The discovery of three lost Salting carpets

In 2007 three Salting carpets were discovered in the Palace of the Dukes of Bragan a, in Guimarães. Two are prayer rugs with central niches, while the third has a central medallion. Finely knotted in wool on a silk foundation, and embellished with metal threads, this is the largest collection of these carpets known outside the Topkapi Saray (Istanbul). For a century, their origin and chronology have been the source of considerable debate. They take their name from the collector George Salting (1835-1909) who donated a significant example to the Victoria & Albert Museum (London), which was originally attributed to 16th-century Iran. However, the vivid colors of this and other carpets, and existence of similar examples in the Topkapi, led later historians to argue that they were Turkish rugs from the 18th or 19th centuries, and possibly even forgeries of classical Persians carpets.

Introduction

They derive their name from the famous Australian-born British art colector, George Salting (1835-1909), who gave a particularly significant example to the Victoria and Albert Museum (London) in 1909 [1]. Defining the “Saltings” as a group has proven a difficult task, but, in general, the presence of a silk foundation (warps and wefts), fine wool pile, high knot count, bright colours and embel ishments in metal thread, along with arabesque designs, religious inscriptions and small dimensions, are regarded as the most significant characteristics [1]. The question of their provenance and chronology has also been the source of continuous debate. At the end of the 19th century, they were original y ascribed by late 19th- and early 20th-century historians to 16th-century Tabriz (Iran) and were recognized to be part of a large group of wel-preserved prayer rugs, with bril iant colours, in the Topkapi Saray1. As some of the religious inscriptions on the prayer rugs are almost exclusively Shi‟ite in nature, it was originally proposed that they had survived in such good condition in Turkey as they were not suitable for use by the Sunni Ottomans. However, their presence in Istanbul encouraged subsequent authors, such as Tattersall (1931), Erdmann (1941), and Rogers (1987), among others, to claim they were 18th- or 19th-century Turkish rugs, and possibly even forgeries of classical Persian carpets. More recently, a new generation of carpet experts has returned to the subject and argued the „Saltings‟ are 16th or 17th century, and that the Topkapi rugs were sent from Safavid Iran to the Ottoman court. Michael Franses proposed that the carpets could have been sent from Shah Tahmasp I (r. 1524-76) to Suleyman the Magnificent in 1556, as suggested by a letter mentioning a gift of “carpets spread on the floor of the mosque for the use of the congregation”, or a decade later, to Sultan Selim II (r. 1566-1574) on the occasion of his accession in 1567 [1]. Jon Thompson has recently suggested, on the basis of an inscription on one of the carpets giving the date of 1590-91 (AH 999)2, that they were a gift from Tahmasp‟s grandson, Shah Abbas I (r. 1587-1629), to the Ottoman Sultan Murad III (r. 1574-95), possibly on the occasion of the Treaty of Istanbul signed in 1590 [2]. Given this complex historiography, resolving the question of the origin and date of the „Saltings‟ is obviously imperative, and the discovery of three carpets in Guimarães offers a unique opportunity to reflect upon these problems. The type of materials used to make a carpet and how they are prepared can reveal groups of carpets and help to establish their origin and/or date [3]. This type of data can also indicate whether these groups reflect a single, coherent production of short duration, or multiple workshops and a wide historical timeline. Only partial scientific information is currently available for the „Saltings‟ [1], and hence, this study takes a multidisciplinary approach to the problems of their origin and date. Analyses are conducted on all of the materials used to make the three Guimarães carpets: namely, their fibres, dyes, mordants, and metal threads, and this data is then compared with published results for related examples documented in the literature to determine whether they are 1) products of Persian or Turkish workshops, and 2) date from the classical period (16th to 17th centuries), or are later, from the 19th or even 20th centuries. For tackling questions of provenance, yel ow and red dyes are especially important, as the former are usually locally produced [4-7], while the latter in some cases have been shown to be cultural y and temporally specific [3]. In previous work on Islamic carpets the red dye obtained from the lac insect, harvested in India and South Asia and widely traded throughout the region, has been shown to be the main colourant for the red ground of 16th- and 17th-century Persian wool carpets [8-10]. The previous detection of this dye in seven prayer carpets in the Topkapi collection would appear to support a Persian (as opposed to Turkish) provenance for the group [10-11]. Identification of the precise insect lac species would provide important evidence for associating the carpets with specific centers of production. For instance, several lac species are found in India (Kerria lacca), while others species are cultivated in China (Kerria chinensis or Kerria mendingonsis) [12] (Appendix I). However, no published study to date has used the HPLC-DAD (High Performance Liquid Chromatography Diode Array) technique to differentiate the various species of lac insects in order to localize precise sources of production, as has occurred, for example, for cochineal or dragon‟s blood [13-14]. Owing to the high complexity of the Kerridae family and the lack of accurate entomological studies on lac species, it was not possible to obtain specimens of the 26 species referred to in literature [6,12,15]. Nevertheless, for the first time, several lac-dye samples from different geographical regions, as well as from different host plants, were analysed using HPLC-DAD and PCA (Principal Component Analyses). This foremost lac-dye comparative database can support and promote further research on lac-dye species, as wel as reveal more detailed information about the source of the red dye and therefore assist with provenance and identification of historical textiles.

