Ashmolean Eastern Art Online, Yousef Jameel Centre for Islamic and Asian Art

As we have seen above, Pfister depended primarily on stylistic comparisons to establish a chronology for the textiles. He chose designs with which he could make visual links to Gujarati Islamic architecture dated from the late fourteenth to the sixteenth century, but also included motifs that he believed to be Hindu in origin, and to have an earlier twelfth- to thirteenth-century source. This particular sequence of chronology has to be abandoned [1]. He excluded any textiles that he considered to be later than approximately 1600 CE. In fact the Newberry Collection provides ample evidence, if it were needed, that there is no specific cut-off date after which Indian textiles were no longer traded to Egypt. However, if one focuses on the type of textile discussed by Pfister, recent opinion among scholars had associated them more specifically with the Mamluk period in Egypt. As long as stylistic comparisons were the primary evidence, the dated architecture of Gujarat and Jain manuscripts remained the major comparative sources. The mosque motifs from Ahmedabad already drawn on by Pfister are generally of the fifteenth and sixteenth centuries, while evidence from Cambay, in particular from the Jami Masjid, is from the fourteenth century. The majority of manuscript paintings are of a fifteenth-century date. At the same time there was also an increase in maritime trade with a shift to the Red Sea, and therefore it seemed reasonable to think of the bulk trade in utilitarian, printed textiles as being particularly associated with these centuries. With recent archaeological evidence and the results of radiocarbon dating on some of the Newberry textiles now available, the question of dates of production and dissemination can be looked at with far more reliable results.

 Radiocarbon dating

Ten textiles were first selected in the summer of 1994 to be analysed by the Radiocarbon Accelerator Unit in the Research Laboratory for Archaeology and the History of Art, Oxford University [2]. A second, smaller group of 6 textiles was chosen in the spring of 1995 and was also tested at the Oxford laboratory; for reasons to be explained below, one of these textiles did not come from the Newberry Collection. The results of the first selection are shown in Table 1 [3]. The radiocarbon testing is a measure of the C-14 (carbon 14) in any sample. Carbon 14 is a radioactive isotope, and therefore decays with time. The older a sample is, the less carbon 14 is in it. There is a margin of error involved, which is called the standard deviation and is expressed by the +/- term. The larger the deviation span shown the less precise the measurement, the smaller the span the more precise it is. For the last 300 years it is not possible to come to closer resolutions. It is primarily for that reason that some of the sample results have a wider margin of error than others, but other factors may also add to the degree of precision or span of deviation.

Table 1. Textiles dated in 1994

Cat. no.          Radiocarbon date

EA1990.143     1335 CE +/- 50

EA1990.153     1525 CE +/- 50

EA1990.247     1010 CE +/- 55

EA1990.305     1255 CE +/- 50

EA1990.480     1690 CE +/- 90

EA1990.712     1595 CE +/- 50

EA1990.804     1460 CE +/- 70

EA1990.1033   1535 CE +/- 50

EA1990.1085   1555 CE +/- 55

EA1990.1123   1450 CE +/- 50

To have to choose 10 fragments for carbon dating from a collection of more than 1,200 is a daunting taks. How can one ensure that those selected actually have a chance of bringing meaningful results, considering the problems that arise with later material? For the first group samples I decided to be at least partly guided by the Indian textiles found at Quesir al-Qadim, in order to have a relatively secure frame of reference. For that reason, [EA1990.143, EA1990.153, EA1990.480, EA1990.1033, and EA1990.1123] were chosen, as they correspond closely to Quesir finds [4].

Quseir al-Qadim as comparative evidence

According to Vogelsang-Eastwood (1990: 15) these textiles are dated to between the mid- thirteenth and the end of the fourteenth century. However, Donald Whitcomb, who was the archaeologist responsible for the Islamic strata of the excavations, has recently further analysed his finds and has come to conclusions that allow for a more narrow time range for the Indian textiles (Whitcomb 1993) [5]. Two areas were excavated in 1980 and 1982, respectively: the so-called Sheikh's house in 1982, and an area to the east of that in the earlier season. The first, preliminary session in 1978 had defined the approximate extent of the Islamic occupation. Coin foinds at the two specific sites (1980, 1982) showed clearly diverging dates. The Sheikh's house produced 15 coins, 5 of pre-1200 date and 10 dated from 1200 to 1250, with no finds beyond 1250. The so-called eastern area produced 30 coins, with the majority of these (21) dated 1350 or later [6]. Whitcomb therefore suggests an Ayyubid occupation around the Sheikh's house which ceased in the mid-thirteenth century, and a Mamluk occupation in the eastern region of the Islamic area at Quesir, of a mid- to late fourteenth- and early fifteenth-century date.

