Archaeometry 48, 2 (2006) 293– 307. Printed in Singapore 48 Archaeometry ARCH 0003-813X © Oxford, 2May University 2006 UK of Oxford, 2006 ORIGINAL An E. J.exploration Tiedemann ARTICLE of and prehistoric K. A. Jakes spinning technology Blackwell Publishing Ltd AN EXPLORATION OF PREHISTORIC SPINNING TECHNOLOGY: SPINNING EFFICIENCY AND TECHNOLOGY TRANSITION* E. J. TIEDEMANN 1418 35th Avenue, Seattle, Washington 98122, USA and K. A. JAKES Department of Consumer and Textile Sciences, The Ohio State University, 1787 Neil Avenue, Columbus, Ohio 43210, USA Thigh-spinning and spindle-spinning, methods for making yarn and string, have been used through the millennia to produce substantial quantities of yarn for textiles. Productivity data were gathered in a replication study of thigh-spun and spindle-spun yarns and were recompiled from the literature. Calculated production rates allow comparison of the spinning methods. Factors influencing production rate include intrinsic fibre properties, fibre preparation and particular details of spinning technique, as well as the experience and motivation of the spinner. The role of efficiency in technology transition between thighspinning and spindle-spinning is discussed. KEYWORDS: SPINNING, SPINDLE, TEXTILE TECHNOLOGY, EXPERIMENTAL ARCHAEOLOGY *Received 5 April 2004; accepted 29 September 2005. INTRODUCTION The spinning of yarn is an essential step in making most textiles. Fibre obtained from plants or animals must be twisted in some manner to make yarns that can be twined or otherwise interlaced into a fabric structure. Thigh-spinning and spindle-spinning are two of the oldest methods for making yarn. Both have been used for millennia to produce substantial quantities of yarn for textiles. Although the time required to produce a textile has been recognized by a few researchers (Drooker 1992; Ericksen et al. 2000), few replication studies have been conducted to evaluate the time cost of each of the steps contributing to the fabric’s completion. Jakes and Ericksen (1997) include crude estimates for yarn preparation by a finger-spinning method. Their work showed that it took twice the amount of time to spin than was required to twine the same length of yarn into fabric. Well-preserved textile artefacts from the Etowah site in Carter County, Georgia, show fine workmanship. No corresponding evidence of yarn-making technology has been recovered at the site. Lack of evidence has made the subject of spinning technology in pre-Columbian eastern North American textiles a matter for speculation (Sibley et al. 1989; Drooker 1992). A comparative study of thigh- and spindle-spinning was designed to answer questions about *Received 5 April 2004; accepted 29 September 2005. © University of Oxford, 2006 294 E. J. Tiedemann and K. A. Jakes spinning technology employed in eastern North America during the Mississippian period (c. ad 800 to ad 1600). The implications for spinning technology in other regions and periods were considered as well. Thigh-spun and spindle-spun yarns were replicated for comparison of production methods and of yarn physical properties. This paper addresses the differences in efficiency for producing thigh-spun and spindle-spun yarns. Although the assumption that spindle-spinning is a faster process is generally accepted, the absolute difference between the spinning methods has not been established. To evaluate the difference between thigh-spinning and spindle-spinning productivity, recorded spinning rates were collected from the literature and spinning rates were measured during replication of yarns. These reveal specific differences in productivity over a spectrum of spinning contexts that are characterized by different fibres and tools. Differences in productivity for novices and experts are also discussed, with an eye towards the perceived benefits or costs of abandoning thigh-spinning in favour of spindle-spinning; that is, adopting a new technology. DESCRIPTION OF SPNNING TECHNOLOGIES Thigh-spinning is a splicing technique that requires no tools. It tends to be used for long bast or inner tree bark fibres, although short wool fibres are also thigh-spun (Samuel 1982, 1985). Figure 1 shows the process of thigh-spinning two-ply yarn. Spinning more than one strand at once and allowing them to twist back on each other into a plied yarn creates a balanced yarn with stable twist that will not kink up or snarl. To add length to the yarn, a spinner splices fresh lengths of fibre into the loose ends of the spun yarn. To spin yarn with a spindle, a spinner draws out a small amount of fibre from a mass of fibre to be twisted into the yarn (Fig. 2). The drawn fibre is attenuated and aligned in a drafting triangle as it is nipped into the yarn. Twist travels from the turning spindle up the yarn to the newly drafted fibre. This process of drawing fibre from a stationary fibre supply into a twisting yarn is called continuous drafting. A spindle is a tool that minimizes the hand motions necessary to put twist into a yarn. Spindle design is highly variable, but all spindles have two essential functions. They store the spun yarn and they keep the yarn turning after being set in motion by the spinner. Spinners apply torque to a spindle in a manner similar to setting a top in motion. A whorl or, alternatively, a wide spot in the shaft, or the built-up mass of spun yarn on the shaft, conserves angular momentum. The inertia provided by the whorl also prevents the freshly spun yarn from untwisting. The spindle shown in Figure 2 has a whorl at the base and a shaft for holding the yarn. In comparison to thigh-spinning, where each turn of the yarn is directly associated with a hand movement, spindle-spinning increases spinners’ efficiency. Because spindles continue to spin after being set into motion with a single hand movement, they decrease the number of hand movements involved in twisting the yarn. Whereas thigh-spun yarns are