Overview
Paul Sérusier’s Boys on a river bank (fig. 1) was purchased for the National Gallery of Victoria through the Felton Bequest in 1948. The painting is dated 1906, but examination has revealed a complete underlayer of paint, carrying an almost identical composition and likely to have been applied at a much earlier time. The texture of this underlayer suggests that the painted canvas was rolled at some stage, with the top layer later being applied over the resulting ridges, losses and cupping in the paint beneath. Technical examination of the work proceeded with the aim of trying to establish a dating for the earlier paint layer. Pigments the artist used in each of the two layers were identified, and were considered in the light of his colour theories and of related shifts in his palette. Sérusier’s contact with Gauguin may also provide some clues about the conception of the composition of the Melbourne picture.
Examining the painting
The investigation of Boys on a river bank was originally prompted by the work’s physical condition: the painting appeared fragile, showing many losses and areas of cupping paint. This apparent fragility, however, was found to be due not to an ongoing loss of paint in the upper paint layer, but to losses and cupping in the paint and ground layers below.
An initial examination of the work in raking light first pointed to the existence of an underlayer, to whose topography the upper layer appeared to conform. The examination revealed a horizontal trend to the deformations in this underlayer, a trend consistent with the canvas’s having been rolled across its width (fig. 2).1 The deformation is not a consequence of the weave of the canvas, which is even in both directions.
There are at least two sets of holes in the tacking edges, indicating that the work was at some stage taken off its stretcher, and subsequently reattached.2 This removal of the canvas from its stretcher was independent of a later removal for restretching.
Closer inspection of the disrupted areas of the painting, using a stereo-microscope, confirmed the existence of an underlayer. This paint layer must already have been hard and cupped when the more fluid top layer was applied over its surface (fig. 3), and a period of time long enough for the lower layer to harden must therefore have elapsed between the two applications. For this reason, and, as we shall see, for reasons associated with the artist’s palette, it can be said that the lower layer predates 1906. The top layer extends not only over cupped paint, but also over losses (figs 3 & 4); this is seen most clearly on the right hand of the figure in the water, where the red-brown paint flows in some areas over flakes of paint already sitting on the canvas, and, elsewhere, over the bare canvas itself (fig. 4).
When comparing the two paint layers under low magnification, one can see that the outlines of the figures and other forms in the lower layer have been retraced in the second application of paint.3 Although the upper paint layer covers the entire canvas, the paint has been applied in such a way that tiny areas of the underlayer remain visible. It is these areas that were examined with the aid of the stereo-microscope. X-radiography, which penetrates a painting’s surface, has revealed that the composition exists in its entirety in the lower layer, with only comparatively minor differences (fig. 5). Importantly, one can also see in the X-radiograph the losses in the underlayer that have been covered by the artist in his reworking of the painting. These losses appear as black areas in the positive print of the X-radiograph (fig. 5).
Although the X-radiograph reveals minor differences between the final version of the painting and the composition seen in the underlayer, in a number of cases it is unclear as to whether or not the changes were deliberate. It is possible that the silhouette of the bush at the top right was reduced to make the form of the boy’s head more pronounced, by showing it against the lighter tone of the water. The torso of the same boy has been straightened on the right side, between the waist and the hip. His feet, together with the right hand and foot of the seated figure, appear not to have been clearly defined in the earlier composition, though here the apparent difference between the two paint layers may in fact relate to densities of paint and to the paint’s penetrability by X-radiography, rather than to a deliberate modification by the artist.4 Drip marks that run towards the right side of the painting, as if it has leant on that edge while some application of paint or ground was wet, are also revealed in the X-radiograph. It is also possible that these drip marks occur on the back of the canvas. In some areas, the cracking and tenting of the paint appears to follow the line of the drips.
Pigment analysis and microscopy
Cross-sections were taken from the painting in order to enable further analysis of the structure and chronology of the paint layers.5 Samples of paint taken for analysis by cross-section are typically almost too small to be seen by the naked eye. Under the stereo-microscope, a section of paint is removed from the edge of a loss or crack, with the aid of a small, fine blade; the sample is set in a resin block, which is then slowly sanded back to the point where the strata of the paint are revealed. This procedure is undertaken using the more high-powered compound microscope to check the surface of the paint as sanding progresses. The samples were selected from the water, the grass, the bush in the top right corner, and one of the figures, to represent the artist’s use of colour throughout the painting.
