Exploring a fifteenth-century garden: a restoration uncovers the past


In 1948 the National Gallery of Victoria acquired from the dealer Tomás Harris in London, on the recommendation of Sir Kenneth Clark, the then Adviser to the Helton Bequest, a fifteenth-century Italian panel painting titled The garden of love (acc. no. 1827/4). The panel was purchased as the work of an artist of the school of Pisanello, but has been offered no fewer than nine attributions since it first appeared in the literature in 1939 (see appendix 1). It is currently catalogued as the work of the Master of the Stories of Helen, an artist from the studio of Antonio Vivarini (c.14 15–c. 1476–84).1 See U. Hoff, European Paintings before 1800 in the National Gallery of Victoria, 4th edn, Melbourne, 1995, pp. 250–2. 

  

The Garden of love belongs to a decorative tradition, alive in fifteenth-century Italy but based in antiquity, within which painted panels, together with wall paintings and textiles (both painted and woven), were used as decorative schemes relating to furniture as well as to architectural spaces. These schemes might depict contemporary activities or historical themes, or they might have particular decorative functions. In the case of the Garden of love, we know of one other panel – a work without figures in its composition – that has a direct relationship to the Melbourne painting (this article will demonstrate that a ‘third’ panel, with two figures, is one and the same as the unpopulated picture). While various suggestions might be offered about the earliest setting in which 1827/4 was displayed, we can only really say that the painting might have belonged to a group consisting of two panels. 

Though 1827/4 has been in the collection of the National Gallery of Victoria for the past forty-six years, it has been on display only for brief periods. The panel’s physical condition has never been well understood, and after a number of attempts to treat the painting a full investigation of its condition began in 1984. The course of this investigation has led to a reassessment of the work. This article will look at 1827/4 as it is today, the changes that have taken place in the picture and how it might have looked before these changes occurred. The Garden of love we have before us now is very different from the painting originally executed. Some of the areas in which it has undergone change have been revealed through documentary evidence, while others remain a matter of speculation, based on our reading of the material evidence presented by the painting itself.

The painting as it appears today 

The composition we see today shows five figures grouped around a central fountain: a man and two women on one side and a woman and a man on the other; the scene occurs in front of an arbour of pink and white roses, which appears to fold in around the figures on either side (figs la & lb). But how much of what we see today dates from the fifteenth century, and how much of the present composition can be safely used to establish areas of comparison with other paintings, whether with regard to iconography or to authorship? 

The Garden of love is made from five planks of timber (spruce), butt-joined vertically to form a support that currently measures 152.5 x 239.0 cm (fig. 2). The vertical dimension is not the original height of the panel, there being clear evidence that the work has been cut down at the top and bottom. Though the left and right edges carry margins of timber about 25 mm wide that were originally left unpainted (they were overpainted in a restoration), the paint film at the top and bottom finishes in a fractured edge where the timber has been roughly sawn. Also notably, the right hand of the figure on the fountain has been cut through at its tip; the base of the bowl of the fountain runs into a curve that begins to turn down at the centre to form a stem but is suddenly terminated; and the figures appear in three-quarter length. Figures painted at three-quarter length are not common in fifteenth-century Italian paintings. When they do occur, they are a consequence of being situated behind painted architectural elements, as in the late- fourteenth-century frescos at the Palazzo Davanzati, Florence, or behind framing elements in large altarpiece settings.2 A number of writers make mention of the Davanzati fresco cycles. For a recent illustration, see A. B. Barriault, Spalliera Paintings of Renaissance Tuscany: Fables of Poets for Patrician Homes, University Park, Pennsylvania, 1994, p. 79. For examples of works by the Vivarini with figures cut by framing elements, see R. Pallucchini, I Vivarini (Antonio, Bartolomeo, Atvise), Venice, [1962], figs 2, 3, 4, 27. 

The panel is currently 5–7 mm thick. It has been thinned from an original thickness of perhaps 50 mm and was previously cradled in three parts. An X-radiograph of the painting also shows series of button infills, suggesting a previous attachment of a horizontal batten across the back.3 Nail holes, such as are indicated by the button infills here, are evident in another panel, also Italian, in the National Gallery of Victoria’s collection. Cola dall’ Amatrice’s The finding of the True Cross (acc. no. 3078/4), a work that consists of three horizontal planks, has nails, nail holes and score-lines on the reverse, suggesting the earlier presence of two vertical braces. This panel, which has retained its original thickness (22 mm), provides valuable evidence about a particular battening procedure.A major treatment to resolve inherent problems with the stability of the panel was carried out at the National Gallery of Victoria in 1992 (see appendix 3). 

