Van Eyck's Washington 'Annunciation': technical evidence for iconographic development - Jan van Eyck's painting

The recent conservation treatment of Jan van Eyck's Annunciation in the National Gallery of Art, Washington, D.C. [ILLUSTRATION FOR FIGURE 1 OMITTED], has uncovered an exceptionally beautiful work. The removal of heavily discolored varnish and overpaint revealed a mastery of space and a play of light that could only be guessed at before.(1) A technical study of the painting undertaken in conjunction with the conservation treatment has afforded new insights into van Eyck's methods for achieving his breathtaking visual effects. A complete report of the technical investigation, which included microscopic examination of the surface, infrared reflectography, and paint analysis, will be published shortly.(2) This technical study also revealed a number of changes made by the artist during the creation of the painting. The present article focuses on those compositional changes that have a bearing on the painting's complex meaning, seeking to lay out the sequence in which van Eyck developed the rich iconography. The implications of these changes for our understanding of van Eyck's iconographic motivations are discussed in Carol Purtle, "Van Eyck's Washington Annunciation: Narrative Time and Metaphoric Tradition," which follows this article.

The technical study has left no doubt about the authorship of The Annunciation. Individual procedures are remarkably similar to the methods used in other works by van Eyck. The Virgin's blue drapery, for example, was built up in a method strikingly similar to that in the altarpiece of the Holy Lamb in Ghent: a first layer in grayish blues establishing light and shade, a second solid blue layer to soften the contrast of the first, and a final, rich blue glaze of ultramarine in a water-based protein medium such as glue.(3) The working methods and the composition of the paint are entirely consistent with what we know of van Eyck's practices in the relatively few works that have been analyzed scientifically.(4) With the exception of the blue glaze, the painting medium seems to be linseed oil; our analysis of The Annunciation to date has not identified water-based protein admixtures to the oil.(5) The pigments are consistent with those of other paintings by van Eyck that have been analyzed.(6)

Close study of the surface disclosed the painter's refined technique. For example, van Eyck prepared the intense red of the flowers of Gabriel's cope (red velvet pile worked against a cloth-of-gold background, [ILLUSTRATION FOR FIGURE 2 OMITTED]) with a preliminary layer of vermilion. The rich texture of the velvet was created with a red lake glaze deepened with a little black.(7) Layers of glaze floated over the vermilion suggest the varied depth of the pile, with superimposed glazes creating the highest pile at the center of the petals.

Artists of an earlier generation, or contemporaries in Italy, would have rendered gold brocade with lines of gilding.(8) In a passage such as the green and gold brocade of Gabriel's dalmatic, however, van Eyck used just two colors of paint, worked wet-into-wet, to create a web of lights, darks, and midtones. He suggested the illusion of a lattice of gold threads by dragging fine strokes of black through still-wet yellow paint; with a few loose hairs of the brush he suggested specks of light glittering on the surface nearby [ILLUSTRATION FOR FIGURE 3 OMITTED].

In his rendering of light van Eyck drew a distinction between the earthly realm and the heavenly. The painting's natural light sources cast shadows, but the seven rays of divine light on which the dove descends to the Virgin do not. Behind the large stool the shadow cast by a primary light source to the viewer's right is softened by light from the upper left clerestory windows. In this diffuse shadow van Eyck worked a subtle brown glaze by tapping with his fingertips.

Van Eyck drew the same distinction between the earthly and the heavenly in his techniques for rendering gold. In the same way that he painted Gabriel's brocade illusionistically, he created the pearl pin in the Virgin's book, Gabriel's jeweled crown and crystal scepter, and even the phrases uttered by the two figures using only paint to suggest the appearance of gold. By contrast he rendered the heavenly rays in gold leaf, using oil mordant gilding in this passage alone. Study of the surface with a stereomicroscope and of a paint cross section taken from one ray shows the traditional technique of oil mordant gilding, which has been observed in other works by van Eyck.(9) Van Eyck's use of actual gold in the rays created a contrast not only technically but also iconographically. While Gabriel was on Earth his angelic vestments, and even the words he uttered, were of this world. Only the rays, rendered in gold, represent the direct intervention of the Deity.

