National Gallery Technical Bulletin Volume 29, 2008
Two Versions of The Fountain of Love by Jean-Honoré Fragonard: A Comparative Study
Mark Leonard, Ashok Roy and Scott Schaefer
When Jean-Honoré Fragonard died on 22 August 1806, at the age of seventy-four, he had worked for the state in curatorial service as one of six original members of the Commission du Musée Central (appointed at the request of Jacques-Louis David) for what would evolve into the Musée du Louvre. Surprisingly, however, the artist’s career as a painter had, in fact, ended, for whatever reasons, almost fifteen years previously. Many probably assumed the artist had died long before, being thought, as he must have been, so much a part of the ancien régime, many of whose lights had been extinguished in various ways by the French Revolution. His obituary in the Journal de Paris on that following Monday reminded its readers that ‘the French school had lost a justly admired painter’. Coresus and Callirhoe (Paris, Louvre), The Fountain of Love (London, Wallace Collection, and Los Angeles, J. Paul Getty Museum; plates 1 and 2) and the Sacrifice of the Rose (several versions are known: Beverly Hills, Resnick Collection; France, private collection; and Buenos Aires, Museo Nacional de Arte Decorativo) were the three paintings singled out for specific mention. The public would have known the first – it was Fragonard’s morceau d’agrément, the painting required of students for acceptance into the Académie – as it was shown at the Salon of 1765 at the Louvre (no. 176). Acquired by the King a few days before the opening of the Salon, Fragonard’s painting was intended as a cartoon for a tapestry at the Gobelins but was never used as such. The other two subjects, painted slightly more than twenty years later, would have been known to the Journal’s reading public primarily through the enormous number of prints (by Regnault and Gérard respectively) made after the paintings, each of which existed in several versions, and all of which had been acquired by private collectors and graced the cabinets of their Parisian hôtels particuliers.
Fragonard began his career in a traditional manner, learning his craft at the Académie, winning the Prix de Rome, studying at the French Academy in Rome for several years (1756–61), and eventually exhibiting at the Salon of 1765 three paintings in three very different genres (better to announce his extraordinary facility): History (Coresus and Callirhoe), Landscape, and Genre (The Parents’ Absence turned to Account, St Petersburg, Hermitage). At the following Salon of 1767 three more genres were selected – Decoration (Group of Putti in the Sky, probably Paris, Louvre), Portrait (Head of an Old Man, possibly Muncie, Indiana, Ball State University Art Gallery), and a selection of finished drawings. Although criticised for not having exhibited at the Salon of 1769, Fragonard chose never to exhibit at the official Salon after 1767 (his work occasionally made an appearance when various printmakers showed their work there based on his compositions). From 1778 to 1786 he would sometimes show his work in public but only at the far less ‘official’ and relatively unregulated short-lived Salon de la Correspondance. For all intents and purposes, Fragonard found his own clients privately rather than relying on state commissions or exposure.
By the time Fragonard painted the various versions of the cabinet-sized ‘love’ pictures – Fountain of Love, Oath of Love, Sacrifice of the Rose and Invocation to Love, to cite only the four subjects – in the 1780s, he had found a ready clientele among the art-buying public. And that public was more than satisfied with the artist’s instinctual ability to move easily between very quickly and loosely painted pictures and those with a more laboured, highly polished surface (an appearance that in the nineteenth century would be called ‘licked’). With this ‘ambidextrous’ facility, the artist could appeal both to clients whose taste was for the sketch as well as to those who were more attracted to a porcelainlike finish. Fragonard’s ability to adapt his skills to these two different modes of painting was not new in the 1780s, however. He had already experimented with both styles as early as the late 1750s, probably while he was in Italy. One has only to compare the two versions of Lost Forfeit (New York, Metropolitan Museum of Art, and St Petersburg, Hermitage). Both are the same size, but painted in his two very different techniques. The various versions of the allegories of love are essentially experiments in these two modes of painting.
The Fountain of Love (note 1) at the Wallace Collection in London has long been recognised and admired as one of the artist’s major works. The reappearance of another version of the composition, now in the J. Paul Getty Museum, Los Angeles (see plates 1 and 2), suggested the project of comparing the two pictures side by side. This became possible in the spring of 2007, when both paintings were brought to the Conservation Studio at the National Gallery for study and treatment (plate 3) (note 2).
Both the Wallace Collection and Getty Museum versions of the picture can be traced back with certainty only to the mid-nineteenth century, when they were in England (for a fuller discussion of the provenance, see below) (note 3). The Wallace Collection picture was acquired by the 4th Marquess of Hertford in 1870, passed on his death to Sir Richard Wallace and formed part of Lady Wallace’s bequest to the Nation in 1897. The Getty version was sold at Christie’s in London in 1913, and soon afterwards entered an American private collection where it was lost to sight. It reappeared on the market in 1999, when it was acquired by the Getty Museum.
