fig. 1 
Flanders

For ninety years the National Gallery of Victoria has been home to one of the most intriguing of all early Netherlandish paintings. In 1922 the small panel painting Virgin and Child (fig. 1) made headlines when it was purchased by the Felton Bequest on the advice of London-based Felton adviser Frank Rinder. The work’s fame rested on the longstanding belief that it was an important work by Jan van Eyck, one of the greatest of early Netherlandish painters. The painting was acquired from the Weld-Blundell family, who had owned it since the early nineteenth century, and to this day it is still often referred to as the ‘Ince Hall Madonna’, after the Weld-Blundell residence.

Virgin and Child made headlines again some thirty-seven years later, when the Gallery re-attributed it to ‘Follower of van Eyck’. The work’s authorship was downgraded following an extensive examination and subsequent conservation treatment at the Institut Royal du Patrimoine Artistique (KIK-IRPA), Brussels, in the late 1950s. The change was prompted by conclusions reached from a series of examinations conducted by Paul Coremans and Martin Davies, along with colleagues from both the KIK-IRPA and the National Gallery in London, which identified numerous characteristics in the painting they considered ‘unEyckian’.1Ursula Hoff & Martin Davies, The National Gallery of Victoria, Melbourne, Centre National de Recherches Primitifs Flamands, Brussels, 1971, pp. 29–40. Their published findings, as well as subsequent work by other researchers, identified inconsistencies of style and execution which do not square with the master’s manner.2Ursula Hoff, European Paintings before 1800 in the National Gallery of Victoria, National Gallery of Victoria, Melbourne, 1995, pp. 103–6. Their conclusion was that the painting was a copy of a lost original by Jan van Eyck.

In recent years, however, some of the findings of the KIK-IRPA’s examination have been called into question, along with the conservation treatment undertaken by Albert Philippot in 1957–58. Some scholars have been open to the idea of returning the painting’s attribution to Jan van Eyck or his assistants.3Hugh Hudson has argued that due to flaws in the technical findings by the committee, the KIK-IRPA report cannot be used as evidence against an attribution to Jan van Eyck or his studio. Furthermore, Hudson argues that the painting can be justifiably attributed to the artist or his studio based on technical evidence, composition and iconography. See Hugh Hudson, Jan van Eyck: The Ince Hall Virgin and Child and the Scientific Examination of Early Netherlandish Painting, VDM Verlag Dr. Müller, Saarbrücken, 2009, pp. 3–4. The work was included in major exhibitions of early Netherlandish painting in Madrid (2001) and Bruges (2002), where it was tentatively attributed to one of van Eyck’s assistants.4Mauro Natale (ed.), El Renacimiento Mediterráneo, Museo Thyssen-Bornemisza, Madrid, 2001, pp. 264–9; Till-Holger Borchert (ed.), The Age of Van Eyck, Thames & Hudson, London, 2002, p. 237.

In the years since the de-attribution, some additional technical examination has provided new insights into the painting.5A summary of recent technical examination is found in John Payne & Carl Villis, ‘Aspectos Técnicos de La Virgen con el Niño de Ince Hall’, in Natale, pp. 266–7. Infra-red reflectograms of the paint surface have revealed greater detail in the underdrawing and minor reworkings, suggesting that the painting is an original composition and not a direct copy of another work.6An examination of recent infra-red reflectography was reported in Hugh Hudson’s Masters thesis, ‘Reexamining van Eyck: a new analysis of the Ince Hall Madonna’, University of Melbourne, 2000, and in Jan van Eyck, 2009, pp. 86–93. Dendrochronological analysis of the oak panel was carried out in 1999 but was not able to match the tree growth rings with dated master chronologies, preventing an estimated felling date for the wood of the panel.7Art Bulletin of Victoria, National Gallery of Victoria, vol. 40, 1999, pp. 44–7. Analysis of the painting’s ground layer through Scanning Electron Microscopy/Energy Dispersive Spectroscopy (SEM/EDS) has confirmed that a chalk (calcium carbonate) preparation was used by the artist.8SEM/EDS analysis of the ground layer was carried out by Deborah Lau, Analytical and Conservation Scientist, CSIRO Australia, in November 2000. Most recently, radiography specialist Roger Marijnissen has described the surface craquelure – which has a horizontal pattern running contrary to the vertical grain of the wood – as a ‘dubious’ feature.9Roger H. Marijnissen, The Masters’ and the Forgers’ Secrets: X-Ray Authentication of Paintings, Mercatorfonds, Brussels, 2009, pp. 186–7.

