The French painter Edouard Vuillard (1868–1940) enjoyed a long career, of which barely the first ten years belong in the nineteenth century. All of the four paintings by him which are kept in the National Gallery were painted well into the twentieth century, and they are currently the latest pictures to hang in Trafalgar Square (note 1). Not distinctively ‘modern’, they nevertheless fit in seamlessly with the Gallery’s nineteenth-century collection. All were executed in the realistic style which Vuillard favoured in the last decades of his life, and which later blurred his reputation as an artist, complicating the perception of his oeuvre as a whole. In the 1890s Vuillard was one of the most progressive artists of his generation, at the forefront of the Nabis movement. Yet his early paintings, with their daring, radical distortions of motifs, had, by the turn of the century, given way to conservative representations of bourgeois society. Moving from the avant-garde works of his youth to the apparently conventional output of his later years, Vuillard seems to have embraced modernity ‘à rebours’.
While better known for his easel paintings – small, boldly synthetic compositions, interiors bathed in a quiet, hushed atmosphere and subtle portraits – Vuillard also distinguished himself as a versatile decorator. His commitment to the ornamental started in the early 1890s, coinciding with his engagement with the Nabis whose main precept, largely inherited from Gauguin, was that above all art should be decorative. Painting should leave the easel and return to the wall, as such reverting to its original status and purpose – its ‘mission décorative’ (note 2). Actively involved with the theatre world (he was a co-founder, collaborator, adviser and graphic designer of the Théâtre de l’OEuvre from 1893) Vuillard applied these principles by painting backdrops for the stage: large, complex, rapidly brushed settings and décors.
In the same decade, starting in 1892, Vuillard started producing large-scale decorations for the private homes of his patrons and friends: screens, canvases or panels, ‘intimate frescoes for apartments’, ‘not executed, as was done in the past, directly on the wall, but on moveable panels set into panelling’ (note 3). Vuillard ended up painting more than 50 such decorations, commissioned from rich clients and later from public institutions. While none of the stage sets survive – since they were soon repainted for the next production, or destroyed – most of these decorations still exist, having been preserved in Paris apartments or ‘hôtels particuliers’, yet for a long time relatively unknown. Large and cumbersome, difficult to move or transfer for exhibition, they were therefore largely out of sight and as a result remained, until recently, the least well-researched part of Vuillard’s production.
Two of the four paintings by Vuillard in the National Gallery are decorative works of this type: ‘La Terrasse at Vasouy’: ‘The Garden’ and ‘The Lunch’ (NG 6388 and NG 6373) (note 4). Painted in 1901 as a single large panel, divided and reworked by the artist in 1935, they are therefore both the earliest and the latest among the Gallery’s Vuillard holdings. The extensive alterations were executed in the typical looser manner of the artist’s late work. As they appear now, the panels have been considered at best a pretty and harmonious decorative ensemble; at worst an unchallenging, unambitious work in a ‘more or less pseudo-impressionistic style’ and, in view of the later alterations, ‘anachronistic [...] a sort of artistic bastard’ (note 5).
Not only did the panels endure major changes and a bad reputation as one of the artist’s least successful achievements, they also suffer – like similar panels elsewhere (note 6) – from poor physical preservation. Painted ‘à la colle’ or in distemper, like most of Vuillard’s large-scale decorative works reworked in the same technique they present specific condition problems, with particularly complex and fragile paint surfaces. Acquired by the National Gallery in the late 1960s and on long-term loan to the Tate between 1997 and 2010, they have, since their return to Trafalgar Square, undergone extensive technical examinations. These have greatly advanced the understanding of what the painting looked like in the first instance. They have also determined the extent of the alterations to both panels, in terms of composition, style and technique; specific attention has been given to the materials used in both 1901 and 1935. A clearer picture emerges of how Vuillard went about painting, dividing and repainting this work, which corroborates the documentary evidence available in the form of archives and photographs, and throws new light on his use of distemper.
‘La Terrasse at Vasouy’ was commissioned by the literary figure, dramatist and critic Jean Schopfer (1868–1931), who from 1904 wrote under the nom de plume Claude Anet (note 7). Cosmopolitan and polyglot, he was born in Switzerland, the son of a wealthy banker and an Anglo-French mother, and studied in Paris. First a journalist and playwright, and later the celebrated author of best-selling and internationally successful novels such as ‘Ariane, Jeune Fille Russe’ (1920) and ‘Mayerling’ (1930), Schopfer was a fascinating figure with a dizzying range of interests (note 8). The cultured dandy also doubled as a great sportsman – in 1892 he won the French Tennis Open – as well as an art critic and collector. As the correspondent of French and American magazines in his thirties and forties, he travelled as far as the United States, Persia and Russia. He collected art and antiquities along the way and brought them back to Paris, where he briefly ran a gallery of Persian art. His interest in art reached far back; before entering the Ecole du Louvre, Schopfer attended the Lycée Condorcet, then a hotbed for young artists and critics. There he first met Vuillard, and made friends with those who in 1891 were to found around Thadée Natanson (1868–1951) ‘La revue blanche’, a lively, cutting edge, avant-garde periodical covering literature, art and music. Both Vuillard and Schopfer contributed to ‘La revue blanche’ in the eleven years of its existence, the latter as a writer – there publishing his first essays and novels – but also as a critic and art historian of the decorative arts.
Schopfer came to own a small easel painting by Vuillard (note 9), but in the light of his own special interest in architecture and design it is likely that the writer rated the artist highest as a decorator. This is suggested by his three successive commissions to him, all of which were for decorative projects. These were destined for Schopfer’s Paris apartment, 132 avenue Victor Hugo, although given their timescale – they were completed respectively in 1895, 1898 and 1901 – they had presumably not been envisaged as an ensemble from the outset. The first of these commissions to Vuillard, and the first instance of collaboration between the two young men, was for a dinner service (note 10). An isolated experiment in Vuillard’s long career, it bore a handpainted decor of ‘Parisiennes’ in fashionable dress, with ruched, plain or patterned blouses and flouncy skirts, ‘weightless figures and hallucinatory floating border motives’ (note 11). Aesthetically close to Vuillard’s depictions of women in his prints of the same year, somewhat amateurish but fresh and highly original, it remains a unique Nabis decoration. Since Schopfer had just got married, this dinner and dessert service for twelve was probably intended as a gift to his bride, the wealthy New York socialite Alice Wetherbee, on the occasion of their wedding (note 12).
Two years later, in an article published in the ‘Architectural Record’, Schopfer campaigned in favour of a decoration adapted to the ‘tastes and requirements’ of the era by illustrating some of Vuillard’s plates from the dinner service (note 13). One should live in a modern decor, he argued, rather than in an environment belonging to another époque; Schopfer’s article praised Vuillard as ‘one of those young painters from whom great things are expected’ (note 14). The friendship between the two men had further flourished during a trip to Italy in January 1898 with Maurice Denis (note 15), and Schopfer’s loyalty to Vuillard soon resulted in a second commission, this time for decorative panels.
The artist had by then developed considerable experience in the genre, having completed some of his most important decorations for such private clients as the Natansons or Dr Louis-Henri Vaquez in the previous years. His first panels for Schopfer, ‘Le Jardin du Relais à Villeneuve-sur-Yonne’, were pendants showing the garden of Thadée Natanson’s house in the village of Villeneuve in Burgundy (note 16). As the unofficial country haunt of ‘La revue blanche’, Le Relais attracted a number of its contributors. Thadée’s wife Misia, Marthe Mellot and the painter Pierre Bonnard feature prominently, relaxing on a bench or rocking chair, while some people wander lazily under the tall trees in an atmosphere of Symbolist languor. Praised as a ‘pastoral symphony’ (note 17), even as ‘masterpieces that Vuillard never surpassed’ (note18), they were installed in the Schopfers’ salon in the apartment at 132 avenue Victor Hugo. Originally envisaged as a single composition, as confirmed by early sketches, but painted as two panels, the ‘Relais’ pendants were joined three years later by ‘La Terrasse at Vasouy’ (note 19). Schopfer’s third commission completed the ensemble, likely to have been conceived to suit the scale and configuration of the spacious Paris apartment: ‘a lively and appropriate ‘plein-air’ backdrop for the young, urban and upwardly mobile Schopfer “ménage”’ (note 20).
