Harvard’s Nocturnes and Whistler’s Sauce

The Harvard Art Museums in Cambridge, Massachusetts, have an especially large collection of works by James McNeill Whistler: more than two hundred objects, in every medium in which the artist worked, including oil paint, watercolour, pastel, etching and lithography. Among these works, the most important are the sixteen oil paintings and twenty-three drawings that were part of a bequest made in 1943 by the American lawyer Grenville L. Winthrop (1864–1943). Winthrop was a graduate of Harvard University and bequeathed his entire collection to the Fogg Museum, one of the three museums that compose the Harvard Art Museums. This included four oil paintings from Whistler’s Nocturnes series: Nocturne in Black and Gold: The Gardens c.1872–5 (YMSM 163), Nocturne in Grey and Gold: Chelsea Snow 1876 (YMSM 174), Nocturne in Black and Gold: Rag Shop, Chelsea c.1878 (YMSM 204) and Nocturne in Blue and Silver c.1871–2 (YMSM 113).1 For brevity, the paintings will be referred to as The Gardens, Chelsea Snow, Rag Shop and Blue and Silver, respectively, throughout this article.

Winthrop especially prized his Whistlers. Three of the nocturnes greeted visitors in the entrance hall of his house at 15 East 81st Street, New York (fig.1), while the fourth hung with other works by Whistler in their very own gallery, known as the ‘Whistler Room’ (fig.2).

When painting his nocturnes in oil, Whistler used a unique preparation and application of his painting medium, which he referred to as his ‘sauce’.2 Whistler’s sauce was made from heavily diluted commercial tube paints. Such a fluid medium facilitated his manipulation of the paint surface, which included scraping or rubbing back layers of paint. Whistler often began his paintings on the easel, but, as one of his assistants recalled, ‘so much “sauce” was used that, frequently, it [the painting] had to be thrown flat on the floor to keep the whole thing from running off’.3 Drip marks can often be seen on Whistler’s works from this period, for example in the bottom left corner of The Gardens (fig.3).

The composition of Whistler’s sauce has been the subject of much debate and inquiry. Several of Whistler’s contemporaries recalled him making the sauce with different ingredients. The artist Walter Sickert (1860–1942) claimed that Whistler thinned his paint with ‘oil and turpentine’.4 Walter Greaves (1846–1930), one of Whistler’s studio assistants, described the sauce as composed of ‘linseed oil and turpentine’.5 Elizabeth Robins Pennell (1855–1936) and Joseph Pennell (1860–1926), Whistler’s first biographers, maintained that it was a mixture of ‘copal, mastic and turpentine’.6 The only known published technical analysis identifying components of Whistler’s sauce was conducted in the mid-1990s, finding evidence for stand oil (linseed oil heated to high temperatures without oxygen), mastic resin and small quantities of a lead drier.7

In late 2020, when it became clear that the Harvard Art Museums would be closed for a prolonged period owing to the COVID-19 pandemic, the opportunity arose to deinstall two of Whistler’s nocturnes. This enabled a cross-comparison of the four nocturnes in the collection to address questions about how Whistler painted his famous night and twilight scenes. The study focused on the composition of the sauce to contribute material analysis that might be compared against primary sources and existing published research.

Technical analysis

Conservation history

Blue and Silver is painted on a wooden panel.8 Prior to Winthrop’s bequest in 1943, a secondary wooden panel and cradle, a grid of wooden slats used to reinforce and preserve the flat surface of the painting, were attached to the verso. Museum conservation records report the removal of an old varnish, inpainting to restore losses and varnishing using synthetic polymeric materials, in 1953.9 These synthetic materials were removed prior to additional inpainting in 1985.10 Chelsea Snow, Rag Shop and The Gardens are painted on canvas and have received no conservation treatment since their arrival at the museum in 1943. There is, however, evidence of earlier interventions, as both Chelsea Snow and Rag Shop have been lined, a process which adheres a secondary canvas support to the verso of the painting. Whistler was actively engaged in preserving the appearance of his paintings, and his correspondence captures his thoughts on how they should look and be cared for. Whistler preferred to have his paintings lined, as well as cleaned and revarnished periodically to preserve their original appearance.11 The Gardens is unlined, providing the rare opportunity to study a nocturne unaffected by lining treatment.