 

                                                                                                                                                               

In addition to the changes affecting the dimensions of the panel, we must also take into account the loss of a figure from the image. Though reference has been made to the significance of the composition having five figures,4 See P. F. Watson, The Garden of Love in Tuscan Art of the Early Renaissance, Philadelphia, 1979, p. 123. the X-radiograph of the painting reveals that in the lower right corner a section of panel with an arched top has been removed and replaced with an insert; above the curved edge of this insert, we see in the radiograph a vermilion hat and part of the forehead of the figure that has been cut out of the painting (fig. 3). Clearly, the Garden of love was once populated by six people, and the hat would suggest a grouping of three men and three women; the original composition was probably, therefore, symmetrical in both the number and gender of its figures. The reconstruction of the corner has been through two stages (figs 5 & 1a) but, most significantly, it has involved a reworking of the lost area of the back of the garment of the woman on the right – the result of this restoration is wholly out of keeping with the treatment of the other draperies. 

 

                                                                                                                                                                As well as this reconstruction of the lower right-hand corner, there is a considerable amount of repair and restoration work in the paint layer of the Garden of love, something that is to be expected in a panel of this age. Some of these repairs will have been a response to problems inherent in the support, since the planks of timber are tangentially cut and will have been prone to warping and, if constrained, splitting, right from the start.5 The position of a plank of wood in relation to the circular growth ring structure of the tree from which it is cut has a fundamental influence on the shape it develops as it dries and changes its moisture content in relation to the moisture content of the air around it. This is a primary factor in the stability of wood panel paintings and, though it was often the case with northern panels that the selection of the plank took into consideration the orientation of the growth rings, it is common for Italian panels to show the opposite. The five planks that make up the support of 1827/4 vary in orientation: the two outer planks are close to quarter-sawn; the central plank has a grain orientation that predisposes it to concave curvature; and the planks on either side are prone to convex curvature. This pattern is evidence of the conventional wisdom of using alternating planks with different directions of warpage in an assembly made up of tangentially cut boards. In the case of a plank that has been tangentially cut, when shrinkage occurs between the growth rings the plank curves; for a plank that has been quarter-sawn, the shrinkage reduces the width but does not bring about curvature. When tangentially cut boards are restrained and cannot curve in response to shrinkage, the stress associated with the change in dimension is released through cracking, as has occurred in 1827/4. Approaches to dealing with the fundamental behaviour of pieces of wood have been many and varied, and form the basis of all decisions about the care and presentation of panel paintings. But among the interferences are some that are primary to our reading of the panel, and these need to be considered in detail. 

Perhaps the most significant is the reconstruction of or the brocades of the figures. The brocades we see today have been achieved not by applying conventional punched gold leaf, or impressed material, but by rubbing a dark paint into incised lines and then glazing over this with a layer of gold powder suspended in a transparent medium. The result of this method, which appears to be without precedent, gives the effect of punched decoration (fig. 4).6 The presence of elemental gold in the glaze was established by Deborah Lau-Greig of the National Gallery of Victoria, using SEM/EDS analysis. Not only must the reconstruction of the brocades have been laborious, but it must also have used materials that would have been costly and difficult to handle. A clear indication that the brocade work in 1827/4 is a relatively recent restoration has come with the discovery of a previously unpublished photograph, taken by Country Life magazine in 1934 when the painting was in the possession of an earlier owner, Mrs Crossley (fig. 5). In this photograph we see not only that the brocades were then different in style from those in the present work but also that there were more of them in place: what is now a plain area in the sleeve of the woman on the right, for example, was intact brocade in 1934. It is also worth comparing at this point the brocades of the figures in the three panels at the Walters Art Gallery, Baltimore, that have been associated with the Melbourne panel. The brocades in the Baltimore pictures offer a similar appearance to those in the Garden of love in 1934. Residues of the original brocade work are visible in the X-radiograph of 1827/4, and although the image is difficult to read we are nevertheless left with a sense that little of the original remains beneath the restored areas.

                                                                                                                                                           We can also see in the unpublished photograph from 1934, as well as in a photograph published that year in Country Life (fig. 6),7 See C. Hussey, ‘Burton Pynsent, Somerset, the Seat of Mrs. Crossley’ [Country Homes, Gardens Old & New], Country Life, 6 October 1934, fig. 13. I am indebted to Mr Burton Fredericksen and Ms Carol Togneri of the Getty Art History Information Program, Santa Monica, California, for drawing our attention to this photograph, and to Ms Camilla Costello, of the Country Life Picture Library, for notifying us of the existence of the unpublished photograph of the painting, held in the Library’s collections. The frame seen on the painting in 1934 is still with the work. Though it is a mixture of tabernacle and entablature styles, reflecting a kind of Pre-Raphaelite taste, it has been retained for historical reasons and because a clear sense of the original setting of the painting has yet to be established. that the balustrade in the painting was formerly reconstructed as a horizontal line; it was later reworked on the present diagonal, an orientation that obeys the perspective of the rest of the painting. We are not able at this stage to date this major restoration precisely, though the Garden of love first appeared on the art market in March 1939 and was published later that year, while in the possession of Tomás Harris, with the restoration having already taken place.8 See E. S. King, ‘The Legend of Paris and Helen’, Journal of the Walters Art Gallery, vol. II, 1939, fig. 9.