The opportunity to bring the two pictures together offered a fascinating glimpse into Fragonard’s ability to work in two seemingly contradictory modes. In spite of the fact that the compositions are virtually identical – minor differences are discussed below – the character of the surfaces in the two pictures is remarkably different. The brushwork in the Getty version is fluid and unrestrained, and as a result the scene takes on qualities that are at the same time both ethereal but dynamic. Many of the details, such as the hands of the putti at the left edge and the leafy trees in the background, are in effect personal artistic notations; Fragonard has worked out his ideas on this version in a rapid, uninhibited manner. By contrast, the Wallace Collection’s picture is painted in a more restrained, finished and distinctly neo-classical mode, a style more in keeping with wider developments in French painting in the 1780s. This difference is seen perhaps most clearly in the female figure: the soft impressions of the face in the Getty version have given way to a classical profile, the arms and legs have become fuller, and the fluttering brushwork of the drapery in the first version has been transformed into a controlled description of form that results in a strong sculptural presence.
In the midst of a verdant forest, a young man and woman eagerly rush forward, their feet just reaching the edge of the basin of a fountain. Putti frolic in the water and billowing spray, and one of them offers a cup of the magical waters for the young lovers to drink. The story of the ‘Garden of Love’, an allegory of the nature and progress of love that has its origins in the poetry of classical antiquity and the Middle Ages, centres on this Fountain of Love. The fountain brings forth the water in which Cupid dips his arrows or from which lovers drink and fall in love. During the eighteenth century, artists came to treat the Fountain of Love almost as a genre subject, with lovers in contemporary dress flirting in a garden around a decorative fountain. With this composition, Fragonard returned the allegory to its more classical origins and imbued it with the thrilling rush of those first beguiling moments of love.
The J. Paul Getty Museum Fountain of Love
After the Getty Museum acquired its version of the subject, the picture was cleaned and restored (it had been covered with an exceptionally darkened, discoloured and heavily toned varnish, as well as extensive re-paints). Before treatment, an X-radiograph was taken, which revealed a number of changes to the composition, most notably Fragonard’s original plan for the two heads at the centre of the composition (plate 4). In the artist’s original conception, the male head was placed closer to the female head, and looked back directly at her (FIG. 1). This also presented a slightly different story for the composition, as the proffered cup would have been intended for the female figure rather than her companion. This initial idea was re-worked by Fragonard: instead, he placed both heads in profile, and slightly further apart, with the result that the cup is now offered to the male figure.
An infrared reflectogram (FIG. 2) shows not only that the position of the head was shifted quite a bit to the left, but also that this shifting may have necessitated a slight reworking of the upper torso of the male figure, as the proper right shoulder appears to have been moved closer to the extended cup. The infrared reflectogram also indicates that the quiver (with its cluster of arrows) in the lower right corner (plate 5) was blocked in fully, but was then partially covered by the base of the fountain, which was painted just over the end of the case (FIG. 3). The presence of pentimenti such as these suggests that the Getty picture was almost certainly the first version of the composition. No similar changes are found in the Wallace Collection picture, indicating that by the time this version was painted Fragonard had finalised the composition. However, over and above the interpretation of pentimenti lies the more interesting question of the strikingly different character of the handling of the two pictures.
The surface of the Getty picture reveals Fragonard not only at the height of his powers, but working in a mode that could be characterised as personal and unrestrained. Because of the exuberant nature of the handling, the painting gives the sense that Fragonard was working purely for his own enjoyment. The painterly appearance of the surface of the Getty picture is the result of a highly fluid technique. The nature of this technique is reflected in a surprisingly complex and multi-layered paint structure as revealed in cross-sections under the microscope. Sampling a picture of this kind is challenging, given the nature of the surface and the lack of large losses or other suitable sample locations. It proved possible, however, to find a number of informative areas for investigation by sampling at the extreme edges of the composition. The same constraints applied to the Wallace Collection version.
The ground on the Getty canvas is a light biscuit-colour, presumably chosen as a means of imparting a sense of ‘rococo’ lightness and luminosity to the paint layers. In each paint cross-section from the picture, the ground layer appears slightly more pink at the top, and slightly more yellow at the bottom; this may be the result of a natural separation that takes place during drying rather than an indication of two separately applied layers. The presence of a variety of coloured pigment particles mixed into the bulk of lead white (note 4) in the ground (which give it a slight coloration) suggests that this layer may have been applied in Fragonard’s studio after the canvas had been stretched, as a commercially applied ground would more likely have been a purer white colour (note 5).