Like the ‘Relais’ panels, ‘La Terrasse at Vasouy’ depicts a sort of latter-day ‘fête champêtre’ on a grand scale. A spacious garden winds back from a lawn to a distant path, with a gate at the back. There, people stroll idly in the distance, while on the right, in the foreground, an elegant company is gathered around a table laden with the remains of an alfresco meal, caught in a happy moment of leisure. These pleasant, idyllic depictions of outdoor scenes stand in stark contrast to Vuillard’s former series of subtle but confined interiors, such as the Natanson’s ‘Album’, or the Vaquez panels (note 21). According to Katherine M. Kuenzli, they mark a transition in his work, presenting, in a post-Nabis, ‘plein-air’ style, ‘an unqualified and wholly sympathetic portrayal of bourgeois leisure’ (note 22).
Much evidence of this new direction can be found in the Schopfer panels. First, they differ from Vuillard’s previous decorative schemes by the specificity of their settings. Like the ‘Relais’ pendants, the National Gallery panels show a real, distinctive site: the villa La Terrasse, also called Les Pavillons de Vasouy, was formed of the remains of an eighteenth-century hunting pavilion (note 23), overlooking the sea at Vasouy, a village near Honfleur on the fashionable Normandy coast. Up on the cliff, above the beach, La Terrasse had been rented in the summer of 1901 by Vuillard’s friends Lucy and Jos Hessel. The latter worked with his uncle and cousins, co-owners of the Bernheim-Jeune gallery, which represented the artist. Although from a different social background, Vuillard was invited to stay with the Hessels, experiencing there the first of his summer ‘villégiatures’ (note 24).
Similarly, the ‘Relais’ and ‘La Terrasse at Vasouy’ represent the first occurrence in Vuillard’s oeuvre of decorations showing deliberate and obvious portraits (note 25). In both cases, these are Schopfer’s friends – various members of the artistic and literary Paris scene, several of them associated with ‘La revue blanche’, and also Vuillard’s patrons. Surprisingly, rather than depicting Schopfer’s own home, here Vuillard painted decorations showing the life setting of his (Schopfer’s) friends; the Schopfers themselves do not even seem to feature at all in the ‘Relais’ panels. It is likely that this and many other such decisions were taken in accordance with Schopfer’s own wishes (note 26).
Following the recent, renewed interest in Vuillard’s decorations, new archival documents have emerged – letters, diaries and photographs (note 27) – confirming that the two National Gallery paintings once formed a single large canvas, and allowing us to retrace its history (note 28).
The Vasouy project was certainly of paramount importance to Vuillard, who listed it among his seven main achievements or events for the year 1901 (note 29). It mattered no less to Schopfer, who went to great lengths to finance it: according to his daughter, the commission of ‘La Terrasse’ was made possible thanks to his proceeds as a tennis champion. In the summer of 1900 the sportsman funded the purchase by organising a tennis tournament in Divonne-les-Bains near Geneva (note 30): ‘With the first money he had earned in Divonne, he [Schopfer] had commissioned from Vuillard an immense panel of four metres by two, which was almost a family portrait.’ (note 31) The large panel was completed and installed by 18 December 1901, as attested by Maurice Denis in a letter to the writer André Gide, referring to the ‘inauguration of Vuillard’s paintings at Monsieur Schopfer’s’, which he attended (note 32). Vuillard’s original red signature from that first composition still shows, clearly legible in the lower left corner of the ‘Garden’ (note 33).
The three panels remained in the apartment for which they were intended for barely more than a year. Engaged in divorce proceedings, and having saved ‘La Terrasse’ for himself in the settlement, Schopfer asked Vuillard whether he would still have a studio where he could ‘house the last large panel which you made for our salon’ and confided: ‘My heart breaks at the idea of abandoning the two other panels which I liked so much – but I don’t want to beg for them.’ (note 34) When the divorce was granted on 23 January 1903, a ruined and dispirited Schopfer moved to a smaller apartment (note 35) while Alice, ‘la belle divorcée,’ (note 36) settled in California with Anne, the couple’s daughter, and most of their furnishings – but without the ‘Relais’ panels, which were given or sold to their friends Emmanuel and Antoine Bibesco (note 37).
It was not until 1909 that Schopfer’s fortunes turned and he could again accommodate ‘La Terrasse at Vasouy’ in his own home. He was by then writing under the name Claude Anet (note 38), as well as acting as a successful dealer (note 39); having settled at 63 rue de Chaillot in 1908, he was able to collect ‘La Terrasse’ from Vuillard a few months later and install it in his new apartment, surrounding it with some plates from the 1895 dinner service, hung in a frieze. In 1911, a year after Schopfer’s remarriage to Clarisse Langlois (note 40), the panel was de-installed yet again. At 108 rue du Bac, a groundfloor apartment, a special wall was built for its display between the salon and the dining room (note 41). ‘This panel took up one wall of the dining room and, the height of luxury when my parents were entertaining, the guests were served in a unique service, fired and painted by Vuillard, especially for my father.’ (note 42) In addition, new decorative paintings were commissioned – from Kerr-Xavier Roussel, Vuillard’s brother-in-law – creating a new context for ‘La Terrasse’ (note 43).
Schopfer died in 1931; in 1935 his widow returned the painting to the artist, who cut it vertically into two halves and reworked it. By 1936 ‘La Terrasse at Vasouy’ had been sold to the Paris dealer Raphaël Gérard, who exhibited the two panels in Vuillard’s major retrospective show two years later as ‘Un jardin en Normandie’ (note 44).
Confusion about their origin seems to have begun from the time of their first public showing. The 1938 exhibition catalogue gave the separated panels the erroneous date of 1898, which, combined with their vague, unspecific title, gave plenty of scope for mistaking them for the two ‘Relais’ panels. The confusion over date seems actually to have originated earlier: in 1914 Achille Segard, mentioning ‘La décoration pour M. Claude Anet’, stated that ‘It dates from 1898. It is composed of three panels’ (note 45), and further lists the paintings: ‘1898 – Three decorative panels executed for M. Claude Anet’ (note 46). Numerous subsequent publications, such as Chastel’s, perpetuated the mistake (note 47). In 1945 Vuillard’s early biographer Claude Roger-Marx refers to the ‘Vasouy’ panel as ‘exhibited around 1905 and reworked in 1936’, when actually, the decorations exhibited at the 1905 Salon d’Automne (alongside the Vaquez panels) were the Relais ones (note 48). To make the matter worse, in his two books published respectively in 1945 and 1948, the divided panels were mistakenly reproduced in their pre-repainting and present state – but in each case paired with their wrong half (note 49). Consequently the confusion deepened, with Schopfer’s daughter describing the panels now in the National Gallery as representing ‘le jardin à Villeneuve sur Yonne’ (note 50), and more recently Bernard Dunstan (note 51) still identifying the ‘Vasouy’ panels as being the ‘Relais’ ones. The mistake was not rectified until the 1980s, by which time the panels had entered the National Gallery where, unsurprisingly, they were acquired and catalogued as the ‘Relais’ panels (note 52).
