Portrait of a Noblewoman, Probably Isabella of Portugal (1397–1472)

The Portrait: The identities of the painter and sitter have long been debated, and while the painter remains uncertain, the noblewoman is now generally considered to be Isabella of Portugal (1397–1472), third wife of Philip the Good, Duke of Burgundy. As the daughter of King John I of Portugal, she was an attractive marriage candidate, but she was still unmarried at the age of thirty-one when she received an official offer in 1429 from Philip, who had sent a delegation of his ambassadors to Portugal (see Lemaire and Henry 1991, pp. 28–31). Among the party was Jan van Eyck, who was commissioned by the duke to paint the likeness of the Portuguese princess and send it back to Burgundy.[1] Unfortunately, only a seventeenth-century drawing survives (see fig. 1 above). In this copy, Isabella wears a dress with a v-shaped neckline, cinched at the waist, and a transparent inset beneath her bodice. This overall appearance is repeated in other portraits of the duchess, including The Met's painting. Most examples also show the divided hennin commonly worn at the Burgundian court. More than ten versions of portraits of Isabella are known, dispersed over different time periods. Two of them are dated relatively securely to her lifetime: the lost Van Eyck portrait, and a painting by the workshop of Rogier van der Weyden, now in the J. Paul Getty Museum, Los Angeles (fig. 2). The Met’s portrait is closely related to the latter work. Two other paintings are most likely later copies after The Met's panel: one is in the Musée Hospice Comtesse, Lille; the other was auctioned at Christie’s, London, in 2004.

Based on an inscription on what he describes as the original frame (see Notes), Herman Nasse (1911) identified the noblewoman as Michelle de France, Philip’s first wife. In a letter to Duveen, Max J. Friedländer (1927) rightfully questioned the basis for this identification, since no one other than Nasse had seen the "original" frame, and the known appearance of Michelle de France does not fit with that of the unknown noblewoman. Instead, Friedländer dated the portrait to 1420 and proposed Margaret of Bavaria, wife of John the Fearless and mother of Philip the Good, as the sitter, based on a later version in Semur-en-Auxois that bears this inscription. This is in contrast to his article from about thirty years earlier (see Friedländer 1900), in which he called The Met's painting an insignificant copy from the end of the fifteenth century. Paul Wescher (1941) was the first to identify the noblewoman as Isabella of Portugal, because of the portrait’s resemblance to the Getty painting. That portrait is usually dated about 1450, which would place it in Isabella’s lifetime. It may in fact have been the same portrait that was described in one of the inventories of Margaret of Austria, great-granddaughter of Isabella, since the clothing of the Getty portrait matches the description.[2] In her première chambre Margaret had thirty-one portraits of members of the House of Habsburg, the House of Burgundy, as well as members of the Spanish and English royal families.[3] A number of these portraits, including those of Philip and Isabella, were circulated widely and copied into the sixteenth century.

The Hidden Virgin and Child: Even without technical examination, there is clearly something beneath the surface of The Met’s portrait—most notably in the area of the noblewoman’s face, where diagonal strokes are visible across the cheek and neck. An unfinished painting of a Virgin and Child, hidden beneath the portrait, was first made visible through an infrared reflectogram published in 1996, when Lorne Campbell ascribed it to a follower of Robert Campin. Since then a new infrared reflectogram has been made, as well as an x-radiograph and elemental mapping by macro x-ray fluorescence (see Technical Notes and figs. 3–8). These greatly improved the legibility of the earlier composition, and in turn, our understanding of the painting. The new infrared reflectogram now clearly shows that the Virgin is looking at Christ, who sits on her lap. Her right hand is extended to the side, possibly holding some object. Christ turns toward his mother, reaching with his hand to her face. Given the rocks in the background, the faint contours of a farmhouse on the Virgin’s left side, and the presence of green pigments found in the bottom of the picture (see Technical Notes), it appears that the Virgin and Child were depicted in a landscape. Possibly the scene was meant to be a variation on the Rest on the Flight into Egypt.