Despite the fact that certain key aspects of the 1957–58 technical examination have been undermined by some of the recent findings, the National Gallery of Victoria has continued to view the Virgin and Child as a work by a follower of Jan van Eyck; not as a copy, but rather as an imitation made later in the fifteenth century. While the 1957–58 technical report has justifiably been challenged, it was never the sole foundation for the de-attribution. In his report accompanying the KIK-IRPA conclusions, Davies emphasised that an attribution of the panel to Jan van Eyck was not excluded by the technical findings, but from a general lack of quality in execution.10Hoff & Davies, p. 47. He neatly summarises the criteria by which any Eyckian work must be assessed: that for a work to be by van Eyck it must do more than ‘give a good general effect, but also pass a rigorous examination of details’.11ibid., p. 46. The particular details of the Melbourne panel reveal qualities that are foreign to the body of securely attributed paintings by van Eyck, suggesting a later date of execution and betraying other sources of influence beyond that of the Bruges painter. The three key aspects examined in this paper are the prevalence of details which appear to derive from securely attributed paintings by Jan van Eyck; the absence of van Eyck’s characteristic spatial planning; and the presence of anachronistic and anomalous visual components.

Similarities to securely dated works by van Eyck

Without question, the NGV’s Virgin and Child bears unmistakable similarities to several of the signed and accepted paintings by the early Netherlandish master. In 1953, before the panel’s reassessment had taken place, Erwin Panofsky analysed a group of widely accepted works by van Eyck, including the unfinished Antwerp Saint Barbara (fig. 3), the Bruges Virgin and Child with Canon van der Paele, St Donatian and St George, the Dresden triptych with Virgin and Child Enthroned, the Frankfurt Virgin and Child Enthroned (Lucca Madonna) (fig. 2), the Washington Annunciation, the Antwerp Virgin at the fountain and the Louvre’s Virgin and Child with Chancellor Rolin, all works spanning the last decade of van Eyck’s life (1431–41).12Saint Barbara, 1437 (Koninklijk Museum voor Schone, Antwerp); Virgin and Child with Canon van der Paele, St Donatian and St. George, 1434–36 (Groeningemuseum, Bruges); the Dresden triptych with Virgin and Child Enthroned, 1437 (Gemäldegalerie Alte Meister, Staatliche Kunstsammlungen Dresden); Virgin and Child Enthroned (Lucca Madonna), c.1437–38 (Städel Museum, Frankfurt am Main); Annunciation, c.1434–36 (National Gallery of Art, Washington, DC); Virgin at the fountain, 1439 (Koninklijk Museum voor Schone Kunsten, Antwerp); Virgin and Child with Chancellor Rolin, c.1430–34 (Louvre, Paris). Erwin Panofsky, Early Netherlandish Painting, Harvard University Press, Cambridge MA, 1953, pp. 191–4. Panofsky attempted to establish a rational chronological sequence to the group by cross-referencing their painting styles and visual motifs against those of the securely dated works, beginning with the ‘early’ Melbourne Virgin, proceeding on the now-discredited inscription on the painting indicating 1433 as its date of execution.13The very old, possibly original painted inscription is inconsistent with van Eyck’s style of lettering and abbreviating. See Hoff & Davies, pp. 45–6. Hudson argues that the signature could be viewed as being by van Eyck despite acknowledging that the spelling ‘EYC’ and the inclusion of ‘BRVGIS’ is not found in his signed work (Jan van Eyck, pp. 150–2). Using this process, Panofsky pointed out several features that the Melbourne panel shares with the others. He referred to similarities between the folds of the Virgin’s mantle and those in the Rolin Virgin, now considered a later work.14Panofsky, p. 193. He also acknowledged the repeated motif of the open Book of Hours in the Antwerp Saint Barbara, another later work, as well as compositional patterns the Melbourne painting shares with the centre panel of the Dresden triptych, 1437, and the Frankfurt Virgin and Child Enthroned, c.1434, a relatively early work.15ibid., pp. 184–5. To reconcile the conflicting compositions and motifs between the group of securely dated works  – including the Melbourne panel, on account of the now-discredited inscription – Panofsky was forced to construct a somewhat uncomfortable sequence of stylistic evolution, including, on account of the Melbourne painting, an ‘archaic’ period between the richly decorative and finely executed paintings, such as the Washington Annunciation (which he dated 1428–29) and the Bruges van der Paele panel, 1436; a different phase to the ‘deliberately archaic’ style of the late Virgin at the fountain, 1439, in Antwerp.16ibid., pp. 183, 193. What concerns us here is not so much Panofsky’s fully cogent rationale in constructing a chronological sequence, but his observation that the Melbourne panel shares an unusually high number of features found in known works by van Eyck.

