National Gallery of Art

In 1906 Abraham Bredius, director of the Mauritshuis in The Hague, traveled to Brussels to examine a collection of drawings owned by the family of Jonkheer Jan de Grez.[1]  [1]
This summary of the discovery of the painting and early reactions to it is taken from Ben Broos’ account in the entry on the painting in Ben Broos and Arthur K. Wheelock Jr., Johannes Vermeer (Washingon DC, and The Hague, 1995–1996), 207. See also Ben Broos, "Vermeer: Malice and Misconception," in Vermeer Studies, ed. Ivan Gaskell and Michiel Jonker (Washington, DC, 1998), 25. There he discovered, hanging high on a wall, a small picture that he surmised might be by Vermeer of Delft. Bredius asked for permission to take down the painting, which he exclaimed to be “very beautiful.” He then asked if the painting could be exhibited at the Mauritshuis, which occurred during the summer of 1907. Bredius’ discovery was received with great acclaim. In 1911, after the death of Jonkheer Jan de Grez, the family sold the painting, and it soon entered the distinguished collection of August Janssen in Amsterdam. After this collector’s death in 1918, the painting was acquired by the Amsterdam art dealer Jacques Goudstikker, and then by M. Knoedler & Co., New York, which subsequently sold it to Joseph E. Widener. On March 1, 1923, the Paris art dealer René Gimpel recorded the transaction in his diary, commenting: “It’s truly one of the master’s most beautiful works.”[2]  [2]
René Gimpel, Diary of an Art Dealer, 1918–1939, trans. John Rosenberg (New York, 1987; first published in 1963), 216.

Despite the enthusiastic reception that this painting received after its discovery in the first decade of the twentieth century, the attribution of this work has frequently been brought into question by later scholars.[3]  [3]
The attribution of this painting to Vermeer was first rejected by P. T. A. Swillens, Johannes Vermeer: Painter of Delft, 1632–1675 (Utrecht, 1950), 64–65. Albert Blankert, Rob Ruurs, and Willem L. van de Watering, Johannes Vermeer van Delft, 1632–1675 (Utrecht and Antwerp, 1975), 108–110, 168, considered the work to be a nineteenth-century imitation. Blankert restated this view in Albert Blankert, Museum Bredius: Catalogus van de schilderijen en tekeningen (The Hague, 1978), 172, and again in Gilles Aillaud, Albert Blankert, and John Michael Montias, Vermeer (Paris, 1986), 200–201. A similar opinion is held by Yvonne Brentjens, “Twee meisjes van Vermeer in Washington,” Tableau 7 (February 1985): 54–58. Arthur K. Wheelock, Jr., review of Albert Blankert, Rob Ruurs, and Willem L. van de Watering, Johannes Vermeer van Delft, 1632–1675 (Utrecht and Antwerp, 1975), Art Bulletin 59 (September 1977): 439–441, argued for the seventeenth-century origin of the painting, placing the work in the circle of Vermeer. He expanded on this theory in Arthur K. Wheelock Jr., “Zur Technik zweier Bilder, die Vermeer zugeschrieben sind,” Maltechnik-Restauro 84 (1978): 242–257, and in Arthur K. Wheelock Jr., Jan Vermeer (New York, 1981), 156. John Michael Montias, Vermeer and His Milieu: A Web of Social History (Princeton, 1989), 265, note 2, proposed that “the painting was begun by Vermeer and finished after his death by an inferior painter, perhaps by Jan Coelenbier, who bought paintings from Vermeer’s widow soon after his death.” Walter Liedtke in Ben P. J. Broos et al., Great Dutch Paintings from America (The Hague and Zwolle, 1990), 43, on the other hand, defends the attribution to Vermeer. For his more recent support for the attribution to Vermeer, see Walter A. Liedtke, Vermeer: The Complete Paintings (Ghent, 2008), no. 25, 140–143. In the 1995 exhibition catalog Johannes Vermeer, organized by the National Gallery of Art, Washington, and the Mauritshuis, The Hague, the attribution of this painting as “Circle of Vermeer” reflected the divergent opinions of the National Gallery of Art and the Mauritshuis. Benjamin Binstock, Vermeer’s Family Secrets: Genius, Discovery, and the Unknown Apprentice (New York and London, 2009), 249–257, considers this painting and Girl with a Red Hat to be self-portraits painted by Vermeer’s daughter, Maria Vermeer. Partially because of their wood supports and similarly small scale, and partially because of subject matter, Girl with a Flute and Girl with the Red Hat  [fig. 1]   [fig. 1] Johannes Vermeer, Girl with the Red Hat, c. 1665/1666, oil on panel, National Gallery of Art, Washington, Andrew W. Mellon Collection, 1937.1.53 have frequently been cited as companion pieces and accepted or rejected together. They may even have been considered companion pieces in the Dissius sale in Amsterdam in 1696.[4]  [4]
The Girl with a Flute measures 20 by 17.8 centimeters; the Girl with the Red Hat  measures 23.2 by 18.1 centimeters. In John Michael Montias, Vermeer and His Milieu: A Web of Social History (Princeton, 1989), 363–364, doc. 339, items 38, 39, and 40 are described, respectively, as “a tronie in antique dress, uncommonly artful”; “Another ditto Vermeer”; and “A pendant of the same.” The unusual costumes in the Girl with the Red Hat and the Girl with a Flute may well have been seen as depicting “antique dress” by the compiler of the catalog. Slight differences in the size of the panels, in the compositional arrangement of the figures, and in the quality of execution have led me to argue in previous publications that the paintings are not companion pieces and that the attribution of the Girl with a Flute to Vermeer could not be maintained.[5]  [5]
See Arthur K. Wheelock Jr., “Zur Technik zweier Bilder, die Vermeer zugeschrieben sind,” Maltechnik-Restauro 84 (1978): 242–257, and Arthur K. Wheelock Jr., Jan Vermeer (New York, 1981), 156, where the painting was designated “Circle of Vermeer.” Subsequently, I have concluded that removing the Girl with a Flute from Vermeer’s oeuvre was too extreme given the complex conservation issues surrounding this image.