Among other aspects of the paint layer to be considered is the substantial overpainting of the roses of the arbour. Though we read the roses as pink, red and white, we need to acknowledge that the white flowers are substantially worn and that those that are pink, red are quite often the result of overpainting. The original construction of the roses is a subtle modelling of crimson glaze over a base of lead white, but only a few of the flowers painted in this way are available as reference points for comparison (fig. 7). 

We should not leave the painting as it is today without looking at the significant strengths in what remains of the original handling. The foliage of the Garden of love is made up of a delightful mixture of three-dimensional illusion and physical three-dimensionality (fig. 8). It has been built with overlapping layers of stems and leaves, painted in a viscous medium that has retained its thickness when dry. The flesh-tones and fountain, by contrast, have been painted in conventional egg tempera by means of tiny, closely worked strokes of colour. Remarkable areas of fine detail remain in the dragons on the fountain (fig. 9), as well as in the hair of the figures. For all the changes to its surface, the panel still carries areas of real enchantment. 

                 

                                                                               

Earlier Contexts

Various suggestions have been made as to the original scheme to which the Garden of love belonged, with most of the literature referring to the group of three panels of almost identical dimensions in the Walters Art Gallery, as well as to two other panels of very similar size.9 The references to the Garden of love in the literature are extensive (see, in particular. King, ‘The Legend of Paris and Helen’; C. L. Joost-Gaugier, Ά Rediscovered Series of Uomini Famosi from Quattrocento Venice’, Art Bulletin, vol. LVIII, no. 2, June 1976, pp. 184–95; F. Zeri, Italian Paintings in the Walters Art Gallery, vol. I, Baltimore, 1976, pp. 237–40; Watson, p. 165; Hoff, pp. 250–2). The present study hopes to shed new light on the historical relationship between these works. 

Some of the changes we have noted in the Melbourne painting can be explained by an association with the panels at the Walters Art Gallery, one of which, the Reception of Helen at Troy by Priam and Hecuba, carries an area of replaced timber in the lower left-hand corner that is exactly the same size as, and a mirror image of, the piece replaced in the Melbourne panel (fig. 10). Not only do the two replaced sections correspond exactly, but also, curiously, an area of pulp infill in the back of the Melbourne picture occurs similarly in the Walters panel.10 For technical descriptions of the Walters panels, see Zeri, pp. 241–2. I am also grateful to Mr Eric Gordon, Conservator of Painting at the Walters Art Gallery, for making reports on the paintings available. In addition, I had the opportunity to view the Baltimore panels at first hand in 1993, and to confirm the observations that are recorded in this article. These observations are compelling evidence that the Melbourne painting was once grouped with the Walters panels in an architectural setting. It is important here to note, however, that the latter works have not been cut down: two of them retain evidence of exposed timber margins on all sides, while the figures depicted are about a third the size of those in the Melbourne panel and are represented at full height.11 See Zeri, vol. I, p. 241. Therefore, although all four panels were almost certainly brought together at some stage, they did not form a group in the first instance. The Melbourne panel must have been cut down to match the height of the Walters panels, and the lost figure at the right would then be a consequence of the assembly of all four panels within an architectural setting.

 

                                                                                                                                                             In reconstructing possible groupings in which the Melbourne panel may originally have been located, reference has been made to two panels of nearly identical dimensions, both of which feature an arbour with a triangulated top edge. In one of these works there are two figures (fig. 11), while in the other there are none (fig. 12). The earliest published reference to these pictures was in 1939, when King noted the panel with two figures in a sale at Christie’s, London, on 9 June 1939 (lot 110).12 King, p. 68, writes: ‘Judging from a photograph, this picture has been decidedly retouched, especially the two figures. Otherwise it corresponds closely in style with [1827/4]’. King also published a photograph of 1827/4, then with Tomás Harris,13 ibid., fig. 9. and mentioned another panel, without figures, that Harris had informed him was in his possession.14 ibid., p. 68 n. 15. The latter work appeared at Sotheby’s, London, on 7 July 1954 (lot 98).15 The picture was listed in the Sotheby’s sale as Ά Hedge of Roses on a Trellis’, by Antonio Vivarini. Its measurements were given as 60″x 91 1/2″. In 1976, Zeri stated that the panel with two figures and the panel without figures were in fact the same work, with the man and woman overpainted in the second picture.16 Zeri, vol. I, p. 237. Watson, p. 165 n. 3, like King, refers to two companion panels to 1827/4, citing the unpopulated picture (which he knew from a photograph in the Frick Art Reference Library, New York) and a ‘third panel’, at Christie’s, London, on 9 June 1939 (the picture with two figures).

