• Textiles and Society

  • Patterns and Weaves: Safavid Lampas and Velvet

  • Textiles and Society
    Carol Bier

    The role and nature of textiles in any society are both pervasive and diverse, for textiles serve everyone in many ways. Textiles are created to serve the daily and ceremonial needs of nearly all individuals, literally from birth to death. In all categories of the world’s population, individuals require textiles, regardless of age, sex, status, belief, or occupation. Yet textiles by their design and function also serve to distinguish among individuals and among groups of individuals, in terms of class, religion, activity, gender, stature, and respect. The basis of fashion, textiles distinguish by means of drapery, form , and cut, by color and texture, by the fibers and how they have been prepared, by processing and finishing of the fabric. Combining aesthetics and technology, the textile arts represent nearly all human activity and express much that is valued in any given society. A major component of material culture, textiles are intended to serve defined purposes. In functional terms, they may be viewed as the products of technology, as cultural artifacts, as works of art, and as objects of trade.

    All textiles, including those that today survive only as fragments, derive from particular places at particular times. Each textile or fragment represents a complex set of human interactions between user and viewer, which implies a social response, and between user and maker, which sometimes results from an economic transaction or transactions among buyer and seller, lender, merchant, and trader. Suppliers of raw materials and market superintendents or commisioners for fair trade and quality control may also have been involved. The quality and apperarance of any given textile results from the contemporaneous conjunction of such factors as desire, taste, knowledge, technology, social traditions, aesthetic preference, style, fashion, political and economic circumstances, market conditions, organization of the household, availability of raw materials, and ecology of surroundings. In short, every textile ever produced from oversized carpets to the tiniest scrap, is the product of its own environment, made and used in response to a particular set of historical circumstances.

    In Iran from the sixteenth through the nineteenth centuries, textiles and the textile arts were prominent in everyday life, in religious ceremony, and at the court. They brought beauty to humble dwellings and to the sparse encampments of migratory pastoralists as well as to the most splendid and sumptuous of palaces. Textiles produced at home by the womenfolk, or in workshops by professional weavers, including those produced both by nomadic and by sedentary peoples, reflect a progression of cultural values during this four-hundred-year period. Maintaining traditional means of expression, the textile arts of Iran from the Safavid through the Qajar periods express adaptation to changing social and economic circumstances that affected aspects of their production, use, and trade. Garments, carpets , and textiles used for interior furnishings and cermonial use produced during this period, demonstarate new concepts of pattern and design, stylistic preferences, and responses to foreign presence and influence.

    One may argue that international trade, increased commercial relations, and burgeoning of regular diplomatic exchange in the siteenth through nineteenth centuries in part both resulted from and effected results in changing roles and relationships among textiles and the people who made them, and those for whom they were made.

    Analysis of the kinds of textiles surviving from the Safavid through Qajar periods, including both those produced in Iran and those imported for use in Iran, provides the basis for provisional systems of classification for the study of textiles that may have broader application. Pertinent functional categories include garments, textiles produced for ceremonial use, fabrics for interior furnishings, containers for storage and transport, and materials for shelter. The study of textile arts from this point of view leads to a better understanding and appreciation of man’s adaptation to his environment. Management of the land by means of animal husbandry and pastoral pursuits, or by cultivation and agricultural pursuits (or a combination of these) yields the raw materials from which nearly all historic textiles are made (wool, cotton, linen, and silk). Insulation from the environment has always been provided by means of shelter and clothing. Tents of the nomads almost always have incorporated some sort of felt or woven fabric for protection from sun, rain, wind, and cold. Felt is produced by a process of compacting fibers rather than weaving them and remains today a chief means of insulation throughout the world. Beyond the purely functional realms of providing subsistence and protection, textiles were also used to ornament man’s surroundings with interior furnishings and fashion, as well as to define and visually distinguish social status.

    Such a functional classification of textiles is external. That is, it can only be arrived at by external considerations involving interpretive decisions on the part of the analyst as to the function of a particular object or group of objects. Another external system of claissification is one based upon attributions. This relies upon interpretive decisions for individual textiles and groups of textiles as to date and place of manufacture, based on comparison with other textiles and by making certain assumptions about links to historical evidence.

    Internal systems of classification, on the other hand, are derived from information inherent in the objects themselves, and may be used in combination with external ones. Internal evidence consists of such information as the type of fiber or filament, their physical characteristics such as spin and ply of the yarn, their arrangement, processing, and finishing. Such information is descriptive, obtained analytically, but does not require the kind of interpretation needed for external systems of classification. For the purposes of this study of Safavid and Qajar textile arts, several broad catergories of textiles have been grouped according to their internal weave structure (e.g., plain weave, brocaded taffeta, double cloth, damask, satin lampas, velvet).1 The term “metal ground” signifies in this study a heterogeneous group of textiles defined stylistically rather than technically, within which different weave structures are present but not all of which are yet thoroughly understood. Although this system of classification is neither consistent nor fully standardized, it has enabled us to form coherent groups sufficient for further comparison and analysis. Other internal systems of classification may be based, for example, upon design and pattern (see below, Sonday, and Bier).

    There are an extraordinary number of sources of information that may be used for the interpretation of textile arts from the Safavid through Qajar periods. Apart from surviving carpets, garments, and textiles, which are often difficult to localize and date, there are illustrations of carpets, garments, and textiles used for interior furnishings and other purposes in arts of the book and in contemporary wall paintings.2 There are also important categories of written materials to be found within court records, European travelers’ accounts, treaties, and trading company archives, which provide direct or peripheral informaion about the textile arts. For the sixteenth and seventeenth centries, court chronicles, royal decrees (farman), even priests’ diaries are of occasional use. European traverlers to Iran in the seventeenth and eighteenth centuries sometimes commented upon the crafts, or remarked upon textiles and carpets available for purchase in the city markets.3 In the eighteenth century there are several eyewitness accounts of the Afghan invasion and its effects of devastatiion on the Safavid capital at Isfahan. Later accounts discuss the decline of crafts, in particular those related to the manufacture of textiles.4 For the nineteenth century the range of sources increases. With the expansion of diplomatic missions and commercial interests, the diplomatic correspondence and consular reports offer exceptional insight into economic conditions with reference to the production of textiles.5 And company archives are a source only recently being explored.6 From the middle of the nineteenth century there are an increasingly known number of photographs of Iran, taken by Europeans and by Iranians, which are available today by way of publication or preserved in several archives.7 For the Qajar period the wealth of oil painting on canvas provides a rich source for the documentation of textile patterns and fashion, supplemented to a lesser extent by representations on a smaller scale in enamel and lacquer. Ethnographic documentation carried out in the 1960s and 1970s among traditional pastoralists offers important information about people and textiles, information often assumed to serve as a model for the past. 8

    Three kinds of problems confront anyone attempting to carry out research in the field of textile arts from the Safavid through Qajar periods. The first kind of difficulty reflects the status of our sources from which to derive information about textiles of these periods. The sheer quantity is vast, but not all sources are presently accessible or otherwise able to be sufficiently utilized. Hundreds, if not thousands, of Persian textiles, garments, and carpets are preserved in private and public collections, as well as many still presumably with the families for whom they were originally intended. Many of the known examples remain unpublished, and nearly all are without inscription or other form of internal documentation to signify historical context. There are few secure links between extant textiles and primary written sources, whether historical, literary, social, religious, or economic. Archives and documents relating to the social and economic history of Iran have not for the most part been previously tapped with specific reference to textile production or patterns of trade. Information from studies in anthropology and literary history and from studies of urban development have not yet been applied to the study of textiles. Primary written sources in Russian and Armenian are also extensive and have not been thoroughly explored. Persian sources are also no doubt of much greater value than has been presently recognized. Documentary sources on the receiving end of textile arts exported from Iran to eastern and western Europe will also surely yield additional information in the future. Marriage contracts and dowry lists, legal, tax and customs records have not been approached from this perspective. Further potential sources of information are no doubt yet to be identified.

    A second kind of difficulty is confronted in the attempt to relate available written historical documentation to the body of extant textiles. This problem is apparent even in the precise identification and delimitation of the most generic names for textiles in the written sources, such as chit, karbas, dibaj, and atlas. Here, ethnographic documentation has offered potential links, but only securely for present usage. Just how far back one may project such usage beyond living memory remains in the realm of speculation. Language usage and the specificity of meaning is surely regional, following the flow of the time, as does artistic style.

