Whistler’s Late Portraits of Ethel and Rosalind Birnie Philip

Whistler met his wife, Beatrice Birnie Philip (1857–1896), also known as Beatrix or Trixie, in the 1870s, when she was married to his friend the architect and designer Edward William Godwin (1833–1886). Like Whistler, Godwin had been greatly inspired by Japanese design. He designed the White House, Whistler’s residence at 35 Tite Street in Chelsea, London, where he lived between 1877 and 1879. Beatrice was the second oldest daughter of the Birnie Philip family, who lived nearby in Merton Villa. The father, John Birnie Philip (1824–1875), was a sculptor best known for his work on the Albert Memorial in Kensington Gardens. Beatrice and their oldest daughter, Constance (1854–1929), were also professional artists.1

After John’s death, his widow Frances (1827–1917) rented out space at Merton Villa as studios. Following Beatrice and Whistler’s marriage in 1888, two of the family’s younger daughters acted as Whistler’s secretaries: Ethel (1861–1920) from 1890 to 1894, and the youngest, Rosalind (1873–1958), from 1896, following Beatrice’s death from cancer. Ethel married the literary critic Charles Whibley in 1894, and, like Whistler and Beatrice, the couple divided their time between London and Paris. Rosalind, the last survivor of the ten siblings, inherited Whistler’s estate in 1903. She made substantial donations to the University of Glasgow in 1935 and 1955 and bequeathed the remainder in 1958. These form the collection of paintings, watercolours and the artist’s materials at the Hunterian, University of Glasgow, and of his private papers and photographs in its Archives and Special Collections department.

Whistler had many opportunities in the 1890s and until his death in 1903 to paint the dark-eyed, dark-haired Birnie Philip sisters, but perhaps little incentive to complete all their portraits, when there was no commission to be fulfilled. There is thus some uncertainty over dates; most of the portraits of the sisters at the Hunterian were never exhibited in Whistler’s lifetime and never sold, and therefore cannot be proved to have existed at a specific date. He painted Ethel nine times and Rosalind at least four times.2 He also made lithographs of Beatrice, capturing smaller and more intimate views of her resting during her illness, and he had painted her during her first marriage in Harmony in Red: Lamplight 1884–6 (YMSM 253; The Hunterian, University of Glasgow).3 The other five sisters do not figure at all in his paintings. A photograph taken between 1903 and 1905 shows four of the sisters, including Ethel and Rosalind, in later life (fig.1), grouped in front of a folding screen by Whistler, Blue and Silver: Screen, with Old Battersea Bridge 1871–3 (YMSM 139; Hunterian).4

The five portraits discussed in this article unquestionably capture Ethel or Rosalind, from the evidence of photographs from this period or later. It is always obvious which of them is depicted, no matter how sketchy or reworked the painting. Whistler mentioned these portraits in passing in his extensive correspondence – with references to making changes to the costume, reworking the pose or simply making good damage resulting from storage – but often without making clear which portrait was being discussed, and therefore which was painted when or over what period of time. While the dates given in the online catalogue raisonné by Margaret F. MacDonald and Grischka Petri are used here for identification of each work, few of them are certain.5

Fewer of Whistler’s portraits have undergone technical investigation than those made between the mid-1860s and the 1880s. Examined at the Hunterian between 2017 and 2024, the five portraits discussed here can be used to clarify the concept of Whistler’s ‘finish’, much discussed in his lifetime and since, and the distinction between an unfinished and therefore unexhibitable painting, as opposed to one that Whistler regarded as complete, saleable and exhibitable, even if contemporary art critics questioned this.

This article discusses one portrait of Rosalind and four portraits of Ethel. The painting of Rosalind is the only portrait of her ever to be exhibited, and therefore the only one with a clearly identifiable history during Whistler’s lifetime. It is almost in the condition in which it left Whistler’s hand since it was not treated by the same conservators who worked on other paintings in the Hunterian in the mid-twentieth century, often without providing images of condition or original framing before treatments. This is very rare for Whistler’s portraits in any collection. The portraits of Ethel were selected as the best candidates from all her portraits for linking specific canvases to the artist's correspondence.