That the top layer of paint was applied considerably later was most dramatically demonstrated by a cross-section taken from the grass on the right-hand side of the painting (fig. 6). In this cross-section one can see a rift through both the lower paint layer and the ground; the paint applied on top has flowed down through the rift – proof that it existed prior to the application of the top layer, which in this area consists of two applications of yellow paint.
Another cross-section, from the centre of the grassy area, clearly reveals the earlier and later greens as having very different compositions (fig. 7), the examination of which may enable these greens to be situated within the context of Sérusier’s colour theories and the shift in his palette over time.
The earlier green is a homogeneous mixture of small particles of green, black and white, with traces of red and dark blue pigments. Elemental analysis using scanning electron microscopy (SEM)6 SEM uses a beam of electrons, rather than light as in a conventional microscope, to produce an image of the sample. Magnification of between 20 000 and 30 000 times can be achieved on some instruments, but for the purposes of the present analysis magnification was in the order of 1000 to 10 000 times. In addition to the visual images produced by the beam of electrons, the X-rays generated during the bombarding of the sample allow elemental analysis to be undertaken. SEM and analysis of samples from Boys on a river bank was generously undertaken by Ian Harrowfield at the CSIRO Minerals Division, Melbourne, and by Deborah Lau-Greig using facilities at the University of Melbourne. confirmed these visual findings, revealing chromium (the major constituent of viridian or chromic oxide green), a carbonaceous black,7 SEM (in this instance) could not detect carbon, but its presence in the black, suggested by the morphology of the pigment particles under optical microscopy, could be deduced from the absence of other elemental data. and lead white. The red seen in the cross-section could not be identified using SEM, because of the practical difficulties in locating the very small red particles within the black and white image on the screen of the microscope.
The later green layer was seen under the optical microscope to contain distinct, needle-shaped particles characteristic of strontium yellow, and large, irregular blue pigment particles. Strontium, confirming the presence of strontium yellow, was subsequently found using SEM, but the blue pigment is yet to be identified.8 The blue was found to contain both cobalt and aluminium, both of which normally occur in cobalt blue, though the cobalt count was very low, and the size and shape of the particles was not characteristic of cobalt blue (this was confirmed by Raman laser spectroscopy undertaken by Dr Kerry Nugent in the Physics Department at the University of Melbourne). The blue material appears to have the same gross characteristics as the glass-based pigment smalt, although silicon, which is a major component of glass, was not located at particle sites in the course of two different analyses. Additionally, a single grain of pigment was tested by Dr Ian Madsen, CSIRO Minerals Division, Melbourne, using X-ray diffraction, to determine whether it was indeed an amorphous glass-based material such as smelt, or a crystalline mineral; however, the tests were inconclusive due to the very small size of the particle. Smalt is usually associated with much earlier use, but it was available in the nineteenth century (see R. Harley, Artists’ Pigments, c.1600–1835, 2nd edn, Butterworth, London, 1982, p. 56).
Dating the underlayer: Sérusier’s colour theories and parallels with Gauguin
Sérusier documented his theories of colour and painting in ABC de la peinture, published in 1921. His use of colour changed throughout his life, notably from an early sombre palette to a more vibrant range of colours later in his painting career.
In the early 1890s, Sérusier favoured a palette based on a dominant colour – used with mixtures of that colour with its opposite (complementary) colour, as well as with black and white – to ensure a harmonious composition. What is revealed in the cross-sections taken from Boys on a river bank, together with what the stereo-microscope shows beneath the top layer of paint,9 See note 3 above. strongly suggests that the initial version of the painting was quite sombre, thus relating to the artist’s use of colour in the early 1890s, though perhaps the underlayer in the Melbourne picture is less monochromatic than some works from Sérusier’s early years.