created in multiple plies at once (e.g., two singles yarns are twisted simultaneously and then twisted back on each other into a two-ply yarn, before moving on to the next length of unspun fibre), spindle-spun yarns can only be made one ply at a time. In order to make a two-ply yarn, two singles yarns must be spun one at a time. In a separate operation, the singles yarns are drawn together analogous to the drawing of fibre and then are spun together on a spindle into a plied yarn. The two spinning methods are radically different solutions to the problem of making long yarn out of short fibres. They differ in the way in which the yarn is lengthened, and the way An exploration of prehistoric spinning technology 295 Figure 1 Thigh-spinning (illustration by Margaret Davidson). (a) The spinner rolls two separate groups of fibres towards her knee, to form two S-twist singles yarns. Circular arrows show the clockwise direction of twist in the forming yarns and the loose fibre ends. The spinner’s left hand pinches off the new singles from the previously finished two-ply yarn. (b) The spinner continues rolling the new singles yarns towards her knee, but has released the pinch between the new singles and the finished two-ply. The singles continue to twist individually in the S-direction, as indicated by the clockwise rotation of the loose fibre ends. At the same time, the singles twist back on each other in the Z-direction, forming the plied yarn. S-twist is being added to the singles by the hand motion, while Z-twist migrates down the two plies from the finished yarn. (c) The spinner rolls the two-ply yarn away from her knee, to increase the twist in the Z-direction. This finishes the two-ply yarn. All the loose fibre ends turn together in the counter-clockwise direction. in which twist is added. Thigh-spun yarns are lengthened by splicing in discrete lengths of fresh fibre. Spindle-spun yarns are lengthened by a continuous draft from a relatively large fibre supply. To twist the yarn, thigh-spinners use friction between a hand and a leg to roll fibres into a yarn. Spindle-spinners apply torque to a spindle shaft often with just a snap of the fingers and the spindle, turning freely, twists the new yarn from a point. 296 E. J. Tiedemann and K. A. Jakes Figure 2 Spindle-spinning (illustration by Margaret Davidson). The spinner draws fibre into the yarn from a relatively large fibre source. The turning of the spindle adds twist to the yarn. The two methods also differ in the process of yarn construction. Two-ply thigh-spun yarns are constructed all at once with two hand motions. The singles are generated simultaneously with one hand motion. The second hand motion plies them together and the spinning on this section of yarn is complete. To make a two-ply yarn with spindle technology, a spinner spins three separate yarns: two singles yarns and the final plied yarn. These two routes to twisted two-ply yarns bear no relation to one another. This description shows that the two technologies resulting in similar finished products are both mechanically and conceptually different. LITERATURE REVIEW Ethnographic and ethnohistoric reports of spinning practices in eastern North America give conflicting evidence. Evidence of thigh-spinning appears shortly after contact and continues well into the 20th century (Smith 1923, 1928, 1932, 1933; Jones 1937, 1946; Kalm 1937 An exploration of prehistoric spinning technology Table 1 297 Relation of spinning time to textile production time Source Spinning method Finished work Spinning time/weaving time Spinning as a percentage of total production time* Jones (1946) Thigh Finger-woven cedar mat 0.22† 10 Bird (1968) Spindle Patterned poncho 0.80‡ – Franquemont (1986) Spindle Patterned llijlla Patterned llijlla Plain-woven poncho Plain-woven costal Plain-woven costal 1.67 1.84 2.80 3.71 4.56 53.6 56.2 69.2 69.1 74.3 *Total production time includes all steps from fibre collection to the completion of the textile. †Spun yarn only bound the edges of the mat. The mat itself was woven from unspun strips of cedar bark; thus the ratio is not comparable to ratios for fabrics made entirely from spun yarns. ‡Calculated from the minimum estimated production time for a poncho. [1753]; Speck 1940; Strachey 1953 [1612]; Swanton 1969), yet other references suggest the use of spindles (Sigüenza y Góngora 1939 [1693]; Adair 1968 [1775]). Scattered collections of spindle whorls have been reported in the American Bottom and have been linked to the onset and increase of spindle use between ad 1000 and 1100 (Alt 1999), but the lack of such evidence at many eastern North American sites is noteworthy. In efforts to quantify the time cost of textile production, a few researchers have documented rates of textile production in ethnographic studies (Jones 1946; Bird 1968; Franquemont 1986). Table 1 shows calculations of spinning time relative to textile production time from these studies, which followed the entire textile production process. The relationships of spinning time to weaving time and spinning time to total production time were calculated for this table from cloth production times presented in the original studies. Because the thigh-spun yarn was only a binding element in a mat woven from cedar strips (Jones 1946), the ratios are not comparable to the spindle-spun ratios, where yarn was the only material used to make the entire finished product. Nevertheless, these data show that the spinning of 20 feet of thigh-spun yarn to secure the edge of a mat required a significant portion of the total production time. For the spindle-spinning examples, the ratios of spinning time to weaving time show that loom-woven fabrics can consume yarn at a faster rate than that at which yarn is produced. The percentage of total production time devoted to spinning shows how spinning rate can dominate the production rate of a woven fabric, especially for plain weave fabrics that can be woven more quickly than intricately patterned weaves. As a necessary and time-consuming component of textile production, spinning activities and spinning technology contain information that potentially bridges the gap between the artefact and human cultural activities of artefact production and use. Because spinning is such a common activity in textile-producing societies, it is a technological variable that provides a window on the daily activities of many workers within a society. For example, Brumfiel (1996, 2001) has used spindle whorl data in an investigation of women’s productivity as an indicator of social agency in Aztec and colonial Mexico. 