• 1  KOVAČIĆ, S., ''Župe splitske nadbiskupije u poljičko-radobiljskom dekanatu g. 1625.'', in Synthesi (...)
• 2  VLAŠIĆ, D., Vizitacije poljičkih župa u XVIII. Saint, Crkva u svijetu, Poljica – Godišnjak Poljičk (...)
• 3  VLAŠIĆ, D., op. cit., 1995, p. 112.  

1The first known reference of the parish church of Saint Michael in Kostanje, a village in the hinterland of Dalmatia, in the part historically known as the Poljica Republic, dates from 1625.1 In the early 18^th century a new church was built. It was consecrated in 1711.2 Archbishop Pacifik Bizza was the first person to describe the church altars with paintings. This was during his visitation in 1748.3

• 4  BOŽANIĆ-BEZIĆ, N., ''Majstori od IX do XIX stoljeća u Dalmaciji II'', in Prilozi povijesti umjetno (...)
• 5  Academician PhD Radoslav Tomić is credited for both dating and attribution of the painting.
• 6  TOMIĆ, Radoslav, paper forthcoming. I am grateful to the author for the data and advice regarding (...)

3This painting was commissioned around 1711, on the occasion of the consecration of the church, and it was intended for the altar of Saint John the Baptist. It is attributed to Francesco Fedrigazzi, a painter from Venice4 who was active in Istria and Dalmatia at the end of the 17^th and in the early 18^th century.5 Between 1691 and 1708 Fedrigazzi lived in Poreč. There he painted his most important work, a copy of the painting The Last Supper by Jacopo Palma the Younger from the Euphrasian Basilica. That painting was intended for the Sinčić Palace. Many of Fedrigazzi's paintings have been identified in the region of Dalmatia, principally in the area of Trogir, where the artist probably lived after 1708.6

• 7  The description of the church altars in the episcopal visitation from 1748 (see note 3) is essenti (...)

4His painting from Kostanje had a turbulent past, having been removed from the church at some point. It was found in the attic of the parish house in Kostanje in 2011, together with two paintings by Filippo Naldi from Florence, dated with 1761.7 Fedrigazzi's painting was found without a stretcher, repeatedly folded and covered in dirt. The left, right and bottom portions of the canvas had been cut off, while the upper part was torn. The paint layer was flaking. Many flakes had detached from the canvas, especially where the canvas was folded.

• 8  Data generously provided by MSc Frane Mihanović, credited for organizing and conducting the radiog (...)

10MSc Frane Mihanović conducted digital radiographic imaging using a mobile X-ray unit (Nanomobil model, Siemens) with exposure conditions of 48 KV and 80 mAs and phosphorus plate as the image receptor. Latent images from phosphorus plates were digitalised using FCR1 FUJI device (FUJIFILM Holdings Corporation). DICOM files were evaluated and processed using Osirix software (Pixmeo); X-ray digital images were merged using Photoshop software (Adobe Systems).8

• 9  Tea Zubin Ferri of the Metris research centre and Marijana Fabečić of the Croatian Conservation In (...)

11Paint cross-section samples were photographed using Olympus BX 51 optical microscope with an integrated XC50 Olympus camera at the laboratory of the METRIS Research Centre for Materials in the Region of Istria in Pula, and at the Natural Science Laboratory of the Croatian Conservation Institute in Zagreb (using an Olympus BX 51 digital camera as well).9

• 10  Tea Zubin Ferri is credited for SEM-EDX cross-section imaging, pigment composition analysis and in (...)
• 11  Data kindly provided by Tea Zubin Ferri. The contrast in BSE images primarily depends on the diffe (...)

12Cross-section samples of yellow, blue, green, red, brown, and flesh colour, together with the ground layer, were imaged and analysed using scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDX). Tea Zubin Ferri performed the analysis at the laboratory of the METRIS research centre.10 As FEI Quanta 250 FEG (Field Emission Gun) scanning electron microscope was used, sample preparation (sputtering) was not necessary. SEM-EDX imaging was performed at low vacuum (100-120 Pa), 10-20 kV high voltage, working distance of 10 mm and spot size between 4 and 6. All paint cross-section samples were imaged using BSED (Backscattered Electron Detector), which facilitated their examination.11

• 12  The presence of lead in the white layer was proven subsequently by EDX analysis.

