- 1 Several specialists who talk about hand-beaten gold (Bigelow et al., 1991) point out two terms in E (...)
1One of the most prestigious creations of its time, the main altarpiece of the Old Cathedral of Coimbra (Portugal) was commissioned in 1499 by Bishop D. Jorge de Almeida. The work was carried out by two migrant masters who came from Southern Netherlands, but worked in the Iberian Peninsula: Olivier de Gand (sculptor) and Jean d’Ypres (painter-gilder) (Fig. 1). This imposing Portuguese-Flemish altarpiece (thirteen meters high) still exhibits the colour scheme of the 16th century, which used abundant gold and blue, typical of flamboyant Gothic. In fact, it was totally or partially coated four times over the five hundred years of its existence. Research into the historical records allows us to date each campaign very precisely: in 1502 (gothic and original polychromy) (Garcia, 1923), in 1583 (refurbishment restricted to a few parts) (Garcia, 1923), in 1685 (new baroque polychromy) (Louro, 1983; Le Gac, 2004a) and in 1900 (complete restoration) (Vasconcelos, 1930). Gold coatings were applied systematically during these four periods and, in all cases, in the form of gold leaf. The use of very thin hand-beaten gold leaf1 was already established on wooden sculptures in Northern Europe during the 11th century (Serck-Dewaide, 1991).
Figure 1: Main altarpiece of the Old Cathedral of Coimbra (1502).
Figure 1 : Retable majeur de la Sé Velha de Coimbra (1502)
© J. Pessoa.
- 2 Natália Alves compiled many deeds executed by notaries related to the gilding of altarpieces in the (...)
- 3 This research, dealing with the qualitative and quantitative analysis by TEM-EDS of the gold alloys (...)
2Numerous publications concerning sacred polychrome artworks contain reports of the use of metallic leaves applied on wooden substrates. The collective work entitled Gilded Wood (Bigelow et al., 1991) remains the seminal publication in this field. However, studies focussing on the quality of the alloys, on the beating process, and on the thickness remain rather scarce, despite the pioneering work undertaken in Portugal by Natalia Alves2 (1989), Maria Teresa Marques3 (1998) and Isabel Ribeiro (Moura et al., 2004).
3The main altarpiece of Coimbra provides us with an excellent opportunity to examine the trade practices of gilders in Portugal between 1500 and 1900, and more specifically:
To study the different techniques applied for gilding the altarpiece.
To verify if the sets of technical particulars from each coating are pertinent to the written historical records and sources on gilding. In addition to medieval and Renaissance treatises, consulted for Western European practices, priority was given to ordinances and contracts written in Portugal from the 16th to the 19th century.
To search for information on the gold products, as well as on their specific output.
To study the caratage (ct) and thickness of the gold leaf wherever possible, taking into account the fact that only traces of the original polychromy and its refurbishment of 1583 remain today, as they were almost completely stripped off in 1685. The baroque re-polychromy, which is the stage most visible today, together with the restoration carried out in 1900 are therefore the two interventions which best allow systematic research.
- 4 By washing, as M.T. Marques (1998) did when studying by TEM-EDS gold leafs sampled from baroque alt (...)
4Twenty samples were taken from the gilded parts of the altarpiece (architectural structure, several statuettes, and statues from the first and third tiers). They were mounted in synthetic resin and prepared as cross-sections. All layers were first examined by optical microscopy and scanning electron microscopy (SEM) for high magnification images. The composition of the gold alloys was ascertained using energy-dispersive X-ray spectroscopy (EDX), which permits major element point analysis and semi-quantitative evaluations. Because the samples were very small (0.5-1 mm3) and the layers so well adhered to one another, the gold leaf could not be separated4 from the surrounding materials, either below (such as bole) or above (such as paint: lead white, verdigris, vermilion). We therefore tried to ascertain the compounds of these surrounding layers by X-ray diffraction (XRD) and by SEM-EDX, and to consider how their elemental composition (Ca, Fe, Pb, Hg, Cu, Sb, Ti, etc.) might interfere with that of the gold leaf and with the level of impurities found within the gold.