Rosalind Birnie Philip

Rosalind’s half-length portrait The Jade Necklace 1896–1900 (YMSM 478; fig.2) was painted in Whistler’s studio at 8 Fitzroy Street, London, in 1896. It was possibly reworked in 1898 and certainly reworked after its exhibition at the Society of Portrait Painters, London, in 1900, with the title Green and Gold: The Lady of the Black Heart.6 Not all reviews were good, leading to Whistler’s decision to rework the portrait with a different costume.7 To this end, he specified a white jacket with ’a black heart', while implying that this differed from the white jacket which a review of the original portrait had described.8 The portrait today shows Rosalind all in black, and wearing a striking necklace, which may have belonged to Beatrice, and which Rosalind would later give to the University of Glasgow (fig.3). It consists of three strings of jade and appears greyish rather than jade green against the black fabric. Rosalind chose The Jade Necklace for the portrait’s title when she included it in the 1905 memorial exhibitions.

Whistler would sketch Rosalind in oil and other media, emphasising her long face and including the same hair style (figs.4–7). Whistler is unlikely to have exhibited these unfinished sketches, none of which depict the costume in detail although none appear to correspond precisely to that in The Jade Necklace.

Evidence of Whistler’s reworking was revealed during examination.9 The pose was subtly reworked at least once, with the proper right shoulder slightly raised and the left lowered, the face narrowed at the chin to make it appear longer, the lips thinned and the nose shortened. Further evidence of reworking is revealed by drying cracks that have now formed in these areas, simultaneously suggesting that more paint was added than had been removed by scraping, and obscuring the form of the earlier features. Ultraviolet light makes it clear that very yellowed and moderately thick varnish is present in two layers: this fills up the cracks and makes the original features even harder to capture by imaging (fig.8a).10 A wide, shoulder-width collar once extended just beyond the shoulders but it was dark grey, not light or white – there is no evidence on the surface that a white jacket was ever painted, nor any convincing evidence that it was effaced either. The dark greenish-grey background has the thickest paint of all, because this was repainted by Whistler: examination in transmitted light shows that Whistler limited his usual practice of rubbing or scraping down paint to the face itself (fig.8b).

The portrait is on a plain-weave canvas with 18 to 20 threads per centimetre, on its original stretcher and unlined. The tacks attaching it to the stretcher are not evenly placed nor used economically, as is usually the case on a canvas supplied by a British colourman such as Winsor and Newton, whose large scale of production led to uniformity.11 Unusually, this canvas is a dark navy colour, the fabric dyed after weaving, not made from blue threads. The undyed threads of the canvas have such varied thickness that they were very likely hand spun, but the cloth was machine woven. The warp thread, the only direction that had a loose thread suitable for removal for fibre identification. resembles ramie more than the flax used in British canvases. This and the handspun threads may indicate a French origin. To the authors’ knowledge the canvas is unique among Whistler’s supports, making it impossible to assert whether it is of British or French origin, given than Whistler worked in both London and Paris in his last decade.12 The canvas has a thin, mid-grey commercial ground, likely based on lead white. Unusually for Whistler’s portraits from any period, the reverse is thinly and fairly evenly painted in a warm dark purplish brown, applied before the canvas was stretched. Both this paint and the navy blue canvas may make an optical contribution to the appearance. Painting on the reverse is rare, and when it occurs it is usually a light grey or offwhite. Whether Whistler painted this unusual canvas and then stretched it – or had someone else stretch it – cannot be determined.