With the technique of infrared reflectography van Eyck's characteristic underdrawing is visible, the drapery defined by clusters of long, parallel strokes, occasionally hatched.(10) The greater part of the underdrawing is fairly fine in character, but some passages seem to have been revised with a broader touch, using a liquid medium. There are many variations in the formal composition from drawing to paint, including shifts in the positions of hands and facial features reminiscent of shifts observed in the recent study of the Arnolfini double portrait in London.(11)

The results of this technical study suggest that before he began work, van Eyck had not fully conceived the complex iconography of The Annunciation as it appears today. Instead, as he worked he developed and amplified the iconographic program in several stages. Since no independent preparatory drawing for this painting survives, the earliest plan for the composition to which we have access is the drawing that van Eyck laid out directly on the white surface of the prepared panel. This underdrawing, now hidden by the subsequent paint layers, can be seen using infrared reflectography [ILLUSTRATION FOR FIGURE 4 OMITTED]. Because the revisions within the underdrawing are formal, rather than iconographic, the underdrawing as a whole is here considered the initial stage of the composition. The artist then painted over the underdrawing, making significant changes. Finally, in certain parts of the painting, van Eyck revised the composition yet again. In these passages he replaced what we will call the intermediate painted stage of the image with further important iconographic elements. This article discusses all the compositional changes for each area of the painting in turn. A summary at the end outlines the sequence in which van Eyck made the changes.

In its final form, the painting depicts the Annunciation in a church interior. Possibly through the architecture itself, and certainly through decorative elements of the interior, van Eyck reiterated the theme of the transition from life under the Old Law to life under the New Dispensation. He contrasted the natural light of the earthly realm, entering from the right, to the divine light streaming in as seven golden rays from the upper left.(12) The moment of the Annunciation is played out with costumes, words, and gestures that would clearly have recalled to the viewer the same scene enacted in the Missa Aurea, or Golden Mass.(13)

The narrow format and the composition, which is open to the right, suggest that this painting was the left wing of a larger work. Any interpretation of the iconographic program must be limited by the recognition that it was originally part of a larger whole. However, the sequence of alterations and additions to the iconography, which can now be seen through the technical study, offers new insights into the planning of that program.(14)

Evidence for Iconographic Modifications

In the underdrawing one can see that van Eyck first planned a somewhat simpler and more architecturally consistent setting for The Annunciation. The basic architectural plan, with rounded clerestory arches above slightly pointed arches below, was intended from the first, as was the placement on the rear wall of a single window above a group of three. However, in the underdrawing the pilasters, now visible only on the left wall, where they divide the arches of the lower story and triforium, ran the entire height of the church from capital almost to ceiling, springing into engaged arches enclosing the clerestory windows. In the rear wall clerestory they formed two blind arches flanking the single central window. The underdrawn pilasters can be seen in an infrared reflectogram between the lower arches of the rear wall, where there are now wall painting roundels of Isaac and Jacob [ILLUSTRATION FOR FIGURES 5, 6 OMITTED]. Van Eyck retained the pilasters of the underdrawing only in painting the side wall, and here only in the lower stories; on the rear wall he eliminated them altogether.

In the underdrawing van Eyck also laid out a consistent system of shadows that indicates a light source to the right of the composition. Bundles of heavy lines lie to the left of the cushioned stool in the foreground and the triforium columns of the rear wall. In the underdrawn arches around the clerestory windows on the side wall a wash indicates dominant shadows cast by the light from the right, rather than from the diffuse light of the clerestory windows.(15) Though the golden rays stream down to the Virgin through these windows, the windows themselves face away from the sun and do not cast an additional set of shadows outside each window's recess.

It is likely that the flat wooden plank ceiling was part of the original conception of The Annunciation. Though the reflectograms show no clear evidence of underdrawing in this area, there may be lines that are obscured by the black-containing paint of the ceiling superimposed over them. There is no evidence of a different ceiling, either in the underdrawing or in a lower paint layer.(16) The ceiling was painted in a simple sequence that finished with the painted decorations beside the beams. Subsequently, a dark brownish black paint covering certain planks (decoration and all) was used to indicate broken planks. Without clear evidence of broken planks in the underdrawing it cannot be determined whether the iconographic detail of the ruined ceiling was an afterthought. However, the simple expediency of the technique employed for painting the gaps suggests that the artist found it efficient to paint the ceiling and then black out a few planks in a compositionally satisfying pattern.

