A STUDY OF THE MATERIALS USED BY MEDIEVAL PERSIAN PAINTERS
NANCY PURINTON, & MARK WATTERS
3 PIGMENT ANALYSIS
3.1 REVIEW OF PREVIOUS WORK
VERY LITTLE substantive chemical or physical analysis of the materials of Persian-manuscript painting has been published. Other individuals have carried out pigment analyses of Islamic manuscript paintings previous to this study. Four unpublished papers discussed here were written by students at Harvard University. A fifth is an article by Gordus (1968), who was mainly interested in demonstrating the advantages of neutron activation analysis. With that analysis, he identified cinnabar, orpiment, and verdigris in an unidentified Persian painting. The sixth paper, by FitzHugh (1988), published in a catalog of the Vever collection at the Arthur M. Sackler Gallery in Washington, D.C., was a study of the pigments in some of the paintings of that collection.
Holmberg's paper (1978), one of the student papers, described the identification of pigments on pages from the Demotte Shahnama, so named because an early 20th-century dealer, Demotte, sold 58 individual pages from an early 14th-century copy of Firdawsi's epic poem, the Shahnama (Book of Kings), to different patrons. Holmberg examined 11 paintings to determine whether they had been retouched. The analytical methods used were microscopic examination, microchemical testing, energy-dispersive x-ray analysis, semiquantitative laser spectrographic analysis, and x-ray diffraction. Miniatures in three separate collections were examined: three in the Freer Gallery of Art, Washington, D.C.; two in the Museum of Fine Arts, Boston; and six in the Fogg Art Museum, now in the Arthur M. Sackler Museum, Harvard University. Holmberg concluded that all the miniatures had been retouched and sometimes she found evidence of more than one layer of retouching.
St. Laurent-Lockwood (1981) sampled blue, white, and yellow from four Persian paintings in the Sackler Museum, Harvard University. She found that the only blue pigment used was ultramarine, the only white was white lead, and the only yellow was orpiment.
The other two student papers were studies of royal Persian paintings and compared royal workshop paintings to provincial ones. One, by Khoury (1985), was an analysis of four royal Persian miniatures from the 16th and 17th centuries. Khoury found that the technique used in the execution of the miniatures was the “classical technique,” except in one case in which the artist was particularly innovative. She found no Indian yellow in the Persian paintings. The fourth student paper by Bailey (1985) compared four paintings, two royal and two provincial, and found that the traditions for both were the same but the provincial works demonstrated the need to economize by using, for example, carbon black to imitate tarnished silver.
3.2 REPORT OF THIS PIGMENT ANALYSIS
According to our observations, pigments were rarely used as pure color; the artists apparently preferred mixing their pigments for more subtle coloristic effects. The pigments used in a medieval Iranian kitabkhana, or workshop, were probably manufactured there. Some of the pigments used by the Persian artists were produced from mineral ores and, in some cases, semiprecious stones. The purity of the color in those pigments would depend on the quality of the natural mineral and the care of the preparation.
The pigments of medieval Persian manuscript painting can be conveniently divided into two chemical groups, inorganic and organic. Apparently, the artists relied much more heavily on inorganic pigments, and some scholars insist that only inorganic pigments were employed in Persian painting (Behzad 1939). The artists must have valued inorganic pigments for their superior covering power and their relative physical and chemical stability in addition to their resistance to fading. Many of the inorganic pigments were prepared from mineral ores, while others were chemically manufactured. In the finest examples of Persian painting, the purity of the colors as well as the precision displayed in the finish attest to the careful preparation of the pigments. Ore that is relatively unadulterated is necessary for the production of a good natural mineral pigment. Generally, the raw material must be fractured and coarsely ground. To purify and reduce the mineral to a powder suitable for use as a pigment, washing in water and levigation—a procedure of grinding and washing in which fine particles float while coarse ones sink—should follow. In most cases, the finest grades of pigment are removed in the first washings and levigations. Alkaline washings can be employed to facilitate the separation of the finest grades as well (Minorsky 1959).
Some earth pigments derived from clays are simply mixed in water. While the desired portion of the clay remains in suspension, sand and heavy impurities sink to the bottom and humus floats to the top. Chemically manufactured pigments often require grinding. When a pigment is produced by altering another finely ground pigment, further grinding may not be necessary. The procedures outlined here are essential to the manual preparation of fine pigments in all parts of the world.