Having begun the face, Whistler scraped back the paint layers around the face and left ear in particular to reveal the top of the canvas weave, a practice Joyce Hill Stoner has termed ‘weavism’.13 The thinned paint around the eyes (fig.9a and b) was applied and then rubbed back, while every small brushstroke to define the shadow of the eyelid or the eye itself was rubbed back and reapplied more than once, until the eyes emerged as distinct tones of paint (fig.9c). Details such as the eyebrows were added with a very fine brush and thinned paint. The surrounding flesh paint, though thin and also rubbed back and heavily reworked, is more substantial than that for the eyes. Whistler must have deemed the lips successful at an early stage of painting, and they were not rubbed back and reworked (fig.9d). Their paint includes lead white, bone black, red lake and vermilion, while for the pale skin Whistler substituted red ochre for vermilion, but added significant amounts of bone black to the shadowed areas. The grey ground makes an optical contribution to the pale skin, both in spots revealed by rubbing down, and through the thin flesh-coloured paint. The lightest areas of the face were applied last in a lighter shade of flesh paint.14 The shadow below the nose and the proper left ear also show severe and localised rubbing down (see fig.8b). Both the thinness of the paint and the process of rubbing down are characteristic of Whistler from c.1870 onwards.

The reworked costume seen today is created from black and dark grey paint, possibly toned with other colours that are not easy to see against the dark surface. The fabric around Rosalind’s throat does include a brown pigment: this area seems to have been reworked less than the rest of the bodice of the dress. Grey rather than black was used for the bodice, while all the beads in the necklace were painted in a much greener paint mixture that likely includes green earth. Applied on top of darker paint, it looks grey, a good match to the surviving necklace. Each bead is painted very thinly, and although the brushstrokes look rapid and free, he depicted its graded bead size accurately, using one carefully premeditated brushstroke for each bead.

The two varnishes, both much yellowed, unevenly brushed on, and highly fluorescent (see fig.8a), likely made of natural resin, the upper toned with a fine black or brown pigment, significantly reduce the subtle contrasts between Whistlers black, grey-black and greenish black paints. The greenish black background is the most affected (see fig.2). The lower varnish may date from the reworking of 1900 while the upper must be posthumous.

Infrared reflectography (IRR) can give valuable information on underlying layers, and hence on earlier versions of the costume (fig.10). There is a suggestion of closely fitted sleeves and a fitted bodice, with the waist more clearly defined and slender than can be judged through the varnish today, even if the span of the waist was not reworked. It is just possible, but not clear, that the right hand, now extending down and beyond the canvas, might have been painted in the original version. Possibly the costume was less dark, but the rendering in IRR of a great deal of black pigment does not suggest anything as light as a mid tone. Surface examination did not indicate any vestige of white fabric in areas with rubbing down, both being harder to discern than the style of the sleeves, even with close looking. The dark background, then not obscured by varnish, possible enhanced the appearance of lightness.

Whistler created a portrait that satisfied him as well as Rosalind, so much that she included it in the 1905 exhibition. The altered dress was so well disguised that it appears more complete – although whether Whistler considered it finished is not known – than many of Whistler’s late full-length portraits of other sitters that he also exhibited.

Ethel Birnie Philip

This article discusses four portraits of Ethel.15 In Sketch for a Portrait of Miss Ethel Philip 1889–96 (YMSM 386; fig.11), hereafter Sketch, Whistler concentrated on Ethel’s face, hat and the high collar of her black walking dress or riding habit. Her white-gloved hands and the (probable) umbrella she holds are more suggested than painted. Ethel is indoors, standing in a green-painted room with a dark floor. The loose brushwork indicates that Whistler was feeling his way towards the optimum placement of the fabric of the skirt relative to Ethel’s tilted-back upper body. As in The Jade Necklace, the Sketch has a grey ground. The vivid green wall was painted by mixing yellow ochre into grey paint, with no further attempt at creating a subtle colour harmony by using yellow ochre in all other areas of the surface, even in small amounts. This lack of harmony could be seen as a strong indicator that this was a sketch not intended to be developed into a finished portrait.

The placing of the head and the style of the hat were carried over into Red and Black: The Fan 1889–96 (YMSM 388; fig.12), hereafter Red and Black, in which Ethel wears a red evening costume and bow in her hair, and is turned slightly more towards the viewer. She also wears a lightweight waist- or hip-length evening cape in black, which is draped over her left shoulder beneath the boa; a catch to fasten the cape was probably concealed beneath. The fabric of the cape hangs down much lower on Ethel’s left, ending beneath the (feathered) fan she holds between her hands, and is only just visible at her right side, hanging behind her a little below waist level.