Attribution and Date: Over the years, the portrait has been attributed to many different artists, from Netherlandish painters such as Henri Bellechose and Daniel Daret, to a French and even a Portuguese artist working at the Burgundian court (see References). All authors, however, seem to agree on a fifteenth-century date. Yet, dendrochronological analysis has indicated that the earliest date the painting could have been created is 1496, with a more plausible date of 1502 onwards (see Technical Notes). However, the confusion is understandable; the identification of Isabella of Portugal makes an early dating more logical and even the scene with the Virgin and Child evokes fifteenth-century compositions like those by Robert Campin. Additionally, the poor condition and the complicated buildup of paint layers (see Technical Notes) present a painting that is particularly hard to read, especially without support from technical examination. A more probable explanation would be that the painting was made by an early sixteenth-century copyist, as is supported by the dating of the wood and handling of the paint (see Technical Notes). Given the close-up manner and static pose in which the unfinished Virgin and Child are depicted, the artist could have been inspired by paintings of Robert Campin and his followers, like the Madonna of Humility with a Crescent Moon (J. Paul Getty Museum, Los Angeles; fig. 9), the Madonna of the Grassy Nook (Staatliche Museen zu Berlin), or the Virgin of the Immaculate Conception (Musée de la Chartreuse, Douai). The landscape, however, is more reminiscent of sixteenth-century compositions, especially the fantastical rocks in the background. Possibly the artist meant for the scene to refer to Campinesque motifs, but updated it using a sixteenth-century landscape. For an unknown reason, this painting was abandoned and at a later moment covered by the portrait of a noblewoman that we see today.

Joyce Klein Koerkamp 2018

[1] There has been some debate about the interpretation of the archival documents and whether or not Van Eyck painted two portraits and sent them to Burgundy separately, one by land and one by sea. So far, most scholars have assumed there were two portraits that are now lost, but Stephan Kemperdick disputes this claim and thinks there was only one portrait. See Stephan Kemperdick, "Early Texts on Some Portraits by Jan van Eyck," Van Eyck Studies: Papers Presented at the Eighteenth Symposium for the Study of Underdrawing and Technology in Painting, Brussels, 19–21 September 2012, Paris, 2017, pp. 310–24.
[2] As suggested by Lorne Campbell and Yvonne Szafran, "The portrait of Isabella of Portugal, Duchess of Burgundy, in the J. Paul Getty Museum," Burlington Magazine 146 (March 2004), pp. 148–57.
[3] See Dagmar Eichberger, Leben mit Kunst, Wirken durch Kunst; Sammelwessen und Hofkunst unter Margarete von Österreich, Regentin der Niederlande, Turnhout 2002, pp. 153–66.

Support: The support is a single oak plank, with the grain oriented vertically. Dendrochronological analysis indicated an earliest possible creation date of 1496, with a more plausible date of 1502 upward.[1] The wood originated in the Baltic/Polish region. It has been thinned to 3/16 inch (4 mm) and cradled. The presence of unpainted margins and traces of a barbe along the left, top, and bottom edges indicate that the original dimensions are preserved there. The original edge is not present on the right side, yet the close cropping of the composition on the other sides suggests that little, if any, of the original painted surface has been trimmed there. A repaired vertical split runs the length of the panel, 2 1/2 inches (6.4 cm) from the left edge. Strips of wood have been added to the sides, measuring 3/16 inch (4 mm) on the left and 1/8 inch (3 mm) on the right.

Preparation: The panel was prepared with an off-white ground. Examination with infrared reflectography revealed that the portrait was underdrawn with a dry material (see fig. 3 above).[2] This underdrawing set out the contours of the sitter’s face, her headdress and robe. There is a stiff quality to the underdrawing that may be indicative of an artist copying the composition from a model. Infrared reflectography also helped visualize the earlier composition of the Virgin and Child, in conjunction with the x-radiograph and elemental mapping with x-ray fluorescence, which will be discussed in greater detail below.