298 E. J. Tiedemann and K. A. Jakes Wherever textiles are made, textiles and textile labour are important components of economic and social activities. Hand-spun, hand-woven textiles have been shown to be commodities in the economic structures of complex states. Pictographic representations of the counts of men’s cloaks in the Codex Mendoza (Berdan and Anawalt 1992; Berdan 1996) document the place of cloth in tribute obligations over the Aztec empire. Early colonial texts from the Andes document how textile labour was used to fulfil tax requirements in the Inca state (Huamán Poma 1978 [1615]; Rowe 1979). These examples link regional and individual labour to state-wide tribute and taxation obligations. In effect, individuals fulfilled their responsibilities by producing specific articles on demand. Thus, textile labour not only meets household needs, but connects household activities to the functioning of the state. A wide range of textile forms and structures have been documented in eastern North America, showing both the diversity of utilitarian textiles at Wickliffe (Drooker 1992), as well as the labour-intensive attributes of textiles found among other artefacts associated with status in the assemblage of Burial 109 at Etowah (Schreffler 1988). The fabrics from Etowah exhibit both very fine yarns and yarns of complex composition, and can be seen as tribute to the dead interred in the burials (Sibley et al. 1996). Shortly after European contact, textiles were common among gifts between peoples: native Americans presented Hernando De Soto with food, skins and blankets, and De Soto presented hosts with beads, pieces of cloth and articles of clothing (Account 1993 [1557]; Rangel 1993 [1851]). The data gathered in this study provide a baseline for enquiry into the cultural context of textile production, for which a theoretical framework has already been developed (Sibley and Jakes 1989; Ericksen et al. 2000). Having recognized the time costs for each step of textile production, Jakes and Ericksen (1997) provide only crude estimates of the time required in finger-spinning fine yarns. Their work has highlighted a need for additional studies of spinning technology and productivity. The present work is also consistent with recommendations in the field of pottery studies that call for detailed examination of chains of production towards the understanding of technology as a social process (Schiffer and Skibo 1997; Sillar and Tite 2000). METHODOLOGY Fibres Bast fibres from four different plants were used as the raw material for the replicated yarns. The plant sources for the fibre were flax (Linum usitatissimum), common milkweed (Asclepias syriaca), Indian hemp (Apocynum cannabinum) and basswood (Tilia americana). Flax is an Old World bast fibre plant, whereas milkweed, Indian hemp and basswood are indigenous fibre plants reported to have been used by Native Americans in eastern North America (Smith 1923, 1928, 1932, 1933; Kalm 1937 [1753]; Jones 1937, 1946; Whitford 1941; Densmore 1974 [1927]; Erichsen-Brown 1979). Flax was selected as a representative of bast fibres used in the eastern hemisphere, so that the findings from this work could be applied to bast fibre spinning regardless of region. It was purchased from a hand-spinning supplier as a prepared fibre, ready for spinning. The milkweed and Indian hemp were collected and prepared from wild plants (Tiedemann 2001). Fibres were removed by hand from the plant stems in an effort to replicate the very fine fibres seen in textiles found at the Etowah site. Only one spinner used basswood, which he had prepared himself for thigh-spinning. An exploration of prehistoric spinning technology 299 Spinners Spinners were selected for their experience in spinning long vegetable fibres and also by evidence that each provided of his or her commitment to teaching and learning spinning techniques. Six spinners participated in the study, either thigh-spinning or spindle-spinning depending on their specialization. One of the authors spun both types of yarns for the study. Two of the spinners have published articles about their work in thigh-spinning (White 1969; Leeds 1999). The other three spinners had been spinning for a minimum of 12 years each and also teach spinning classes. All of the spinners except the author were paid a modest fee for their participation. With the exception of the basswood yarn, which was spun in the preliminary stages of research, the spinners were given instructions to spin fine yarns. They were also shown a photograph of textiles from Etowah Mound C no. 843, so that they would understand the focus of the replication study. In short, the spinning consultants were explicitly asked to create yarns of the order of fineness of the Etowah textiles, with a specific set of fibres. The purpose of explaining the desired yarn, rather than leaving the spinners to make any yarn, was to generate yarns that represented individual solutions to the problem of making fine yarns with specific fibres and specific technologies. All spinners made two-ply yarns. A two-ply structure was chosen to replicate the two-ply yarns that occur in eastern North American archaeological textiles. The process of thigh-spinning by rolling separate strands of fibre in