Whistler’s correspondence with Kennedy refers to three portraits of Ethel in his London studio that Kennedy had contemplated purchasing, seemingly Rose et Or: La Tulipe 1892–6 (YMSM 418; fig.13), another portrait of Ethel, Mother of Pearl and Silver: The Andalusian 1891–1900 (YMSM 378; Hunterian), and another in which she wore black.16 Exactly which painting is referred to in the letters has always been unclear. Correspondence in 1896 between Whistler and the art dealer Edward Guthrie Kennedy (1849–1932), records Kennedy’s discovery that Whistler had reworked, and to his mind ’spoiled’ the ’full length [portrait] in pink’ that Kennedy had offered to buy. Offended by this reaction, Whistler withdrew the picture from sale. This account may relate to either work. Furthermore, the same portrait in the correspondence was described both as ’pink’ and as ’red’, again introducing the idea of reworking to change the colour of the costume in one of Ethel's portraits. Technical study suggests that this must be Rose et Or.

Examination and technical imaging identified no substantive changes made by Whistler to the pose or the costume in Red and Black.17 The final application of paint for the dark red background, which may be a dull crimson floor-length curtain meeting a featureless brown or grey floor, has slightly slimmed the form, narrowed the width of the skirt and shortened the cape. The proper left eye was originally lower, but the pose of the head was not altered, having been worked out already in the sketch. This minor refinement of a pose is so ubiquitous in Whistler’s portraiture that it cannot be classed as significant reworking.

The dark flounce on the bottom hem of the cape remains the most difficult part of the dress to interpret when it is viewed closely: it may denote the placement of darker paint to balance the composition, and an intention to turn this into a flounce or black trimming. The black feather boa that is wound fully round Ethel’s neck reaches at least to the level of the hem on the proper right side, with the other end disappearing from sight behind her. Its full length remains unknown; boas were made in various lengths in the 1890s, the longest intended to touch a hem with both ends.18 The proper left arm must have been held just behind the body, out of sight of the painter. There is no evidence that the left hand was ever intended to be visible.

Red and Black evolved over several years. Whistler may still have been working on it at the time of his wife’s fatal illness in 1896. Red and Black was first exhibited shortly after Whistler’s death, in 1903, at the 13th Exhibition of the Royal Society of Portrait Painters in London. The label on the back is therefore likely in Rosalind’s hand: ‘Rouge et Noir. L’Eventail / (unfinished) / J. McNeill Whistler’. The painting is now lined and not on its original stretcher.19 Rosalind had a number of canvases restored for the memorial exhibitions. The replacement stretcher is too large by a few millimetres on all sides, and the original canvas area, just about trimmed of the tacking margins that ran round the sides of the original stretcher, sits within its area.

The ground on the canvas now looks like warm grey, darker than that for Sketch, although the fact that the painting was lined makes it impossible to tell whether it was originally a warm or a cool grey. The paint was thinly applied, without impasto anywhere, and must have been thinned considerably with turpentine. The face was rubbed down at least once and reworked in the same position. The iris of the proper left eye was lower at an earlier stage, but the nose was confidently modelled, with a single long, lighter, brushstroke added last for a highlight. The face is the only area rich in lead white, and traces of vermilion, red earth and umber could be detected here. The wet red paint mixture of the lips was wiped back and they have an unresolved appearance. The costume was painted throughout in the same colours: a translucent brownish-black and an opaque orange-red. Vermilion and red earth pigments are present in the red paint, while the ‘black’ paint includes brown umber as well as bone black and is thus warm in tone.

The red must have been applied first to establish the figure, then the dark boa and cape were applied on top, slightly overlapping the dress in the case of the cape. Whistler applied several brushstrokes of red and warm black in succession to establish the feathery outline of the boa, each with some overlap of the other colour. The last brushstroke applied to the bodice were red ones, for the dress rather than the black boa, although in reality the black boa lay on top of the red dress fabric. The red background – a mixture of red earth, with some vermilion and bone black and minor amounts of cadmium yellow – was painted up to the figure and the unfinished fan, as was the floor, with added umber to give a more yellow cast and some contrast with the wall colour in the background. That Whistler sought to establish colour harmony in this painting can be seen in its severely limited palette, with vermilion, red earth, umber, possibly brown earth, and bone black present in all the colours he mixed, even the flesh tones.