As  for the question of  the chronology  of  the  „Saltings‟, dye  analysis  is  also useful,  as  the 

identification of synthetic dyes would provide good evidence for a late date, as demonstrated by Halpine for the Gobelins tapestry-upholstered furniture in the National Gallery of Art (Washington) [16], while AMS Radiocarbon Dating (C14-AMS) can confirm an earlier one. Previous C14-AMS analyses conducted on a smal number of „Salting‟ carpets [9] have indicated a 16th- to 17th-century date for them, but for unknown reasons a number of measurements gave “too high a carbon age” (circa 30% of the C14-AMS total analysis performed), and this question needs to be resolved [17]. The composition and morphology of the metal threads is also relevant for establishing a date, although work in this area is still in its infancy. Final y, the Guimarães carpets reveal an intense abrash (dappled tones) which is usual y attributed to subtle changes in dye recipes, and an indicator of the presence natural dyes. When present in moderation, abrash has been considered a sign of high artistic quality [18], but, in the case of the Guimarães rugs, it could also be a consequence of degradation effects, owing to their exhibition under excessive light conditions for many years [19]. Mordant analysis and quantification of the dye content is a first step for resolving this question.

Technique

The structure of a carpet consists of warps, wefts and knots, and two processes are combined, weaving and knotting. The warps are the vertical threads that run the length of the loom and, when cut, form the fringes at the ends; the wefts run horizontally across the warp. Patterns are created by the pile-knots which are wound around one or two warps. After each row of knots is complete, the wefts are inserted across the carpet, and beaten down tightly with a comb to form a strong, resistant fabric. As each coloured knot is applied in sequence, along successive rows, carpet patterns are constructed using a pointil ist technique, and involve laborious pre-planning of the design [20]. The level of detail achieved in the pattern depends on the volume of the interlaced structure created by the warp, weft and knot, and hence, when the fibre diameter is low, designs with greater detail are possible. In the finest carpets, silk is used for all three threads, owing to its high strength in relation to its diameter. While horizontal and vertical designs can be created easily using the grid of warp and weft, diagonal and curving lines require a relative high knot density (more than 3,500 knots/dm2), and silk is often used for the warp (and weft) in wool carpets to enhance the knot count. This is the case for the „Saltings‟, and an S-twisted silk thread is used for both the warp and weft to increase the knot count to render very detailed designs, including the tightly curving and scrolling vines which are typical of their designs. The knots are wrapped around two warps in an asymmetrical format, open to the left (figure 1), which also enhances the detail, as these knots are more compact and the pile is more evenly spaced. In addition to knotting, carpet patterns can also be enhanced with decoration added in tapestry technique. Metal ic threads are especial y popular, and in the „Saltings‟, decorative threads comprising a silk core wrapped with a metal strip are introduced in selected areas, such as the inscriptions.

Style and Decoration

Although the precise date of the „Salting‟ carpets has been debated for over a century, their design characteristics reflect Safavid art, which was given a major impetus under Shah Tahmasp I (1514 - 1576), who was an important patron of the art of the book, but also commissioned magnificent carpets and other objects [2]. Under his patronage, artists from Herat migrated to Tabriz, resulting in a fusion of the two artistic styles prominent in Iran prior to the rise of the Safavids: the art of the Timurids and Turkmans. This new style features tchi clouds, angels, pheasants and multiple layers of winding scrolls as major decorative elements [2]. Figures and animals were also used increasingly, and integrated into compositions with geometric framing devices, such as elliptical and star-shaped medallions, cartouches, quatrefoils and poly-lobed forms, as seen in the borders of the „Salting‟ carpets. The splendour of this style stagnated after Shah Tahmasp became disinterested in the visual arts around 1555, and was only revived again with the rise of Shah Abbas I (1587-1629) [2]. At this time, the artists of the re-established court library began to generate a new decorative style of floral and foliate decoration, which, in accordance with the Shah‟s taste and political programme, was implemented across a wide range of arts, from il umination and bookbinding to textiles, carpets, and architectural tiles. In contrast to previous periods, the design elements are larger in scale, bolder in form and more homogeneous, and the same split-leaf arabesque can be found in bookbinding, a „Salting‟ carpet, and building decoration (figures 2 to 4). A new colour palette was also introduced under Shah Abbas. In textiles, there was a preference for yellows, while the dark blues and reds typical of the 16th century gave way to a palette of softer shades of beige, pink, light green, and light blue [20]. An elaborate use of metal thread for decorative elements, as wel as large areas of the background, to create a visual effect of splendor and luxury were also typical of the Shah Abbas period [22]. The „Saltings‟ are enigmatic in sharing characteristics of both styles. On the one hand, they have poly-lobed medallions, cartouches, and a rich colour palette consistent with the Tahmasp style, and on the other, they have large-format arabesque scrolls and extensive use of metal thread usual y associated with textiles produced under Shah Abbas. This mixture of styles is in part responsible, along with the bright colours of the Topkapi carpets, for associating the „Saltings‟ with later Turkish production and for questioning their authenticity.

