The Center for Hellenic Studies

Picturing Homeric Weaving

Susan T. Edmunds


In discussion of Homer and the lyric poets, Gregory Nagy has shown how “the idea of making song is expressed metaphorically through the idea of making fabric.” [1] The process of weaving would have been completely familiar to the average person in Greek antiquity, yet it is not generally familiar to us. We touch woven cloth almost constantly but have little opportunity to observe its making. Moreover, methods of weaving were different in antiquity from what they are today. So while Nagy’s discussion illuminates the metaphor with which the poet explains his art, the metaphor itself may have little illuminating power for a modern reader. My goal here is to shed light on the craft of weaving as it can have been understood by the audiences of a living tradition of Homeric poetry.
I use the expression Homeric weaving to convey a certain detachment from time and place within the history of textile technology. By ‘weaving’ I mean both the process and the product—although the ‘picturing’ I present will necessarily focus on the former. [2] Whether Penelope’s loom looked just like Circe’s loom or whether they used the same techniques or produced similar cloth are not questions that ought to be posed. Ancient audiences, like any audience, would respond to the Homeric poems with their own mental picturing, which in the case of weaving imagery would have varied according to the local weaving traditions with which each audience member was familiar. Nevertheless, there would be certain constants—notably the warp-weighted loom, which was the dominant weaving technology in Greece and adjacent areas from at least the late Neolithic into classical times. [3] The Homeric poems portray weaving as heroic, magnificent, clever, valuable, the womanly counterpart to warfare. [4] It was the work of elite women: Helen, Andromache, Penelope, Arete, as well as goddesses. Circe and Calypso wove, to say nothing of Athene herself, warrior and weaver both. They wove patterned cloth which, in the case of the first three, expressed their own qualities, as well as their relationship to particular men. Helen weaves the story of the Trojan War, [5] Andromache weaves flowery love charms, not knowing that Hector is dead, [6] and Penelope weaves a stratagem to forestall betrayal of Odysseus. [7] For Homeric women, however, their skill as weavers and the great value of cloth made them vulnerable to ending up as booty as the result of heroic activity among men. While no part of this poetic tradition of Homeric weaving can be taken as historical evidence, the Iliad and the Odyssey seem to reflect certain realities in the history of textile production. In the Bronze Age, textiles were among the most valuable trade goods. [8] Cloth production was highly organized in palace workshops controlled by elite members of the society. [9] Elaborately patterned textiles were certainly produced and probably were a means of recording culturally significant information. [10]
Understanding of the ancient Greek warp-weighted loom has increased greatly over the past 75 years or so. In 1936 Grace Crowfoot brought her extensive experience as a weaver, textile archaeologist, and ethnologist to the subject. While she did not know of the still-living tradition of the Scandinavian warp-weighted loom, her essay, which lists most of the important references to the loom in Greek texts and some ancient Greek textile vocabulary, remains a good, brief introduction to the subject. [11] In the 1950s, Marta Hoffmann found women in western Norway and others in Finland who could demonstrate the operation of the warp-weighted loom as they had learned it from their elders. Her book is a detailed study of the history and technology of the warp-weighted loom and includes a discussion of ancient Greek iconographic evidence. [12] Elizabeth Barber, like Hoffmann and Crowfoot a weaver as well as a scholar, published a magisterial study of textile production in the Eastern Mediterranean in the Neolithic and Bronze Ages. [13] Textile archaeology, once more or less ignored in the hunt for collection-ready stuff like pottery and gold, is now an active and fruitful field. It has produced, among many fine examples, the experimental studies done at the Centre for Textile Research of the Danish National Research Foundation. [14] This institution has also sponsored conferences and publications on ancient textiles. [15]
My principal concern in this essay is with the process of weaving on the warp-weighted loom, more or less from the weaver’s point of view. My goal is to help the reader form a mental image of technical aspects of this type of weaving as an aid to understanding literary references. It is thus a picturing with quite a narrow focus. It is specifically not an attempt to make replicas of any possibly historic tools or textiles (although I have constructed and used a warp-weighted loom, which appears in some of the figures). The first section covers some fundamental aspects of weaving in general. Subsequent sections discuss the warp-weighted loom and its mechanics. I compare features of looms in images from the eighth to the fourth centuries BCE to those of the warp-weighted loom that was used in Scandinavia up into modern times and offer several different mechanical possibilities. I then try to bring into focus the action of the weaver’s hands and the tools used, especially the pin beater (kerkis). Finally, I discuss some aspects of pattern weaving.

What is Weaving?

For the purposes of this essay, weaving will be understood as an interlacement of two sets of threads at more or less right angles to each other within a system that holds one set of those threads under tension. The threads held under tension are called warp threads or the warp. The threads that run at right angles to the warp are called weft threads or the weft. [16] A loom is, first of all, a frame or some other system that holds the warp threads under tension. The weft threads intertwine with the warp in one or more of an infinite variety of patterns that produce weave structures.
The simplest of weave structures is plain weave, generally described as weaving in which the weft passes over one warp thread and under the next across the whole warp and then returns, passing over each thread it previously went under and vice-versa. [17] While correct in one sense, this description is misleading in two ways. First it implies that in the process of weaving, the weft is taken over and under, over and under, the warp threads, which is rarely the case. Rather, the warp threads are separated to form a shed, a passage for the weft, as described below. Second, in appearance, a given cloth woven in plain weave may be weft-faced, warp-faced, or balanced. Each looks quite different: in a weft-faced weave the warp stays stretched and is covered by the weft, which is indeed positioned over one warp thread and under the next; in a warp-faced weave the weft remains straight and the closely spaced warp threads are deflected over one shot of weft and under the next; and in a balanced weave both warp threads and weft threads are bent around each other, creating a checkerboard pattern (See Figure 1). Weave structures other than plain weave are used for decorative purposes or to change the quality of the cloth, making it warmer and thicker, for example, or more lace-like. [18]
Some weaving that may appear complex, such as pictorial tapestry, is not a complex weave structure. Tapestry is weft-faced plain weave with discontinuous threads. That is, a weft thread does not pass across the whole width of the warp before turning back in the opposite direction, but differently colored threads each turn back and forth in their own separate areas according to the needs of the design (see Figure 2). Patterns in weaving can also be created through the use of extra threads, for example a supplementary weft. [19] The supplementary weft is not part of the essential structure of the cloth, but embellishes it much like embroidery; unlike embroidery, however, it is introduced as part of the weaving process (see Figure 3).
Weave structures are generally controlled by a shedding device. A shed is the space between the warps through which the weft passes, and a shedding device is the means of creating that space. In order to weave at all, at least two different sheds must be created for a single set of warp threads. Thus, in plain weave, a shedding device causes every even-numbered thread to separate from every odd-numbered thread in one direction for the first shed and to separate in the opposite direction for the second shed. In more complex weave structures, shedding devices separate the warp threads in the same fashion, but in more complex patterns. Sometimes the weaver creates a shed with her hands or with a tool in one small section of the warp at a time. [20]
It is a simple matter to create one shed by inserting a bar (the shed bar) transversely through the warp, picking up every second thread (see Figure 4). The shed bar remains in place as the weaving progresses so that the weaver can easily find that shed, which is sometimes called the natural shed. However, it is physically impossible to create a second shed with a second bar of the same sort. A second shed—called the counter-shed—can be picked up, with the fingers or a tool, each time it is needed, or heddles can be used. Heddles, which take various forms, hold one set of warp threads in such a way that they can be moved past the other set (or sets), thus creating a counter-shed (also called the heddle shed). In modern looms, heddles are generally held in frames (or harnesses) that move in relationship to each other to create sheds. On simpler looms, heddles are held on a heddle bar. They may be individual loops of string or one continuous string looped or knotted around the heddle bar. In either case, one loop goes around each individual warp thread of the counter-shed (see Figure 5). Complex weave structures often involve the use of many different sheds.
Most of the very few textile fragments that survive from Greek antiquity are woven in plain weave. Some however are woven in twill, a weave structure in which the weft passes under and over two or more warp threads at a time, in various patterns. A common twill, and one used in Greek antiquity is 2/2 twill, which involves four different sheds with warps moving in pairs. Each weft thread passes under two warp threads and over the next two, with each subsequent weft thread offset by one. [21] Cloth woven in 2/2 twill has diagonal lines or ridges, such as those seen in denim fabric (see Figure 6).

What is a Warp-Weighted Loom?

Though we speak of looms as things, and modern industrial looms generally are great clanking assemblages of precisely manufactured parts that have much the same presence whether there is cloth on them or not, a loom is essentially more a method or a process than a thing. A so-called ‘primitive loom’ may consist of nothing but a few sticks and bits of string in addition to the cloth or imminent cloth being woven. It is customary to say that cloth is woven ‘on a loom’, but once the cloth is woven and removed, what is left is hardly a loom. The loom materializes temporarily in the process of weaving and is not necessarily separable in thought from either process or product, as is evident in the introduction of Penelope’s ruse.
στησαμένη μέγαν ἱστὸν ἐνὶ μεγάροισιν ὕφαινε,
λεπτὸν καί περίμετρον·
Setting up in the palace a great histon, she wove
a fine and very wide (histon)
Odyssey ii 94–95 [22]
The word histon in the first line must be translated ‘loom’. In the second line, where it is understood as the object of huphaine (‘she wove’) and modified by lepton (‘fine’) and perimetron (‘very wide’) it must mean ‘weaving’ in the sense of ‘fabric’.
A ‘primitive’ loom does not necessarily result in the production of primitive cloth. To some degree, the reverse may be true. Some of the most sophisticated weaving the world has known was done in pre-Columbian Peru on back-strap looms, which consist of a few sticks and pieces of thread in addition to the threads being woven. The warp is tensioned by the weaver’s body. By analogy, a hollow log may be beaten with sticks and the sophistication and intricacy of the percussive music produced has little to do with the log. Aside from the physics of sound and the psychological and physical constants of human hearing, it has everything to do with a society in which rhythm-making is practiced and appreciated and a tradition in which rhythmic forms have developed. The warp-weighted loom, likewise, is best approached as a method or tradition of weaving. It is a method that actually involves two separate looms: one (a band loom) for creating an array of warp threads bound together at one end by a heading band (or, alternatively, arranged on a heading cord); and another (the warp-weighted loom) on which that array of warps is suspended from a horizontal beam, the warp threads tensioned with weights, and the weft introduced to make cloth. The heading band and several different means of shedding in the warp-weighted loom—one generally accepted and three introduced here to address puzzles in the iconographic evidence—are discussed below.
The warp-weighted loom belongs to Europe and the Near East. Its use can be dated to the early Neolithic and it probably began much earlier. Loom weights, which may be ordinary field stones or purpose-made weights of shaped stone, ceramic, or metal, have the archaeological value of durability. Loom weights (as well as quite sophisticated cloth fragments) have been found among other debris left by Neolithic pile dwellers of Switzerland and from even earlier sites in central Europe. [23] They were discovered at Troy in levels that can be dated to 3,000 to 2700 BCE. [24] Hoffmann mentions a possible find of loom weights in Jericho at a level dated to the seventh millennium. [25] Other loom types, probably as old, were in use around the Mediterranean: the ground loom, in which the warps are stretched between beams pegged out on the ground, is especially suitable for a desert climate and nomadic life-style; a vertical loom in which the warp is stretched between two horizontal beams was in use in Egypt somewhat later than the ground loom but became standard, supplanting the warp-weighted loom throughout the region by Roman times. [26] There is no direct evidence for the back-strap loom in Europe or the Middle East, though Barber takes it, especially in band-loom form, as the natural precursor for both the ground loom and the warp-weighted loom. [27]
In various cultures different loom types seem to be associated particularly with male or with female weavers. In early Egyptian representations of weaving, ground looms were used and the weavers are women. Male weavers show up once the vertical two-beam loom is introduced. [28] Similarly in modern Guatemala, women typically use back-strap looms while weavers using horizontal looms with treadles are typically male. Hoffmann notes that in Europe at the beginning of the Middle Ages a shift occurred from the use of the warp-weighted loom for commercial weaving to use of the horizontal loom with treadles. This shift coincided with another: weaving for trade ceased being done by women in the workshops of manor houses and came into the hands of professional male weavers organized into guilds. [29] All of the weavers with first-hand knowledge of the warp-weighted loom whom Hoffmann found in Norway and Finland in the mid-twentieth century were women. She speculates that the warp-weighted loom is essentially a woman’s tradition.
For the Greeks, at least through Classical times, the warp-weighted loom was the primary, if not the only, type of loom in use. We know of warp-weighted looms from many findings of loom weights, from vase paintings, and from literary references. Herodotus writes of the Egyptians, who used the two-beam vertical loom, that they do many things contrariwise, including weaving:
ὑφαίνουσι δὲ οἱ μὲν ἄλλοι ἄνω τὴν κρόκην ὠθέοντες, Αἰγύπτιοι δὲ κάτω.
Other nations weave by beating the weft upward; the Egyptians downward.
Herodotus II.35.10
In other words, weaving by beating the weft upward seemed to Herodotus the normal way to it. Beating against gravity requires extra effort and is tolerable only if gravity is doing some other important work for the weaver, such as tensioning the warps. The weft is beaten upward only on the vertical warp-weighted loom, the type of loom that was familiar to Herodotus.Compounding the perversion of the Egyptians, from the Greek point of view, the weavers there were men:
ὢ πάντ’ ἐκείνω τοῖς ἐν Αἰγύπτῳ νόμοις
φύσιν κατεικασθέντε καὶ βίου τροφάς·
ἐκεῖ γὰρ οἱ μὲν ἄρσενες κατὰ στέγας
θακοῦσιν ἱστουργοῦντες, αἱ δὲ σύννομοι
τἄξω βίου τροφεῖα πορσυνουσ’ἀεί.
O, true image of the ways of Egypt that they show in their nature and their life! For there the men stay at home weaving, but the wives go forth to win the daily bread.
Sophocles Oedipus at Colonus 337–342 [30]
While there is evidence of male weavers in Greek antiquity, the Homeric weaver under discussion here, the archetypal user of the warp-weighted loom, is certainly female. [31]
The warp-weighted loom was gradually abandoned throughout Europe in favor of other technologies. As mentioned above, it survived into modern times in parts of Scandinavia and is still used among the Sami people, though perhaps mostly out of a desire to preserve a cultural heritage. Because it fell out of use, it has been assumed that it was always an inferior technology. Those who think so particularly mention that one has to stand to use it, though I have often thought after a few hours of seated weaving that walking up and down at the loom would be far easier on the body over time. The warp-weighted loom offers a distinct advantage over other looms in maintaining a constant and even tension on the warp threads, no matter how they are manipulated. In pattern weaving, especially, where weft threads of different sizes may be introduced, often with uneven distribution, warp-weighting is an elegant solution to a sometimes difficult problem.