Differences in the handling of paint can be discerned between the left and rear walls of the church. The left wall was rendered in a single thin layer of loosely brushed tan paint without revisions; this is part of the first painted version, the intermediate stage of this part of the composition. Most of the rear wall, with its frescoed decoration, is much more solidly painted, with more than one layer of paint. There is evidence that this thicker paint structure resulted from van Eyck's repainting the rear wall in the course of producing the final stage of the composition.

In the area of the Jacob and Isaac roundels between the arches of the rear wall, examination with a stereomicroscope shows two superimposed painted images. The first painted image replaced the pilasters of the underdrawing with a continuous light tan paint like that of the side wall, relieved only by a fine red painted decoration around each arch. In the final version the roundels, with their Old Testament prefiguration of the Annunciation, were painted over this simpler decoration. In a photomicrograph taken below and to the left of the Jacob roundel, this series of layers is visible [ILLUSTRATION FOR FIGURE 7 OMITTED]. The light buff tone seen intermittently is the first painted wall; traces of a red line around the arch, which was the first decoration, can be seen in the lower left of the detail. Van Eyck used a darker grayish paint, which now forms the rear wall, to cover the earlier version with its red decoration; he then painted the roundels, including the red border, a portion of which is seen at the upper right of the detail.

The wall paintings on the upper wall, which depict the presentation of the infant Moses to Pharaoh's daughter and Moses receiving the commandments, probably also referred to Old Testament prefigurations of Christ.(17) Here, too, van Eyck first painted a light buff rear wall; in the final revision he covered the upper part of the wall with the same darker grayish paint seen lower on the wall, then he painted the Moses frescoes. A photomicrograph of that area [ILLUSTRATION FOR FIGURE 8 OMITTED] shows the foot of Pharaoh's daughter. Here the original lighter wall color can be seen only through an incised line that van Eyck scored lightly through the dark gray paint to guide his placement of the figures; the pointed toe overlaps the line.

There may also have been a formal motivation behind the repainting of the rear wall. The light buff of the wall color in the first painted version was almost identical to the tone of the left wall. Its replacement with a darker gray no doubt clarified the space, further distinguishing the rear wall from the left and emphasizing the spatial recession.

In moving from underdrawing to paint, van Eyck also modified the capitals of the columns and the structural pier. When painting the foliate piers he made minor formal alterations, slightly expanding similar patterns in the underdrawing. In the capital of the structural pier and the capital partially visible on the right, the underdrawing recorded a simple pattern of diminishing cornices. However, when he painted those elements the artist replaced the simple cornices with iconographic details not present in the underdrawing: a battle scene on the pier and on the smaller capital a struggle between two figures.

Iconographic changes to the foreground of The Annunciation center on the design of the inlaid stone (niello) floor and on the vase of lilies. The stool also was modified, the painted version considerably enlarging the underdrawn form. It is not clear, however, whether this alteration should be read as emphasizing an element of the symbolic program or as a formal compositional modification.

The niello floor, as we see it today, presents a complex program combining roundels with the signs of the zodiac and narrative panels including episodes from the lives of Samson and David, who were frequently cited as heroes prefiguring events in the life of Christ [ILLUSTRATION FOR FIGURE 9 OMITTED]. In the infrared reflectogram, however, quite a different pattern is visible, with diamond shapes, lobed forms, and a grid that suggests tiles [ILLUSTRATION FOR FIGURE 10 OMITTED]. This initial floor design would have been a carpet or decorative tile pattern such as those in the Dresden triptych of the Enthroned Madonna. Possibly the tan paint that serves as the base tone of the present floor originally was intended as the first paint layer of the carpet or tiles. In any case, none of that design was carried out in paint, and the brown-black niello pattern was painted in its place.