Designing the scientific study

To identify the components of Whistler’s sauce, the paint stratigraphy and composition as well as the surface coatings of the four nocturnes were studied. This combined approach allowed organic media, in particular natural resins, incorporated in the body of the paint, to be differentiated from any present as a final varnish coating, and offered the opportunity to look for evidence of the sauce being used. A surface scraping of the varnish layers was taken alongside two small samples from each painting. One sample was prepared as a cross-section to view the layers of the painting, and the other was used for medium analysis to identify the paint composition.12 Sampling was limited to the edges of the paintings where existing damage was present. This, along with the small number of samples studied, limits the possibility of making comprehensive conclusions about a whole painting. However, it does provide a glimpse into Whistler’s working practice for the areas studied, which were divided into two categories: light passages and dark passages. This draws on a distinction made by Whistler himself when titling his nocturnes, with ‘grey’ or ‘blue’ used to describe paintings lighter in tone, and ‘black’ for the darker paintings.

Light passages

The paintings Blue and Silver and Chelsea Snow were used to investigate how Whistler painted light passages in his nocturnes. Blue and Silver represents a scene at Chelsea on the River Thames in London, looking towards Battersea and depicted in the diffuse light of the setting sun (fig.4). In Chelsea Snow, a ground covering of snow and a grey sky provide the light passages, which surround dark buildings and trees on a street in Chelsea (fig.5).

The cross-section prepared from Blue and Silver reveals the layers of paint hidden below the surface (fig.6). The lowest layer, applied directly onto the panel, is a ground layer rich in lead white, with sparse particles of baryte and calcite. The coarse irregularity of the lead white pigment suggests the paint was mixed and applied by the artist. On top of this are single layers of purple-grey and blue paint, and finally a surface varnish layer. Both paint layers contain a more finely ground lead white pigment, mixed with bone black, red ochre, baryte and calcite for the purple-grey and ultramarine and cobalt blue for the blue.13

In the cross-section, a translucent medium is visibly swirled into the blue paint layer. This is clearer in the detail of the visible image, shown alongside the black and white backscattered electron image of the same area, where the swirls of medium appear black against the bright white pigment particles (fig.6b and c).14 Pigment particles appear to have been occasionally drawn into the medium, which suggests the medium was applied into the paint with the sweeping motion of a brush. This could be the result of Whistler adding more medium to his paint to make it more fluid but not completely homogenising the mixture, resulting in the swirling pattern that remains visible.

Analysis of the surface scraping, using thermally assisted hydrolysis and methylation pyrolysis gas chromatography-mass spectrometry (THM-Py-GCMS), revealed no evidence of natural resins and identified butyl methacrylate, the synthetic polymer resin used to inpaint losses in 1985.15 Analysis of the sample containing all the paint layers, using the same methods as the surface sample, identified the same synthetic polymer resin, but oil and Pinaceae resin were also present.

Pinaceae resin is a natural resin that comes from evergreen trees in the Pinaceae family, including species of larch, fir, spruce and pine. Two products commonly used by artists can be made from Pinaceae resin: turpentine and colophony (rosin). Turpentine is the liquid obtained from distillation of the resin, and colophony is the solid residue left behind from that process. Compositionally the two products are very similar and cannot be differentiated with ease (if at all). Regardless, the identification of Pinaceae resin in the paint sample but not in the varnish sample indicates that the Pinaceae resin occurs only in the paint layers, likely the blue paint layer, where the cross-section shows the addition of a translucent medium to the paint film. This identification could be consistent with accounts that describe Whistler’s use of turpentine in his sauce. Its use would result in a more fluid paint medium, which is a characteristic of Whistler’s paint.

Taken from an area of the grey sky, the cross-section prepared from Chelsea Snow has similarities to that from Blue and Silver (fig.7). A pale grey paint layer, made from a mixture of ultramarine, lamp black, lead white, baryte, calcite, clays and celestine, represents the artist-applied ground.16 The incorporation of canvas fibres into this layer confirms it as the lowest paint layer applied to the canvas. Above this is a second pale grey paint layer, a mixture of lead white, cobalt blue, bone black and gypsum. The boundary between this layer and the ground can be difficult to discern in the visible image due to their similarity in colour and tone, but it can be more clearly seen when viewed with ultraviolet light and in the backscattered electron image (fig.7b and c). This second layer contains swirls of medium into the paint, as seen for Blue and Silver, as well as a large pool of medium pigmented with ochre.

In visible light, a relatively thick, transparent layer sits on the surface of the sample. When viewed with ultraviolet light, this is shown to consist of three layers. First, applied over the grey paint layer, is an medium-rich layer with a blue fluorescence that contains occasional particles of calcite, clays and an unidentified lead phase.17 Above this sits another medium-rich layer that exhibits a yellow-green fluorescence and contains scattered particles of ochre, calcite and lead white. Unevenly applied and each significantly thinner than the underlying pigment-rich paint layer, these two media-rich layers were likely applied with a more fluid consistency. A non-fluorescent varnish layer is present at the top of the sample. The presence of dirt on the surface of this layer can be used to trace the upper boundary.