Paint Layers: An earlier composition depicting a Virgin and Child in a rocky landscape lies beneath the portrait of the noblewoman. The earlier composition can be visualized using a combination of analytical techniques, principally infrared reflectography and elemental mapping by macro x-ray fluorescence (macro-XRF).[3] The infrared reflectogram shows the distribution of a carbon-containing black paint, applied loosely with a broad brush (fig. 3). This paint was used to establish contour lines and to begin shading some passages, for example, in the Virgin’s proper right forearm and along the rocky outcroppings at the left. However, other parts of the composition were only roughly sketched with this black paint: see the abbreviated strokes of the Child’s limbs and the Virgin’s right hand. Macro-XRF revealed that the artist also blocked in some areas with color. The elemental distribution maps show the presence of an iron-containing pigment (fig. 5) and a lead-containing pigment in the Virgin’s robe (fig. 6), clearly not associated with the noblewoman’s portrait. Macro-XRF also registered a mercury-containing pigment in this area (fig. 7), some of which is likely associated with the noblewoman’s dress but also extends to the left of the dress, corresponding with the Virgin’s robe. The presence of these elements suggests that a mixture of vermilion, lead white and brown earth were used to paint the Virgin’s red robe. Some bright red can be glimpsed beneath the present composition, particularly where the blue paint of the background at left has a pronounced craquelure. The lead distribution map also registered a small amount of lead in the fleshtones of the Virgin and Child, but it is difficult to distinguish how much lead is associated with the earlier composition as opposed to the portrait of the noblewoman, which contains a large amount of lead white throughout. Macro-XRF also revealed a swathe of a copper-containing material near the bottom of the composition (fig. 8), closely following the lower contour of the Virgin’s drapery. Green paint can be observed beneath the noblewoman’s brocade dress there; this green could be a grassy foreground in the earlier composition. No other elements of a notable amount were found with macro-XRF.

Technical examination combined with analytical imaging pointed to the most likely scenario being one in which a black or grey paint was first used to sketch out the composition of the Virgin and Child and to introduce some initial grisaille shading, and then a few select colors were laid in, including the red of the Virgin’s drapery, the green in the foreground, and perhaps some whites or fleshtones in the figures. The varying level of finish seen in the infrared combined with the limited amount of color present strongly suggests that the composition was not brought beyond the preliminary stages of painting.

No intermediary paint layers were detected between this earlier composition and the portrait of the noblewoman. Despite the apparent absence of an intermediary layer, the portrait was underdrawn, as described above. The presence of an underdrawing without an intermediary layer to conceal the earlier composition offers further evidence that the Virgin and Child was abandoned in the preliminary stages: technically it would be very difficult to produce a legible underdrawing atop a fully worked-up vibrant oil painting, but it is conceivable that an artist—and perhaps one who did not have the time to wait for an intermediary layer to dry—may have drawn directly on top of a thinly painted sketch.

Comparison of the underdrawing to the painted surface revealed that a few changes were made when the underdrawing of the noblewoman was worked up with paint. The contours of the hennin were slightly shifted, the woman’s chin--rather undersized in the underdrawing--was enlarged, and minor adjustments were made to the folds of the brocade dress. The neckline was painted slightly lower and with a dashed line, but the solid underdrawn line is now slightly evident through the aged oil paint.

The presence of the earlier composition just beneath the surface compromises the legibility of the portrait. The increased translucency of the aged oil paint, in addition to some abrasion of the uppermost layers of paint, renders the underlying composition faintly apparent. Furthermore, the colors used in the earlier composition are evident in spots due to extensive cracking of the upper paint layers, some of which even follow the forms of the underlying composition. This phenomenon is most disturbing in the diagonal curves that run through the noblewoman’s jaw and upper neck.

The paint is handled quite thickly in the portrait, with slightly raised brushstrokes and opaque paint mixtures, at odds with the glazes typical of fifteenth-century technique, particularly in the fleshtones. While a dense paint application may have been necessary to conceal the initial composition, it is also characteristic of later painting technique from the beginning to middle of the sixteenth century. The portrait of Isabella of Portugal attributed to the workshop of Rogier van der Weyden (J. Paul Getty Museum, Los Angeles; fig. 2) is a useful comparison when considering the technique.

While the heavy-handed technique may be attributed to the presence of the earlier portrait beneath, it is also clear that the portrait was painted by a less refined artist or copyist. The seemingly translucent white fabric that tops the woman’s divided hennin was rendered using an opaque light blue, which negates any translucent effect. Furthermore, the golden highlights of the brocade, painted using what appears to be lead-tin yellow, are a simplistic rendering of how light would have hit the folds. The pattern of the brocade does not follow the curves of the drapery, perhaps indicating that a stencil was used to transfer the pattern from a pattern book (fig. 10).