Organic materials constituted the second major group of pigments. They must have gained acceptance as painters came to value an enhanced variety of color harmonies. Many of these pigments were derived from extractions of vegetable matter or from insects. In some cases, the raw organic material, when properly powdered, could serve directly as a pigment.
A tendency to fade and a lack of covering power characterize organic pigments, and most were likely relegated to a subordinate role in painting.
Although pigments known as lakes have been used in Western artworks, the use of lakes in Persian painting has not been studied. To produce a lake, the raw material, usually an intensely colored dye or stain, was precipitated on a finely powdered, colorless or white inert substrate such as calcium carbonate (chalk or limestone) in order to function as a pigment. The final pigment color could vary depending on the substrate. The gradually expanding range of colors found in Persian paintings from the 15th and 16th century and the ready availability of organic lake pigments used at that time in the textile industry strongly support the hypothesis that these pigments were employed in manuscript paintings (Wulff 1966, 189–92).
3.2.1 Blue Pigments
Most scholars of Persian painting have acknowledged natural ultramarine as the most important blue pigment in the Persian painter's palette (Welch 1972; Laurie 1935; Martin 1912, 108). The use of ultramarine was mentioned by Sadiqi Bek and in the appendix to Qadi Ahmad's treatise. Gettens and Stout (1966) reported early occurrences of ultramarine as a pigment in the 6th and 7th centuries in the Iranian plateau. Lapis lazuli, the mineral from which natural ultramarine is prepared, was readily available from the mines of Badakhshan along the shores of the upper Oxus (in modern northern Afghanistan), a source known since antiquity (Laufer 1919, 520). The brilliant blue of ultramarine was extracted in an alkaline solution (lye) from a kneaded wax resin ball containing powdered lapis lazuli. The blue was then thoroughly washed to remove any residual alkalinity. Laurie (1939) suggested that the improved method of preparation for ultramarine was first developed in Europe and later transmitted to Iran in the 15th century. However, he provided no conclusive evidence as a basis for his suggestion.
In this investigation ultramarine blue was found to be one of the two most commonly used pigments, white lead being the other (see table 1). The ultramarine was very finely ground for Rustam Approaching Tents from a 16th-century copy of the Shahnama at LACMA (M.73.5.592). It was also found in a more coarsely ground form in a chapter heading from a Khamsa of Nizami (M.73.5.606). Both were executed in the 16th century, so it appears that different types of preparation methods could have existed in different workshops or for different purposes.
TABLE 1 Pigments Identified in 19 Medieval Persian Paintings in the Los Angeles County Museum of Art
Another blue pigment used in Persian painting was azurite. Holmberg (1978) found azurite in the Demotte Shahnama, made in the 14th century. The paintings analyzed in her study date from the 15th through the 17th centuries, and azurite was found in six of them. It may have been more widely used in earlier times according to Laurie (1935), who described a blue in a 14th-century Persian manuscript as azurite but without any analytical evidence. Farooqi (1977) did not even list it as part of the Persian or Indian palette. Vajifdar (1981) did, however, include azurite in his list of Persian pigments.
The Chinese, who employed azurite more than any other blue pigment, may have exported it to Iran (Schafer 1963). Wulff (1966, 3) has mentioned that northern Iran is rich in copper carbonate ores. Since azurite is a basic carbonate of copper, the raw material for the pigment may have been obtained there. Ground comparatively coarsely, azurite provides a deep blue pigment; it loses its color when ground too finely.
Another blue pigment, indigo, is a vegetable extraction from Indigofera tinctoria and other plants and was introduced into Iran from India in the middle of the 6th century according to Laufer (1919, 370). It had long supplied textile dyers with a deep blue stain. As a blue pigment, it is relatively stable and has good covering power. The use of indigo as a pigment was documented for neighboring regions by Gettens and Stout (1966), suggesting its importance in the Persian palette.
Laurie (1935) also listed two organic blues he felt may have been employed in Persian painting. One was indigo, and the other was more of a purple, the Tyrian purple made from shellfish. Vajifdar (1981) and Farooqi (1977) mentioned indigo. It cannot be certain that the organic blue found in this study is indigo, but the 16th-century scene of a Prince and Princess in a Garden (M.73.5.16) does contain an organic blue pigment. This evidence is contrary to Behzad's (1939) contention that only mineral colors were used in Persian paintings.