Indeed, further motifs found in the Melbourne panel can be identified in other van Eyck paintings. The overall composition, with the central placement of the seated Virgin and Child over a Persian carpet with a baldachin behind them, in an interior with a window to the left, derives from the Frankfurt Virgin and Child Enthroned (Lucca Madonna) panel (fig. 2). Further motifs, such as the candelabrum, brass laver and glass bottle, also quote the Frankfurt Virgin. The cropped and foreshortened side window appears in the Arnolfini double portrait of 1434 (fig. 4), along with the bare wood floor.17These features appear to be present in the lost Woman at her bath, reproduced in Willem van Haecht’s The Constcamer of Cornelius van der Geest, 1628 (Rubens House, Antwerp). The NGV Virgin’s distinctive red mantle, which spills out over most of the floor in the foreground of the painting, creates a broad pyramidal shape comparable in its form, generous size and turn of folds to the Antwerp Saint Barbara (fig. 3).18Panofsky, p. 185. The torso, shoulders and arms of the Christ Child derive from that of the Child in the central panel of the Dresden triptych (figs 5 & 6); the legs from the van der Paele panel; and the head possibly from the now-lost Van Maelbeke Virgin.19The Van Maelbeke Virgin, formerly of the church of St Martin’s in Ypres, has been lost since the 1790s and is only known to us by later drawings, such as the one in the Germanisches Nationalmuseum, Nuremberg. See Susan Jones, ‘New evidence for the date, function and historical significance of Jan van Eyck’s “Van Maelbeke Virgin”’, Burlington Magazine, vol. 148, no. 1235, Feb. 2006, pp. 73–81. The face of the NGV’s Virgin closely resembles that of the Virgin in the Antwerp Virgin at the fountain (figs 7 & 8).

The fact that this relatively simple image contains so many familiar compositional elements from other paintings places it at odds with each of the accepted autograph works by van Eyck. If we were to examine any other van Eyck Marian painting in search of a similar tendency to directly quote and repeat key compositional elements from other works, we would find very little beyond the discreet re-use of design motifs found in the architecture, landscape and fabrics.20Susan Jones, ‘The use of patterns by Jan van Eyck’s assistants and followers’, in Susan Foister, Sue Jones & Delphine Cool (eds), Investigating Jan van Eyck, Brepols, London & Turnhout, 2000, pp. 197–207; see also Lisa Monnas, Merchants, Princes and Painters, Yale University Press, New Haven & London, pp. 121–3. The securely attributed works show an artist approaching each new composition with an appetite for innovation rather than a will to expediently recycle poses and body parts for later works. For instance, the pose of the Child in other Marian interiors, such as the Dresden, Frankfurt, Rolin, and van der Paele panels, show the Child oriented in distinctly different positions. In each, the design of the Virgin’s red mantle is equally distinct. Susan Jones has commented that the recycling of van Eyck’s motifs and designs was carried on, in a relatively undiluted form, by his followers in Bruges until around 1460, after which time the influence of Rogier van der Weyden begins to be felt through the arrival of painters such as Hans Memling.21Susan Jones, ‘The use of patterns by Jan van Eyck’s assistants and followers’, p. 204. This observation appears particularly pertinent in regards to other features of the painting which signify a departure from van Eyck’s style and motifs.

The absence of van Eyck’s characteristic spatial planning

Van Eyck’s observational and technical skill and his attention to detail are evident throughout each of his completed works. When the viewer examines each detail thematically – the creation of interior architectural space, the study of light on the surfaces of forms, the choices in iconographic representation, or the technical virtuosity in applying paint – van Eyck emerges as a painter who constructed images in a deliberate, informed and incredibly diligent manner. The finished compositions contain no wasted space or meaningless motifs. Van Eyck’s interiors achieve a believable illusion of figures and objects in interior spaces through an efficient, empirically devised perspective, an acutely sensitive use of light and shadow, and the calculated positioning of objects in space to reveal their spatial interrelationships.

This is most clearly shown in the Frankfurt Virgin and Child Enthroned (Lucca Madonna) (fig. 2). This painting, which demonstrates van Eyck’s symmetrical compositional planning at its most rigorous, gives sophistication and subtlety to a relatively simple image through a well-applied combination of refined pictorial devices. Utilising a compositional format which recalls Giotto’s allegorical figure Justice, c.1305, in the Scrovegni Chapel in Padua, the enthroned Virgin and Child are centrally positioned in a tiny ecclesiastical interior, possibly a chapel.22Gordon Morrison believes that the Virgin and Child of the Frankfurt panel are placed in a ‘domestic interior’, noting that there is a category of such interiors that would have platforms, thrones, arched windows and vaulted ceilings, and would be found in the homes of the nobility. I have used the term ‘ecclesiastical interior’ to distinguish it from a domestic interior which illustrates a room with more prosaic functional elements, such as a fireplace, bed, mirror, chest or bench. The perspective is determined primarily by the tiled floor, which provides orthogonal lines, and the intersecting horizontal lines of the tiles, which help the eye estimate the degree of recession into space. Additional orthogonal lines are created by the carpet, the platform upon which the Virgin’s throne rests, and the window sill, shelves and canopy over her. Equally important to the believable illusion of the space surrounding the Virgin is the use of light. Entering through the arched window and circular window above it, full daylight meets the Virgin and Child from the left, and upper left, and covers more than two thirds of her face. We recognise from the recession of the floor tiles that the throne sits far back enough for the Virgin’s face to receive the light from the windows. We also recognise from the dark line of shadow on the right side of the wall (cast by the throne and baldachin) that the throne is positioned close to the back wall. The shadows cast by the lion statuette on the far right, and its platform, carry over to the right wall, further satisfying our eye that the figures and throne convincingly inhabit this interior space.