one direction and then allowing them to twist back on each other results in a balanced two- or more ply yarn, depending on the number of strands with which the spinner works. Spindle-spinning results initially in a singles yarn that must be spun again with another singles yarn to make a two-ply yarn. As spinners were interviewed prior to their spinning contribution, it became clear that the thigh-spinners had individual approaches to making the yarn that was requested. One, when asked to spin fine yarn, preferred to twist the yarn in his fingers rather than use the thigh-spinning method. Another had developed his own variation of the thigh-spinning method, which he preferred. Rather than directing these spinners to spin in the same manner that she had researched and learned to spin, the author chose to accept the other spinners’ individual solutions to the yarn-making problem and incorporate them into the research. The spindle-spinners used their own spindles. All three used high whorl spindles with hooks. Spindle dimensions are given in Table 2. Yarn production rates Yarn production rates were calculated in metres per minute (m min−1) from the length of yarn spun during a 10 min period. Each spinner was asked to spin each fibre for 10 min after he or she had spent a few minutes getting used to spinning with that fibre. The yarn was tied at the beginning and end of the timed period with small lengths of red embroidery floss. As a demonstration, at least one timed spinning was conducted during the author’s interview with the spinner. The spinners then carried out the remaining timed spinnings on their own. The spindle-spinners were timed for both singles spinning and plying, because the overall spinning rate includes both operations. The marked lengths of yarn were measured later by the author. Average spinning rates were calculated for each spinner over all three fibre types. The combined average spindle-spun yarn production rate (Rtwo-ply) was computed from the average E. J. Tiedemann and K. A. Jakes 300 Table 2 Yarn production rates Raw spinning rates† (m min−1) Spinner* Mean rate (m min−1) T1 Fx Fx I M 0.1265 0.101 0.108 0.0905 0.107 T2 Fx I M 0.1575 0.1945 0.177 0.176 T3 B 0.170 0.170 F4 Fx I 0.08 0.0565 0.0683 Average‡ 0.130 Spinner* Spindle singles Raw spinning rates† (m min−1) Spindle ply Mean rate (m min−1) S1 Fx Fx I M 0.6172 0.493 0.38 0.415 0.476 S5 Fx I M 0.778 0.528 0.505 0.604 Fx I M 1.435 0.881 1.146 1.154 S6 Raw plying rates† (m min−1) Fx Fx I M 1.1565 1.3315 1.077 1.103 I M 0.767 0.882 Fx 1.094 M 0.932 Average‡ 0.745 Combined average spindle-spun yarn production rate (Rtwo-ply) = 0.271 Spindle dimensions Mean rate (m min−1) Mass (g) Whorl diameter (cm) Length (cm) 1.167 25.7 5.2 23 0.825 25.5 8.0 26 1.013 30.4 12.0 30 1.002 *Key to spinners: The first character designates the spinning method: T for thigh-spinning, F for finger-spinning and S for spindlespinning. The second character distinguishes the individual spinners, 1– 6. Spinner number 1 spun both thigh-spun and spindle-spun yarns, whereas the other spinners only spun one type of yarn. †Measurement of spinning rates using different types of fibre: B, basswood; Fx, flax; I, Indian hemp, M, milkweed. ‡Averages are unweighted averages over the spinners in each group. singles spinning and plying rates, given in m min−1, where Rs is the average singles spinning rate and Rp is the average plying rate: Rtwo- ply = 1 .  1 1 1 R + R + R   s s p The equation takes into account that three operations must be performed in order to make a two-ply, spindle-spun yarn. For every metre of plied yarn, 2 m of singles yarns must be spun An exploration of prehistoric spinning technology 301 and 1 m must be plied. To find the spinning rate for a two-ply yarn in m min−1, the number of minutes required to complete all the spinning for 1 m of plied yarn are summed; that is, twice the singles spinning time plus the plying time. The number of minutes to spin 1 m is equal to the inverse of the measured spinning rate. The result in minutes per metre is then inverted to yield Rtwo-ply in m min−1. RESULTS Yarn production rates Results for the 10 min rate-determining spinnings appear in Table 2. Although each spinner was asked to spin each fibre for 10 min to determine the spinning rates over all the fibres, several spinners forgot one timing. Spinner number 3 spun only a single repetition of basswood fibre. Averages were calculated despite the missing data. As expected, thigh-spinning and finger-spinning were slower than spindle-spinning. Spindle-spinning was 2.1 times as fast as the average of all the non-spindle techniques. If the slower finger-spinning is eliminated in the calculation of the average, then the average thighspinning rate was 0.151 m min−1 and spindle-spinning was 1.8 times faster. The thigh-spinning method used by T2 to achieve the fastest rate was a personal variation on the traditionally reported method. T3, however, using the traditional method, was only slightly slower. DISCUSSION Table 3 summarizes spinning-rate data collected from the literature of hand-spinning studies. One entry, Mountford (pers. comm., 2004), was obtained from a 10 min timing following two telephone interviews in March of 2004. Mountford is a professional thigh-spinner, who supplies yarn to weavers for Chilkat dancing blankets. Her instructive thigh-spinning article of 1985 was published under the name Samuel. The reported spinning rates have been standardized here for ease of comparison to units of m min−1. These data, organized from slowest to fastest, show differences that cannot necessarily be attributed to spinners’ varying levels of experience. For instance, the Bird (1968), Franquemont (1986) and Lopéz A. (1985, cited in Vreeland 1986) averages were calculated from large samples taken from large populations of active spinners. These rates are the slowest spindle-spinning rates that appear in the