Several natural resin varnishes overlie the painting, applied both after the early lining and after a wax impregnation treatment and re-varnishing, which the Hunterian’s conservation record dates to the 1960s.20 Such a treatment produces the visual effects of lining a canvas painting, but without the addition of a secondary canvas. Why this treatment was deemed necessary is unclear. These varnishes have yellowed, combining with the added adhesive and wax to give the picture a slightly greenish cast, most obviously in the floor. This acts in opposition to the warm red and brown-black colour harmony Whistler created. The position of the fan was resolved, but, as noted above, the dark trim or flounce on the hem was not; viewed close up, both may merit the description of ‘unfinished’, as does the poor resolution of the flooring and dull crimson curtain. Although the costume could have been refined Whistler stopped painting it at this point, satisfied with the portrait and adding his butterfly signature at left, below Ethel’s waist. Rosalind also deemed the portrait acceptable for exhibition, albeit ‘unfinished’ by the standards of the time, as evidenced by her label.

Harmony in Black: Portrait of Miss Ethel Philip 1894–6 (YMSM 419; fig.14) also depicts Ethel’s right profile, which she likely considered her best side. A study in black and dark tones that is relieved only by Ethel’s pale face, the portrait is very hard to read, even when unframed and strongly illuminated. That the varnish is glossy also makes the paint surface difficult to see clearly, even under magnification. An early photograph of the painting gives the clearest impression of her costume (fig.15). The orthochromatic film used for indoor photography at the time provided an image that resembles a contemporary infrared photograph, revealing mixtures of black paint with different proportions of black and other pigments more clearly.21 This photograph, alongside varied illumination of the painting, indicates that Ethel wore a heavy, floor-length, decorated black coat.22 The back of the coat and the extended split sleeves have a pattern made from dark grey brushstrokes, which might indicate embroidered or brocaded fabric, and the coat style is edge-to-edge, not buttoned or otherwise fastened down the front. A black feather boa is visible at Ethel’s neck within the standing, rounded collar of the coat, its visible end reaching her waist at the front, which means it was a different, shorter accessory than the floor-length one depicted in Red and Black. Her feet are not visible, but her right hand, in a short, cream-coloured glove, just extends through the split sleeve, and she carries a pinkish-cream closed fan with a ribbon looped over her wrist. She wears a red ribbon, but she does not wear a hat or a veil although otherwise dressed for outdoors. There is no evidence that the costume was modified during painting, although the position of Ethel’s back, just below the shoulders, was slightly adjusted, possibly to emphasise the waist.

The background is plain black and featureless. Ethel is in an interior with a yellow-brown floor, possibly wood, although no planks can be seen. There is no skirting board depicted but it is clear that behind her, to the right of the painting, two walls meet. Diffuse lighting that could have come from a window falls from behind her, causing the skirt to cast a gentle shadow on the floor in front of her. Halfway up the canvas, to the left and behind Ethel, is a very large, stylised red butterfly signature consisting of seven brushstrokes. The rightmost brushstroke overlies the long, thin black stroke that defines the back of the coat. The lower three brushstrokes of the butterfly are lightly applied and fuzzy, but the top four appear to have been reinforced. All lie beneath several layers of varnish, and likely at the same level in the stratigraphy, suggesting that all the brushstrokes are Whistler’s own. Traces of paint just visible with a light angled carefully onto the surface indicate that another butterfly was placed just above this one but was later almost completely wiped off. The butterfly is not obvious in the early photograph, but its orthochromatic film would have underemphasised reds, so it cannot be concluded that the photograph predated the application of either butterfly.

