Spindle Whorls

Spindle whorls, or malacates in both Spanish and Nahuatl, are perforated disks used in the production of spun fiber. Handspinning involves a wooden spindle and whorl, with the whorl acting as a flywheel to maintain inertia when the spindle is given a sharp twist. Spinning in pre-Columbian Mesoamerica was done using two principal methods: "drop-spinning," where the spindle and whorl were spun from a standing position, drawing and twisting the fiber as the spindle dropped to the ground; and "supported-spinning," where the tip of the spindle was spun in a small bowl. The two techniques were used to produce different quality thread, and were better suited for different fiber materials (S. McCafferty and G. McCafferty n.d.; Parsons 1972b; Smith and Hirth 1988; Voorhies 1989). In contemporary Mexico, maguey is usually spun using the drop-spinning method due to the relative coarseness of the raw fiber and its long staple length (Parsons and Parsons 1990). Cotton is usually spun using the supported-spinning method because of its short staple length. It should be noted, however, that maguey can be re-spun using supported spinning to refine the quality of the thread, and multiple threads may be plied together in a subsequent production step using the drop-spinning method.

Spindle whorls are the most common evidence for textile production in the archaeological record of pre-Columbian Mexico. They were usually made of baked clay, but occasionally occur made of stone or bone. Ethnographic studies have also described whorls of wood, rubber, and sun-dried clay (Garcia Valencia 1975; Hochberg 1980). Whorls are a fairly common artifact class from Postclassic sites in central Mexico, but are rare in earlier contexts, probably because they were made of non-durable materials. Although whorls have often been mentioned in site reports, detailed descriptions are seldomly presented. Jorge Enciso (1971) published illustrations of decorated whorls from private collections, with the designs classified by motif.

In the first rigorous study of spindle whorls, Mary H. Parsons (1972b) analyzed whorls found in surface collections from the Teotihuacan and Texcoco valleys in the Basin of Mexico. She used attributes of diameter, weight, and hole size to classify whorls into three main types, which were further divided into groups based on such characteristics as form and decorative techniques. The major distinction found was between large (Types 1 and 2) and small (Type 3) whorls, which Parsons interpreted as relating to the difference between the production of maguey and cotton thread, respectively.

The economic role of textile production has been discussed by Smith and Hirth (1988) for sites in western Morelos, where high frequencies of small whorls were found in a region known for its tribute payments to the Aztec state in spun cotton and woven garments. Elizabeth Brumfiel (1991) considered the relative frequencies of whorls from Early and Late Aztec period sites in the southeast Basin of Mexico from the perspective of changing strategies of domestic production in relation to tribute demands by the Aztec state.

As indicated, the difference between small and large whorls may relate to the material being spun, and therefore to factors of economics and availability. However, since the size distinction relates more directly to the different methods of spinning, the spatial distribution of whorls probably relates to the areas where specific types of spinning were practiced. Supported spinning was commonly practiced within the household compound, and therefore small whorls should be more likely to have been used and lost in that context. Drop-spinning was practiced in a wider range of spatial contexts, such as walking within the community or to and from the fields. Large whorls, then, should be found in extra-mural contexts (i.e., outside of the household compound).

Intentional discard of whorls also occurred with other classes of domestic refuse, and also as grave goods. According to ethnohistoric evidence, a woman's weaving tools were burned at the time of her death (Sahagún 1950-1982, Book 2:138). The presence of whole whorls in midden deposits may therefore indicate the death of a spinner, as potential evidence for the domestic life cycle. A mass burial found in San Andrés Cholula included a large number of spinning tools (whorls and spinning bowls), often as the only grave good associated with retainers of the central individual (Suárez 1990; McCafferty 1992b).