It is interesting that most of the textile finds were made at the latter site, and are therefore associated with the Mamluk occupation. There is a discrepancy in the number of textile fragments reported by Whitcomb and Vogelsang-Eastwood; in his paper Whitcomb mentions 53 Indian textiles, 7 of them from the Sheikh’s house and 46 from the eastern area. Vogelsang-Eastwood, however, gives a total of 69 fragments, which breaks down as follows:

1978   1

1980   46

1982   22

One of the textiles from the 1980 excavation (and included in the number) is of silk with a tie-dye (bandhani) design. As Whitcomb’s reassessment of his data was still in draft form in 1993, it seems likely that his record was less complete than Vogelsang-Eastwood’s. She gives full photographs and accession numbers for the fragments, so that the information one can gain from her publication is considerably richer. The evidence from her report confirms Whitcomb’s comment that more Indian textiles were present at the Mamluk site than at the 1982 excavation at the Sheikh’s house, but it also shows that the fabrics were more common at the Ayyubid location than Whitcomb reports.

It is, in my opinion, not possible to distinguish the fragments from each other according to the two areas of excavation. There is no distinctly different patterning, although the colours of the textiles show some variation of distribution, as shown in Table 2. In other words, the 1980 finds from the eastern area, and presumably of a later date, produced 28 blue fragments, generally with traces of indigotin, and 15 fragments that were dyed any shade of red, achieved with the use of mordant [7]. The textiles reported from 1982, on the other hand, were predominantly mordant-dyed, i.e. a shade of red. Only 9 out of 22 fragments had been dyed blue. Apart from the single fragment from the 1978 preliminary excavation, only 3 textiles showed an application of both blue and red colorants. The total number found, though, is too small to come to any conclusions about preferred colours or designs.

Table 2. Distribution of Quesir al-Qadim textiles according to colour

Year     Site date          Red          Blue         Red and blue

1978    1200-1400          -              -                  1

1980    1350-                15            28                 3

1982    1200-1250          13            9                  -

How do the Quseir fragments compare to the C-14 dates? Three of the samples [EA1990.143, EA1990.153, and EA1990.1033] were chosen because they were close to Quseir finds from the 1980 excavation, presumably of late fourteenth- to early fifteenth-century date [EA1990.143, EA1990.153, and EA1990.1026]. Our analysis gave the dates of 1335 CE, 1525 CE, and 1535 CE, all +/-50. In other words, the medium values of the three fragments are on either side of the chronology of the eastern area, although 135 is close enough to be considered contemporary with Mamluk Quseir. The medium value of [EA1990.153] is approximately 100 years later than the supposed abandonment of Quseir as an international harbour. However, it is surprisingly close in design to the Quseir 18 cat. no. (Ill. 5), down to minute details of band arrangement. Even a small dot added to the continuous vine is found in both textiles. It is of course possible for a specific combination of decorative elements to be in use over some time, and we have evidence for that with our material for other designs as well. But in this case the two textiles are so specifically close that my inclination is to doubt that there are 100 years between their production. From the preliminary excavation reports for the harbour, it is not quite clear to me how complete the abandonment of the area was after the harbour ceased to be used for the Indian trade. Is it possible that some of the textile finds may actually have a later date?

By contrast the third date of 1535 CE +/-50, which also is later than the date proposed by Whitcomb for the Islamic occupation of Quseir, is less of a problem for the comparison to Quseir fragments [EA1990.1026]. Although the leaves and small berries appearing in the border have parallels in Quseir (Vogelsang-Eastwood 1990: cat. nos. 32, 36), there is sufficient difference to allow for a lapse in time. The centre field of the Newberry textile has a pattern that is related to the Quseir cat. no. 1, of a red and blue dyed fragment, so far not published with an excavation location. This particular design can appear in many variations, from lavishly ornate arrangements to highly stylized and abstract versions, as is discussed in Chapter 7. In one form or another it continued to be in use into the nineteenth century.

However, Quseir cat. no. 1 and Newberry [EA1990.1123] are another instance of complete overlap of designs (Ill. 6, Pl. 10 [EA1990.1116]). Once again, the similarity is so close that the fragments could have come from the same piece of cloth. The Newberry textile is dated to 1450 CE +/- 50, which still fits into Whitcomb’s dating of the eastern area, especially if one is willing to accept the earlier possibility for the carbon-dated textile.