When this same visual analysis is applied to the NGV’s Virgin and Child (fig. 1), the illusion of the Virgin sitting within the interior space is not quite believable. This is due to the absence of any rational perspective arrangement and integrated relationships of light and shadow. Orthogonal lines in this painting are created by the lines of the floorboards, windows, shelves, table, rug and canopy, but they do not have any of the coordinated arrangement or focus that we see in works by van Eyck, which all show a disciplined plan of spatial recession. In the Melbourne painting, each object recedes at a different angle to others with no suggestion of common vanishing points (fig. 9). This tendency is alien to van Eyck’s working technique and has more in common with the less controlled spatial recession found in the work of Robert Campin, van der Weyden and their numerous followers in the second half of the fifteenth century.23Van Eyck’s perspective strategies are discussed in a yet to be published article by the author, entitled ‘Perspective and pictorial construction in the paintings of Jan van Eyck and his contemporaries in early Netherlandish painting’. Because of this lack of perspectival information, one cannot determine exactly how far back into space the baldachin sits, or its position relative to the side window. Nor is there information provided about the relationship between the figures and the interior from the light which fills the space. Is the light from the side window – which appears to be coming from behind the two figures – responsible for the almost frontally lit Virgin and Child? Why is the wall behind the cloth of estate as dark on the left, where it is closest to the window, as it is on the right? Where are the relationships of shadows and tonal transitions that appear in the Frankfurt panel? Without these we cannot understand exactly where the cloth hangs in relation to the chest and candelabrum to the right, or the bench and bottle at left, and an unresolved amount of space exists behind the baldachin. Without a rationally constructed interior or the creation of a relationship between light and shadow, the eye is able to recognise the individual parts of the image but is unable to process the sum into a fully integrated composition.

Anachronistic and anomalous visual components

Certain pentimenti from the Melbourne Virgin and Child were revealed by recent infra-red reflectography, with one detail showing a curious type of alteration (fig. 10).24The infra-red vidicon reflectogram of the Melbourne painting was captured and assembled at the National Gallery of Victoria by Michael Varcoe-Cocks, Conservator of Paintings 1850–1950, in 2000. Just beneath the canopy of the baldachin the reflectograms reveal the forms of turned-up curtains, suggesting the artist had earlier intended to have rolled-up curtains suspended from the front of the baldachin. This is a critical piece of evidence, suggesting that the painting is not a direct copy of another Jan van Eyck painting.25This pentimento was first identified by Hugh Hudson. See ‘Re-examining van Eyck’, p. 40. However, it presents a new anomaly in that baldachins of this type are not characteristic of van Eyck’s period. Rolled-up curtains for bed canopies are commonly seen in paintings from the first part of the fifteenth century, including van Eyck’s Arnolfini double portrait (fig. 4), and in many paintings by van der Weyden. Their use for baldachins in Netherlandish painting, however, is very rare, especially in images of the Virgin before 1450.26The Virgin Enthroned with Saints Jerome and Francis, 1457, by Petrus Christus, in the Städelsches Kunstinstitut, Frankfurt, shows a baldachin with unfurled curtains wrapped around the crystal pillars of an unusual throne. Hanging baldachin curtains are also seen in the Madonna and Child Enthroned, by the mid fifteenth-century Middle Rhine painter the Master of the Darmstadt Passion, in the Berlin Gemäldegalerie. One of the very few examples of this type of baldachin is found in a late fifteenth-century painting attributed to the Master of the Legend of Saint Barbara, an imitator of van der Weyden, though it is for the papal throne of Nicholas V, not the Virgin’s.27The coronation of Henry II and the consecration of his sword, by the Master of the Legend of Saint Barbara, Germanisches Nationalmuseum, Nürnberg. Another papal throne with a curtained baldachin is seen in a late fifteenth-century panel of the The legend of St Romuald, by Colijn de Coter and assistants in the Mechelen Cathedral.