 

More recently, the unpopulated panel again appeared on the art market, at Sotheby’s, New York, on 10 January 1991 (lot 9), having been in the possession of Maximilian Etchecopar of Buenos Aires;17 At Sotheby’s, the picture was described as ‘Garden Scene with Roses Growing up Trellises: The Garden of Love’ (tempera on panel, 152.5 x 233.5 cm, attributed to the Master of the Stories of Helen (early 1440s–c.l470)). it was then acquired by the Venice art dealer Pietro Scarpa,18 While with Scarpa, the panel reappeared on the art market, in an advertisement for the 11th International Antiques Fair, Milan. It was listed as ‘A Hedge of Roses in a Garden of Love’ and was attributed to Antonio Vivarini or Antonio da Negroponte (Apollo, vol. CXXXV, no. 362, April 1992, suppl. p. 24). who has since made available an X-radiograph of the panel (fig. 13). While there is no evidence of figures in the radiograph, if we study the photographs of the Scarpa picture and the composition with two figures we see a direct correspondence in the leaves of the upper edge of the arbour, an identical positioning of the flowers and a correspondence in the joins and cracks in the panels. Instead of the figures having been overpainted, therefore, they have been removed at some time between the Christie’s sale in June 1939 and King’s communication with Tomás Harris – the two panels are indeed one and the same.19 The changes to this panel, like those to 1827/4, can be dated to 1939. The Crossley panel (later to become the Melbourne panel) was sold in March that year and the panel with two figures was sold in June. Both works are referred to by King in his 1939 article; the Melbourne panel was at this point recorded as with Tomás Harris, the work without figures also being noted as with Harris but not available as a photograph. We can therefore assume that Harris had had major restoration done on the latter panel in the few months between the June sale and the publication of King’s article (with its reference to a panel without figures), and on 1827/4 between the March sale and the appearance of King’s photograph and description. Whether it is any longer possible to establish the originality or otherwise of the two lost figures is open to question. 

The Scarpa panel has not been examined directly by the author; however, a former colleague from the National Gallery of Victoria, Ms Jane Clark, had the opportunity to view it in 1993. It is possible that the panel did not form part of the original grouping to which 1827/4 belonged, but that it was made up at a later date to complete a setting (perhaps the setting that included the Walters panels).20 Dr Scarpa does not believe that his panel has been cut down. In addition to the dimensions and subject matter of the Melbourne and Scarpa panels, there is some further evidence in the X-radiographs of these works to draw them together: the sky in both pictures has been brushed vigorously with a layer of lead white to create the triangulated form of the upper edge of the arbour, this being followed later by the leaves (figs 14 & 15). Stylistically, then, the arbours have obvious links, but how securely the two panels can be linked together may rely on comparing them directly, since they have both been through a number of extensive restorations.

The Original Construction 

Reconstructing the likely original form of the Garden of love is made possible by a careful examination of the material and archival evidence before us. 

 

As previously noted, the panel consists of five vertical planks of timber, still at their original width; however, given that the panel has been cut down, we should allow for the figures once to have been full-length, and for the panel formerly to have had a height closer to 190 cm. The ground layer of the painting has been identified as hydrated calcium sulphate and appears under magnification to consist of the two layers suggested in the fifteenth century by Cennino: gesso grosso and gesso sottile.21 Cennino d’Andrea Cennini, Il Libro dell’Arte, trans. D. V. Thompson, New York, 1960. For a detailed discussion of the preparation of Italian panel paintings, see Art in the Making: Italian Painting before 1400, London, 1989. A distinction is commonly drawn between the northern European use of calcium carbonate (chalk) as a filler in grounds and the southern European use of calcium sulphate (gypsum, gesso). Further distinctions are made between gesso grosso and gesso sottile, and then between the levels of hydration of the calcium sulphate form present in the ground. For a considerably more involved discussion of the differences between the forms in which calcium sulphate was used in grounds in Italian pictures, and for a tabulation of works according to level of hydration, see R. J. Gettens & Μ. E. Mrose, ‘Calcium Sulphate Minerals in the Grounds of Italian Paintings’, Studies in Conservation, vol. 1, no. 4, 1954, pp. 174–89. Deborah Lau-Greig, of the National Gallery of Victoria, has carried out thermal analysis on the ground in the Garden of Love and has found it to be the hydrated form of calcium sulphate. However, questions have been raised about the level of hydration remaining after processing of the mineral and its reuse (involving the addition of water) to form the gesso, and the issue requires further clarification. Of additional interest in the present discussion is the nature of the materials used in the Garden of love, with particular reference to the foliage. Further detailed analysis of the materials used in the painting is currently in progress, and the distinctions between the materials and techniques of northern and southern Europe in the fifteenth century are of particular interest here. In the past there has been a tendency to draw clear distinctions between the techniques practised in north and south – egg tempera with a gesso ground in Italian works and an oily medium with a chalk ground in Flemish pictures – but the reality is not as clear-cut, certainly not by the end of the fifteenth century. In the case of the Garden of love there may be some logic to hypothesizing an association of northern and southern technique, with the source of this approach to be found in the structure of the Vivarini studio, and in the individuals at work within it. The image commences with drawing, which we can see at its most detailed in the fountain (fig. 16) (perhaps suggesting that its origins lie in a copy-book) and also in pentimenti in the skirt of the figure on the fountain. The drawing appears to have been put down with a brush, and while it is extensive in the fountain it has not been used to define the edges of the foliage. On the other hand, it most likely has been employed to articulate the major forms of the figures and the architecture. 