Operation of the Ancient Greek Warp-Weighted Loom

This discussion refers to eleven surviving images of Greek warp-weighted looms. [32] I attempt to make sense of them as operational looms and I take the details seriously, trying not to fall back on the claim that the painter either did not know exactly what looms looked like or chose to represent a loom in some fanciful way. On the other hand, every detail cannot be taken literally. For example, some images show very widely spaced warps, each with its own individual loom weight. The warps would always be tied to the weights in bunches. No matter how coarse or how widely spaced the warp threads, it would be impossible for the painter to have painted each one, no matter how fine a brush he might have. [33] Faced with a choice, the painters (with one exception) rejected the option of representing warp threads as a mass and chose instead to portray the functional relationship of warp thread and loom weight. It is also reasonable to assume that some details of looms and various weaving tools were omitted from the ancient representations, which were not—after all—intended as operational diagrams.
I have redrawn the images of the looms after photographs of the vases, or in one case the stone stele, that bear the original image. My drawings show only the loom, except where a weaver was depicted in the process of weaving. In two cases, I have used drawings from the original publication of the image because I could not get a photograph that gave sufficient information. I have made every effort to represent faithfully each loom as depicted, though I have omitted details that are not useful for understanding the loom, like detectable brush strokes within a shape. Full references to the representations of ancient looms are listed in the Appendix along with links to on-line photographs of the objects on which they appear, if they are available. For convenience in the following discussion, I call the loom images ‘the Cypriot loom’ (see Figure 7), ‘the Corinthian looms’ (see Figure 8), ‘the Metropolitan Museum loom’ (see Figure 9), ‘the Chiusi loom’ (see Figure 10), ‘the Pisticci loom’ (see Figure 11), ‘the Attic loom’ (see Figure 12), ‘the British Museum loom’ (see Figure 13), ‘the Sackler Museum loom’ (see Figure 14), [34] ‘the Mississippi loom’ (see Figure 15), ‘the Ashmolean loom’ (see Figure 16), and the ‘Thessalian loom’ (see Figure 17).
As might be expected from the inseparability of loom and weaving process, all the images of looms include cloth. The normal length of a peplos, and probably most clothing woven on the warp-weighted loom was about 4 to 5 feet wide and 5 to seven feet long. [35] Cloth of that length could not be made without rolling it up as it was being woven and, indeed, in most of the loom images, it is clear that a length of already woven cloth is wound on the top beam of the loom, the ‘cloth beam’. [36] The ruse that Penelope used to delay answering her suitors—weaving by day and unweaving by night so that the finishing of the work would be indefinitely delayed—is much more plausible if one knows that the cloth was wound up as it was woven and thus could never be visible on the loom in its entirety. It would be almost impossible to unroll it until the whole cloth was completed and ready to be removed from the loom.
Some of the following discussion and some of the figures are based on what might loosely be called ‘experimental archaeology’. That is, I have constructed a warp-weighted loom along the lines of looms found in the iconographic evidence, and I have tested various possible ways of weaving on them. I should be clear that this exploration does not meet accepted standards for the scientific use of experimental archaeology. [37] The materials used were chosen for clarity in demonstration and for availability and not for historical accuracy. My purpose here is to demonstrate, in broad outlines, the possibility of various weaving methods that make sense of the iconographic evidence.

The Heading Band

As mentioned above, the first step in setting up a warp-weighted loom (or setting up a weaving) is making a heading band. The heading band (also sometimes called a starting band) is made on a band loom, although this ‘loom’ may be no more than a small number of threads tied between two uprights (or otherwise tensioned) and fitted with a shedding device. It is not known what kind of tensioning and heddling devices the ancient Greeks used in band looms—although bands were certainly woven for many purposes in addition to the organizing of warp threads for the larger loom. [38] A rigid heddle device is one option, and I have used it for convenience (see Figure 18). [39] The warps of the band loom must be somewhat longer than the width of the planned cloth to be woven on the warp-weighted loom. The wefts for the band loom will be the warps for the planned cloth. Loops of this weft-soon-to-be-warp are drawn through shed and counter-shed in the band-loom warps and extended to a length that will accommodate the weaving of the planned cloth (see Figure 19). The warp-faced band, with its distinctive ribbing, will become a decorative strip along one side of the eventual cloth (see Figure 20). [40] The character of the heading band—its materials, its fineness, its length, its organization—is the guiding structure of the finished cloth. To reverse the metaphor Nagy delineates in discussing the Panathenaic Festival as humnos, the heading band is the beginning of the fabric just as “the specially intricate humnos is the beginning of the rest of the song.” [41]