The baldachin of the NGV’s Virgin and Child presents another significant problem of the image and its attribution, in that its cloth of estate is uncharacteristic of Jan van Eyck. The cloths of estate in van Eyck’s Marian paintings are made of lampas silk. Although the design for each one varies, they all conform to certain conventions: the symmetrical, geometric patterns all create a distinct frame around the head of the Virgin, and all have fold crease lines which show off the rich sheen of the fabric. The gold brocaded velvet of the cloth in the Melbourne Virgin, however, is flat and without fold creases. According to textile historian Lisa Monnas, this type of fabric is not inconsistent with paintings from the 1430s or 1440s, but in this example it is shown with incorrect repeats of the pomegranate vine pattern and an ambiguous construction, making it unclear whether the fabric is intended to show one or two loom widths.28Lisa Monnas, in correspondence with the author, 24 Nov. 2012, NGV conservation files. The asymmetrical, serpentine pattern does not frame the head of the Virgin; instead, from behind it initiates a large ‘S’ movement, which is carried through the arm supporting the manuscript and through the drapery of the mantle.29Also noted in Charles D. Cuttler, Northern Painting from Pucelle to Bruegel: Fourteenth, Fifteenth, and Sixteenth Centuries, Holt, Rinehart and Winston, New York, 1968, p. 96. As a key compositional device, this large flow of curved lines is also uncharacteristic of van Eyck; in all other examples the human forms are contained within a grid-like pattern created by the cloth pattern and the architecture they inhabit. More tellingly, the utilisation of an asymmetrical pattern for the cloth of estate is not characteristic of those found in baldachins or cloths of estate in Netherlandish painting of the first half of the fifteenth century. Cloths utilising similar broad swirling patterns are more common in later fifteenth-century paintings, notably in numerous works by Memling, such as his The Virgin and Child with St Anthony Abbot and a donor, 1472 (National Gallery of Canada, Ottawa), and in others such as Hugo van der Goes’ Adoration of the kings, c.1475–80 (Hermitage, Saint Petersburg), and in the works of lesser known painters such as the Master of 1499.30A problem with the design of the cloth of estate was also identified in Elisabeth Dhanens, Hubert and Jan Van Eyck, Alpine Fine Arts, New York, 1980, p. 367. See also Marijnissen, p. 187. Likewise, the absence of a golden or jewelled trim on the edges of the Virgin’s mantle is inconsistent with Eyckian convention and is more commonly found in paintings from the second half of the fifteenth century.31Holy family in a domestic interior, c.1470, by Petrus Christus (Nelson-Atkins Museum of Art, Kansas City) contains a Virgin with a trimless red mantle.

Another anomalous feature of the Melbourne panel is the side window (fig. 12). Missing from this detail are the internally-hinged wooden shutters that appear in virtually all other Netherlandish fifteenth-century domestic interiors. Instead, the shutters and their hinges in the Melbourne Virgin are located on the exterior of the window frame; that is, they are pushed outwards to open.32This feature was noted in Hoff & Davies, p. 48. While there are countless examples of Netherlandish interiors of the first half of the fifteenth century with internally opening shutters, there do not appear to be any examples of ones that swing outwards in the paintings of van Eyck, his contemporaries or even their followers. Instead, they tend to appear occasionally in Westphalian paintings from 1450 onwards.33Hoff & Davies noted outward shutters in the Annunciation from the Liesborn Altarpiece, c. 1475, by the Master of Liesborn (National Gallery, London). Other examples include Johann Koerbecke, Christ before Pilate, 1457, from the Marienfeld Altarpiece (Landesmuseum, Münster), and a pen drawing by the Housebook Master, Bathhouse scene, c.1475 (Housebook, p. 19a., Graf von Waldburg-Wolfegg Collection, Schloss Wolfegg, Germany). Another alien feature is the biconical glass flask resting on the bench before the window. This was a distinctly German bottle type which, according to H. E. Henkes, was frequently painted by German painters, but not by their Flemish counterparts.34This information was provided by Dr H. E. Henkes, Assistant Keeper, Department of Decorative Art of the Museum Boymans-van Beuningen, Rotterdam. Correspondence with Emma Devapriam, 11 Dec. 1991, NGV curatorial files.

A relative lack of quality in execution and paint handling in the Melbourne panel has been regularly noted over many years.35A summary of opinions on the painting over the past 110 years is found in Hoff & Davies, pp. 103–5. During the late nineteenth and early twentieth centuries numerous conflicting judgements were made; we cannot be sure how accessible the painting was to these writers, or how their views may have been affected by the work’s state of preservation. Several opinions in favour of a van Eyck attribution were given with the qualification that the work had, in their opinion, been significantly overpainted.36Panofsky, p. 183; Cuttler, p. 96; and others, noted in Hoff & Davies, pp. 103–5. This view is hard to reconcile with the moderately abraded but essentially intact painting we have today. Microscope examination of the painting and a photograph of it made during the 1957–58 treatment reveal abrasion of the red glaze of the Virgin’s mantle, paint losses and abrasion around the lower edges and the floorboards, plus minor paint losses associated with the paint craquelure.37A photograph of the painting taken after varnish and overpaint removal in 1955 has been published in Hoff & Davies, plate 1. Some repainting has been identified in the keys of the cupboard at right, and the dark lines of detail in the eyes and mouths of the figures. However, it can be affirmed that apart from the damage and repair to the area of the wood floorboards, the painting’s appearance is not significantly compromised by its condition or by restoration treatment. This is important in establishing whether the painting’s condition might explain its acknowledged deficiencies of execution and finish. These deficiencies are evident over important parts of the composition, most especially the carpet.