It is perhaps surprising to realize that the Marian symbol of the majolica vase and lilies was not planned in the underdrawing [ILLUSTRATION FOR FIGURES 11, 12 OMITTED]. As Carol Purtle has pointed out, the vase of lilies is one of several details, including the Virgin's gesture and the book lying on a fold of her mantle, that are typically found in illuminations of the Annunciations from the Boucicaut workshop.(18) But where those other small-scale features of the composition were planned in the underdrawing, there is evidence that the lilies were a late addition. The infrared reflectogram shows only the image of the painted flowers. With the stereomicroscope, one can see that unlike the Virgin's hands, whose space was reserved during the execution of her mantle, the lilies were painted over the mantle. Blue paint can be seen below the warm white paint of the lilies, both in a paint cross section and through paint losses in the flowers.(19) In the same way, the blue and white vase was placed over the completed niello floor and covers the brown-black paint of the design in that area. Artists seem to have recognized that as oil paints increase in transparency over time, dark paints can begin to show through white paints laid over them. For this reason, it was common to reserve spaces in dark-colored areas for light-colored details, rather than overlapping them. In this case, however, instead of leaving an open area where the vase was to appear, van Eyck first brought the design of the niello floor right up to the edge of the stool. This brown-black design is now visible through the vase, giving it a ghostly impression.(20) It is conceivable that the vase of lilies appeared in a preparatory study and that van Eyck kept the feature in mind without transferring it to the underdrawing. But both the absence of underdrawing and the addition of flowers and vase over detailed paint suggest that in painting the lilies van Eyck modified a passage he had once considered complete.

Technical study cannot conclusively establish the sequence of van Eyck's modifications to the rear wall and to the foreground because there is no paint overlap between these two areas. Within each of these two areas, however, there is evidence for moments at which the artist considered that part of the composition completed. While van Eyck's superimposition of the vase over the detailed floor suggests that he once considered the foreground complete without the lilies, the technique of the seven golden rays reflects the artist's ongoing revisions to the rear wall. These rays were rendered using gold leaf. Whereas water gilding, which is familiar from the backgrounds of Italian tempera paintings, is laid down before any paint is applied, mordant gilding was the artist's final step, applied over completed paint. Lines incised through finished paint layers laid out the path of each ray. Roughly following these lines the artist laid down the rounded, raised line of oil-based yellow mordant, and when it had dried to tackiness applied the gold leaf. Because these rays, entirely typical of van Eyck's technique, lie over all the paint layers of the rear wall, it is clear that throughout his experimentation with this passage, van Eyck did not consider the composition complete until he had painted the composition that appears today.

A Sequence for Van Eyck's Development of The Annunciation

From the evidence outlined above, we can summarize the stages in which van Eyck first conceived, then expanded the symbolic program of The Annunciation. The initial stage, the underdrawing, included a uniform architecture with pilasters on the rear wall continuing the pattern of the left wall, but with the general scheme of rounded arches above pointed ones already established. The floor was decorated with a carpet and possibly ornamental tiles. This initial stage included the figures of Mary and Gabriel in essentially their final forms, as well as the cushioned stool in the foreground. The pattern of shadows indicated in the underdrawing of the architecture background, the draperies, and the stool consistently recorded a light source on the right. Thus, the major themes of the architectural character, the Annunciation as enacted in the Missa Aurea, and the fall of natural light from the right were all included in van Eyck's initial conception of the painting.

As he painted, van Eyck revised the composition more than once. The important changes from the underdrawing to paint suggest that he may have shown the complete underdrawing to the patron for approval before proceeding.(21) Later he repainted certain passages, revising the composition yet again. On the rear wall the intermediate painted stage included a simple decorative scheme. Here a fine red decoration painted around the arches without pilasters relieved flat surfaces rendered in a light tan paint, very close in tone to the side wall.

For the final version van Eyck repainted the greater part of the rear wall, significantly enriching the iconographic program with Old Testament references in the wall paintings on the rear wall depicting the blessing of Jacob by Isaac and scenes from the life of Moses. In the foreground van Eyck also introduced Old Testament references, painting over the patterned floor of the underdrawing with the inlaid niello floor of the final composition. He also considerably enlarged the stool planned in the underdrawing. In painting the foreground, van Eyck apparently did not initially anticipate including the majolica vase of lilies but, instead, later superimposed this detail over the Virgin's mantle and the decorated floor.