When the ‘Lunch’ was first proposed for acquisition by the dealer César de Hauke it was still rightly described as ‘Le Déjeuner dans un Jardin en Normandie’ (note 53), but it did not take long for the mix-up to creep in, and for the considered picture to be referred to as le ‘Déjeuner à Villeneuve-sur-Yonne’ (note 54). Nonetheless, the National Gallery is to be credited for having reunited the two panels, which had been dispersed in different private collections since 1938.Their purchase proved a slow and painstaking process, as more prestigious acquisitions took precedence – when the ‘Lunch’ was first considered in 1962, Philip Hendy, the Gallery’s Director, noted that ‘attention was diverted to [Cézanne’s] ‘Dans le Parc du Château Noir’, and when we next wanted to consider it, his ‘Grandes Baigneuses’ offered themselves’ (note 55). Yet, as the closest thing to a large Impressionist ‘plein-air’ scene that could be aimed for, ‘La Terrasse’ was regarded as highly desirable – ‘of real importance as an acquisition, being a portrayal of French life of a type painted by many of the most important later 19th C French painters, examples by the most prominent painters were no longer available’; it was stressed that ‘the picture would go far to fill the gap’ (note 56).
Evidently, the simultaneous availability of both panels on the market contributed to the Gallery’s decision to proceed: while the ‘Lunch’, then with a Paris dealer, was being considered for acquisition, the ‘Garden’ also came to the Gallery’s attention, for sale at Galerie Beyeler in Switzerland. Expert opinions were invoked as this ‘unique opportunity’ offered itself of reassembling what Claude Roger-Marx described as a ‘masterpiece never to be surpassed’ (note 57). In any case, funds were insufficient for the purchase of both panels. Providentially, in 1964 a British buyer was found for the ‘Garden’, Major D. Allnatt, who on the Gallery’s shrewd advice placed the panel on long-term loan to Trafalgar Square (note 58). This prompted the Trustees to complete the acquisition of the ‘Lunch’ in 1966. When Major Allnatt died four years later, the Gallery was able to purchase the ‘Garden’.
All through these long and arduous negotiations, the prospect of reuniting the pictures was considered with great excitement; the possibility of reassembling the panels was even briefly entertained. As soon as the ‘Lunch’ arrived in the Gallery in March 1966, rejoining the ‘Garden’ for the first time in 30 years, plans were made to ‘unframe both pictures and put them together, in order to see whether they could in fact be joined again’ (note 59). The idea was dropped; arguably, this would have been contrary to Vuillard’s intentions: the artist had turned the panels into two distinct compositions, each of them coherent single units, with the repainting ‘actually carried out with a view of parting them, indeed in order to enable them to be sold separately’ (note 60).
Immediately prior to acquisition, the ‘Lunch’ was examined by the National Gallery’s chief restorer and declared ‘in very good condition’ (note 61). Yet evidence of cracking and flaking was noted as early as 1973, and both panels have since posed ongoing conservation problems (note 62). As part of an exchange scheme, the paintings were sent in May 1997 to the Tate Gallery, where they remained until February 2010 – with the exception of a brief trip to the Royal Academy in early 2004, for the London showing of the most important Vuillard retrospective to date, to which they had been exceptionally loaned (note 63).
Despite the changes made to the format of the ‘Vasouy’ panels, and their obvious fragility, they have, overall, encountered little in the way of external intervention. The ‘Lunch’ was lined prior to entering the National Gallery Collection, no doubt to arrest the problems of paint flaking, and also received a thin coating of rabbit skin glue in order to ‘brighten the colours’ (note 64). A thin fluorescing coating can be seen in some cross-sections, and some drip marks are visible on the painting when the surface is examined with ultraviolet light. Only two very tiny non-original restorations were observed and these were readily distinguished from Vuillard’s own ‘repaints’. The ‘Garden’ panel appears relatively untouched, although some consolidation of the flaking has been undertaken in the past.
Like most of Vuillard’s decorative works, ‘La Terrasse’ was painted in distemper (glue-size), which accounts for much of its history of cracking and flaking: the brittle medium was thickly applied and reapplied, causing a poor adhesion between the paint layers. Experimental, and untested by time, this was nonetheless the technique Vuillard most favoured. He presumably made the large single panel in the studio on rue Nollet which he had been renting since March1900 (note 65). Sketches and drawings probably made ‘sur le motif ’, as well as photographs he took himself during that summer, indicate Vuillard’s degree of deliberation before he established the final composition (note 66). Surviving oil studies focus on the gravel path and its surrounding vegetation (note 67); these seem to have helped Vuillard to define the layout of the garden. Yet the artist relied instead on his photographs of the actual event when planning the figures and their spatial organisation around the table, borrowing isolated elements from several of these images and combining them in his final design (note 68).
Once a composition was established Vuillard would sometimes map it on a full-size sheet of paper in chalk before transferring it to the canvas in paint (note 69); yet there is no evidence of this procedure here (note 70). Vuillard was naturally drawn to using distemper; it was a technique he had employed extensively while painting stage sets, where large areas had to be covered quickly, and where non-reflective surfaces were needed. Such requirements did not apply to domestic decorations intended for private homes, yet Vuillard remained attracted to the aesthetic qualities of the medium, its ‘magical flashes of light with astoundingly matt colours’ (note 71). Aware that oil painting might be more permanent, but willing to ‘remove the easy brilliance of the colours’ (note 72), the artist was seeking matt effects while trying to preserve the painting’s brightness and luminosity – a ‘subdued inner glow’ (note 73), a ‘muffled resonance which surprises’ (note 74).
In the distemper method (‘peinture à la colle’) as used by Vuillard in 1901, dry pigments are bound by the artist in a glue medium which must be applied warm. Analysis has confirmed the presence of a proteinaceous binder for both painting campaigns on the ‘Vasouy’ panels (note 75). Vuillard’s distemper technique was closely described by Jacques Salomon, the artist’s nephew by marriage and a fellow painter, and, though his account is not always accurate, this gives some insight into Vuillard’s working methods and resulting visual effects. Salomon states that the artist did not employ the ‘colle de peau’ (hide glue) used for theatre decorations, as this medium only lasted a few days before going off, preferring instead to employ ‘Tottin’, a rabbit skin glue (note 76), which could be bought from the druggist in sheets (note 77).
Due to the demands of industrial manufacture for a more flexible animal glue, numerous factories producing rabbit skin glue, such as La Maison Chardin, were established during the nineteenth century just outside Paris on the Seine (note 78). Totin Frères was perhaps the best-known manufacturer of rabbit skin glue, and the sheets of glue stamped with their logo were known as ‘colle de peau Totin’ (note 79). The Paris ‘Annuaire Almanach’ for 1901 includes a large advert for Totin Frères under the listing ‘colle sèche de peau de lapin’ (dry rabbit skin glue), stating that the company had taken over from Bonnefoy et Winder and giving a Montreuil-sous-Bois (Seine) address (note 80). Recently, researchers at Lille University analysed a 100-year-old sample of gilder’s glue (‘colle à doreurs’) from ‘Maison Totin-Frères’ using proteomics and identified peptides of rabbit collagen, seeming to confirm that it is a rabbit skin glue (note 81). ‘Totin’ is said to be more transparent and matt than other skin glues and thus suited Vuillard’s aesthetic (note 82).