In the initial layer of Boys on a river bank the colours also appear to be mixed in the manner in which Sérusier made up his colours in the early 1890s. For example, a cross-section taken through the bush shows a mixture of blue-green, white, black and a small amount of dark blue, plus the complementary red-brown. A further cross-section shows the colour used for the water of the river, a pale green overall, to be mixed in a similar way, though with more white; it contains the dominant colour green, with traces of yellow pigment, as well as black – and perhaps dark blue – plus white.10 SEM of the lower paint layer found chromium both in isolation, indicating chromic oxide (green), and in association with lead, indicating lead chromate (yellow). In 1896 Sérusier abandoned his earlier theories, now propounding a system of colour selection based on a colour wheel used in combination with a complex construction of triangles. 11 P. Sérusier, ABC de la peinture (1921), rev. edn, Paris, 1942, pp. 76–9.
The principal colours in the top layer of Boys on a river bank – the red-orange of the fleshtones, the bright green of the grass, and the blue-purple of the bush and smock – appear consistent with this method of choosing colour. In 1896, however, Sérusier also wrote of another system of colour selection, as a simple alternative to that requiring the triangular construction over the colour wheel. This second system involved working primarily with a dominant colour, its opposite, a colour between the two and a ‘dissonance’ made with the dominant colour plus white. The colours in the top layer of Boys on a river bank may equally be aligned with this system. The fact that Sérusier exercised a certain flexibility in applying his principles of colour selection makes it difficult to place the palette in the Melbourne painting definitively within the framework of a specific theory. Suffice it to say that the colours are strongly in keeping with the shift away from the palette governed by Sérusier’s early theory – the palette with which the lower paint layer corresponds.
As noted above, cross-sections show that the colours chosen for the later layer are markedly different, in tone and brightness, from those beneath. The green of the grass in the later application is the most notably different colour (fig. 7), but in addition the pale green of the water has been made significantly brighter, and the blue-green of the bush has been rendered darker and in greater contrast with its surroundings.
Sérusier’s work was so diverse throughout his career that attempting to date the underlayer of Boys on a river bank by locating the composition within the context of the artist’s oeuvre as a whole proves to be an impossible task. Moreover, although the stretcher used for this painting is one Sérusier favoured in his earlier years, it was also used later in his career, and thus is not a useful means of dating the underlayer. Interestingly, Sous la lampe, a painting with the same dimensions and carrying the same date as Boys on a river bank, also has a textured appearance, at first suggesting that the final composition was painted over an earlier version. Examination and analysis of this picture, however, show the paint to have been directly applied to the canvas in one layer, despite the textured surface.12 Tuulikki Kilpinen, head of conservation at the Ateneum in Helsinki, where the painting is held, very kindly examined the work in detail and arranged for analysis by Pixe/Pigme at the University of Helsinki. In this method, the paint on the surface of a work is analysed without the need for taking a sample, though the painting has to be transported to the laboratory for the analysis to be carried out. The technique is similar to SEM, but uses a beam of protons, rather than electrons, to generate X-rays and gamma rays in the paint at the selected site. This process enables the elements – and thus the pigments – in the paint layer to be identified.
An interesting clue to the conception of Boys on a river bank may lie with Sérusier’s contact with Paul Gauguin. Gauguin and Sérusier knew each other, and both painted at Pont-Aven in the 1880s.13 A lesson Sérusier had with Gauguin in October 1888 is documented in chronologies of his life (see J. Warnod & M. A. Anquetil, L’Eclatement de l’impressionisme (exh. cat.), Musée Départemental du Prieuré, Saint-Germain-en-Laye, 1982, p. 135. According to Boyle-Turner, there is some evidence that Sérusier also had contact with Gauguin in 1893 (C. Boyle-Turner, Paul Sérusier, Ann Arbor, Michigan, c.1983, p. 92). In 1888, Gauguin painted two works of boys bathing – Young Bretons bathing (Hamburger Kunsthalle) and Children wrestling (private collection).14 See R. Pickvance, Gauguin and the Pont-Aven School (exh. cat.), Art Gallery of New South Wales, Sydney, 1994, cat. no. 5, repr. p. 33. The latter is markedly similar to Boys on a river bank.15 It was Children wrestling, viewed by the author in the exhibition Gauguin and the Pont-Aven School at the Art Gallery of New South Wales, Sydney, in 1994 (see note 14 above), that initiated further analysis of Boys on a river bank. Gauguin describes his painting in a letter to van Gogh in July 1888:
I have just finished a Brittany fight that you will like, I’m sure. Two young boys in blue and vermilion bathing-trunks. Another in the top right getting out of the water. Green lawn, pure veronese-green graduating to chrome yellow without execution like Japanese crepon prints. At the top a cascade of boiling water, white pink and rainbow-coloured at the edge near the frame. At the bottom, a white patch, a black hat, and a blue smock.16 Paul Gauguin, letter to Vincent van Gogh, July 1888, cited in Pickvance, cat. no. 5.