3.2.2 Red Pigments
Another organic pigments found during this analysis was a red used in a Youth Pouring Wine M.73.5.570). This painting is unusual among the paintings in this study because it was produced as an independent work of art in 1681 and not as an illustration in a narrative manuscript. We cannot determine what type of red organic pigment this is, but there are several possibilities.
Wulff (1966, 189–90) has commented on the presence of the kermes (Kermes vemilio) and the cochineal (Coccus cacti) insects in parts of Persia. The red colorant carmine was extracted from the bodies of the dead female insects and precipitated on a white or colorless inert substance to produce kermes (grain or crimson) lake and cochineal lake. The kermes lake is one of the most ancient of the natural dyestuffs, and cochineal lakes were imported to Europe from the New World in the 16th century (Gettens and Stout 1966). The root of the madder plant (Rubia tinctorum), which was cultivated in Persia, provided the deep red color of madder lake (Wulff 1966, 190). Another red lake, imported from India, derived its color from the exudation of the lac insect (Coccus laccae) (Schweppe and Roosen-Runge 1986). As already noted, whether Persian painters made use of organic colorants as lakes or in pure form has not been investigated.
The other red pigments found in our study were all inorganic and are also mentioned in the literature. Cinnabar or vermilion (red mercuric sulfide, HgS) can be produced by grinding the mineral cinnabar (referred to as cinnabar) or it can be manufactured from mercury and sulfur (to make vermilion) (Gettens et al. 1972). These were the most widely used red pigments found in this study, but they have been combined in table 1 because the process used to indicate the mercury-containing pigment (x-ray fluorescence) did not allow for differentiation. Rich in sulfide ores, Iran may well have supplied the demand for vermilion within its own borders (Wulff 1966, 3). It has been found in many works of art, even in antiquity (Filippakis et al. 1979).
The appendix to Qadi Ahmad's treatise offered instructions for making the red from cinnabar, and Sadiqi Bek described the manufacture of vermilion by pounding mercury and sulfur together before heating (Minorsky 1959; Dickson and Welch 1981). From earlier times, the pigment had also been manufactured from mercury and sulfur (Kopp 1843–47; Cennini 1960, 24). This red has long been valued for its stability and excellent hiding power. Many other authors have referred to the pigment, so this information reconfirms the general opinion regarding the importance of vermilion (or the natural form in the mineral cinnabar) for Persian artists.
Several iron oxide reds (earths or ochres) may have been employed by Persian artists, including a brick red, sometimes called Venetian red, and a deep purple-red rich in the mineral hematite, also known as Indian red. This is an ochre, one of the earth pigments which include clays, ochres, siennas, and umbers. Ochres consist of silica and clay, and red ochre is colored primarily by anhydrous iron oxide (the mineral hematite). Extensive deposits of iron oxide exist all over the world, one particularly good-quality source is the island of Hormuz in the Persian Gulf (Gettens and Stout 1966). Iron oxide has been used extensively as a pigment throughout history and prehistory, and it is no surprise that it is found in these paintings. It is mixed with white lead and vermilion in two paintings (M.73.5.412 and M.73.5.437) to give a richer, cooler red. Of the red pigments, iron oxide red was the second most frequently occurring red pigment in this study.
Red lead is the only other red pigment identified in these paintings. Laurie (1935) considered it to be a particularly fine orange type of red lead and observed it in Persian paintings as early as the 13th century. Red lead is produced by roasting white lead. The result is red tetroxide of lead (Pb3O4) that is, as Laurie correctly identified it, a very finely textured orange-red pigment. Red lead is less stable than the other reds found in these paintings, a characteristic that might explain why it is used less often. It was identified in only five of the paintings studied here, and in none of those was it the only red pigment used.
Another red pigment mentioned in the literature is realgar (Vajifdar 1981). This is an orange-red sulfide of arsenic that is closely related chemically to and associated in nature with orpiment. Qadi Ahmad's appendix mentioned a “red orpiment” that resulted when yellow orpiment is ground exceptionally fine. He claimed that the more it was ground, the redder it became (Minorsky 1959). The pigment described is more likely to be realgar, the natural sulfide of arsenic. This is a separate pigment and does not result from grinding orpiment. It is not as plentiful in nature as orpiment, and it seems reasonable to expect that it would be found less often as a pigment. It was not found in this study.