 

                                                                                                                                                                                                                                          

The development of the paint layer begins with the sky, which as we have noted has been brushed vigorously into the triangulated form that defines the positioning of the foliage (fig. 14); however, the sky does not extend behind the figure on the fountain. The sticks that form the structure of the trellis have been put in place next, followed by the background green, which has been painted up to the edges of the trellis, the fountain and the figures. The first leaves and stems have then been added, with up to three layers of overlapping foliage being developed. Many of the flowers have been painted on top of the foliage, with only minimal reference to the arrangement of the stems and leaves. It is not clear whether the fountain or the figures were next, but it is likely that the figures were the last major feature of the painting. 

The painting has been built in a progressive and sequential way – clearly allowing room for more than one hand to have been involved, in a studio production. Even so, there is evidence of the picture’s having been pulled together with final adjustments. We can see in the area around both hands of the woman on the right, for example, that the reserves left for the flesh-tones were larger than the hands that eventually went into them, and that the resulting gaps around the hands have had to be filled in (fig. 17). In addition, the dark shadows under the trellis were painted after the figures but before the thorns on the stems (fig. 18), and there have been adjustments to the hair of the woman to the left of the fountain, as well as to the hat of the man behind her. 

A conception of the painting in its original form is offered in figure 19. The reconstruction has been made by overlapping the existing composition with tracings of garments and figures from the Walters panels – which correspond closely with the forms in the Melbourne panel – in order to extend the existing five figures and to replace the one that has been lost at the right. The fountain has also been extended from the curve that remains at the base of the present bowl, suggesting the top of a stem. The rest of the base, the tiled foreground, and the choice of figure for the man on the right are the author’s invention.   

 

 When we consider the symmetry of the original composition of six figures, we can also identify the axes around which the image is constructed. The original composition would have been symmetrical not only about the vertical axis but also about the horizontal: the lower of the two fruit at the centre of the fountain defines the centre of the image. A simple, but nevertheless convincing, perspective grid for the painting can be established once we have this knowledge. The perspective of the painting in its present form is awkward when the work is viewed set low, but we should consider it placed on the wall at a height that allows us to situate ourselves well forward on the plane on which the full-length figures would stand, with the fountain seeming to tower above.22 For an analysis of the three registers of Italian decorative painting – cassoni, spalliere and cornice – see Barriault, Spalliera Paintings of Renaissance Tuscany. The scale of the Garden of Love, and its perspective construction, give credence to the notion that it is a picture that was intended to be situated in the cornice register. 

Though this study reveals that the panel at the National Gallery of Victoria has more to tell than could have been known when the painting was acquired, and that there remain areas where our knowledge is still incomplete, it is hoped that it will nevertheless offer a view of the picture that can reliably be used for comparison and further study. It can sometimes be difficult for us to accept the kinds of changes in a painting that are considered here, and we therefore need to be confident that our conclusions are based in the material substance of the work before us. For all its problems, however, the Garden of love remains an enchanting and curious example of an area of artistic activity that is not always well understood and that warrants our continued attention. 

John Payne, Senior Painting Conservator, National Gallery of Victoria (in 1994).    

Appendix 1 

Chronological history of attributions proposed for 1827/4 

1939    CARLO CRIVELLI (Wooley & Wallis, 22–24 March 1939 (lot 475)).

School of Antonio Vivarini (E. S. King, ‘The Legend of Paris and Helen’, Journal of the Walters Art Gallery, vol. II, 1939, p. 67 n. 15; King accepts Berenson’s earlier attribution of the panels at the Walters Art Gallery, Baltimore (see appendix 2), to the studio of Vivarini (p. 55), and states that he believes 1827/4 to have come from the same workshop (p. 67 n. 15))

1946     ANTONIO ROSSA DA CADORE (R. Longhi, Viatico per cinque secoli di pittura, Venice, 1946, pp. 46–7). 

1948     SCHOOL OF PISANELLO (Catalogue of the National Gallery of Victoria, rev. edn, Melbourne, 1948, p. 114). 

1949     BENEDETTO BEMBO (C. L. Ragghianti, in La critica d’arte, vol. VIII, 1949, p. 298 n. 29). 

1953     DARIO DA PORDENONE (L. Coletti, Pittura veneta del quattrocento, Storia della pittura italiana, Novara, 1953, p. 84). 