15It was presumed that the layer contained lead, which would have attenuated X-ray radiation and reduced the readability of the respective parts of the digital radiograph.12 However, the digital radiograph shows clearly discernible figuration in almost the entire surface of the painting. The areas with a higher degree of X-ray absorption (shown in white or lightershades of grey) mainly correspond with the light painted parts of the painting (flesh tones), which contain lead white. Nevertheless, it is notable that the contrast of the bottom half of the digital radiograph is lower than would be expected, which may indicate the presence of a layer of high X-ray attenuation. In addition, a difference in the degree of X-ray absorption between the upper and the lower portions of the painting is apparent: darker shades of grey and higher contrast (lower absorption) are observable in the upper half of the radiograph, while the lower half shows much lighter shades and lower contrast (higher absorption).

• 13  SEM-EDX analysis showed that the landscape was painted using green earth, a pigment consisting of (...)

16This corresponds to the stratigraphy of the paint layer. For example, dark green trees surrounding the images of Christ and Saint John the Baptist appear quite light in the X-ray digital image, indicating a higher degree of X-ray absorption due to the layer of lead white. The contrast of the radiograph is lower in these areas: the treetop in the background of Saint John the Baptist is hardly discernible.13 The presence of a white paint layer was determined by cross-section analysis. On the other hand, the light-painted sky with yellow clouds in the top part of the painting appears surprisingly dark in the X-ray digital image, which may be explained by the absence of high absorbing layer of lead white in this area. This layer is not present in cross-sections of the samples taken from this part of the painting.

• 14  HELWIG, K., ''Iron Oxide Pigments, Natural and Synthetic'', in Artist's Pigments, A Handbook of Th (...)
• 15  HELWIG, K., op. cit., 2007, p. 89.

22The reddish-brown preparatory layer is mainly composed of iron oxides with aluminosilicates. In addition to aluminium, silicon, calcium, magnesium, manganese, potassium and sulphur, chemical elements originating from accessory minerals commonly present in natural earth pigments from iron oxide deposits,14 traces of titanium are also present. Iron oxide pigments can contain a significant amount of titanium-containing minerals, which often occur in rocks and soils.15

• 16  HELWIG, K., op. cit., 2007, p. 51.
• 17  MERRIFIELD, M. Ph., Original Treatises, Dating from the XIIth to the XVIIIth Centuries, on the Art (...)
• 18  Cf. HRADIL, D.; GRYGAR, T.; HRADILOVÁ, J.; BEZDIČKA, P., ''Clay and iron oxide pigments in the his (...)
• 19  Cf. GROEN, K. M., ''Earth Matters, The origin of the material used for the preparation of the Nigh (...)
• 20  Cf. HELWIG, K., op. cit., 2007, p. 89, Fig. 32. d, an X-ray energy spectrum of a red earth from It (...)

23Coloured grounds (yellow-brown and red-brown), consisting mainly of earth colours with the addition of black and white pigments, were developed in northern Italy in the 16^th century.16 During the 17^th and the 18^th century they were predominant in Europe. For instance, Giovanni Battista Volpato's 17^th-century manuscript lists a recipe for preparatory layer made of finely ground argil, red earth and umber.17 Argil is generally of aluminosilicate composition. The most noticeable constituents of Fedrigazzi's ground layer are large, coarse orange grains (in some places yellow), which mostly contain Al, Si and Fe, possibly ground argil. Red earth commonly consists mainly of iron oxide (containing clay and quartz in various ratios, depending on the source). The colour in natural reds comes from hematite; coarse-grained, crystalline violet-red particles are visible by optical microscopy in brownish clayey matrix (large Al and Si peaks) of Fedrigazzi's ground, as well, though natural or artificial hematite could have been added intentionally.18 Large, colourless particles of quartz (relic mineral) are also visible. EDX analysis has confirmed high silicon and oxygen content.19 EDX spectrum of an Italian red earth shows considerable similarities in elemental composition with the spectrum of brownish matrix from the ground layer of Fedrigazzi's painting,20 which is not surprising, as he is of Italian descent. To conclude, a combination of these or similar earth pigments could have been used for the ground layer of the Kostanje painting.