3.2.3 Yellow Pigments
A variety of yellow pigments were mentioned in the literature. Some, like orpiment and yellow ochre, were from mineral sources. Others, like gamboge and Indian yellow, were organic pigments (Minorsky 1959; Laurie 1935, 1939; Vajifdar 1981). Other inorganic yellow pigments, like massicot, could be manufactured. Not all the yellows mentioned in the literature were detected in this research.
The most popular yellow pigment was orpiment, affirming Gettens and Stout's (1966) opinion that orpiment as a yellow was widely used, especially in the East. Wulff's (1966, 3) comment on Iran's wealth of sulfide ores suggests that orpiment, the yellow sulfide of arsenic, was available from local sources. The mention of orpiment in the appendix of Qadi Ahmad's treatise provides additional evidence of its importance in the Persian palette (Minorsky 1959). This pigment is no longer used because it is poisonous (Gettens and Stout 1966). Orpiment may have fallen out of favor because of a technical problem as well, for it can cause white lead to discolor and turn gray (White 1984). The two pigments do not need to be mixed together or applied on top of each other for this discoloration to occur; it can occur even if the orpiment is only adjacent to the area of white lead. An example of the discoloration of adjacent paint is seen in the faces of the Flaying Scene (M.73.5.437). In the painting Laila and Majnun at School (M.73.5.417, fig. 2) orpiment and white lead were mixed together before application, and the yellow robe worn by a figure on the right side of the schoolroom shows discoloration. This problem of discoloration did not deter artists from using orpiment, however. It was used throughout the three centuries investigated.
Laila and Majnun at School, miniature from manuscript of Khamsa by Nizami, Persia, Shiraz, 924 A.H. (A.D. 1517). Opaque watercolors on paper, 24.0 × 14.0 cm. Los Angeles County Museum of Art, Nasli M. Heeramaneck Collection, Gift of Joan Palensky. M.73.5.417
The other yellow pigment mentioned in the literature was yellow ochre, and it was found in only three paintings. All three were earlier paintings from the 15th and 16th centuries, depicting The Appearance of Sakyamuni (M.73.5.412), Laila and Majnun at School (M.73.5.417), and Lady with a Fan (M.73.5.587). We cannot be sure if there is any significance to the fact that yellow ochre was found only in earlier paintings. Further study may disclose any possible trends. According to this research, Persian artists preferred inorganic yellow pigments, possibly due to their availability and color stability. The problems mentioned with orpiment notwithstanding, its color is stable.
Organic yellow pigments were not found in this analysis, but they may have been used. Indian yellow, a brilliant, stable, yellow pigment prepared in Bengal from the urine of cows fed on mango leaves, may have been used by Persian painters. Saffron, a rich golden yellow powder from the flower of the plant, Crocus sativa, that was cultivated from Sassanian times in Iran, probably could have served directly as a pigment (Laufer 1919, 320). The Persian painters may have found this powder particularly useful in combination with gold pigment. Because saffron has always been dear, less expensive colors such as the orange from safflower (Carthamus tinctorius) or the yellow from shoots and roots of tumeric (Curcuma domestica and Curcuma longa) were often combined with saffron as adulterants (Laufer 1919, 309–10; Wulff 1966, 191). Wulff (1961) has indicated that both colors were employed in Persian textile production as well. Iranian textile dyers employed yet another yellow prepared from Persian berries (Rhamnus infectorius) as a lake; it might have been used as a pigment, too.
3.2.4 Green Pigments
The most common green pigment mentioned in the literature on Persian painting techniques is verdigris (Dickson and Welch 1981; Laurie 1935; Vajifdar 1981). Historically, verdigris is a general term for green corrosion products that form on copper, brass, or bronze. Sadiqi Bek gave the most complete instructions for the making of verdigris. One must dig a pit two meters deep and in it bury copper plates immersed in wine vinegar for a month (Dickson and Welch 1981). Other recipes include vinegar and copper but can include salt and sugar (Farooqi 1977). The various recipes produce verdigris with different chemical compositions. Kühn (1970) found that verdigris could contain one copper acetate or a mixture of different copper acetates. Depending upon its chemical composition, verdigris can have a basic or neutral pH. The neutral verdigris (Cu(CH3COO)2H2O) is produced by grinding basic verdigris in strong acetic acid; this could be accomplished by recipes that include vinegar. Banik et al. (1981) investigated the deterioration of green copper pigments and the destructive effect on paper supports. In several of the paintings examined, copper greens have turned the paper below it brown and finally made it so weak that it cracked and broke apart.