Sérusier has tackled the grass in much the same way as did Gauguin. Indeed, the Melbourne painting contains many elements similar to those in Gauguin’s composition, indicating, in my view, a direct relationship between the two.
Dating the upper layer: the signature
The physical characteristics of the signature raise the possibility that it postdates the application of the top paint layer. Under magnification, the signature is seen to have blurred edges, normally the result of paint being applied wet-in-wet, that is, over a wet paint layer (fig. 8). However, the blue paint used for the signature is of a wash-like consistency, and the blurred appearance at the edges of the letters is in fact almost certainly due to the yellow paint beneath being seen through its very thin layers. It appears that the painting was ‘dry’ when the signature was applied: the thin paint of the signature has run from the ridges of the paint below it – indicating that the paint film was sufficiently firm to hold its ridges at the time the signature was inscribed. There exists, then, the possibility that the upper layer may not be exactly contemporary with the signature, and may have been painted at an earlier time.
Summary
Two distinct paint layers exist in Boys on a river bank. The lower layer certainly predates 1906, the date inscribed, with the signature, in the lower right-hand corner of the work. The approach to colour in this underlayer also appears to coincide with Sérusier’s colour theories of the early 1890s, but not with those dating from after the mid-1890s; the initial composition of the painting may therefore date from the earlier period. The colours in the later layer of the composition appear to relate to Sérusier’s theories of the mid to late 1890s. Additionally, the physical characteristics of the signature suggest the possibility that the upper layer was painted earlier than the signature. This in turn raises the question of when the later layer might have been laid down. It could have been applied prior to 1906, but how much earlier? And what is the significance of the similarities between Sérusier’s colours – and indeed his whole composition – and Gauguin’s treatment of a comparable subject in 1888?
If we accept that Sérusier’s practice was consistent with his colour theories, the upper layer may date from as early as 1896 – or perhaps from even earlier if the artist wrote of his theories after developing them in his painting practice. The visual links with Gauguin’s Children wrestling of 1888 open the possibility that Sérusier’s work was painted around this time, though this hypothesis does not account for the fact that his use of a brighter palette is generally held to date from the mid-1890s.17 For further discussion of Boys on a River Bank, see S. Dean, European Paintings of the 19th and Early 20th Centuries in the National Gallery of Victoria, Melbourne, 1995, p. 109.
Linda Waters, Paintings Conservator, National Gallery of Victoria (in 1997).
Notes
1 The deformation is not a consequence of the weave of the canvas, which is even in both directions.
2 This removal of the canvas from its stretcher was independent of a later removal for restretching.
3 Although the upper paint layer covers the entire canvas, the paint has been applied in such a way that tiny areas of the underlayer remain visible. It is these areas that were examined with the aid of the stereo-microscope.
4 Drip marks that run towards the right side of the painting, as if it has leant on that edge while some application of paint or ground was wet, are also revealed in the X-radiograph. It is also possible that these drip marks occur on the back of the canvas. In some areas, the cracking and tenting of the paint appears to follow the line of the drips.
5 Samples of paint taken for analysis by cross-section are typically almost too small to be seen by the naked eye. Under the stereo-microscope, a section of paint is removed from the edge of a loss or crack, with the aid of a small, fine blade; the sample is set in a resin block, which is then slowly sanded back to the point where the strata of the paint are revealed. This procedure is undertaken using the more high-powered compound microscope to check the surface of the paint as sanding progresses.