    A third kind of problem encountered in research is in the analysis of the textiles themselves. Consistency is desired in analysis and classification, yet this is hard to achieve. This situation results in part from the extraordinary diversity of woven materials in general. But it is expecially difficult for the documentation and description of the compound wovern structures that reached such high levels of complexity in the Safavid through Qajar periods. Terminology appropriate for categorizing coherent groups of textiles to use as a basis for comparative research has not yet been sucessfully achieved.9 Careful scrutiny of technical analyses and documentation may offer needed information to be able to determine patterns of development within the textile arts of Iran and neighboring lands. Working hypotheses about workshop and trade practices must then be tested against both written and material evidence.

    The directions and opportunities for future research in this field are plentiful. The important role of the textile industries within the economic history of Iran deserves closer attention and more precise definition.10 Access to local and foreign sources of materials for the production of textiles (silk, wool, cotton, linen, and dyestuffs) in different priods, and fluctuations in their availability under differing political and economic circumstances, must have had direct impact on local textile production, but this has never really been studied.

    The process of drawloom weaving and its technology is not thoroughly understood or documented for these periods. Treatises have yet to be located for fuller documentation of the use of dyes and for the processes of commercial weaving, as well as for the manufacture or embellishment of textiles by other means (e.g., embroidery, applique, crotcheting, knitting, felt-making, accessorization). Written sources should be thoroughly combed for citations or mention of innovation, improvements, or other modificatiions in the technology or its application that may have had an effect upon textile design and patterning.

    Other areas of research not yet adequately explored concern craft organization from the Safavid through Qajar periods.11 The relationship between weaving and the metaphysical concepts and philosophy of the guilds remains particularly opaque. Difficulties of interpretation in this area at any but the most general level are extreme, but certain assumptions lead us to expect a direct link between philosophical principles and their application to the practice of weaving.

    The whole question of rug production requires extensive review and investigation. The rubrics in Oriental carpet literature for “Indo-Persian”, “Herat,” “Indo-Ispahan,”12 reveal the difficulties inherent in identifying a secure place of manufacture for groups of carpets related in style but differing in structural characteristics that probably reflect differences in place and processes of production. The sources of inspiration for design and layout of the pattern, the processes of dyeing, and instigation of production by commission or commercial interests, all resulted from a matrix of cultural and historical circumstances present when these carpets were produced. Yet our understanding of this matrix and the resulting artifacts, made for a diversity of markets from east to west, remains in embryonic form.

    The series of essays presented here, and the catalogue of textile arts that follows, represents an initial effort to bring together recent research from different disciplinary perspectives drawing upon both written and material sources. The information as it is presented will hopefully be refined and revised in the near and distant future, but it may serve in the meantime as a foundation upon which to build or to reconstruct a better understanding of textiles and society in Iran from the Safavid through the Qajar periods.

    1For definitions of testile terms as they are used in the chapters and catalogue of this book, see Glossary.

    2The subject of Safavid painting has received considerable attention in the past fifteen years. Recent reference works include Martin B. Dickson and Stuart Cary Welch, The Houghton Shahnameh 2 vols. (Cambridge: Fogg Art Museum, Harvard University Press, 1981); B.W. Robinson, “Persian and pre-Mughal Indian Painting,” in Islamic Painting and the Arts of the Book, ed. B,W, Robinson, (London: Faber and Faber, 1976); idem, Persian Paintings in the India Office Library: a Descriptive Catalogue( London: Sotheby Parke Benet, 1976) idem, Persian Paintings in the John Rylands Library (London: Sotheby Parke Bernet, 1980); Norah M. Titley, Persian Painting and Its Influence on the Art of Turkey and India: The British Library Collection (Austin: University of Texas Press, 1983); Anthony Welch, “Painting and Patronage under Shah ‘Abbas I”, Iranian Studies 7 (1974): 458-507; Stuart Cary Welch, Wonders of the Age: Masterpieces of Early Safavid Painting, 1501-1576 (Cambridge: Fogg Art Museum, Harvard University Press, 1979); idem, Persian Painting: Five Royal Safavid Manuscripts of the Sixteenth Century (New York: George Braziller, 1976). For Savafid wall-painting, see Ernst Grube, “Wall Paintings in the Seventeenth Century Monuments of Isfahan,” Iranian Studies 7 (1974): 511-42.

    3.Of European travelers’ accounts, those most useful for detailed descriptions of the textile industries and their products are the works of Sir John (Jean) Chardin, Les Voyages du Chevalier Chardin, ed. L. Langles, 10 vols. (Paris, 1811); Jean Baptiste Tavernier, Les six voyages, 3 vols. (Paris); Sir Thomas Herbert, Travels in Persia, 1627-1629 ed. and abrid. W. Foster (New York, 1929.

    4See Laurence Lockhart, The Fall of the Safavi Dynasty and the Afghan Occupation of Persia (Cambridge University Press, 1958) and idem, Persian Cities (London, 1969.)

    5 Abbas Amanat,ed., Cities and Trade: Consul Abbott on the Economy and Society of Iran, 1847-1866 (London: Ithaca Press, 1983). See also Charles Issawi, ed., The Economic History of Iran, 1800-1914 (Chicago: University of Chicago Press, 1971).

    6Annette Ittig, “The Kirmani Boom: A Study in Carpet Entrepreneurship,” Oriental Carpet and Textiles Studies I (1985): 111-23. For earlier discussion of the industry and its development, see A. Cecil Edwards, The Persian Carpet: A Survey of the Carpet-Weaving Industry of Persia (London: Duckworth, 1953).

    7See Angelo M. Piemontese, “The Photograph Album of the Italian Diplomatic Mission to Persia, Summer 1862,” East and West 22 n.s. 3-4 (1972): 249-90; Jennifer Scarce, Isfahan in Camera: 19th Century Persia through the Photographs of Ernst Hoeltzer, Art and Archaeology Research Papers. A special issue (London, 1976); Donna Stein, “Early Photography in Iran,” History of Photography (October-December 1983), pp. 257-91; Iraj Afshar, “Some Remarks on the Early History of Photography in Iran,” in Qajar Iran: Political, Social an Cultural Change, 1800-1925, ed. Edmund Bosworth and Carole Hillenbrand (Edinburgh: Edinburgh University Press, 1983), pp. 261-90.

    8See Allen Zagarell, Prehistory of the Northeast Bahtiyari Mountains, Iran: The Rise of a Highland Way of Life, Beihefte zum Tubinger Atlas des vorderen Orient, B. 42 (1982), for discussion of ethnography and the interpretation of archaeological assemblages.

    9The early attempt of Nancy A. Reath and Eleanor B. Sachs, Persian Textiles and their Technique from the Sixth to the Eighteenth Centuries Including a System for General Textile Classification (New Haven: Pennsylvania Museum of Art, 1937), remains a standard for Safavid textiles. Irene Emery, The Primary Structures of Fabrics: An Illustrated Classification (Washing, D.C.: The Textile Museum, 1980), is of more general application. Ann Pollard Rowe, “After Emery: Further Considerations of Fabric Classification and Terminology,” The Textile Musuem Journal 23 (1984): 53-71, extends Emery’s system to descriptions of compound weaves. For other terminologies, see Dorothy K. Burnham, Warp and Weft: A Textile Terminology (Toronto: Royal Ontario Museum, 1980).

    10John Emerson, “Some European Sources on the Economic Structure of Persia between 1630 and 1690” (Ph.D. diss., Cambridge University, 1971) and Linda K. Steinmann, “Shah Abbas I and the Royal Silk Trade” (Ph.D. diss, New York University, 1986) each approcach this for understanding the economy.

    11These issues are addressed in Willem Floor, “The Guilds of Qajar Persia” (Ph.D. diss., Leiden University, 1971), and idem, “The Traditional Crafts and Modern Industry of Qajar Iran, “ (forthcoming). See also Mehdi Keyvani, Artisans and Guild Life in the Later Safavid Period: Contributions to the Social-Economic History of Persia (Berlin: Klaus Schwarz Verlag, 1982).