Salomon refers to Vuillard’s distemper technique as ‘sa cuisine’, as ‘it demanded an elaborate apparatus: a spirit lamp and electric stove, quantities of pots and pans, a host of boxes and bags of powder’ (note 83) The sheets of glue were soaked for twelve hours, then taken out and put in a water bath with four or five times their volume of water. For large works, Vuillard would mix his colours in advance, using many pots for all the mixtures and keeping them constantly in a water bath to prevent the glue from setting (note 84). Distemper appears to have been a laborious and difficult medium to master, particularly since it was difficult to judge the final colour as it would change considerably upon drying and was affected by the colour of any underlayer (note 85). Consequently, if Vuillard ran out of a particular hue he often tested the new mixture on pieces of paper, which he dried over an electric fire to check the colour match (note 86).
Cross-sections, as well as surface examination of the paintings, show numerous voids, probably air bubbles, which may have been caused by vigorous stirring and mixing of the paints. The impasted picture surface is lumpy, and sampling reveals agglomerations of pigment which have not been properly dispersed in the abruptly mixed paints. The red lake pigments seem to be particularly prone to forming agglomerations. A collection of pigments gathered from Vuillard’s studio after his death have been in the National Gallery since the early 1970s.These were stored in pepper pots, and it is easy to imagine the artist carefully shaking in a little pigment from these pots to adjust his colour (note 87). Sampling shows how very complex these pigment mixtures were: individual layers often contain eight to ten different pigments, indicating how the artist attempted to compose subtle colours. As observed by Dunstan, Vuillard’s early distemper paintings were quite thinly painted with relatively smooth surfaces which did not show much evidence of overpainting (note 88). This can be seen in some of the areas which appear to have escaped much of the reworking of 1935. Here the underbound paint almost acts like a stain on the coarse, unprimed canvas (note 89). No doubt the absorbency and texture of the support was critical in achieving the tapestry-like effects prized by the artist in this period of his career (note 90).
When Vuillard was asked which pigments he used for his distemper paintings he said: ‘The worst… ones which I bought at the corner druggist: ‘les verts anglais’, ‘le bleu charron’ and ‘le blanc de Meudon’, in cakes’ (note 91). This off-hand statement appears to be completely at odds with the findings from ‘La Terrasse at Vasouy’. Rather than using cheap, low-quality pigments Vuillard appears to have employed expensive and cutting-edge colours which were unlikely to have been available at the local druggist. For example, cobalt blue and cobalt violet, both used by Vuillard, were, apart from one of the carmine lake pigments, the most expensive dry colours listed in the Lefranc catalogue (note 92). Also, Vuillard’s primary yellow pigment, cadmium yellow, was quite expensive at 14 francs for 100 grams – relative to 2 francs 65 for chrome yellow and 3 francs 50 for Naples yellow. In fact, Vuillard specifically records in several diary entries buying colours from the Lefranc shop (note 93). It is interesting to note that most of the pigments listed under the section ‘couleurs sèches’ were sold as fine powders, although a few were available as cakes, such as lead white, which was sold ‘en pain’, possibly due to its known toxicity. The distemper medium would seem to be challenging enough to control without the additional difficulty of breaking up cakes of pigment, and it is more likely that he used powdered colours, especially as he seems to have used pepper pots to store them.
It is not completely certain which pigment Vuillard’s term ‘English green’ describes. Fiedler and Bayard’s study of contemporary literature showed that it was sometimes used to denote both emerald and Scheele’s green (note 94). No copper arsenite pigments were identified on the ‘Lunch’, and only Rinman’s or cobalt green and viridian were observed in samples (note 95). It is perhaps possible that there could be some confusion between Rinman’s green and emerald green as they appear similar in cross-section, despite early comments about the poor tinting strength of cobalt green. However, under ‘couleurs sèches’ the 1928 Lefranc catalogue separately lists ‘vert anglais’ (six types), ‘vert de cobalt’, ‘vert de Scheele’, ‘vert Véronèse’ and ‘vert émeraude’, amongst others. ‘Vert de cobalt’, along with ‘vert émeraude’ (viridian), is the most expensive green listed. Church, author of the 1890 publication ‘The Chemistry of Paints and Painting’, states that ‘when properly prepared, cobalt green is a pigment of great beauty and power’ and that it ‘is, in fact, one of the too-rare pigments which is at once chemically and artistically perfect’ (note 96). However, ‘one sample of deep transparent green which I obtained from a Paris colour-manufacturer contained both viridian and ultramarine, added to enrich the colour of the cobalt green which formed the basis of the pigment’ (note 97). It is interesting to note that cobalt green on ‘La Terrasse at Vasouy’ was usually identified in conjunction with synthetic ultramarine and a little viridian, which might suggest a manufacturer’s mixture. Recent analysis of a set of Lefranc oil paints from the early 1930s showed the ‘vert anglais no. 5’ to contain a mixture of Prussian blue and yellow ochre. This combination has been found on the panels, but on the whole Vuillard appears to have relied on green pigments rather than mixed greens (note 98).
‘Bleu charron’ seems, traditionally, to refer to a colour based on woad dye, but has more recently been associated with Prussian blue mixed with barium sulphate (note 99). Prussian blue was identified along with synthetic ultramarine in the mixed green of the standing man’s waistcoat at the extreme right of ‘La Terrasse’, and this seemed to be associated with chlorine, suggesting the pigment was prepared with the ‘indirect method’ (note 100). There was a good deal of barium sulphate found in this sample, but this was present in quite large agglomerations and associated with zinc, suggesting the use of lithopone rather than an extender for Prussian blue (note 101). Lefranc lists both ‘bleu de Prusse’ and ‘bleu mineral’ in the 1928 catalogue of dry pigments; the latter name usually suggests additions of barium sulphate and kaolin to the Prussian blue pigment (note 102).
‘Blanc de Meudon’ is calcium carbonate. This could act as a white pigment in a glue medium but does not appear to have been used as such in ‘La Terrasse at Vasouy’. Calcium carbonate was detected in some paint layers but is likely to have been added as an extender, probably by the pigment manufacturer, or as a natural contaminant of some of the earth colours used. Rather, Vuillard seems to have used either zinc white, lithopone or titanium white. The newly discovered ‘oxyde de titane non chimiquement pur’ (probably anatase with barium sulphate) was listed as an oil tube colour in the 1928 Lefranc catalogue and priced at 9 francs for a No. 10 tube, the same as the better quality lead white ‘blanc d’argent’, whereas zinc white was a little cheaper at 7 francs 80. Titanium white was not listed in the dry pigments section for this year.
Vuillard avoided lead white in both painting campaigns. This may have been due to its poisonous nature – he also appears to have avoided the arsenic containing emerald green – or the pigment’s vulnerability to blackening when not used in an oil medium. On balance, however, it is probably more likely that the artist preferred the purer white colour of zinc and titanium pigments over the warmer tint of lead white.
Although vermilion has been identified on the ‘Terrace’ panels, Vuillard seems to have relied primarily on strongly coloured red earth and red lakes for his red pigments, and vermilion is found only as a minor addition to paint mixtures. This might be due to the difficult working properties of vermilion in a glue medium, with its tendency for poor wetting and, again, its toxicity. At least two types of red lake were observed in the samples: one dark red without marked fluorescence and the other a bright red with a strong orange fluorescence under UV illumination. High-performance liquid chromatographic (HPLC) analysis of the highly fluorescent red lake identified a pseudopurpurin-rich madder, probably made from a derivative of natural madder such as Kopp’s purpurin (note 103). Analysis of the darker red lake using attenuated total reflectance – Fourier transform infrared (ATR–FTIR) spectroscopy suggests that this is based on alizarin, available in synthetic form from the later nineteenth century and typified by alizarin crimson (note 104). Peaks for aluminium and sulphur were observed by SEM–EDX in both the lake pigments, suggesting that, in each case, the substrate may have been a type of light hydrated alumina (a basic aluminium sulphate) very commonly identified in late nineteenth- and twentieth-century French lake pigments (note 105). SEM–EDX analysis of the substrate of the natural madder pigment also revealed the presence of zinc in slightly higher concentration than in the sample as a whole. This probably indicates the use of some zinc white as an extender in the pigment to give a light, but bright colour. The same analysis also revealed the presence of calcium in the dark red lake. A calcium salt is typically added during the manufacture of alizarin red pigments.
