Now that we have laid the theoretical and empirical groundwork for an evolutionary approach to the corrugation problem, we can turn to the goal of explanation. The logical structure of evolutionary theory and explanations derived from that theory, discussed in Chapter 3, requires that both replication and variation exist in the empirical entities under investigation. The presence of replication and variation establishes the necessary and sufficient conditions for descent with modification. Once these conditions exist, differential replication of variation can occur through a variety of stochastic and deterministic sorting mechanisms. The nature of the interaction of different variants with their environment determines which sorting mechanisms are causally relevant to explanations of any particular empirical case.

Evidence regarding the development and spread of corrugation in the American Southwest, presented in Chapter 5, indicates that corrugation developed through a sequence of innovations that introduced new variation into the repertoire of techniques used to make utility ware vessels. The evidence also documents that these innovations were copied with sufficient fidelity to produce clear lineages of descent or replication of corrugation techniques. Our current knowledge suggests that a variety of replicators may have been involved in the spread of corrugation. These possible replicators include the pottery itself, the knowledge of how to make different forms of corrugation, the meaning attached to corrugation, the biological reproduction of the people possessing different utility ware technologies, and larger scale sociocultural systems. This diversity of potentially relevant replicators suggests that adopting multiple frames of reference may be necessary for formulating a complete account of changes in corrugated pottery in the Southwest.

Evidence presented in Chapters 6 and 7 indicates that characteristics of the utility ware environment and the ways different plain and corrugated utility wares interacted with their environments changed during the development and spread of corrugation. This, together with the rapid adoption of neck banding and full-body corrugation, suggests that deterministic sorting mechanisms were probably involved in at least some aspects of the shift from plain to corrugated utility wares. However, these observations do not rule out the possibility of a significant role for stochastic sorting mechanisms as well.

The evidence presented in the previous three chapters also shows that the question "why corrugation?" with which I began this research is too simplistic to serve as a basis for developing explanations of the complex history of corrugation in the American Southwest. We can not understand why ancestral Puebloan populations used full-body, indented corrugation on their utility wares without understanding how and why this technique of corrugation developed in the first place. In addition, it is quite possible that explaining different aspects of the corrugation problem will require shifting our frame of reference and employing alternative sorting mechanisms. To more adequately approach the explanation of corrugation, I have divided the problem into four parts based on our current understanding of the rise and fall of corrugation and the context in which these changes took place. These four parts are the initial adoption of neck banding, the experimentation and elaboration of neck banding, the adoption of full-body corrugation, and the eventual return to plain-surfaced utility wares among Pueblo populations. In addressing each of these components of the corrugation problem, I begin by summarizing the evidence from Chapters 5 through 7. I then attempt to identify potentially causal relationships from different frames of references that may account for the patterns of differential persistence and spread of particular corrugation techniques.

Advent of Neck Banding

The technique of leaving structural coils unobliterated first appeared on jar necks in the American Southwest during the seventh century AD. Before this, all utility wares were scraped plain. This earliest neck banding occurred in brown ware pottery from the southern Mogollon region of New Mexico. At the same time, utility pottery produced in the Puebloan areas to the north consisted of scraped plain gray wares made primarily in the forms of bowls and narrow-mouth jars or ollas. Formal cooking jars, identified by the co-occurrence of a wide-mouth jar form with substantial soot accumulations, were relatively rare in these early Basketmaker assemblages. In addition, cooking with pottery involved a wider variety of vessel forms than in later times. Both these lines of evidence suggest that cooking in pottery jars was neither as common nor as formal as it would become later.