literature, but inexperience is not a likely explanation for slower production. Very short fibres, such as cotton, flax tow and some wools, require more turns per inch to hold the fibres together in a yarn. Spinners will need more time to insert more twist as fibre length decreases. The difference between the Peruvian cotton and wool spinning rates, therefore, is not unexpected, because short cotton fibres require much more twist than wool fibres. Flax line is the fibre used for the second fastest spindle-spinning rate and hemp (we can only assume long fibres) is the fastest. Fibre length, where given, shows a positive correlation with increased spindle-spinning rate. The quality of fibre preparation also influences both the rate of drafting and the evenness of the yarn. In the Finnish study of wheel-spinning, the observed difference between flax and wool spinning rates was regarded by the author as very narrow, because the quality of the available wool fibre had not been as good as the quality of the flax fibre (Vallinheimo 1956). Alternatively, Peruvian cotton spinners recognize that poorly prepared cotton results in lumpy yarn that breaks easily (Vreeland 1986). Thus spinning rates are at least partially dependent on E. J. Tiedemann and K. A. Jakes 302 Table 3 A comparison of the hand-spinning rates reported in the literature Source Spinner(s) and region Spinning method Fibre Jones (1946) Samuel (1982) Mountford (pers. comm., 2004) López 1985, in Vreeland (1986) Bird (1968) Franquemont (1986) Schwarz (1947) Bird (1968) Franquemont (1986) Schwarz (1947) Vallinheimo (1956) Vallinheimo (1956) 2 spinners Ontario, Canada Pacific coast, Canada 1 spinner, Canada Thigh Thigh Thigh Basswood Cedar and wool Wool 0.05 0.25 0.38 50 spinners Mórrope, Peru Spindle, no hook Cotton 0.65 100 spinners, Pisac, Peru 53 spinners, Chinchero, Peru Bukovina, 1912 6 spinners Pisac, Peru 10 spinners, Chinchero, Peru Southern Italy 1 spinner Suojärvi, Finland 21 student spinners, Finland Spindle, unspecified Spindle, no hook Spindle, no hook Spindle, unspecified Spindle, no hook Spindle, with hook Spindle, no hook Wheel† Vallinheimo (1956) Szolnoky 1950, in Endrei (1968) Vallinheimo (1956) 1 spinner, Suojärvi, Finland 1 spinner, Nagylóc, Hungary Wool Wool – Plying Plying – Flax tow Wool Flax line Spindle, no hook Flax line Spindle, unspecified Hemp Spinning teacher, Finland Wheel† Vallinheimo (1956) Champion spinner, Ruovesi, Finland Wheel† Wool Flax line Wool Spinning rate* (m min−1) 1.04 1.08 1.00–1.40 1.59 1.81 1.83 2.00–2.15 2.21 2.30 2.40 2.63 4.40 4.73 7.19 *Rates have been converted from various units to m min−1 for comparison. †Wheels are treadle driven, with a bobbin and flyer spinning system. intrinsic fibre properties, as well as work invested in fibre preparation that is appreciated, but is not directly measurable, at the time of spinning. Differences in the manipulation of the fibre, yarn and spindle may also affect spinning rates. A study of native cotton processing and spinning on the North Coast of Peru documents differences in technique for North Coast cotton spinning and Highland wool spinning. The North Coast spinners hold the spindle at the base in their right hands and extend it horizontally in front of them to spin an S-spun cotton yarn. The Highland spinners hold the spindle at the top, letting it hang vertically beneath their right hands to spin a Z-spun wool yarn (Vreeland 1986). Other possible variations in spindle-spinning include spindle design—for example, high whorl, low whorl, no whorl or supported—the use of a hook or a half-hitch knot to secure the yarn to the tip of the spindle, and the method of setting the spindle in motion. Although experience can be measured in years of work or quantity of yarn produced, human motivation is harder to explain. An individual spinner may have interest in spinning only as fast as his or her neighbours, or he or she may be interested in spinning faster for economic benefit or for competition. These are just a few intangibles that, nevertheless, affect rates of spinning. Spinning rates representative of a particular production environment have been calculated from large samples of spinners still engaged in daily production, as in the Bird, López and Franquemont studies. Such samples included the fast and the slow in a range of possible spinning rates. They give a general picture of the rate of yarn production within a population using similar tools and having similar expectations. An exploration of prehistoric spinning technology 303 Comparison of the spinning rates reported in this study (Table 2) with the rates reported in prior studies (Table 3) shows substantial differences. The thigh-spinners spun 2–3 times faster than the rate reported in Jones (1946), but slower than the rates reported by Samuel (1982) and Mountford (pers. comm., 2004). The spindle-spinners were slower than all previously reported rates. The average spindle rate of 0.745 m min−1 for singles compares only to the previously reported cotton-spinning rate of 0.65 m min−1. The plying rates collected here were also slower than the rate of 1.81 m min−1 reported in Franquemont (1986). The differences between the spinning rates collected in this study and the spinning rates reported elsewhere probably reflect the degeneration of the craft that Endrei (1968) suggested in his report on wheel-spinning. None of the spinners in this study spins daily for household needs. Similarly, Jones (1946) reported that the women whose thigh-spinning rates he recorded were reviving a little used skill at his request. The difference in productivity between the spindlespinners recruited for this study and spindle-spinners in other studies was especially marked. These results can be explained in terms of Vallinheimo’s (1956) study of spinning productivity. Her comparison of spindle- and wheel-spinning productivity showed that the level of experience and motivation of