    12Ellen S. Smart and Daniel S. Walker, Pride of Princes: Indian Art of the Mughal Era in the Cincinnati Art Museum (Cincinnati: Cincinnati Art Museum, 1985), pp. 84-88; Wilhelm von Bode and Ernst Kuhnel, Antique Rugs from the Near East, transl. Charles Grant Ellis, 4th ed. (Ithaca: Cornell University Press, 1984), pp. 118-24; Donald King and David Sylvester, The Eastern Carpet in the Western World from the 15th to the 17th Century (London: Arts Council of Great Britain, 1983)), pp. 96-98; Murray L. Eiland, Chinese and Exotic Rugs (Boston: New York Graphic Society, 1979), pp. 145-57; Friedrich Spuhler, Islamic Carpets and Textiles in the Keir Collection (London: Faber and Faber, 1978), pp. 90-94; K. Brisch, “Indischer spiralranken Teppich,” Berliner Museen, Berichte aus den preussischen Kunstammlungen 3(1975); May H. Beattie, The Thyssen-Bornemisza Collection of Oriental Rugs (Castagnola: Villa Favorita, 1972), pp. 39-65; Kurt Erdmann, Oriental Carpets, trans. Charles Grant Ellis (New York: Universe Books, 1960), pp. 41-45,57. See also Charles Grant Ellis, “The Portuguese Carpets of Guirat,” in Islamic Art in The Metropolitan Museum of Art, ed. Richard Ettinghausen (New York: The Metropolitan Museum of Art, 1972), pp. 267-89.

    Pattern and Weaves: Safavid Lampas and Velvet
    Milton Sonday

    The highest achievements of Safavid silk designers and weavers are preserved in a limited number of figured velvet and lampas1 weaves. Aside from their sumptuous surfaces and monumental patterns, they are ideal for the insights they provide for understanding technology and aesthetics. Each is an easily identified compound weave in an advanced degree of complexity; velvet is the use of supplementary warps in a foundation weave, and lampas a particular combination of two weaves-one weave the foundation and the other a supplementary. The more complex a weave structure, the more features it has for comparative analysis. The production of each of these weaves occurs over sufficiently wide coordinates of time and space that Safavid velvet and lampas weaves can be seen in an international context. In addition, the motifs created for these weaves are uniquely Safavid.

    Other figured weaves are not included in this brief essay because an insufficient number survive, or their technical features are too limited to provide meaningful data for comparative analysis. For example, only two damasks have been published as Safavid, although a few others are so attributed (cat. no. 8).2 While a fair number of brocaded3 plain weaves (cat. nos. 12, 56), brocaded satin weaves and float-patterned satin weaves4 survive, they are technically relatively straightforward. Double cloths, including those which are brocaded as well as those with areas in which warps and wefts are integrated (cat. nos. 25,26), have few technical details which help survive their types.

    A large number of fabrics are grouped together on the basis of design, such as floral patterns on a gold or silver background. This group presents special problems and can only be mentioned briefly in this essay. Some fabrics in this large group fall into a structural category that has an ancient history in the Islamic world: plain weave with complemantary wefts and inner warps5 (e.g. cat. nos. 48,62, 63). The majority, however, although very similar in appearance, have technical features that suggest they belong in another catergory. All in this second group can be said to have been woven according to a common structural concept: the use of supplementary wefts (e.g. cat. nos. 19, 21, 54).6 There are many interesting variations in technigue in this so-called metal background group. They await further study before they can be placed securely in a historic or geographic frame of reference.

    There is considerable disagreement as to which brocaded plain weaves, brocaded satin weaves, float-patterned satin weaves, double cloths, and the various types of the metal background group are Safavid and which are Mughal. By comparison there is little disagreement as to the Safavid origin of the velvet and lampas weaves included in this catalogue. I will try to identify those features of design and structure which make a group of velvet and lampas weaves uniquely Safavid. It is wise to suppose that each weave is likely to have its own history and development.


    Patterns can be composed in two ways: either as continuous patterns, with elements of their compositions arranged to be repeated endlessly in all directions or as pictorial patterns, with elements of their compositions arranged to be contained within and related to a top, bottom, and sides. All too often what survives of a historic fabric is a fragment that is too small to give an impression of what the original looked like. Fragments must, therefore, if possible, be placed within the context of either continuous or pictorial compositions. Frequently one can reconstruct the composition of a continous pattern from one or more fragments, provided one understands the principle upon which such patterns were laid out.

    A continous pattern can theoretically extend endlessly in all directions. Its success is judged by assessing the use of formal elements of composition with an understanding that the pattern is indeed continuous. Nothing should hold the eye in one place too long and thus prevent it from scanning the entire pattern.

    Historically almost all continuous patterns ere precisely laid out in a repeat format. In order to analyze the layout of the repeats of a continuous pattern, one must first derermine the unit of the repeat. Repeat units are designed with no boundaries so that elements of the composition interact when adjoined edge to edge. Only when repeat units re repeated can the rhythms and balances of the larger continuous pattern be perceived. As a result, repeat units of the best patterns are hardly noticed. In this lies the special skill and sucess of the designer of continuous patterns.

    The delineation of a repeat unit is determined by the production technique employed; in this essay it is weaving. A repeat unit for weaving must be squared off so that its sides are parallel with warps and its top and bottom parallel with wefts (see fig. 1a-f). Motifs of all woven continuous patterns must be placed within a squared unit, with the understanding that when repeated by a loom, it becomes part of the larger continuous pattern. This squared unit is the technical repeat unit.

    Some technical repeat units can be subdivided into smaller areas, each of which has within its squared-off boundaries all the elements of the pattern. These are minimum pattern areas. They can be laid out free of the restrictions that a loom’s mechanism places on the technical repeat unit.

    A loom is able to mechanically repeat a weave structure and a pattern in only two ways. Technical repeat units must be designed with this in mind. The two weaving repeat systems are straight repeat and pointed repeat. A straight repeat is one in which a technical unit is repeated straight across and straight up and down, with no alteration in direction. A point repeat is one in which a technical unit is reversed to form mirror images along vertical and / or horizontal axes. These two systems apply to all types of looms.

    Woven repeats are controlled on the loom by first threading warps in the appropriate harness,7 either straight or pointed, and second, by opening sheds or cords in the appropriate harness. The width of a repeat unit is limited to the number of warps available to control one technical repeat unit.8 The length of a technical repeat unit has no such restrictions.

    Even though continuous patterns were designed in the abstract for endless coverage in all directions, in actuality they were produced in fabric as lengths. If properly designed and produced, lengths can be joined along their vertical edges to extend pattern coverage. Selvedges, therefore, are an important factor in the planning of a design and the reconstruction of a continuous pattern.

    Case by case discussion of several Safavid patterns illustrates the basic principles of woven repeat systems, including the importance of selvedges.

    The correct visual orientation of the velvet panel “Women in a Landscape” (cat. no, 2) is horizontal, which means that the warps of the velvet are at right angles to the standing figures. At first glance it appears to be a panel with well-defined pattern boundaries. However, the composition is only partially pictorial. A closer study of this and other extant panels reveals that the four figures are but one unit of a straight repeat. For example, in cat. no. 2, a small section of the edge of the jar held in the outstretched hand of the woman on the far left can be seen again on the right edge of the velvet. One technical repeat unit fills almost the entire width of the velvet. The technical repeat unit is 216 cm long and 72 cm wide, with the full selvedge-to-selvedge width measuring 74.5 cm.9 Lengths of this velvet were not intended to be joined selvedge to selvedge. The extraordinary length of the repeat, that the technical repeat unit fills the width of the velvet and is at right angles to the warp, the subtle use of color, and extensive warp-pile substitution (explained in the section on velvet) make this one of the world’s highest achievements in weaving.

    The velvet fragment showing a hunting scene (cat. no. 34), almost encompasses an entire technical repeat unit of a straight repeat. This is clearly indicated by the bits of motifs that are cut off at one edge and appear at a corresponding edge. The repeat measures 39.5 cm x 22 cm. Since none of the three known fragments of this pattern has a selvedge, the continuous pattern cannot be reconstructed as a selvedge-to-selvedge width. The reconstruction has been left with sides ragged, as determined by the narrative arrangement of motifs. In figure 2, all identical motifs are positioned directly above and across from each other, which is characteristic of a straight repeat. The technical repeat unit, if squared off, would have within it everything needed to repeat the pattern. In this case the technical repeat unit and the minimum pattern area are the same.

    See cat. nos. 1 and 37 for other fabrics with straight repeats in which the technical repeat unit and the minimum pattern area are the same.