The Guimarães Carpets

Two of the three „Salting‟ carpets found at the Palace of the Dukes of Bragan a (Guimarães) are small prayer rugs, symmetrical along the longitudinal axis, with a niche occurring in the central field, a wide colour palette, religious inscriptions, and are profusely decorated with metal-wrapped thread. The first rug, previously in the collection of Vital Benguiat and known as the “Benguiat Prayer Rug” (figure 5), has vine scrolls decorating the red field, with the upper corners filled with irregularly-shaped areas of colour with short inscriptions, and a continuous inscription outlining the central niche. The main border has inscribed cartouches in red and green which alternate with lobed medallions on a beige ground. This carpet is related stylistically to six rugs preserved in the Topkapi Saray Museum (Istanbul), and five in private collections or of unknown provenance.3 The carpet is in a poor state of conservation, and exhibits intense abrash, visible mainly in the blue, green and red areas of the upper corners. The second carpet, known as the “Duff Prayer Rug” (figure 6), has a field of interlocking hexagons, with the upper corners fil ed with vine scrolls, while inscribed cartouches alternating with medallions in the upper half of the border. The inner border and niche also contain running inscriptions. Red predominates as the ground colour throughout. The hexagon field of this carpet is found in only one other example, in the Topkapi Saray (Istanbul)4, but the format of the border with inscriptions located exclusively in the upper half can be seen in ten additional carpets, now in the Metropolitan Museum of Art (New York), Walters Art Gallery (Baltimore), Topkapi (Istanbul), and several private col ections5. This carpet is also in a poor state of conservation and the abrash is located almost exclusively in the lower main border. The third, and largest, carpet is an entirely new addition to the corpus of „Salting‟ carpets and a very important discovery. It was previously in the col ection of Messrs Perez (London) and published by Stanley Reed (1967), but not included in the study of Franses and Mil s (1999), and only finally recognized in the study presented here. It has a central medallion (as opposed to a niche) (figure 7), and an exceptional y high knot count (11.155 knots/dm2), considering that the pile is wool and not silk. This knot density is more than twice that of the famous Ardabil carpet (V&A) and closer to rugs made exclusively of silk, such as the „Kashans‟ [20]. The central field has a dark red ground covered with tchi clouds, a blue medallion in the centre, and green corner medallions. In the main border, lobed medallions alternate with the inscribed cartouches on a beige ground. The inscription appears to be a Persian poem, which describes it as “precious royal carpet… under the feet of such King Suleyman” (see appendix I). Reed has proposed that it was dedicated to the Persian Shah Suleiman (r.1667-1697) [22]. This poem is potential extremely important historical evidence and is currently being considered for greater clarity for this study by Professor Wheeler Thackston (Harvard University) (forthcoming information). While the „Salting‟ prayer or niche rugs are well defined as a group stylistical y, the medallion group remains to be properly described, and few parallels can be found for the field design and inscription of the Guimarãs carpet, in contrast to the prayer rugs. In general terms, the quarter medallions decoration in the field and of the border can be related to four recorded examples in the Hermitage (St. Petersburg), Musée Historique des Tissus (Lyon), and Musée des Arts Décoratifs (Paris)6. It is also in a poor state of conservation, and abrash is visible throughout the carpet: in the central field and main border, as well as al the colours present.