The earliest appearance of neck banding in the northern Southwest occurred in the early to middle part of the eighth century, 50 to 100 years after its emergence in the southern Mogollon region. This early date for neck-banded pottery applies to the southern edge of the Colorado Plateau, areas west of the Chuska Mountains, and the northern San Juan basin, including the Mesa Verde region. However, east of the Chuska Mountains in the southern San Juan basin and the Rio Grande valley, neck banding did not appear until the latter half of the ninth century. Deteriorating climatic conditions during the late ninth and early tenth centuries brought about a substantial southward movement of people out of the Mesa Verde region and other parts of the northern southwest. Contemporary population increases in the south including the southern San Juan basin, the southern Chuska Valley, and parts of the Rio Grande region, all east of the Chuska Mountains, are almost certainly a result of immigration from the areas being partially or wholly abandoned. This migration coincides with the first appearance of neck banding in the areas east of the Chuska Mountains.

Regardless of the region or the age of first appearance, the earliest neck banding consists of relatively wide, flattened coils applied without overlap (i.e., filleted). Rather than being an example of independent invention or convergent evolution, the strong similarities in all of these early neck-banded forms suggest that the innovation of leaving coils unobliterated probably spread north from the Mogollon region through migration, diffusion or a combination of the two. Regardless of how the trait spread, its delayed appearance in the southern San Juan basin and Rio Grande valley indicates that these regions may have been isolated from the contacts necessary for transmission until the middle of the ninth century.

At the time neck banding was introduced into the manufacturing repertoire of ancestral Puebloan peoples of the Colorado Plateau, these potters were forming their vessels by stacking up relatively wide coils or slabs. Leaving some of these construction elements unobliterated on the necks of jars added little or nothing to the cost or performance of the vessels. Because the exposed coils were wide and not overlapped, very little additional roughness was created on the exterior surface that could have improved handling. Although the weld between exposed, filleted coils is weaker than between coils that have been scraped together, the necks of vessels are not exposed to intense stress during most uses. Thus, the reduced strength of neck-banding probably did not present a significant problem. Consequently, the initial adoption of neck banding may have been as a decorative elaboration involving little or no change in labor investment.

The preceding paragraphs conform well to a traditional approach to explaining the rise of neck banding that combines culture historical and processual notions. A style of decoration, neck banding, appeared, perhaps by chance, and spread over a vast area through the migration of people and the transmission of the trait through various forms of interaction and communication. Although this may be a satisfactory account in some ways, important questions remain unanswered. For example, why neck banding and not some other decorative techniques such as incisions, impressions and appliquŽs? Further, why did neck banding spread when it offered no apparent advantage to the makers and users of the pottery? Kidder (1936) tried to answer such questions by stating that people are simply compelled to decorate. However, this assertion does not advance our understanding of the situation. Does an evolutionary approach offer any further insight into these questions?

Let us look first at the question of why neck banding and not other decorative techniques. At the time neck banding first arose in the southern Mogollon region and spread to the western Anasazi areas, other techniques for creating decorative elaborations do not appear to have been part of the repertoire of these potters. Incising did appear later in these areas, but only after neck banding was well established. In the Rio Grande and Gallina areas to the east, incising and neck banding appeared at about the same time, and, in these areas, incised and punched decorations competed well with neck banding for many years. This suggests that the dominance of neck banding was simply a historical accident. However, incising also arose in the Mogollon area and competed with neck banding quite well even after neck banding had already become established.

Perhaps subtle distinctions in the cost of the various techniques for decorating utility wares played a role in their differential success in particular areas. Neck banding requires no additional cost. In fact, neck banding may actually save a very small amount of time during the forming of the vessel in comparison to entirely plain vessels by eliminating the need for obliterating the coils in the neck of jars. In contrast, the other possible decorative techniques all require some additional effort after the entire vessel has been scraped plain, and added technology in the form of tools, materials and techniques used to create the decorations. Although these cost differences were probably quite small, they may have been sufficient to inhibit, but not completely exclude, the encroachment of the other decorative techniques after neck banding was already well established.