    In the next three examples the technical repeat unit has within it two minimum pattern areas, one a mirror image of the other, one on top of the other. Figure 3a shows all the elements of the composition of the “Dragon Slayer” (cat. nos. 33, 60, 61) arranged in a pictorial manner. Figure 3b shows the same elements squared off as a rectangle measuring 32 cm x 25 cm, the sides determined by a selvedge preserved in one of the fragments. This is the minimum pattern area. Delineation of top and bottom is arbitrary, but such boundaries are best coordinated with motifs and their placement. Since a repeat unit has within it all the elements of a composition arranged so the edges can meet and continue the pattern, cutting a mofif in half has no fundamental effect on the following analysis.

    The squared-off minimum pattern area of figure 3 b can be repeated only if another, as in figure 3c, is placed in mirror image directly above it. Now twice as high, 64.5 cm, this is the technical repeat unit. This is the portion of the continuous pattern that was programmed into the pattern section of the loom. Only as a straight repeat could the pattern be woven with offset motifs facing in opposite diretions in alternate rows. The technical repeat unit has within it two minimum pattern areas, one on top of the other, one a mirror image of the other as in figure 1b.

    The reconstruction of this continuous pattern as a selvedge-to-selvedge width has three technical repeat units across the width (fig. 3d). This weft dimension corresponds to the widths of other velvets, both intact and similarly reconstructed. Measurements and reconstructions indicate that a Safavid velvet might have been about 72 cm wide, plus or minus 2.5 cm or more.

    When mirror images of the minimum pattern area are placed one above the other, the axis on which they are turned is an imaginary one running through their centers. This imaginary axis is a useful guide in reconstructing such pattern layouts, because a selvedge is likely to be either on the edge of the minimum pattern area or on the imaginary center line. These are the only vertical lines that are possible in such a layout of minimum pattern areas.

    The continuous pattern of the velvet with hunt scenes (cat. nos. 30,31,32), can now be read as a straight repeat with two minimum pattern areas, one mirror image placed above the other (fig. 4). The technical repeat unit measures 116 cm x 67.5 cm and fills the entire width of the velvet. The pattern was carefully planned so that lengths, when properly joined, continue the pattern laterally.

    The narrower a repeat, the more difficult it is to determine the width of a fabric. However, if we accept the proposed widths of lampas and velvet weaves as fairly standard for the Safavid peropd, reasonably accurate reconstructions are possible. The number of repeat units within a width varies from fabric to fabric as demonstrated in the reconstructions for this catalogue.

    There are enough surviving fragments, some with selvedges, of the lampas weave showing a courtier leading a captive (cat. no. 59), for it to be reconstructed as a full width (fig. 5). The technical unit of the straight repeat measures 92 cm x 36 cm and has two minimum pattern areas within it, one a mirror image of the other, one above the other. Two technical repaeat units make up the width. One repeat would have been too narrow and three too wide for the Safavid period. When compared with surviving full widths and other reconstructed continuous patterns, a measurement of 72 cm seems reasonable and reinforces the observation that a Safavid lampas weave is on the average, 68 cm wide.

    Other fabrics with straight repeats of technical units having mirror images of minimum pattern areas placed one above the other are cat. nos. 22 and 27.

    In the next example, the technical repeat unit has within it one minimum pattern area, which is offset (fig. 6a). In spite of the fact that all surviving fragments of this version of “Khusraw sees Shirin Bathing” (cat. no. 29) are extremely worn, a full width can be reconstructed, because a selvedge is preserved in one fragment.10 The width of the minimum pattern area is 17 cm, and four across provide a fabric of 68.5 cm, an appropriate width for the Safavid peroid. The minimum pattern area shown in figure 6b is 112.5 cm long. The technical repeat unit is as long as the minimum pattern area and twice as wide, this dimension allowing for the offsetting of two minimum areas (see fig. 1c). The technical repeat unit was woven as a straight repeat, and there are two such units with the width of the fabric.

    Four fragments of another version of “Krusraw sees Shirin Bathing” with the same repeat system, are known in velvet. The pole warps of this velvet were cut and they cover the entire surface (cat. no. 28; Cleveland Museum of Art 44.499a,b; The Metropolitan Museum of Art 1978.60). There are no cypress trees in this version of the pattern, which gives it a different visual character. Reconstruction of a full width is not possible, since none of the fragments has any indication of selvedge (fig. 7). Sides have been left ragged as determined by the narrative arrangment of motifs.11

    Mechanically woven mirror images, or point repeats, have axes of symmetry that fall on the edges of the technical repeat units. The technical units of such repeats are planned knowing the pattern will not be complete without mirror images. The area created by the full number of reversals of the technical repeat is called the maximum pattern area. It is identified in figure 1d-f by the heavy outline. Axes of symmetry are a dominant visual feature of such repeats, although they may not always be mechanical axes of symmetry. Vertical axes, those parallel with the warp, ideally fall on both selvedges so that lengths of fabric can be joined to extend the width of the pattern.

    The lampas weave (cat. no. 36), has vertical and horizontal axes of mechanical symmetry, or point repeats. These axes create a grid throughout the design. One section squared off by the grid is the technical repeat unit, here a rectangle 12.2 cm x 7 cm. In this example the minimum pattern area and the technical repeat unit are the same. One vertical and one horizontal turn place four technical repeat units in four different positions. These four units as a group make a maximum pattern area twice as high and twice as wide as a single technical repeat unit (see fig. 1f). An amusing weaving mistake can be seen on one horizontal axis. The horizontal axis of the mirror image of the pattern occurs twice, as if the weavers went out to lunch and, upon resuming weaving, inadvertently reversed the pattern direction. Quickly realizing their mistake, they again reversed the pattern, this time to the proper direction. Repeats with vertical horizontal axes of symmetry do not appear to have been very common Safavid lampas and velvet weaves.

    Two velvet fragments in The Textile Museum (3.336) have vertical mechanical axes of symmetry only. Each technical repeat unit has two minimum pattern areas within it, one a mirror image of the other , one above the other (see fig. 1e). Its minimum pattern area, shown in figure 8a, measures 30.5 cm x 17 cm. The technical repeat unit is, of course, twice as high. Four technical repeat units make up the fabric’s width of 68.5 cm (fig. 8b). The “Falconer” 9cat. no. 10) is a fragment of a velvet with this pattern layout and a vertically symmetrical repeat system. It would have been similar to the length of velvet with standing women in the Royal Ontario Museum (962.60.1) and another in the Keir Collection (no. 107).

    The pattern of the velvet (cat. no. 9) has a strong ogival lattice superimposed on a thin vine bearing bold flowers (fig. 9a). The ogival lattice has the same flower between the points of its ogees. This ogival lattice with blossoms is vertically symmetrical. Two other blossoms on a thin vine are positioned within each ogee. One is on the axes of symmetry of the ogival lattice, the other at right angles to them. That this pattern was woven as a straight repeat is indicated by the fact that the horizontal blossoms are not vertically symmetrical and face left in one row and right in another. The minimum pattern area is shown in figure 9b. There are two technical repeat units within the 68.5 cm reconstructed width, each measuring 71 cm x 34.5 cm.

    Analysis of this pattern shows that the technical repeat unit is actually made of two superimposed minimum pattern areas, one for the ogival lattice and another for the thin floral vine. The ogival lattice can be reduced to a minimum pattern area 17 cm wide, the sides defined by vertical axes of symmetry (fig. 9c, top). There are two minimum pattern areas, one a mirror image of the other, one above the other, with the imaginary central axis running through them. The floral vine can be reduced to a minimum pattern area of 34.5 cm wide (fig. 9c, bottom). The two minimum pattern areas were placed one above the other in the position each occupies in the continuous pattern. The vine’s minimum pattern area is twice as wide as that of the ogival lattice. While the smaller minimum pattern area of the ogival lattice is aligned on a selvedge, the wider floral vine’s minimum pattern area is not.

    Separating these two minimum pattern areas and placing them in the positions they occupy in the continuous pattern shows how their imaginary central axes (or center lines) and edges are aligned (fig. 9c). The center lines and the lines falling on the edges of the two minimum pattern areas are equidistant and create vertical sections 8.5 wide. By shifting the floral vine’s minimum pattern area one section to the right, its right edge is aligned with the right edge of the ogival lattice’s minimum pattern area. Both areas can be now superimposed to form the true minimum pattern area. This area can therefore be divided into four vertical sections 8.5 cm wide and the 68.5 cm-wide fabric into eight sections, numbered from right to left 1,2,3,4,1,2,3,4 (fig.9b,c).

    The velvet discussed earlier (fig. 8a, b), which was woven with a vertically pointed repeat system, also has an ogival lattice superimposed on a thin vine. However, both can be contained within the same minimum pattern area.