Dyes and Mordants

The colours used in 16th- and 17th-century Persian and Turkish carpets, referred to here as “classical”, were obtained from natural sources, namely plants or insects. It was only in the 19th century, with the discovery of mauve dye by Perkin (1856) and several coal-tar-based dyes, that synthetic dyes were adopted in Iran and Turkey to produce bright and vivid colours [10]. Regional taste also played a factor in colour choices and in the composition of the colour palette. The combination of contrasting primary colours is a general feature of Turkish carpets, while in many classical Persian carpets red dominates the central field and blue or green are used for the border, and in the decoration, secondary colours are juxtaposed beside primary ones, such as orange with blue or green with yel ow [20]. Published analyses of the dyes used in Islamic carpets also reveal a number of interesting trends in the choice of dyestuffs in these two regions, which are useful for studying the „Salting‟ carpets [10-11]. The number of analyses is too small to make firm conclusions, but some general observations can be made. For the reds, the most common dye found in 16th- and 17th-century Turkish carpets was extracted from the roots of the madder plant (Rubiciaea family). It occurs more rarely in classical Persian carpets, there it is exclusively used for the oranges [8-9]. The main chromophores of madder are alizarin and purpurin, both compounds based on 1,2 dihydroxy anthraquinone chromophore, table 1. In addition, dyes prepared from coccid insects were popular [23]; Porphyrophora genera have been found in Persian silk carpets, such as the “small silk Kashans” [8,23]. Other red dye sources of animal origin were used in Iran and India, such as lac-dye (Kerria genera) [3], not found in Turkish carpets, along with cochineal [11]. Lac dye and cochineal are easily distinguished by the presence of laccaic acids and carminic acid chromophores, respectively (table 1). Lac-dye has been identified in a large number of classical Persian carpets [8-9] usual y in the red ground; its presence in the „Saltings‟ is potentially an important indicator for determining their provenance and for this reason is discussed extensively below [1]. The blues in historical carpets from both Iran and Turkey use exclusively indigo dye extracted mainly from the indigo shrub (Indigofera tinctoria L.) or from dyer‟s woad (Isatis tinctoria L.) [12,18]. The main chromophore of indigo is indigotin (figure 8) and its identification in historical rugs is an important indicator of authenticity, as indigo was the most successful and prolific natural blue prior to the invention of synthetic dyes in the 19th century. The yellows in classical Persian carpets were generally obtained from local plants, such as asbarg (Delphinium semibarbatum) which is composed primarily by quercitin, kaempferol and isorhamnetin chromophores. Weld (R. Luteola), rich in luteolin-7-O-glycoside was also found in Persian carpets (figure 9). The use of related plants with luteolin glycoside derivatives [3,11] have been identified in both Persian and Turkish rugs. The luteolin-based compounds are very well documented and considered characteristic of an ideal dye7, and thus commonly identified in high quality textile products [4]. By contrast, the yel ows found in historical Turkish rugs do not reflect the use of asbarg, as it does not occur in Turkey; according to the literature, it is unlikely that this plant was ever imported into Turkey, as more than enough indigenous yel ow dye plants are available for use [11,18]. Curiously, the oranges in Persian carpets do not include lac-dye, but, instead, madder mixed with a yel ow dye. As for Turkish rugs, the results obtained do not offer a precise conclusion; although it is generally agreed that some Turkish plants such as chrysanthemus (Chrysanthemus coronarium L.) or dyer‟s sumac (Cotinus coggygria SCOP.) are used to obtain this colour [18]. Greens in both Turkish and Persian rugs are obtained by using a mixture of indigo with a yel ow dye [11]. In both Iran and Turkey, the browns and blacks are obtained using tannins from local sources, for example, knobby or gallnut oak (Quercus spp.). Beige colours were often obtained with natural wool in both Persian [8-9] and Turkish Carpets [18]. However, in some Persian carpets, analyses of beige areas have shown the presence of the juglone chromophore probably obtained from the walnut tree (Juglans regia L.) (figure 10). This dyestuff does not appear to have been identified in Turkish carpets [11]. It should be mentioned that some of the dye analyses reported in the literature were performed with extraction methods based on strong acids (usually with HCl) [7,24] and, even with Thin Layer Chromatography (TLC) [11]. TLC was used before HPLC-DAD, but does not provide complete results, as it is not possible to identify the precise chromophores and hence the correct dye-source [4]. The HCl method is an aggressive extraction method which can promote the degradation of chromophores in some dyes. In yellow dyes, it can cause decomposition of the glycosidic dye compounds to their parent aglycons, with consequent loss of information about the original dye source. For example there are numerous plants with different luteolin glycoside derivatives such as weld (Reseda luteola L.), dyer‟s greenwood (Genista tinctoria L.), true chamomile (Matricaria chamomil a L.) or three-leafed sage (Salvia Triloba L.) [18]. When these plants are submitted to a HCl extraction, the various luteolin glycosides present are destroyed during the process and usual y only luteolin is detected in the final dye extract. Frequently, the identification of luteolin in dyed yellow fibres is attributed to the presence of weld, one of the most stable yel ow dyes. Nevertheless, when the HCl extraction method is used, it is not possible to exclude the utilization of other related plants such as the ones mentioned previously. Therefore, mild extraction procedures such as those used in this work, and sucessfully reported in literature, are necessary to obtain greater information about the dye source [4,7].

Mordants

In order to attach these natural dyes to the textile fibre to ensure they are fixed permanently and will not bleed, a metal ic ion (mordant) was usually applied in both Persian and Turkish carpets (figure 11). The mordant acts as an intermediary between the fibre and the dyestuff, and is usually obtained from a mineral source (salts in the form of crystals, efflorescence or crusts, mud, metal oxides). The choice of mordant can determine both the wash-resistance of a colour and its intensity [12,18]. The presence of aluminium, iron and copper as mordents has been observed in both classical Persian and Turkish carpets for different colours [9,18]. For the yellows and reds usually an alum mordant was used, while in the dark browns the presence of an iron mordant has been detected [9,18]. For some beige colours, no mordant was detected at all. Also in blue fibres dyed with indigo, no mordant is applied, as indigo is a vat dye which precipitates in the fibres through an oxidation- reduction reaction [4,8].

Lac Dye Library

The Indian and South Asian origin of lac-dye would appear to support a Persian (as opposed to Turkish) provenance for the „Salting‟ carpets. However, no study to date has used HPLC-DAD to differentiate lac-insect species to identify locations of dye production. Indeed, of the lac species of Asian origin belonging to the Kerria genera (ca. 26 species) [15,25], only Kerria lacca Kerr has been characterized using HPLC-DAD [6,12,26]. In this study, for the first time, 82 lac-insect historical samples of unknown species Kerria genera and 2 samples from Paratachardina genera, with diverse provenance, as wel as from different host plants, were analyzed using HPLC-DAD. This data base methodology is applied here for the first time to lac-dye represents a first step towards developing necessary research on lac-dye species, to reveal greater information about the precise source of the red dye and therefore assist with provenance identification of historical textiles.