Although a combination of history, or chance, and subtle cost differences may account for the dominance of neck banding in the northern Southwest, we still must answer the question of why neck banding or any other decorative technique was adopted at all. The temporal and spatial patterns of appearance of neck banding in the northern Southwest and other evidence presented in Chapter 5 indicate that migration played a prominent role in the spread of the technique. As people moved to new areas, they brought with them the techniques they learned in other areas. However, unless the immigrants completely replaced the indigenous populations, which seems unlikely given the continuities in other local material culture, neck banding must have also spread by diffusion from the immigrants to the indigenous people. In evolutionary terms, neck banding apparently replicated independently of, or in addition to, the replication of biological organisms and ethnic groups.

It is difficult to identify anything inherent in neck banding itself that would promote its replication over entirely plain-surfaced utility wares. Neck-banded vessels were not superior to plain vessels in some way that would have triggered people's cognitive algorithms for evaluating and rationally selecting innovations on the basis of performance benefits or cost reductions. However, I can imagine three other possible explanations that I present briefly here. Unfortunately, thoroughly testing these hypotheses will have to wait until we know more about the migrations that took place during the eighth and ninth centuries, and the nature of any adaptive differences and patterns of interaction that may have existed between migrant and indigenous populations.

Perhaps there was something in the social context in which neck banding spread, or in the meaning ascribed to neck banding that contributed to its adoption. At the time neck banding spread into the northern Southwest, most people lived in small, dispersed settlements, which probably consisted mainly of close kin. Under these conditions, exogamous marriages are essential for the long-term viability of populations. Arranging such marriages requires the establishment and maintenance of relationships with people in other settlements. As migrant populations moved into new areas, perhaps the back and forth transfer of innocuous, decorative traits signaled a willingness to engage in cooperative interactions including intermarriage. If the migrant and indigenous populations were small, establishing such cooperative relationships may have been essential for biological survival. Consequently, the spread of neck banding and possibly other decorative traits may have indirectly enhanced the biological reproductive success of both migrant and indigenous populations. However, the spread of neck banding occurred too fast for these possible impacts on biological reproductive success to account for the adoption of neck banding directly. Instead, the environment with which utility wares interacted changed in a way that promoted the replication of neck banding because it facilitated potentially difficult, yet essential, social interactions.

Another possibility is that, for one reason or another, the migrant populations were seen as especially good models by the indigenous populations. Perhaps the migrant populations were particularly successful relative to the indigenous people because of some adaptive advantage. If so, this may have resulted in an indirect bias (sensu Boyd and Richerson 1985:241-279) toward the adoption of traits possessed by the migrant population irrespective of their adaptive value. This could have created a strong selection for neck banding as a cultural replicator despite its lack of direct impact on biological fitness.

It is also possible that the spread of neck banding was a consequence of drift rather than selective sorting. In other words, stochastic processes acting in small populations led to the fixation of neck banding even though it was selectively neutral. In this case, patterns of interaction and the chance adoption of traits because of these interactions led to the spread of neck banding. In encounters between people possessing neck-banded and scraped plain technologies, the simplest model of transmission indicates that each individual or group would have a 50% chance of adopting the technology used by the other. Under these conditions, a purely random increase in the abundance of neck banding could have led to its complete adoption in small populations as an increasing proportion of interactions resulted in the transmission or replication of neck banding. In addition, if the migrating populations entering the northern Southwest were large relative to the indigenous population, then neck banding could have spread relatively rapidly as most encounters would have involved exposure to neck banding.

Experimentation with Neck Banding

During the ninth and tenth centuries in the Mesa Verde region and other areas of the Southwest, neck banding increased in frequency and diversity. The use of neck banding increased both in terms of the relative frequency of wide-mouth jars possessing the trait and the extent of vessel surface with exposed coils. By the late ninth century, most wide-mouth jars displayed exposed coils, and on some, these coils extended down past the neck and onto the upper body of the vessel. We also see at this time the beginnings of variation in the techniques of neck banding with the appearance of narrower filleted coils in the early ninth century and slight to moderate overlapping of adjacent coils in the late ninth century.