A number of factors point to seventeenth-century origins for the Girl with a Flute and, indeed, relate the work intimately with Vermeer’s other paintings. Technically, examination of the panel using DendrochronologyDendrochronology
 A method of dating wood by examining the annual growth rings. has determined a felling date in the early 1650s.[6]  [6]
See Technical Summary for Girl with a Flute. A paint sample taken from a yellow highlight on the girl’s left sleeve, moreover, indicates the use of seventeenth-century pigments characteristic of Vermeer’s paintings: natural ultramarine, azurite, and lead-tin yellow.[7]  [7]
Hermann Kühn, “A Study of the Pigments and the Grounds Used by Jan Vermeer,” in National Gallery of Art Report and Studies in the History of Art, vol. 2 (Washington, DC, 1968), 194. These pigments were prevalent in the seventeenth century but not at later dates. Natural ultramarine, one of Vermeer’s favorite pigments, is produced from the semiprecious stone lapis lazuli. It was an expensive pigment, prized as much for its intrinsic value as for the luminosity of its blue hue. Around 1830 an artificial means of producing ultramarine was invented in France, which soon supplanted the more expensive natural ultramarine in artists’ palettes. Azurite never disappeared as completely as did natural ultramarine from artists’ palettes, but it is infrequently found after the seventeenth century. Lead-tin yellow, another pigment frequently found in Vermeer’s paintings, was replaced by Naples yellow after the seventeenth century. It seems to have been unknown from the mid-eighteenth century until it was rediscovered in 1940. Stylistically, the jacket worn by the girl is comparable to jackets seen in other works from the late 1650s to the mid-1660s, for example,  Woman Holding a Balance in the National Gallery of Art andThe Concert in the Isabella Stewart Gardner Museum, Boston. Other artists, particularly Gerard ter Borch the Younger (Dutch, 1617 - 1681), Gabriel Metsu (Dutch, 1629 - 1667), and Frans van Mieris (Dutch, 1635 - 1681), also depict women in similar costumes.

