National Gallery of Art

As she leans over the gate of a wooden fence a young girl stares directly at the viewer. In her left hand is a broom. The fence appears to surround a well, whose dark, round form is visible in the foreground. The well is flanked by a large overturned bucket on the right and a dark object, perhaps a trough, on the left. While the girl’s form is lit strongly from the left, the dark background and even the area around the well remain relatively undefined and obscured in shadow.[1]  [1]
This entry is a revised version of the text that appeared in the Nationalmuseum catalog of Rembrandt och Hans Tid [Rembrandt and His Age] (Stockholm 1992), no. 83, and the symposium papers published thereafter (Arthur K. Wheelock Jr., “A Girl with a Broom: A Problem of Attribution,” in Görel Cavalli-Björkman, ed., Rembrandt and His Pupils [Stockholm, 1993], 142–155). I have benefited greatly from my many conversations with Susanna Pauli Griswold about the issues discussed in this entry. I would also like to thank Dennis Weller and Melanie Gifford for their helpful comments.

A Girl with a Broom, in large part because of the appealing features of the young girl and the genre-like character of the subject, has long been admired as one of Rembrandt’s most sensitive depictions of figures from his immediate environs. This attractive model has been identified repeatedly as a young servant girl in Rembrandt’s household, but her identity remains unknown.[2]  [2]
This identification was first proposed by Émile Michel, Rembrandt: Sa vie, son oeuvre et son temps, 2 vols. (New York, 1893), 1:75. It was reiterated by, among others, Otto Benesch, “The Rembrandt Paintings in the National Gallery of Art,” Art Quarterly 6 (Winter 1943): 26. Geertje Dirckx could not have served as the model; having been born around 1610, she would have been too old to be this sitter, who is probably about twelve to fifteen years old. Hendrickje Stoffels, who was born in 1626, and who entered Rembrandt’s household around 1647, would also have been too old. The model who posed for Girl with a Broom probably also posed for Girl at the Window, 1645 [fig. 1]   [fig. 1] Rembrandt van Rijn, Girl at a Window, 1645, oil on canvas, Dulwich Picture Gallery, London. Photo reproduced by permission of the Trustees of Dulwich Picture Gallery. The girls have comparable hairstyles; they have relatively broad faces with widely separated eyes and low, flat eyebrows; their noses, the tips of which have a slightly bulbous appearance, are similar; and finally, their broadly formed lips are virtually identical.[3]  [3]
Computer examinations of the physical characteristics of the heads in these two paintings have been undertaken at the National Gallery of Art. The results have reinforced the notion that the model was identical. I am particularly indebted to Ambrose Liao and Donna Mann for their enthusiastic research on this project.

Whether this work was meant as a portrait or as a genre scene has been a matter of some discussion. Should it have been possible to identify the girl, the painting would almost certainly be classified as a portrait because of the frontal pose and careful depiction of the features.[4]  [4]
See, for example, Rembrandt’s Titus at His Desk, 1655 (Museum Boijmans Van Beuningen, Rotterdam, inv. no. 512), which would probably be classified as a genre scene where the sitter not known. Nevertheless, the setting and accoutrements give the painting the character of a genre scene, albeit one that is not fully explained to the viewer. Why, for example, is the girl holding the broom while leaning over the wall surrounding the well, and does the prominently placed bucket have any iconographic significance?[5]  [5]
Susan Koslow, “Frans Hals’s Fisherboys: Exemplars of Idleness,” Art Bulletin 57 (September 1975): 429, has associated the crossed-arm pose of the girl with idleness. This interpretation, however, is not convincing. The type of well depicted appears to be similar to that in The Village Holiday by Daniel Teniers the Younger (1610–1690) (Virginia Museum of Fine Arts, Richmond, no. 56–23). In this painting a broom and a bucket stand adjacent to the well.