The Frame of the Loom and Loom Mechanics

Once made, the heading band is attached to the cloth beam, which is then fitted to the top of two other beams, which will be the uprights of the loom. Near the center (from top to bottom) of the uprights, one or more other beams are attached parallel to the cloth beam. Their function relates to shedding and will be discussed below. Then the whole is leaned against a wall or otherwise made to stand more or less upright. [42] The warp threads, which would have been secured in some way to prevent them from tangling, are loosened from their anti-tangling arrangements, disposed over the center cross-beam or beams, and attached in bunches to weights in order to hold them taut.
The disposition of the warps in relation to the center cross-beam(s) and the making and use of heddles is discussed in what follows. Grace Crowfoot, Marta Hoffmann, Elizabeth Barber, and others with a hands-on knowledge of weaving have argued convincingly and at length that ancient Greeks understood and used heddles. It seems unnecessary to debate that point further. Exactly how they used heddles is not resolved, however. I will describe four different possible models for creating heddle sheds, each of which involves the weaver in a different sort of action.
Almost the only thing we know from Greek literature about the weaver’s action is that she walks back and forth at the loom. [43] She might also sing as she does so. [44] The Iliad gives us one other clue: during the funeral games for Patroklos, Odysseus is very close behind Aias in a running race…
ἄγχι μάλ’, ὡς ὅτε τίς τε γυναικὸς ἐυζώνοιο
στήθεὸς ἐστι κανών, ὅν τ’ εὖ μάλα χερσὶ τανύσσῃ
πηνίον ἐξέλκουσα παρὲκ μίτον, ἀγχόθι δ’ ἴσχει
στήθεος· …
... very close, as is the weaving rod to the breast of a fair-belted woman when she deftly draws it in her hands, pulling the bobbin through the warp, and holds (the weaving rod) near to her breast…
Iliad 23.760–763
What the weaving rod (kanōn) might be and how it might be used will be considered for each of the four models.
The Northern European Model
Current understanding of the ancient Greek warp-weighted loom owes much to evidence from the northern rim of Europe gathered over the past 150 years or so, including the still-living tradition studied by Hoffmann in the 1950s. [45] Hoffmann’s informants in the west of Norway still used warp-weighted looms for making åklær, woolen coverlets woven in traditional patterns (see Figure 21). Despite differences in the appearance of the northern European loom and the ancient Greek loom, it has been assumed that there were no great differences in operation. Barber’s schematic drawing of a notional ancient Greek warp-weighted loom, juxtaposed to Hoffmann’s schematic drawing of a notional northern European warp-weighted loom, gives an overview of the generally accepted understanding of the loom’s mechanics (see Figure 22). Barber notes “the higher shed bar and flimsier heddle bar(s) in the ancient version.” [46] In another version of the same drawing, Barber labels each of these flimsier heddle bars kanōn—indicating that this is the bar that the weaver draws toward her breast when she passes the weft bobbin through the shed. [47]
The shedding in the ancient Greek loom and in the northern European warp-weighted loom is presented as identical. The loom frame is supported by a wall or other structure and leans backward, tilting away from the weaver. The warp hangs from a cloth beam and is tensioned by weights. The warp threads are arranged to hang alternately in front of and behind a crossbeam attached to the uprights—the shed bar. [48] In the northern loom the shed bar is low, just above the weights; in the Greek loom, it is a center cross-beam, halfway between the top and bottom of the uprights. Loops of thread—the heddles—are attached to the warp threads that hang behind the shed bar and to a moveable heddle bar or, in Barber’s drawing of the Greek loom, two heddle bars, each extending about half the width of the warp (see Figure 23). In both looms, one shed, the natural shed, is formed by the combined force of gravity and the shed bar which holds half the warps further forward than the rest. The counter-shed is formed when the heddle bar (which in this model is the kanōn) is pulled forward and the rear warps pass by the front ones (see Figure 24). Weaving is accomplished by drawing the weft alternately through the natural shed and the counter-shed and beating it upward so that each weft row lies neatly against the previous rows. One important difference from the Northern European is that the Greek loom lacks the “strong wooden forks sticking out about the middle of the loom uprights” [49] that support the heddle bar when it is in its lifted position. As Barber explains, it would not be possible to lift a stout heddle bar extending across the width of the loom, along with its attached loom weights, with one hand while passing the weft with the other. Thus in her drawing, the heddle bars are light and extend only partway across the warp. They could be managed by one weaver walking back and forth in front of the loom.
Barber’s schematic drawing shows a loom that is functional and all its features can plausibly be seen in images from antiquity. No single ancient representation of a loom has that configuration of features, however. Barber’s schematic loom has one center-cross beam and two rows of loom weights. The Cypriot loom, the Metropolitan Museum loom, and the Attic loom all have one center cross-beam and one row of loom weights (see Figure 7, Figure 9, and Figure 12). The Chiusi loom, and the four Boeotian looms all clearly have two center cross-beams and two rows of loom weights (see Figure 10, Figure 13, Figure 14, Figure 15, and Figure 16). The Pisticci loom probably does, too, although the drawing, and most likely the image itself, allow various interpretations (see Figure 11). The Thessalian loom has no visible center cross-bars or loom weights. Only the Corinthian looms (see Figure 8) fail to fit the pattern. Both clearly have two center cross-bars and—apparently—one row of loom weights. There are a lot of marks in these paintings that are difficult to interpret. [50]
In the looms with two center cross-beams, the upper beam has often been taken for a heddle bar. Against this notion stand the facts that (1) such a large heavy heddle bar would need to be supported (both against the uprights while the natural shed was in use and away from the uprights when the heddle shed was in use) and no such supports appear in any image from Greek antiquity and (2) in the Chiusi loom the joinery of the loom is indicated and both center cross-beams, as well as three other crossbeams on the loom, are shown joined to the uprights in exactly the same way. The upper one cannot be interpreted as a moveable heddle bar without assuming misrepresentation on the part of the painter.
Painterly misrepresentation has been appealed to in explaining the differences in the loom weights also. The arrangement of loom weights in two rows is known from archaeological finds. [51] For looms that work like those in Barber’s and Hoffmann’s schematic drawings, two rows of weights are a mechanical necessity. If front and back warp threads were attached to the same weight, when the back ones were lifted, the tension on the front ones would be released and the shed would be lost. To avoid the unsatisfying conclusion that the painter of the Cypriot loom, the Metropolitan Museum loom, and the Attic loom all chose not to depict the weights accurately, we might conclude that the surviving images represent two distinct types of looms: looms with one center crossbeam and one row of loom weights, and looms with two center crossbeams and two rows of loom weights. [52] The Corinthian looms, if painted accurately, may represent a third type.
Two questions arise. First, what was the function of the second cross-beam in looms with two rows of loom weights? And, second, how was shedding accomplished in looms with one row of loom weights?
A possible answer to the second question is that there was no mechanical shedding in looms with one center cross-beam and one row of loom weights. Crowfoot, who believes that the Greeks did understand the use of heddles, thinks that heddles may not have been used in looms when tapestry was being woven. [53] The reasoning is that the weft is turning within such small spaces that heddles are more trouble than they are worth, and in fact some present-day tapestry weavers choose not to use heddles. Most do, however, since heddles make the work go faster, especially if there are relatively large areas of a single color, as seems to have been the case in ancient tapestry. Moreover weaving that included tapestry may not have been woven entirely in tapestry. [54] And, to look again at the iconography, the large bobbin in the hand of one of the weavers on the Metropolitan Museum loom suggests that the cloth she is weaving is not tapestry but mostly or entirely of one color. The weft apparently travels at least half-way across the width of the loom, where she would either meet the weft put in by her fellow weaver or hand over the bobbin to her (see Figure 9).
The High-Warp Tapestry Loom Model
A possible purpose of the second cross-beam is to function as a stationary heddle bar. A heddle bar of this type is characteristic of the vertical two-beam tapestry loom traditionally used in the Gobelins Tapestry Works in Paris—the so called high-warp (haute-lisse) loom. Like the warp-weighted loom, the high-warp tapestry loom uses a shed bar to maintain one shed. Unlike on the warp-weighted loom, the weft is beaten downward and the weaving builds from bottom to top. Typically, the seated weaver weaves back and forth in one area at a time, reaching up with one hand to engage the heddles as needed in the area being worked on (see Figure 25). [55] With the other hand, the weaver passes the bobbin through the shed, and then beats down the weft with the pointed end of the bobbin (see Figure 26). The tapestry weaver has no need to open a shed across the entire warp at once since the weft in tapestry weaving travels only a short distance in either direction.
In the warp-weighted loom, the upper of two center cross-beams can function as a stationary heddle bar (see Figure 27). The shed is created by grasping the heddles with the hand or lifting them with a rod (see Figure 28). Like the tapestry weaver, the weaver at the warp-weighted loom does not need to open the shed all the way across the loom. Even if the weft is continuous across the width of the cloth, the weaver travels along with it and can open the shed as she goes. Or she might weave the heddle shed in a two step process: first by opening the shed in small sections (with her hand, or a stick, as shown in Figure 28) and picking up a manageable number of warp threads on a rod. This rod would be the kanōn and is not the same as the stick used to open the small section of shed. The second step would be to open the shed with the kanōn, holding it close to her breast as she passes the weft. Although the weaver on the warp-weighted loom must beat upward, she would not have to reach upward (as the high-warp tapestry loom weaver must) to secure the shed, since the shed bar and heddles would be about at her waist.
The Inkle Loom Model
A variation on the high-warp tapestry loom model involves a quite different method of shed creation, which is used in some inkle looms (a kind of band loom). In this method, the heddles themselves and the warp thread attached to them remain stationery while the other set of warp threads is moved from one side of them to the other (see Figure 29). The natural shed is exactly the same as for the high-warp tapestry loom model. The front warp threads are held forward by the lower center cross-beam. The back warp threads are attached to the upper center cross-beam with string heddles. To form the counter-shed, the weaver pushes the front warp threads backward through the rear warp threads, which are prevented by the heddles from going more than a certain distance back (see Figure 30). The weaver picks up the group of warps behind which she will pass the weft, perhaps transferring them to a weaving rod (kanōn) as described for the previous model. (Again, the stick in the figures should not be confused with the kanōn.) Of all the models discussed, this is the easiest to use. It is also easy to set up, since the heddles can be alternately looped like those in the next model. The looms with two center cross-beams and two rows of loom weights, for example the Chiusi loom or any of the four Boeotian looms, would work well, with no particular modifications, if configured with stationary heddles (see Figure 10 and Figures 13, 14, 15, and 16).
The Alternately Looped Heddle Model
The alternately looped heddle model is an adaptation of a method sometimes used on Central American back-strap looms to create both a shed and, alternatively, half that shed on the same heddle bar. (Lifting every other thread in one shed can be useful in some types of weaving with a supplementary weft). [56] On a warp-weighted loom, this method can be used to create two different sheds on one heddle bar. Thus four sheds can be created on a loom with two stationary heddle bars. (Four or more sheds can also be created on the Northern European warp-weighted loom by adding an additional heddle bar for each additional shed.) Twill requires at least three different sheds, though four-shed twill is, as Barber says, more “expectable” because the warp-weighted loom lends itself to an even division of the warp. [57] Although most of the few fragments of cloth that have survived from ancient Greece were woven in some form of plain weave, there are examples of twill and possibly of other weave structures that would require more than two sheds. [58]
To construct heddles with alternate loops, a continuous heddle thread is passed up and down in front of the heddle bar, looping around one warp above the bar, the next one below, the next one above, and so forth, alternating in this way across the warp (see Figure 31). This method of constructing heddles is very quick and easy compared with, for example, the knitting or crocheting of heddles done on the Northern European looms. [59] Securing the sheds is more laborious than with the other models. First, there is no natural shed. One shed can be opened easily when the upper heddle loops are lifted or a stick placed beside the heddle bar is pushed back. This action can be performed with one hand and the weft passed with the other, or all the warps for that shed might be picked up on a kanōn before the weft is passed. Securing the second shed is a two-step process, however. When the stick is pushed back and the upper shed is formed, a counter-shed forms underneath it (see Figure 32). In order to weave in this shed, a second stick or rod must be inserted through it and the upper shed must be released (and allowed to disappear). As with the high-warp model and the inkle model, the weaver might work along the warp, picking up sections of shed on a longer rod (the kanōn). That rod would hold only the counter-shed, through which the weft can be passed. Because there is no natural shed, this method requires (for plain weave) just one center cross-beam and also just one row of loom weights. What is more, there is no need for a loom configured in this way to lean; it can stand quite upright.
Let us look again at the Metropolitan Museum loom (see Figure 9). It seems to stand upright. It has one center cross-beam and one row of loom weights. The weaver on the right has both hands on the rod that spans her half of the loom. This rod could be the kanōn she is inserting to hold one or the other shed. Some elements in the drawing are, of course, disputable. Do the warp threads as painted on the vase actually hang behind the center cross-beam? Is the weft bobbin actually held within them? The drawing is necessarily an interpretation. (An image of the vase painting can be found on the Museum’s website at The cross hatches along the beam are clearly present. Might they represent heddles, simplified for the sake of clarity, as the widely spaced warp threads certainly are a simplification? Despite these uncertainties, the alternately looped heddle model is far more consistent with the details of this vase painting than is the Northern European model. As mentioned above, the Cypriot loom and the Attic loom are of the same type (see Figure 7 and Figure 12).
There may well be other ways to provide shedding on warp-weighted looms with one center cross-beam and one row of loom weights or on those with two center cross-beams and two rows of loom weights. It could also be argued on the contrary that any loom configuration other than that seen in the living tradition is unlikely, given the generally conservative nature of the craft. [60] Certainly most scholars seem to accept the northern European model as providing the best information available concerning the operation of the ancient Greek warp-weighted loom despite the iconographic evidence. All in all, we probably do best to heed Barber’s admonition, made in regard to another puzzle, “I do not know what the solution may be, although I can think up several scenarios. In any case, it would seem to be one more little bit of evidence that the ancients knew and used more loom types than we currently think they did.” [61]

Loom Weights

As noted above, loom weights are attached to the warps before heddles are made. While the primary function of loom weights is to tension the warps, they also have a role in spacing warp threads. As also noted, loom weights are the only part of warp-weighted looms that have survived from antiquity and there are numerous finds. The weights are of many different sizes and shapes—sometimes even where their disposition makes clear that different sizes and shapes were used simultaneously on the same loom. Recent work in experimental archaeology is providing insight into what must have been a complex functional interrelationship among size, weight, and shape of loom weights in producing textiles of any given type. [62] Part of the weaver’s expertise was to know how to choose the weight, number, and shape of weights that would allow her to produce a cloth of a specific type and quality.

What is a Kerkis and How is it Used?

Not a Shuttle

‘Shuttle’ is a customary but inaccurate translation of kerkis. The purpose of a shuttle is to carry the weft as it introduced into the warp. Among our images of warp-weighted looms from antiquity, several show weft wrapped around some kind of stick or bobbin. Each stick would, of course, hold a separate color of yarn, so in a multicolored weaving there would be many such sticks in use. The extra beam at the top of the Chiusi loom seems to have been designed for the purpose of holding just such an array of small bobbins (see Figure 10). The Ashmolean loom shows a weft bobbin caught between warps, as a weaver might leave it when interrupted at her work (see Figure 16). In the Metropolitan Museum loom, three large weft bobbins can be seen, one in use and two awaiting their turn (see Figure 9). [63] The similarity in shape between the weft wrapped on these bobbins and the spindle in use on another part of the image on Metropolitan Museum vase suggests that such bobbins were (at least sometimes) the shafts of the spindles on which the yarn was originally spun (see Figure 33). [64] A passage in Plato makes clear that a kerkis is a different object from a spindle: each is listed separately as a tool for textile production. [65] Moreover, though a weaver might have multiple weft bobbins for any one weaving project, she would typically have one kerkis.
Most of the Greek words for the structural parts of the warp-weighted loom have non-Indo-European roots. Kerkis, however, has cognates in other Indo-European languages. [66]
*krek- ‘beat the weft with a stick’. ON hræll (< *krokilos) ‘pin beater’ (stick used by a weaver to beat the weft home)…Gk κρέκω ‘strike (the web), weave; pluck a stringed instrument’, κρόξ ‘warp, thread (of the warp)’, κροκύς ‘tuft of wool’, κερκίς ‘pin beater’. A word of the west and center of the IE world. The basic meaning of this lexeme is ‘to beat the weft home with a small stick that has the weft wound on it’… [English ‘reel’ is also a cognate.] The twanging action by which the pin was used to disentangle the warp led in Greek to a transfer to twanging a stringed instrument. [67]
In another context, Barber says, “Among the Greeks, the kerkis seems at least sometimes to have carried the weft on it…” but gives no evidence. [68] If the kerkis did have weft wrapped around it, then it functioned partly as a shuttle, but it was not a shuttle that could be thrown across the warp. [69] A shuttle used on a horizontal, treadle-operated loom generally has a smooth, flat bottom and may be shaped like a boat which holds a bobbin that rotates as the weft is drawn off it. Or it may be shaped like a ski with a kind of pedestal on top of it around which the yarn is wrapped. The shuttle is ‘thrown’ (more accurately, slid) from one side of the web to the other. Shuttles of this type are useless with a warp-weighted loom (see Figure 34).
A more accurate (though less poetic) translation of kerkis is pin beater. [70] A pin beater is a tool used on looms that do not have the reed typical of horizontal treadle-operated looms. The reed is like a comb that has both sides closed. The warp threads run through it. It is swung toward the fell of the cloth to beating in the newly placed weft. LSJ (s.v. κρέκω) gives another possible English translation: ‘weaver’s sley’. A ‘sley’ (or ‘slay’) is, literally, ‘the thing that strikes’ the weft. The sley for the horizontal treadle loom was originally made of reeds and thus the whole device came to be called ‘the reed’ (see Figure 35). The word sley remains in modern usage, as far as I know, only in ‘sley hook’, the device used to pull each warp thread through the reed in the process of dressing the loom. Roughly speaking, the pin beater does the work of the reed, but it does somewhat more and somewhat less, as discussed below.
The shape and size of the Greek pin beater is not known. Plato mentions the existence of different forms appropriate to weaving different types of fabric. [71] Other objects called kerkis in ancient Greek (for example: a measuring rod, a wedge-shaped division of seats in the theatre) suggest that it was stick-like or pointed at one end or both. Hoffman shows an Icelandic pin beater made of whale bone that looks like a flattened stick, pointed at both ends, 21 centimeters long and perhaps two or three centimeters wide. [72] Another (shown in a nineteenth century drawing of an Icelandic loom, held between the warps) is similar but curved, like an elongated crescent moon. [73] Crowfoot, who had visited with ‘primitive’ weavers throughout the eastern Mediterranean in the 1920s and 1930s, says of the pin beater, “I know it chiefly in the form of the gazelle horn, but I have also seen a straight wooden pin used, and, in Palestine, an iron pin, slightly hooked, which is evidently a substitute for the curved horn.” [74] Plato, in the passage concerning their different forms, describes pin beaters as made of wood. Calypso has a golden one. [75]