In addition to the functional role of spindle whorls in the production of spun fiber, ethnohistoric accounts also suggest a symbolic role in the construction of female identity (S. McCafferty and G. McCafferty 1991). Spindles and whorls were presented to baby girls during their bathing ceremony (Sahagún 1950-1982, Book 6:201). During a girl's education she was admonished to "pay heed to, apply yourself to, the work of women, to the spindle, the batten" (Sahagún 1950-1982, Book 6:96). The principle deities of the Mother Goddess complex (including Toci, Tlazolteotl and Xochiquetzal) were closely affiliated with spinning and weaving as symbolic of sexual reproduction and also female production (Sullivan 1982; S. McCafferty and G. McCafferty 1991). As diagnostic attributes of their costumes they wore spindles, whorls and spun fiber stuck into their hair (Figure 147). This trait was emblematic of affiliation with the Mother Goddess complex and resources of female power. The use of spindles in the headdress was shared as a pan-Mesoamerican trait of goddess impersonators/supernaturals in Aztec, Borgia Group, Mixtec, and Maya pictorial manuscripts. Interestingly, prestigious women from the Mixteca de la Costa weaving town of Jamiltepec, Oaxaca, are known today as "malacateras" (loosely translated as "spinners"), and they continue to wear spindles in their hair.

Figure 147. The goddess Tlazolteotl with spindles in her headdress (after Codex Borgia)

In summary, the research potential of spindle whorl analysis includes both their functional use in textile production and symbolic role in the construction of female gender identity. Since one of the criteria suggested by the ethnohistoric model for identifying pre-Columbian households was the presence of both male and female activities, evidence for textile production as a traditionally female activity provides potentially important information. Furthermore, the spatial distribution of whorls within the UA-1 site could potentially relate to female work space as well as inform on where different spinning techniques were practiced.

A total of 133 whorls (including fragments) were recovered at UA-1 (Figures 148-149). When comparing this total with other contemporary sites (e.g., Coxcatlán [Sisson 1973, 1974, n.d.] or Tetla-11 [Norr 1987]), this is a relatively high figure. A comparably high number of whorls (n=77) were recovered in the UA-79 excavations (Lind 1979a), dating to the Late Postclassic period, and the UA-79 whorls provide a useful comparison for the analysis of size, form, and decorative style (Limón n.d.; S. McCafferty and G. McCafferty 1991, n.d.).

A detailed analysis of the UA-1 whorls is being prepared for separate publication (S. McCafferty and G. McCafferty n.d.), but a catalogue of the whorls is presented in Appendix A. The present discussion summarizes some of the

Figure 148. Spindle whorls from UA-1

Figure 149. Spindle whorls from UA-1

results of that analysis, particularly as they relate to the functional context of the UA-1 structures.

Several indices were used to characterize the whorls, including diameter, height, weight, hole size, volume, and shape. This variety of dimensions is important because such characteristics as diameter and height affect the speed and duration of spin of the spindle. A wider, disc-shaped whorl will produce a slower spin for a longer period of time. A taller, bead-like whorl will produce a faster rotation that will not last as long. As an illustration of this principle, consider a figure skater revolving slowly with arms outstretched, but accelerating as he/she brings his/her arms in. The shape measurement, defined as the ratio of height to diameter, relates to these dual variables, with a lower value indicating a more disc-like whorl, while a higher measure indicates a more bead-like whorl.

The weight of a whorl is also considered a significant variable in determining if drop-spinning or supported-spinning were practiced. A whorl that is too light is impractical for drop-spinning, while one that is too heavy does not work well for supported-spinning. Whorls used for supported-spinning of cotton by contemporary hand-spinners weigh about 1 ounce (ca. 30 g) (Linder and Linder 1977), and are about the size of a walnut (Hochberg 1980). Curiously, both in size and weight these ethnographic whorls are significantly larger than the small "cotton" whorls identified by Parsons (1972b), suggesting that the small whorls may have had a more specialized function, perhaps in the respinning of cotton and/or maguey, or for other materials such as rabbit or human hair (see S. McCafferty and G. McCafferty n.d. for a discussion of materials spun in pre-Columbian Mexico).

The hole size measurement relates to the perforation in the center of the whorl, through which the spindle shaft would pass. Top and bottom hole sizes vary slightly, probably as a result of wear, but also provide evidence for how the whorl was situated on the shaft. This dimension relates in part to the size of the spindle, with a larger hole size relating to a thicker, and therefore probably longer shaft. But hole size can also reflect whorl position on the pointed shaft, as whorls with small holes could be fitted onto the tip of a large spindle.