Slightly surprising was the late date of 1690 CE +/-90 given for Newberry [EA1990.480]. It was chosen as a comparison to the Quseir fragment cat. no. 56, which is from the Sheikh’s house and therefore is likely to have a thirteenth-century date. Both textiles are mordant-dyed with a stepped diamond design, and they seemed sufficiently similar to expect a closer date. However, it may be another instance of continuity over a considerable time span, in this case possibly as long as four to five centuries [8].

[EA1990.1085] was also chosen to be analysed because it was exceptionally close to a fragment from an archaeological find, although one without a datable context. A single tab found at the Nubian site of Qasr Ibrim is so close to the border tabs on two large textiles in the Newberry Collection that the Qasr Ibrim fragment visually completely integrates with the Newberry piece if it is placed on top of it [EA1990.1073]. The result of the carbon dating gives 1555 CE +/-55 for the Newberry textile. So far no chronology has been given to the Qasr Ibrim finds, as much of the site is extremely difficult to date. A total of 71 printed textiles have been found at Qasr Ibrim, virtually all of them in very poor condition.

Dating particular designs

Of the four remaining samples analysed in this first group, three were decided on for rather more subjective reasons. Newberry [EA1990.804] was chosen because it shows a type of design that also relates to textiles traded to South-East Asia. The trumpet-shaped and paired leaves, combined with two versions of quatrefoils, are common motifs on cloths that have survived from Sulawesi, where they were taken by traders, presumably in the late seventeenth or eighteenth century. Quite often the cloths also include small geese encircling rosettes ([EA1995.61] Ill. 7). My own response to these textiles was that their patterns were possibly fully developed by the fifteenth century, as their individual motifs relate closely to details found in Jain manuscripts of the fourteenth and fifteenth centuries. This speculative opinion seemed worth while investigating. The carbon-dated fragment had a result of 1460 CE +/-70, which confirmed the hypothesis; it also provided an impetus to speculate further about the actual age of some of the textiles traded to Indonesia, with a surprising result, as will be seen below.

[EA1990.305] is one of several fragments with a design that is visually particularly interesting. I think it is very likely to be related to a motif that appears often on dress painted in Jain manuscripts, as seen in the lower garment worn by the reclining figure of [EA1987.33]. It may occasionally also have affected designs used in a strictly Islamic context. For example, it could have been the source for a curious spiral design found in the ceiling paintings of an early sixteenth-century mosque in Rada', Yemen, a connection that I have taken up and developed elsewhere (Barnes forthcoming) (Ill. 8). As the textiles that survive from Egypt were also traded to the Yemen, it is distinctly possible that their designs had an impact on local decorative arts. The carbon-dating result of 1255 CE +/-55 shows that the design was established by the thirteenth century. The early date was a surprise, as the dimensions of the fabric are unusually large. Most of our textile remains are fragments that would fit on to an A4 size sheet, but this one has a dimension of 78 cm. x 53 cm.

[EA1990.712] has the pan-Asian motif of linked circles containing a rosette (or lotus), found as a surface ornament in any number of media and techniques. This particular example was patterned with a stamped and hand-painted mordant, rather than stamped resist, as the reverse shows considerably less dye saturation than the surface.9 From Quseir al-Qadim we apparently have no textiles that were patterned by printing the mordant, rather than a resist, and, as will be discussed in Chapter 6, which deals with mordant- and resist-dyeing, it is possible that mordant-printing (rather than resist-printing followed by mordant application or immersion) was a later development. For that reason alone it was potentially of interest to choose this sample. In addition, the space between the lotus and the circle outline is densely filled with white tendrils, comparable to filler tendrils found in north-west Indian textiles exported to South-East Asia, presumably in the seventeenth to eighteenth centuries. Once again, the result of 1595 CE +/-50 fits both a post-Quseir date and the time for which we know of a thriving trade with textiles to South-East Asia that show at least parts of the same pattern, in particular the densely placed filler tendrils.

[EA1990.247] was chosen for two reasons. Along one side of the large fragment are the remains of an inscription which reads ‘... to its owner...' (Ill. 9), clearly meant to be ‘eternal blessings to its owner’, as one can see from similar sayings on more complete textiles (Pfister 1938: 69) [10]. When the inscription was read, my colleague James Allan suggested a Fatimid date for it, although this seemed initially unlikely, as being too early. Instead I thought it more likely to be yet a further example of the continuity of a formalized design. However, when I began to look more closely at the history of Jain manuscript painting, as it is so clearly relevant to the printed textiles, there seemed to be a curious stylistic similarity between the painted wooden book covers of the early manuscripts and this type of textile, of which several examples survive in various collections [11].