Lac Insects

Lac-dye was highly appreciated in Asia and is documented in India as a textile dye, from at least the 4th century BC [12]. Lac-dye was also known and used by the Romans for dyeing purposes since the 2nd century AD, which suggests that lac was probably one of the first insect dyes to be used in West Asia [26]. The dyestuff is obtained from female scale insects of the Kerridae family (figure 12) which contains nine genera and approximately 100 described species. Lac-dye can be obtained only from Kerria and Paratachardina genera [15]. The Kerria genera comprises 26 species [15], which can be found mainly in India, China, Taiwan, Sri Lanka, Australia, and Pakistan, and the Kerria lacca Kerr species is the most important (Appendix II). There are several publications about the Kerria species [5,15,28]; however, none of them have used molecular taxonomic studies for confirmative species identification. According to Penny Gul an9 it is essential that fresh specimens are collected from Asia and molecular taxonomic studies are performed to correctly identify the lac insect species as has occurred, for example, for the Paratachardina species [29-30]. The Paratachardina genera comprises a total of nine species, which have been recently revised, mainly from China, India, Sri Lanka, Philippines and Papua New Guinea [29]. However, species belonging to Paratachardina genera do not produce lac-dye of commercial importance and mainly Kerria spp. are exploited [15]. Nevertheless, as the Paratachardina genera is the only one for which accurate taxonomic studies have been published, it was included in this study and specimens were analysed. Several entomologists supplied five Kerria spp and two Paratachardina spp correctly-identified insect-specimens for HPLC-DAD analysis. As the taxonomy of the Kerria genera is not well established (Appendix II), it was not possible to obtain reliable insect-specimens for all of the 26 Kerria species described in the literature. Hence, in this study, as a first approach to studying the Kerria insects, 76 unknown historical lac-dye insect sources from different geographical regions (India, Australia, Taiwan, Vietnam, Laos, Singapore, Pakistan, Bangladesh Sri-Lanka and others), as wel as from different host plants (Ficus spp., Schleichera spp., Butea spp., Zizyphus spp. and Shorea sp.) (figure 13), were characterized by HPLC-DAD and submitted to PCA analysis. These 76 historical sources of lac-dye, dating from the 19th century and belonging to the Royal Botanic Garden at Kew, provide a good starting point for establishing similarities between the sources and for identifying insect groups according to their provenance and host plants. In future research, these insect samples should be submitted to molecular taxonomic studies and compared with the 26 Kerria species correctly- identified by entomologists.

Lac-Dye

The process of production of lac dye involves thousands of lac insects attach themselves to twigs and the fleshy young leaves of suitable host plants (figure 13), from Moraceae, Sapindaceae or Dipterocarpaceae families, usually in November [31-32]. By March, each insect covers itself with a predator protection layer, exhibiting the appearance of a red oval-shaped, smoothly polished lifeless sack entirely filled with a beautiful red liquid [33]. The coated branches of the host plants are cut and harvested as sticklac, which contains both resinous matter and the red organic dye composed of laccaic acids. In order to obtain the pure deep red colouring matter suitable for dyeing purposes, known as lac-dye, it is necessary to perform a simple water extraction process. In extractions with 100% water, the principal chromophores of lac-dye, namely laccaic acid A, B, C and E, are obtained in the final water extract (table 2). On the other hand, ethanolic extractions of the sticklac product wil remove only the resinous matter which is composed mainly of yellow compounds (table 2). Lac comes onto the market as a raw product, occurring in many forms; among the most common are the already mentioned sticklac, grain/seed lac (material removed from the twigs), and a subsequent variety which is fused and cast into moulds, caked lac or lac dye (table 2) [33]. The pale perforated kind, in which no insect exists, constitutes the material for shel ac employed for varnish- making.

Experimental

A detailed characterization of the materials present in the three Guimarães carpets was performed using different methods. The fibres were identified by optical microscopy in longitudinal section; the colours were characterized using colorimetric measurements; all dyes were identified by High- Performance Liquid Chromatography with Diode Array Detector (HPLC-DAD) and when necessary with Mass Spectrometry (LC-MS); all mordants were analysed by Inductively Coupled Plasma with Atomic Emission Spectrometry (ICP-AES); and the metal threads were characterized by Energy Dispersive X-ray Fluorescence (µED-XRF) and Scanning Electron Microscopy (SEM-EDX). C14-AMS analysis was performed in one sample from the medallion carpet. Prior to HPLC-DAD identification of the red dyes, several historical lac-dye sources were analyzed by HPLC-DAD and submitted to a PCA analysis. Also several parameters in the samples preparation were tested in order to obtain HPLC chromatograms with good resolution. For more details about experimental proceedings see Appendix III.