Recent scholars have doubted the attribution to Rembrandt and some have even speculated that the painting is eighteenth century in origin.[6]  [6]
Virtually all scholars since Abraham Bredius, Rembrandt: The Complete Edition of the Paintings, revised by Horst Gerson (London, 1969), 580, no. 378, have doubted the attribution to Rembrandt. Because A Girl with a Broom has a distinguished provenance that reaches back to 1678, when it is almost certainly listed in the inventory of the collection of an acquaintance of Rembrandt’s, Herman Becker, the latter suggestion is clearly unacceptable. Even though the painting was attributed to Rembrandt when it was in Becker’s collection, its style differs in enough fundamental ways from that of Rembrandt’s authentic paintings to warrant the doubts mentioned in the literature.

The primary reason that A Girl with a Broom has been associated with eighteenth-century images is its physical appearance. The surface is deformed in areas, particularly in the face and hands, by pronounced WrinklingWrinkling
 Small furrows and ridges in paint or fabric due to shrinkage, folding, or compression. of the paint similar to that found in certain English paintings of the eighteenth century [fig. 2]   [fig. 2] Raking light, Rembrandt Workshop (Possibly Carel Fabritius), A Girl with a Broom, probably begun 1646/1648 and completed 1651, oil on canvas, National Gallery of Art, Washington, Andrew W. Mellon Collection, 1937.1.74.[7]  [7]
Abraham Bredius, Rembrandt: The Complete Edition of the Paintings, revised by Horst Gerson (London, 1969), 580, no. 378 wrote: “The surface is composed of small particles of paint curling slightly at the edges, such as one observes on pictures which have been exposed to extraordinary heat or on pictures of the eighteenth century. The latter possibility, in the present state of Rembrandt research, should not be excluded.” The issue was further taken up by Hubert von Sonnenburg, “Maltechnische Gesichtspunkte zur Rembrandtforschung,” Maltechnik-Restauro 82 (1976): 12. Von Sonnenburg associated the “gerunzelte Farbschicht” with that found in eighteenth-century English paintings. This effect, he wrote, resulted from an excess of drying oil or from the character of the medium itself. He questioned whether the painting had been made by a follower of Rembrandt and called for a serious scientific analysis of the work. This effect had, until the painting’s conservation treatment in 1991–1992, been exacerbated by the thick layers of pigmented varnish. Technical analysis undertaken at the time of the treatment indicated that the wrinkling in the surface resulted from the interference of an underlying paint layer that had not sufficiently dried. The X-radiographs [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.] reveals that the girl’s face was painted over an earlier head looking upward to the right [fig. 3]   [fig. 3] X-radiograph composite, Rembrandt Workshop (Possibly Carel Fabritius), A Girl with a Broom, probably begun 1646/1648 and completed 1651, oil on canvas, National Gallery of Art, Washington, Andrew W. Mellon Collection, 1937.1.74. To judge from the X-radiograph, the lead white modeling around the nose and cheek of the underlying head is quite dense. Little or no wrinkling appears on the surface image covering these areas of the underlying image. The wrinkling on the surface is most pronounced where it overlaps transparent areas of the underlying images, such as eye sockets. It thus appears that these shaded areas were modeled in dark, medium-rich glazes that had not yet dried at the time the top layers were applied.[8]  [8]
I would particularly like to thank Karin Groen, who analyzed a group of samples taken from this painting and confirmed the assessment of the problem developed by the Scientific Research department at the National Gallery of Art (letter, December 4, 1992, in NGA curatorial files). She specifically noted that medium-rich paint (high oil content) can be observed in many of the layers. A dark brown underlayer, sandwiched between medium-rich layers, contains manganese, probably in the form of umber, which promotes a fine type of wrinkling. Layers near the surface contain cobalt, which promotes surface drying. Once the surface dries prior to the drying of the underlying layers, wrinkling of the paint occurs. She also noted the presence of vermilion near the proper right hand that belonged to the later change in the composition.