• 21  WINTER, J.; WEST FITZHUGH, E., ''Pigments Based on Carbon'', in Artist's Pigments, A Handbook of T (...)
• 22  Cf. SPRING, M.; GROUT, R.; WHITE, R., 'Black Earths' – A Study of Unusual Black and Dark Grey Pigm (...)
• 23  SPRING, M.; GROUT, R.; WHITE, R., op. cit., 2003, p. 96.

24Black particles visible under the optical microscope exhibit splintery, angular and often elongated form, resembling those of charcoal black prepared from wood or fruit stone char.21 However, these particles have very high sulphur content, as observable in the sulphur EDX map of the brown paint cross-section (Fig. 11 b). Therefore, this is probably a sulphur-rich coal-type black; in some cases, this organic black pigment may cleave in a way that the particle shape is resembling a charcoal black, but no wood cell structure is observed.22 Sulphur-rich coal black falls in the class of black earth pigments, which comprise a number of different black minerals.23

• 24  THOMPSON, D. V., The Materials and Techniques of Medieval Painting, Dover Publications, Inc., New (...)
• 25  Today it is assumed that both massicot (or masticot, which is the name found in the manuscripts of (...)

27Composition of the yellow paint in the sample taken from the clouds above St. Mary is nearly identical to that of the white underpainting: EDX spectrum of yellow paint layer shows large lead peak, with significant carbon and oxygen peaks. This might indicate lead white mixed with one of the yellow lead pigments, which were particularly used in medieval Italy.24 Pigments from this group may contain tin or antimony, but EDX analysis has not confirmed their presence. Therefore, it can be assumed that the yellow pigment that Fedrigazzi used is massicot (lead oxide, PbO), a pigment that has been produced by calcination of lead white since the antiquity, as an intermediate product in the preparation of red lead.25 Traces of silicon, aluminium and iron are also present; hence small amounts of some kind of yellow earth or yellow lake are possible as well.

• 26  See, for example: SCHWEPPE, H., ''Indigo and Woad'', in Artist's Pigments, A Handbook of Their His (...)
• 27  The earliest recorded examples of indigo in oil medium are 13^th century Norwegian panel paintings. (...)
• 28  Though analytical evidence suggest that painters have continued to use it, indigo apparently drops (...)
• 29  While indigo was identified in only very few works by 15^th and 16^th century Northern European pain (...)
• 30  EIKEMA HOMMES, M. H. van, op. cit., 2002, p. 120, 121.
• 31  In all indigo dye plants, the main precursor of blue colorant indigotin is water-soluble indican. (...)
• 32  EIKEMA HOMMES, M. H. van, op. cit., 2002, 111, 119, 141.

29Although organic compounds cannot be identified by EDX analysis, the prevalence of use of indigo in historical painting, in comparison to other blue dyestuff sources, makes it the most probable option. The majority of published occurences of indigo in painting date from 13^th to 19^th century.26 A growing number of analytical reports suggest that indigo was quite commonly used in 15^th to 17^th-century oil painting, particularly in the 17^th century.27 The pigment was less frequently employed in 18^th and 19^th centuries.28 Notably, indigo has been documented in works by Italian 15^th and 16^th century painters, most of them being active in Venice, the main port of import for Indian indigo at the time.29 Due to increasing import of tropical indigo during the 17^th century, contemporary painting manuals almost exclusively refer to this dyestuff source.30 The main sources of tropical indigo are plants of the genus Indigofera. In Europe, the native woad plant, Isatis tinctoria L., was the most important source for indigo until the 17^th century.31 Studies have shown that markedly high amounts of both phosphorous and calcium might imply that the pigment derives from woad indigo.32 However, no traces of phosphorous were detected in the blue paint from Kostanje painting. Considering the date of origin of the painting and Venetian origin of Francesco Fedrigazzi, a few references on the use of imported, tropical indigo in oil painting are following.

• 33  EIKEMA HOMMES, M. H. van, op. cit., 2002, p. 145-155. In her dissertation, van Eikema Hommes lists (...)
• 34  EIKEMA HOMMES, M. H. van, op. cit., 2002, p. 150, 151, 152.
• 35  EIKEMA HOMMES, M. H. van, op. cit., 2002, p. 135, 136.