Malachite is found in nature as a mineral. It is a basic copper carbonate, chemically similar to azurite and occurring with it but with different proportions of carbonate to hydroxide. The synthetic product is known as green verditer (Banik et al. 1981). Malachite, although a natural mineral, can cause degradation in a paper substrate at elevated temperatures (Banik et al. 1981). There was only one questionable occurrence of this pigment in the LACMA paintings. Holmberg (1978) did not find malachite in the Demotte Shahnama.
Green earth pigment may have been employed by Persian painters as well. Its simple preparation and its availability in the form of green clays (or earth) have contributed to its early and continuing widespread use in all parts of the world (Gettens and Stout 1966). Green earth was not found in this study or by FitzHugh (1988) in the Vever paintings.
We did identify some other copper greens in the paintings from the LACMA collection. One of the unusual greens, a basic copper sulfate identified by means of x-ray diffraction, was brochantite. Our observations are preliminary and tentative, but the brochantite seems to have caused less damage to the paper substrate than the basic copper chlorides, atacamite and/or paratacamite. The painting that contains brochantite (M.83.27.2) has been mounted so the reverse of the painting cannot be examined. Plesters has also identified brochantite by x-ray diffraction in a sample of powdered artificial green supplied from old stock given to her by an English colorman (Gettens and FitzHugh 1974).
Farooqi (1977) gave the ingredients for an “artificial verdigris” as orpiment and indigo. Such a pigment, a green that was a mixture of orpiment and an organic blue pigment, was found in the trees of the image Prince and Princess in a Garden (M.73.5.16). In that same painting, however, a copper green was found in the rug used by the royal couple. The paper beneath the copper green pigment has turned brown while the paper under the “artificial verdigris” has not. A similar situation with regard to degraded paper substrate in conjunction with two green paints was observed in Laila and Majnun at School (M.73.5.417). The chemistry of copper green pigments, and their effect on paper, require further research.
3.2.5 White and Black Pigments
The two most essential pigments in any painter's palette are black and white. The unanimous choice for white in these paintings was white lead. This finding is consistent with treatises, which mentioned white lead as the only white used (Minorsky 1959; Dickson and Welch 1981). Other writers, however, listed other possibilities: whiting, chalk, and zinc white (Laurie 1935, 1939; Vajifdar 1981; Farooqi 1977). Whiting and chalk, are two natural forms of calcium carbonate (Gettens and Stout 1966). Zinc was detected in one of the LACMA paintings (M.73.5.417) with x-ray fluorescence spectrometry, but only in a tiny and insignificant amount. We found titanium only in inpainted areas of one painting (M.83.27.2). Zinc and titanium whites were not commercially available until the 19th and 20th centuries respectively (Gettens and Stout 1966) and so would be found in inpainted areas. St. Laurent-Lockwood (1981) also found white lead to be the only choice of Persian painters.
Chandra (1949, 19–20) suggested that zinc white might have been available in Iran long before its initial mass production in Europe at the beginning of the 19th century. He used the term safeda, however, which could refer to zinc white or to white lead, so it is impossible to determine which pigment he meant (FitzHugh 1988).
Black pigments are mentioned by Laurie (1939) in his discussion of the Persian painters' palette. He thought the material of the black pigment was lampblack, an understandable conclusion since lampblack is the only black mentioned in some treatises. Bukhari (1963) and Vajifdar (1981) mention lampblack as a black pigment. In the appendix to Qadi Ahmad's treatise, however, are four recipes for black inks (Minorsky 1959). Two require lampblack, or soot, combined with a gum in addition to other ingredients. The lampblack was produced by collecting the soot deposited on an inverted earthenware bowl from a lamp burning hemp oil. Baking the collected soot in the center of a dough ball extracted oil residues and resulted in a pigment consisting largely of amorphous carbon, a deep black powder. The other two recipes require charred wheat starch or tin combined with quicksilver (mercury) to provide a black substance (Minorsky 1959).