    Patterns with a superimposition of lattices and vines were popular in Islamic lands in the sixteenth century as well as in the Christian Meditteranean. Charles Grant Ellis isolated five levels in a sixteenth-century Safavid carpet in the Musee des Gobelins.12 Carpet designs are not restricted by a mechanical repeat system. This seems to be the most complex pattern in Safavid art. The inventive layout of the two minimum pattern areas of the velvet, which was designed within a technical repeat system (cat. no. 9), reinforces Ellis’ suggestion that Safavid multilevel designs are unequaled.

    The system of pattern analysis applied in this essay to Safavid patterns increases appreciation of woven continuous patterns and opens new areas for research. The precision needed for the reconstruction and analysis of the last velvet pattern discussed indicates the importance of grids in pattern analysis and the actual designing of complex patterns in general. All woven patterns, simple as well as complex, conform to the grid imposed by a loom’s built-in repeating system and ultimately to the right interlacing of warps and wefts. Grids are therefore, necessary in all aspects of plain and figured weaving.

    All the repeat layouts just discussed were in general use worldwide in the sixteenth century. In the Safavid period there appears to have been a fondness for placing two mirror images of the minimum pattern area one above the other, within the technical repeat unit in both straight and pointed repeat systems. This layout may have appealed to the silk designers working in Iran in the sixteenth and seventeenth centuries and may help characterize their approach to design.

    A significant feature of Safavid continuous patterns in lampas and velvet weaves is the use of the human figure and narrative subject matter. Motifs are distinguished by clarity of drawing and can be identified as Safavid by the style of motifs such as birds, leaves, trees, flowers. No single motif is overly highlighted in their continuous patterns and there are practically no open spaces. Nor are there grand, large-scale movements in spite of the length of some of their repeats. Rather there is an equal distribution of points of emphasis. Perhaps in an effort to enliven the static nature of their pattern, a high number of wefts of different colors were used in certain lamlpas weaves and the substitution of pile warps was invented in velvet weaving.

    One must often visually reconstruct colors that have faded, such as red and yellow, or deteriorated, such as black, in order to fully appreciate the original vibrancy and strength of a Safqavid silk. Mary McWilliams has noted that Safavid velvet and lamps weaves attributed to the sixteenth century are characterized by bright, clear colors and strong value contrasts. Lighter colors, such as salmon red-orange, pale green and pale blue, were introcuced in the seventeenth century and appear alongside the bright colors in a transitional phase. By the end of the Safavid period the paler colors, closer in value, dominate the palette. Consistent with this development is the use of black or another dark color for outlines in the earlier silks. By the end of the Safavid period outlines are less emphatic due to the use of a lighter color palette.


    In this section, cat. nos. 1 and 58 will be the example for the following discussion. The plate accompanying cat. no. 1 shows a diagram of the front and back of the structure based on this Safavid lampas.13 Figures 10a,b show the front and back of the head of a bird superimposed on the cypress with diagram in figure 10c.

    A lampas weave is a particular combination of two weaves, each having its own warps and wefts. One weave is the foundation, here a 4&1 satin weave with an interruption of two.14 The other weave is supplementary to it, with its own warps and wefts, here a 1Z3 twill weave.15

    On the front of this lampas and others like it there is a contrast between the warp-float face of the foundation satin weave and the weft-float face of the supplementary twill weave. In Safavid examples the warp-float face of the satin-weave foundation is used for the background of the pattern and the weft-float face of the supplementary twill for the motifs. The back of the fabric is dominated by the warp-float face of the twill, a 3S1 twill, and its many wefts, which completely cover the weft-float face of the foundation satin weave.

    The foundation satin weave in this diagram and fabric has yellow warps and wefts. The supplementary twill weave has five wefts in each shed, each a different color. This particular type of lampas is characterized by the fact that the warp of the supplementary twill weave (in the diagram colored pink, but in the fabric yellow) not only interlaces with its own five wefts in 1Z3 twill, but also with the weft of the foundation weave in the same twill interlacing sequence (see n. 17 for a variation). In this lampas there are five foundation satin-weave warps to one warp of the supplementary twill weave, or a warp proportiion of 5:1.

    Wefts of the supplementary twill weave are brought to the front of the lampas one at a time to produce the pattern, and in these areas is the weft-float face of the twill. When not needed on the front as required by the pattern, they remain on the back. Whether on front or back, twill interlacing is maintained.

    The wefts of the supplementary twill weave are brought to the front of the lampas one at a time to produce the pattern, and in these areas is the weft-float face of the twill. When not needed on the front as required by the pattern, they remain on the back. Whether on front or back, twill interlacing is maintained.

    The wefts of the supplementary twill that are on the front for purposes of pattern are separated from those on the back by warps of the foundation weave. Therefore foundation warps control pattern, which in this example is in groups of seven. A weft of the supplementary twill weave, when it is on the front of foundation warps, floats over no less than seven foundation warps, or multiples of seven, before being transferred to the back. Groups of seven foundation warps remain constant throughout the length of the fabric, each group of seven being a warp-pattern step. Warp-pattern steps are controlled individually by the pattern harness of the loom, in this case a drawloom.

    The fact that the foundation weave is a middle plane pierced by the wefts of the supplementary weave is one of the distinguishing features of a lampas and differentiates it from a double cloth.16

    During weaving, one weft of the satin weave is followed by each of the wefts of the supplementary twill weave in their respective order; in this diagram there are five in the order of green, orange, violet, blue, and yellow. The weaver controls the sheds of both the satin and twill weaves and their interlaced connection. The selection of twill wefts to pattern is programmed into and achieved by the pattern harness of the drawloom, which is worked by an assistant perched on top or at the side of the loom. A separate pattern selection or pattern shed is required for each of the five wefts in the supplementary twill weave. In this diagram and fabric, the pattern selection for the entire group of five wefts is completed and repeated in the next group of five. This means there is a weft-pattern step of two.

    The warps of the foundation satin weave interlace only with their own wefts, whereas the warps of the supplementary twill weave interlace with their own five wefts plus the weft of the foundation weave. When compared in profile, the warps of the supplementary twill weave curve more than the warps of the foundation weave and have a greater “take-up.” Therefore the loom on which this lampas and others like it were woven had two warp beams to accomodate the differing amounts of warp being taken up or wound off each warp beam in the process of weaving. Judging the relative amounts of tension on each warp and the maintenance of both tensions requires skill and experience on the part of the weaver and anyone who studies a lampas.

    Tension on the foundation warps must be set so the structure, in this case satin, is properly warp-faced. Tension on the supplementary warp must be tight enough to create an even surface when its wefts are on the front, but not so tight as to pull its wefts deeply into the foundation weave. Tensions vary from structure to structure and more importantly from one period and culture to another. In general, in a Safavid lampas the supplementary weave is bound relatively tightly to the foundation weave.

    The aim in all figurative weaving is to produce the pattern in the same proportion as set out by the designer. A circle is to be round, not oval, and a square equilateral. Maintaining proper tension plays an important role in creating and maintaining correct proportion. Other factors include the proportion of foundation warps to warps of the supplementary weave, the number of foundation warps in the warp-pattern step. The result can be studied by measuring and counting the grid made by the warp-and weft-pattern steps. The grid may or may noy define equilateral units. In this diagram and fabric the warp pattern step of seven foundation warps and the weft-pattern step of two wefts create a square. The entire pattern can be visualized as a square grid, each square being one warp and one weft-pattern step. The pattern can be reproduced by filling in the squares of the grid with various colors. The black outline of the bird’s head seen in figure 10a is shown as a pattern step grid in figure 10c. The distance between the thick lines of the grid reperesent one centimeter in the fabric.

    The number of warp-pattern steps is related to the width of a technical repeat unit in that the pattern harness of the drawloom was set up to control a specific number of warp-pattern steps. In this example, if there are 15 warp-pattern steps per centimeter and the width of the technical repeat unit is 36 centimeters wide, the technical repeat unit then has 540 warp-pattern steps. With two technical repeat units within the selvedge-to-selvedge width, 1080 were needed for the fabric’s width. With seven warps per warp-pattern step, there are 7560 foundation warps in the fabric. This count does not include those warps needed for the left and right edge, which in a Safavid lampas is usually an unpatterned 4&1 satin-weave4 stripe about .5 cm wide. Nor does this count include mistakes and the warps of the supplementary weave.