Results and Discussion

The two Guimarães prayer carpets, Benguiat (PD77) and Duff (PD78), share features commonly identified in the „Salting‟ group, namely the presence of a bright colour palette, silk foundation, wool pile, metal-wrapped silk threads, arabesques designs, religious inscriptions and small dimensions. The Medallion carpet, by contrast, is distinguished by the absence of religious features, in both its composition which lacks a prayer niche and inscriptions which are not excerpts from the Koran but probably a secular poem. Analyses undertaken thus far on the „Salting‟ group have focused mainly on the prayer type, and hence, the study undertaken here of the Medal ion carpet offers new and important data for appreciating this under-studied type. With the aim of tacking the questions posed at the start of this study, concerning the date and provenance of the carpets, the results of the diverse analytical techniques utilized here (described above) are integrated and presented below, according to the standard method of preparing the materials for constructing a carpet: beginning with the fibres, followed by the colours and dyes (with special emphasis given here to lac-dye), and ending with the metal ic threads, which were usual y purchased from an external source outside the sphere of the carpet-making workshop. As C14 offers exclusively evidence for dating, it is dealt with at the end.

Metal Threads

Microscopic analyses indicated that the metal threads used in the three Guimarães carpets comprise a metal strip wound (S-direction) around a yellow silk core, so that the edges of the tape abut but do not overlap. This allows the colour of the substrate – in this case yellow silk -to play a greater role in the final visual effect, enhancing the golden sheen of the thread (figures 1 and 2, appendix VIII). This technique gives a smooth, as opposed to a rough, irregular surface to the thread [38]. Measurements for the metal strip for the Benguiat Rug (c. 0.23 mm wide and 0.015 mm thick) are in accordance with previous results reported for „Salting‟ carpets (ca. 0.25 mm wide) [38].

XRF and SEM analyses demonstrate that the metal lamina of the Guimarães threads is composed 

mainly of silver (Ag), gold (Au) and copper (Cu) (figures 2 and 3, appendix VIII). In the prayer rugs, Ag was major element (circa 80%), followed by Au (circa 10%), Cu (circa 2%), with vestigial amounts of tin (Sn) and lead (Pb). Previous results for the „Saltings‟, however, identified the presence of Pb, but not Sn [38]. The SEM analyses also show considerable degradation of the exterior surface, particularly in places where sulphur (S) and chlorine (Cl) were identified and probably the result of Ag corrosion products (figures 5 and 6, appendix VIII). By contrast, the inside of the strip, which is in direct contact with the silk core, and therefore protected from contact with environmental conditions, and is better preserved. Analyses here revealed the presence of Au in higher amounts, in contrast to the exterior where it presumably had been worn away. It is difficult to establish from these analyses with certainty how the strips were made, but it is possible to identify the presence of a silver strip with a layer of gold. This is consistent with the method previously proposed by Eiland for the „Saltings‟ and „Polonaise‟ carpets, in which a silver sheet is covered with a thin layer of mercury (Hg) to create an amalgam for gilding, and then forged to a very thin gauge and hand-cut [38]. However, confirming the use of this technique requires further research.

Concluions

The publication of the „Salting‟ carpets in 1999 represented an important step in presenting the first corpus of identified carpets. However, doubts remained about their precise date and provenance. The rediscovery of two prayer carpets in Guimarães, and a third Medallion carpet which was not included in the corpus in 1999, offered a significant opportunity for returning to these questions using an interdisciplinary approach and the application of a wide range of analytical techniques. The dyes identified in the three Guimarães carpets point consistently to an Iranian provenance. Use of lac-dye in the reds and pinks, but madder (and not lac) in the oranges, is consistent with other results for carpets associated with Iran in Portuguese collections. Together, these studies reinforce the notion that this palette is a classical characteristic of Persian (and not Turkish) dyeing practices, especially as the Medal ion Carpet has an inscription written in Persian. Through the development of the HPLC-DAD database for lac-dye, it was possible to distinguish Kerria and Paratachadina genera, for the first time, and to confirm that only Kerria has sufficient red-dye chromophores to be used for dyeing textiles. It was possible to discriminate the samples from the Kew Garden col ection, according to their composition and eventually their origin. The red textile samples from the Guimarães rugs appear to be more similar to insect samples from Pakistan, in close proximity to Iran. Further development of the database is required to establish more rigorous comparisons between historical textiles and lac-insect sources. This can only be achieved with the col aboration of entomologists, and the identification of reliable Kerria insect samples which have been submitted previously to molecular taxonomic studies, and for which their precise origin and host plants are wel documented. For the browns and yellows in the carpets, tannins and probably weld, respectively, were found. As these are usual y obtained from local sources, further research could offer additional information for narrowing the provenance of the carpets. In this respect, it is noteworthy that the brown used in the Medal ion Carpet is comprised of ellagic acid, not encountered in the Guimarães prayer rugs or other „Saltings‟ analysed [11], but previously identified in the „small silk Kashan‟ in the Museu Nacional de Machado Castro [20]. This latter carpet can be dated to the second half of the 16th century and represents the height of Safavid carpet production. The presence of this compound draws attention to the importance of further study of the browns in Islamic carpets, something which has rarely been considered in the past, owing to the apparent insignificance of this colour to the overall visual effect. No significant difference, however, was observed between the three carpets in relation to the mordants used, which it was hoped would be a distinguishing feature of provenance. The abrash observed reflects differences in dye procedures, and the concentration of alum present, and not degradation for light exposure. As for the chronology of the carpets, al the data presented here supports an early date, such as the exclusive use of natural dyes and the non-industrial construction techniques used to make the metal threads, which have also been identified in other classical Persian carpets. Moreover, as “a carpet can only be as old as its design” [10], al of the stylistic features are also consistent with Safavid art. Although these few elements are significant, they are not necessarily conclusive, of course, and only the AMS Radiocarbon results for Medal ion Carpet confirms a more precise date between the late 15th and mid-17th century. This result is extremely important, as it confirms that it is an authentic Safavid rug, as expected from its extraordinarily high quality (11.155 knots/dm2), rarely seen in wool carpets. However, Reed‟s proposal that the carpet was intended to be placed “under the feet” of the Persian Shah Suleiman I (r. 1666-1694) is not directly confirmed, as the dates of this Shah are slightly outside the parameters of the AMS Radiocarbon results. Instead, a date for the carpet in the 16th century seems more likely, which raises the possibility that it could have been made especial y for the Ottoman Sultan Suleyman the Magnificent (r. 1520-1566), and possibly sent in 1556, as original y proposed by Mills and Franses [1]. This proposal, if it can be supported by additional scientific or historical evidence, would make this carpet extremely important and only the second example, besides the Ardebil (V&A), to offer a historical name and related date.