This carpet is uncharacteristic of those found in van Eyck’s paintings, both in design and most especially in execution. Each of van Eyck’s three centrally-positioned Virgin and Child paintings contain richly coloured carpets of different origins with complex geometrical patterns and heavy bordered edges (fig. 2). These borders, which create three or more strong lines of colour, have a critical perspectival function and feature a glowing vermilion red that gives them clarity and richness. The inner fields, containing medallions and geometrical designs, are characterised by van Eyck’s hallmark precision of handling, complexity of design and refined execution. The Melbourne painting’s carpet (fig. 13), which appears to be a Caucasian design with gammadion cross and animal emblems, is little more than a sketch; its decorative design is applied with a surprisingly crude sgraffito technique into the brown-red body of paint, a technique which appears entirely foreign to van Eyck’s practice.38Hudson, Jan van Eyck, pp. 106–7, argues that the use of sgraffito for the brush suspended on the wall in the Arnolfini double portrait is a comparable example of this technique. This view is not shared by the author. Furthermore, Lisa Monnas has noted that the baldachins and carpets in the Marian paintings by van Eyck are generally harmonised in their iconography and colouring, with the colours green, white and red (representing the three theological virtues) present through each.39Lisa Monnas, ‘Silk textiles in the paintings of Jan van Eyck’, in Foister, Jones & Cool, p. 155; and Monnas, Merchants, Princes and Painters, pp. 124–6. No such nuance is present in the Melbourne panel, in which the baldachin, figures and carpet are painted in clashing colour combinations: green and gold in the baldachin, blue and red in the Virgin’s gown and mantle, and brown and black in the carpet.

The NGV’s painting does display passages of refined brush-work, particularly in the Virgin’s red mantle and in the description of the gold threads of the brocade; however, other areas are painted in a heavy-handed manner, at least by the standards of van Eyck. A comparison of the mullioned window frames of the Melbourne panel and the Arnolfini double portrait is useful (figs 11 & 12). An appreciable gap in subtlety is at once evident between these two passages, both in terms of how the light is observed and how it is rendered in paint. A bright light passes through the Arnolfini window, casting gently receding shadows against the window frame and embrasure. Van Eyck has noted details such as the refraction of light as it passes through the glazed upper window, fine surface texture of the brick window frame and the wood of the embrasure. In the NGV’s Virgin and Child the light and shadows appear exaggerated and simplified by comparison, with relatively little attention to light diffusion and tonal gradation. This gap in handling is also evident in the brushwork in the same passage, where the brightest tones of lead white are laid with a thick, brushy modelling absent in the Arnolfini double portrait and other van Eyck interiors. Though the difference in scale of these two passages needs to be recognised – the London panel is around three times taller – a similar difference in quality is also apparent when comparisons are made with details of similar sized panels, such as the Dresden triptych or the Virgin at the fountain.

Conclusion

The information discussed here, along with research by several writers over the past fifty years, presents a complex and elusive debate about an enigmatic painting. The work demands a searching examination to draw out explanations of the persistent (though not always fully articulated) doubts that have followed it over the past century. This is not a simple task, as the painting does not yield its secrets easily and has continued to throw up new clues which confound any simple dismissal.

Paint-layer pentimenti and areas of underdrawing in the Melbourne Virgin and Child suggest that it has original elements, and therefore cannot be a direct copy. Facing this evidence, we are confronted by only two options for attribution: the painting must either be a product of Jan van Eyck and/or his studio, or is an attempt at an Eyckian image by a later imitator. Within the context of the features established here as uncharacteristic of van Eyck or his generation – such as the relatively undeveloped arrangement of light and shade, the absence of a cohesive perspective plan, the choice of fabric for the cloth of estate, and the uneven quality of execution – there are too many arguments against an attribution to van Eyck for it to be sustained. For the painting to be included into the group of accepted works by the master we would be forced to create a new class of van Eyck painting: one that does not exactly fit the criteria by which we usually define his work. Similarly, an attribution to van Eyck’s studio appears equally improbable due to the presence of details and a compositional planning process more commonly found in a later period or in other schools of Northern European painting.

We must therefore conclude that the painting was constructed by an imitator to resemble a work by van Eyck as closely as possible. If this is the case, the artist must have had access to paintings by the master, as well as to copy drawings made by assistants or followers, suggesting that a Flemish painter operating within reach of Bruges, or in a neighbouring region such as the Upper Rhine, was responsible.40Susan Jones has commented that the non-Eyckian details of the Melbourne panel suggest that the painter worked from a compositional copy drawing (from a lost original by Jan van Eyck), which did not contain detailed descriptions of critical elements – such as the pattern for the cloth of estate – thereby forcing the artist to revert to pictorial conventions from his own place and time. She is in agreement with Martin Davies that the Melbourne panel shares a common source as the work of the same theme by the Master of the Madonna of Covarrubias, and that the Covarrubias painting is probably closer to the lost original, owing to the broad compositional scheme it shares with van Eyck’s Saint Barbara: a tripartite vertical division and the placement of the Virgin in the lower half of the composition. Susan Jones, correspondence with the author, 6 May 2013. Some writers have speculated that the work may have originated in southern Europe, but this argument is impossible to sustain considering the presence of the chalk ground layer and given the knowledge this painter must have had of the motifs and manner of the master’s work.41Craig Harbison, Van Eyck: The Play of Realism, Reaktion Books, London, 1991, p. 95; Hoff & Davies, p. 40. The painting’s anomalous, non-Eyckian features appear to fit more comfortably in the context of another, later generation of Netherlandish or possibly northern German painters of the second half of the fifteenth century who were indebted not only to Jan van Eyck, but also, indirectly, as the spatial construction and anomalous details suggest, to Tournai painters such as Campin and van der Weyden.