In all the black paint samples in this study only one black pigment, charcoal, was identified by polarizing light microscopy. Charcoal is not mentioned as a pigment in the literature on Persian painting techniques; it is mentioned only as a material used in underdrawings. Charcoal was used in the form of a charred tamarind twig as a drawing tool or powdered and then pounced (by means of a cloth bag) through pricked stencils used to transfer drawings (Titley 1983, 216). With powdered charcoal already available to them, it makes sense that Persian painters would use it as a pigment.
In her analysis of the pigments in the Demotte ShahnamaHolmberg (1978) also suggested that the black pigment was charcoal rather than lampblack. Perhaps lampblack was not used as a black paint in paintings or in illuminations, but the recipes mentioned are only for black ink. Was it used as an ink? We could not explore the question extensively, but the ink sample examined was lampblack. Therefore, there may well have been separate types of black color used in separate workshops. Perhaps the finer texture of the lampblack made a superior ink for writing, while the coarser-grained (and more easily produced) charcoal made a better color for the painter. The particles that result from lampblack, made by collecting the soot of burned oil, are tiny and round, while the particles from ground charcoal are larger and more fractured in appearance under the microscope (Gettens and Stout 1966). Using a polarized light microscope to view particle morphology, we concluded that the paintings examined for this study did not contain lampblack as a pigment.
Metallic powders crowned the precious palette of Persian miniature painting. Iran was rich in alluvial gold and copper ore (Wulff 1966, 2, 13–14; Laufer 1919, 510–12). The mines of Badakhshan supplied Iran with silver. Two forms of colors are recorded as being used: ground particles in a binding medium and metallic foil applied on a binding medium (Dickson and Welch 1981). The paintings examined did not contain gold leaf used as a color. Gold was used, but in the form of ground particles in a medium. This finding corroborates Laurie's (1935) observations that a gold paint was used on most Persian paintings.
The preparation of the gold paint was a laborious process. First, gold leaf was made by pounding gold between layers of deerskin until the gold was extremely thin. Then the gold leaf was pounded on iron or stone with glue until it formed paste. This paste was dissolved in water and allowed to settle. The water was poured off, and the sediment was mixed with dry glue and saffron. The gold paint was then ready for use. Another possible preparation of powdered gold included mixing the gold with honey or glue between one's fingers, which could take hours, and then dissolving the resulting paste in water. After the gold was applied with a brush, it was “cooked,” meaning burnished, until it was a bright glossy gold surface (Behzad 1939). St. Laurent-Lockwood (1981) describes the mixing of gold with copper to make the gold a warmer color and with silver to make it cooler. In Youth Pouring Wine (M.73.5.570), gold was mixed with silver for the vessel the youth holds, which is a noticeably cooler gold than the gold of his boots. This metal combination was compared with other gold areas in other paintings by x-ray fluorescence and confirmed.
Laurie (1935) mentioned a “heavy dull grey gold” used in Persian paintings of the 13th century. Although he observed it in later works, it was rare in subsequent centuries. An example of this gray-gold may be present in a Battle Scene from the Big Head Shahnama (M.75.24). In this painting a charcoal gray was applied over the gold helmets of the warriors.
Silver was also used, but is now only seen as black in the paintings because it is tarnished. The water in the image of King and Sage in Landscape (M.73.5.29) is silver, as are the gray steps in Zahhak Enthroned with Two Sisters (M.83.27.2). Holmberg (1978) also found tarnished silver during her study. When the silver is as tarnished as the examples in these paintings, the techniques used for this analysis cannot determine whether the silver was applied as particles in a medium or as silver leaf. The Persians must have known that silver tarnished because Bailey (1985) found carbon black where silver would have been expected in a composition from Shiraz. This Shiraz painting could have been copied from an older, tarnished model.
Any identification of pigments employed in Persian painting by simple visual examination is highly unreliable. Differences in the source and manufacture of each pigment, which produce variations in the finished color, complicate the problem of identification. The large body of technical information on painting techniques, tools, and pigments used in Western paintings has proved valuable in dating, authenticating, and distinguishing between original painting and restoration. Persian painting techniques and tools are so different from Western painting traditions that a new body of information needs to be assembled before it can be applied to the study of Persian paintings.