    Similarly, the number of wet-pattern steps can be counted in the length of the technical repeat unit. Multiplying this number by five, one for each of the wefts in the supplementary weave (six if the lampas had been brocaded) gives the total number of pattern controls and sheds opened by the pattern harness of the drawloom.

    The counts of warp-and weft-pattern steps per centimeter, which in many Safavid lampas weaves is low, help to convey the relative fineness or coarseness of a pattern’s curves. A pattern grid with low counts gives a pattern a jagged appearance even in a fabric with densely set warps and compactly beaten weft. High warp and weft counts do not necessarily mean a fine pattern step count.

    The silk warps of this fabric were set very close together; the fabric was tightly woven, and with five pattern colors carried in each shed of the supplementary twill weave throughout it entire length, the result is a sturdy, heavy weight fabric with a lively colored pattern on a smooth, lustrous satin background. The weight of the fabric is no doubt an indication that it was intended to be used either for furnishings or for tailored garments in which draping and gathering of fabric were not essential. Having as many as five colors in each shed of the supplementary weave throughout the length of the fabric gave the designer great flexibility in the placement of colors; this was further expanded by the option of brocading others. The low count of the pattern grid gives the pattern a boldness it otherwise would not have and allows it to read at a distance.

    Lampas weaves such as this were often brocaded, a fact not shown in the diagram, nor an attribute of this particular fabric. Brocaded wefts in these lampas weaves can be considered to be one of the wefts of the supplementary twill weave, but discontinuous, and in the same shed as the continuous wefts of the twill. The weft that was brocaded was usually a silk wrapped with a narrow strip of metal foil, which in many fabrics has almost completely worn off so its use easily escapes notice.

    Judging by those lampas weaves that survive as full selvedge-to-selvedge widths, as well as those with large-scale patterns that can be reconstructed, the standard width of a Safavid lampas may have been 68 cm plus or minus 2.5 cm.

    All the Safavid lampas weaves studied in preparation for this project have satin-weave foundations. Most have 1&3 twill-weave supplementary weaves. A few have 1&2 twill-weave supplementary weaves. Warp proportions are sometimes 4:1. One has been found so far in which the warp of the supplementary weave interlaces with the satin weave foundation in a plain-weave interlacing sequence and with its own wefts in a 1&3 twill weave.17

    Although there are exceptions, research carried out for this project suggests that the general features that best characterize Safavid lampas weaves are these: a high degree of structural uniformity, heavy weight, a high number of colors in the supplementary weave, a low-pattern grid-count, a standard width of about 68 cm and, of course, distinctive Safavid style motifs.

    Structural variations in lampas weaving, such as the few just noted, are not unusual within the wider geographic and historic frame of such weaves. What such variations mean within a Safavid context is as yet unexplained, but they seem to be diagnotstic features and may indicate workshop distribution and differences as well as production time-spans within Iran. Parallels between structure, motif, and pattern layout may or may not exist and cannot be made until many more lampas weaves are analyzed in as detailed a manner as the example presented here. Comparison of structural features of Iranian lampas weaves produced prior to the Safavid period, as well as those produced in neighboring areas in all periods, will help place them in a clearer historic and international context.18 Even if historic documents provide measurements and describe materials, colors, and surfaces, they are unlikely to reveal the precise structural features of a weave. Therefore technical analysis of surviving fabrics is essential, and such studies must be integrated with information provided in historic documents.

    To summarize, lampas weaves were in use as early as the thirteenth century, and by the founding of the Safavid dynasty they were being woven in a wide gegraphic range in bothe Islamic and Christian silk-weaving centers. The earliest lampas weaves have a supplementary plain weave held to a plain weave foundation. Variations in the interlacec connection of the two weaves appear early in their history. Their evolution can be traced by the increasing number of shafts used for the foundation and supplementary weaves-4&1 satin weave for the foundation and 1&3 twill weave for the supplementary weave were standard in the sixteenth century, Safavid products are straightforward structures with apparently few variations. Bright, vibrant colors and bold, large-scale patterns are their most distinctive features.


    In spite of their luxurious surfaces and sometimes complex structures, velvets were woven following a relatively simple concept: the introduction of supplementary warps into a foundation weave to produce pile. Cat. no. 60 shows a diagram of the front and back of what may be considered a typical Safavid velvet structure. Velvets of the type represented by these diagrams, such as those, for example, in cat. nos. 2,10, and 15, can be identified by the weave of the foundation, the manner in which the supplementary or pile warps are secured in the foundation, how may pile warps there are in one pole unit, where they are used to produce pile, plus the use and binding of supplementary wefts.

    The foundation weave in the velvet represented by these diagrams is a 4&1 satin with an interruption of two. Three pile warps together make up a pile unit and in this diagram there are four such units. The warps of each unit were entered into the loom between six foundation warps, yielding a warp proportion of 6:1, which in these velvets is constant throughout the width of the favric. Any of the three warps in each pile unit could have been pulled to the front of the fabric to make a pile loop. Which one of the three was pulled was determined by the pattern. All the loops were then cut, leaving tufts of an even height.

    Pile warps were held in place by the tightness of the satin-weave foundation with its closely packed warps and by the two wefts of the satin weave between which loops were pulled up. These two wefts, one inserted before the loop and the other after, are, in this velvet, in the same shed of the satin-weave foundation.This pairing of foundation wefts occurs in every other shed of the satin weave, with wefts in alternate, or nonpile, sheds serving a slightly different function.

    The thickness of foundation wefts, the tightness of the weave, the density of the foundation warps and warp proportion, determine the proportion of the pattern grid, which in this velvet is square. Each square could have been filled with a pile loop. With the pattern grid superimposed on pile units regularly spaced acrosss the width of the fabric, it was possible to fill each square of the grid with one of the three warps in the pile unit to produce a pattern. The higher the count the finer the pattern.

    The most obvious feature of this velvet is the contrast between the cut pile of the pattern and the flat surface of the background. Loops have been pulled up to create only the motifs of the pattern, not the background. Velvets with a contrast between pile and nonpile areas are commonly called “voided”. Not all Safavid velvets with a 4&1 satin weave foundation and proportion of 6:1 are voided (see cat. no. 28).

    A great deal of silk is used in this and all velvets due to the amount needed in each pile warp to produce pile. These warps are generally thick, often paired, so that after the loops are cut, the filaments separate and splay to form a full tuft and dense coverage of an area of cut pile. Because each pile warp used silk at a different rate, each has its own take-up rate and source; a spool for each of the pile colors requires an enormous rack behind the loom. Patterns as elaborate and long as these were programmmed in the pattern -harness of the drawloom, which was controlled by the pattern draw-person working in tandem with the weaver, as is shown in illustrations of Eatern and Western drawlooms. A velvet loom can thus be imagined without having to propose its specific details.

    The wefts in the nonpile sheds of the foundation satin weave also consist of a pair, but a pair with a front-to-back relationship. It is between these pairs that all the pile warps pass and are thereby held to the back of the fabric when not being pulled through the structure to form pile on the front.

    The use of two types of paired foundation wefts, one vertically aligned and the other horizontally aligned, is one of the subtle “tricks” of velvet-weaving and an indication of sophisticated technique.

    Due to the tightness of the weave, the high desity of foundation warps, and because of the nature of satin to hide its own weft and anything behind it, the dense, uniform, warp-float faced satin weave appears as the smooth background of the pattern in this voided velvet. The pile warps, when not used for pile, do not blemish the smooth satin surface. Extra insurance for the hiding of pile warps is provided by the front weft of the horizontally aligned pair through which all the pile warps pass.

    However in none of the voided Safavid velvets that I know of is the satin weave fully exposed. Sucessful as the satin weave is its concealing capacity, it was nevertheless covered or faced with narrow strips of metal foil, as indicated in this diagram, or silk wrapped with narrow strips of metal foil. These supplementary facing wefts effectively cover the satin weave and increase the luster, impact, and expense of such velvets. The choice of yellow for the foundation warps provides added depth and warmth to a gold facing.

    These wefts continue unbroken through entire pile motifs, or if wide, well within them. Facing wefts, which are a flat strip of metal foil, appear to have been cut off at the end of each weft passage. A foil-wrapped silk facing weft was not cut off, but carried to the weft passage going in the opposite direction. Thus, facing wefts never penetrate the fabric, but sit on the front. Every third foundation warp binds the facing wefts to the front of the velvet, in this example in a 1Z4 twill interlacing order. In the diagram the facing wefts are cut off in the lower right corner to expose the yellow satin-weave foundation. In those velvets with a 4&1 satin-weave foundation with an interruption of one, the facing wfefts are bound by every third warp in a 1S4 twill order.