The 1928 Lefranc catalogue lists a range of dry madder pigments, separating out those based on synthetic alizarin from ‘laques de garance naturelle’, prepared from the madder root. The synthetic lakes include blue, orange and yellow alizarin lakes, but also the more conventional scarlet and crimson. The natural madder lakes were separated into ranges for oil and for watercolour, a greater number of hues of rose colour being available for oil than for watercolour. Interestingly, standard alizarin-based lakes at 9 francs 50 centimes for 100 grams were nearly three times the price of the natural madder lakes at 3 francs 75 centimes.
Three different types of cadmium yellow were identified in sampling: a light, a dark and a very pale pigment with a strong orange-red fluorescence which was associated with zinc sulphate (note 106). Six shades of cadmium yellow were listed in the dry pigments section of the Lefranc catalogue of 1928.
Vuillard seems therefore to have used far more refined pigments than he pretended to when painting the ‘Terrasse’; technical analysis reveals the complexity of his palette, while an invaluable description of the original panel’s colour harmonies (which contemporary photographs fail to convey) can be found in Achille Segard’s book, ‘Peintres d’aujourd’hui: les Décorateurs’. Writing in 1914, not only did Segard offer a precise account of the 1901 composition – off-centre, with the almost life-size figures grouped on one side – but also, crucially, he dwelt on the palette used by Vuillard, giving a fair idea of what the large panel must have looked like. The table is ‘covered with a grey and red large checked tablecloth’; one can see ‘lots of greys and browns in the clothes, several yellow straw hats, a green chair’. ‘To our left a wide winding path, grey-pink’; the garden forms ‘a vast greenery enriched with the yellows and brown-pinks of certain types of trees’ (note 107).
This colour scheme can still be observed on the right tacking edge of the ‘Garden’, where the large canvas was cut and a few centimetres of the composition folded around the stretcher. Three red dots which were part of the 1901 composition, belonging to the flowerbed lining the path, are clearly visible. As confirmed by X-radiography, this flowerbed was originally higher (at about feet height of the lady in a red skirt), and looked neater and denser, with more intricate brushstrokes. Vuillard experimented with their aspect and direction, in order to convey variations of texture: dots for the flowerbed, horizontal strokes for the path and vertical for the grass. The photographs and X-rays show his complex and closeknit brushwork, with small, overlapping daubs of paint, ‘très en tapisserie’, according to Segard. In the mid-to-late 1890s, the parallel with textile was often made by Vuillard’s critics, his paintings praised for their ‘tapestry effects’ (note 108). The figures look almost woven into the canvas; with their matte surfaces and variations of hues, they display tactile quality and textural richness.
Segard adds that ‘brushstrokes are applied flat with no care for blending or modelling’, and deems drawing ‘inexistent’, but describes the strong patterns which enliven the picture: ‘a checked table carpet or a stripy cloth [...] completing the colour harmony’ (note 109). Whether fabric or foliage – the ‘little-individualised tree species’ – for Vuillard everything becomes pattern; even faces. In ‘La Terrasse’ these are hardly characterised at all: ‘Faces hardly exist. They are coloured marks’, yet they are definitely portraits; Segard concedes that ‘one recognises certain personalities, notably M. Tristan Bernard, cut off by the frame’, standing in profile on the right (note 110).
The playwright and humorist Tristan Bernard (1866–1947), a friend of Vuillard’s since their time at the Lycée Condorcet, was a contributor to ‘La revue blanche’, of which this lunch scene seems to be an informal gathering: all sitters identified in the 1901 panel either belong to that circle or have a connection with it (note 111). Next to Bernard, from right to left, stands Louis Schopfer, brother of Jean, with his wife ‘Bob’, seated and turning her back to us. In the early 1890s Louis Schopfer shared a small flat with his brother on Quai Voltaire, prior to becoming a banker; there, three times a week, they would entertain their Nabis artist friends (note 112). Right behind Mrs Louis Schopfer’s profile are Lise and Léon Blum, barely visible, half emerging from Bob’s elaborate hat. Now remembered as a major historical figure – he was to become France’s first Socialist Prime Minister in the late 1930s (note 113) – Blum was in the 1890s a young engaged intellectual who used ‘La revue blanche’ as a tribune for his anarchist ideas (note 114). Appointed as the magazine’s managing editor from April 1894, he also contributed non-political texts such as, in 1899, a review of Jean Schopfer’s first book (note 115). Further to the left, at the table, sitting next to Lise Blum, Misia Natanson (1872–1950) turns away to chat to the dramatist Romain Coolus. Misia Godebska, a highly gifted pianist of Polish descent, married Thadée Natanson in 1893. Flirtatious and talented, she soon became the ‘dame de beauté des Nabis’, and a recurrent motif in Vuillard’s interiors as well as his decorations – gracing, among others, the left panel of the ‘Relais’. The ‘sirène et muse’ (note 116), nicknamed the ‘Pompadour of “La revue blanche”’, was also to inspire some famous posters and portraits by Bonnard, Toulouse-Lautrec and Renoir. Her lunch companion Romain Coolus (1868–1952), another Lycée Condorcet alumnus and later a celebrated writer, was himself part of the ‘revue blanche’ clique (note 117). Jean Schopfer himself stands at the back of the table, behind an unidentified female guest, possibly Marcelle Aron, Tristan Bernard’s wife (note 118); finally, to the left, an almost life-size Alice Schopfer is being led away into the garden by Pierre Bonnard (1867–1947). Bonnard, whose tall silhouette was actually borrowed from a photograph taken that summer, not at La Terrasse but in nearby Criqueboeuf, was also one of Vuillard’s closest friends, as well as, later, Schopfer’s collaborator (note 119). Within the adjoining ‘jardin’, two young children are playing on the grass: Anne (Jean and Alice Schopfer’s daughter) possibly based on a photograph, wearing what looks like a ‘chapeau Jean Bart’ (note 120); and her young cousin Hélène (Louis and Bob’s baby daughter, also called ‘Biche’), behind a wheelbarrow (note 121). The little girls are are being looked after by a maid, or nanny (note 122), who sits in what looked like a wicker chair, as can be observed in the photographs of the 1901 panel and on the X-rays.
By 1901 ‘La revue blanche’ was in decline, and this milieu was about to disintegrate (note 123); marriages would soon break up, and friendships dissolve – the flamboyant Misia, the object of Vuillard’s infatuation during the Nabis years, is here a discreet presence, relegated to the back of the table. Yet the 1901 picture exuded a sense of bohemian camaraderie. This dazzling reunion was the re-creation of a very intellectual and Parisian social circle; since it was painted in the studio, it was also a Paris re-creation of a countryside scene, executed in a particularly non-naturalistic style. The dense brushwork would have made it look somewhat confined, like a Vuillard interior: an airless ‘plein-air’ scene. Vuillard’s gardens are often described as ‘intimités en plein air’ or ‘intimités collectives’ (note 124); and Claude Roger-Marx wrote that for Vuillard, ‘everything is an interior, gardens, theatre sets, Parisian squares with their furniture of trees, benches, street lamps.’ (note 125); the Vasouy garden was no exception.