Carl Villis, Conservator of European Paintings before 1800, NGV (in 2013).

Acknowledgements

I wish to thank John Payne, Irena Zdanowicz, Ted Gott, Gordon Morrison, Laurie Benson, Jane Devery, Elena Taylor, Lisa Monnas and Susan Jones for their generous assistance in preparing this article.

 

Notes

1        Ursula Hoff & Martin Davies, The National Gallery of Victoria, Melbourne, Centre National de Recherches Primitifs Flamands, Brussels, 1971, pp. 29–40.

2        Ursula Hoff, European Paintings before 1800 in the National Gallery of Victoria, National Gallery of Victoria, Melbourne, 1995, pp. 103–6.

3        Hugh Hudson has argued that due to flaws in the technical findings by the committee, the KIK-IRPA report cannot be used as evidence against an attribution to Jan van Eyck or his studio. Furthermore, Hudson argues that the painting can be justifiably attributed to the artist or his studio based on technical evidence, composition and iconography. See Hugh Hudson, Jan van Eyck: The Ince Hall Virgin and Child and the Scientific Examination of Early Netherlandish Painting, VDM Verlag Dr. Müller, Saarbrücken, 2009, pp. 3–4.

4        Mauro Natale (ed.), El Renacimiento Mediterráneo, Museo Thyssen-Bornemisza, Madrid, 2001, pp. 264–9; Till-Holger Borchert (ed.), The Age of Van Eyck, Thames & Hudson, London, 2002, p. 237.

5        A summary of recent technical examination is found in John Payne & Carl Villis, ‘Aspectos Técnicos de La Virgen con el Niño de Ince Hall’, in Natale, pp. 266–7.

6        An examination of recent infra-red reflectography was reported in Hugh Hudson’s Masters thesis, ‘Reexamining van Eyck: a new analysis of the Ince Hall Madonna’, University of Melbourne, 2000, and in Jan van Eyck, 2009, pp. 86–93. 

7        Art Bulletin of Victoria, National Gallery of Victoria, vol. 40, 1999,
pp. 44–7.

8        SEM/EDS analysis of the ground layer was carried out by Deborah Lau, Analytical and Conservation Scientist, CSIRO Australia, in November 2000.

9        Roger H. Marijnissen, The Masters’ and the Forgers’ Secrets: X-Ray Authentication of Paintings, Mercatorfonds, Brussels, 2009, pp. 186–7.

10      Hoff & Davies, p. 47.

11      ibid., p. 46.

12      Saint Barbara, 1437 (Koninklijk Museum voor Schone, Antwerp); Virgin and Child with Canon van der Paele, St Donatian and St. George, 1434–36 (Groeningemuseum, Bruges); the Dresden triptych with Virgin and Child Enthroned, 1437 (Gemäldegalerie Alte Meister, Staatliche Kunstsammlungen Dresden); Virgin and Child Enthroned (Lucca Madonna), c.1437–38 (Städel Museum, Frankfurt am Main); Annunciation, c.1434–36 (National Gallery of Art, Washington, DC); Virgin at the fountain, 1439 (Koninklijk Museum voor Schone Kunsten, Antwerp); Virgin and Child with Chancellor Rolin, c.1430–34 (Louvre, Paris). Erwin Panofsky, Early Netherlandish Painting, Harvard University Press, Cambridge MA, 1953, pp. 191–4.

13      The very old, possibly original painted inscription is inconsistent with van Eyck’s style of lettering and abbreviating. See Hoff & Davies, pp. 45–6. Hudson argues that the signature could be viewed as being by van Eyck despite acknowledging that the spelling ‘EYC’ and the inclusion of ‘BRVGIS’ is not found in his signed work (Jan van Eyck, pp. 150–2).

14      Panofsky, p. 193.

15      ibid., pp. 184–5.

16      ibid., pp. 183, 193.

17      These features appear to be present in the lost Woman at her bath, reproduced in Willem van Haecht’s The Constcamer of Cornelius van der Geest, 1628 (Rubens House, Antwerp).

18      Panofsky, p. 185.

19      The Van Maelbeke Virgin, formerly of the church of St Martin’s in Ypres, has been lost since the 1790s and is only known to us by later drawings, such as the one in the Germanisches Nationalmuseum, Nuremberg. See Susan Jones, ‘New evidence for the date, function and historical significance of Jan van Eyck’s “Van Maelbeke Virgin”’, Burlington Magazine, vol. 148, no. 1235, Feb. 2006, pp. 73–81.

20      Susan Jones, ‘The use of patterns by Jan van Eyck’s assistants and followers’, in Susan Foister, Sue Jones & Delphine Cool (eds), Investigating Jan van Eyck, Brepols, London & Turnhout, 2000, pp. 197–207; see also Lisa Monnas, Merchants, Princes and Painters, Yale University Press, New Haven &  London, pp. 121–3.