30Painters principally used indigo pigment to depict folds in drapery; in oil paintings dating from 15^th and 16^th centuries, the pigment was seldom used in uppermost paint layers. From the second half of the 17^th century, due to better quality of imported material and developed methods of pigment preparation (i. e. purification and calcination), indigo was no longer restricted to underpaint. When using the pigment in top layers, contemporary texts recommend mixing it with lead white and/or chalk for modelling blue colour.33 Fedrigazzi’s dark blue paint contains both calcium carbonate and lead white in significant amounts. Siccative properties of lead white could also be the reason for its use – since indigo oil paint dries very slowly, it was often advised to mix it with pre-processed drying oil and siccatives.34 Large relative amount of aluminium suggests the addition of calcified rock alum, which was one of the historically employed methods for purifying indigo and improving its permanence.35

• 36  SCHWEPPE, H., op. cit., 1997, p. 97.
• 37  Indigo and turnsole were predominant medieval sources for organic blue colour, hence their names w (...)

31Schweppe states that emission-spectrographic analysis can provide ''negative evidence for indigo''. The presence of iron would indicate Prussian blue, while copper, cobalt or sodium would point to azurite, cobalt blue or ultramarine, respectively.36 Since other historical sources of organic blue dyestuff have been documented in manuscripts and (only exceptionally) detected in easel paintings,37 further analysis is required to prove the presence of indigo or to point to a different blue colorant.

• 38  THOMPSON, D. V., op. cit., 1956, p. 163.

34As noted earlier, green earth is a pigment composed of clay minerals. It has a dull colour that can range from bright, bluish or greenish grey to dark olive green. Because of its transparency and low intensity, green earth was rarely used to paint landscapes in oil paintings.38 It is quite surprising that the colour that Fedrigazzi used to paint the landscape in the Kostanje painting is quite intense, with tones ranging from deep blue-green to warmer ochre-green.

• 39  GRISSOM, C. A., ''Green Earth'', in Artist's Pigments, A Handbook of Their History and Characteris (...)

35The explanation might lie in his painting technique – in the use of white underpainting, to be more precise. By applying the green earth as a glaze over a white surface, maximum saturation (chroma) is achieved, with a significant increase in excitation purity and lightness. In some parts of the Kostanje painting the green colour has a yellowish tone. This happens when the pigment (green earth) is glazed over or mixed with white.39

• 40  GRISSOM, C. A., op. cit., 1986, p. 146.

36Undiminished intensity of green colour in Fedrigazzi's painting could be attributed to stability of green earth, which is a permanent pigment, resistant to light and air.40 Darkening occurs only when oil or varnish penetrate the pigment particles. The fact that Fedrigazzi's painting was not varnished supports this assumption.

• 41  Cf. GRISSOM, C. A., op. cit., 1986, p. 150, 151, Fig. 9, and GETTENS, R. J., STOUT, G. L., Paintin (...)

37Microscopically, the samples of green colour from Fedrigazzi's painting indeed exhibit the characteristics of minerals that constitute green earth. The layer of green colour is a mixture of translucent, rounded particles of different shades of green, in places blue, with traces of yellow or brown earth.41

• 42  GRISSOM, C. A., op. cit., 1986, p. 148, 149, 143. The most significant use of green earth has been (...)

38Because of their similar chemical composition and microscopic characteristics, SEM-EDX analysis cannot distinguish glauconite (warmer tones, occuring more frequently, often containing impurities) from celadonite (bluish tones, less common, comes in a more pure form, famous historical source in Verona).42

• 43  Historically, the preparation of lake pigments from organic dyestuff was based on the precipitatio (...)

40The colorant in the uppermost pale purplish glaze could not be identified by EDX. This indicates organic dyestuff. The presence of aluminium and calcium points to a lake pigment substrate (hydrated alumina and calcium carbonate).43

• 44  SCHWEPPE, H.; ROOSEN-RUNGE, H., ''Carmine – Cochineal Carmine and Kermes Carmine'', in Artist's Pi (...)
• 45  SCHWEPPE, H.; WINTER, J., ''Madder and Alizarin'', in Artist's Pigments, A Handbook of Their Histo (...)
• 46  SCHWEPPE, H.; ROOSEN-RUNGE, H., ''Carmine – Cochineal Carmine and Kermes Carmine'' in op. cit., 26 (...)