Sophie Scully 2018

[1] Wood identification and dendrochronological analysis completed by Dr. Peter Klein, Universität Hamburg, report dated May 1997. The report can be found in the files of the Department of Paintings Conservation. “The youngest heartwood ring was formed out in the year 1485. Regarding the sapwood statistic of Eastern Europe an earliest felling date can be derived for the year 1494, more plausible is a felling date between 1498..1500….1504 + x. With a minimum of 2 years for seasoning an earliest creation of the painting is possible from 1496 onwards. Under the assumption of a median of 15 sapwood rings and 2 years for seasoning, as probably usual in the 14th/15th century, a creation is plausible from 1502 onwards.”
[2] Infrared reflectography completed with a Merlin Indigo InGaAs near-infrared camera with a StingRay macro lens customized for the wavelengths covered by the camera, 0.9 to 1.7 microns, 2014.
[3] XRF mapping was carried out in 2014 by Silvia A. Centeno, Research Scientist in the Department of Scientific Research, and Giulia Olmeda, who at the time the painting was analyzed was a visiting graduate student from the Università degli Studi di Padova, in collaboration with Geert Van der Snickt and Koen Janssens, University of Antwerp. We are thankful to Joris Dik, Delft University of Technology, and Koen Janssens, for the loan of an XRF scanner prototype to The Met.

George Salting, London (by 1899–at least 1900, as Early Flemish, Michele de France); [Agnew, London, by 1908–at least 1909]; Friedrich Wilhelm, Freiherr von Bissing, Munich and The Hague (by 1910–26, as Burgundian, about 1420; his sale, Frederik Muller, Amsterdam, November 30, 1926, no. 6, as French, 15th century, Portrait of a Noblewoman, for 27,000 guilders to Duveen); [Duveen, London and New York, 1926–28; sold to Rosenfeld]; Ernst Rosenfeld, New York (1928–d. in or before 1937; his estate, 1937, as French, about 1440; sold to Knoedler); [Knoedler, New York, from 1937; sold to Ryan]; Clendenin J. Ryan, New York (until d. 1939; his estate sale, Parke-Bernet, New York, January 19–20, 1940, no. 219, for $14,000 to Knoedler and Zinser); [Knoedler and R. Zinser, New York, 1940; sold for $16,800 to Harkness]; Mrs. Edward S. (Mary Stillman) Harkness, New York (1940–d. 1950)

London. New Gallery. "Exhibition of Pictures by Masters of the Flemish and British Schools," 1899–1900, no. 34 (as Early Flemish School, Michele de France, lent by George Salting).

New York. M. Knoedler & Co. "Fifteenth Century Portraits," April 15–27, 1935, no. 3 (as "Portrait of a Noble Lady, presumed to be Michelle de France," by a Master of the Burgundian School, about 1420, lent by Ernst Rosenfeld).

[Max J.] Friedländer. "Die Leihausstellung der New Gallery in London, Januar–März 1900.—Hauptsächlich niederländische Gemälde des XV. und XVI. Jahrhunderts." Repertorium für Kunstwissenschaft 23 (1900), p. 251, refers to it as an insignificant copy from the end of the 15th century.

Hermann Nasse. "Gemälde aus der Sammlung des Univ.-Professors Dr. Freih. Fr. W. von Bissing zu München." Münchner Jahrbuch der bildenden Kunst 6 (1911), pp. 231–32, fig. 1 (inscription on frame) and fig. 16, records the inscription on the original frame, "Michelle de France . . . [see Notes]" and identifies the sitter as Michelle de France, wife of Philip the Good, Duke of Burgundy; considers it a Burgundian work from about 1420 or a later copy.

Hermann Nasse. "Un portrait de Michelle de France dans la collection du Bn de Bissing à Munich." Revue archéologique, 4th ser., 19 (May–June 1912), pp. 406–12, ill., is certain that the subject is Michelle de France, Duchess of Burgundy; based on the costume dates it to the 1st quarter of the 15th century and attributes it to the workshop of Henri Bellechose; states that the picture has been carefully cleaned and restored since Friedländer saw it in 1900.