    In some velvets a supplementary weft was inserted between the two metal facing wefts and pulled up to form loops. Although shown in the diagram as open loops, they often crossed to form closed loops. They were bound to the front ot the fabric by those warps which bound both the preceding and following facing weft, resulting, in this example, in 2Z3 twill binding order. Usually a foil-wrapped silk, these wefts were used sparingly for small details such as the centers of flowers or jewelry, as in fugure 11.

    In cat. no. 2, the looped supplementary weft was carried from one weft passage to another by having been pushed to the back of the fabric where it appears as a short spanned float between one passage and the next (not illustrated). This is the only weft in these velvets which moves in and out of the structure. It was no doubt made to penetrate the fabric to avoid the bump which would have been caused by carrying this heavy, foil-wrapped silk weft from one widely spaced passage to another.

    The diagrams with cat.no. 60 show a small scetion of a velvet with three pile warps forming a unit, each of the four units having three warps of the same three colors. In this small section having a pattern grid with four squares across and three up and down, each square could have been filled with any one of the three colors. Additional colors could have been used without adding or subtracting from the number in each unit by simply changing the color of one or all of the three warps in a unit. Such color changes, if maintained throughout the length of the fabric, form stripes, the edges of which ideally line up with motifs falling on or close to the edges of the striped color changes, not beyond.

    Safavid velvet weavers, however, were freed from restrictions imposed by stripes by taking pile warps out of units and substituting them with warps of other colors. In this way units of pile warps contained three warps in all areas of the fabric, and colors could be changed at will. These color changes could be as wide as an entire motif, or limited to only a detail within a motif, and maintained as long as needed, or until a new color was required. Short ends of the warps taken in and out of the structure are left on the backs of such velvets, and color changes give them the appearance of patchwork when seen on the back, as in figure 12. On the front, colors are constantly and freely changing, with no hint of where they were actually taken out or introduced.

    In many velvets black was used for outlines and details such as hair, and, because it is needed throughout the pattern, it was one of the three warps in each pile unit across the full width of the fabric and was never taken out. The other two colors in each unit were substituted, so that none of the figures in the next repeat are colored in the same way. Pile-warp substitution was used in the majority of velvets represented by this diagram, but pile-warp substitution itself was not shown in this diagram. The flexibility of coloring can be seen in velvets (cat. nos. 2,35).

    The back of a velvet is dominated by its pile warps. They are attached to the fabric where they were pulled through the structure to produce pile and, in this example, by the horizontally aligned pair of foundation wefts through which they all pass. The front weft of the pair prevents the pile warps from being seen on the front. The back weft of the pair is in every other shed of the weft-float face of the satin weave foundation and is the only foundation weft that can be seen on the back. When looking at the back of the velvet these wefts appear to be bound as a 1&4 twill, the direction being determined by the interruption of the satin weave. The two diagrams at the top of figure 13 show the wefts on the back bound in an S-direction, here a 1&4 satin weave with an interruption of one. The two diagrams at the bottom of figure 13 show a Z-binding direction, here a 1&4 satin weave with an interruption of two. The wefts on the back hold nonpile-producing warps to the velvet. The nonpile-producing warps are deflected along the twill direction because they are under less tension than those producing pile.

    The effect of an S-direction binding on the pile warps, set to the proportion of six foundation to one pile unit, as in this example, is diagrammed in figure 13, top left. The thickest lines reperesent deflected nonpile-producing warps. Medium weight lines represent a pole-producing warp that is held in verticlal alignment and under tension by having been pulled through to the front. Fine lines are the foundation warps and alternate foundation wefts (the back of the horizontally aligned pair) which appear on the back. Nonpile-producing warps are deflected into the S-direction by five successively bound wefts. They are abruptly forced into the counter Z-direction and brought back into alignment before the continuation of the S-binding of the next five wefts.

    Counter deflection occurs when the diagonals of the binding sequence cross a pile-warp unit. These abrupt counter deflections catch the light differently, causing their alignment to stand out. On the backs of those velvets with warp proportion of 6:1, alignment is either S- or Z- depending on the interruption of the 4&1 satin weave foundation. See figure 13, top left and bottom left.

    In those velvets ith five foundation warps to one pile unit, pile warps not used for pile are deflected in the S- or Z-direction, depending on the interruption of the 4&1 satin weave foundation, but the abrupt counter deflection occurs in horizontal rows, as in figure 13, top right and bottom right.

    Almost twenty years ago, when we worked together in The Textile Museum, Louisa Bellinger shared with me the same observations about the deflection of pile warps on the backs of velvets with 4&1 satin weave foundations, saying that Safavid velvets have diagonals on the back and a warp proportion 6:1; Indian velvets have no diagonals, and a warp proportion 5:1. This has been a clue to geographic origin. There is no advantage of one proportion or the other, and there is no reason why both Iranian and Indian velvet weavers would not have used them both. However, this is a significant technical feature when one considers the conservative nature of the textile industry in general. Although styles may change, technical details such as proportion tend to remain the same when conditions supporting production are stable, as they were in Iran and India in the sixteenth and seventeenth centuries.

    The only Safavid velvet I am aware of with a foundation weave other than a satin weave is cat. no. 34, which has a plain weave foundation of which there is a diagram. Both pattern and background are cut pile of an even height covering the entire front. One pile unit alternates with two foundation warps. The two foundation warps alternate one thick, one thin. All pile warp units contain one green and one white warp. The unique feature of this velvet is that the foundation wefts are maintained on two levels by the thick foundation warps (actually tripled silk warps), held under high tension throughout the structure. The alternate thin foundation warps (actually a single silk warp) are allowed to curve in the interlacing, having been held under less tension. The wefts on the back are the pair in one shed of the plain weave through which pile warp loops were pulled. The wefts on the front are single over which all pile warps pass, thereby being firmly locked into the structure. Since the entire front is covered with cut pile it was not necessary to hide the underlying structure and the pile warps passing over the weft on the front.19 It is tempting to suggest that this velvet might be earlier that those with satin-weave foundations. In this catalogue it has been dated to the reign of Shah Tahmasp I (1524-76) on the basis of turban.

    Judging by those velvet weaves that survive as full selvedge-to-selvedge widths, as well as those with large-scale patterns that can be reconstructed, the standard width of a Safavid velvet may have been 72 cm plus or minus 2.5 cm.

    Velvets, that is fabrics with pile produced by a supplementary warp, were woven in the Eastern Mediterranean using linen as early as the fourth or fifth centuries A.D.,20 often conbined with tapestry weaving using purple wool. There appears to be a gap in time between these linen velvets and the fourteenth-century Italian and Spanish velvets that used silk.21 Satin-weave foundations were used for silk velvets throughout the fifteenth century in Italy and Spain as well as in Northern Europe. It was in this century that all the technigues that appear in Safavid examples, plus other technigues as well, were developed, including supplementary metallic facing wefts, supplementary metallic weft loops, and repeats up to one meter or more, some of which filled the entire width of the velvet.22

    While the origins of velvet-weaving in Iran are unknown, the earliest surviving examples are sixteenth century. It is possible that the specialized techniques needed to produce velvet were introduced into Iran from Europe, either through direct contact or through intermediaries such as the Ottomans. Ottoman velvets sometimes show the influence of their Italian neighbors23 and vice versa. It may be significant in this respect that European Christians were established in Iran in the fourteenth century.24 The Christian clergy may have played an important role in introducing European velvets produced in Europe. The great mystery is what developed in Iran during the fifteenth century. Harold Burnham’s thesis that this specialized weaving technique was introduced into China by the Portuguese shortly after their expansion into the Far East in the sixteenth century, 25 is yet another incentive to suppose it was introduced at an as-yet-unkown date to Safavid weavers as well.

    No matter what the history of velvet weaving in Iran is, Safavid weavers developed a technique uniquely their own: pile-warp substitution. Themes for their patterns were drawn from their own history and culture as well as from the international crosscurrents in Iran at the time.

    A thorough understanding of Safavid velvet must await detailed studies of Turkish, Indian, and additional European velvets that were woven before, during, and after the Safavid period. For example, there are also velvets with twill-weave foundations. Some of them are Indian and as early as the seventeenth century, while others seem to be Iranian of the nineteenth or twentieth centuries.26 None of the velvets with a twill-weave foundation appears to be Safavid. Observations such as this, however, tentative, indicate that Safavid velvets were consistent in structure and quality, which in turn indicates centralized and controlled production within the Safavid state.