The Characteristic Tool of the Weaver

The kerkis, or pin beater, was the characteristic tool of the weaver. Again, in the Cratylus, after Socrates has said that “a name is an instrument of teaching and of distinguishing natures, as the kerkis is of distinguishing the threads of the web…”
ΣΩ. Ὑφαντικόν δέ γε ἡ κερκίς;
ΕΡΜ. Πῶς δ’ οὔ;
Socrates: And the kerkis is the instrument of the weaver?
Hermogenes: Assuredly.
Plato Cratylus 388c [76]
In the Ion of Euripides, Kreusa describes as an ἐκδίδαγμα κερκίδος (ekdidagma kerkidos ‘a sampler from the pin beater’) the cloth which identifies Ion as her child. [77] Similarly in Theocritus Helen is praised for weaving with her pin beater finer cloth than anyone. [78] Most poignantly, Andromache, as she weaves a robe, bids her maids to prepare a bath for Hector.
νηπίη, οὐδ’ ἐνόησεν ὅ μιν μάλα τῆλε λοετρῶν
χερσὶν Ἀχιλλῆος δάμασε γλαυκῶπις Ἀθήνη.
κωκυτοῦ δ’ ἤκουσε καὶ οἰμωγῆς ἀπὸ πύργου·
τῆς δ’ ἐλελίχθη γυῖα, χαμαὶ δέ οἱ ἔκπεσε κερκίς.
unwitting one, nor did she know that far from all baths flashing-eyed Athene had vanquished him at the hands of Achilles. But she heard the shrieks and groans from the wall, and her limbs reeled, and from her hand the pin beater fell to the floor.
Iliad 445–448
Andromache is marked in this passage by the word νηπίη (nēpiē), as not just unwitting, but doomed, just as Patroklos is marked as doomed when he asks Achilles permission to go into battle:
Ὥς φάτο λισσόμενος μέγα νήπιος· ἦ γὰρ ἔμελλεν
οἷ αὐτῷ θάνατόν τε κακὸν καὶ κῆρα λιτέσθαι.
So he spoke in prayer, great fool that he was, for it was certain to be his own evil death and fate for which he prayed.
Iliad 16.46–47
As her kerkis falls to the floor, Andromache’s life as the mistress of her own workshop is over. She may weave again, but as a slave. [79]
Identification of the pin beater as the characteristic tool of the weaver shows up on the northern rim of Europe also. Hoffmann cites an old rhyme from the Færoe Islands, where weaving was done on warp-weighted looms into the early nineteenth century. The rhyme is translated: “all the weaving women are coming, each with her pin-beater.” [80]

What the Kerkis Does

As suggested by its root meaning, one function of the pin beater is to beat the weft into place. Another is to even out the warp threads by strumming across them. A third likely use is to pick the shed, especially in pattern weaving. The character of the Stranger in Plato’s Statesman, using an analogy to make a point about the concepts of combining and separating, says that both spinning and weaving consist partly in ‘taking apart the things that lie close together’: in the case of spinning, the preparatory process of combing the wool fibers, and in the case of weaving ‘half the work of the kerkis’. [81] ‘Taking apart the things that lie close together’ could refer to picking the shed, driving the weft home, and strumming to even the warp spacing. The other half of the work of weaving is εὐθυπλοκία κρόκης καὶ στήμονος (euthuplokia krokēs kai stēmonos) ‘even intertwinement of warp and weft’ include driving home the weft and, again, strumming the warp thread.
Beating the Weft into Place
An aspect of weaving that requires experience and skill is placing the weft across the warp so that it is neither too tight nor too loose. If the weft is put into the warp in exactly a straight line, the edges of the cloth will draw in so that the cloth becomes progressively narrower as it is woven; if it is too loose, the cloth will be uneven and lumpy. Thus as the weft is drawn through the shed it is initially pushed up at intervals so that short sections hang loose within the shed. Once it is evenly distributed to the weaver’s satisfaction, the pin beater is used again to thrust the whole thread into place next to the previous passage of the weft. In this action, the kerkis could be said to divide the warps as it slips between them (see Figure 36). Once the whole weft row is in place, it must be beaten home, which is possibly another task for the pin beater. This combining action might also be done by a weaver’s sword (see Figure 37). Aristotle distinguishes between spathēsis (use of the sword) and kerkisis (use of the pin beater), the former being a ‘pushing together’ and the latter a ‘pushing apart.’ [82] In this conception, the sword takes over one function that the pin beater might have in one type of weaving, such as pattern weaving, and the pin beater itself becomes more specialized. The use of the weaving sword appears also in this passage from Aeschylus:
ἰδοῦ δ’ ὕφασμα τοῦτο, σῆς ἔργον χερός,
σπάθης τε πληγὰς ἠδὲ θήρειον γραφήν.
Look at this piece of weaving, the work of your hand,
at the strokes of the weaving sword, and the picture of the beasts upon it! [83]
Aeschylus Choephoroi 231–232
The beating in of the weft is a mark of the weaver’s hand, a sign of her skill or perhaps even a signature to the knowing eye. For the initial beating in, the pin beater must be smooth and pointed. To do the work of the sword, it is best if flattened and perhaps somewhat curved.
Strumming the Warp
Cloth that is woven on looms that have no reed, like back-strap looms or, especially, band looms, is often warp-faced. With warp-faced weaving, the weaver does not have the problem of keeping the warps evenly spaced apart for they are close together by design. But weaving balanced cloth (where warp and weft have equal spacing) on looms without a reed requires other methods. The threads must be carefully placed and strategically beaten in. [84] Even after they are beaten in, there is an unevenness that no amount of beating can fix. The solution is to strum across the warps with the hand or the pin beater, or even to strike the warps with a stick so that they vibrate enough to shake themselves into an orderly row (see Figure 36, numbers 3 and 4). They are thus properly combined. The weaver on the left in the Metropolitan Museum loom illustration may be striking the warp threads in this way, or she may be using the stick as a sword (see Figure 9). If used as a sword, the stick goes between two sets of warp threads, as the weft does; if used to strike, the stick is on the outside, striking across all the warp threads in a given area at once.
Strumming the warp threads addresses another problem as well. When the shed is changed and the warp threads cross past each other, they often tend to stick together, particularly if they are wool. Strumming causes them to separate and release themselves back in place. It is for this strumming action that the pin beater (or, by extension, the loom) is called ‘tuneful.’ [85] Strumming may also cause the loom weights to knock together (as would changing the shed), which might or might not be perceived as musical. For strumming, the pin beater should be smooth and somewhat flattened. For beating, longer and more flexible is better.
Picking a Shed
In pattern weaving, the weaver often creates a temporary shed, section by section, with her hand or a tool. A pin beater for this operation could be small and have some flatness to it with tapering or pointed ends (see Figure 38). It will be awkward if it is too long. If a warp-weighted loom is equipped with alternate-loop heddles, as proposed above for looms with one center cross-bar and one row of loom weights, a tool is also necessary to manipulate the heddles to secure the shed. For this, a larger pin beater would be needed—though not so large it could not also beat in the weft and strum the warps perfectly well (see Figure 32, where a small stick is used for the purpose). Here, the pin beater clearly serves as a separator. Despite their various shapes, to accommodate various purposes, we can trust Plato that all the different pin beaters mentioned above “have the true form of the pin beater.” [86]
Manipulation of the pin beater, whether for separating or combining, requires the kind of skill that comes from long experience. In the words of Mathilde Landercy La κερκίς paraît être l’instrument principal du tissage, celui dont le maniement est le plus difficile. La κέρκισις constitue vraiment, semble-t-il, le travail spécifique du tisserand, la pierre de touche du bon artisan.(“The kerkis appears to be the principal tool of weaving, that which is most difficult to use. Use of the kerkis surely seems to be a skill specific to the weaver and the touchstone of the skilled artisan.”)

Pattern Weaving

I follow Nagy in translating the Greek word poikilos (literally ‘many-colored’), in its role as descriptor of fabric, as ‘pattern-woven.’ [87] Pattern weaving, as the name implies, is the introduction of a pattern of some kind (other than simply the pattern of interlacement of the threads) into a cloth during the weaving process. By pattern, I mean either a regular, overall, wallpaper-type pattern or pictorial figures that might be representative or abstract. While pattern weaving can be monochrome (if patterns are produced by changes in texture), it is typically colored. Pattern weaving can be loom-controlled or manually controlled. A loom-controlled pattern is one that is produced by the set-up of the loom. The diagonal stripes in straight twill are an example, but twill can easily be manipulated to create wider stripes, zigzags, or diamond shapes. Other weave structures offer a limitless array of wallpaper-type patterning. Manually controlled patterns are those that the weaver introduces by manipulating the weaving without essentially changing the overall weave structure. Weft stripes are the simplest example. The weaver can, at any point in the process of introducing the weft, decide to change to another weft of a different color. Next simplest is warp stripes, which take more planning, as their introduction would (for the warp-weighted loom) be part of the construction of the heading band. When warp stripes and weft stripes are combined, the result is plaid. Some weave structures, among them tapestry and certain supplementary weft techniques, enable the weaver to produce practically any imaginable image. Others lend themselves especially to stylized images of particular shapes.
Poikilos has often been translated ‘embroidered.’ Although embroidery was known in Greek antiquity, when a cloth is described as poikilos in Homer or in tragedy, pattern weaving—not embroidery—is meant. [88] Although surviving textiles are so few, it is clear that pattern weaving of various types was very common, from the Bronze Age and through classical times. As Barber writes “the Minoans in particular depicted themselves as wearing multicolored cloth in designs so varied and so elegant as to make the owner of a modern handweaving boutique envious.” [89] Diane Lee Carroll, in the early 1960’s, studied patterned textiles depicted in Greek (and Etruscan) art from the eighth to the second centuries BCE, comparing them to extant textile fragments and concluded that by and large pattern-woven textiles as painted on, for example, vase paintings provide good evidence for what was in fact produced and worn, especially in the fourth and fifth centuries BCE, for which there is the most evidence. [90]
In a well-illustrated discussion of what kind of weaving Penelope might have been doing, Barber shows that first, in order for her ruse to be credible, the funeral shroud Penelope was weaving for Laertes must have been patterned. Pattern weaving is time consuming. Second, the weave structure for such a ceremonial cloth was probably either tapestry or supplementary weft. Third, pictorial cloth was typically arranged in horizontal bands of figures or friezes. Fourth, from comparative evidence, it is entirely credible that culturally important narratives would be recorded in a ceremonial cloth, as Helen in Iliad III 125–128 is said to weave the story of the Trojan War. And fifth, the weaving of such an icon is a quasi-priestly office and would, in Homeric times, necessarily be in the hands of an elite woman rather than a band of slaves. [91] I cannot do better here than to refer the reader to Barber’s chapter on pattern weaving. I will add some brief observations regarding three weave structures: tapestry, supplementary weft, and double cloth. I will conclude with a discussion of the education of weavers that I believe adds strength to Nagy’s notion of “a parallelism between poetry and fabric-work as prime media of mythmaking.” [92]


‘Tapestry’ is a term often used for textiles that are not, in fact, tapestry-woven in a technical sense. The Bayeux Tapestry, which is embroidered, is an example. Another is a set of weavings found in Överhogdal Sweden in 1910 (see Figure 38, above, and the discussion below under ‘Supplementary Weft’). True tapestry weave, as explained above, is a weft-faced plain weave in which wefts weave back and forth, each in its own area of color. In modern tapestry-woven cloth, the entire cloth is tapestry-woven, so that the warp does not show anywhere (at least on a finished piece). Coptic tapestries, woven in early Christian Egypt, however, were woven as medallions or decorations within plain-weave clothing. It is entirely possible that there was a long tradition in Greece of partially tapestry-woven clothing.
Although virtually any image can be woven in tapestry, tapestry lends itself well to horizontals and diagonals. Since each weft is turning in its own area, a vertical edge of any figure within an image will create a slit, unless that slit is addressed by one technique or another. For example, the edge might be sewn, either in the course of weaving or afterward, or the weft threads of adjoining figures might both turn around the same warp thread, creating a kind of muffling of the clarity of the edge (see Figure 39). In a small area, slits are inconsequential and can be ignored. Thus tapestry is very suitable for small figures (like those in the weaving on the Chiusi Loom shown in Figure 10) or for figures based on diagonals. It is not as suitable for the squared meander patterns that figure so prominently on funerary urns of the Geometric period.
Tapestry can be woven so that it is almost perfectly reversible (although the great pictorial tapestries of Medieval and Renaissance times were not woven that way and it is uncommon to do so today).