Salomon Reinach. Répertoire de peintures du moyen age et de la renaissance (1280–1580). Vol. 4, Paris, 1918, p. 552, ill. (engraving) , as Burgundian school (after Jan van Eyck?).

Seymour de Ricci. "A Flemish Triptych for Melbourne, II." Burlington Magazine 40 (April 1922), p. 166, discusses versions of the portrait with different identifications of the sitter, maintaining, however, the identification as Michelle de France.

Max J. Friedländer. Letter to Mr. Lowengard [Duveen Bros.]. April 21, 1927, notes that a poor copy of this portrait, reproduced in O. Rubbrecht (L'origine du type familiar de la maison de Hapsbourg, 1910, fig. 9) bears the inscription "Margarata Alberti Dudis in Bavari . . . Philippi Boni mater," and concludes that this is a portrait of the wife of Jean sans Peur [or Margaret of Bavaria]; mentions another portrait of this princess on a tombstone at Dijon (Rubbrecht, fig. 8); attributes the MMA example to a painter occupied at the French Burgundian Court, about 1420; observes that although mostly Netherlanders were active at the court, the boundaries between French and Netherlandish art were not clearly drawn during this period.

Georges Hulin de Loo. Le Maître de Flemalle (typescript of notes for lectures delivered at the centre "Primitifs flamands," Brussels, in 1939). 1939 [information reported by Lorne Campbell, Ref. 1974], associates it with the work of Campin and tentatively suggests that the sitter might be Margaret of Burgundy, Countess of Hainaut and Holland, who died in 1441.

Public Sale. Park-Bernet, New York. January 19–20, 1940, p. 84, no. 219, ill. (photogravure), catalogues the picture.

Paul Wescher. "Das höfische Bildnis von Philip dem Guten bis Karl V." Pantheon 18 (1941), pp. 196, 202, ill., states that on the reverse of this panel the subject is referred to as Michelle de France, but identifies her in his article as Isabella of Portugal, the same sitter that appears in the portrait ascribed to Rogier van der Weyden in the Rockefeller collection (now J. Paul Getty Museum, Los Angeles, as Workshop of Rogier van der Weyden).

Reynaldo dos Santos. Letter to Theodore Rousseau. December 20, 1951, considers this a portrait of Isabella of Portugal before her marriage to Philip the Good of Burgundy, finding confirmation in her resemblance to the older woman in the Rockefeller [later Getty] portrait; states that if our portrait was indeed made before her marriage, then it must have been painted by a Portuguese artist; notes that more certain attribution is made difficult by the rarity of Portuguese paintings from the first part of the 15th century.

Erwin Panofsky. Early Netherlandish Painting: Its Origins and Character. Cambridge, Mass., 1953, vol. 1, p. 478 n. 4 to p. 293, calls this picture a free copy of the Rockefeller [later Getty] panel minus the inscription.

Reynaldo dos Santos. "Retratos do século XV." Belas Artes, 2nd ser., no. 5 (1953), pp. 34–35, fig. 1, sees a resemblance in features to the portrait, earlier in date, of D. João I in the Museu Nacional de Arte Antiga, Lisbon, suggesting that our portrait may represent his daughter, Isabella; comments on the similarity of the sitter here to that in the Rockefeller [later Getty] portrait.

Josephine L. Allen and Elizabeth E. Gardner. A Concise Catalogue of the European Paintings in The Metropolitan Museum of Art. New York, 1954, p. 78.

Helen Comstock. "The Connoisseur in America: Portrait of Isabella of Portugal, Duchess of Burgundy." Connoisseur 137 (March 1956), p. 71, tentatively identifies the sitter as Isabella of Portugal based on her resemblance to a 17th-century variant inscribed with her name in the Hammer sale (Heberle, Cologne, October 5–6, 1894, no. 151, ill.); rejects identification of either sitter with that of the subject represented in the Rogerian Portrait of a Lady in the Rockefeller Collection [now Getty Museum]; notes that Isabella appears with other members of the house of Burgundy in the central panel of a triptych, "The Feeding of the Five Thousand," by the Master of the Legend of Saint Catherine, in Melbourne, Australia [National Gallery of Victoria]; says our portrait may have been executed before Isabella left Portugal in 1430 and considers it a work of the Portuguese School, either a Northern or Flemish artist working in Portugal or a Portuguese student of some member of the school of Bruges.