Harmony in Black has a medium- to fine-weight canvas that is very slubby and textured, with a plain, very open weave that suggests it was made in France.23 French canvases were quite readily available in London by this date, so this does not imply that the canvas was purchased in France or, by implication, painted there. The canvas was glue-paste lined to a finer canvas a long time ago and likely after Whistler’s lifetime, perhaps in response to the water damage visible on the reverse. The tacking margins that had gone around the sides of the original stretcher were trimmed off all the sides, and the cut edges of the original canvas were taped to neaten them, a common practice by liners in Whistler’s lifetime and for a few decades after that. Non-original varnish seals down the tape, in the picture plane. The lining canvas might have a zinc white-based ground, since zinc was detectable from the reverse but in none of the paint analysed from the front.24 The stretcher is solid and well made but may not be original.

The canvas features a dark brown, opaque imprimatura – an initial layer of colour applied to the ground supplied on the canvas – that includes lead white, chalk, barium sulphate, silicate-based extenders, Mars brown and a small amount of Mars yellow, both of which are manufactured and more intensely coloured forms of earth pigment, but not umber, a brown earth pigment which contains an additional metallic element. Only obvious under magnification, there is also a thin but opaque pale pink or mushroom-pink preparatory layer beneath this, made from lead white, red lake, bone black and deep red or brown Mars colours, with chalk added. There are three tack-head-sized circles in the paint in the yellow-brown floor, arranged vertically but slightly askew, in line with the inner edge of the stretcher bar, and thus well into the sight area. If these are ground-down tacks, they might have only gone through the imprimatura or the preparatory layer, since they underlie Whistler’s paint. Their purpose is unclear. This could hint at the use of a temporary stretcher during painting. If Whistler used a temporary stretcher, he widened the format, bringing the tacks into view. Examination of other earlier Whistler portraits has not suggested he did this habitually, although it is notable that the late portraits have varied widths but similar heights, and therefore generally required custom-made stretchers rather than standard formats.

The paint for the yellow-brown floor in Harmony in Black includes lead white, bone black, Mars brown and some Mars yellow. When examined with a microscope, the face in particular shows evidence of rubbing down to remove earlier paint, and this has in different areas exposed the canvas tops, the pale pink preparatory layer or the dark brown imprimatura. The shadows around the eye are particularly thinly painted and consist mostly of brown imprimatura rather than flesh-coloured paint. The pale paint on the cheeks is the thicker and more opaque area of the face, but still thinly applied. Whistler painted Rosalind’s face in The Jade Necklace in the same way. The front hair is only lightly suggested on Ethel’s forehead, with a small amount of thinned paint. Her pale lips were painted with at least one pinkish red lake (fig.16), not the brighter vermilion Whistler often used for lips, but there is no evidence of colour loss here as can occur when red lake is used for flesh painting. The paint of the background and the greater accumulation of paint for the coat were both rubbed down to reveal the brown imprimatura, but only very rarely down to the canvas tops, in the course of painting. Both these apparently black areas have numerous extremely light scumbles of dark grey paint, present not just in the grey embroidery of the coat but throughout the darker areas too. These scumbles are hard to discern due to the accumulated effects of now-yellowed varnish applied after the canvas was lined, the remains of earlier varnishes beneath it, and the presence of a thick, glossy modern varnish that may be made of polyvinyl acetate and which is scratched from dusting the surface as well as from trapping dust and fibres.

The paint is thin almost everywhere, and thus the weave of the canvas at low magnification is visible almost everywhere, since even areas that has paint more thickly applied such as the face, were then rubbed down. All the black areas include bone black; all the red areas, such as both butterfly signatures and the red ribbon, include vermilion; the floor must include brown earth to achieve this surface colour; and Ethel’s dark brown hair includes umber. This, with the red lake pigments chosen for her pale skin and lips, is the same, very narrow palette shared by Red and Black and Ethel’s portrait Rose et Or: La Tulipe.