Although the existence of an earlier form beneath the girl’s head is fairly easy to distinguish in the X-radiographs, evidence of an underlying layer is more difficult to discern for the rest of the body. Nevertheless, an earlier shape for the blouse, blocked in with paints with little density, can be distinguished in various places.[9]  [9]
The X-radiographs [see X-radiography] measure only the relative density of metal-based paints, hence other components of the initial paint layer could exist that cannot be read with this examination procedure. More information could possibly be gained through examination with neutron autoradiography. The most obvious of these is along the outer contour of the girl’s right sleeve. An earlier layer, probably the same, can also be made out under the handle of the broom both in the X-radiographs and with the naked eye. Also visible through the brown color of the broom handle is the full extent of the girl’s thumb.[10]  [10]
The thumb is also visible in the X-radiographs. Since the girl’s hands have surface distortions much as those found in the head, underlying paint layers here must have had paint characteristics similar to those in the shaded portions of the earlier head.

Whatever the explanation for the unusual nature of the paint in the flesh areas, neither technical nor visual evidence provides an argument for removing A Girl with a Broom from the immediate orbit of Rembrandt.[11]  [11]
Although a comparable wrinkling effect is not found in the impastos of paintings by Rembrandt, similar problems do exist in the backgrounds of at least two of his works, Abduction of Proserpine (Gemäldegalerie, Berlin; Br. 463), and Alexander (City Art Gallery and Museum, Glasgow; Br. 480). Not only is the image appealing in subject matter, the modeling of the features is sensitively rendered, and the folds in the girl’s white blouse are executed with great bravura.

Notwithstanding the inherent qualities of A Girl with a Broom, a close comparison of it with two comparable paintings by Rembrandt—Girl at a Window, 1645, in Dulwich [fig. 1]   [fig. 1] Rembrandt van Rijn, Girl at a Window, 1645, oil on canvas, Dulwich Picture Gallery, London. Photo reproduced by permission of the Trustees of Dulwich Picture Gallery and Servant Girl at a Window, 1651, in Stockholm [fig. 4]   [fig. 4] Rembrandt van Rijn, Servant Girl at a Window, 1651, oil on canvas, Nationalmuseum, Stockholm—points out differences that clearly call into question the attribution to Rembrandt. The centrally placed figure remains isolated in the composition and does not activate the surrounding space as do the girls in the Dulwich and Stockholm paintings. Specifically, in comparison to the Dulwich painting, the modeling of the blouse in A Girl with a Broom is much freer, even in the folds of her right sleeve that are similar in character. Whereas in Girl at a Window Rembrandt created the illusion that the cloth actually rises and turns over upon itself, the folds in A Girl with a Broom have been formed with distinctive brushstrokes highlighting the uppermost ridges of the fabric. Nothing in the Dulwich painting is comparable to the extremely expressive brushwork in the left sleeve, where chiaroscuro effects are achieved by highlighting illuminated folds with slashing strokes of white impasto. Finally, while the blouse is more freely rendered here, the girl’s features are not modeled with the same degree of plasticity as they are in the Dulwich painting. There, Rembrandt boldly modeled the eyes, nose, and mouth with nuanced strokes that clearly convey the structure of the girl’s head. In the subject's face in A Girl with a Broom, as well as in her blouse, paint is more at the service of light than of structure. Accents effectively highlight the hair, forehead, nose, and upper lip, but they are not used to create underlying form. The difference in approach is most distinctly seen in the area of the right eye, where a general halftone shadow does little to suggest three-dimensional character. Instead, the eye’s structure, particularly the upper eyelid, is created with painted lines.