6 SEM uses a beam of electrons, rather than light as in a conventional microscope, to produce an image of the sample. Magnification of between 20 000 and 30 000 times can be achieved on some instruments, but for the purposes of the present analysis magnification was in the order of 1000 to 10 000 times. In addition to the visual images produced by the beam of electrons, the X-rays generated during the bombarding of the sample allow elemental analysis to be undertaken. SEM and analysis of samples from Boys on a river bank was generously undertaken by Ian Harrowfield at the CSIRO Minerals Division, Melbourne, and by Deborah Lau-Greig using facilities at the University of Melbourne.
7 SEM (in this instance) could not detect carbon, but its presence in the black, suggested by the morphology of the pigment particles under optical microscopy, could be deduced from the absence of other elemental data.
8 The blue was found to contain both cobalt and aluminium, both of which normally occur in cobalt blue, though the cobalt count was very low, and the size and shape of the particles was not characteristic of cobalt blue (this was confirmed by Raman laser spectroscopy undertaken by Dr Kerry Nugent in the Physics Department at the University of Melbourne). The blue material appears to have the same gross characteristics as the glass-based pigment smalt, although silicon, which is a major component of glass, was not located at particle sites in the course of two different analyses. Additionally, a single grain of pigment was tested by Dr Ian Madsen, CSIRO Minerals Division, Melbourne, using X-ray diffraction, to determine whether it was indeed an amorphous glass-based material such as smelt, or a crystalline mineral; however, the tests were inconclusive due to the very small size of the particle. Smalt is usually associated with much earlier use, but it was available in the nineteenth century (see R. Harley, Artists’ Pigments, c.1600–1835, 2nd edn, Butterworth, London, 1982, p. 56).
9 See note 3 above.
10 SEM of the lower paint layer found chromium both in isolation, indicating chromic oxide (green), and in association with lead, indicating lead chromate (yellow).
11 P. Sérusier, ABC de la peinture (1921), rev. edn, Paris, 1942, pp. 76–9.
12 Tuulikki Kilpinen, head of conservation at the Ateneum in Helsinki, where the painting is held, very kindly examined the work in detail and arranged for analysis by Pixe/Pigme at the University of Helsinki. In this method, the paint on the surface of a work is analysed without the need for taking a sample, though the painting has to be transported to the laboratory for the analysis to be carried out. The technique is similar to SEM, but uses a beam of protons, rather than electrons, to generate X-rays and gamma rays in the paint at the selected site. This process enables the elements – and thus the pigments – in the paint layer to be identified.
13 A lesson Sérusier had with Gauguin in October 1888 is documented in chronologies of his life (see J. Warnod & M. A. Anquetil, L’Eclatement de l’impressionisme (exh. cat.), Musée Départemental du Prieuré, Saint-Germain-en-Laye, 1982, p. 135. According to Boyle-Turner, there is some evidence that Sérusier also had contact with Gauguin in 1893 (C. Boyle-Turner, Paul Sérusier, Ann Arbor, Michigan, c.1983, p. 92).
14 See R. Pickvance, Gauguin and the Pont-Aven School (exh. cat.), Art Gallery of New South Wales, Sydney, 1994, cat. no. 5, repr. p. 33. 15 It was Children Wrestling, viewed by the author in the exhibition Gauguin and the Pont-Aven School at the Art Gallery of New South Wales, Sydney, in 1994 (see note 14 above), that initiated further analysis of Boys on a River Bank. 16 Paul Gauguin, letter to Vincent van Gogh, July 1888, cited in Pickvance, cat. no. 5. 17 For further discussion of Boys on a River Bank, see S. Dean, European Paintings of the 19th and Early 20th Centuries in the National Gallery of Victoria, Melbourne, 1995, p. 109.
15 It was Children wrestling, viewed by the author in the exhibition Gauguin and the Pont-Aven School at the Art Gallery of New South Wales, Sydney, in 1994 (see note 14 above), that initiated further analysis of Boys on a river bank.
16 Paul Gauguin, letter to Vincent van Gogh, July 1888, cited in Pickvance, cat. no. 5.
17 For further discussion of Boys on a river bank, see S. Dean, European Paintings of the 19th and Early 20th Centuries in the National Gallery of Victoria, Melbourne, 1995, p. 109.