the spinner affect productivity. A spindle-spinner from Suojärvi, Finland, spun flax yarn from a well-prepared distaff at 2.40 m min−1. With a wheel, the winner of a 1954 national level Finnish spinning competition spun wool yarn at a rate of 7.19 m min−1. In contrast, 21 students in a home economics teaching program with 2 years’ wheel-spinning experience and a practice schedule of 4 –5 h per week nearly matched the productivity of the Suojärvi spindle-spinner, with an average flax spinning rate of 2.30 m min−1. Their spinning teacher spun flax at a rate of 4.73 m min−1 and wool at 4.40 m min−1. These results show that spinning on a wheel rather than a spindle does not necessarily increase the rate of yarn production. Vallinheimo comments that only with ‘Mühe und Not’ (effort and necessity) does a wheel-spinner improve upon the efficiency of a spindle-spinner. Just as Vallinheimo’s (1956) study of wheel-spinners with varying levels of experience showed that spinning rates of practised spinners do not compare with the rates of highly experienced spinners, the results of this study show that practised spindle-spinners do not approach the production rates of highly experienced spindle-spinners. Although the primary lesson from these results is that the production rates of modern hobby spinners do not reflect the maximum possible production rates of the technology, the results still have merit. The spindle-spinning results show the production rates that might be expected of spinners adopting spindle technology. The combined average spindle-spun yarn production rate achieved by the spinners in this study was 0.271 m min−1. The maximum recorded thighspun production rate of 0.38 m min−1 (Mountford pers. comm., 2004) surpasses their productivity by 40%. In this context, spindle-spinning compares rather poorly to thigh-spinning. Only when data collected from more experienced spinners are compared does spindle-spinning show greater productivity than thigh-spinning. Calculating Rtwo-ply using Franquemont’s average spinning and plying rates (Table 3) gives a rate of 0.416 m min−1 for two-ply spindle-spun wool yarn—only slightly faster than Mountford’s thigh-spun rate. For flax line, the two-ply spindle-spinning rate, calculated with a singles yarn spinning rate of 2 m min−1 and an assumed minimum plying rate of 2 m min−1, equals 0.667 m min−1. Vallinheimo concluded from her results that the spinning wheel is not absolutely faster than the spindle. Likewise, it appears here that the spindle is not absolutely faster than thigh-spinning, but requires dedication and necessity to surpass thigh-spun productivity. The maximum differences between thigh-spinning and spindle-spinning rates were best expressed in the literature, where spinners with some economic dependence on their spinning 304 E. J. Tiedemann and K. A. Jakes were evaluated. For them, spindle-spinning two-ply yarns can be almost two times faster than thigh-spinning them. This production rate study shows, however, that spindle-spinning would not have this advantage immediately after it was adopted. The practised, but part-time, spindlespinners evaluated in this research could not spin as fast as a more experienced thigh-spinner. The large difference between incipient spindle-spinning productivity and established spindlespinning productivity suggests that thigh-spinners making a transition to spindle-spinning would not have seen an immediate increase in their productivity. If a marked increase in productivity would not be noticed, it becomes important to find other reasons why spinners might adopt spindles. Using singles yarns rather than plied yarns in textile manufacture would have accentuated the speed of spindle-spun production over thighspun production. In spindle-spinning, making two or more yarns and then plying them together doubles, at the very least, the time investment in a given length of yarn. Spindle-spun singles may be produced at a rate six times faster than a two-ply, thigh-spun yarn. Even the slower spindle-spinners from this study produced singles faster than the fastest thigh-spun rate. This advantage would not be relevant where textiles continued to be made of two-ply yarns. Because the inherent properties of singles and plied yarns are different enough to affect the properties of the final textile, it is reasonable to expect that a preference for two-ply yarns could outweigh concerns over yarn production rate. Singles yarns tend to be weaker and more irregular than plied yarn. Yarn construction also affects the feel and drape of the final textile. Because singles yarns are not an interchangeable alternative to plied yarns, a switch from plied to singles yarns would result in a new kind of textile. If increased spinning rate is abandoned as the assumed chief advantage of the spindle, one remaining spindle property that sets the two methods apart is the spindle’s reliance on continuous drafting as the means of lengthening the yarn. Adoption of spindles may have followed from a need to handle fibres differently rather than a need to make more yarn. As suggested by Barber (1991), one possible reason for the introduction of spindles is that they initially conferred an advantage in the spinning of short fibres such as cotton and wool. Without a tangible leap in productivity, spinners working almost exclusively with long bast fibres and an effective thigh-spun splicing technology might have had little reason to invent or adopt spindles and continuous drafting. CONCLUSIONS Dedicated spindle users produce yarn at a faster rate than dedicated thigh-spinners, but new and sporadic spindle-spinners do not share this advantage. In the context of technology transition, the rate of newly adopted spindle-spinning would not necessarily seem faster to an accomplished, life-long thigh-spinner. A period of equivalence between the rates of spindlespinning and thigh-spinning can be expected to add to the