Significant stylistic differences also exist between A Girl with a Broom and Rembrandt’s Servant Girl at a Window [fig. 4]   [fig. 4] Rembrandt van Rijn, Servant Girl at a Window, 1651, oil on canvas, Nationalmuseum, Stockholm, even though the two works are dated the same year. The young woman represented in this latter painting is possibly, although not necessarily, the same; the pose, however, like that of the girl in the Dulwich painting, appears more natural and organic than in the Washington painting, where the girl’s head seems too large for her body. The subject's face in the Stockholm Servant Girl is more freely brushed than that in the Washington painting and modeling is achieved with quick and certain strokes. Accents of light help enliven her form, particularly around the eyes, in a way that is absent in A Girl with a Broom. The blouse, red jacket, and right hand of the servant girl in the Stockholm painting are also modeled with broad strokes that are quite consistent throughout and help create the painting’s harmonious effect. In the Washington painting, on the other hand, while the brushwork of the sleeves is bold and vigorous, that of the face and hands is relatively restrained, and that used to paint the broom is comparatively timid.

The contrasts in manner of execution between A Girl with a Broom and both of these related paintings are so intrinsic to an artistic approach that it seems improbable that A Girl with a Broom was executed by the same hand. The differences between the Washington and Dulwich paintings are such that it does not seem possible to account for them by differences of date, even if the Dulwich painting were executed in 1645 and the Gallery’s painting in 1651. It is even more improbable that Rembrandt would have created such different images as the Washington and Stockholm paintings in the same year. The signature and date of A Girl with a Broom, moreover, are certainly suspect. Although there is no evidence to suggest that they have been added at a later date, they are written in an uncharacteristic form, placed, as they are, around the circular inner edge of the well.[12]  [12]
The signature appears to be integral with the paint surface, and no varnish has been found between it and the underlying paint. Should there have been no date inscribed on the painting, the similarity in the age, hairstyle, and general appearance of the girl in the Washington and Dulwich paintings would have called for a projected date for A Girl with a Broom of 1646/1648, only a few years after the Dulwich Girl.[13]  [13]
It is a curious coincidence that the Stockholm Servant Girl at a Window is also dated 1651. Both paintings were in France in the eighteenth century, as was the Dulwich painting. One of these three paintings may have been the work described by Roger de Piles in the preface to his Cours de Peinture par Principes (Paris, 1708), 10–11, as quoted in Seymour Slive, Rembrandt and His Critics, 1630–1730 (The Hague, 1953), 129: “Rembrandt diverted himself one day by making a portrait of his servant in order to exhibit it at his window and deceive the eyes of the pedestrians. . . . While in Holland I was curious to see the portrait. I found it painted well and with great strength. I bought it and still exhibit it in an important position in my cabinet.” One possible explanation for the discrepancies of date and style, given the existence of an earlier image, is that the painting was begun in the late 1640s and only finished in 1651. This work, thus, may be one other example of a painting executed over an extended period of time (see, among the Rembrandt paintings in the Gallery’s collection: Saskia van Uylenburgh, the Wife of the Artist, The Apostle Paul, and The Descent from the Cross.

Few specifics are known about the nature of Rembrandt’s workshop in the late 1640s and early 1650s. Samuel van Hoogstraten (1627–1678), in his Inleyding tot de Hooge Schoole der Schilderkonst (Rotterdam, 1678), indicates that he was active in the master’s workshop before he returned to his native city of Dordrecht in April 1648. The fellow students he mentions were Carel Fabritius (Dutch, c. 1622 - 1654) and Abraham Furnerius (c. 1628–1654). Among other artists working with Rembrandt in the late 1640s were Karel van der Pluym (1625–1672), Constantijn van Renesse (Dutch, 1626 - 1680), and Nicolaes Maes (Dutch, 1634 - 1693). It seems probable that Willem Drost (Dutch, active c. 1650 - active 1655) and Abraham van Dijck (1635/1636–1672) also became Rembrandt pupils around 1650, although nothing certain is known about their relationship to Rembrandt. Indeed, many questions remain about paintings from Rembrandt’s workshop around 1650 (see, for example, Portrait of Rembrandt) because it is extremely difficult to establish the independent identities of Rembrandt’s pupils during these years. Nothing in the oeuvres of artists known or thought to have been working with Rembrandt in the early 1650s can be effectively compared either thematically or stylistically to this work. A more probable date, in terms of the manner of execution, appears to be the late 1640s.