Rose et Or looks most like a fully finished portrait, at a viewing distance of 3 to 4 metres. Ethel’s posture implies that she stood on a low dais, likely concealed by the train of her light pink evening gown, on a brown floor, a couple of metres in front of a crimson curtain. Facing the corner of the room she turns her head towards the viewer. For Harmony in Black she was standing directly on the floor, not on a dais. In this portrait there may be a deep skirting board, more pink in tone than the floor, which was probably made of light-toned wood. On close inspection Rose et Or shows evidence of Whistler having refined Ethel’s pose and altered her figure. He narrowed the waist at both sides, which is the most significant change. He also slimmed the left shoulder and made the right shoulder more sloping; at both shoulders, he painted over an earlier, red gauzy fabric with paler and pinker tones. The deeper pink puffy sleeves meet a pale, flesh-coloured bodice, the fabric of which also underlay the gauzy sleeves and skirt. There is some retouching at the left side of the slimmed waist and hips. Whistler painted the forearms after narrowing the waist. While the hands remain unfinished, the pose of the arms is clear. The left arm was altered until it became too slim, but this was not corrected. Infrared reflectography (IRR) did not provide much more information than could be seen by close looking.

Long, tight cuffs in a very pale pink, striped, gauzy fabric extend from the sleeves to or beyond the elbow, and the long gloves Ethel wears extend over these cuffs. The dress appears to be made from sheer and gauzy fabrics stitched over opaque and flesh-coloured fabric beneath the bodice and skirt. Close examination confirms that Whistler built up the layers of fabric in paint: the skirt consists of at least two layers of paler pink paint, depicting a gauzy fabric, over another, similarly gauzy layer in red, with the flesh-coloured opaque fabric painted beneath. The scarf, which combines red and pink gauze fabrics and appears to have been starched, is the most detailed part of the costume, which was steadily developed but not altered much in style nor in colour during the painting process. The ambiguity of what Ethel holds in her hands was not resolved by the 2023 study. It looks as if only a little paint was ever applied here, and this was then wiped away. It is possible Ethel did not hold anything at all but simply clasped her hands behind her. Other details such as the red decoration in her hair – perhaps a comb, feather or lacquered sticks – have defied resolution. At the top right corner, something was so thoroughly effaced from the background that it is unrecognisable now.

Cusping of the fine canvas for Rose et Or on all four sides, now lined into the picture plane, indicates the painting originally had a slightly smaller format;25 at the same time, the tacking margins were – unusually – cut off unevenly with fairly small scissors. This suggests that lining led to the use of the non-original, slightly larger frame that is around the painting today. The lining canvas is even finer than Whistler’s canvas, and it probably has a ground of lead white and zinc white on the side that is now inside, like Harmony in Black.

The floor on which Ethel stands consists of the brown imprimatura alone, made from lead white, chalk, yellow ochre and umber. The crimson background includes red ochre, umber and chalk, the paint applied more thickly here than elsewhere, although it is still thin by conventional standards. All areas of the costume are thinly painted. The skirt, sleeves and the scarf all include lead white, vermilion and red ochre; a false colour infrared image suggests that this range of pigments is common to the whole costume.26 Certainly the flesh paint includes the same mixture. A red lake pigment (not madder, from its absence of fluorescence) could be seen in the pale bodice under magnification, and it may be widespread in the pink areas of the costume although it is not discernible in deeper colours. Black pigment is present in even the palest shades of pink paint. There are no runs of liquid medium, although areas of the skirt are clearly medium-rich, and there is no clear evidence for the addition of what Whistler referred to as his ‘sauce’.27 The face and hair were more extensively rubbed down than the costume or background, and were repainted thinly, probably several times. In numerous places, this has revealed the canvas tops or the imprimatura. Early varnishing, which could have been done by Whistler himself, took place while the canvas was in a frame, and residues of at least one early varnish remain. The topmost varnish may be made polyvinyl acetate, available after his lifetime. Their combined effect is not visually disturbing, since they are collectively little yellowed in comparison to those in the other paintings discussed here. The title Rose et Or (Rose-pink and Gold) could in fact indicate that varnishes, since removed, had yellowed considerably, altering the overall tonality and giving it a warmer, perhaps golden tone that Whistler himself had not painted. A typed label on the reverse, and a handwritten one in the hand of Rosalind, suggest that ‘Or’ in the title was used only after Whistler’s death, that is, at a period when the varnish had had time to grow yellow.