Supplementary Weft

Supplementary weft techniques can easily be mistaken for embroidery because, unlike tapestry weave, if the wefts that create the image are removed, a cloth remains. Typically in the weaving process a shot or two of a ground weft (usually identical to the warp thread) is put in, followed by a shot of the supplementary weft, which may take any of a number of different forms. The so-called Överhogdal Tapestries, which date from the Viking Age (800–1100 C.E.) and include Norse as well as Christian imagery, are a good example. [93] These works are particularly interesting for this discussion because they were woven on a warp-weighted loom (see Figure 40). The weave structure is a form of soumak, a supplementary weft technique that involves looping the colored weft around small numbers of warp threads. The Överhogdal Tapestries were woven in fairly narrow bands (with the images on their sides, from the weaver’s point of view) and then sewn together—a practice that might millennia earlier have given rise to the frieze organization seen on Greek vase paintings and some of the textiles pictured on them. Greek looms were wider, however, and the frieze-like bands may always have been woven continuously, with the figures upright from the weaver’s point of view. When fabric is rolled on a cloth beam as it is woven, only a portion of the weaving is visible at any one time. The ‘story’ would naturally run across the cloth in bands, and each band would be visible to the weaver in its entirety during the weaving process.
There are many different types of supplementary weft and they can easily be combined in the same fabric (see Figure 41). A characteristic of some supplementary weft techniques is a constant busyness within the design, the result of problems of snagging if the weft floats too far on the back (or the front) of the cloth. Barber suggests that the images on the large Attic funerary urns of the Geometric Period are modeled on textiles woven in supplementary weft techniques. [94] (A good example is a krater shown on the website of the Metropolitan Museum: Textiles woven in supplementary weft techniques are sometimes reversible, but usually the patterns are illegible on the back side.

Double Weave

The cloth into which Helen weaves the story of the Trojan War and that into which Andromache sprinkles flowers as charms are both called diplax. [95] Nagy translates diplax as ‘fabric that folds in two.’ [96] He makes clear that he means a fabric that folds once; a fabric with two layers. [97] Without taking issue with his overall argument, I would like to add that, to a weaver, ‘a fabric that has two layers’ signifies a weave structure called double weave. [98] In a double-woven cloth, two layers of cloth are woven simultaneously, one layer behind the other, from the weaver’s point of view. Depending on how the weft or wefts travel at the edges of the cloth, the two layers can, when taken off the loom, be two entirely separate cloths (though it would generally be easier to weave each cloth separately), or they can be attached at one or both sides. If they are attached at one side, a single layer of cloth can be woven that is twice as wide as the loom it is woven on. In this case the edge where the two layers met would become a natural fold in the center of the cloth. (Because of technical difficulties of double weave, it is highly unlikely that such a double-wide cloth would be pattern-woven, unless in simple stripes or plaid.) If the two layers are attached at both sides, the resulting two-layered cloth can be both light and warm and thus quite luxurious. It can also be a different color on each side, which may be what is portrayed in a number of vase paintings of the archaic period. Certain figures wear cloaks that are formally folded over a shoulder or an arm, displaying the underside of the fabric, which appears to be a different color from the side worn facing outwards. [99] However, the aspect of double weave most relevant here is that, at any point in a double cloth, the two layers can change places with each other, interlocking the layers and offering a fine way to form patterns in two colors (see Figure 42).
With four sheds, two plain-weave layers can be woven. Thus any loom on which one can weave two-two twill can be used for double-weave. There are four sets of warp threads, working in pairs, and typically the two pairs are of contrasting colors. [100] Figures in the patterning can be picked up and woven so that they become part of the opposite side of the fabric. Unlike tapestry and supplementary weft, double weave offers no impediment to weaving vertical lines or large blocks of color. A tunic, woven probably in highland Peru in the late 15th to early 16th century C.E., shows the complexity of which double weave is capable. [101] The part of the piece that is tapestry-woven is a good demonstration of the way in which tapestry weave favors the diagonal (see Figure 43).

The Education of Weavers and Formulaic Narration

Weaving traditions evolve as the result of countless repetitions and countless innovations at the hands of countless weavers and ancillary workers. Over time, the fit between methods and materials is gradually refined, along with the fit between practical needs, aesthetic expressions, and agendas for communication. In the case of wool cloth, the type of wool, the way it is treated before spinning and the way it is spun all affect how it can be woven and how the finished cloth will feel. The case is similar with any other fiber. Many heading bands must be made and the warp on them woven off before it is known what the optimal ratio is between the thickness of the heading band warp and the thickness of its weft (the warp of the eventual cloth). That ratio affects the spacing of the warps, which determines how warp and weft interact in the cloth to be woven and whether the edges of the cloth draw in or maintain an even spread. The optimal heaviness and distribution of the loom weights can also be discovered only through countless efforts of trial and error. The final result of the weaver’s effort, especially with wool, cannot be seen until the cloth is removed from the loom and goes through a fulling process. Moreover, design vocabularies develop along with weave structures. All of this traditional knowledge has to be absorbed for a weaver to have proficiency in her craft. In addition, manipulation of weave structures to produce patterns is a complex process, requiring fairly sophisticated mathematical skills as well as manual dexterity.
Homeric weaving, that is weaving as portrayed in the Homeric epics, is a reflection of a very old and highly developed tradition. In book six of the Iliad, Hector asks his mother, the queen, to make a particular offering to Athena, one designed to turn around the fortunes of the war and save the city of Troy. He asks her to choose whichever of her robes is dearest to her.
τῶν ἕν’ ἀειραμένη Ἑκάβη φέρε δῶρον Ἀθηνῃ,
ὃς κάλλιστος ἔην ποικίλμασιν ἠδὲ μέγιστος,
ἁστὴρ δ’ ὣς ἀπέλαμπεν· ἔκειτο δὲ νείατος ἄλλων.
Of these Hecabe took one, and brought it as an offering for Athene, the one that was fairest in its designs and amplest, and shone like a star, and lay beneath the rest.
Iliad VI 293–295
In a bow to a great weaving tradition, the poem mentions that Hecabe’s most precious textiles were made by Sidonian women. Textiles are among the valuable gifts given to Odysseus in his travels, especially the gifts from the Phaeacians. While we can only imagine these superb textiles, it is just possible that some of them survived into the Archaic Greek period in which the Homeric poems were taking the shape in which we know them. [102] At any rate, the high Homeric valuation of textiles reflects historical reality in the eastern Mediterranean in the Bronze Age. [103] In Minoan and Mycenaean times, textile production was highly organized and controlled. This was a textile tradition not unlike that of highlands of South America in pre-Columbian times. Many more textiles survive to us from that tradition, offering some points of comparison. Many pre-Columbian textiles are fine beyond belief, as if indeed made by gods or heroes (if one measures the limit of skill for humans by that of humans living today). Although the palace organization and wealth are long gone, there are weavers even now in the high Andes who have inherited remnants of that weaving skill and sophistication.
Anyone in an industrial society now who learns hand weaving learns techniques. The same was true of the tapestry weavers of late Medieval and Renaissance Europe. They learned, for example, how to weave diagonals, how to weave circles, how to shade one color into the next—all in order to reproduce painterly images that they would read from a drawing or diagram prepared by a designer, commonly called a cartoon. It did not matter whether or not the whole image was visible to the weaver; it was only necessary to follow the cartoon. In societies that carry on indigenous traditions of figured weaving, however, weavers start at a very young age and learn patterns, or motifs. Clearly our Homeric weavers learned this way. [104] Childhood samplers, saved for their sentimental value, make an appearance in Greek tragedy. In Euripides Ion, for example, Ion shows Kreousa the weaving she had left in a casket with him when she abandoned him: [105]
Κρ. σκέψασθ’ ὃ παῖς ποτ’ οὖσ’ ὕφασμ’ ὕφην’ ἐγώ.
Ιων ποῖόν τι; πολλὰ παρθένων ὑφάσματα.
Κρ. οὐ τέλεον, οἷον δ’ ἐκδίδαγμα κερκίδος.
Ιων μορφὴν ἔχον τίν’; ὥς με μὴ ταύτῃ λάβῃς.
Κρ. Γοργὼν μὲν ἐν μέσοισιν ἠτρίοις πέπλων.
Ιων ὧ Ζεῦ, τίς ἡμᾶς ἐκκυνηγετεῖ πότμος;
Κρ. κεκρασπέδωται δ’ ὄφεσιν αἰγίδος τρόπον.
Ιων ἰδού.
τόδ’ ἔσθ’ ὕφασμα †θέσφαθ’ ὡς εὑρίσκομεν†
Κρ. ὧ χρόνιον ἱστῶν παρθένευμα τῶν ἐμῶν.

Kr. Look for a piece of weaving which I did while just a child
Ion What sort of weaving? Young girls do lots of weaving.
Kr. It is incomplete, like a sampler from the pin-beater.
Ion What figure does it have on it? I ask so you don’t take me in with this.
Kr. There is a Gorgon in the center threads of the material.
Ion O Zeus, what destiny seeks me out like a hunter?
Kr. It is edged with snakes in the manner of the aegis.
Ion Look! Here is the piece of weaving †how we discover oracles!†
Kr. Ah, girlish work of my loom seen after so long.
Ion 1417–1425 [106]
Kreousa notes that she did this weaving when she was a child, and that it was “like a sampler from the pin beater.” In this sampler, she was practicing the weaving of a gorgon, filling out the edges with snakes—in other words, one of a number of traditional figures for a weaver. Presumably, she would have woven others in the same way. Once the traditional formulaic elements are learned, they can be combined within traditional frameworks. A look at how traditional weavers in highland Bolivia learn their craft may help in understanding the Homeric weaver.
In the late 1990s, Gary Urton apprenticed himself to skilled weavers in a Bolivian village as part of a study of mathematical knowledge in traditional communities. [107] He chose weavers as his subject because weaving necessarily involves not only counting, but other mental manipulations of numbers and shapes. In order to better understand what they were doing, he chose to learn something of their craft, which he found to be a humbling process. The weavers in this community were all women and, after they got over their hilarity that a man should want to learn to weave, they were gracious teachers. His principal teacher was a 12 year-old girl, whose explanations he could more readily grasp because she was still very much a learner herself. After much effort, Urton managed to produce the simplest traditional pattern, such as a young child in the community might make. As he reports, these children observe the weaving process from the time they are born. When very small, they play at weaving, and by the time they are four or five years old they can weave the simplest patterns in narrow bands. The younger children learn by imitating the older children and competing with each other. Girls begin more formal training in early adolescence.
Most of the operations involved in weaving geometrical, animal, and other patterns are learned by young girls as numerical formulas, or “recipes,” passed on from older, more experienced weavers to younger ones. By practicing these recipes, they gradually are able to reproduce them, not as a matter of conscious manipulation, but of habit—of the smooth flow of body movements and the feel of proper rhythms. [108]
Some but not all of the women in the community eventually become master weavers, who are respected as community leaders. Urton describes their skill:
Since most textiles bear a number of different design elements arranged next to each other across the width of the warp threads of a fabric, the weaver must know what set of threads must be selected for each of the several designs of a fabric with each pass of the weft thread. While some of these “picking sequences” repeat after as few as six or eight passes of the weft, on more complex designs it takes some twenty or thirty (or more) passes of the weft to begin to repeat a design. All of the information on the proper picking sequence for each design of a fabric is held in the head of the weaver. However, a master weaver…generally does not consciously repeat to herself, or become aware of, this information as she weaves. Weaving is usually performed extremely quickly by the master weaver. Her movements are performed automatically, without counting to herself, according to hand and body routines of movement that become habitual through long training. [109]
In explaining this ‘automatic movement’, Urton compares the unconscious movements of a skilled typist.
Women in this community weave on back-strap looms and the weaving is warp-faced. Warping is a particular skill and must be adapted for each particular weaving. Often a master weaver is called in to do the warping for another weaver’s project. A master weaver holds that rank…
…not just because she is proficient at counting threads, but also because she combines that technical skill with aesthetic considerations based on producing pleasing and compelling arrangements of threads through the manipulation of color and space. The fact that the spatial arrangements are products of different thread counts provides the link between precise knowledge and aesthetics that is the hallmark of a master weaver. [110]
By analogy, we can say that Homeric weavers—Penelope, Helen, Andromache, Arete, as well as their divine counterparts Calypso and Circe—were master weavers by virtue of extensive training, long experience, keen intellect, and fine taste. Like Athene herself, they were embodiments of a kind of skill that was synonymous with intelligence. When Menelaos and Odysseus came to Troy on an embassy to retrieve Helen, they began to “weave a web of words and stratagems” to get her back. [111] The metaphor does not mean that their plans are like a web in which something can be caught. Rather, they had an outcome in mind and sought to create an intricate set of conditions from which it would follow if skillful steps were taken.
We may now be in a better position to understand from the other side (so to speak) the weaving metaphor that Nagy has explicated so fully. In discussion of the Panathenaic festival, he argues “that epic narration is visualized not only generally as the craft of weaving but also specifically as the specialized craft of pattern-weaving.” [112] The comparison is not between the humble craft of producing cloth and the great epics. In this metaphor, epic poetry gains luster from comparison with a craft that was its counterpart as a bearer of history and tradition. Weaving is the craft and activity of Athene. [113] It requires her intelligence. It carries her prestige.
Nagy shows that Homeric pattern weaving is like Homeric poetry in being essentially narrative. [114] Urton’s description of how pattern weaving is learned in highland Bolivia suggests that pattern weaving is also like epic narration in being formulaic. When Helen wove the battles of the Trojans and Achaeans into a robe, she would have been using a particular vocabulary of woven images that she had at her command. [115] Like words or phrases ‘fitting into’ the hexameter line, the figures that weavers learned to weave would have been ‘formulas’ that must fit into a certain pattern of threads. As Urton describes, the weaver would know in her body the operations required to cause a certain image to emerge after a certain number of passes of the weft. [116] And in her artistry, she would know how to combine them into a narration. Each narration, within a living tradition, would be unique even though made of ‘formulaic’ elements. [117]