A cross-section sample from the dull mid-brown area at the extreme left edge, just above the fountain (plate 6), reveals a transparent layer that was applied directly on top of the ground (although it does not appear consistently in all of the sample sections). At first sight it has the appearance of containing a resinous material, perhaps a type of varnish, but it does not show a stronger fluorescence in ultraviolet illumination under the microscope when compared to the surrounding paint layers. It could well be similar to Fragonard’s paint medium. GC-MS analysis detected both heat-bodied linseed oil and a significant quantity of fir balsam, such as Strasbourg turpentine from the Abies alba species, in each of the paint samples from the Getty painting (note 6). It has not been possible to ascertain the distribution of these different materials within the layer structure observed in the cross-sections. However, the lack of distinct fluorescent layers may suggest that the fir balsam was included as an addition to the paint medium rather than applied as intermediate layers of varnish. One possible explanation of this unusual technique, for which there is some evidence in the contemporary literature, is that Fragonard applied a layer of the pure paint medium on to the priming in order to seal the surface and prevent his very medium-rich subsequent paint layers from sinking into the ground (note 7). A number of cross-sections show successive (and fairly numerous) layers of alternately opaque and translucent (or medium-rich) paint (see, for example, plate 7). Although there is the appearance of effortless facility in this type of approach, it should be noted that the painting may not have been executed all in one session; there may have been a need for drying time after the application of the more translucent layers.
The presence of alternating layers of medium-rich and more opaque pigment-laden paint, and the inclusion of fir balsam, may help to explain the unusually ‘wrinkled’ appearance of the surface of the picture (plate 8). The various paint layers would have dried at different rates, not only as the picture was being created, but over a period of time as well. Drying cracks appear to have formed below the upper surface, and in some areas the underlying layers appear to have pushed upwards through those cracks. The upper surface behaved in a somewhat elastic manner, stretching over the underlying craquelure (and not breaking to form a similar network of cracks), but still taking on a distorted appearance as a result of the related protrusions (note 8).
Although it may be tempting to describe the Getty painting as a preparatory sketch of sorts, that would be a misunderstanding of Fragonard’s work and methods. Not only is the picture more complex and more highly finished, from a purely technical standpoint, than Fragonard’s true ‘sketches’, it is also more complex – and more fully realised – in its underlying conception as well as in its finished form.
The Wallace Collection Fountain of Love
The Wallace Collection picture presents a nearly identical compositional scene, but in this case Fragonard chose a more restrained mode of expression that is in remarkable contrast to the highly personal nature of the Getty picture. The differences between the two pictures are most apparent through the comparison of particular details. When the draperies of the female figures are compared, for example, it can be seen that the loose handling of the surface in the Getty version (plate 9) has given way to a carefully controlled and constructed sense of form in the Wallace Collection version (plate 11) which resonates with references to classical sculpture. In the same way, the poignantly expressive quality of the face of the female figure in the Getty version (plate 10) has been transformed into an idealised classicising profile in the Wallace Collection picture that, once again, is intentionally reminiscent of ancient sculpture (plate 12).
As noted above, the surface appearance of the Getty picture reflects the materials and techniques revealed in paint cross-sections; similar investigation of the layer structure in the Wallace Collection version also reveals insights into the underlying handling and its impact upon the character of the final surface. The highly polished appearance of the Wallace Collection Fountain of Love came about through the use of a subtly different range of materials to those employed for the Getty picture, handled in a significantly different manner. GC-MS analysis of samples from the Wallace picture again found heat-bodied linseed oil as the principal binder. Although trace quantities of fir balsam were also detected, the proportions seemed too low to attribute any significance other than the use of this material in Fragonard’s studio. In the Getty painting, the amounts were significant.
Paint samples were limited to the perimeter of the painting, since no suitable sites were available within the composition. However, it was possible to co-ordinate the sample sites with equivalent locations in the Getty picture to allow for as much direct comparison as possible between the two.
Paint cross-sections showed that the canvas of the Wallace Collection picture had been prepared with a bright white ground, without any trace of the scattering of coloured pigment particles found in the priming on the Getty canvas (plate 13). The pure, brilliant reflective nature of this preparation suggests that it might have been a commercially applied layer and not a product of the studio, although this cannot be proved. A second ground layer was applied on top of the white priming. This upper ground is an intense red colour (appearing almost the colour of vermilion in cross-sections, although it is composed of a crystalline red earth pigment mixed also with some lead; see plate 13) (note 9). It is logical to assume that this upper layer was probably applied in the studio rather than representing part of a commercial double ground (note 10).
As might be expected from its powerful colour, the red ground plays a visual role, lending a slightly warmer character to the flesh tones, particularly of the female figure. However, this tonal influence is a subtle one, due to the opacity and density of the overlying paint layers.
The paint layers revealed in the cross-sections are thin and dense (see for example plate 14), in marked contrast to the translucent, medium-rich paint structure found in the Getty picture. The highly finished character of the surface is the result of the restrained, controlled application of one or two opaque layers, rather than the fluid multi-layered system found in the first version of the composition. It is debatable whether or not this was a deliberate attempt on Fragonard’s part to respond to the changing taste for neo-classical painting that had developed in late eighteenth-century Paris, largely a consequence of the profound impact of Jacques-Louis David’s highly successful output. It could be argued additionally that the Wallace Collection picture represents Fragonard returning to a mode of painting more in keeping with the principles fostered by the Académie – a way of painting that he had chosen to eschew earlier in his career. In spite of the speculative nature of those arguments, it is clear that the artist’s choice for a different mode of expression in this version of the composition was an entirely intentional and planned one.