5Depending on the type of surface to be covered (architectural structure or statuary of the altarpiece), the means employed, and the final effect required, two recurrent gilding techniques can be distinguished:
Water gilding, on bole or ‘assiette à dorer’, composed of ochre and/or clays bound in a proteinaceous medium (usually an animal glue or egg white), whose presence is meant to allow the gold leaf to be fixed by moistening the bole. Once dry, the gilded part can be polished with an animal tooth or an agate stone to obtain a brilliant and shiny finish, hence its name of ‘burnished gilding’. For this method, the wooden substrate has to be properly primed with a thick white ground, which offers better compressive strength during the burnishing. In Southern Europe, such a whiting consists of calcium sulphate in its anhydrite form (gesso) and/or bihydrite form (gypsum), applied with a proteinaceous binder.
- 5 Even though the surface of a non-burnished water gilding could achieve a poor reflective effect as (...)
Oil gilding, over mordant or ‘or de couleur’, composed of a fat layer made of a drying-oil containing pigments rich in metallic oxides. The gold leaf was laid on when this layer was nearly dry. It cannot be polished, hence its generally accepted name of ‘matte gilding’5. This technique requires less care in preparation and execution.
In both cases, there will be an influence on the tone of the gold from the coloured underlayers.
- 6 ‘Appliqué relief brocade’ refers to both the imitation of sumptuous brocades and the technique itse (...)
6In 1502 we observe: 1) water gilding on the structure of the altarpiece, where the gold leaf is applied to an extremely fine bole of a very pale orange colour (Fig. 2); 2) oil gilding on tin leaf in the production of ‘appliqué relief brocades’6 which still exist in the niches in the first, third and fourth tier, and that perhaps once also covered the rich garments of the figures. Here, the gold leaf is laid on a translucent and reddish oil layer containing lead-based pigments. While a translucent red glaze, essentially organic, brings out the contrast of the brocade pattern in the first tier, we found evidence of a translucent greenish copper glaze in one cross-section from the third tier.
7In 1583, we observe only oil gilding, applied on the architectural structure without prior whiting, to refresh some parts. The topography of the gold leaf is very irregular. It has been attached to a thick dark red layer containing coarsely ground lead-based pigments, clays and iron-oxides (Fig. 2). The entire process clearly indicates that Fernão da Costa, who carried out the work in only eight days, did not have enough time to execute the gilding more carefully by applying water gilding and covering larger areas.
8In 1685, as stipulated in the contract of 1684, all the carved elements of the structure and the statuary were treated with burnished gilding (with the exception of flesh tone, hair, and beard), which was laid on an orange bole (Fig. 2). Water gilding was all the more imperative for the clothing once the techniqueof sgraffito (‘estofado’ in Portuguese) was also required by the commissioner. This other technique, closely connected with gilding (described in detail by Nunes  and Pacheco ), reproduced the brocade fabrics still in fashion. It involved painting a colour in egg tempera over burnished gilding and stripping off the paint selectively in different design patterns to expose the shiny gold beneath (Serck-Dewaide, 1991). A small raised hand moulded texture, meant to imitate to perfection the weft looped weave of Italian or Spanish brocades, was applied to complete the surfaces of the clothing (Le Gac, 2004b). Such final decorations are made in wax, hence water repellent; therefore, they are the only ones covered with oil gilding (not mentioned in the contract). Between the wax and the gold leaf, we found a very fine ‘or de couleur’, whose complex elemental composition led us to believe that palette scrapings could have been boiled with oil.
- 7 We verify thus that gilders continued to use an ‘or de couleur’, whereas, at the end of the 18th ce (...)
9In 1900, during the restoration, 1) three movable groups of the predella were completely or partially replaced with new elements and then water gilded. The gold leaf was laid on a pinkish bole and carefully polished. 2) Matte gilding was used on the damaged elements of the architectural structure (moulding, stars glued on the flat areas, or small fixed figures). This matte gilding is applied on a thick oily layer of a light yellow colour7, rich in lead, barite, cadmium and chrome compounds, which reveal the wide range of pigments available. In places, there are indications of a hurried application and lack of care: pigmented oil splashes over the blue background, and wrinkles on the gold leaf (it appears to have been laid on too soon, before the ground had dried out sufficiently). In contrast, the water gilding has been applied with a high degree of craftsmanship. Whether this indicates two separate gilders, or is a factor of viewing distance (the oil gilding is much further away), it is not clear.
Figure 2: Cross-section of the gilding (R-O1). Optical microscope, magnification 100x.
Figure 2 : Coupe transversale d’échantillon (R-O1) comportant trois dorures successives (MO 100x).