It seems clear that our considerations of the rich iconographic program of Jan van Eyck's Annunciation must address its evolution as well as the final image. This complex process, during which the image was enriched with successive layers of symbolic meaning, is not unique to The Annunciation. In the Ghent Altarpiece an element as significant as the fountain at the center of The Adoration of the Lamb seems to be an addition made well into the painting process. The fountain does not appear in the underdrawing, and it lies over green underpaint that makes no provision for this feature.(22) In the Virgin with Chancellor Rolin, in Paris, a purse hanging at the waist of the patron was eliminated and the Child's gesture of benediction was added in the transition from underdrawing to paint.(23) The most recent technical study of the Arnolfini portrait at the National Gallery in London also has revealed alterations that go beyond formal modifications. Elements such as the dog and the tiny figure of Saint Margaret on a chair finial, which have been seen as important to the symbolic program, were not underdrawn but were painted in later.(24) While such modifications, made during the painting process after the underdrawing was completed, could reflect an ongoing development of the iconographic program by the artist himself, it seems likely that they reflect consultations during the creation of the painting with a patron or advisor.

Technical study of The Annunciation not only has enriched our understanding of the development of the iconography of this complex painting but also suggests that similar future studies may reveal such amplifications in a wider range of van Eyck's work.

Frequently Cited Sources

Coremans, Paul B. et al., Les primitifs flamands, 3d ser., Contributions a l'etude des primitifs flamands, vol. 2, L'Agneau Mystique au laboratoire: Examen et traitement (Antwerp: De Sikkel, 1953).

Hand, John, and Martha Wolff, Early Netherlandish Painting: The Collections of the National Gallery of Art Systematic Catalogue (Washington, D.C.: Cambridge University Press, 1986).

Notes

I have enjoyed my many discussions of The Annunciation with David Bull, chairman of painting conservation, and with John Hand, curator of Northern Renaissance paintings, both of the National Gallery of Art. I am grateful to Barbara Berrie, Julien Chapuis, Molly Faries, and an anonymous Art Bulletin reader for thoughtful comments on drafts of this article. I also wish to thank my colleagues at the Electron Microscope Central Facility of the University of Maryland and in the departments of Visual Services, Painting Conservation, and Scientific Research at the National Gallery for their assistance. Most particularly I would like to thank Carol Purtle for the many insights she brought to our collaborative exploration of this painting.

1. National Gallery of Art, Andrew W. Mellon Collection, acc. no. 1937.1.39, oil on canvas transferred from panel, 90.2 x 27.1 cm. After the treatment was completed in 1994, it is clear that a remarkable subtlety of painting survived a 19th-century transfer from the original panel support to canvas. For a full discussion of the recent conservation treatment by David Bull, chairman of painting conservation at the National Gallery of Art, and evidence for previous restorations of The Annunciation, see David Bull, "The Cleaning and Restoration of Jan van Eyck's Annunciation," Studies in the History of Art: Conservation Research (forthcoming).

2. See E. Melanie Gifford, "Jan van Eyck's Washington Annunciation: A Technical Study," Studies in the History of Art: Conservation Research (forthcoming).

The painting was first studied using infrared reflectography, and extensive examination of the magnified paint surface was carried out using a stereomicroscope. (X-radiography was of only limited use because of the interference of a heavy, white lead adhesive added during the transfer from panel to canvas in the 19th century; see Bull (as in n. 1). Catherine Metzger, Elizabeth Walmsley, and Colin Fletcher carried out infrared reflectography with a Kodak platinum silicide camera configured to 1.5-2.0 microns and assembled infrared reflectograms with Adobe Photoshop averaging individual images from eight frames.

A limited number of paint samples was taken for analysis by light microscopy and scanning electron microscopy. Light microscopy was carried out with Leitz Orthoplan research microscopes equipped for visible light and autofluorescent examination. The painting's layer structure was studied through microscopic paint cross sections examined by reflected light. Biological stains were used to survey paint cross sections for painting medium. Pigments were identified by polarized light microscopy of dispersed paint samples. Pigment identifications were confirmed by energy-dispersive X-ray analysis of paint cross sections using a JEOL electron microprobe at the facilities of the Electron Microscope Central Facility of the University of Maryland at College Park. I am most grateful to Myron Eugene Taylor, director of the facility, and to Helaleh Maghsoudlou, graduate research assistant, for their aid.