One unusual aspect of the girl’s wardrobe is the hat she wears. No exact equivalent exists in any other painting of the period, although similar wide-brimmed hats are found frequently in Dutch prints and drawings of working-class women.[8]  [8]
A. M. Louise E. Mulder-Erkelens, keeper of textiles, Rijksmuseum, Amsterdam, has suggested (letter of May 7, 1974, to A. B. de Vries, copy in NGA curatorial files) that the hat may have been intended to suggest some “archaic or exotic characteristics.” She related it to hats seen on gypsies and shepherdesses in works by Abraham Bloemaert (1566–1651) and Karel van Mander (1548–1606). She also noted that artists often kept unusual headgear in their studios that could assist in giving chiaroscuro effects to the model’s face. See Sturla J. Gudlaugsson, Ikonographische Studien über die holländische Malerei und das Theater des 17. Jahrhunderts (Berlin, 1938), 21. Similar wide-brimmed hats are found frequently in works by Rembrandt and his school. See Julius S. Held, Rembrandt’s “Aristotle” and Other Rembrandt Studies (Princeton, 1969), 11–12. This hat, however, has an oriental character that may relate to a vogue for oriental dress apparent in the latter decades of the seventeenth century.[9]  [9]
See Seymour Slive, “A Family Portrait by Nicolaes Maes,” Fogg Art Museum Annual Report (1957–1958): 32–39. Chinese hats were generally constructed of woven bamboo. This one appears to have been modified by the addition of a gray, white, and black fabric covering, presumably to enhance its appearance.[10]  [10]
Thomas Lawton, formerly assistant director, Freer Gallery of Art, Washington, DC, has been most helpful in analyzing the nature of this hat. Indeed, this strange hat actually reinforces the argument that the origins of this painting are seventeenth-century. It would be extremely unlikely for an artist of a later period to include such a hat in a painting that purported to be a Vermeer.

The Girl with a Flute and the Girl with the Red Hat are so close in concept that one must assume that they were conceived at approximately the same time, most likely in the mid-to-late 1660s. In each painting the young woman looks toward the viewer with an expectant expression, her eyes alert, her mouth half open. Each wears an exotic hat, sits in a chair with lion finials, and leans on one arm. Behind each of them hangs a tapestry of which only a fragment is visible. In each picture, light entering from the sitter’s left, an unusual feature in Vermeer paintings, strikes the girl’s left cheek, nose, and chin.

The manner in which optical effects of color are exploited in the two works is also comparable and characteristic of Vermeer. In each painting the face is shaded with a thin green glaze pulled over the flesh tones, a technique Vermeer developed more extensively in his later works. Colored highlights are a distinctive characteristic of Vermeer’s style, and in the Girl with a Flute the mouth is accented with a turquoise green highlight in a manner comparable to the pink highlight Vermeer applied to the mouth of the Girl with the Red Hat. The actual color of the highlight is similar to the green accent in the eye of the Girl with the Red Hat. Finally, the sunlit blue jackets worn by the two girls are similarly animated by numerous yellow highlights.

Despite the many stylistic and technical similarities between these paintings, significant differences in quality do exist. To begin with, the Girl with a Flute is a less successful composition. Whereas the pose of the girl in Girl with the Red Hat, as she turns and rests her arm over the back of her chair, subtly integrates suggestions of movement and stability, the frontal pose of the girl in Girl with a Flute is relatively flat and immobile. Her hat, left shoulder, and right hand are awkwardly cut by the edge of the panel.[11]  [11]
There is no indication that the panel has been trimmed, as was first suggested by Wilhelm Martin, “La Jeune Fille à la Flute de Vermeer de Delft,” L'Arte Flamand e Hollandais 8 (July 1907): 20–23, and Wilhelm Martin, “Jan Vermeer van Delft. Het Meisje met de Fluit,” Onze Kunst 12 (1907): 20, who thought the painting to be a fragment. Not only is the back of the panel beveled along all four edges, but also the paint along the edges does not appear fractured in a way that would suggest it had been trimmed. The flute she holds, which is actually a recorder, is curiously undefined and seems inaccurately rendered.[12]  [12]
I am most grateful to Helen Hollis, formerly of the Division of Musical Instruments, Smithsonian Institution, Washington, DC, for her observations on the nature of musical instruments in Vermeer’s oeuvre and on the specific character of the “flute” in this painting. Although its fipple mouthpiece is correctly indicated by the double highlight, the air hole below the mouthpiece is placed off-line. As seen in the recorder hanging on the wall in a painting by Judith Leyster (Dutch, 1609 - 1660), it should lie on an axis with the upper lip of the mouthpiece (Nationalmuseum, Stockholm, inv. no. NM 1126). The finger holes seen below the girl’s hand are turned even farther off this axis, although such a placement would be allowable if the recorder were composed of two sections.