10Up until the 18th century, gold leaf came from mint gold coins rather than from raw gold, described as ‘or vierge’. Sources, either concerning Portugal or other countries in Europe, refer to the use of “ducados” (Cennini, ca. 1400), “scudi” (Vasari, 1550), “cruzados” (Livro dos Regimentos 1572), “piastres” or “anciennes monnaies d’Espagne” (Diderot and d’Alembert, 1752), or very simply to “moedas de ouro” (Brandão, 1985). While some coins were undoubtedly pure gold, others would have been alloys. This suggests that goldbeaters must have known how to refine debased coin species by cupellation and quarting (Guerra and Gondonneau, 1998).
11Early sources indicate that the thickness of the metal leaves was of great importance because: 1) it affects the spreading capacity and performance of the leaf; 2) a very thin leaf creates an illusion at a lower cost, since it can be applied to materials of inferior quality, giving the impression that they are made out of solid gold or silver; 3) while thicker leaves are well adapted to flat surfaces, thinner leaves are more suitable for carved elements.
Table 1 : Weight* of gold used for the beating process and output by square meters obtained with the beaten gold leaves, according to several documentary sources.
Tableau 1 : Poids* d’or utilisé pour la batte, et « rendement superficiel » des feuilles battues à partir de quelques sources anciennes.
voir PDF en annexe
See PDF below.
12Gold and silver, in thinly beaten leaves, will cover larger areas with a much reduced weight of metal. Table 1 summarizes the documents where we found one or more references to the type of gold used (mint gold coins or raw gold), the weight of gold required for the beating process, the number of leaves obtained and/or their dimensions. The number of leaves obtained from an approximate weight of 28 g increases significantly over time. The number of leaves rises from 500-700 in the 1st century (Plinius) to 1190 in the 14th century (Cennini), to 1366 in 1550 (Vasari), to 1600-2000 in 1690 (Furetière), until 2407-2880 in 1750 (Savary des Bruslons). The increase of the area covered is also significant, successively from 726-931 m2 in the 16th century (Vasari, 1550), to 1114 m2 during the 17th century (Furetière, 1690), and 1587-1899 m2 in the 18th century (Savary des Bruslons, 1750).
13The observation of all the gold leaves of the main altarpiece of Coimbra with SEM (magnification up to 5000x) reveals thicknesses clearly below 1 µm, varying between 0.24 and 0.45 µm. These dimensions are not absolute, since there are other parameters to consider: 1) gold, even burnished, presents an irregular surface on a micrometric scale; 2) if flaking off, the gold leaf seems to be thicker than in reality; 3) overlaps can falsify the findings as well; 4) if sample cross-sections are not precisely positioned, any slight deviation can be sufficient to modify the apparent thickness of all the layers.
14Although Cennini (ca. 1400) recommended “the use of thicker gold on panels and thinner gold on mouldings and leaves”, we have not studied this aspect. In the altarpiece, the flat surfaces are reserved for the application of the colour blue. The other parts of the architectural structure and the statuary are all carved in the round, resulting in more or less complex volumes and an unclear demarcation between flat and rounded areas.
- 8 Fully aware that “Gold in its perfect pure form is soft” and that “the more metals are hard, the mo (...)
15On the entire set of results obtained by SEM-EDX, Au is present in the highest quantity. Its purity is such that, between 1502 and 1685, gold appears to have been the sole metal acceptable. The elemental composition of the gold shows very minor amounts of Cu and sometimes, but rarely, Ag. Most of the time, the readings for Cu and Ag fall below the margin of error. Nevertheless, every time a peak for Cu or Ag appeared in the spectra, even if very weak, those elements were taken into account, in case they were present as traces. The original gilding (1502) has a high Au content (wt%), more than 99%; the one from 1583 shows a little Cu (one analysis, among the four we carried out, indicated that it contains 1.6% Cu); and Cu was found as well in 1685, from circa 0.5% to 4.5%, according to the specific decorative surfaces (water or oil gilded). These results indicate that either high purity native gold alloys were used for making the gold leaf, when Cu reaches a concentration of ca. 1% (Guerra and Calligaro, 2003), or that Cu was intentionally introduced in the fabrication of the very thin leaf to modify its hardness8, and, perhaps, colour. The last, 1900 gold leaf is of lower quality. It clearly corresponds to an Au-Ag-Cu alloy – the so-called ‘ternary alloy’ obtained when the three metals are intentionally mixed together – in which Au is nearing 83%, and which is approximately equivalent to a 20 ct gold alloy.