Paint media were analyzed by Suzanne Quillen Lomax and Susana Halpine. For the identification of oils and resins, hydrolyzed and methylated samples were analyzed by gas chromatography (using a Perkin Elmer Autosystem gas chromatograph) and gas chromatography-mass spectrometry (using a Varian 3700 gas chromatograph and Finnigan Ion Trap Mass Spectrometer). For the identification of protein media such as egg or glue, samples were prepared by cold-water extraction followed by hydrolysis and PITC derivitization, then analyzed for amino acids by high-pressure liquid chromatography (using a Hewlett-Packard 1090 HPLC system).

3. Samples of the midtone blue of the Ghent Altarpiece have a similar structure. See Coremans et al., 99, pls. XI,3, XXIV, 1.

4. Coremans et al.; Paul Coremans, "La technique des 'Primitifs flamand's: Etude scientifique des materiaux, de la structure et de la technique picturale. III. Van Eyck: L'Adoration de l'Agneau Mystique Gand: Cathedrale Saint-Bavon," Studies in Conservation 1 (1954): 145-61; Leopold Kockaert, "Note sur les emulsions des primitifs flamands," Bulletin de l'Institut Royal du Patrimoine Artistique 14 (1973-74): 133-39; Leopold Kockaert and Monique Verrier, "Application des colorations a l'identification des liants de Van Eyck," Bulletin de l'Institut Royal du Patrimoine Artistique 17 (1978-79): 122-27; J.R.J. van Asperen de Boer, "A Scientific Re-examination of the Ghent Altarpiece," Oud Holland 93 (1979): 141-214; Pim Brinkman et al., "Het Lam Godsretabel van Van Eyck: Een heronderzoek naar de materialen en schildermethoden," pt. 1, "De plamuur, de isolatielaag, de tekening en de grondtonen," and Pim Brinkman et al., pt. 2, "De hoofdkleuren blauw, groen, geel en rood," Bulletin de l'Institut Royal du Patrimoine Artistique 20 (1984): 137-66, and 22 (1988-89): 26-49; Leopold Kockaert, P. Gausset, and M. Dubi-Rucquoy, "Detection of Ovalbumin in Paint Media by Immunofluorescence," Studies in Conservation 34 (1989): 183-88; J.R.J. van Asperen de Boer, Bernard Ridderbos, and Manya Zeldenrust, "Portrait of a Man with a Ring by Jan van Eyck," Bulletin van het Rijksmuseum 39 (1991): 8-35; Emil Bosshard, "The Examination of the Thyssen-Bornemisza Annunciation," Apollo 136 (1992): 4-11.

5. Samples available for medium analysis were necessarily very small. Using gas chromatography-mass spectrometry (see n. 2 above), Suzanne Quillen Lomax identified fatty acid ratios typical of linseed oil in a sample from the architecture of the rear wall, as well as some evidence of a diterpenoid resin. Samples from the green glaze of Gabriel's dalmatic and from the heavily damaged blue of the Virgin's mantle also showed evidence of a diterpenoid resin but did not have identifiable fatty acid ratios; these samples were more difficult to interpret, probably because of the extremely small size of the samples and old damage to the blue paint (see unpublished analysis report dated Sep. 21, 1993, on file at the Scientific Research Department, National Gallery of Art, Washington, D.C.). Using high-pressure liquid chromatography (see n. 2 above) Susana Halpine found no evidence of any protein binder, such as glue or egg, in any of the samples (see unpublished analysis report dated July 28, 1993, on file at the Scientific Research Department, National Gallery of Art).

6. The palette includes white lead, lead-tin yellow, ultramarine, red lake, vermilion, a copper-green glaze, earth colors, and black.

7. The layer structure of the red, including the use of black pigment in the deepest red glazes, is comparable to that observed in other paintings by van Eyck. See, for example, Brinkman et al., 1988-89 (as in n. 4), 36.

8. Brocade could be represented either by applying paint over a water-gilded surface or by applying mordant gilding or shell gold over a painted surface. See the discussion of gilding techniques in David Bomford et al., Art in the Making: Italian Painting before 1400, exh. cat., National Gallery, London, 1989, 21-26, 43-48. An example of a gilded brocade in an Italian painting contemporary with van Eyck is described in Martin Wyld and Joyce Plesters, "Some Panels from Sassetta's Sansepolcro Altarpiece," National Gallery Technical Bulletin [1] (1977): 3-17. An example in an altarpiece by a northern artist a generation before van Eyck is described in Hans Nieuwdorp, Regine Guislain-Witterman, and Leopold Kockaert, "Het pre-Eyckiaanse vierluik Antwerp-Baltimore," Bulletin de l'Institut Royal du Patrimoine Artistique 20 (1984-85): 70-98; and E. Melanie Gifford, "A Pre-Eyckian Altarpiece in the Context of European Painting Materials and Techniques around 1400," in Flanders in a European Perspective: Manuscript Illumination around 1400 in Flanders and Abroad; Proceedings of the International Colloquium; Leuven, 7-10 September 1993, ed. Maurits Smeyers and Bert Cardon (Louvain: Uitgeverij Peeters, 1995), 357-70.