There is much that can never be known about weaving in Greek antiquity. Nevertheless, an overall view of what is known, and an acquaintance with the basic mechanical operations of weaving, can perhaps enliven the experience of readers of ancient Greek when they run across references to weaving or weaving metaphors. These I have tried to provide. Above all, I offer this essay in the hope that it will help readers more fully appreciate the metaphorical connection between weaving and epic poetry that Gregory Nagy has explored in such illuminating depth. [118]

List of Figures

Figure 1: Three Examples of Plain Weave
Figure 2: Tapestry Weave
Figure 3: Supplementary Weft
Figure 4: Shedding Basics (1)
Figure 5: Shedding Basics (2)
Figure 6: Example and Diagram of 2/2 Straight Twill
Figure 7: The Cypriot Loom
Figure 8: The Corinthian Looms
Figure 9: The Metropolitan Museum Loom
Figure 10: The Chiusi Loom
Figure 11: The Pisticci Loom
Figure 12: The Attic Loom
Figure 13: The British Museum Loom
Figure 14: The Sackler Museum Loom
Figure 15: The Mississippi Loom
Figure 16: The Ashmolean Loom
Figure 17: The Thessalian Loom
Figure 18: A Rigid Heddle Shedding Device
Figure 19: A Heading Band under Construction
Figure 20: Detail of a Heading Band
Figure 21: The Northern European and the Ancient Greek Warp-Weighted Looms
Figure 22: Comparative Diagrams of the Warp-Weighted Loom
Figure 23: Northern European Model (Loom Set-up)
Figure 24: Northern European Model (Shedding)
Figure 25: High-Warp Tapestry Loom
Figure 26: High-Warp Tapestry Loom (Detail)
Figure 27: Stationary Heddle Bar (Natural Shed)
Figure 28: Stationary Heddle Bar (Counter Shed)
Figure 29: An Inkle Loom with Stationary Heddles
Figure 30: Inkle-type Shedding on a Warp-Weighted Loom
Figure 31: Alternately Looped Heddles
Figure 32: Shedding with Alternately Looped Heddles
Figure 33: Spinner from the Metropolitan Museum Loom Vase Painting
Figure 34: Shuttle on a Modern Horizontal Loom
Figure 35: Reed and Beater on a Modern Horizontal Loom
Figure 36: Actions of the Pin Beater
Figure 37: Action of Pin Beater or Sword
Figure 38: Pattern Weaving on a Warp-Weighted Loom (1)
Figure 39: Tapestry Sample Woven on a Warp-Weighted Loom
Figure 40: Pattern Weaving on a Warp-Weighted Loom (2)
Figure 41: A Supplementary Weft Sampler Woven on a Warp-Weighted Loom
Figure 42: A Heading Band and Double-Weave Sampler Woven on a Warp-Weighted Loom
Figure 43: Andean Tunic

Appendix: Sources and notes for Figures 7–17

Note: Pictures of nine of the eleven loom images listed below appear in Flores 2006, which is available on-line, at no cost, at The pages are not numbered, but the images can easily be found by scrolling through. All but three are together, more or less in the middle of the text; the other three are at the end. Those not included are the Cypriot Loom and the Thessalian Loom.

Figure 7: The Cypriot Loom

The original image is on a dish from Cyprus (late 8th century BCE) now in the collection of the Akademishes Kunstmuseum of the University of Bonn. It was first published in Aspris 1996. Other bibliography is in Smith 2009: 86 n25. A small image of the dish is displayed on the website of the Akademishes Kunstmuseum (
The image on the dish includes two large birds perched on the left of the loom, a rectangular object (identified by Aspris as a type of platter [119] ) above the loom, and another rectangular object (ladder-like, with triangles between the rungs of the ladder [120] ) adjacent to or attached to the right of the loom. Curiously, there are almost identical objects of this type depicted on the Attic geometric funerary urn at the Metropolitan Museum referenced in the section on supplementary weft (
This loom has one center cross beam and one row of loom weights. On the loom is an apparently finished weaving, though there seems to be nothing rolled on the cloth beam.

Figure 8: The Corinthian Looms

These two looms appear on an araballos from Corinth (ca. 600 BCE) published by Weinberg and Weinberg. [121] The drawing is by Dr. Marian Welker. The drawing is also shown at
The vessel is very small, the painting not in perfect condition and the details unclear. Nothing useful could come of redrawing it. The scene is taken to represent the rivalry between Arachne and Athene. There are two looms shown. In neither is it possible to distinguish cloth rolled on the upper beam. The cloth on both looms appears as cross-hatches, presumably representing some kind of patterning. Both looms have two center cross-beams and a curious array of loom weights. Some, but certainly not all of the loom weights appear to be attached to more than one warp thread—not in bunches but separated, as if they might rock with a change of shed. Three large objects under the loom on the left appear not to be weights, or at least not functioning weights. Some of the weights on that loom look horizontal. Could they be crescent-shaped loom weights? [122] Some loom weights were fitted with metal or wood rods, which passed through a hole in the top. [123] Although such rods might be represented here, it seems unlikely. [124]
One of the figures working at the loom has a stick in each hand; the other may be holding a two-pronged object in each hand or only in her right and something else in her left. It is hard to determine where her arm ends and the thing she is holding begins.

Figure 9: The Metropolitan Museum Loom

The loom is on a small lekythos (ca. 560 BCE) in the collection of the Metropolitan Museum of Art. The image on this vessel shows women engaged in various stages of cloth production from weighing out wool, preparing it and spinning it, to weaving and folding up a finished cloth. An image of the vessel, showing the loom, is on the museum’s web site:
The loom has one center cross-beam and one row of loom weights. Just above the center cross-beam are two light rods, one extending from the left-most edge of the warp to the center of the loom, and the other from the right-most edge of the warp to the center. The weaver on the right appears to be holding this stick (rather than the weft bobbin just above it). The weaver on the left holds a stick, with which she is apparently beating in the weft. Other weft bobbins rest in the crisscrossed lashing at the junction of the uprights and the center crossbeam. Another stick is held in the lashing where the upright meets the cloth beam on the right.
This vase painting seems very precise in its rendering of the loom. Hoffmann objects that the beams are impossibly slender and that “nothing prevents the beam from rotating; in practice, this would result in the weights dragging the cloth down and unwinding it;” and that “the crosses do not indicate a realistic means of fixing the beam to the uprights; the beam can be rotated, and it would require a support more substantial than a crossed rope.” [125] I would answer that, while most of the Scandinavian looms are made of hefty logs, one Lappish loom pictured in Hoffmann’s own book is made of far more slender beams than that on the Metropolitan Museum lekythos. [126] Moreover, if the uprights and cloth beam are notched (like Lincoln Logs) so that they fit together, (1) the cloth beam will not rotate under the pull of gravity; (2) the cloth beam will be supported by the notches in the uprights and the crisscross lashing will easily keep it in place; and (3) the loom-weights would have to be removed and tied on again when the cloth was rolled on the beam and with that done, two women (perhaps standing on stools) could easily untie the beam, roll up the cloth, rejoin the notches and retie it. Alternatively, the loom could be laid down for this procedure and stood back up again for the weights to be tied on. I have constructed a loom with such notches and cross-ties. The beams did not slip out of place, the cloth beam did not bend, and I was able to roll up the cloth by sliding the loom into a horizontal position.
Weights of just this shape have been found—even one (now in the British Museum) with its metal ring still intact. [127]

Figure 10: The Chiusi Loom

The original image is on a Greek skyphos (ca. 430 BCE), sometimes called “the Penelope vase,” found at Chiusi, Italy, and now in that city’s archaeological museum. An image of the vase can be seen at the following link: A drawing of the entire image with figures of Penelope and Telemachus is on the Perseus Project website:
This loom has two center cross-beams as well as an additional beam across the top, above the cloth beam, and another somewhat below the cloth beam. The purpose of the latter is a mystery to me. The top beam seems to have holes in which weft bobbins can be placed to be ready at hand—a feature any tapestry weaver would immediately appreciate. The joinery of uprights and crossbeams is marked with a square, which Hoffmann says “must be purely ornamental.” [128] If, like the Metropolitan Museum loom, however, the beams were notched Lincoln log fashion, the horizontal lines of the squares would be the edge of the notch in the upright and any gap between it and the cross beam; the vertical lines of the squares would be lashing running vertically around the crossbeam in front and horizontally around the upright in back—a more ornamental choice than the crisscross lashing on the Metropolitan Museum loom, in keeping with this elegant loom. A line midway between the fell of the cloth and the upper cross beam may depict a light rod like those on the Metropolitan Museum loom.
The disposition of the pattern weaving is unlikely to be based on observation, but this is after all a representation of a legendary time. The patterned portion rolled on the beam would be the reverse of the side toward the weaver as she weaves and seems to have no relationship, as far as textile design goes, to the frieze of winged creatures.
The loom weights are not visible in photographs of the vase itself, but apparently are clearly there as drawn, though difficult to see straight-on.

Figure 11: The Pisticci Loom

The image is from a Greek vase (5th century BCE) found at Pisticci, Italy. The drawing is after Quagliati 1904: 199 figure 4. The entire drawing can be seen in Flores 2006 as well as in Barber 1991:111.
This side-view of a loom is unique. This loom is at an angle to the floor and is perhaps supported by the ceiling, as well as a strut that runs out at an angle from the upright midway between the cloth beam and the center cross beam or beams. This image is also unique in representing the warp threads as a mass. Whether there are two center cross-beams or one is not clear from the drawing, although the width of the warp mass suggests there may be. The more or less trapezoidal shapes at either side of the mass of warps apparently represent two rows of loom weights. It seems highly unlikely that the cloth beam could be suspended from the uprights as shown.

Figure 12: The Attic Loom

The original is part of a domestic scene on a hydria (ca. 450-404 BCE) now in the collection of the Sackler Museum, Harvard University. An image on the museum’s website shows the vase, with part of the loom showing: More information on the vase and parts of the loom image are available on the Perseus Project website at
This loom has one center crossbeam. Although they do not show in the photograph or the drawing, there is apparently one row of loom weights. The loom is supported by a strut similar to that in the Pisticci Loom. The cross beams are lashed to the uprights in a crisscross configuration as on the Metropolitan Museum Loom. Some of the woven cloth is rolled on the cloth beam, and the patterning on the border shown on the front of the cloth does not show on the rolled cloth.