This Paris-made picture of a Paris crowd was destined for a Paris flat; the Schopfer interior on Avenue Victor Hugo reflected the couple’s cultured and eclectic tastes: modern paintings, Louis XVI furniture, exotic rugs and cushions. The grand salon measured 3.23m floor to ceiling, with skirting all around the room, of about a metre high – as can be deduced from the 1901 photographs, which left Vuillard with a surface of about 2.2 m in height by 3.8 m in width to cover (note 126). The artist actually enjoyed the challenge of working within particular limitations, ‘drawing all kind of stimulants from the constraints imposed by a specific location – bedroom, salon, library, theatre foyer’. He ‘invested in these commissions the best that he had seen, felt or dreamt’(note 127) and excelled at creating decorations in relation to the room’s architecture. His distemper technique helped unify the whole space – the panels showed the same matte surface as the plaster skirtings and ceilings: ‘The link with the architectural setting is immediately established thanks to the appearance of the medium, which is reminiscent of plaster or stone.’ (note 128)
Rather than looking back to the Renaissance mural, with its sense of perspective, its illusionistic depth and games of ‘trompe-l’oeil’ (note 129), the Schopfer decorations were reminiscent of the work of Pierre Puvis de Chavannes (1824–1898). Puvis’s death in 1898 had attracted renewed attention for his austere ‘murailles’ (stonework), revered by two generations of artists for their matt effect, even light, and soft and muted tones (note 130). Vuillard borrowed from the master his ‘minor tonalities, in a flat ensemble which respects the wall like a thick fabric, in a totally new way’ (note 131). Not only do the Vuillard panels show the aspect of a woven textile, they also reproduce the planar structure of a tapestry, composed by zones of contrasted patterns, creating separate fields on the immense surface.
Yet unlike tapestries or eighteenth-century ornamental ‘verdures’, the Schopfer panels did not aim to imitate a painting (note 132), but formed a ‘grande décoration’ – inseparable from the room for which they were intended, and inseparable from each other. The ‘Terrasse’ was conceived and painted as a complement to the ‘Relais’ pendants, which it was supposed to join, in a harmonious, unified ensemble. The three panels were similar in height (note 133); they portrayed, in part, the same people (Bonnard, Misia) and showed an identical composition: the edge of the house visible on one side; a large seated figure, turning away, at the far right, and left, closing the image, a sandy, sinuous path, tall trees, banks of shrubs and small figures in the distance. They belonged to the same scheme, with the large panel echoing and balancing the pendants.
It has been plausibly suggested that the ‘Relais’ and ‘Terrasse’ panels would have hung next to each other, probably at right angles, separated only by a door into a small antechamber (note 134). Vuillard must have wanted to stress the continuity between the three works, in order to confer on the decor as much coherence as possible. For this reason, as seen in the detail from the large panel, the artist originally gave the Vasouy house the same plain red brick edge as the one visible on the Relais external walls, in the Schopfer left pendant, ‘Woman Reading on a Bench’. Vuillard’s 1901 photographs of the Vasouy lunch scene clearly show a distinctive brick pattern alternating three layers of dark red brick, and three layers of pale bricks, which Vuillard chose not to replicate in the painting – for the sake of unity with the ‘Relais’ pendants. This idea of rendering uniform the three decorative panels destined to hang in the same room must have been intentional from the outset, as attested by one of Vuillard’s oil studies. Linked with the 1901 panel, this study was presumably an early thought, with a house wall – just visible on the left – presenting a distinctive plain brick edge.
However the panels now show the ‘alternating’, stripy brickwork, faithful to the appearance of the house; when Vuillard divided and reworked the composition in 1935, he reverted to his photographs of his 1901 ‘villégiature’ at Vasouy, basing his alterations on what he saw on the snapshots. There was no need to harmonise it with the ‘Relais’ panels any more; ‘La Terrasse’ had long left the salon for which it had been painted in the first place, and had long been separated from the pendants it was originally meant to complement. Clarisse Schopfer, who had inherited the large panel on Schopfer’s death in 1931, had to leave the couple’s flat on the rue du Bac during the year 1935 (note 135) and asked Vuillard to divide the painting. An entry in the artist’s diary indicates that he was consulted in the autumn of 1934, and that the couple’s daughter, Leïla, might have also been involved in the discussion: ‘... question of the large C.A. panel to cut; Lelia [sic], decision’ (note 136).
It was suggested that the division had actually been Vuillard’s idea, to help Clarisse accommodate ‘La Terrasse’ in her new, smaller home, but it is more likely that mother and daughter had intended to part with the work, and had it cut into two in order to make it more saleable (note 137). Having seen many of his decorative ensembles separated and sold individually, Vuillard would not have found the request disconcerting. The artist returned to earlier work late in life on numerous occasions, reworking his paintings considerably, completing them or changing their appearance altogether (note 138).
The artist was probably quite keen to embark on this particular project. As the Schopfers’ close friend for more than three decades (note 139) (even Schopfer’s brother remained an acquaintance) (note 140), he had had numerous occasions to inspect his large panel in their home. Often invited to lunch at the Anets’, he had grown increasingly dissatisfied with it, writing in 1919 after such a visit: ‘Good impression of Roussel’s paintings… bad effect of my panel next to Bonnard’s portrait; my ignorance of drawing’ (note 141); and again in 1926: ‘went directly to lunch at Claude Anet’s, magnificent Bonnards, my gloomy old panel…’ (note 142) Clarisse’s request must have been a good opportunity for Vuillard to return to his picture, improving it (Bonnard was missing a foot in the 1901 panel) and bringing it up to his new standards of work.
This would prove a long process, of which Vuillard’s diary provides an invaluable day-by-day account. Seemingly his only significant project in the year 1935, he worked on it relentlessly, from at least February until late October (note 143). Photographs taken in Vuillard’s studio at place Vintimille while the work was still in progress show the divided panels placed on easels (note 144). They seem to record a late stage of the reworking, with the alterations already in place, indicating that Vuillard was already well into the project.
Faced with this kind of challenge, the artist would follow a familiar procedure, first looking for documents, souvenirs and sketches relating to the ‘gloomy old panel’, and getting ready. On 16 February he writes: ‘Looked out documents for Schopfer panels, sketches and snapshots [...] At home, looked out old sketches, cleaned palette.’ (note 145) The next day seems to have been spent doing the same (note 146), searching through well-maintained files and archives: for his main projects Vuillard would keep a ‘considerable documentation in the sketches and pastels, sometimes offered to the client, or carefully placed in dossiers, with the secret hope of possibly reworking the painting’ (note 147) As he remarked himself, the ‘great value of these first studies after nature’ lay in ‘the ingenuity of observation’ (note 148) – in their freshness and spontaneity.
As part of his working process he also made drawings, frequently based on the photographs he had found, thus testing new ideas; on 13 April, ‘Tea on my own, worked Anet; drew on dark paper after photograph, black and white [...] sun, back home, worked alone Anet’ (note 149); again on 22 June: ‘Drew studies of foliage, first on paper, struggle to copy the form in the photograph’ (note 150). Vuillard must have had his photographs at hand as he was working on the divided canvases, going back and forth from canvas to photographs, treating these quite casually in the mess of his studio – as attested by the numerous smears of paint covering his 1901 snapshots, which appear to correspond to the paint he used in the reworkings. These images were used as aides-mémoires, as a support to document and revive his memories (note 151). Vuillard’s greater dependency on black-and-white photographs results, in the 1935 panels, in new tonal arrangements, sharper contrasts of light and dark and better-defined forms. The trees and bushes in the ‘Garden’ are scrupulously based on Vuillard’s 1901 photographs, as is the background of the ‘Lunch’, with its interstices of sky between leaves and branches. Dappled light pierces through what was, in the old panel, a dense and compact wall of foliage. The still life has been modified, with some elements added to match the table arrangement on the 1901 photograph, and the tablecloth now shows a smaller check pattern based on the same photograph.