Although no documentary proof has survived that clarifies the different roles of student and assistant in Rembrandt’s workshop during the 1640s, the more advanced of his students, for example Hoogstraten and Fabritius, would have worked as assistants in the workshop after they finished their apprenticeship.[14]  [14]
Fabritius seems to have studied with Rembrandt in the early 1640s before returning to Midden-Beemster in 1643. Virtually nothing is known about him during the late 1640s, but it seems unlikely that he remained in Midden-Beemster the entire time without continuing his contact with Rembrandt in Amsterdam. Midden-Beemster is only about thirty kilometers from Amsterdam and was a community that had many ties with Amsterdam. In 1648 or 1649 Fabritius painted the portrait of a wealthy Amsterdam silk merchant, Abraham de Potter (Rijksmuseum, Amsterdam, inv. no. A1591). By 1650 he had moved to Delft. For further information on Fabritius see Christopher Brown, Carel Fabritius (Oxford, 1981), and Frederik J. Duparc, Carel Fabritius, 1622–1654 (The Hague and Schwerin, 2004). In all likelihood they continued to help execute paintings that would be sold under Rembrandt’s name, even after they had begun working independently and signing their own works.[15]  [15]
In this respect their relationship to Rembrandt would have been much the same as that of Anthony van Dyck to Peter Paul Rubens during the late 1610s. In those years Van Dyck simultaneously painted in Rubens’ style when working in Rubens’ studio and in his own personal style when painting in his own workshop. Paintings created for Rembrandt’s workshop, to judge from those that have recently been attributed to these artists, would often be free adaptations of Rembrandt’s own compositions. These works, once accepted by the master as worthy of his production, would be inscribed with his signature and the date.

A Girl with a Broom fits into this scenario. It is one of a number of paintings loosely derived from Rembrandt’s Girl at a Window in Dulwich. Hoogstraten was particularly fond of this compositional type, if one is to judge from his late 1640s Young Man in a Hat, at a Half-Door in the Hermitage.[16]  [16]
Young Man in a Hat, at a Half-Door is not signed. It was first attributed to Hoogstraten by Werner Sumowski, Gemälde der Rembrandt-Schüler, 5 vols. (Landau i.d. Pfalz, 1983), 2:1339, no. 856. The painting was also cataloged as by Hoogstraten in Christopher Brown, Jan Kelch, and Pieter van Thiel, Rembrandt: The Master and His Workshop: Paintings (New Haven and London, 1991), 356, no. 74. The quality of this work, however, is comparatively mediocre, and it is impossible to reconcile the simplistic handling of paint seen here with that found in A Girl with a Broom. A much finer painting of a comparable type that has been attributed to Hoogstraten, Young Woman at an Open Half-Door, signed and dated Rembrandt 1645 [fig. 5]   [fig. 5] Rembrandt van Rijn (or follower), Young Woman at an Open Half-Door, 1645, oil on canvas, Mr. and Mrs. Martin A. Ryerson Collection, 1894.1022. Photo © 1994, The Art Institute of Chicago. All Rights Reserved, is also executed in a manner distinctively different from that of A Girl with a Broom.[17]  [17]
The painting was included in Christopher Brown, Jan Kelch, and Pieter van Thiel, Rembrandt: The Master and His Workshop: Paintings (New Haven and London, 1991), 350, no. 72, as by Hoogstraten. I would like to thank Martha Wolff at the Art Institute for her observations about the differences in technique between these two paintings and for sending me detailed photographs of the Chicago painting. In addition to the Chicago painting, another Rembrandt School painting from this period, depicting a young boy leaning against a metal railing, is in the Cincinnati Art Museum. See Mary Ann Scott, Dutch, Flemish, and German Paintings in the Cincinnati Art Museum (Cincinnati, 1987), 107–110, no. 38. As is evident in comparisons of the hands ([fig. 6]   [fig. 6] Detail of hands, Rembrandt Workshop (Possibly Carel Fabritius), A Girl with a Broom, probably begun 1646/1648 and completed 1651, oil on canvas, National Gallery of Art, Washington, Andrew W. Mellon Collection, 1937.1.74 and [fig. 7]   [fig. 7] Detail of left hand, Rembrandt van Rijn (or follower), Young Woman at an Open Half-Door, 1645, oil on canvas, Mr. and Mrs. Martin A. Ryerson Collection, 1894.1022. Photo © 1994, The Art Institute of Chicago. All Rights Reserved), the forms in the Chicago painting are executed in a far crisper manner, with flatter planes of color and fewer nuances of shading. Differences in character between the white sleeves of the girl in the Washington painting and the white shirt of the girl in the Chicago painting also point out that the Gallery's A Girl with a Broom was executed by an artistic personality that favored a freer, more painterly approach.