Figure 13: The British Museum Loom

This skyphos (like Figure 14, Figure 15, and Figure 17) is in the rather slap-dash Cabirion style. Circe and Odysseus are shown to the left of the loom and a boar is to the right of it. Cloth is rolled on the cloth beam. There are two center cross-beams and two rows of loom weights.

Figure 14: The Sackler Museum Loom

The original is on a Boeotian skyphos (late 5th century BCE) in the collection of the Sackler Museum at Harvard University. An image of the vessel is on the museum’s web site:
Like Figure 13, Figure 15, and Figure 16, a Cabirion-style vessel from Boeotia. A seated Penelope and two men are to the right of the loom. The essential features of the loom are the same: cloth rolled on the cloth beam, two center cross-beams, and two rows of loom weights. This loom also has brackets that support the loom from the rear. It is clear that we are looking at the front of the loom, since the cloth would hang on the rear side of the roll.

Figure 15: The Mississippi Loom

The original is on a scyphos (late 5th century BCE) in the collection of the University Art Museum, University of Mississippi. An image of the vessel is on the website of the Perseus Project:
Like Figure 13, Figure 14, and Figure 16, a Boeotian scyphos in the Cabirion style. Circe and Odysseus are to the right of the loom, which has the same configuration: rolled cloth on the cloth beam, two center cross-beams, and two rows of loom weights. Supporting brackets turn either outward or to the rear.

Figure 16: The Ashmolean Loom

From a painting on a scyphos (4th century BCE) in the collection of the Ashmolean Museum, Oxford University. An image of the vessel can be seen on the website for the museum’s print service:
Like the previous three figures, a vessel from Boeotia in the Cabirion style. Odysseus and Circe stand to the left of the loom, which like the other three has two center cross-beams and two rows of loom weights as well as cloth rolled on the cloth beam. This loom also has a weft bobbin held in the warps and the weft thread is visible, running from the fell of the cloth to the bobbin. The leaning of the loom to the side may be an attempt to convey its leaning towards the rear.

Figure 17: The Thessalian Loom

This loom appears in a relief on a marble stele from Thessaly (late 4th century BCE), in the collection of the National Archaeological Museum, Athens. A photograph of the stele is on the website of the Perseus Project:
The stele shows Odysseus being discovered by Eurykleia, while Penelope stands behind at her loom. The cloth beam shows rolled cloth, but no center cross-beam is visible. Penelope holds a weft bobbin in her right hand and her left hand is raised toward her weaving. The drawing shows an object in her left hand which may or may not actually be on the relief, though it appears to be in the photograph.


Aspris, M. 1996. “Ein zyprischer Teller mit der Darstellung eines Webstuhls.” Bonner Jahrbücher 1996:1–10.
Barber E.J.W. 1991. Prehistoric Textiles: The Development of Cloth in the Neolithic and Bronze Ages with Special Reference to the Aegean. Princeton, NJ.
Barber, E.J.W. and Adams, D.Q., s.v. “Textile Preparation” in Mallory and Adams 1997: 572.
Broudy, E. 1979. The Book of Looms. Hanover, NH.
Brown, R. 1995. The Weaving, Spinning, and Dyeing Book, ed. 2, New York.
Burke, B. 2010. From Minos to Midas: Ancient Cloth Production in the Aegean and in Anatolia. Ancient Textiles Series 7. Oxford.
Cadogan, G., Whitley, J., Kopaka, K., and Iacovou, M. eds. In press . Parallel Lives: Ancient Island Societies in Crete and Cyprus. Athens.
Carroll, D.L. 1965. Patterned Textiles in Greek Art: A Study of their Designs in Relationship to Real Textiles and to Local and Period Styles. Dissertation, University of California, Los Angeles.
Crowfoot, G. 1936-7. “Of the Warp-Weighted Loom.” Annual of the British School at Athens 37:36–47.
Davidson, G.R. 1952. Corinth: Results of Excavations Conducted by the American School of Classical Studies at Athens XII: The Minor Objects.
Davidson, G.R. and Thompson, D.B. 1943. Small Objects from the Pnyx I. Hesperia Supplement 7.
Edmunds, S.T. 1990. Homeric Nēpios. Harvard dissertations in classics. New York.
Flores, R.R. 2006. Reconstructing the Ancient Greek Warp Weighted Loom. Honors thesis for Graduation in the Mitte Honors Program, Texas State University.
Gow, A.S.F. 1965. Theocritus. Cambridge.
Hamilton, E. and Cairns, H. eds. 1961. The Collected Dialogues of Plato, Including the Letters. New York.
Heubeck, A., West, S., and Hainsworth, J.B. 1988. A Commentary on Homer’s Odyssey, Volume 1: Introduction and Books i-viii. Oxford.
Hoffman, M. 1964. The Warp-Weighted Loom: Studies in the History and Technology of an Ancient Implement. Oslo (Reprinted, 1974, in the United States by Robin and Russ Handweavers).
Jebb, R.C. 1965. Sophocles Part II. The Oedipus Coloneus. Amsterdam.
Karouzou, S. 1968. National Archaeological Museum Collection of Sculpture: A Catalogue. Athens.
Landercy, M. 1933. “La destination de la kerkis dans le tissage en Grèce aux IV siècle.” L’Antiquité classique II. Gent.
Lee, K.H. 1997. Euripides: Ion. Warminster.
Lloyd-Jones, H. 1970. The Libation Bearers by Aeschylus. Englewood Cliffs, New Jersey.
Mallory, J.P. and Adams, D.Q. eds., 1997. Encyclopedia of Indo-European Culture. London.
Mårtensson, L., Andersson, E., Nosch, M.-L., and Batzer, A. 2007. “Technical Reports in Experimental Archaeology” Part 3 (on loom weights) and Part 4 (on spools).
Michel, C. and Nosch, M.-L. eds. 2010. Textile Terminologies in the Ancient Near East and Mediterranean from the Third to the First Millenia BC. Oxford.
Murray, A.T. and Dimock, G.E. transl. 1995. Homer Odyssey, volumes I and II. The Loeb Classical Library, Cambridge, Mass.
Murray, A.T. and Wyatt, W.F. transl. 1999. Homer Iliad, volumes I and II. The Loeb Classical Library, Cambridge, Mass.
Nagy, G. 2002 (date of original publication). Plato's Rhapsody and Homer's Music: The Poetics of the Panathenaic Festival in Classical Athens. Online edition at
Nagy, G. 2008. Homer the Classic. On-line edition at Washington.
Nagy, G. 2009. Homer the Preclassic. On-line edition at Washington.
Oscarsson, U. 2010. De gåtfulla Överhogdalsbonaderna (The Enigmatic Överhogdal Tapestries). Jamtli, Sweden.
Quagliati, Q. 1904. “Pisticci: tombe lucane con ceramiche greche.” Notizie degli Scavi di Antichità I:196–208.
Robert, C. 1900. “Die Fusswaschung des Odysseus auf zwei Reliefs des fünften Jahrhunderts.” Athenische Mitteilungen 25: 325–38.
Rogers, P. W. 2007. Cloth and Clothing in Early Anglo-Saxon England: AD 450–700. York.
Smith, J.S. 2009. Art and Society in Cyprus from the Bronze Age into the Iron Age. Cambridge.
Smith, J.S. and Tzachili, I. “Cloth in Crete and Cyprus,” in Cadogan, G. et al. eds. In press.
Snyder, J. M. 1981. “The Web of Song: Weaving Imagery in Homer and the Lyric Poets.” Classical Journal 76.3:193–196.
Tuck, A. 2006. “Singing the Rug: Patterned Textiles and the Origins of Indo-European Metrical Poetry.” American Journal of Archaeology 110.
Urton, G. 1997. The Social Life of Numbers: A Quechua Ontology of Numbers and Philosophy of Arithmetic. With the collaboration of Primitivo Nina Llanos. Austin, Texas.
Wace, A.J.B., 1948. “Weaving or Embroidery?” American Journal of Archaeology, 52.1:51–55.
Weinberg, G.D. and Weinberg, S.S. 1956. “Arachne of Lydia at Corinth.” In Weinberg S.S. 1956: 262–267.
Weinberg, S.S., ed. 1956. The Aegean and the Near East; Studies Presented to Hetty Goldman. New York.
Wisti Lassen, A. 2007 “Weaving with Crescent Shaped Loom Weights” Lejre Forsøgscenters Forskninglegater til Kulturhistoriske Eksperimenter (Projekttitel: “Et Forsøg Med Bananformede Vævevægte” Journalnummer HAF 09/07) Copenhagen.