Among the small population remaining in the Mesa Verde region in the middle tenth century, variation in the techniques of neck-banding continued to increase with the addition of rippling and sporadic indenting of exposed coils. This elaboration usually occurred as a narrow decorative band just below the rim. The relative frequency of narrow and substantially overlapped coils also increased, becoming the most abundant form, and the use of exposed coils on the upper body of vessels below the neck continued to expand. Incising of coil junctures, primarily on narrow filleted coils, also increased during this time, resulting in greater surface roughness.

In the southern San Juan basin area, the middle to late tenth century also exhibits a zenith in variation of neck-banding and corrugation techniques. Although the first use of wide, filleted neckbands occurred much later in the southern San Juan basin, utility ware type descriptions for the area indicate a similar pattern of development to that seen in the Mesa Verde region. In fact, the routine use of systematic indenting may be earlier in the southern San Juan basin than other areas of the northern Southwest.

Although the goal of these innovations appears to have been decorative embellishment, producing narrower, more elaborate neckbands rather than improved performance, it was with these innovations that neck-banding began to affect the cost and performance of vessels. One means of producing narrow bands, the overlapping of adjacent coils, created a rougher surface and increased the strength of coil welds in comparison to the various forms of filleted neckbands. The rougher surface would have made handling the vessels easier particularly when they were hot. The increased weld strength of overlapped coils probably also facilitated the extension of exposed coils farther down the vessel body. The application of exposed coils below the vessel neck and, more importantly, below the upper level of the heated liquid contents of the vessel would have resulted in a slight loss in heating efficiency, but also significantly improved cooking control through reduced boilovers. The addition of systematic indenting or pinching of substantially overlapped coils further improved the strength of welds between adjacent coils, and increased the external surface area of the vessel, which may have resulted in even greater control over cooking. However, the use of more, narrow or substantially overlapped coils over a greater portion of the surface would have also increased the cost of vessels by requiring more time in the forming stage of manufacture.

The use of utility wares also appears to have changed during the latter half of the ninth century and the tenth century, at least in the Mesa Verde region. The increasing focus of utility ware production on a single vessel form, wide-mouth jars, and the growing relative abundance of cooking-related use alterations attest to the expanded importance of cooking in the use of utility wares. Not only does cooking appear to have become the primary use of utility wares in the Mesa Verde region by the tenth century, but the intensity and formality of use in cooking may have also increased. This increased focus on cooking in jars may have resulted from the introduction into the Southwest of Mais de Ocho, a variety of corn that is more suitable for milling than were earlier flint varieties. If more corn was being milled, as an increase in the relative abundance of milling stones suggests, rather than roasted on the cob, then the cooking of corn meal into mush or stew was a natural consequence in the absence of technology for bread making.

Increased moist cooking of agricultural products high in starch such as corn and beans would have had important consequences for populations employing this food processing strategy. Nutritional studies of the effects of cooking on starch digestibility demonstrate that cooking significantly increases the digestibility of starches (Bishnoi and Khetarpaul 1993; Collings et al. 1981; Jenkins et al. 1982; Stahl 1989; Wandsnider 1997). In an unprocessed form, starch occurs as more or less complex glucose-based polymers. These complex molecules resist hydrolysis by our digestive enzymes, and, consequently, can pass through our intestinal tract relatively undigested. Cooking, particularly moist cooking, aids digestion by breaking down the complex starch molecules, through gelatinization and pasting, into their simpler and more readily hydrolyzed constituents. In addition, moist cooking removes or neutralizes antinutrients, such as phytic acid, polyphenols, and saponins, which inhibit absorption. Consequently, boiling or simmering of food, particularly for long periods, would have increased the nutritional yield of a given quantity of food. Studies show a 50 to 100 percent increase in in vitro starch digestibility for intensely cooked grains over uncooked starches with the amount of increase depending on the kinds of precooking processing, such as soaking (Bishnoi and Khetarpaul 1993; Hellendoorn et al. 1970). In addition, porridge produced by this cooking method could have been used to wean infants at a younger age which may have resulted in shorter birth spacing, or increased workload for women in activities other than child rearing (Buikstra et al. 1987; Nerlove 1974).