The artist in Rembrandt’s circle during this period who was most capable of both the nuanced modeling of the face and hands and the rough bravura brushwork found in the sleeves was Carel Fabritius, but specific comparisons with other works by him are difficult to make because few paintings can be firmly attributed to him during the mid-1640s. Thus only a tentative attribution to him is suggested.[18]  [18]
In 1993, at my suggestion, the attribution of this painting was changed from “Rembrandt van Rijn” to “Carel Fabritius and Rembrandt Workshop,” and the painting was exhibited as such in Stockholm (Rembrandt och Hans Tid [Rembrandt and His Age] [Stockholm, 1992], no. 83). The Fabritius attribution, however, was not generally accepted. A number of colleagues felt that insufficient comparative material existed to make a firm attribution. Walter Liedtke, “Stockholm: Rembrandt and His Age” (review of the exhibition Rembrandt och Hans Tid), The Burlington Magazine 124 (December 1992): 829–830, believes that the artist of the Chicago painting (fig. 5), which he attributes to Samuel van Hoogstraten, also executed A Girl with a Broom. Egbert Haverkamp-Begemann (personal communication, 1993) would prefer to leave the attribution of the Washington painting as “anonymous.” One of the few comparisons to Fabritius’ work that can be made is to his evocative Self-Portrait, c. 1645–1648 [fig. 8]   [fig. 8] Carel Fabritius, Self-Portrait, c. 1645–1648, oil on canvas, Museum Boijmans Van Beuningen, Rotterdam. Photo: Studio Tromp, Rotterdam. Although the modeling of the face of the girl in A Girl with a Broom is more nuanced than that of the Self-Portrait, where modeling is achieved with vigorously applied broken impastos, these differences may well relate to different artistic intents. The boldly uncompromising application of paint in the Self-Portrait was clearly intended to help characterize the artist’s personality, whereas the careful modeling in the girl’s face was appropriate to her sex and age. The character of the brushwork in the faces of these two paintings, indeed, is far more comparable than one might initially suspect. In both instances paint is densely applied with broad, interlocking brushstrokes that model facets or planes of the face. Similarly placed accents, moreover, help define the cheekbone and nose. A specific point of comparison is in the structure of the eyes: in each instance the upper portions of the relatively large, flat, almond-shaped eyes are defined by a black line rather than by modulations in tone. This particular manner of articulating eyes is not found in paintings by other artists in Rembrandt’s circle.