21      Susan Jones, ‘The use of patterns by Jan van Eyck’s assistants and followers’, p. 204.

22      Gordon Morrison believes that the Virgin and Child of the Frankfurt panel are placed in a ‘domestic interior’, noting that there is a category of such  interiors that would have platforms, thrones, arched windows and vaulted ceilings, and would be found in the homes of the nobility. I have used the term ‘ecclesiastical interior’ to distinguish it from a domestic interior which illustrates a room with more prosaic functional elements, such as a fireplace, bed, mirror, chest or bench.

23      Van Eyck’s perspective strategies are discussed in a yet to be published article by the author, entitled ‘Perspective and pictorial construction in the paintings of Jan van Eyck and his contemporaries in early Netherlandish painting’.

24      The infra-red vidicon reflectogram of the Melbourne painting was captured and assembled at the National Gallery of Victoria by Michael Varcoe-Cocks, Conservator of Paintings 1850–1950, in 2000.

25      This pentimento was first identified by Hugh Hudson. See ‘Re-examining van Eyck’, p. 40.

26      The Virgin Enthroned with Saints Jerome and Francis, 1457, by Petrus Christus, in the Städelsches Kunstinstitut, Frankfurt, shows a baldachin with unfurled curtains wrapped around the crystal pillars of an unusual throne. Hanging baldachin curtains are also seen in the Madonna and Child Enthroned, by the mid fifteenth-century Middle Rhine painter the Master of the Darmstadt Passion, in the Berlin Gemäldegalerie.

27      The coronation of Henry II and the consecration of his sword, by the Master of the Legend of Saint Barbara, Germanisches Nationalmuseum, Nürnberg. Another papal throne with a curtained baldachin is seen in a late fifteenth-century panel of the The legend of St Romuald, by Colijn de Coter and assistants in the Mechelen Cathedral.

28      Lisa Monnas, in correspondence with the author, 24 Nov. 2012, NGV conservation files.

29      Also noted in Charles D. Cuttler, Northern Painting from Pucelle to Bruegel: Fourteenth, Fifteenth, and Sixteenth Centuries, Holt, Rinehart and Winston, New York, 1968, p. 96.

30      A problem with the design of the cloth of estate was also identified in Elisabeth Dhanens, Hubert and Jan Van Eyck, Alpine Fine Arts, New York, 1980, p. 367. See also Marijnissen, p. 187.

31      Holy family in a domestic interior, c.1470, by Petrus Christus (Nelson-Atkins Museum of Art, Kansas City) contains a Virgin with a trimless red mantle.

32      This feature was noted in Hoff & Davies, p. 48.

33      Hoff & Davies noted outward shutters in the Annunciation from the Liesborn Altarpiece, c. 1475, by the Master of Liesborn (National Gallery, London). Other examples include Johann Koerbecke, Christ before Pilate, 1457, from the Marienfeld Altarpiece (Landesmuseum, Münster), and a pen drawing by the Housebook Master, Bathhouse scene, c.1475 (Housebook, p. 19a., Graf von Waldburg-Wolfegg Collection, Schloss Wolfegg, Germany).

34      This information was provided by Dr H. E. Henkes, Assistant Keeper, Department of Decorative Art of the Museum Boymans-van Beuningen, Rotterdam. Correspondence with Emma Devapriam, 11 Dec. 1991, NGV curatorial files.

35      A summary of opinions on the painting over the past 110 years is found in Hoff & Davies, pp. 103–5.

36      Panofsky, p. 183; Cuttler, p. 96; and others, noted in Hoff & Davies, pp. 103–5.

37      A photograph of the painting taken after varnish and overpaint removal in 1955 has been published in Hoff & Davies, plate 1.

38      Hudson, Jan van Eyck, pp. 106–7, argues that the use of sgraffito for the brush suspended on the wall in the Arnolfini double portrait is a comparable example of this technique. This view is not shared by the author.

39      Lisa Monnas, ‘Silk textiles in the paintings of Jan van Eyck’, in Foister, Jones & Cool, p. 155; and Monnas, Merchants, Princes and Painters, pp. 124–6.

40      Susan Jones has commented that the non-Eyckian details of the Melbourne panel suggest that the painter worked from a compositional copy drawing (from a lost original by Jan van Eyck), which did not contain detailed descriptions of critical elements – such as the pattern for the cloth of estate – thereby forcing the artist to revert to pictorial conventions from his own place and time. She is in agreement with Martin Davies that the Melbourne panel shares a common source as the work of the same theme by the Master of the Madonna of Covarrubias, and that the Covarrubias painting is probably closer to the lost original, owing to the broad compositional scheme it shares with van Eyck’s Saint Barbara: a tripartite vertical division and the placement of the Virgin in the lower half of the composition. Susan Jones, correspondence with the author, 6 May 2013.

41      Craig Harbison, Van Eyck: The Play of Realism, Reaktion Books, London, 1991, p. 95; Hoff & Davies, p. 40.