Even though the addition of a wedge of grey paint in the right corner of the ‘Garden’ seems actually to reinforce the link between the divided works, the panels had by then become two independent compositions. Delivered from the constraints of having to make them match with any pendants or interior, Vuillard could allow himself to make full, faithful use of his 1901 photographs. These must have come in handy when trying to solve his new compositional problems. The new, smaller format of the ‘Lunch’ panel imposed a tighter check pattern on the tablecloth, less overpowering; the change of design to vertical also forced Vuillard to modify the garden layout quite extensively. Formerly a flat, horizontal strip, the gravel path has been reworked and given a firmer outline, now cutting the picture diagonally, zigzagging towards the lower right. Rather than preserving the original garden, structured in parallel bands stressing its planarity, as can be observed in Vuillard’s ‘Jardins Publics’ (note 152), Vuillard recreated the hollows and recesses along the path that can be seen in his own photograph. The result is a more conventional image, typical of Vuillard’s late style, with an illusion of depth which was deliberately absent from the 1901 panel.
Once, in June, Clarisse Schopfer visited Vuillard in the studio to check on progress, and declared herself ‘happy’ (note 153); Vuillard does not report any more specific comment. With a loose brief, if any, the artist was probably given free rein to do whatever he had in mind. Yet his diary reveals that the project was not straightforward – a painstaking enterprise for the ageing artist, who month after month struggled to bring it to completion, with an increasing sense of boredom. In April he ‘tries with difficulty to return to Cl. Anet… Jacques R stays without daring to speak to me; impatience within myself to get back to work or at least being able to think quietly […] Returns to work, struggle […]’ (note 154); in May he feels ‘once more dazed, attempt to work, difficult’ (note 155).
Yet as Vuillard was making progress, moments of lassitude alternated with bouts of satisfaction: in April he describes the ‘...rather good effect panels C.A.’ (note 156); and writes in June: ‘back to work on panel C. Anet with good spirit’ (note 157). According to the diary, the artist started with the left half, from March until about May, focusing on the garden first, its foliage and flowers, the sky and trees of the path; then adding the tall female figure in white, Lucy Hessel in a ‘robe blanche’, based on a photograph (note 158). The picture of the ‘Garden’ on an easel in Vuillard’s studio shows that the artist later returned to this figure, giving her face more precise features, and modifying the dress’s left sleeve and décolletage: the loose, toga-like smock was transformed into a more fitted gown, with a marked waist.
By May he had tackled the ‘Lunch’, was ‘struggling to rework [the] cushion. Foliage effect of the previous day, effect on the tablecloth’ (note 159), ‘reworking [the] background of a panel C. Anet, dog and leg’ (note 160). A dog was already present in the 1901 panel, yet closer to the right, its muzzle only reaching the middle of the green chair where Bob is seated; Vuillard shifted it to the left. On 28 May he ‘reworks Bob’s hat’ (note 161), and on 13 June retouched the background foliage (note 162). Then followed what looks like a long summer break, and not until late September did he resume work on the panels (note 163). The pace then increased, with the artist spending ‘grandes journées’ (long days) on them, focusing on the sky, the white dress and the ‘greenery at left’ (note 164). Stressful, intensive days of labour were barely interrupted by a visit to Madame Anet on 9 October (note 165), or a quick nap; not even the artist’s impending flu (note 166). Vuillard was evidently working towards a deadline, the panels’ ‘imminent departure’ (note 167); this happened on 31 October 1935. On that day the ‘last touches’ are applied – on that occasion presumably both ‘Garden’ and ‘Lunch’ were signed again, in brown (note 168) – and the panels are removed and taken to Clarisse’s, place de Breteuil (note 169).
As in 1901, in 1935 Vuillard had used distemper or ‘peinture à la colle’ for reworking the paintings. In all their precision, Vuillard’s diary entries make no reference to his ‘cuisine’ around the panels; we know he used a palette (note 170), and a note for 13 May 1935 mentions him buying his colours from Lefranc (note 171); no more detail is given. Chastel wrote of Vuillard that ‘he will never produce paintings succulent to the touch and smooth under the fingers’ (note 172), which is particularly true of the 1935 picture surface.
In contrast to the initial composition, the distemper paint of the  second campaign is bodied and thickly applied, creating its own texture and concealing the canvas weave. Some areas of paint are quite gelatinous in appearance, suggesting that Vuillard had heated the glue for a long time. Jacques Salomon refers to this in his description of Vuillard at work, stating that ‘often, with the evaporation of water, the glue will thicken until it becomes sticky, but, carried away by his work, he will use it as is, welcoming some of this matter which became, he said, rock hard, even cracking, applied over layers that had not had time to dry.’ (note 173) According to Salomon, Vuillard sometimes removed areas of the composition when reworking by soaking the paint in hot water and scraping it down with a knife (note 174). No particular evidence of this was found on the ‘Vasouy’ panels, and the X-radiograph revealed that many of the elements from the 1901 composition were existent beneath the 1935 reworking. There are areas in the X-radiograph which are thin and difficult to interpret, but these could equally be explained by the absence of X-ray-dense pigments, particularly lead white, and the thinness of the initial paint application rather than scraping down. In fact, there are numerous areas where it appears that the artist has had to ‘retouch’ his work due to flaking and loss. This is quite evident in areas of Tristan Bernard’s waistcoat, at the extreme right of the ‘Lunch’ panel, where paint strokes can be seen running over impasted areas and into troughs which show as loss on the X-radiograph. We know from Vuillard’s diary for 18 October 1935 that this particular figure gave him trouble: ‘...day spent on my own on panels C. A. [...] yellow tones, the figure of Tristan’ (note 175). Areas in the foliage background can also be seen to have been ‘restored’ by the painter, and here Vuillard appears to have used a palette knife in order to match the thickness of the surrounding paint.
Since no distinctive interface or isolating layer was identified which would separate out the two campaigns, Vuillard’s later use of the newly discovered titanium white pigment proved invaluable in assessing the extent of his 1935 alterations. Titanium dioxide was first developed in 1908 and produced on a small scale in 1915. By 1916 composite anatase pigments with barium sulphate were commercially available in Norway and the United States, although these were limited until late 1918 and not available in France before 1922 (note 176). A good quality pure anatase pigment was first commercially produced in France in 1923 (note 177) and titanium white is listed as an oil tube paint in the 1928 Lefranc catalogue. Raman analysis undertaken on samples containing titanium white identified the pigment as the anatase form (note 178), helping to give us a ‘terminus post quem’ date of about 1923 for any layers containing titanium white. By contrast, zinc white, often observed due to its distinct fluorescence in ultraviolet light, was used in the lower layers of all samples taken, thus confirming a distinct change in the choice of whites between the 1901 and 1935 campaign. It is interesting to note that the titanium white was usually found to be associated with barium sulphate.
Otherwise, Vuillard appears to have used broadly the same palette except perhaps increasing his use of cobalt blue and cobalt violet pigments (note 179), and tending to prefer viridian over cobalt green. Naples yellow appears to have been used only in the first composition and is replaced with a pale cadmium yellow in the 1935 campaign. A little strontium yellow was also identified in the 1901 painting, whereas only cadmium yellows and yellow ochre were identified in the later reworking.