As a result of the high proportion of lead white in the lower ground, and a proportion of lead white in the upper priming, the X-ray image of the Wallace Collection picture registers very little of the figurative paint layers. Infrared reflectography, however, is more revealing and gives a very clear indication of the composition, and, as noted above, does not show the presence of any pentimenti, suggesting that the compositional details were not altered in the course of painting in this, the presumed second version of the subject.
Comparison of the two versions
Although Fragonard’s manner of painting in the two versions of The Fountain of Love is markedly different, as noted earlier, the materials used in their making – particularly the palettes employed – are very similar. The difference in the surfaces and appearances of the two paintings is a fundamental consequence of the difference of technique employed, the ground colour and, to a lesser degree, the weight of canvas used (note 11). Only a limited number of samples were taken to investigate the paint layer structures involved, but they reveal also the majority of, if not all, the pigments employed (note 12). The more sombre-toned paint layers tend to be fairly complex mixtures of pigment, while the lighter parts of the compositions, the flesh paints and so on, are based on lead white or Naples yellow, used pure, or combined with perhaps just one other tinting pigment such as vermilion or an ochre. The full palettes are quite restricted, however, and appear to be made up only of lead white, a carbon black pigment, a variety of natural earth colours (particularly yellows and reds), vermilion (note 13), and a bright opaque primrose yellow identified as Naples yellow (lead antimonate) (note 14). Similar pigment mixtures were found in the grey-blue paints of the sky (largely carbon black with white) and the foliage greens (mixtures of yellow and black pigment) in both pictures. However, small quantities of Prussian blue were identified by FTIR in samples from the Getty painting and a few blue particles were observed (but not identified) in one sample from the Wallace picture. The only clear difference in palette between the two pictures is the presence of a deep redbrown lake pigment used in mixed paints in the Wallace Collection version (see plate 14) (note 15).
It is natural to assume that the composition of the Getty painting was transferred to the Wallace Collection canvas, but there is no evidence to suggest that this was done or how it might have been accomplished. There are no visible indications of underdrawing in either picture, nor are any full-scale compositional drawings known (note 16). However, it is revealing to overlay the compositions of the two paintings with the aid of computer imaging. Comparative points can be taken from similar details in each painting, and by means of imaging software (note 17) compositional alignments can be identified. If, for example, two comparison points are chosen from the female figures, such as the tips of the noses and the tips of the index fingers of the outstretched hands, and the areas between those two points are compared, the outlines of the upper torso and profile head are almost an identical match (plate 15). The only details that are not perfectly aligned are the forehead, hair and floral wreath of the female figure in the Wallace Collection picture; Fragonard chose to heighten these details in the second version, thus providing a more sculptural focus at the very centre of the scene. It is equally interesting to note that although the heads and torsos of the figures are positioned identically, the legs of the figures, albeit similar in profile, do not align perfectly (plate 16). The legs in the Wallace Collection picture appear to have been shifted slightly to the right and upwards, perhaps to increase the impression of fleeting movement. If a tracing had been used, this effect could have been achieved simply by shifting the paper a little to the right and upwards. Similar kinds of small compositional shifts occur as well in other areas of the picture, ranging from the precise placing of the various groups of putti to the angle of the profile of the edge of the fountain to the left. Since full-scale compositional drawings by Fragonard have not been identified, it is likely that after finalising the composition in the Getty picture, a tracing or other mechanical means of transfer was used to place the composition on to the Wallace Collection canvas.
Fragonard’s different approach to the handling of the Wallace Collection version also resulted in both the addition and the elimination of some details. The proper left foot of the male figure was added to the Wallace Collection picture, and this proved to be a subtle yet important detail, as the presence of the second foot helps to anchor the figure in space. The branch of the tree at the far right became more elaborate in the Wallace Collection picture, and is in keeping with the general effect of the landscape there, being painted with a heightened sense of depth and atmosphere. In the Getty picture, the hands of the putti were not fully developed, appearing as rapid shorthand notations, indicating their position and giving a sense of movement; in the Wallace Collection picture, however, they became fully articulated, completing the sculptural presence of the forms.
It is perhaps most telling, though, that nearly all of the water spilling from the fountain in the Getty version was omitted in the Wallace Collection composition. While the painterly mode of the Getty version allowed Fragonard to depict the rushing movement of the water with a rapid and fluid movement of the brush, the more restrained mode of the Wallace picture precluded such open, free brushwork, so the details involving cascading water were simply eliminated. Unfortunately, by eliminating this feature of the first design, the movement of the hands of the putti in the fountain at the left edge became less meaningful. Lastly, indications of a quiver and a plant in the lower right corner of the Getty version were replaced in the Wallace Collection picture with a prominent and carefully executed signature (plate 17). The formalised appearance of the signature, which includes a highlighted flourish on the extended tail of the letter ‘g’, hinting at the illusion of a classical relief sculpture, neatly underscores the essence of the difference between the two pictures.