1957     STUDIO OF ANTONIO VIVARIN (Β. Berenson, Italian Pictures of the Renaissance: A List of the Principal Artists and Their Works with an Index of Places – Venetian School, vol. I, rev. edn, London, 1957, p. 198). 

1961     SCHOOL OF ANTONIO VIVARINI (U. Hoff, Catalogue of European Paintings before Eighteen Hundred, National Gallery of Victoria, Melbourne, 1961, pp. 135–7). 

1962     MASTER OF THE STORIES OF HELEN (R. Pallucchini, I Vivarini (Antonio, Bartolomeo, Alvise), Venice, [1962], p. 86; Pallucchini gives the name ‘Master of the Stories of Helen’ to an unknown artist of the immediate circle of Vivarini, working in the 1460s or 1470s).

NORTH ITALIAN (M. Levey, ‘National Gallery of Victoria, Melbourne: Catalogue of European Paintings before Eighteen Hundred by Ursula Hoff’ [Book Reviews], Museums Journal, vol. 62, no. 2, September 1962, p. 125). 

1968     DARIO DA PORDENONE (Laura Marussigh, letter to Dr Ursula Hoff, 27 December 1968 (National Gallery of Victoria files)). 

1976     MASTER OF THE STORIES OF HELEN (F. Zeri, Italian Paintings in the Walters Art Gallery, vol. I, ed. U. E. McCracken, Baltimore, 1976, p. 237). 

STUDIO OF ANTONIO VIVARINI (C. L. Joost-Gaugier, ‘A Rediscovered Series of Uomini Famosi from Quattrocento Venice’, Art Bulletin, vol. LVIII, no. 2, June 1976, p. 193). 

1979     ANTONIO DA NEGROPONTE (P. F. Watson, The Garden of Love in Tuscan Art of the Early Renaissance, Philadelphia, 1979, p. 122). 

1988     ANTONIO DA NEGROPONTE (Dr Mauro Lucco, Università degli Studi di Bologna, conversation with the author, 1988; Dr Lucco viewed 1827/4 in Melbourne and said he believed the Walters panels to be by the same hand). 

1989     MASTER OF THE STORIES OF HELEN (P. Tomory & R. Gaston, Summary Catalogue: European Paintings before 1800 in Australian and New Zealand Public Collections, Sydney, 1989, no. 249). 

1995     STUDIO OF ANTONIO VIVARINI (Master of the stories of Helen) (U. Hoff, European Paintings before 1800 in the National Gallery of Victoria, 4th edn, Melbourne, 1995, pp. 250–2).    

Appendix 2  

Table of panels relevant to this discussion   

Six paintings relevant to this discussion are described in the literature: the three panels at the Walters Art Gallery, Baltimore, which show scenes from the Legend of Paris and Helen; and three paintings in the Garden of love series. The present whereabouts and other basic data pertaining to each of these panels are provided belowTable of panels relevant to this discussion

(image of table pending)

Appendix 3  

Physical structure and structural treatments to May 1992  

     

The central planks of the panel are not quarter-sawn and must have been unstable from the beginning: it is likely that the panel cracked or showed signs of stress very early in its life. The first major interference to the structure of the panel must have involved cutting it down at top and bottom and removing a section at the lower right-hand corner. The most probable date for the second major interference is early this century, when the painting moved to England. It is likely that what was received in England was a panel that had been cut down at the top and bottom, had a large section cut out at the bottom right, and was possibly cracked in a number of places. We have no documentation to support this view; however, we do know what action was taken with respect to the panel at this time.     

The company of W. Morril, a noted lining and cradling exponent who at times worked on items from the collections of the London National Gallery, undertook a structural repair of the panel, which was first divided into three sections by separating it at the joins at either side of the central plank. Each section was then thinned to between 5 and 7 mm, the missing section at the lower right was replaced and three separate closed cradles were applied to the reverse. The timber used for the cradling was mahogany. The three separated sections were then reassembled along the join lines. The edges of the joins were planed to create a good surface for gluing, and as a result some of the paint layer was removed (in the X-radiograph of 1827/4 we can see a step in the rim of the base of the fountain). In addition, the pieces were reassembled out of plane, leaving an interval along both joins that later had to be filled in and overpainted. There is evidence, in the form of a large area of sawdust, pulp and glue infill on the reverse of the two planks of the left side, that the thinning process was problematic; when the panel’s cradle was removed in 1992 it was found that the saw used in the thinning appeared to have dug deeply into the panel at the point where the infill is now located, causing an area of radical thinning. Curiously, a similar area occurs in the Reception of Helen at Troy, the Walters panel with the cut-out corner (see appendix 2).   