The handling of the paint in the Girl with a Flute is also less assured than in the Girl with the Red Hat. In particular, the integration of tones and color in the Girl with a Flute lacks the cohesiveness characteristic of Vermeer. Flesh tones in the girl’s face are not modulated with the same degree of refinement. Transitions between the shadow of the eye and the sunlit cheek, between the shaded and unshaded portions of the chin, and the areas between the nose and mouth are rather abrupt.[13]  [13]
These abrupt transitions between areas are accentuated in the X-radiograph of the Girl with a Flute   . The thumbnail of the girl’s ill-proportioned hand is indicated by a uniformly dense paint whereas during the mid-1660s Vermeer generally would accent only a portion of a nail with a light highlight. Finally, the uniformly thin necklace has none of the modulations of accent and tone that Vermeer delighted in rendering.

Comparisons of the lion finials in the Girl with a Flute [fig. 2]   [fig. 2] Detail of lion-head finial, Attributed to Johannes Vermeer, Girl with a Flute, probably 1665/1675, oil on panel, National Gallery of Art, Washington, Widener Collection, 1942.9.98 and the Girl with the Red Hat [fig. 3]   [fig. 3] Detail of lion-head finial, Johannes Vermeer, Girl with the Red Hat, c. 1665/1666, oil on panel, National Gallery of Art, Washington, Andrew W. Mellon Collection, 1937.1.53 also point out the relatively unrefined brushwork of the former. The lion finial in the Girl with the Red Hat is modeled wet-into-wet by subtle variations in the weight and thickness of the strokes, whereas the finial in the Girl with a Flute does not have the same degree of articulation. The essential vocabulary of thin diffused strokes superimposed by opaque highlights is the same, but the lines necessary to create a sense of volume and form are less successfully integrated.[14]  [14]
Microscopic examination of the chair finial reveals that the surface is filled with small particles of foreign matter embedded in the paint. This foreign matter, whether it be dust, brush hairs, or wood splinters, is found throughout the paint. In only one other work by Vermeer have I noted similar foreign matter embedded in the paint: The Guitar Player (Iveagh Bequest, Kenwood, London, datable about 1672).

Finally, although in both instances the girls’ blue jackets are animated with diffused yellow highlights, in the Girl with the Red Hat the diffused highlights are grouped with a certain optical logic lacking in this work. To heighten the blue color on the shoulder of the Girl with the Red Hat, for example, Vermeer first highlighted the area with light blue strokes and then superimposed a sequence of yellow strokes over the blues. He painted the ridges of the highlighted folds with opaque yellow strokes. The jacket of the girl in Girl with a Flute is painted in a similar technique, but surety of the execution is lacking.

Despite such distinctions in quality it seems unadvisable to remove Girl with a Flute from Vermeer’s oeuvre, since painting techniques are similar to those in the Girl with a Red Hat and other Vermeer’s paintings. The soft modeling of the yellow highlights on girl’s blue jacket, for example, is similar to the character of the blue and yellow edging on the yellow material that hangs from the turban in The Girl with a Pearl Earring from the mid-1660s (Mauritshuis, The Hague).[15]  [15]
See inventory no. 670, c. 1665, from the Mauritshuis, The Hague. By the end of the 1660s, moreover, Vermeer began to create more abrupt transitions in his modeling that are not unrelated to the way in which the face in this painting is handled.