The Regnault print
On 18 November 1785, the Journal de Paris (p. 1327) announced for sale the print after Fragonard’s The Fountain of Love (plate 18) (note 18). It is, in fact, with the sale of the prints after his paintings that a great part of Fragonard’s fame and fortune was made. Furthermore, with the publication of the prints he was assured copyright of his compositions. A second notice appeared the next month in the Mercure de France (24 December 1785, p. 618). The printmaker was Nicolas-François Regnault (1746–c.1810), a close contemporary of the painter. Regnault had become a master at stipple etching, the medium in which he reproduced several of Fragonard’s paintings. Although colour printing was much in vogue at the time, this particular technique produces the tonal effects that best capture the brilliant chiaroscuro of Fragonard’s original painting. When in 1787, Marguerite Gérard (1768–1837), Fragonard’s sister-inlaw (having married the printmaker Henri Gérard who produced the print after the Sacrifice of the Rose), painted The Interested Student (note 19), she depicted her heroine seated in the midst of the bric-à-brac of a shop holding in her hands a framed and glazed copy of Regnault’s print. Appropriately, given the subject of the print, a statue in plaster (or marble) of two embracing cupids on a draped stand overlooks this connoisseur, who has perched her plumed bonnet on their heads. Almost a century later, Regnault’s print after Fragonard makes a telling appearance in Jean-Léon Gérome’s small portrait (or genre painting) called In the Antechamber, 1879 (note 20). From the subject of the print, we can surmise what the seated gentleman in Gérome’s painting is anxiously waiting for. Indeed, Regnault’s print after The Fountain of Love was most successful and served as a source for innumerable craftsmen working in porcelain, silver, wood and other materials from the latter part of the eighteenth to the end of the nineteenth century.
While it might be imagined that it would be easier for Regnault to replicate the harder, more highly finished version of The Fountain of Love, that is, the Wallace Collection version, it seems that the print bears a closer relationship to the more painterly Getty version. The print is believed to have been produced in 1785, and Cuzin puts the Getty picture just before 1785 and the Wallace Collection picture just after 1785 (note 21). In the stipple etching of 1785 Regnault adapts, but does not repeat, the small flowering plant in the lower right corner of the Getty painting (behind which he eliminates the small red quiver with arrows, presumably dropped by one of the winged putti). The water spilling from the fountain, less easily reproduced in the ‘licked’ type of surface of the Wallace Collection painting and therefore eliminated, is comprehensively adopted in the print from the Getty picture. By contrast, the weightiness of the two figures seems to be captured not from the more ethereal Getty version but rather from the Wallace picture. What conclusion can be drawn from Regnault’s print derived from Fragonard’s composition? The most probable explanation is that Regnault had access to both pictures before they left Fragonard’s studio. The print is an amalgam of details drawn from both paintings with some adaptations made, probably in collaboration with Fragonard himself, so that the well-known reproductive print became a unique variation on both compositions.
Generous support for this project was provided by the J. Paul Getty Trust, the National Gallery, and the Friends of Heritage Preservation, a Los Angeles-based group dedicated to the conservation of important works of art world-wide.
We are grateful to a number of colleagues at the National Gallery and at the Wallace Collection for their assistance. At the National Gallery, Joseph Padfield provided the comparative imaging overlays, Rachel Billinge the infrared reflectography, and David Peggie and Rachel Morrison carried out medium analyses. Humphrey Wine offered suggestions on the text. At the Wallace Collection, support for the project was provided by Rosalind Savill, Jeremy Warren, Stephen Duffy and Jo Hedley.
Appendix: Corrected provenance of the two versions of The Fountain of Love
Getty version: oil on canvas, 64 x 52.4 cm, in a period neo-classical frame.
Possibly Charles-Nicolas Duclos-Dufresnoy (1733– 1794), notaire, pre-empted by the state but restored to his natural son Charles-Athanase Walckenaer (1771– 1856) on 22 July 1795, by whom sold; Paris, 18–21 August 1795, lot 28 (see Wallace version) [see Note A below]; possibly Villeminot, payeur générale de la marine, by whom sold; Paris, 25 May 1807, lot 21 (see Wallace version) [see Note B below]; Mr Robert Hollond (1808–1877) and his wife Ellen Julia Hollond, née Teed (1822–1884) [see Note C below], Stanmore Hall (not Castle), Stanmore, Middlesex, London, Paris and Cannes by about 1840/60? and by inheritance to his nephew John R. Hollond, Wonham, Bampton, Devonshire, whose estate was sold at Christie’s London, 11 April 1913, lot 50 (1000 guineas), to Thomas Agnew & Sons, Ltd (stock no. 4202), by whom sold on 21 October 1913 to Henry Reinhardt & Sons, New York, by whom sold between 1913 and 1915 to John North Willys (1873–1935), Toledo, Ohio; by inheritance to his daughter Mrs Virginia Willys Lucom, Palm Beach, Florida, and by inheritance to her husband Mr Wilson C. ‘Buzz’ Lucom (died 2006), Palm Beach and Panama City; by whom sold at Christie’s New York, 29 January 1999, lot 167, to Simon Dickinson, Inc., New York; by whom sold to the J. Paul Getty Museum in the summer of 1999 (Acc. no. 99. PA.30).