==

Bibliography

1. Mil s, J., and M. Franses. “Salting Carpets.” Oriental Carpets and Textile Studies. Danville, Califórnia, (1999). 2. Canby, S. R. Shah ´Abbas, The remaking of Iran . London: The British Museum, (2009). 3. Walker, D. “ exhibition catalogue.” Flowers underfoot – Indian Carpets of the Mughal Era. New York: The Metropolitan Museum of Art., (1997). 4. Ferreira, E., et al. “The natural constituents of historical textile dyes.” The Royal Society of Chemistry, 33, (2004), 329-336. 5. Wouters, J. “High Performance Liquid Chromatography of Antraquinones: Analysis of Plant and Insect Extracts and Dyed Textiles .” Studies in Conservation, vol. 30, No. 3, August (1985), 119-128. 6. Wouters, J. and Verhecken A. “The scale insect (Homoptera: Coccoidea). Species recognition by HPLC and diode-array analysis of the dyestuffs.” Annls Soc. ent Fr. (N.S.), 25 (4), (1989), 393 - 410. 7. Zhang, X., Laursen, R. A. “Development of Mild Extraction Methods for the Analysis of Natural Dyes in Textiles of Historical Interest Using LC-Diode Array Detector-MS.” Analytical Chemistry, 77 (7) , (2005), 2022 - 2025. 8. Valssassina, M., et al. “The Colour of Carpets.” In The Oriental Carpet in Portugal: Carpets and Paintings, 15th-18th centuries, exhibition catalogue, from J. Hallett and T. Pacheco. (eds.) Pereira. Lisbon: Museu Nacional de Arte Antiga., (2007). 9. Armindo, E., et al. “A Persian Carpet‟s Paradise Garden: Discovering Historical and Technical Aspects Through Carpet Conservation and Restoration.” ICOM-CC 15th Triennal conference New Delhi. (2008), 960-966. 10. Böhmer, H. and Enez, N. “A Topkapi Prayer Rug: Arguments for a Persian Origin.” HALI, 41, (1988), 11-13. 11. Enez, N. “Dye Research on the Prayer Rugs of Topkapi Col ection.” Oriental Carpets and Textile Studies. California: Berkeley, (1993), 191-204. 12. Cardon, D. Natural dyes. London: Archetype publications, (2007). 13. Wouters, J.and Verhecken A. “Potential Taxonomic Ppplications of HPLC Analyses of Coccoidea Pigments (Homoptera: Sternorhyncha).” Belg. J. Zool., 121, (1991), 211 - 225. 14. Sousa, M. M., et al. “Flavylium Chromophores as Species Markers for Dragon's Blood Resins from Dracaena and Daemonorops Trees.” Journal of chromatography A (volume 1209), (2008), 153-61. 15. Sharma, K. K., et al. “Role of lac culture in biodiversity conservation: issues at stake and conservation strategy.” Current Science, 91 (7), 10 de October (2006), 894-898. 16. Halpine, S. M. “An Improve Dye and Lake Pigment Analysis Method for High-Performance Liquid Chromatography and Diode-Array Detector.” Studies in Conservation, vol. 41, No. 2, (1996), 76-94. 17. Murray, Jr. Eiland. “The Saltings Reconsidered.” Oriental Carpets and Textile Studies, V, 2nd part. Danvil e, Califórnia, (1999). 18. Böhmer, H. Koekboya Natural Dyes and Textiles. Germany, (2002). 19. Frade, C., P. Cruz, E. Lopes, M. Sousa, J. Hal ett, Santos, R. and A. Aguiar-Ricardo. “Cleaning Classical Persian Carpets with Silk and Precious Metal Thread: Conservation and Ethical Considerations.” Lisbon: (accepted) to ICOM-CC 16th Triennial Meeting, (2011). 20. Hal ett, J. and Pereira, T. P. (eds.). The Oriental Carpet in Portugal: Carpets and Paintings, 15th- 18th centuries. Lisbon: Museu Nacional de Arte Antiga, (2007). 21. Wearden, J. Oriental Carpets and Their Structure Highlights from the V&A Col ection. London: V&A Publications, (2003). 22. Stanley, R. Oriental Rugs and Carpets, Pleasures and Treasures. London: George Weidenfeld and Nicolson Ltd, (1967). 23. Serrano, A. “Cochineal, A Precious Source of Red: Cochineal Dyes Identification by High Performance Liquid Chromatography with Diode Array Detection and Principal Component Analysis Master Dissertation, (2010). 