isenbrant’s house. These drawings carry different names, like “patronen” or “proiectien” and in another document the word “bewerpen” was used. In short, artists copied drawings from each other, even charged money for it, as Albert Cornelisz had done to Benson. However, we do not always need the physical existence of a drawing or cartoon, to understand that they were used in the production process. As Maryan Ainsworth made clear, the popular Virgin and Child with the milk soup was reproduced with the aid of a cartoon, as could be judged from the IRR mosaics.xxx The Virgin’s face in the Holy Family, including the veil, was not only used for the various versions of the Holy Family, but we encounter exactly the same model with exactly the same veil, that is folded in exactly the same pattern in the various versions of the Virgin and Child with the milk soup. David must have had a model drawing for this particular type in his studio, or perhaps a painted modello, which could be used again and again.xxxi However, the model was known outside David’s studio as well, since it was re-used in the two versions of the Holy Family in Munich and Cologne. xxxii Moreover, this model was re-used in other compositions as well. A painting that was last seen at a Christie’s sale in 1995 portrays the Virgin and Child before a landscape.xxxiii Although the Virgin does not wear a veil, her face and that of the Christ Child, are exact repetitions of the couple in the Munich Holy Family.xxxiv Two smaller versions of the Virgin and Child before a landscape are to be found in Museum Catharijneconvent in Utrecht and in the Museu Nacional de Machado de Castro in Coimbra (Portugal), although in these versions the Virgin does wear a transparent veil.xxxv Although these two pictures, too, have been given to Adriaen Isenbrant, they stem from different hands than the larger Virgin and Child before a landscape, a conclusion that is most certainly in line with the heterogeneous character of the Isenbrant Group.

Peter van den Brink

We are also grateful to Till-Holger Borchert, Director of Musea Brugge, Bruges, for independently endorsing the attribution after examining the painting at first hand.

*For footnotes please see the printed copy of the catalogue.