[ back ] 1. Nagy 2008:2§92.
[ back ] 2. I am responsible for all photographs and drawings unless otherwise noted.
[ back ] 3. I use the term ‘warp-weighted loom’ to refer to a type of vertical loom shown on Greek vase paintings, the characteristics of which are fully discussed in this essay. Almost any type of loom can be tensioned with weights, but only the type under discussion here is generally called a ‘warp-weighted loom’.
[ back ] 4. Iliad VI 490–493.
[ back ] 5. Iliad III 125–128.
[ back ] 6. Iliad XXII 440–441. See Nagy 2009 II§376.
[ back ] 7. Odyssey ii 93–110 and xix137–144. The Homeric text does not say explicitly that Penelope’s weaving is patterned, but that it is can be assumed because of the length of time it takes.
[ back ] 8. Smith and Tzachili (in press).
[ back ] 9. Burke 2010 gives a good summary.
[ back ] 10. Barber 1991:311–382.
[ back ] 11. Crowfoot 1936/1937.
[ back ] 12. Hoffmann 1964.
[ back ] 13. Barber 1991.
[ back ] 14. Technical reports with good photographs of warp-weighted looms in operation can be seen on the website of the University of Copenhagen under The Danish National Research Foundation’s Centre for Textile Research, http:/
[ back ] 15. Of particular interest here are Michel and Nosch 2010 and Burke 2010.
[ back ] 16. Weft is the term generally used by English-speaking weavers today. Woof is a less commonly used synonym.
[ back ] 17. A common synonym for plain weave is tabby.
[ back ] 18. True lace is made by a different process.
[ back ] 19. There can also be supplementary warp, complementary warp or weft, and various other methods of bringing extra threads into the weaving. Some of these methods are beyond the scope of this paper; others are discussed below.
[ back ] 20. For the purposes of this discussion, I use the feminine form of pronouns in referring to a weaver or weavers. There were male weavers in antiquity, as there are today, but the weaver of Homeric poetry is always female.
[ back ] 21. The warp threads for 2/2 twill can be numbered 1, 2, 3, 4, 1, 2, 3, 4, etc. For the first passage of the weft, a shed is formed by lifting all threads numbered 1 and 2; for the second weft passage, all threads 2 and 3 are lifted; for the third, all threads 3 and 4; and for the fourth, all threads 4 and 1.
[ back ] 22. Cf. Odyssey xix 139–140. In all translations of the Iliad and the Odyssey, I have relied on the Loeb Classical Library editions (Murray and Wyatt, transl., in the case of the Iliad and Murray and Dimock, transl., in the case of the Odyssey), modifying as necessary to emphasize particular points.
[ back ] 23. Barber 1991:387 and Appendix A, which is a table of archaeological findings of weights that appear to be loom weights.
[ back ] 24. Hoffmann 1964:18.
[ back ] 25. Hoffmann 1964:18.
[ back ] 26. This vertical loom is referred to as a ‘two-beam loom’. Warp-weighted looms sometimes have two beams, but only one of them is involved in tensioning the warp threads.
[ back ] 27. Barber 1991:254. See also all of Chapters 3 and 11 in Barber 1991 for a full discussion of these issues.
[ back ] 28. Barber 1991:113.
[ back ] 29. Hoffmann 1964:232–237.
[ back ] 30. Translation: Jebb 1965, with modifications.
[ back ] 31. For evidence of male weavers, see Plato Phaedo 87b–88c, also Hippias Minor 368c; possibly also Hesiod Works and Days 537–538.
[ back ] 32. Barber 1991:92n7 lists several other fragmentary pieces of evidence. (I was not able to acquire sufficiently good photographs of these fragments to make drawings, or even to judge whether they offered useful information.) A different device that appears in vase paintings is sometimes called a hand loom. For example, a search of the Perseus Digital Library ( for ‘loom’ turns up a vase painting in which a woman holds an object apparently made of sticks bound together in a triangular shape. Within that triangle is a roughly trapezoidal frame on which there is something that might be cloth. Though this object is described as a ‘hand loom’, it is most likely a frame for a type of netting called sprang. See Barber 1991:122–124; Michel and Nosch 2010: Figure 3.17d and caption.
[ back ] 33. A coarsely woven rug would have, at a minimum, five or six warp threads per inch; any cloth that would shine like a star would probably have had 20, 30, or more warp threads per inch.
[ back ] 34. There are actually two vases in the Sackler Museum at Harvard University that portray looms. The one I call ‘the Attic loom’ is the other.
[ back ] 35. For the size of cloth woven on the warp-weighted loom in Greek antiquity, see Barber 1991:103–106 and 363.
[ back ] 36. ‘Cloth beam’ is standard terminology for the beam on any loom in which the cloth is wound during the weaving process.
[ back ] 37. For a clear articulation of such standards, see Mårtensson et al. 2005/2006:3.
[ back ] 38. See Barber 1991:116–118. A heading cord can be used instead of a heading band, in which case, the construction of the warp might be carried on differently, for example as described in Hoffmann 1964:127 and 132 Figure 6.
[ back ] 39. I know of no evidence of the use of rigid heddles in ancient Greece. They may have been known to the Romans and certainly were used throughout Europe by medieval times. They seem to have come to the Americas from Europe. They were commonly used for making heading bands for warp-weighted looms in Finland.
[ back ] 40. Such bands can be seen in some of the figures on the Parthenon frieze. See Barber 1991:271, 272 Figure12.4; 361 Figure 16.1.
[ back ] 41. Nagy 2002: Chapter 3 “The Greek word that we may translate as ‘heading band’ is diasma (cf. exastis ‘selvedge’). In Hesychius, diasma is glossed as phareos arkhê, that is, ‘the beginning of the fabric’. So also, I propose, the specially intricate humnos is the beginning of the rest of the song.”
[ back ] 42. The degree of uprightness has been a subject of much discussion, as if it could be settled for all warp-weighted looms in one stroke. I will mention various options in discussion of the images of looms from Greek antiquity.
[ back ] 43. See, for example, Iliad I 21.
[ back ] 44. Odyssey vii 61–62.
[ back ] 45. Hoffmann 1964:23–150. Hoffmann also discusses in detail the history of use of the warp-weighted loom in Europe.
[ back ] 46. Barber 1991:111 Figure 3.27 caption.
[ back ] 47. Barber 1991:270 Figure 12.3. The diagram of the loom is the same, but Greek words for the parts of the loom are given.
[ back ] 48. In this configuration of the loom, the warp threads in front of the shed bar must be attached to the bar by some kind of lashing (as shown in Barber’s diagram) and the warp threads behind the bar secured to it with a lacing cord (as shown in Hoffmann’s diagram of the northern European loom). Otherwise, it is difficult for the weaver to maintain the evenness of the cloth.
[ back ] 49. Barber 1991:110.
[ back ] 50. In the loom on the right, there are what look like horizontal rods above most of the loom weights, which resemble the bobbins used to store extra warp in several archaeological experiments. (Mårtensson et al., Wisti Lassen 2007). In the loom in Figure 24, the photograph does not clearly show that there are two rows of loom weights because the two rows are hung at about the same height. It may have been the practice in antiquity to hang the front loom weights at a different height from the rear ones, but it may also have been the practice of painters to show clearly that there were two rows of loom weights by portraying them at different heights. The rather ragged single row of loom weights may only show that the painter of the Corinthian looms did not follow this convention.
[ back ] 51. Barber 1991:93–103, 389–390.
[ back ] 52. The weights on the Metropolitan Museum loom are a known type, with an iron ring at the top to which the warp threads are attached. There is an example in the British Museum (see Broudy 1979: Figure 2-8).
[ back ] 53. Crowfoot, 1936–1937:43.
[ back ] 54. Coptic tapestry weavings, for example, which survive in abundance, are typically embedded in cloth woven in a balanced plain weave.
[ back ] 55. In the large loom shown in Figure 25 and Figure 26, the heddle bar is stationary within the weaving process, even though it is moveable to bring it into the weaver’s reach when needed. In smaller Gobelin-type looms, the heddle bar is one rod extending across the entire loom and fixed at both ends.
[ back ] 56. Brown 1995:70–71, 80–81.
[ back ] 57. Barber 1991:167.
[ back ] 58. Barber 1991:313–314, 370n; Carroll 1965:11.
[ back ] 59. Hoffmann 1964:42–43, 45 Figure 12.
[ back ] 60. A counterargument could be made that the mechanics of the Northern European warp-weighted loom, which involve the lifting of a heavy rod and a set of weights, is a laborious method of recreating the mechanics of a horizontal loom with harnesses, which was well-known in Europe from Medieval times onward.
[ back ] 61. Barber 1991:204–205n6.
[ back ] 62. Mårtensson et al. 2007; Wisti Lassen 2007 and forthcoming.
[ back ] 63. My drawing of the bobbin shows it caught between the warps, like the bobbin in the Ashmolean loom (see Figure 16). In the vase painting, it appears to be floating with nothing really holding it up.
[ back ] 64. A ceramic spindle whorl will stay on an appropriately shaped wooden spindle simply by being pushed tightly on to it, and it can be pulled off when no longer needed. The convenience to the weaver of not having to rewind bobbins is great.
[ back ] 65. Plato Statesman 281e.
[ back ] 66. Barber 1991:279–280.
[ back ] 67. Barber, E.J.W. and Adams, D.Q., s.v. “Textile Preparation” in Mallory and Adams 1997:572.
[ back ] 68. Barber 1991:273. She cites here also the English cognate ‘reel’.
[ back ] 69. There are, as might be expected, many types of shuttles, often fashioned from whatever is at hand. Weavers on back-strap looms often wind weft around a slender stick in a figure eight fashion so that it forms a rather long, sausage-shaped mass.
[ back ] 70. Landercy, Crowfoot, and Barber agree on identifying the kerkis as pin beater (Landercy 1933; Crowfoot 1936/1937:44–46; Barber 1991:273–274).
[ back ] 71. Plato Cratylus 389b–c: “And whatever kerkides are wanted, for the manufacture of garments, thin or thick, of flaxen, woolen, or other material, ought all of them to have the true form of the kerkis, and whatever is the kerkis best adapted to each kind of work, that ought to be the form which the maker produces in each case?...[the maker must know how to express the natural form in the appropriate material, for example] how to put into wood forms of kerkides adapted by nature to their uses.” Translation by Jowett, B. in Hamilton and Cairns 1961.
[ back ] 72. Hoffmann 1964:140, Figure 61.
[ back ] 73. Hoffmann 1964:116, Figure 53(6).
[ back ] 74. Crowfoot 1936–1937:44.
[ back ] 75. Odyssey v.62.
[ back ] 76. Translation by Jowett, B. in Hamilton and Cairns 1961.
[ back ] 77. Euripides Ion 1419.
[ back ] 78. Theocritus 18.33–34.
[ back ] 79. For a full discussion of nēpios and its implications in the Iliad and Odyssey, see Edmunds 1990. On Andromache as mistress of her own workshop, see Iliad VI 490–492.
[ back ] 80. Hoffmann 1964:126. ‘Pin beater’ here translates hræll, which as seen above, is cognate with kerkis.
[ back ] 81. Plato Statesman 282b.
[ back ] 82. Aristotle Physics 243b11.
[ back ] 83. Translation Lloyd-Jones 1970, with modifications.
[ back ] 84. As any beginning weaver has had to learn, even a reed does not completely save the cloth from becoming progressively narrower as the weaving progresses. The weaver must learn how to space the thread and make adjustments for each new set of materials used.
[ back ] 85. See Aristophanes Frogs 231–232, Euripides Iphigenia in Tauris 222–224.
[ back ] 86. Plato Cratylus 389b–c.
[ back ] 87. Nagy 2008ch4, 2009ch10.
[ back ] 88. As Wace pointed out (Wace 1948) the text of Homer alone is enough to establish the fact. See also Barber 1991:359n2.
[ back ] 89. Barber 1991:314.
[ back ] 90. Carroll 1965.
[ back ] 91. Barber 1991:358–382.
[ back ] 92. Nagy 2002: Chapter 3.
[ back ] 93. Pictures of the images on the ‘tapestries’ can be seen at:
[ back ] 94. Barber 1991:366.
[ back ] 95. Iliad III 125–127 and XXII.440–441, respectively.
[ back ] 96. Nagy 2009:II§373.
[ back ] 97. Nagy 2009:II§386.
[ back ] 98. The terms ‘double weave’ and ‘double cloth’ are used synonymously. ‘True double cloth’ is something different, as is ‘double-faced cloth,’ neither of which will be discussed here.
[ back ] 99. See, for example, the second figure from the left on an amphora now in the Museum of Fine Arts Boston (accession n. 98.918)
[ back ] 100. Triple weave is possible, or even more layers, but unlikely on a warp-weighted loom.
[ back ] 101. Unfortunately, the colors of the two layers are (or, more likely, have become) quite close, making the pattern somewhat difficult to see.
[ back ] 102. Barber 1991:280–282.
[ back ] 103. Smith and Tzachili (in press); Burke 2010; Michel and Nosch, eds. 2010.
[ back ] 104. Production practices may have been different by the time of Theocritus. Remarks of Praxinoa (Theocritus 15.80–81) about certain figured textiles—ποῖαί σφ’ ἐπόνασαν ἔριθοι,/ποῖοι ζωογράφοι τἀκριβέα γράμματ’ ἔγραψαν (“what workers they must have been that made them, and what artists that drew the lines so true”)—suggest that the weavers and the designers were different from each other. Translation: Gow 1965.
[ back ] 105. For other examples, see Aeschylus Choephoroi 231–232, Euripides Iphigenia in Tauris 814–817, Ion 1489–1490.
[ back ] 106. Translation: Lee 1997 (modified).
[ back ] 107. Urton 1997.
[ back ] 108. Urton 1997:125.
[ back ] 109. Urton 1997:117–118.
[ back ] 110. Urton 1997:124.
[ back ] 111. Iliad III.212.
[ back ] 112. Nagy 2009:II§385.
[ back ] 113. See, for example, Hesiod Works and Days 64.
[ back ] 114. For example, Nagy 2009:II§384.
[ back ] 115. For a completely different type of woven image on the same theme, see the fifteenth century tapestry, now in the Metropolitan Museum in New York called “The Battle with the Sagittary and the Conference at Achilles' Tent (from Scenes from the Story of the Trojan War)” at the following link:*&when=A.D.+1400-1600&what=Textiles|Tapestries&pos=78#.
[ back ] 116. In discussing Homeric poetry and mnemonic chanting among some weavers, Anthony Tuck suggests what might have been an additional method for weavers to organize the knowledge required to produce certain patterns. The particular examples he gives are too simple to be convincing, but a more complex relationship between song and the process of weaving is not unknown among weavers. See Tuck 2006.
[ back ] 117. Bhakti Ziek studied traditional weaving in rural Guatemala in the 1960s. Each village had its own style of supplementary weft pattern weaving that was so distinctive it was possible to identify what village a woman was from by the woven patterns on her clothing. One young weaver, on being asked how it felt to her to be weaving designs just like her mother’s, replied indignantly that her weaving was not at all like her mother’s. (Bhakti Ziek, personal communication.)
[ back ] 118. In addition to Greg, who seems to have become interested in weaving about the same time I did, I wish to thank Deborah Holcomb, Michael Holcomb, Louise Wheatley, and Prudence Jones, with whom I have shared an interest in the warp-weighted loom; Liza Cleland, Deborah Holcomb, Agnete Wisti Lassen, and Joanna Smith for steering me toward very useful resources; Sarah Kass for her critique of my early attempts at constructing the figures; Lowell Edmunds, Bhakti Ziek, and Michael Holcomb for their very useful editorial comments; and Lowell Edmunds for expert advice on the conventions of scholarly works and for invaluable help in obtaining books and articles.
[ back ] 119. Aspris 1996.
[ back ] 120. Very similar objects appear in the Geometric krater at the Metropolitan Museum, referenced in the text in the section on supplementary weft. This krater can be seen at The picture must be magnified to see the objects mentioned.
[ back ] 121. Weinberg, G.D. and Weinberg S.S. 1956:263 Figure 1.
[ back ] 122. See Wisti Lassen 2007.
[ back ] 123. Davidson and Thompson 1943:68-69n58.
[ back ] 124. Davidson 1952. Images of weights fitted with rods occur as stamps on certain of the weights themselves (one of many types of images found stamped on loom weights). The rods look short and rather slight and were used for tying the warp to the weight.
[ back ] 125. Hoffmann 1964:308.
[ back ] 126. Hoffmann 1964:61 Figure 22.
[ back ] 127. Broudy 1974:28 Figure 2-8.
[ back ] 128. Hoffmann 1964:306.