Wallace Collection version: oil on canvas, 63.5 x 50.7 cm, in the Demidoff frame of about 1840.
Possibly Charles-Nicolas Duclos-Dufresnoy (1733– 1794), notaire, pre-empted by the state but restored to his natural son Charles-Athanase Walckenaer (1771– 1856) on 22 July 1795, by whom sold; Paris, 18–21 August 1795, lot 28 (23 pouces x 18 pouces, 9 lignes = approximately 62.26 x 50.76 cm) [see Note A below]; possibly Villeminot, payeur générale de la marine, by whom sold; Paris, 25 May 1807, lot 21 (64 x 56 cm) [see Note B below]; Count Nicolai Demidoff (1773–1828), Moscow and San Donato; by inheritance to his son, Count Anatole Demidoff (1812–1870), San Donato and Paris; by whom sold at San Donato Sale, Paris, 26 February 1870, lot 106, to Richard Seymour-Conway (1800–1870), 4th Marquess of Hertford, London and Paris; by descent to Sir Richard Wallace (1818–1890) and then to his wife Lady Wallace (1819–1897), London; by whom bequeathed to the Nation in 1897 as part of the Wallace Collection (Acc. no. P394).
The Wallace picture has been published extensively since the nineteenth century. The Getty Fountain of Love, known only very recently, has been published in the following: R. Portalis, Honoré Fragonard, sa vie et son oeuvre, 1889, p. 277; A. Dayot and L. Vaillat, L’oeuvre de J.B.S. Chardin and J.-H. Fragonard, 1908, p. xvii, under no. 127; The Wallace Collection Catalogue: Pictures and Drawings, 1913, pp. 76–7 under P394; Wallace Collection Catalogue; Pictures and Drawings, 1928, p. 104; L. Réau, Fragonard; sa vie et son oeuvre, 1956, p. 163; G. Mandel, L’opera completa di Fragonard, 1972, p. 109, no. 518a; J.P. Cuzin, Jean-Honoré Fragonard, sa vie et son oeuvre, 1987, p. 332 under no. 373; J. Ingamells, The Wallace Collection, Catalogue of Pictures, III. French before 1815, 1989, pp. 154–5, no. 394 under versions; P. Rosenberg, Tout l’oeuvre peint de Fragonard, 1989, no. 409B; J.P. Cuzin, ‘Fragonard: quelques nouveautés et quelques questions’, Mélanges en homage à Pierre Rosenberg, 2001, pp. 174–6, figs 12-13 (the first publication of the Getty painting with illustration; and noted as the first version of the composition); S. Duffy and J. Hedley, The Wallace Collection’s Pictures; a complete catalogue, 2004, pp. 156–7, under P394 (the Getty painting as the first version of the composition); Andrei Molotiu, Fragonard’s Allegories of Love, 2007, passim (but with incorrect provenance given).
Note A: Colin B. Bailey, Patriotic Taste; Collecting Modern Art in Pre-Revolutionary Paris, 2002, pp. 131–2, esp. p. 280, footnote 9. This supersedes his ‘Madame Du Barry and Greuze’, Burlington Magazine, 136, November 1994, pp. 758–9. J. P. Cuzin, ‘Fragonard: quelques nouveautés et quelques questions’, Mélanges en homage à Pierre Rosenberg, 2001, pp. 174–6, esp. p. 175 where he also mentions that the dimensions do not rule out the possibility that the Getty picture may be the one sold in the Duclos-Dufresnoy sale.
Note B: S. Duffy and R. Wenley, Anatole Demidoff: Prince of San Donato (Collectors of the Wallace Collection, I), exhibition catalogue, Wallace Collection, London, 1994, pp. 60–1, cat. no. 22, for the Wallace version Demidoff provenance.
Note C: Ellen Julia Hollond, a fascinating salon hostess (in Paris and London), philanthropist and author (see Patrick Waddingham in the Dictionary of National Biography online 13570), owned many pictures by both old masters and contemporary artists as is evidenced in the Christie’s sale (see provenance of the Getty painting) of her husband’s nephew’s estate (he inherited their collection, they having died childless). During her lifetime she gave to the National Gallery François Boucher’s jewel-like Pan and Syrinx (NG 1090), 1759, and bequeathed Ary Scheffer’s portrait of herself (NG 1169) and his Saints Augustine and Monica (NG 1170), 1854 (Julia Hollond was the model for Saint Monica). Robert Hollond’s portrait was painted by John Hollins in his A Consultation Prior to the Aerial Voyage to Weilburgh (NPG 4710) of 1836 in the National Portrait Gallery.