Skeletal evidence indicates that Pueblo populations in the northern Southwest during the ninth century may have experienced both better nutrition and early weaning. Ninth and early tenth century populations in the Mesa Verde and Black Mesa (northeastern Arizona) areas had slightly greater life expectancy (higher mean age at death) than later populations in the same areas suggesting better nutrition (Martin 1994; Martin et al. 1991; Nelson, Martin, Swedlund, Fish, and Armelagos 1994; Stodder 1987). Younger ages for the formation of tooth enamel defects or hypoplasias, which are frequently associated with weaning stress in young children, during the ninth century in comparison to the twelfth and thirteenth centuries in the Mesa Verde region indicate that early weaning may well have been a common practice during the earlier period (Stodder 1987). A similar pattern in hypoplasias among early and late populations on Black Mesa in northeastern Arizona (Martin et al. 1991) suggests that early weaning may have been common during the ninth and possibly tenth centuries in the northern Southwest. Both better nutrition and shorter birth spacing would have contributed to the tremendous population growth that occurred in the Mesa Verde region during the ninth century (Schlanger 1986, 1988; Wilshusen and Varien 1996). Unfortunately, a very sparse skeletal record from the preceding Basketmaker periods precludes comparing the data from the ninth and tenth centuries to earlier populations.

In addition to affecting the reproductive success of the biological populations, the increased importance of moist cooking in the ninth and tenth centuries changed the environment with which the utility ware vessels interacted. Greater use in cooking meant that thermal stresses became a more significant factor in determining the use-life, and performance during cooking may have begun to outweigh other performance characteristics. Because of their impacts on cooking control and ease of handling, increasing the surface area of exposed coils and extending the use of exposed coils farther down the vessel body may have come under heightened positive selection during this time.

Although some of the innovations in neck banding introduced during the ninth and tenth centuries could have enhanced the reproductive success of the people that used them, these changes are difficult to explain in terms of these biological impacts alone. Actual increases in fecundity cannot explain the changes in neck banding because there was not sufficient time for significant population growth to occur. Too few biological generations could have occurred during the later half of the ninth century and the tenth century for the effects of moist cooking and improved cooking control to be felt unless these activities improved reproductive success by a factor of 4 or more. Although settlement data indicate that populations did grow in some areas during this time (e.g., Schlanger 1986), we lack estimates of actual changes in fecundity. However, in a study of skeletal populations from the Midwest that date to before, during, and after an increase in the consumption of boiled starchy seeds, Buikstra and others (Buikstra and Konigsberg 1985; Buikstra et al 1986, 1987) estimate a fertility increase of approximately 1.5 associated with this change in diet. It is very unlikely that a similar change in diet together with the beneficial innovations in neck banding could have resulted in substantially greater fertility in the Southwest.

By shifting the frame of reference to the cognitive and behavioral aspects of the organisms rather than their actual reproductive success, we gain more insight into this period of increased innovation in neck banding. One might begin with the common assumption that necessity is the mother of technological innovation, and try to explain the experimentation with neck banding as a response to the altered use environment. As moist cooking with ceramic vessels became more important, the potters tinkered with how they made their pots in an attempt to improve their performance. However, both overlapping of adjacent coils and indenting first appear as decorative elaborations. Overlapping coils was just one way potters used to produce narrower neckbands, and indenting was initially sporadic and limited to only a few of the exposed coils. Although some of the innovations in neck banding did have consequences for how the vessels performed during use, these consequences were likely unintended. In addition, these innovations in neck banding techniques also increased the cost of producing the vessels without any offsetting improvements in durability. Thus, it appears that the changes in neck banding techniques that occurred during the ninth and tenth centuries were not conscious design adjustments motivated by utilitarian or economic concerns. Perhaps only the extension of exposed coils below the neck of vessels can be accounted for in these cost-benefit terms.