41Red lake pigments, such as carmine from kermes scale insect (Kermes vermilio Planchon) or dyestuff obtained from brazilwood (Caesalpinia spp.), have been prepared by precipitation from aqueous extract using alum since the Middle Ages.44 Colorants obtained from madder (Rubia tinctorum L.) and other plants of the Rubiaceae family have also long been employed as lake pigments, commonly on an aluminium substrate.45 Lake pigments that are produced in this way give translucent films when mixed in oil.  It is for this reason that they were mainly used as glazes.46 Thin homogeneous translucent top layer in the paint sample taken from St. John the Baptist's robe is also applied in the form of glaze.

• 47  KIRBY, J.; SPRING, M.; HIGGITT, C., ''The Technology of Red Lake Pigment Manufacture: Study of the (...)
• 48  KIRBY, J.; SPRING, M.; HIGGITT, C., op. cit., 2005, p. 81.

42Although the origin of the red lake dyestuff cannot be determined using EDX spectroscopy, some conclusions may be drawn from the presence and proportion, or the absence of certain elements in the lake substrate. Studies have shown that both plants and scale insects from which the dyestuffs are extracted can contribute a number of elements: S, P, Si, Ca, K, Cl, Mg, Na and Cu, the most important in insects being P, S, K and Cl.47 If a significant amount of phosphorus is present in a lake substrate, the dyestuff is probably derived from insects.48 However, no traces of phosphorus were determined in the layer of red lake from the Kostanje painting.

• 49  KIRBY, J.; SPRING, M.; HIGGITT, C., op. cit., 2005, p. 71, 75, 76, 80.

43Furthermore, when the dyestuff is extracted directly from raw materials, it is much more likely that the elements present in the raw material will remain in the precipitated pigment (as opposed to the use of coloured textile shearings for the extraction of the dyestuff). A number of historical recipes suggest that lake pigments from lac and brazilwood were often prepared by direct extraction from source materials. Recipes for extraction of dyestuff from brazilwood list calcium carbonate (usually chalk) as a common ingredient, thus large amount of calcium, but also aluminium, can be expected in brazilwood lakes.49 The most significant elements in Fedrigazzi's red lake are, indeed, aluminium and calcium, as observable in the aluminium and calcium EDX maps (Fig. 9 b). This red lake also contains traces of silicon, chlorine and potassium, elements that may indicate the presence of a dyestuff of vegetable origin, extracted directly from the raw material.

• 50  Using, for example, ninhydrin test. KIRBY, J.; SPRING, M.; HIGGITT, C., op. cit., 2005, p. 76, 77, (...)

44The red lake from the Kostanje painting contains no traces of sulphur. Comparatively high sulphur content could have indicated a dyestuff obtained from the wool dyed with madder, especially if the presence of proteins was confirmed.50 Since no dull orange fluorescence typical of madder lakes was observed under ultraviolet light, it is safe to conclude that Fedrigazzi did not use madder.

• 51  THOMPSON, D. V., op. cit., 1956, p. 120, 121.
• 52  THOMPSON, D. V., op. cit., 1956, p. 120, 121. See also KIRBY, J.; WHITE, R., ''The Identification (...)
• 53  BERRIE, B. H., ''Rethinking the History of Artists' Pigments Through Chemical Analysis'', in Annua (...)
• 54  KIRBY, J.; SPRING, M.; HIGGITT, C., op. cit., 2005, p. 83. Brazilwood lake has been confirmed in t (...)

46Yet, a few remarks have to be made. According to some authors, brazilwood colour was vastly used for painting and for dyeing in the Middle Ages due to its cheapness and ease of use.51 Following the Spanish conquest of Mexico in 1512, carmine derived from the New World cochineal (Dactylopius coccus Costa) gradually replaced kermes carmine and brazilwood in textile dyeing and painting, becoming widespread during the 17^th century.52 Although the use of brazilwood lake has often been noted in manuscripts, the pigment seems to have scarcely been detected by chemical analyses.53 Furthermore, examples of paintings with brazilwood lake used as an underlayer have been documented, the reason for both probably being the well known impermanence of the dyestuff.54

• 55  See KIRBY, J., WHITE, R., op. cit., 1996, p. 70, 71 and KIRBY, J.; SPRING, M.; HIGGITT, C., op. ci (...)

47Bearing in mind that Francesco Fedrigazzi was an Italian painter, it should be noted that the New World cochineal dyestuff has been reported in red lakes from the majority of examined 17^th and 18^th-century Italian paintings.55 However, different analytical techniques should be employed to confirm brazilin or substances that would indicate other dyestuff sources.