16In interpreting the results from the perspective of the History of Art in general and of the history of the altarpiece in particular, it is useful to consider the following aspects:
- 9 Different types of gold leaf were fabricated at that time and are specially mentioned to be used fo (...)
17At the end of the 15th century, and according to guild regulations, for Brussels in 1454 and Antwerp in 1470 (Pagès-Camagna, 2002) for example, only fine gold could be used by the painter-gilders on altarpieces. Jean d’Ypres would not only have been fully aware of such restrictions, but would have also imported this practice with him (as he did for many other techniques applied by the workshops of Brabant that we found on the altarpiece of Coimbra). In Portugal, however, the control over materials was the privilege of the commissioner. Taking into account the awe-inspiring spectacle Bishop D. Jorge de Almeida sought to produce with his imposing project (Vasconcelos, 1930), he had no reason not to impose the greatest refinement, and therefore the use of a very pure gold9. This aspect has been verified by the present study.
18In 1583, the amounts of money assigned to the purchase of materials and the manpower necessary for refurbishing the altarpiece were limited. Only 1100 real were allocated for the gold leaf. At that time, a “milheiro” (a “thousand”, i.e. ten booklets of a hundred gold leaves each) cost between 3000 and 5000 real, depending on where it was obtained (Lisbon or Oporto), but perhaps also on how the gold was produced, its thickness and fineness. As the analyses show, very pure gold has been utilized, indicating that Fernão da Costa did not compromise by using a low-grade and cheaper alloy; he rather respected the transcendental meaning of the most precious metal above all others in the religious context in which it had to be used.
- 10 In French, as well as in Portuguese, the word ‘gold’ (‘or’ / ‘ouro’) in use at this time means ‘gil (...)
- 11 Until the 18th century, the sworn assayers from The Royal Mint used the touchstone (a small tablet (...)
- 12 Among hundreds of contracts they red, Natália Alves found only three (dated 1712 and 1737) that ref (...)
19At the end of the 17th century, the gilding of the altarpiece was a matter of priority for Bishop D. João de Mello, and while the changes represented the cutting edge of the new baroque polychromy, the dichromatic (gold and blue) character of the gothic coating was preserved. The contract signed in 1684 insisted on the gold being “subido”, i.e. of a very high grade, “bem corado”, colourful, and “brunido”, burnished on all the parts to be gilded (carved elements of the structure and statuary). These adjectives, used to characterize “the gold”10 occur in a great number of Portuguese contracts of the 17th and 18th centuries (Brandão, 1984-1985; Alves, 1989; Le Gac in Serck-Dewaide et al., 2004). The contract is very explicit: Manuel da Costa Pereira, the painter-gilder who was in charge, had to use fine gold, of a saturated yellow, and obtain a brilliant finish using the water-gilding technique. The mention of the colour is significant because, up to the 18th century, the quality of gold or silver was very often evaluated based on their visual aspect11. Any reference to colour would also be intended to differentiate ‘fine gold’ (24 ct gold) from ‘common gold’ (23.25 ct gold, redder, containing up to 12 grains of silver [2.08%] and 6 grains of copper [1.04%]), and from ‘pale gold or green gold’ (16 ct gold, a lot less yellow, containing up to a third of silver). These three types of beaten gold leaf were officially recognised at the beginning of the 18th century (Savary des Bruslons, 1723; Diderot and d’Alembert, 1752). It should be pointed out again that the gold caratage is not specified in the altarpiece contract. The mention of the gold caratage is practically – not to say totally – omitted in the contracts of this period12; hence the importance of finding a deed signed by the same artist in 1686 for the gilding of another altarpiece, in which it was stipulated “to use a pure twenty four-carat gold similar to the one employed on the main altarpiece of the Cathedral [of Coimbra]” (Garcia, 1923). This demand is all the more exceptional because it refers to a perfectly pure gold. According to Furetière (1690) and Savary des Bruslons (1723), this was extremely difficult to obtain: “gold can never be refined up to this [level] – 24 carats – there are always some small fractions missing”. Diderot and d’Alembert (1768) add: “it is enough that the test tube reports 23 carats to the gold leaf be reputed fine, as the missing weight cannot be prejudicial to the highest-grade gold”.