However, strict cost-benefit assessments of individual neck banding variants is not the only calculus that may have been involved in bringing about this period of heightened experimentation. In the remainder of this section, I discuss four models that specify different conditions under which the rate of innovation can increase without regard to the specific utilitarian benefits accrued by individual innovations themselves. In the first model, population size is the sole determinant of innovation rate. The second model examines the relationships between risk sensitivity, innovation rates, and productivity. The third model focuses on the meaning of innovation in a social context and the conditions that can produce fluctuations in the cost of innovations both within and between social groups. Finally, the fourth model considers the role of learning curves in determining innovation rates.

Both logic and data indicate that a relationship exists between population size and innovation rate. If innovations occur randomly and at some more or less constant per capita rate, then the more people there are the more innovations there are likely to be. A similar exponential curve for global population increases and increases in the number of innovations suggests that this logical relationship may have been an important cause of gross changes in frequency of innovations over the course of human history. However, I am doubtful, based on current evidence, that this relationship held for the period of increased innovation in neck banding techniques during the ninth and tenth centuries. Demographic reconstructions indicate that the period between AD 880 and, at least, 940 saw substantial population decline in the Mesa Verde region and rapid population growth in the southern San Juan basin (Dean et al. 1994; Wilshusen and Varien 1996). The Mesa Verde region assemblage analyzed for this study from the middle of the tenth century (5MT8371) shows the greatest variety of neck banding techniques. Published data suggest that similar increases in the variety of neck banding occurred in the southern San Juan basin at the same time. Given that migrations account for at least some of the regional differences in population trajectories, it is difficult to obtain a clear picture of overall population size for this period. However, I find no clear evidence to suggest that the late ninth and early tenth centuries was a period of sizable population growth for the northern Southwest as a whole. Consequently, it does not appear at this time that the increase in neck banding innovations during this time could have been caused directly by population growth in the absence of an increase in the per capita innovation rate.

The second model suggests that the per capita innovation rate may have increased during the late ninth and early tenth century as a result of changes in the sensitivity to risk on the part of populations in the northern Southwest during that time. Recently, Fitzhugh (1999) has attempted to model innovation rate as dependent not on population size, but on the productivity and/or consumption of resources. Based on microeconomic and ecological theory, this approach models marginal utilities, the relationship between a unit of production and the utility or value derived from it, as a sigmoid or logistic function shown in Figure 47 (see also Blurton Jones 1987; Boone 1992; Kohler and Van West 1996; Smith 1988; Smith and Boyd 1990; Winterhalder 1986, 1990). In the lower portion of the curve, marginal utility increases as productivity increases until the inflection point after which marginal utility decreases as productivity increases. The risk sensitivity of individuals, defined as susceptibility to variance in outcomes, is determined by their position on the marginal utility curve (Figure 47). An individual with productivity in the area of increasing returns will be more risk-prone while an individual in the area of diminishing returns will be more risk-averse. This is because a greater variance under conditions of increasing returns is more likely to yield beneficial outcomes than a higher variance in the region of diminishing returns.

Figure 47. A model that shows the relationship between sensitivity to the risks of innovation and productivity under conditions of increasing and diminishing returns.

In Fitzhugh's (1999) formulation, people experiencing low productivity, and thus in the increasing returns portion of the marginal utility function, are more likely to increase variance through innovation than people with higher productivity and diminishing returns. Consequently, innovation rates should be relatively high during periods of resource stress, as long as sufficient energy and time are available for investment in innovations, and relatively low during periods of high productivity. This prediction matches the temporal pattern of innovation in neck banding. During the late ninth and early tenth centuries, environmental degradation reduced agricultural productivity across the northern Southwest resulting in southward migrations and the abandonment of some areas. This is also the period of greatest innovation in neck banding techniques. The stress on pottery producers and consumers may have also been high during this period. Increased reliance on moist cooking during the ninth and tenth centuries probably reduced the use-life of cooking pots, as evidenced by a possible increase in accumulation rate (see Figure 20). This would have required a greater investment in pottery production to maintain household inventories at a suitable level. This heightened demand for pottery production at a time of decline in agricultural productivity may have focused at least some of the increased innovation on the pottery itself. However, it is unclear how the apparent decorative nature of most of the neck banding innovations fits with this model.