While the addition of medium to the paint is clearly visible in cross-section, analysis of the remaining samples for Chelsea Snow failed to elucidate what the source might be. Analysis of both the surface scraping and the sample containing all the layers identified Pinaceae resin and oil. Both samples also contained a compound that may be related to mastic; however, the use of the resin could not be definitively confirmed.18 If present, mastic is likely in the thinly applied layer just below the varnish, based on the characteristic yellow-green fluorescence observed in cross-section. This layer was probably unintentionally captured during sampling of the varnish, explaining its presence in the analysis of the surface scraping.

Dark passages

The scenes depicted in The Gardens (fig.8) and Rag Shop (fig.9) are almost obscured when viewed today. The paintings were already dark in tone due to their depiction of night scenes, but this has been exacerbated by ageing of the materials, which have darkened and increased in transparency. The cross-section prepared from The Gardens is immediately visually different from those described for the light passages above (fig.10). Instead of a traditional pigment-rich ground, a thick application of a transparent medium, with a blue fluorescence in ultraviolet light, surrounds the canvas fibres. In areas of damage and loss across the whole painting, this medium-rich layer can be seen applied directly on the canvas (fig.11).

Above this, layers of paint and media have been unevenly applied, creating swirls that often sweep into one another. This is best viewed with ultraviolet illumination (fig.10b). The irregular, uneven layers observed in this cross-section illustrate Whistler’s distinct approach to paint application in the dark passages of his nocturnes, whereby he used dilute paints that tended to run, drip and pool in an uncontrollable manner. These layers are sparsely pigmented with bone black, ochre, rare vermilion, gypsum, baryte and a lead phase.19

Analysis of the surface scraping by THM-py-GCMS revealed the presence of unrefined bleached shellac.20 Shellac is a resin excreted by the female of the lac insect. It is typically a dark orange colour and finds use predominantly in furniture coatings. However, it can be bleached to remove the colour and make it suitable for use as a picture varnish. Bleached shellac may be differentiated from non-bleached shellac based upon the identification of chlorinated marker compounds by THM-py-GCMS. Additionally, Pinaceae resin and oil were identified in the varnish sample. Bleached shellac, Pinaceae resin and oil were also identified in the sample containing the paint layers.

A chip of paint that dislodged from the edge of the painting provided the opportunity to study a second area of the composition.21 The sample does not capture all the layers present, instead the cross-section is dominated by a single medium-rich layer, that fluoresces when viewed in ultraviolet light, containing pigments in a variety of colours, including bone black, ultramarine, cadmium yellow, ochre, umber, chrome oxide green, gypsum and baryte (fig.12).22 Towards the bottom of this layer, sparsely pigmented areas are visible. The fluorescence of the pigmented and effectively unpigmented areas is the same, suggesting the use of the same medium, identified to be bleached shellac. The boundary between the pigmented and unpigmented areas is blurred. This may suggest they were applied as two layers in quick succession, or that the variation is the result of incomplete mixing. The bottom of the sample bears traces of an underlying paint layer, which contains zinc white and ultramarine mixed with zinc yellow, sparse vermilion and baryte.23 A very thin, unpigmented layer sits on the surface, presumably a varnish layer.

Elemental mapping was used to track the distribution of bleached shellac throughout the cross-sections of The Gardens (fig.13). Using the map for chlorine, which remains within the resin after bleaching, it was confirmed that bleached shellac is the transparent medium that surrounds the canvas fibres in the original sample. It is also present to a lesser extent, and in varying concentrations, in the upper layers. Chlorine is evenly distributed across the thick medium-rich layer of the second sample, including both the pigmented and unpigmented areas, indicating both areas utilise bleached shellac.24