The recent treatment of the panel began in 1979 with the cleaning of the sky. A change in staff at the National Gallery of Victoria in 1984 led to a reassessment of the treatment, and a thorough examination of the painting. The cradling on the work was removed at the Gallery in 1992, revealing the impressed mark of W. MORRIL on the horizontal battens (fig. 20). A parquet of balsawood blocks was then attached to the reverse. The removal of the cradling was undertaken as a workshop with Bettina Jessell of the Art Restoration Centre, Washington DC, whose visit was funded by the Business Council of the National Gallery of Victoria and The Art Foundation of Victoria.   

 

The choice of the balsa backing was made after extensive deliberation, and consideration of diverse and sometimes contradictory opinions on the best method of dealing with such a large and yet very thin panel. Since 1992, the success of the treatment has been constantly under examination. After a year the painting took on an undulating curvature that corresponds to the projected intention of the original structure. The central plank has assumed a concave curve, while those either side have a convex curvature. 

The treatment of 1827/4 has continued since 1992, with selective varnish removal, reduction of overpainting, reworking of old repairs and retouching of exposed damages. The painting has been varnished with MS2A and retouched with pigments in Paraloid B72.   

Notes 

1     See U. Hoff, European Paintings before 1800 in the National Gallery of Victoria, 4th edn, Melbourne, 1995, pp. 250–2.

 2     A number of writers make mention of the Davanzati fresco cycles. For a recent illustration, see A. B. Barriault, Spalliera Paintings of Renaissance Tuscany: Fables of Poets for Patrician Homes, University Park, Pennsylvania, 1994, p. 79. For examples of works by the Vivarini with figures cut by framing elements, see R. Pallucchini, I Vivarini (Antonio, Bartolomeo, Alvise), Venice, [1962], figs 2, 3, 4, 27. 

3     Nail holes, such as are indicated by the button infills here, are evident in another panel, also Italian, in the National Gallery of Victoria’s collection. Cola dall’ Amatrice’s The Finding of the True Cross (acc. no. 3078/4), a work that consists of three horizontal planks, has nails, nail holes and score-lines on the reverse, suggesting the earlier presence of two vertical braces. This panel, which has retained its original thickness (22 mm), provides valuable evidence about a particular battening procedure. 

4     See P. F. Watson, The Garden of Love in Tuscan Art of the Early Renaissance, Philadelphia, 1979, p. 123. 

5     The position of a plank of wood in relation to the circular growth ring structure of the tree from which it is cut has a fundamental influence on the shape it develops as it dries and changes its moisture content in relation to the moisture content of the air around it. This is a primary factor in the stability of wood panel paintings and, though it was often the case with northern panels that the selection of the plank took into consideration the orientation of the growth rings, it is common for Italian panels to show the opposite. The five planks that make up the support of 1827/4 vary in orientation: the two outer planks are close to quarter-sawn; the central plank has a grain orientation that predisposes it to concave curvature; and the planks on either side are prone to convex curvature. This pattern is evidence of the conventional wisdom of using alternating planks with different directions of warpage in an assembly made up of tangentially cut boards. In the case of a plank that has been tangentially cut, when shrinkage occurs between the growth rings the plank curves; for a plank that has been quarter-sawn, the shrinkage reduces the width but does not bring about curvature. When tangentially cut boards are restrained and cannot curve in response to shrinkage, the stress associated with the change in dimension is released through cracking, as has occurred in 1827/4. Approaches to dealing with the fundamental behaviour of pieces of wood have been many and varied, and form the basis of all decisions about the care and presentation of panel paintings. 

6     The presence of elemental gold in the glaze was established by Deborah Lau-Greig of the National Gallery of Victoria, using SEM/EDS analysis. Not only must the reconstruction of the brocades have been laborious, but it must also have used materials that would have been costly and difficult to handle. 

7      See C. Hussey, ‘Burton Pynsent, Somerset, the Seat of Mrs. Crossley’ [Country Homes, Gardens Old & New], Country Life, 6 October 1934, fig. 13. I am indebted to Mr Burton Fredericksen and Ms Carol Togneri of the Getty Art History Information Program, Santa Monica, California, for drawing our attention to this photograph, and to Ms Camilla Costello, of the Country Life Picture Library, for notifying us of the existence of the unpublished photograph of the painting, held in the Library’s collections. The frame seen on the painting in 1934 is still with the work. Though it is a mixture of tabernacle and entablature styles, reflecting a kind of Pre-Raphaelite taste, it has been retained for historical reasons and because a clear sense of the original setting of the painting has yet to be established. 