One other painting can be brought into this discussion, a Portrait of a Woman attributed to Fabritius by the Rembrandt Research Project (RRP).[19]  [19]
Stichting Foundation Rembrandt Research Project, A Corpus of Rembrandt Paintings, vol. 3, 1635–1642, ed. Josua Bruyn et al. (Dordrecht, Boston, and London, 1989), C107. The painting and its pendant (Br. 251), which are traditionally identified as portraits of Adriaentje Hollaer and her husband, the painter Hendrick Martensz Sorgh, are in the collection of the Duke and Duchess of Westminster. See also Abraham Bredius, Rembrandt: The Complete Edition of the Paintings, revised by Horst Gerson (London, 1969), 291, no. 370. Although this painting is signed and dated “Rembrandt.f/1647,” the RRP has concluded that it was executed by Fabritius around 1642. Whether or not such a redating is justified, and I would maintain that the date on the painting reflects the period of its execution, the attribution of this portrait to Fabritius is convincing. The differences in style between the carefully modeled head of this woman and Fabritius’ more broadly and roughly executed Rotterdam Self-Portrait, however, demonstrate the range of techniques Fabritius was capable of during these years. The girl's head in A Girl with a Broom falls somewhere between these two works. The hands of the woman in Fabritius' Portrait and those of the subject in A Girl with a Broom also show marked similarities. In both instances they are modeled with interlocking planes of color that are generally brushed across the forms, particularly the fingers, rather than along their length.[20]  [20]
For a detail photograph of the hands of the Portrait of a Woman see Stichting Foundation Rembrandt Research Project, A Corpus of Rembrandt Paintings, vol. 3, 1635–1642, ed. Josua Bruyn et al. (Dordrecht, Boston, and London, 1989), 677, fig. 4.

The hypothesis that A Girl with a Broom could have been created during the mid-to-late 1640s by Fabritius in response to Rembrandt’s Girl at a Window, however, needs to remain extremely tentative because of the 1651 date inscribed on the painting. Fabritius almost certainly would not have added the signature and date because he had moved to Delft in 1650. It is possible, however, that the image was reworked and brought to completion by another artist at this date. The basis for this hypothesis is the stylistic discrepancy that exists between the execution of the broom, the bucket, and even the fence surrounding the well, and that of the figure. Neither the broom nor the bucket is executed with the same surety as the figure itself. The tentative brushstrokes do not model the forms with bold accents comparable to those found on the girl’s blouse. The relationships of scale between the girl and these objects are also peculiarly discordant.

Technical evidence seems to support the hypothesis that the broom may have been worked up after the initial blocking in of the figure had occurred. As has been mentioned, an earlier form of the blouse and the girl’s left thumb were painted under the broomstick. Whether or not the broom was part of the original concept is of some debate. In the X-radiographs (see [fig. 3]   [fig. 3] X-radiograph composite, Rembrandt Workshop (Possibly Carel Fabritius), A Girl with a Broom, probably begun 1646/1648 and completed 1651, oil on canvas, National Gallery of Art, Washington, Andrew W. Mellon Collection, 1937.1.74) there is the appearance of a reserve left for the broom. The area of little density within the costume, however, would not have been blocked in with dense paints since it conforms to the position of her red shoulder strap. To the right of the broom this red is painted over a dark layer, while to the left of the broom the red is painted over the white shirt, which may be an indication that it was applied as a result of a design change. Immediately above the shoulder is a dark area in the X-radiographs that seems to conform to the shape of a portion of the broomstick. Whether this diagonal shape is a reserve is also difficult to determine, in part because it abuts another dark area adjacent to the girl’s head that has no logical relationship to the final image.[21]  [21]
The only possibility that I can come up with is that the combined forms may have been a reserve for an implement with a horizontal piece at the end of the handle. In any event, the definition of the “reserve” that seems to correspond to the shape of the broom has been enhanced on the left by the paints containing lead white that were used at the last stage of execution to silhouette the figure against the dark background (and to cover PentimentiPentimenti
 An alteration made by the artist to an area that was already painted. in the girl’s shirt).

One bit of technical evidence that links the signature and date, the broom, and the bucket concerns their distinctive reddish orange accents, which have a vermilion component. Similar accents also appear on the girl’s curls and on her shoulder to the left of the broom, indicating that these other areas of the painting may have been finalized at this time as well.[22]  [22]
This observation has been confirmed through Karin Groen’s analysis of the paint layers. See note 8. Just why A Girl with a Broom would have been worked on at two different stages is not known, although it may well be that the painting was not originally brought to completion because distortions in the surface from the wrinkling paint had quickly developed.

Arthur K. Wheelock Jr.

April 24, 2014