One complicating factor in trying to come to a determination about the attribution of this painting is that its surface is not in good condition (see Technical Summary). More important, however, the X-radiograph [fig. 4]   [fig. 4] X-radiograph composite, Attributed to Johannes Vermeer, Girl with a Flute, probably 1665/1675, oil on panel, National Gallery of Art, Washington, Widener Collection, 1942.9.98 [see X-radiographyX-radiography
 A photographic or digital image analysis method that visually records an object's ability to absorb or transmit x-rays. The differential absorption pattern is useful for examining an object's internal structure as well as for comparing the variation in pigment types.] and the infrared reflectogram [fig. 5]   [fig. 5] Infrared reflectogram, Attributed to Johannes Vermeer, Girl with a Flute, probably 1665/1675, oil on panel, National Gallery of Art, Washington, Widener Collection, 1942.9.98 [see Infrared ReflectographyInfrared Reflectography
 A photographic or digital image analysis method which captures the absorption/emission characteristics of reflected infrared radiation. The absorption of infrared wavelengths varies for different pigments, so the resultant image can help distinguish the pigments that have been used in the painting or underdrawing.] and andindicate that the composition was reworked in the seventeenth century. The patterns of the collar folds on both shoulders were altered, the size of the left cuff reduced, and the contour of the right arm changed. Other alterations include the addition of the fur trim on the front of the jacket and a reduction in the size of the hat. Finally, the girl’s finger that rests on the recorder was also apparently added, a change that may have occurred when the position of the flute was moved to the left. This change is evident in the reflectogram [fig. 5]   [fig. 5] Infrared reflectogram, Attributed to Johannes Vermeer, Girl with a Flute, probably 1665/1675, oil on panel, National Gallery of Art, Washington, Widener Collection, 1942.9.98.

It appears that when the painting was reworked, the initial composition was still largely at the blocking-in stage.[16]  [16]
The thinness of the execution on the figure’s proper right shoulder and arm is probably indicative of the level to which the painting was initially brought. The reworkings slightly altered the woman’s pose by lowering her left shoulder and adjusting the position of the cuff. As a consequence, she no longer appears to lean to such a degree on her left arm. Although the reasons for the reworking of this painting are not known, they suggest that the painting was not brought to completion and that damages subsequently occurred to the original design layer.[17]  [17]
It is unclear whether these damages occurred before or after the painting was reworked. The x-radiograph reveals that quite defined losses exist under the white collar on the girl’s left shoulder [fig. 4]   [fig. 4] X-radiograph composite, Attributed to Johannes Vermeer, Girl with a Flute, probably 1665/1675, oil on panel, National Gallery of Art, Washington, Widener Collection, 1942.9.98. Other losses exist below her left eye, between her nose and mouth, and on her cuffs and right hand. Just why these losses occurred is not known. Perhaps some inherent problem of adhesion existed between the paint layers and the ground. This latter explanation might account for some of the peculiar alligator crackle pattern that occurs in the paint on the woman’s cuff and in the thin blues of her jacket.

Technical evidence indicates that the alterations were made soon after Vermeer blocked in the initial composition since paint characteristics on the surface reflect those of the underlying layer. It is conceivable and, indeed, probable that the alterations were made by someone other than Vermeer, perhaps to prepare the work for sale after his death, even though the artist is not known to have had students or other close followers. Montias suggests that the revisions were the work of Jan Coelenbier (1600 or 1610-1677), who purchased paintings in 1676 from Catharina Bolnes soon after Vermeer’s death.[18]  [18]
John Michael Montias, Vermeer and His Milieu: A Web of Social History (Princeton, 1989), 338, doc. 362. As these paintings were to be auctioned the following year, Coelenbier may have tried to bring the work to completion to secure a higher price. Whether Montias’ hypothesis proves true, the artist who reworked Girl with a Flute certainly knew Vermeer’s paintings from the late 1660s and early 1670s, for he incorporated a number of stylistic features from this period of the artist’s career.

The complex issues surrounding the attribution of this little painting can be summarized as follows: the general character, appearance, and even painting techniques found in the Girl with a Flute relate closely to Vermeer’s work, specifically to the Girl with the Red Hat, but the quality of execution does not appear to be of the same high level expected from this master. While it seems that Vermeer initially blocked in the painting, which was executed in the mid-1660s, the image was extensively revised at a somewhat later date, probably by another hand. The unsatisfactory condition of the painting, largely as a result of AbrasionAbrasion
 A gradual loss of material on the surface. It can be caused by rubbing, wearing, or scraping against itself or another material. It may be a deteriorative process that occurs over time as a result of weathering or handling or it may be due to a deliberate attempt to smooth the material., is not only detrimental to the appearance of the image but also complicates any assessment of the work’s attribution. It seems appropriate to indicate the uncertainty surrounding the work’s attribution by designating it “Attributed to Vermeer.”

Arthur K. Wheelock Jr.

April 24, 2014