costs of innovation. Although spindlespinning offers an ultimate increase in productivity, transitions between thigh-spinning and spindle-spinning may have greater learning and acceptance barriers than are often imagined. These independent findings support Barber’s (1991) suggestion, based on archaeological remains in dynastic Egypt, that the choice of raw materials in the form of short fibres rather than long fibres may have a great influence in the initial development of spindle-spinning. The Egyptian example, in which flax fibres were spliced into singles yarns before being plied on a spindle, shows the uncertain boundary of transition between spinning technologies. That spindles and thigh-spinning have coexisted shows how methods of handling materials can persist despite the concomitant use of more efficient technology. Although long fibres such as flax fibres have been An exploration of prehistoric spinning technology 305 shown here to have the fastest potential spindle-spinning rates, that potential was not realized in dynastic Egypt. Only in retrospect can it be assumed that spindles are faster than thigh-spinning. Documented coexistence of splicing and spindle-spinning also suggests that there are many possible paths to the acceptance of spindle technology. The work patterns of spinners and weavers can be expected to affect productivity, but they fall outside the reach of m min−1 productivity measurements. Spinning and weaving are typically pursued intermittently with other household responsibilities. In Franquemont’s study, one weaver commissioned the yarn for her llijlla to be spun by another woman, who specialized in spinning for others. Another woman worked from a pre-existing stock of yarn that she replaced as it was used (Franquemont 1986). Weavers reported that they could expect to spend up to 6 months working on a poncho, and never less than 3 months (Bird 1968). Textile production exists within a universe of time constraints, individual preferences and social relationships. Changes in spinning technology will have occurred in a complex environment, where any number of variables in the chain of production could have been modified before one yarn-making technology was abandoned in favour of another. In this research, yarn production has been shown to be a bottleneck in the textile chain of production. Thus spinning practices should be sensitive to increases in demand for textiles, and increasing demands on production should be recognized as motivators for spinners. Although the previous discussion calls into question the notion of simple efficiency-based transitions from thigh- to spindle-spinning, it remains clear that spindles offer a significant increase in productivity to dedicated users. Yet the investment necessary to realize greater spindle productivity highlights the difficulty faced by spinners adopting spindles to replace an embedded thigh-spinning technology. An understanding of the barriers to this transition will allow researchers to look for the forces that drove the transition, rather than accepting innovation and adoption or simple diffusion explanations of technology change. Spinning activities have great potential as archaeological indicators. Because spinning can be isolated as a rate-determining step in textile production, it should be sensitive to time constraints and increasing needs for efficiency. Wholesale changes in spinning technology will correspond with changes in the need for and use of textiles. With textiles having been a mainstay of gift giving, tribute and taxation relationships in so many societies, spinning practices and technology can reveal developments in and pressures on such arrangements. Concepts that might be pursued in further research include the following. A change in social organization that emphasizes accumulation and/or redistribution of textile wealth may be accompanied by the appearance of spindle-spinning where it was not present before. Where thigh-spinning is practised, any change in general use or distribution of textiles that stresses the household production system may result in the development of spindle-spinning. Although other reasons for the use of spindle- or thigh-spinning could be envisioned and then tested, this paper has pursued one path, that of production efficiency. In general, spinning consumes a great deal of time and is practised daily by a large segment of the population in societies that rely upon hand-spinning for all of their textiles. When understood as a dynamic labour variable affected by raw materials as well as social pressures, spinning activities can open a window on the daily lives and labour of a large segment of most pre-industrial societies. ACKNOWLEDGEMENTS This research was supported by a Lucy R. Sibley Research Award, an Ohio State University Graduate School Alumni Research Award and an Anita McCormick Fellowship. 306 E. J. Tiedemann and K. A. Jakes REFERENCES The account by a gentleman from Elvas, 1993 [1557], True relation of the hardships suffered by Governor Hernando de Soto (trans. and ed. J. A. Robertson), in The Desoto chronicles, vol. 1 (eds. L. A. Clayton, V. J. Knight Jr and E. C. Moore), 19–219, University of Alabama Press, Tuscaloosa. Adair, J., 1968 [1775], The history of the American Indians, Johnson Reprint Corporation, New York. Alt, S., 1999, Spindle whorls and fiber production at early Cahokian settlements, Southeastern Archaeology, 18(2), 124 –34. Barber, E. W., 1991, Prehistoric textiles: the development of cloth in the Neolithic and Bronze Ages with special reference to the Aegean, Princeton University Press, Princeton, NJ. Berdan, F. F., 1996, The tributary provinces, in Aztec imperial strategies (F. F. Berdan, R. E. Blanton, E. H. Boone, M. G. Hodge, M. E. Smith and E. Umberger), 115–35, Dumbarton Oaks, Washington, DC. Berdan, F. F., and Anawalt, P. R., 1992, The Codex Mendoza, University of California Press, Berkeley. Bird, J. B., 