Since the size of the hole detracts from the mass of a whorl, a separate measurement was created: volume. In this case, volume was estimated as the volume of a cylinder using whorl diameter and height, minus the volume of the cylinder defined by the hole size. This measure has several advantages over weight, particularly because it provides a measure comparable to weight that can be estimated even from fragmentary whorls.

The diameter of the UA-1 spindle whorls range from 14 to 74 mm, falling into three modes (Figure 150). The first

Figure 150. Diameter of UA-1 spindle whorls

mode, between 23-31 mm, corresponds with small whorls identified by Parsons (1972b), Norr (1987) and Sisson (n.d.). A second mode ranged from 32-37 mm, representing the greatest number of UA-1 whorls. Sisson (n.d.) described a weak second mode that averaged 34 mm in diameter, but neither Parsons or Norr identified this pattern. The third mode found in the UA-1 data is a dispersed group that ranges from 38 to 74 mm, with the greatest concentration at 50-51 mm. This peak corresponds to Parsons' Type 1, which she grouped with large whorls; neither Sisson or Norr had a significant number of this size whorls. In comparing the UA-1 whorls with those found at UA-79, a similar pattern emerged in the first two modes, but the greatest concentration of large whorls at UA-79 occurred between 55-59 mm.

Height measurements were taken on the UA-1 whorls (Figure 151), and when plotted they formed a bell-shaped curve ranging from 5-18 mm, with a pronounced peak between 8-10 mm. A weak second mode ranged from 20-25 mm. A similar pattern was found by Norr (1986). At UA-79, the greatest peak ocurred between 8-12 mm, and a strong second mode occurred between 19-30 mm.

In combining these two dimensions into the shape measurement, the greatest peak (representing almost 50% of the UA-1 whorls) was at .25, indicating that disk-like whorls with a height to diameter ratio of 1:4 predominate (Figure 152). As described above, these would produce a long, controlled rotation on the spindle. The same peak is present for the UA-79 whorls, but there is a relative increase in whorls with a taller shape, at the 1:2 ratio.

The great majority of the UA-1 whorls (80%) weighed 5 g or less, with the remaining whorls ranging to a maximum weight of 51 g. Parsons' (1972b) Type 3 whorls were concentrated between 1-10 g, while Type 1 weighed between 21-60 g, and Type 2 weighed from 50-80 g. The majority of whorls from Tetla-11 clustered between 2-18 g, with the greatest

Figure 151. Height of UA-1 spindle whorls

Figure 152. Shape of UA-1 spindle whorls

peak at 11 g, but with occasional outliers to 52g (Norr 1987). Smith and Hirth (1988) described a similar pattern from Morelos. The whorls from UA-79 fall into a bimodal distribution, with the main cluster between 3-11 g, and a dispersed group from 45-91 g. In terms of weight, it is notable that the UA-1 whorls are consistently lighter than whorls from other assemblages, and that there is a very high percentage of these light whorls, especially when compared with the Valley of Mexico data.

At UA-1, hole diameter is plotted as a bell shaped curve ranging from 2-13 mm, with the greatest peak at 7-8 mm (Figure 153). This is very different from Parsons' data from the Valley of Mexico, where the highest concentration was at 3 mm, in whorls belonging to Type 3. The larger types (Types 1 and 2) had their maximum peaks at 8-9 and 10 mm, respectively. Whorls from Tetla-ll had hole diameters usually ranging from 3-7 mm, with the peak at 5 mm (Norr 1987). No clear pattern was apparent among whorls from UA-79, where the greatest concentration occurred at 3 mm but hole diameter remained fairly consistent through 13 mm. The significance of this data is that while the UA-1 whorls were consistently lighter than the Valley of Mexico whorls, even for Parsons' small whorls, the shaft holes were larger. It is impossible to conclusively explain this discrepency between weight and hole size, but it is possible that it

Figure 153. Hole size of UA-1 spindle whorls

relates to whorl position on the spindle shaft, and therefore distinctive spinning techniques.