    The consistency of velvet and lampas weave structures, the predominant use of mirror images in a technical repeat unit in both straight and pointed repeat systems, and the reliance on stock motifs point to a tightly organized stable silk weaving industry and patrons rich enough to support it. Details of structure show that production of Safavid velvet and lampas was extremely labor-intensive and required great skill. This is best demonstrated in the quality of color, the high number of colors incorporated in one shed of the supplementary weave of a lampas, the substitution of pile warps in velvet, and the length of some of the repeats-216 cm long in the “Four Women in a Garden” (cat. no. 2). It is clear that the best Safavid velvets and lampas weaves are among the highest quality textiles of the sixteenth and seventeenth centuries and represent to me the highest achievement in the arts of Safavid figurative weaving at this time.

    In general, the designs for Safavid velvets appear stronger and more convincing than those for lampas weaves. Might this relative difference be related to the hypothesis that the weaving of sumptuous velvets was introduced into Iran after initial development in the Mediterranean? Or was it simply that lampas weaves had an older and perhaps faded tradition in Iran and that velvets were more expensive? It may be that the innovative designs needed for velvets were to reflect the glory of the Safavids.

    1 Lampas is a term recommended by the Centre International d’Etude des Textiles Anciens (CIETA), in Vocabulary of Technical Terms (Lyons: CIETA, 1964), p.28. It is defined in Dorothy K. Burnham, Warp and Weft, A Textile Terminology (Toronto: Royal Ontario Musuem, 1980), p. 82. Irene Emery introduced it only in preliminary fashion under the heading “Integrated Weave Structures,” in The Primary Structures of Fabrics (Wasgington, D.C.: The Textile Museum, 1966), pp. 159-60. Figured is a term for any fabric patterned in the weave and is useful in avoiding reference to a specific loom type.

    2 The first is the red and white “prisoner” silk in the Yale University Art Gallery(1937-4628); the second is a small-scale Chinese style pattern in blue and white silk, examples of which are preserved in the Textile Musuem (3.214), the Yale University Art Gallery (1937-4629), and the Museum of Fine Arts, Boston (36.570). Both are a 4 &1 satin weave. Damask, a simple weave having one set of warps and one set of wefts, is the juxtaposition of two faces of a two-faced weave, such as 3 & 1 twill, 4 & 1 satin, or 7&1 satin.

    3 French, trame brochee. Emery, Primary Structures, “discontinuous supplementary weft, “ pp. 171-72.

    4 French, lisere. What is meant here is a simple weave, one having one set of warps and one set of wefts patterned by floats, as opposed to a simple weave patterned by a continuous supplementary weft, in French, lance.

    5French, taquete. Emery, Primary Structures,pp. 150-51, described them as a compound weave with “3-span floats in alternate alignment.”

    6Ann Rowe places them in the compound complementary weft group. See Ann Rowe, “After Emery: Further Considerations of Fabric Classification and Terminology,” The Textile Museum Journal 23 (1984): 62, figs. 10-13.

    7Harness in this context does not mean shaft, but everything needed to control warp movement. For example, a multishaft loom is a single harness loom; a drawloom has two harnesses, one for the weave structure and another for the pattern. See Burnham, Warp and Weft,drawloom, p. 48, harness, p. 69.

    8 In the discussion of lampas in the next section, see warp-pattern step, and in the section after that, on velvet, see pile warp units.

    9 Measurements of historic fabrics and reconstructions of historic patterns are never precise due to the changes in scale that occur along a length during the weaving and the stretching that occurs after use. Warp direction measurements are losted first, weft direction measurements second.

    10Friedrich Spuhler, Islamic Carpets and Textiles in the Keir Collection (London: Faber and Faber, 1978) no. 95. Observed by Lucy Maitland, 1987

    11The two versions of “Khusraw sees Shirin Bathing” could also have been described as having horizontally offset repeat units as implied by the ragged edges of the reconstruction. To imagine such an interpretation, rotate fig. 1c90 degrees. No matter how an offset is interpreted in a woven continuous pattern (either vertically or horizontally), the two positions of the minimum pattern area occupy the same square unit measure comprising the technical repeat unit needed, which is always woven in a straight repeat system.

    12Charles Grant Ellis, “The System of Multiple Levels,” A Survey of Persian Art (Oxford University Press, , 1967) vol. 14 3172-83. The five levels, each meticulously drawn and illustrated in a different color, are shown in various combinations.

    13Louise Bellinger chose this fabric to illustrate a Safavid lampas in a small exhibition in The Textile Museum in 1964 for which I prepared the prototype of this diagramming style. It is to her I dedicate these essays on the structure.

    14Interruption is a term used to describe satin weaves, referring to the number of warps between successive binding points following the Z-diagonal on the warp-float face and weft-float face of the satin weave. Thus a 4&1 satin weave (the warp-float face) with an interruption of two is, on the reverse, a 1&4 satin weave( the weft-float face) wtih an interruption of one. These can be written 4&1(2) and 1&4(1).

    15This symbol indicates that the warp is over one weft, the diagonal of the twill is in the Z-direction and the weft floats over 3 warps. This, then, is the weft-float face of the twill. The reverse would be written:3S1 twill.

    16 The distinguishing feature of a double cloth is that the two weaves “intercross.”

    17 The fabric studied is a fragment in the Los Angeles County Museum of Art (M, 73.5.637), other pieces of which are preserved in The Victoria and Albert Museum (T.718c-1899), and the Deutsches Textilmuseum, Krefeld (01302). The pattern is composed of four offset cartouches, each containing a garden scene, and four panels, each containing two animals.

    18Identifying Timurid lampas weaves has intrigued scholars for the past century and there have been periodic changes in attribution. The most recent work is that of Anne E. Wardwell, “Silk and Gold/Silver Textiles from the late Mongol to early Timurid Period from Central Asia and the Near East,” forthcoming in Islamic Art, 1988.

    19If this structure is turned 180 degrees and wefts become warps, it suggests the structure of pile carpets with warps on two levels. Might there be a relationship? Noted by Peggy Osterkamp working with Milton Sonday.

    20Albert Frank Kendrick,Catalogue of Textiles from Burying Grounds in Egypt,3 vols. (published under the authority of His Majest’s Stationery Office, London, 1920-22) vols. 1 and 2; vol. 2, nos. 301 (Victoria and Albert 708-1886), 302 (ibid.,682-1886),303 (ibid., 691-1886). That these are velvets is not noted in the catalogue, but were so noted and diagrammed by Frances Morris in her personal copy which is now in the Textile Department of the Cooper-Hewitt Museum.

    21Milton Sonday, “What Can We Learn from a Fabric About the Loom on Which It Might Have Been Woven?” Irene Emery Roundtable on Museum Textiles Washington, D.C.: The Textile Museum,1979),p.254, figs. 16,17. See also Lisa Monnas, “Developments in Figured Velvet Weaving in Italy During the 14th Century, Bulletin de Liaison du CIETA,nos. 63-64 (1986) pp 63-100.

    22 Chiara Buss et al., Tessuti serici italiani 1450-1530 (Milsn:Electa, 1983).

    23 Adele Coulin Weibel, Two Thousand Years of Textiles (The Detroit Institute of Arts, 1952). 153.

    24 See Peter Jackson and Laurence Lockhart, eds., The Cambridge History of Iran, vol. 6 (Cambridge University Press, 1986), pp. 373-74. That the Mongols opened the way to the East in the 13th and 14th centuries is outlined by Daniel J. Boorstin in The Discoverers, A History of Man’s Search to Know His World and Himself (New York: Vantage Books, 1985): “Then for a single century, from about 1250 to about 1350, that curtain was lifted, and there was direct human contact between Europe and China. During this interlude the bolder and more enterprising Italian merchants no longer had to wait until their exotic goods reached Aleppo, Damascus, or Alexandria. Now they themselves took caravans across the Silk Road to the cities of India and China, where they could hear Christian missionaries, Frankish and Italian friars, saying mass...” p. 125.

    25 Harold Burnham, Chinese Velvets: A Technical Study, Royal Ontario Museum, Art and Archaeology Division, Occasional Paper, vol. 2 (1959): 1-64.

    26 Spuhler, Islamic Carpets, p. 192.

    Back to Top
     About Us   Site Map   Textile History   Tell A Friend   Services   Bibliography   Links   Contact us