Conclusion

Materials analysis and close examination indicate that each of the portraits discussed here was painted with a limited palette of colours, upon a grey ground or imprimatura, and that even areas that were not reworked were often rubbed down to reveal it. Bone black features in all colours depicted, while red ochre and vermilion, with some use of red lake, served for flesh paint, red and pink fabric, and backgrounds. The palettes for all the portraits are very similar, with only a few additional colours that loosely suggest the decorative scheme of the room depicted. In these late portraits, Whistler employed the techniques he had developed in the 1870s for portraits such as Miss May Alexander 1874–5 (YMSM 127; Tate N05964), with the brushstrokes often even lighter and even more suggestive of jade, brocade or feathers than in that decade.28

Rose et or and Harmony in Black have a similar history of early treatment including lining, that may have been posthumous and done at Rosalind’s instigation. Whistler’s correspondence with Kennedy refers to a late portrait of Ethel that was reworked. Technical examination implies that this must be Rose et Or rather than Harmony in Black, since only the former shows evidence of significant reworking including changes to the costume as well as minor improvements to the pose of the sitter.

The Jade Necklace, which has undergone less posthumous intervention than most of Whistler’s portraits from the 1890s in the Hunterian’s collection, reveals more clearly how skilfully he could depict detail in a predominantly black costume, and how accurate and expressive his single and rapid brushstrokes could be in the achievement of a finished and characterful likeness, even if at close quarters they suggest a painting still in progress. The single brushstrokes used by the painter John Singer Sargent (1856–1925) to denote a jewel or the gleam of a well-polished black shoe often have the same expressive quality when examined at close range, but in contrast have never drawn accusations of lack of finish. That Rosalind regarded The Jade Necklace as fully acceptable and exhibitable is evidenced by her inclusion of it in Whistler’s memorial exhibitions.

Whistler would have expected his full-length portraits to be appreciated in a gallery or a formal domestic setting, hence at a distance of three or four metres. Close examination ignores this obvious fact. At the correct viewing distance, the lack of resolution at the hem level in the floor-length gowns and coats, of Ethel’s hands and left arm in Rose et Or or of Rosalind’s waist in The Jade Necklace, all concerning to a close observer, become unnoticeable. The sitter’s appearance and mood, the latter echoed by the title Whistler gave to the painting, dominate the viewer’s perception. The canvas looks aesthetically finished. While the butterfly signatures in Harmony in Black and Rose et Or also disappear at a viewing distance, this is caused by accumulated, yellowed varnishes and could be reversed by varnish removal. The technical findings of this study would assist with planning such a treatment. The butterfly signatures would have been more obvious in Whistler’s lifetime, and integral to the aesthetic of each portrait.

These thoughtfully placed butterflies were his mark of approval of a portrait that was not only a good and characterful likeness, exhibitable and hence ready for a discerning purchaser, but also a successful collaboration between sitter and artist. This artistic collaboration towards a finished portrait has been discussed extensively in the context of Whistler’s portraits of his romantic partner and model Joanna Hiffernan (1843–1886) in the 1860s, and was explored in the 2022 exhibition Whistler and the Woman in White.29 The later portraits of his romantic partner in middle life, the artist Maud Franklin (1857–1939), have been interpreted in the same light.30 Whistler’s portraits of two of his younger sisters-in-law can be regarded in the same way.

Sketches in watercolour of his wife Beatrice following their marriage reflect their happy relationship.31 Sketches and portraits of his sisters-in-law Rosalind and Ethel Birnie Philip survive in larger number than portraits of Franklin or youthful ones of Whistler’s half-siblings; they hint at the collaboration and rapport between artist and sitter that were necessary ingredients for a successful and exhibitable likeness. Rosalind worked with Whistler as secretary and sitter, but she was also his friend. Her labelling of Red and Black as unfinished but exhibitable illustrates their shared understanding of Whistler’s aesthetic aims. ‘Finish’ in a very late Whistler oil portrait is defined not so much by materiality or coeval concepts of a completed portrait as by his acknowledgement of a job well done, with the addition of his butterfly signature as a visible seal of approval and completion.