The cross-section prepared from Rag Shop differs from The Gardens as it contains a more ‘traditional’ ground of lead white, calcite and baryte (fig.14). Although no canvas fibres were captured in the cross-section, the shape of the lowermost layer in the cross-section follows the contours of the canvas weave, suggesting that it is the ground layer on the canvas. The mixing of a black pigment, likely lamp black, in lead white for this layer indicates an artist-applied ground. Atop the ground is a brown paint layer that was revealed to contain multiple layers when viewed with ultraviolet illumination (fig.14b). The thickest layer, applied on top of the ground, has a transparent quality in visible light but becomes opaque, with some associated fluorescence, with ultraviolet illumination. These observations suggest a medium-rich layer. This layer contains a scattering of pigment particles: sparse bone black, ochre, gypsum, calcite, baryte and likely lead white. This layer also contains dissolved zinc, suggesting the use of a drier. Above this are multiple thin, unevenly applied layers with differing pigment morphology and concentrations, visualised more easily in the backscattered electron image (fig.14c). The thick, sparsely pigmented layer appears very dark in the backscattered electron image, especially in comparison to the ground layer below it. The layer applied directly above has a much higher percentage of pigment, with lead white, bone black, ochre and baryte. The bright white spots visible within this layer in the backscattered electron image correspond to the lead white pigment particles. This is overlain by another thin medium-rich layer with no obvious pigment particles. The top layer is medium-rich, with comparatively more ochre and bone black and less lead than the previous lead-rich layer. Occasional baryte and calcite are also present in the top layer.

Analysis of the surface scraping identified the presence of unrefined bleached shellac, as seen for The Gardens, in the varnish. Additionally, Pinaceae resin and oil were identified. Bleached shellac was not present in the sample containing the paint layers, where only Pinaceae resin and oil were identified. The varnish and paint samples were taken from different locations on the painting, and this analysis suggests that the bleached shellac is not present over the entire painting. Perhaps it was instead selectively applied or incompletely removed during an undocumented cleaning. The presence of chlorine was not identified in any of the medium-rich layers in the cross-section with elemental mapping, which rules out the presence of bleached shellac within the paint layers.

Conclusions from analysis

Whistler’s use of his sauce enabled him to work wet-in-wet and execute his paintings quickly. The perceived lack of finish in the nocturnes rankled many of Whistler’s critics, in particular John Ruskin. In 1877 Whistler exhibited eight paintings at an exhibition at the Grosvenor Gallery in London. Among the works were his Nocturne in Black and Gold: The Falling Rocket 1875 (YMSM 170; Detroit Institute of Arts), as well as Blue and Silver. Reviewing the exhibition, Ruskin declared: ‘I have seen, and heard, much of Cockney impudence before now; but never expected to hear a coxcomb ask two hundred guineas for flinging a pot of paint in the public’s face’.25 Whistler, claiming the piece amounted to libel, sued for damages.

During the trial the following year, Whistler was asked repeatedly by Ruskin’s lawyer John Holker how long it took him to complete his paintings. With regard to Blue and Silver, Whistler answered straightforwardly: ‘I completed the mass of the picture in one day, after having arranged the idea in my mind’.26 When he was asked how long he worked on The Falling Rocket, Whistler explained that it had taken two days to paint it. Ruskin’s lawyer challenged him: ‘The labor of two days is that for which you ask two hundred guineas?’ Whistler’s reply was met with applause from the gallery: ‘No. I ask it for the knowledge I have gained in the work of a lifetime’.27 It was this knowledge that provided Whistler the recipe for his sauce – a recipe whose ingredients his contemporaries could never quite agree on – and an understanding of the way paint and various media move, behave and might be manipulated to match his vision. This study of the four nocturne paintings in the Harvard Art Museums has provided insight into the composition of Whistler’s painting medium and the unique ways in which he painted with it.

Based on the analysis conducted here (see table 1), Pinaceae resin may be suggested to be part of Whistler’s ‘sauce’. Using contemporary written accounts as a guide, it seems likely that the presence of Pinaceae resin indicates turpentine, the use of which would have resulted in the fluid paint medium that is a characteristic of Whistler’s work. Analysis also indicated the presence of oil, which would be expected with the use of oil paint from a tube. It is difficult to conclude whether additional oil was added to the sauce as written accounts suggest, but it remains a possibility. No evidence for copal was found during this study. It may be that the recipe for Whistler’s sauce was not fixed, with evidence found for the incorporation of bleached shellac and perhaps mastic into the paint in some, but not all, of the nocturnes.

Cross-sections suggest that Whistler used the sauce in two different ways. To create the light passages, Whistler mixed his medium with a high concentration of pigment, whereas to create dark passages, the medium forms the basis of the paint, adding to this only a small amount of pigment. In the lighter compositions (or passages) of Blue and Silver and Chelsea Snow, single, thick, pigment-rich layers were applied. In these layers, pools or waves of added medium are visible, suggesting incomplete mixing. This contrasts with the darker compositions of The Gardens and Rag Shop, where multiple medium-rich layers are applied. The thickness of these layers varies greatly depending on the area sampled.

The thick application of bleached shellac to the canvas instead of a more traditional pigment-rich ground in The Gardens remains one of the most unexpected discoveries of this study, and this unusual application adds more intrigue to Whistler’s painting practice.