Mark Leonard is Head of Paintings Conservation and Scott Schaefer is Senior Curator of Paintings at the J. Paul Getty Museum.
1 Both versions of the picture appear to be on standard size canvases, quite close to a ‘toile dite de douze’ (No.12: 22.5 x 18.5 in. ) available in two weights of canvas weave. No.15 is also fairly close. See Paillot de Montabert, ‘Tarif et Mesures des Toiles Préparées’, Traité Complet de la Peinture, Vol. 9, Paris 1829, p. 147.
2 In May 2007, both pictures were brought to the conservation studio at the National Gallery, London. Mark Leonard, the Conservator of Paintings from the Getty Museum, cleaned and restored the Wallace Collection picture (which had been covered with a darkened varnish and discoloured retouches), and worked with the conservation scientists at the Gallery on a technical study of both paintings. The paintings were then exhibited publicly in London at Hertford House until October 2007 when the Getty version travelled on to an exhibition, Consuming Passion: Fragonard’s Allegories of Love, at the Sterling and Francine Clark Art Institute in Williamstown, Massachusetts. It returned to Los Angeles with the same exhibition in early 2008.
3 Fragonard’s version of The Fountain of Love, now in the J. Paul Getty Museum, was first seen and identified by Scott Schaefer while on a visit to Palm Beach, Florida, as Senior Vice-President of Sotheby’s New York in the winter of 1996–7. He noticed the painting hanging high above a very tall cabinet in the dining room of the Villa Flora (an Addison Mizner house originally built for Edward Shearson in 1923). Although Mr Wilson Lucom, the owner, thought it ‘only a copy’, Schaefer asked that it be brought down so that he could examine it more closely. He subsequently informed the owner that the picture was, in his opinion, another autograph version of the picture in the Wallace Collection by Fragonard himself and would do very well at auction; it was later sold by Christie’s New York in 1999. Even after its recent reappearance, nowhere is the provenance of the picture listed correctly. For this reason the present authors have wished to give the known provenances of both versions of the picture here (see above). Both the Wallace Collection and Getty versions can only be traced back with certainty to the mid-nineteenth century. The earlier provenance, traditionally given for the Wallace Collection version, could just as easily relate to the Getty example, given the closeness of the dimensions (also corrected here) and the subject matter.
4 Lead white was identified in the ground layer by EDX analysis in the SEM; no other white components were present. The coloured pigment particles are earths and carbon black. Lead white was also identified by the same means in the fluttering white draperies of the female figures in both versions.
5 According to treatises of the period, there was some uncertainty as to the best methods of preparing canvases for oil painting and the materials to employ. Some authors ‘hedged their bets’ and described the advantages and disadvantages of various methods. A primary purpose for the ground layer, particularly a lower ground, where more than one layer was involved, was to fill the canvas weave with a layer of good mechanical strength and body. Lead white in oil works extremely well for this purpose and has been found as the priming on canvas for French eighteenth-century paintings, becoming the norm a few decades after the date of the Fragonards discussed here. Similarly, coloured grounds in one or more layers were also common, and to judge from a survey of paintings in the National Gallery collection, no characteristic or standard method of priming emerged in France in the eighteenth century, although ground colours in the warmer range of hues are more common than others, particularly red-browns, dull reds and pinkish tones. For further comments on French eighteenth-century ground types and other aspects of technique, see A. Massing, ‘French Painting Technique in the Seventeenth and Early Eighteenth Centuries and De La Fontaine’s Académie de la peinture (Paris 1679)’, in E. Hermens, ed., Looking Through Paintings: The Study of Painting Techniques in Support of Art Historical Research, Baarn, The Netherlands, and London 1998, pp. 319–90; also, E. O’Donoghue, R. Romero and J. Dik, ‘French Eighteenth-Century Painting Techniques’, in A. Roy and P. Smith, eds, Painting Techniques: History, Materials and Studio Practice, Dublin Congress of the IIC, London 1998, pp. 185–9.
6 GC-MS analysis found fatty acid methyl esters in proportions consistent with heat-bodied linseed oil. Fir balsam was indicated by the presence of the characteristic norambreinolide component (B+ 123, M+ 250) together with methyl dehydroabietate and corresponding oxidation products. Various dehydrodehydroabietate and hydroxydehydroabietate components were also observed (a series of components with B+ 237).
7 It was common practice at this period to work up the composition of a painting with a preliminary lay-in of relatively dark monochrome paints, often quite translucent, representing a so-called ébauche, perhaps over an underdrawing. Interestingly, although J.-B. Oudry refers to this method of composing a picture, he also recommends a layer of varnish applied as a preliminary to further painting in order that the next layers of paint may glide on; this is not a widely used technique, but he suggests that it allows greater ease of handling. Fragonard may have been influenced by a combination of these methods. See J.-B. Oudry, ‘Discours sur la pratique de la peinture: ébaucher, peindre à fond et retoucher’ (original lecture: 1752) published in E. Piot, Le Cabinet de l’amateur, nouvelle série 1o année: années 1862–3, Paris 1863, pp. 107–17, esp. p. 110.