Detailed analyses of the UA-1 whorls indicate that they do not conform well with the big/little "rule of thumb" developed by Parsons, in part because they fall into a trimodal rather than a bimodal distribution. Careful consideration of the dimensions recorded from other sites (such as Coxcatlán and Tetla-11), indicate other differences between these whorl assemblages and the Valley of Mexico materials. The techniques of hand-spinning obviously require a more complex model to account for all of the potential variation.

To better illustrate this potential variation, the volume measurement provides a standard that incorporates diameter, height, and hole diameter. In the UA-1 whorls, the volume measurement indicated three distinct modes, between 4000-10,000 mm3, 14,000-24,000 mm3, and 36,000- 42,000 mm3 (Figure 154). By far the greatest concentration is in the first group.

Figure 154. Volume of UA-1 spindle whorls

The UA-79 whorls also show a trimodal distribution in terms of volume. Yet while the first group has the same range, the second and third groups do not correspond to the modes found at UA-1. Since there is a temporal difference between the two assemblages, it is likely that technological change may have affected the design of the two larger groups of whorls. In a comparison of whorls from primary depositional contexts (the UA-1 trash midden and UA-79 middens f-10 and f-12), this pattern is also apparent.

Another method of characterizing whorls from Cholula is based on form, using a typology developed for the UA-1 whorls but that has been expanded to include the materials from UA-79 (S. McCafferty and G. McCafferty n.d.). Twelve basic forms were identified, with subtypes described for several of these (Appendix B). At UA-1, Form Type 1 was the most common form found, particularly in subtypes 1a, 1b, and 1c. Types 2 (a and b) and 3a were also abundant. In contrast, at UA-79 whorls of Type 1 were a minor component, while Form 3 was completely absent. Type 2 was the predominant type present, with Types 9a and 10 also common. This comparison of whorl forms between the two assemblages further indicates a diachronic change between the Late Tlachihualtepetl phase and the Late Cholollan phase.

Finally, the spindle whorls from UA-1 were usually decorated, either with mold-impressed motifs, post-firing incision, or painted designs. Occasionally, mold-impressed whorls were further elaborated with the application of black bitumen (chapopote). Design motifs found at UA-1 included many geometric motifs, especially hatched semi-circles that form a star pattern (Figure 155). Other motifs included zoomorphic and occasional floral patterns (Figures 156-157). In contrast, at UA-79 the predominant motif was floral, particularly in patterns interpreted as marigold flowers.

A second stylistic distinction between the two assemblages was in the use of bitumen paint. At UA-1, 11.5% (n=15) of the whorls showed remnants of bitumen. These were always small whorls, and almost always had mold-impressed decoration, often with the bitumen partially obscuring the design. Stratigraphically, bitumen was more common in the lower levels (14.8%) as opposed to only 8.7% in levels I and II. The possibility that the popularity of bitumen covered whorls decreased in the Late Postclassic period is supported by the relative scarcity of bitumen from UA-79.

The significance of the bitumen coating is problematic. Since the raw material originates on the Gulf Coast, it has been suggested that whorls with this coating indicate exchange with the coast (Parsons 1972b). No functional significance for bitumen on whorls is known, and instead it is likely that the coating relates to symbolic identification with the Mother Goddess complex (S. McCafferty and G. McCafferty 1991). The use of bitumen paint was associated with the goddess Tlazolteotl, who was often portrayed with

Figure 155. Geometric motifs on UA-1 spindle whorls

Figure 156. Zoomorphic motifs on UA-1 spindle whorls

Figure 157. Floral motifs on UA-1 spindle whorls

bitumen paint around her mouth. The chewing of bitumen, in the form of chicle, was closely associated with female gender identity and sexual status (Sahagún 1950-1982, Book 10:89-90; S. McCafferty and G. McCafferty 1991). It is therefore argued that the presence of bitumen on whorls relates not so much to regional identity, but to female gender identity and affiliation with the Mother Goddess cult, particularly with Tlazolteotl (S. McCafferty and G. McCafferty 1991).