20We therefore conclude, considering the results we obtained by SEM-EDX, that Manuel da Costa Pereira respected the terms of the contract. Taking into account the enormous amount of gold present, the gilding shows great consistency, both in quality and execution, despite the participation of many individuals organised in a workshop, as was usual at this period for large-scale operations. Such homogeneity is dependent first on the goldbeater, in Portugal, who had to produce all the “milheiros” necessary for the gilding of a single artwork (Brandão, 1985; Alves, 1989). It seems, however, that the differences observed in the gold caratage between water gilding and oil gilding were likely due to the painter-gilder himself or one of his craftsmen. The results for oil gilding clearly show that it was a deliberate choice to use a low-grade gold on statuettes, where gilding was required on the small wax ornaments (which had been first applied to the garments of the small figures). We were not able to establish if the use of ca. 21.5 ct gold leaf was a matter of economy – a low-grade gold cannot be noticed on very small objects – or if it was rather an aesthetic concern, that of differentiating burnished gold from matte gold.
Table 2: Gold alloys for indoor exposure, available at A Favrel Lisbonense in 1901.
Tableau 2 : Alliage d’or en feuilles pour dorure d’intérieur, disponibles chez A Favrel Lisbonense en 1901.
- 13 The list of colours available in England in 1929, published by Andrew Lins (1991), shows that the d (...)
21The caratage of the gilding used in the 1900 restoration does not comply with the purest commercial grade accessible at the time (a 23 ct gold), that was especially manufactured, among other gold alloys for indoor exposure, in Lisbon, by AFavrel Lisbonense (Varella, 1901) (Table 2). Both caratages identified by SEM-EDX (20 ct and c. 19.5 ct) correspond to Varella’s yellow greenish gold alloys: either “cidrão claro” (“light citron” – 20 ct gold) or “verde 1ª” (“extra fine green” – 19 ct gold)13. Once again, was the quality of the gold leaf chosen for an economic reason or an aesthetic one? The gold alloy used in the gilded surfaces would have resulted in an obvious change of hue, from an original warmer tone to a pale one, reinforced by the very light yellow ground over which the gold leaf was laid. While the restoration appears well blended nowadays, we wondered whether, at the time of application, there had been an attempt to visually ‘integrate’ the new gilding. In the samples of the most recent oil gilding, we did not find any tinted layer above it which would have provided a certain patina.
22These preliminary results, compared with the precisely dated documents to which we had access, and with published studies on the circulation of gold in the past (Godinho, 1971; Guerra and Calligaro, 2003), led us to formulate some theories about the provenance of this metal: for the first metal leaf (1502), it is possible that the gold came from Northern Africa, for the second (1583) from Western Africa, for the third (1685) via the Spanish from the mines of Colombia, and for the fourth (1900) from Brazilian ores. It proved difficult to confirm the provenance of the gold in this study for various reasons: the very thin layers of gold (0.24-0.45µm) available for study, and their location, surrounded by materials whose elemental composition could compromise the characterisation of trace elements in the gold. Indicators, such as the presence of Pt in the gold leaf from 1685, or Pd in the leaf from 1900 (Guerra and Calligaro, 2003), could provide a certain provenance. However, attempts to identify them using particle induced X-ray emission (PIXE) or synchrotron radiation X-ray fluorescence spectroscopy (SR-XRF) did not yet provide satisfactory results. This study is currently being pursued.
23On the main altarpiece of the cathedral of Coimbra, different gilders applied the procedures of their time while perpetuating an age-old craft tradition that changed little since the one described by Cennini (ca. 1400). This study has revealed new information, but in order to understand how widespread these practices were, we must rely on further research. We look forward to further information on three Oporto gilded altarpieces provided in the study of Bidarra et al. (2009) in order to broaden our understanding of the gilding techniques used in Portugal, particularly during the baroque period.
The authors are extremely grateful to Isabel Nogueira, of the Instituto Superior Técnico / Instituto de Ciência e Engenharia de Materiais e Superfícies (Lisbon), for providing the analytical data by SEM-EDX, to Maria José Vinagre, of the Departamento de Estudo dos Materiais / Instituto dos Museus e da Conservação (Lisbon), for providing the analytical data by XRD, and to Maria Filomena Guerra, of the Centre de Recherche et de Restauration des Musées de France, for accepting the challenge to try to measure the concentration of characteristic trace elements in the gold leaf (by PIXE, with the AGLAE accelerator). They especially thank Marie-Louise Hugon and Josephine de Linde for translating this paper from French into English, and Leslie Carlyle for technical content editing.