8 In a technical study of the larger finished canvas version of Fragonard’s Le Verrou (canvas, 73 x 92 cm; c.1778) in the Louvre (for which a small sketch on wood panel exists, 26.3 x 39.5 cm), the authors J. Ligot and L. Faillant- Dumas comment on the extensive cracking in the paint layers of the canvas picture, identify different forms of craquelure and cite a variety of separate causes, including drying defects and later treatments: see J. Ligot and L. Faillant- Dumas, ‘Le Verrou examiné au laboratoire’, La Revue du Louvre et des Musées de France, 39, 1, 1989, pp. 80–3.
9 The materials in both ground layers were identified by EDX analysis.
10 The cross-sections do not reveal any visible interface between the two ground layers, which might suggest the passage of some significant drying time between the application of the layers. However, there are ways of rubbing down a dried ground before applying the next which would probably eliminate any evidence of this kind.
11 Canvas weave thread counts on the X-radiographs gave on average: 19 threads/cm (vertical) and 18 threads/cm (horizontal) for the Getty painting, and 14 threads/cm (vertical) and 13 threads/cm (horizontal) for the Wallace Collection version.
12 Samples taken around the edges may not in general be representative of all materials used in the painting. However, given the heterogenous (and multilayered) nature of the paint layers and the precaution of a surface survey of both pictures with the stereomicroscope, it seems likely that in this study the full palettes can be defined fairly reliably. The overall (restricted) palette is similar to that reported for Le Verrou, see Ligot and Faillant-Dumas 1989 (cited in note 8), p. 83.
13 A few particles of vermilion were detected in mixed paints from the edges of the compositions; it seems likely that vermilion is the tinting pigment employed for the rosier flesh tones in both versions.
14 Lead and antimony, with virtually no tin, was confirmed by EDX in samples from each painting: bright impasto yellow foreground, right (Getty); strong pale yellow, right (Wallace Collection).
15 Analysis of individual particles of this red-brown lake pigment by EDX, with EDX-mapping, showed the presence of aluminium, potassium and sulphur, arising from the substrate; no analysis of the lake dyestuff proved possible in the small heterogenous samples available.
16 The use of compositional drawings for complete paintings does not seem to be a general feature of Fragonard’s working method. However, two versions of comparable size, one highly finished, the other summary and sketch-like, of The Visit to the Nursery, Washington DC, National Gallery of Art, painted before 1784 (canvas, 73 x 91 cm, highly finished and location unknown; canvas, 64 x 79.5 cm, ‘sketch’) appear to have a ‘small’ drawing of the whole composition associated with the paintings, but its role is not clear. See Jean- Honoré Fragonard (1732-1806): Orígenes e influencias. De Rembrandt al siglo XXI, exh. cat. Fundación ‘la Caixa’, Barcelona 2006, pp. 96–7; the drawing (private collection) is fig. 48, p. 97. See also Cuzin 2001 (cited in Note A), cats 123–5 (the images are wrongly captioned there).
17 The blended images or overlays were produced by combining each image using VIPS image-processing software, which is available to download at no charge from the VIPS website: http://www.vips.ecs.soton.ac.uk (accessed 16 November 2007). Documentation of the software and its installation is also available on the site as well as examples of the overlay process used for the Fragonard combined images contained in this article.
18 See Victor Carlson in Regency to Empire: French Printmaking 1715-1814, exh. cat., Baltimore Museum of Art and the Minneapolis Institute of Arts, 1985, pp. 256–7, cat. no. 89; Aimer en France 1780-1800, exh. cat., Bibliothèque Municipale et Inter-Universitaire, Clermont-Ferrand, 1977, p. 65, cat. no. 66
19 Present location unknown – the painting is known only through Gérard (or Géraud) Vidal’s (1742–1801) engraving after it.
20 G.M. Ackerman, Jean-Léon Gérome, London 2000, p. 296, cat. no. 275, illus.
21 Cuzin 2001 (cited in Note A), pp. 174–6.
This volume of the Technical Bulletin has been funded by the American Friends of the National Gallery, London with a generous donation from Mrs Charles Wrightsman.
Series editor Ashok Roy
© National Gallery Company Limited 2008
All rights reserved. No part of this publication may be transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without the prior permission in writing of the publisher.
First published in Great Britain in 2008 by National Gallery Company Limited St Vincent House, 30 Orange Street
London WC2H 7HH
British Library Cataloguing in Publication Data
A catalogue record for this journal is available from the British Library
ISBN 978 1 85709 419 0
ISSN 0140 7430
Project manager Jan Green
Editor Diana Davies
Designer Heather Bowen
Picture research Karolina Majewska
Production Jane Hyne and Penny Le Tissier
Repro by Alta Image, London
Printed in Italy by Conti Tipocolor