The whorl assemblage from UA-1 also provided evidence against a Gulf Coast origin for bitumen-covered whorls. One group of five whorls had an identical mold-impressed pattern, probably made from a single mold at or near the Structure 1 compound. The pattern and dimensions of the whorl mold found in Structure 1 Room 2 (recorded on the Object Card) closely matched those of the whorls, but the mold was missing from the lab at the time of this analysis so it was impossible to verify this association. The five whorls showed differential use of bitumen: three members of the group had bitumen coating but two others did not. This evidence indicates that at least in this case the bitumen was not applied on the Gulf Coast, but rather the raw material was available locally. Furthermore, this suggests that the ideological significance of the bitumen decoration was shared by the occupants of UA-1.

The spatial distribution of whorls from UA-1 is also useful for identifying use and discard areas. Because of the proposed distinction between large whorls used for drop-spinning and small whorls used for supported-spinning, the whorls were differentiated on the basis of size, with whorls having a diameter of less than 4 cm considered as small and those with a diameter greater than 4 cm considered as large. As with the other artifact classes, a further distinction was made between the upper two levels, and Levels 3 and below which were considered primary depositional contexts (Figure 158).

The relative frequencies of small and large whorls varied between the two contexts, with large whorls more frequent in the upper levels (33% as opposed to 24%). In

Figure 158. Spatial distribution of spindle whorls

the upper levels, whorls were distributed fairly evenly, with only weak clustering apparent. In contrast, whorl distribution in the primary contexts shows strong clustering in the Trash Midden and in Well 1, and with almost all other whorls found in contact with the structure floor. In the upper levels large whorls were randomly distributed, whereas in the occupational context the large whorls were only found in the midden and well, while small whorls were also found on the floor.

These patterns support the hypothesis that whorl distribution varied based on context, with larger whorls more likely to enter the archaeological record in mixed contexts, as they were more often used in open areas outside of the household compound. Small whorls, used in supported spinning, were most often used within the household compound. The whorls recovered from the floor contact may relate to areas where spinning took place at the time that Structure 1 was abandoned.

The UA-1 excavations provide a rich sample of spindle whorl data that can be used to infer the role of spinning and weaving activities in the structural compound. Since spinning and weaving were activities closely associated with female production and also gender identity, this artifact class provides some of the best evidence for establishing the presence of women within the Structure 1 compound. Furthermore, the high number of whorls indicates that spinning, at least, was an important activity in the compound, either for meeting tribute demands, ritual gift-giving, or for market exchange.

The UA-1 spindle whorls provide information on diachronic change both in whorl morphology and in the design elements encoded onto the whorls. This data is first indicated through stratigraphic variation within the site, but is supported through comparisons with Late Postclassic whorls from UA-79 and also the mass burial found in San Andrés Cholula (Suárez 1990). Whereas small whorls were the most common both at UA-1 and UA-79, larger whorls increased in frequency in later contexts. This may indicate a greater importance of maguey fiber during later periods, perhaps as a result of Aztec market strategies.

Several important differences were identified between the Cholula whorls from UA-1 and UA-79 and the Valley of Mexico whorls presented by Parsons (1972b). Perhaps the most important is the trimodal distribution found in whorl diameter and volume measurements. Based on the assumption that whorl size and weight relates to functional characteristics of the spinning process, it is likely that the Cholula data reflects a more complex set of techniques. This is also found in the evidence of weight and hole diameter, since the UA-1 whorls were consistently lighter than the Valley of Mexico whorls, yet they had larger hole diameters, suggesting the use of larger spindles, or a different position on the spindle. To resolve these problems, additional spindle whorl data from carefully controlled contexts are needed.

Finally, these comparisons relate to symbolic elements encoded onto the whorls. During the occupation levels of UA-1, bitumen and other motifs associated with the goddess Tlazolteotl were common in the whorl assemblage. In later contexts, however, floral motifs were more common, suggesting a possible religious transformation to increased importance of the central Mexican Mother Goddess complex, including Toci and Xochiquetzal (S. McCafferty and G. McCafferty 1991). Since this change also relates to ethnic affiliations, the spindle whorl data supports the general ethnohistoric and archaeological evidence for ethnic change in Cholula (McCafferty 1989b).