The early Jain books, painted on palm leaf and with narrow wooden covers, are dated to the eleventh century, with some chronologically on either side. Their covers usually show very thick, fleshy vines, with lotus blossoms in profile, and occasionally with lively animals—elephants, monkeys, geese—set between the tendrils (Ill. 10). In particular the generally fleshy and organic appearance, but also the big lotus blossom in profile, provided a point in common with the textile design. I remain sceptical about using stylistic comparisons as a tool for dating the printed textiles, as continuity was and is the norm rather than the exception. However, where additional evidence also points towards a particular period of production, the art historian’s ‘gut reaction’ deserves to be put to the test. Both epigraphic and stylistic comparisons proved valid, as the result of 1010 CE+/-55 has shown. At least one of the Indian textiles surviving from Egypt was of an earlier date than so far accepted.

The second group of C-14 dates

Following these results, in the spring of 1995 I chose a second set of fragments for testing. The results are as shown in Table 3. The first five textiles are illustrated in [EA1990.136, EA1990.320, EA1990.361, EA1990.404, and EA1990.950]. Their dates mostly confirm the presence of the material in Mamluk times. [EA1990.361] is very similar to architectural designs found in both Jain and Muslim buildings of Gujarat (Ill. 11), while the trees of [EA1990.950] could have come from a Jain manuscript. In addition, the results for [EA1990.320] and acc. no. EA1995.61 deserve to be emphasized, as they expand on the first set of samples tested.

Table 3. Textiles dated in 1995.

Cat. No.         Radiocarbon date

EA1990.136    1400 CE +/-40

EA1990.320    1060 CE +/- 40

EA1990.361    1420 CE +/- 40

EA1990.404    1600 CE +/- 40

EA1990.950    1410 CE +/- 40

EA1995.61      1400 CE +/- 40

The date for EA1995.61 came as a surprise. It is not an Indo-Egyptian fragment, but a large textile traded from India to South-East Asia and collected recently on Sulawesi, where cloths of this type have long been in use as ceremonial fabrics. Its design clearly relates to [EA1990.804], with a fifteenth-century radiocarbon date. Up to now it has always been assumed that the earliest Indian textiles surviving from Indonesia were not likely to be older than the late seventeenth or eighteenth century [12]. The date of 1400 CE given to this cloth calls for a reassessment of the surviving evidence for the South-East Asian trade. In no way is it in contradiction to the historical documentation, however. We know that when Portuguese traders in the early sixteenth century first observed the merchandise moving throughout the Indian Ocean, the importance of textiles from Gujarat in South-East Asia was immediately noticed.

A textile identified in historical documents?

From the first group analysed, the result of 1010 CE +/-55 for [EA1990.247] was by far the most surprising. As [EA1990.321] has a design that is very close to it, this was chosen for the second set of samples. The result gave an eleventh-century date for this fragment as well. To have a similar date confirmed, for a textile related in design, is extremely important. I am in no doubt that both textiles are Indian in origin. An eleventh-century identification predates any so far accepted chronology for the Indo-Egyptian cotton fragments. It is not in conflict, however, with historical documents.

Goitein’s analysis of the Cairo Genizah papers refers to Indian cotton imports, although his historical analysis has so far not been used extensively by scholars interested in the Indian textile connection with Egypt, as his sources seemed to predate the surviving textiles. This view can now be reconsidered, and Fatimid Cairo may become directly relevant to our interpretation. Prior to receiving the results of the C-14 analysis, I had noticed one type of textile in particular in Goitein’s discussion: a type of cloth called mihbas (pl. mahabis) [13]. In the volume on daily life in A Mediterranean Society he says:

A textile constantly mentioned as a main export from India to the West, and especially from the north- Indian province of Nahrawara-Gujerat, requires some attention. It was named mihbas (more commonly occurring in the plural mahabis), which might be translated as wrapper. It was traded in scores, consisting of twenty complete or half thawbs, a word usually denoting the robe worn by everyone, male or female, but also the piece of material needed for a robe. The mahabis were also used for the manufacture of pillows; red and black are mentioned as their colors, and their average price, approximately 1 Adenese, or 1/3 Egyptian dinar, makes it more than likely that they were made of cotton, the most important of Indian textiles. (1983: 171)[14]

Although not specifically referred to as being of cotton (i.e. Arabic qutn), he thought it likely that these import cloths from north-western India actually were of cotton fibre. This can hardly be doubted, as cotton fabric was and is the major export textile from Gujarat. The only alternative, silk, was too expensive to be referred to here. These textiles, however, although not expensive and certainly not of luxurious quality, were in common use. This is an exceptionally important reference, as it may describe Indian printed textiles.