Kurt Bauch. "Bildnisse des Jan van Eyck." Heidelberger Akademie der Wissenschaften (1961/62), p. 107, fig. 11, observes that the picture appears to represent Isabella of Portugal, and must have been made by one of her Netherlandish court painters whose other works remain unknown.

Maria Julieta Ruival. Unpublished manuscript. 1966, dates our picture about 1415 and suggests the painter was a Portuguese artist working in Flanders; identifies the subject here and in the Rockefeller [Getty] portrait as Isabella of Bavaria; notes her resemblance to the sitter in a portrait in the Louvre, Paris, whom she identifies as Isabella of Bavaria based on her similarity to the sitter in a portrait inscribed with her name in the Stoclet collection, Brussels.

Lorne Campbell. Unpublished notes. 1972, notes that several versions of this portrait exist, including one at the Musée des Beaux-Arts, Lille, "which forms part of a series of reliable portraits of the Counts and Countesses of Flanders, and which is inscribed with the name of Isabella of Portugal, Duchess of Burgundy"; believes there can be no doubt that the sitter here is Isabella; finds attribution of our panel to a Portuguese painter unreasonable, since, as the most repeated likeness of the Duchess, the original must surely have been the work of one of the Burgundian court painters in the Netherlands of "?ca. 1440," based on the costume; observes that the portrait "may be entirely repainted, except in restricted areas of the flesh".

Lorne Campbell. Unpublished catalogue entry. May 20, 1974, sees in this portrait an affinity with the works of Campin and Jacques Daret and suggests it is the work of Daniel Daret, who was appointed court painter in 1449; lists six other variants, noting that the inscriptions on them agree only in identifying the sitter as a Duchess of Burgundy; dates the costume from about 1450 and concludes that the sitter can only be Isabella of Portugal, who married Philip the Good in 1429, and who died in 1472; notes that later unscrupulous copyists, "when unable to secure authentic portrait of the personages whom they wished to depict, used available portraits of different personages".

Jeffrey Chipps Smith. "The Artistic Patronage of Philip the Good, Duke of Burgundy (1419–1467)." PhD diss., Columbia University, 1979, pp. 324–25 n. 101, fig. 302, states that the portrait is known in at least six versions, of which this is the finest, and that comparison with the Rockefeller or Dijon portraits confirms that the sitter is Isabella.

Lorne Campbell. Unpublished notes. 1981, states that infrared reflectography reveals that this portrait was painted over an unfinished picture of the Virgin and Child, "which was also in the style of the 'Master of Flémalle'"; believes the original, and also this replica, were by artists from his circle and were probably painted about 1440.

Claudine Lemaire and Michèle Henry. Isabelle de Portugal, duchesse de Bourgogne, 1397–1471. Exh. cat., Bibliothèque Royale Albert Ier. Brussels, 1991, p. 149, calls the Getty picture a presumed portrait of Isabella of Portugal, a late replica by an unknown artist based on a lost original by Rogier van der Weyden; considers the The Met's panel a mediocre interpretation of Rogier's original, although the coiffure, dress, and smile of the sitter are reminiscent of the sitter in the Getty work.

Katharine Baetjer. European Paintings in The Metropolitan Museum of Art by Artists Born Before 1865: A Summary Catalogue. New York, 1995, p. 249, ill. p. 248.

Lorne Campbell. "Campin's Portraits." Robert Campin: New Directions in Scholarship. Ed. Susan Foister and Susie Nash. [Turnhout, Belgium], 1996, p. 132, figs. 19, 20, relates it to Jacques Daret; believes the "Virgin and Child" underneath relates to a work of the Master of Flémalle and that this may indicate its origin in the workshop of one of the Master's followers.

Diane Wolfthal and Cathy Metzger. Los Angeles Museums. Brussels, 2014, pp. 202, 205 n. 23.

Inscribed on original frame: "Michelle de France. Fille de Charles VI Roy de France et d'Isabeau de Bavière. Mariée en juin 1409 à Philippe le Bon Duc de Bourgogne" (see Nasse 1912).