8     See E. S. King, ‘The Legend of Paris and Helen’, Journal of the Walters Art Gallery, vol. II, 1939, fig. 9. 

9    The references to the Garden of Love in the literature are extensive (see, in particular. King, ‘The Legend of Paris and Helen’; C. L. Joost-Gaugier, ‘A Rediscovered Series of Uomini Famosi from Quattrocento Venice’, Art Bulletin, vol. LVIII, no. 2, June 1976, pp. 184–95; F. Zeri, Italian Paintings in the Walters Art Gallery, vol. I, Baltimore, 1976, pp. 237–40; Watson, p. 165; Hoff, pp. 250–2). 

10    For technical descriptions of the Walters panels, see Zeri, pp. 241–2. I am also grateful to Mr Eric Gordon, Conservator of Painting at the Walters Art Gallery, for making reports on the paintings available. In addition, I had the opportunity to view the Baltimore panels at first hand in 1993, and to confirm the observations that are recorded in this article. 

11    See Zeri, vol. I, p. 241.

 12    King, p. 68, writes: ‘Judging from a photograph, this picture has been decidedly retouched, especially the two figures. Otherwise it corresponds closely in style with [1827/4]’. 

13    ibid., fig. 9. 

14    ibid., p. 68 n. 15. 

15    The picture was listed in the Sotheby’s sale as ‘A Hedge of Roses on a Trellis’, by Antonio Vivarini. Its measurements were given as 60″x 91 1/2″.

16    Zeri, vol. I, p. 237. Watson, p. 165 n. 3, like King, refers to two companion panels to 1827/4, citing the unpopulated picture (which he knew from a photograph in the Frick Art Reference Library, New York) and a ‘third panel’, at Christie’s, London, on 9 June 1939 (the picture with two figures). 

17    At Sotheby’s, the picture was described as ‘Garden Scene with Roses Growing up Trellises: The Garden of Love’ (tempera on panel, 152.5 x 233.5 cm, attributed to the Master of the Stories of Helen (early 1440s–c.l470)). 

18    While with Scarpa, the panel reappeared on the art market, in an advertisement for the 11th International Antiques Fair, Milan. It was listed as ‘A Hedge of Roses in a Garden of Love’ and was attributed to Antonio Vivarini or Antonio da Negroponte (Apollo, vol. CXXXV, no. 362, April 1992, suppl. p. 24). 

19    The changes to this panel, like those to 1827/4, can be dated to 1939. The Crossley panel (later to become the Melbourne panel) was sold in March that year and the panel with two figures was sold in June. Both works are referred to by King in his 1939 article; the Melbourne panel was at this point recorded as with Tomás Harris, the work without figures also being noted as with Harris but not available as a photograph. We can therefore assume that Harris had had major restoration done on the latter panel in the few months between the June sale and the publication of King’s article (with its reference to a panel without figures), and on 1827/4 between the March sale and the appearance of King’s photograph and description. 

20    Dr Scarpa does not believe that his panel has been cut down. 

21    Cennino d’Andrea Cennini, Il Libro dell’Arte, trans. D. V. Thompson, New York, 1960. For a detailed discussion of the preparation of Italian panel paintings, see Art in the Making: Italian Painting before 1400, London, 1989. A distinction is commonly drawn between the northern European use of calcium carbonate (chalk) as a filler in grounds and the southern European use of calcium sulphate (gypsum, gesso). Further distinctions are made between gesso grosso and gesso sottile, and then between the levels of hydration of the calcium sulphate form present in the ground. For a considerably more involved discussion of the differences between the forms in which calcium sulphate was used in grounds in Italian pictures, and for a tabulation of works according to level of hydration, see R. J. Gettens & Μ. E. Mrose, ‘Calcium Sulphate Minerals in the Grounds of Italian Paintings’, Studies in Conservation, vol. 1, no. 4, 1954, pp. 174–89. Deborah Lau-Greig, of the National Gallery of Victoria, has carried out thermal analysis on the ground in the Garden of Love and has found it to be the hydrated form of calcium sulphate. However, questions have been raised about the level of hydration remaining after processing of the mineral and its reuse (involving the addition of water) to form the gesso, and the issue requires further clarification. Of additional interest in the present discussion is the nature of the materials used in the Garden of Love, with particular reference to the foliage. Further detailed analysis of the materials used in the painting is currently in progress, and the distinctions between the materials and techniques of northern and southern Europe in the fifteenth century are of particular interest here. In the past there has been a tendency to draw clear distinctions between the techniques practised in north and south – egg tempera with a gesso ground in Italian works and an oily medium with a chalk ground in Flemish pictures – but the reality is not as clear-cut, certainly not by the end of the fifteenth century. In the case of the Garden of Love there may be some logic to hypothesizing an association of northern and southern technique, with the source of this approach to be found in the structure of the Vivarini studio, and in the individuals at work within it.

 22    For an analysis of the three registers of Italian decorative painting – cassoni, spalliere and cornice – see Barriault, Spalliera Paintings of Renaissance Tuscany. The scale of the Garden of Love, and its perspective construction, give credence to the notion that it is a picture that was intended to be situated in the cornice register.