24. Valianou, L., et al. “Comparison of Extraction Methods for the Analysis of Natural Dyes in Historical Textiles by High-Performance Liquid Chromatography.” Anal. Bioanal. Chem., 395 (7), (2009), 2175 - 2189. 25. United States Department of Agriculture - Systematic Entomology Laboratory. http://www.sel.barc.usda.gov/catalogs/kerri da/KerriaAl .htm (accessed in April, 2010). 26. Hisao, O., et al. “Separation of lac dye components by high-speed counter-current chromatography,.” Journal Chromatography A (813), (1998), 71-77. 27. Enez, N. and Böhmer, H. “Ottoman Textiles: Dye analysis, results and interpretation.” Dyes in History and Archeology 14. Amesterdam, (1996), 39-44. 28. Karapanagiotis, I. and Chryssoulakis, Y. “Investigation of red natural dyes used in historical objects by HPLC-DAD-MS.” Annali di Chimica, 95, (2005). 29. Kondo, T. and Gullan, P. “Taxonomic review of the lac insect genus Paratachardina Balachowsky (Hemiptera: Coccoidea: Kerri dae), with a revised key to genera of Kerri dae and description of two new species.” ZOOTAXA, 17 de October (2007), 1-41. 30. Schroer, S. et al. “The genetic diversity, relationships, and potential for biological control of the lobate lac scale, Paratachardina pseudolobata Kondo & Gul an (Hemiptera: Coccoidea: Kerri dea).” Elsevier, Biological Control, 46, (2008), 256-266. 31. Hofenk de Graaff, J. H. The Colourful Past – Origins, Chemistry and Identification of Natural Dyestuffs. Switzerland: Abbeg-Stiftung, (2004). 32. Phipps, E. Cochineal red: The Art History of a Color, New York: Metropolitan Museum of Art, (2010). 33. The Natural organic colouring matters, Monographs on industrial chemistry. Edited by Sir Edward Thorpe. 34. Szostek, B., J. Orska-Gawrys, I. Surowiec, e M. Trojanowicz. “Investigation of natural dyes occurring in historical Coptic textiles by high-perfomance liquid chromatography with UV-VIs and mass spectrometric detection.” Journal of Chromatography A, 1012, (2003), 179-192. 35. Valssassina, M. “Work in progress.” comunica ão pessoal. Caparica, (2007). 36. Balfour-Paul, J. Indigo. London: Archetype Publications, (2006). 37. Karapanagiotis, I., Minopoulou, E., Valianou, L., Daniilia, S., Chryssoulakis, Y. “Investigation of the colourants used in icons of the Cretan School of iconography.” Analytica Chimica Acta 647, (2009), 231-342. 38. Eiland III, M. L., “Analysis of Metal Threads: New Light on the Salting Problem”. Oriental Carpets and Textile Studies. Danvil e : International Conference on Oriental Carpets, (1999), Vol. V. 39. Tímar-Balázsy, A. and Eastop, D. “Metal threads.”Chemical Principles of Textile Conservation, de A. and Eastop, D. Tímar-Balázsy, 128-138. Butterworth Heinemann, (2004). 40. Zhang, X., Boytner, R., Cabrera, J., Laursen, R., “Identification of yel ow dye types in pre- colombian Andean textiles”, Analytical Chemistry, 79 (4) (2007), 1575 - 1582. 41. Marques, R., Sousa, M., Oliveira, M., Melo, M., “Characterization of weld (Reseda Luteola L.) and spurge flax (Daphne gnidium L.) by high performance liquid chromatography-diode array detection- mass spectrometry in Arraiolos historical textiles” Journal of Chromatography A, 1216 (2009) 1395 - 1402. 42. Naes, T., Isaksson, T., Fearn, T., Davies, T. Multivariate Calibration and Classification. Chichester

NIR Publications.

43. Preisner, O. E., Lopes, J. A., Guiomar, R., Machado, J., Menezes, J. C., “Uncertainty assessment in FT-IR spectroscopy based bacteria classification models”, Chemom. Intell. Lab. Syst. 94 (1) (2008) 33 – 42. 44. Stuiver, M. e Reimer, P.J. Extended 14C Data Base and Revised CALIB 3.0 14C Age Calibration. Radiocarbon 35 (1), (1993), 215-230.