The visible encrusting and heavy impasto on most of the picture surface could be justified by the superimposition of several layers of paint, or by Vuillard’s use of new methods – employing the same materials in a new way, credited to a new manner and different handling. The painting’s tapestry effect is now lost in favour of broader, more dynamic brushwork, with longer strokes, thick or fluid, and a more ‘Impressionist’ style, as seen in the vegetation. Vuillard applied large dabs of rich blue paint to the tree trunk and plants in the foreground of the ‘Garden’, and the same blue to the background foliage in the ‘Lunch’ – maybe the ‘feuilles arabesques’ mentioned in the diary (note 180). In both panels, large square strokes highlight the voids: Vuillard opens up the space to allow more sky to show through the foliage.
The Impressionist-inspired brushwork and ‘muffled tension of light, the golden yellow contrejours and the numerous reflections’ (note 181) mark his return to a previous mode of realism. A diary entry in May 1935 mentions Monet, whose influence is manifest in the way Vuillard captures the effect of dappled daylight and shadows on foliage and fabrics (note 182). A parallel can be made between the Schopfer panels and Monet’s ‘Women in the Garden’ of 1867 (note 183): ‘Monet’s pattern of striped and spotted costumes, the elegant arabesques described by figures, skirts and foliage’ find here a ‘flattened, airless echo’, in the opinion of Andrew Carnuff Ritchie (note 184). Claude Roger-Marx observes in the ‘Lunch’ the same proximity, ‘by the execution and the subject, of certain Monets, and even more, certain Renoirs, with its lightness and liveliness. […] The filiation with the Impressionists is here asserted more directly. The seated woman, her back turned, has the cheeky grace of a Renoir; the standing couple […] reminds of the ‘Danse à la Campagne’, whereas the setting itself is close to Monet’s large outdoor scenes from the time of the ‘Déjeuner sur l’Herbe’ in the museum at Frankfurt.’ (note 185)
Significantly, Vuillard’s diary also does not mention any of his alterations to the portraits contained in the painting. Most protagonists from the original panel were recast into the newly created ones, but with some degree of updating. Following the latest fashion, ‘Bob’ Schopfer and Misia are given new hats (the latter a new dress too), and Romain Coolus is wearing a cap which matches his new beige-coloured jacket. Other changes were made for less superficial reasons. Alice Schopfer, who in the 1901 panel leant backwards with a radiant expression, as if bursting with laughter, now adopts a more restrained pose, and a discreet smile; the Schopfer couple had divorced three decades earlier, Jean was dead and there was no need to give Alice a beaming grin.
Others have simply disappeared: Vuillard obliterated Léon Blum and his wife Lise, now absent from the divided panels. By 1935 the young literary critic had become a key, highly controversial socialist figure, about to make history the next year when elected France’s first Jewish prime minister. His presence in the painting now conferred on it a strong political undertone which, given the more conservative political allegiance of Vuillard’s late patrons and clients, could have made it more difficult to sell. Léon Blum was painted over with foliage, and his wife Lise (who had died in 1931) was also erased, replaced with a portrait of Lucy Hessel.
As the wife of the dealer Jos Hessel, cousin of the prominent Bernheims, Lucy Hessel had introduced Vuillard to a new social network, more conservative than the Natansons’: wealthy aristocrats, bankers and businessmen from the upper strata of the French bourgeoisie – ‘le milieu Hessel’ (note 186). Yet the 1935 paintings are not the portrait of this new circle, but a personal homage to Lucy herself. Though absent from the 1901 composition – she was then merely the mistress of the owner of the villa where a luncheon party was given, where Vuillard was still a newcomer – Lucy now features prominently in the divided paintings. Vuillard inserted her in the ‘Garden’ panel, where she stands tall and elegant wearing a flattering white dress, based on a 1901 photograph taken at Vasouy. In the ‘Lunch’, she sits, hand on her chin, a remote, pensive presence hidden behind Bob Schopfer’s large, fashionable hat. The dog in the lower left can be read as another reference to Lucy (note 187): Basto, the ‘stocky beige and white griffon’ of the Normand summers (note 188) has been replaced by another collie from the same ‘dog dynasty’ (note 189), and given a place of choice in the foreground. Nicknamed ‘le Dragon’, Lucy Hessel was haughty, opinionated, but kind and generous. After Vuillard lost his mother she had become the most important woman in his life, whatever the nature of their relationship; she remained his closest friend and lifelong ‘confidante’, as well as a muse and a model.
Another homage to Lucy may be found in the form of one of the small figures in the background of the ‘Garden’. There, Vuillard has reworked the most distant group of strollers on the path – painting over the originally heavy silhouette in dark suit to the left replacing it with another male figure wearing what looks like a képi and a blue uniform, unmistakably a soldier. Associated with the First World War, the ‘bleu horizon’ colour had been adopted by troops since early 1915. The shade of blue we see here is rather brighter than the pale ‘horizon’ shade, but in the distemper technique colours alter as they dry, particularly when applied on top of previous layers, as is the case here: Vuillard may have intended to use a strictly ‘horizon’ tone. This particular choice of colour reveals that this figure is likely to depict a First World War soldier; presumably a French ‘officier d’infanterie’ wearing a fitted blue ‘vareuse’ (jacket), a ‘culotte’ (trousers), what looks like ‘bandes molletières’ (puttees) (note 190) and faun-coloured ‘brodequins’ (laced boots). Also, the clearly distinguishable cane the soldier holds in his hand could refer to his military grade; or, as the attribute of the blind or visually impaired then as now, it may indicate that this serviceman was a ‘mutilé de guerre’ (disabled war veteran), having lost his sight in combat.
Vuillard’s extensive oeuvre contains hardly any depiction of military men, with the exception of a painting, the ‘Interrogation of the Prisoner’ (note 191), some related studies and two large pastel portraits of a soldier named Lucien Grandjean (?–1921), done in 1915 (note 192). Grandjean was a ‘mutilé de guerre’, blinded in the trenches with ypérite or mustard gas, whom Lucy Hessel had taken under her wing (note 193). His ‘marraine de guerre’ from 1915 (note 194), she later took him as the couple’s adoptive child. One of these two portraits shows Grandjean sitting (note 195), wearing a képi similar to the one worn by the soldier on the ‘Garden’ panel, and holding, likewise, his cane in front of him. This eccentric character, formerly an acrobat, was dear to both Lucy Hessel and Vuillard, who served as Grandjean’s witness at his wedding on 2 March 1916, and became his daughter’s godfather; this little girl, Lulu Grandjean, features in many of his paintings (note 196). Grandjean died in 1921; in 1935 Lulu was officially adopted by the Hessels. It is therefore not unthinkable that, when reworking ‘La Terrasse’ that same year, the artist might have wanted to complete his tribute to Lucy Hessel by adding the figure of this friend, who had connections with both her and Vuillard. It has indeed been suggested that the figure facing the soldier is the artist, depicting himself standing in profile, as he often did, with his dark hat, coat and distinctive red beard (note 197).
Whether these identifications are correct or not, the presence of a First World War soldier in the ‘Garden’ panel indicates that, despite Vuillard’s heavy reliance on photographs of the particular event – a summer day in the garden of La Terrasse in 1901 – he did not aim to recreate the actual luncheon scene, no more than he intended to situate his reworked pictures in the present. By 1935, the French army had imposed khaki as the regular colour for its troops (note 198), Vuillard’s beard had gone white and Lucy Hessel did not look so slender and youthful any more (note 199). The time depicted in the present paintings seems to lie somewhere in between; the picture of a particular memory, onto which a new layer of memories superimposed itself, following Bergson’s adage: ‘perceiving ends up being nothing more than an opportunity to remember’ (note 200). Stylistically, also, the paintings look back to the past with their late, regressive, much-criticised Impressionist style. Yet they have acquired a new, more reflexive dimension; ‘in the end, [Vuillard] painted – about what he saw – a kind of sentimental transfiguration of reality. He painted his pleasure of painting.’ (note 201)