9. The technique of oil mordant gilding is described in the introduction to Bomford et al. (as in n. 8), 43-48. A very similar structure to that of the rays in The Annunciation was observed in the Ghent Altarpiece; see Coremans et al., 110-11.

10. Molly Faries examined the underdrawing of The Annunciation with infrared reflectography using a vidicon camera in 1981-82 during her tenure as Ailsa Mellon Bruce Senior Fellow at the Center for Advanced Study in the Visual Arts at the National Gallery of Art. Some of her reflectograms and her observations were published in John Hand's catalogue entry on the painting in Hand and Wolff, 75-86. After the recent cleaning, the painting was reexamined using a platinum silicide camera (see n. 2 above), sensitive at longer wavelengths than a vidicon camera. This technical advantage, and the removal of old repaints, allowed additional features of the underdrawing to be seen.

11. Rachel Billinge and Lorne Campbell, "The Infra-red Reflectograms of Jan van Eyck's Portrait of Giovanni(?) Arnolfini and His Wife Giovanna Cenami (?)," National Gallery Technical Bulletin 16 (1995): 47-60.

12. John Hand discusses the symbolic program in his catalogue entry in Hand and Wolff, 75-86.

13. Carol J. Purtle, The Marian Paintings of Jan van Eyck (Princeton: Princeton University Press, 1982), 46-50.

14. See Purtle's article following in this issue for a new discussion of the iconographic program of The Annunciation in light of the evidence of the changes revealed by the technical study.

15. The washed shadows seem to be part of the underdrawing rather than an early paint layer. In the upper part of the painting the wash is disrupted by what could be interpreted as water spots. Through a tiny paint loss corresponding to a wash passage visible in the reflectograms, a loosely brushed black layer can be seen lying directly on the ground, below the much thicker paint layers.

16. A diagonal form, seen in the reflectograms to the right, has yet to be explained, but it does not seem to be a clear indication of a different architectural form, as are the pilasters and arches on the walls below.

17. Hand and Wolff, 79-80.

18. Purtle (as in n. 13), 40-44.

19. There is no evidence that the vase of lilies was an addition by a later artist; the paint cross section reveals no features between the paint layers of the mantle and the lily to suggest that much time passed before the addition of the lilies.

20. This effect, which has been commented on in the literature (John Ward, "Hidden Symbolism in Jan van Eyck's Annunciations," Art Bulletin 57 [1975]: 197 n. 10), is almost certainly unintended and should not be interpreted as part of the iconographic program.

21. In their study of the underdrawing of the Virgin with Chancellor Rolin, J.R.J. van Asperen de Boer and Molly Faries came to a similar conclusion: "La Vierge au Chancelier Rolin de Van Eyck: Examen au moyen de la reflectographie a l'infrarouge," Revue du Louvre 40 (1990): 37-49.

22. Van Asperen de Boer, 1979 (as in n. 4), 194-95, 201, figs. 59, 60. This observation was not made by Coremans in his discussion of the central panel (Coremans et al., 107-17), but the infrared photography available at that time could not penetrate green paint, and no paint samples were taken from the fountain.

23. Van Asperen de Boer and Faries (as in n. 21), 37-49.

24. Billinge and Campbell (as in n. 11), 47, 50, 59, figs. 1, 3.

E. Melanie Gifford is research conservator for painting technology at the National Gallery of Art. Her research focuses on artistic style and painting technique in northern Europe. She is currently writing a book on the development of naturalistic landscape painting in the Netherlands in the seventeenth century [Scientific Research Department, National Gallery of Art, Washington, D.C. 20565].

COPYRIGHT 1999 College Art Association
COPYRIGHT 2004 Gale Group