1968, Handspun yarn production rates in the Cuzco region of Peru, Textile Museum Journal, 2(3), 9–16. Brumfiel, E. M., 1996, The quality of tribute cloth: the place of evidence in archaeological argument, American Antiquity, 61(3), 453 – 62. Brumfiel, E. M., 2001, Asking about Aztec gender: the historical and archaeological evidence, in Gender in prehispanic America: a symposium at Dumbarton Oaks, 12 and 13 October 1996 (ed. C. F. Klein), 57–85, Dumbarton Oaks, Washington, DC. Densmore, F., 1974 [1927], How Indians use wild plants for food, medicine, and crafts, Dover, New York. Drooker, P. B., 1992, Mississippian village textiles at Wyckliffe, The University of Alabama Press, Tuscaloosa. Endrei, W., 1968, L’évolution des techniques du filage et du tissage du moyenâge à la révolution industrielle, Mouton, Paris. Erichsen-Brown, C., 1979, Medicinal and other uses of North American plants, Dover, New York. Ericksen, A. G., Jakes, K. A., and Wimberley, V. S., 2000, Prehistoric textiles: production, function, semiotics, in Beyond cloth and cordage (eds. P. B. Drooker and L. D. Webster), 69–83, University of Utah Press, Salt Lake City. Franquemont, E. M., 1986, Cloth production rates in Chinchero, Peru, in The Junius B. Bird Conference on Andean Textiles (ed. A. P. Rowe), 309–30, The Textile Museum, Washington, DC. Huamán Poma, 1978 [1615], Letter to a king (trans. and ed. C. Dilke), E. P. Dutton, New York. Jakes, K. A., and Ericksen, A. G., 1997, Socioeconomic implications of prehistoric textile production in the eastern woodlands, in Materials issues in art and archaeology V (eds. P. Vandiver, J. Druzik, J. F. Merkel and J. Stewart), 281–6, Materials Research Society Proceedings 462, Materials Research Society, Pittsburgh, PA. Jones, V. H., 1937, Notes on the preparation and the uses of basswood fiber by the Indians of the Great Lakes region, Papers of the Michigan Academy of Sciences, Arts, and Letters, 22, 1–14. Jones, V. H., 1946, Notes on the manufacture of cedar-bark mats by the Chippewa Indians, Papers of the Michigan Academy of Sciences, Arts, and Letters, 32, 341–63. Kalm, P., 1937 [1753], Peter Kalm’s travels in North America, vol. 1, Wilson-Erickson, New York. Leeds, J. F., 1999, Making fast cordage, Bulletin of Primitive Technology, 17 (Spring), 37–42. Rangel, R., 1993 [1851], Account of the northern conquest and discovery of Hernando de Soto (trans. J. E. Worth), in The Desoto chronicles, vol. 1 (eds. L. A. Clayton, V. J. Knight Jr and E. C. Moore), 251–306, University of Alabama Press, Tuscaloosa. Rowe, J. H., 1979, Standardization in Inca tapestry tunics, in The Junius B. Bird Pre-Columbian Textile Conference (eds. A. P. Rowe, E. P. Benson and A. Schaffer), 239–64, The Textile Museum, Washington, DC. Samuel, A., 1985, Chilkat spinning, Threads, 1(October/ November), 55–9. Samuel, C., 1982, The Chilkat dancing blanket, Pacific Search Press, Seattle. Schiffer, M. B., and Skibo, J. M., 1997, The explanation of artifact variability, American Antiquity, 62(1), 27–50. Schreffler, V. S., 1988, Burial status differentiation as evidence by fabrics from Etowah Mound C, Georgia, Ph.D. thesis, The Ohio State University. Schwarz, A., 1947, The increased effect of the spindle, Ciba Review, 59, 21–65. Sibley, L. R., and Jakes, K. A., 1989, Etowah textile remains and cultural context: a model for inference, Clothing and Textile Research Journal, 7(2), 37–45. Sibley, L. R., Jakes, K. A., and Song, C., 1989, Fiber and yarn processing by prehistoric people of North America: examples from Etowah, Ars Textrina, 11, 191–209. Sibley, L. R., Jakes, K. A., and Larson, L. H., 1996, Inferring behavior and function from an Etowah fabric, in A most indispensable art: Native fiber industries from eastern North America (ed. J. B. Petersen), 73–87, University of Tennessee Press, Knoxville. An exploration of prehistoric spinning technology 307 Sigüenza y Góngora, C. d., 1939 [1693], Instructions to and journal of Don Carlos de Sigüenza y Góngora, in Spanish approach to Pensacola, 1689–1693 (ed. I. A. Leonard), 152–92, The Quivira Society, Albuquerque, NM. Sillar, B., and Tite, M. S., 2000, The challenge of ‘technological choices’ for materials science approaches in archaeology, Archaeometry, 42, 2–20. Smith, H. H., 1923, Ethnobotany of the Menomeni Indians, Bulletin of the Public Museum of the City of Milwaukee, 4(1), 1–82. Smith, H. H., 1928, Ethnobotany of the Meskwaki Indians, Bulletin of the Public Museum of the City of Milwaukee, 4(2), 189–274. Smith, H. H., 1932, Ethnobotany of the Ojibwe Indians, Bulletin of the Public Museum of the City of Milwaukee, 4(3), 348–433. Smith, H. H., 1933, Ethnobotany of the Forest Potawatomi Indians, Bulletin of the Public Museum of the City of Milwaukee, 7(1), 32–127. Speck, F. G., 1940, Penobscot man: the life history of a forest tribe in Maine, University of Pennsylvania Press, Philadelphia. Strachey, W., 1953 [1612], The historie of travell into Virginia Britania, The Hakluyt Society, London. Swanton, J. R., 1969, The Indians of the southeastern United States, Scholarly Press, Grosse Pointe, MI. Tiedemann, E. J., 2001, Characterization of prehistoric spinning technology: toward the determination of spinning practices employed in Mississippian textiles, Ph.D. thesis, The Ohio State University. Vallinheimo, V., 1956, Das spinnen in Finnland, Kansatieteellinen Arkisto 11, Suomen Muinaismuistoyhdistys, Helsinki. Vreeland, J. M. Jr, 1986, Cotton spinning and processing on the Peruvian North Coast, in The Junius B. Bird Conference on Andean Textiles (ed. A. P. Rowe), 363–83, The Textile Museum, Washington, DC. White, J. K., 1969, Twined bags and pouches of the eastern Woodlands, Handweaver and Craftsman, 20(3), 8–10, 36–7. Whitford, A. C., 1941, Textile fibers used in eastern aboriginal North America, Anthropological Papers of the American Museum of Natural History 38, part 1, American Museum of Natural History, New York.