To add to the evidence produced by Goitein, mahabis are also mentioned by Serjeant with regard to Aden, where they are referred to in connection with the harbour tax imposed on import goods from India (1972: 130 n. 64). Goitein relates mahabis to Arabic habasa, ‘to wrap’. It is revealing that in addition to their use as dress items, he has also found references to the cloth as used for the making of pillows. This completely fits with the surviving textile evidence. I also think that Goitein’s description of the cloth, as being red and black, is not in contradiction with our material. There generally is not a clear distinction made between black and blue, especially when a deeply dyed indigo cloth is referred to.

The Fustat-C textile

Finally, another possible piece of evidence actually comes from the archaeological work at Fustat itself. Although the overall site is chronologically difficult, the last season in 1980 at so-called Fustat-C produced apparently reliable data, as it was a rare undisturbed area, or so it was claimed at the time by the excavation team led by George Scanlon and Wladyslaw Kubiak (Scanlon and Kubiak 1989). As mentioned in the previous chapter, one blue and white textile fragment was found at Fustat-C that could have been Indian. Its design is not directly similar to [EA1990.247], but the density of weave, the thread quality, and intensity of blue are close to the Newberry fragment [15].

The textile find has been dated to the late eleventh century or earlier, but this evidence has so far been treated with some caution. Although Louise Mackie in her discussion of the fragment suggests that it could be Indian, up to now this would have been the only surviving piece, and the single fragment together with an unusually early date raised more questions and doubts than it confirmed possible answers. The radiocarbon results for the Newberry fragments, however, should make us alert to the textile finds of Fustat-C, and specifically to the one example of resist-printed fabric.

It may also bring us to take another look at a cotton fragment in the Textile Museum in Washington (acc. no. 6. 115). This is a resist-printed textile, dyed blue and with a design of small arches very similar to the design of the Fustat-C fragment, as well as an inscription. It has been published by Kühnel, and he has given it a tenth-century date on stylistic grounds; the inscription he identified as early kufi, corresponding closely to Samarra pottery ware (Kühnel and Bellinger 1952: 99-100). His interpretation was never taken up and developed, and it may indeed be off the mark. However, the similarity to the Fustat-C textile is striking.

With the radiocarbon analysis now pushing back the date for the earliest surviving Indo-Egyptian textiles to the eleventh century, the Fustat-C evidence is presented in a completely new light. Although by no means conclusive until a similar analysis can be done on it, our result certainly adds weight to the excavation report. It would be most welcome if, after all the doubts cast on the chronology of Fustat finds, it were possible to confirm textile data from this site, as it was once the hub of Asian trade with the Mediterranean world, a trade in which the textile industry played a major economic part.

Summary of evidence

The results and comparisons discussed here provide evidence that may be small if considered in isolation, but which, if pieced together, sketches a picture that gains considerably in depth. To conclude, we now have some reliable dates for the textile trade that make it possible to move beyond a stylistic comparison, and which also allow for a critical reassessment of some of the recent archaeological finds. Regarding the latter, the immediately most challenging comparison is with the single fragment found at Fustat-C; the results for [EA1990.247] and [EA1990.320] support the initial archaeological report. Of course one cannot paint a broad picture on the evidence of three fragments. But it is hoped that they provide the basis for further work.

The radiocarbon results may also be useful for a full, final report on Quseir al-Qadim, as they provide comparative evidence. In general there is no contradiction in the two groups of material, i.e. the Quseir al-Qadim textiles and the C-14 analysed Newberry fragments. There are, however, some questions that remain open from the preliminary reports for the site, regarding in particular the last period of occupation, and these have been highlighted, in my opinion, by the results of the C-14 analysis.

In general the ‘hard’ evidence, both from the scientific analysis and the archaeological records, supports rather than contradicts the opinion previously reached by stylistic comparisons. It is quite clear now that the textile trade between India and Egypt was very active during Mamluk times; this confirms the historical and art historical evidence. The doubts that John Irwin cast on the early date of the printed fragments are to some degree removed (Irwin and Hall 1971: 4-13). His emphasis on the importance of continuity, though, is certainly valid [16]. In addition, however, we can be virtually certain that the Mamluk trade only continued an exchange in goods that went back to Fatimid times. While there were historical documents to this effect, in particular in the Genizah papers, we now have evidence for the material surviving from as early as the eleventh century. Against this chronological background, let us now turn to take a close look at the evidence surviving in the collection itself. The actual material composition of the fabric will be first considered, before turning to the techniques used to create the designs. Then the designs and their arrangements are discussed before the material is placed into a wider frame of reference.