The third model is based on the handicap, or costly signaling theory developed by the biologist Amotz Zahavi (1987, 1991; Zahavi and Zahavi 1997) and formalized by Graffen (1990a, 1990b). This theory attempts to explain elaborate displays or decorations in the biological and cultural worlds as signals of competitive ability or some other hidden feature of the signaler. The amount of investment made in signaling varies in proportion to the ability of other potential competitors, the composition of the audience, and the likely payoff from signaling. The advantage of costly signals is in the response that they engender in the receivers, which depends on their ability to interpret the signals accurately. If costly signals result in avoidance of unnecessary conflict or promote beneficial cooperation, then there can be an advantage for both the senders and the receivers of the signals.

Recently, some anthropologists have begun to use this theory to explain variation in elaboration or waste exhibited by human behavior (e.g., Bird and Smith 1999; Boone 1997, 1999; Neiman 1997, 1999). Although most of these applications have focused on monuments and other massively elaborate constructions, there is no reason for the theory of costly signaling to be limited to such cases. Even small differences in cost can produce the same effects under the right conditions. The increased investment in neck banding innovations, particularly during the tenth century, may represent just such a case.

Using narrower coils, overlapping adjacent coils and applying indents and incisions all add to the cost of manufacturing neck-banded pottery, and thus could provide a basis for costly signaling. Although we lack data on spatial variation in the use of these different techniques that would allow us to assess the applicability of costly signaling, temporal trends in the investment in these techniques do appear to match the costly signaling model. In the Mesa Verde region, investment in neck banding elaboration increased gradually during the ninth century and then rose dramatically during the tenth century. During the late ninth century, the first large villages formed which probably included unrelated or distantly related households. This period of village formation could have elevated the potential for competition because the likelihood of competition among individuals and groups increases with the decreased degree of biological relationship. Under these conditions, we would expect an increase in costly signaling, perhaps with the senders being the potters and the receivers being potential mates.

The significant increase in the variety of neck banding during the tenth century, including substantial decrease in exposed coil height and the first use of indenting, also occurred during a period when the importance of costly signaling was probably on the rise. The large-scale migrations at the end of the ninth century and the apparent coalescence of populations in the southern San Juan basin brought many people together who had previously been isolated from one another. Again, the association of unrelated individuals with no historical knowledge of one another could have prompted an increased investment in costly elaborations to signal competitive advantage and skill. Comparisons of the energy investments in pottery and possibly other material culture from different households and settlements occupied at the same time could be used to test this hypothesis more fully.

The final model of change in innovation rate draws on the metaphor of an adaptive landscape or design space to explain technology trajectories or learning curves. Economists have noted for some time that technological innovations follow a consistent trajectory with a rapid innovation rate occurring early in the history of a new technology followed by a diminished rate of innovation despite continued investment. Recently, Stuart Kauffman and his colleagues at the Santa Fe Institute have attempted to explain this pattern of innovation as a consequence of adaptive search on a rugged but correlated landscape formed by the space of possible designs (Kauffman 1995:201-206; Kauffman and MacReady 1995; Kauffman et al. 1998). The basic idea is that after an innovation opens a new area of design space, the innovation rate rises as people explore that new space in a more or less random fashion. These innovations during this initial period of exploration tend to come from the top down, finding major variations on the theme that involve changes in several attributes at once. These innovations represent relatively long jumps across the space of possible designs. However, as people discover these new design themes, it becomes progressively harder to make further innovations. Consequently, innovation becomes increasingly difficult and more focused on small changes that involve very short moves in design space. This pattern in the introduction of variation is seen in both technological and biological change and probably results from more general constraints on the adaptive exploration of rugged yet correlated landscapes.

It is possible that the initial development of neck banding opened up a new area of design space in the construction of ceramic vessels among early Puebloan populations. Leaving coils unobliterated moved from being in the domain of the possible to that of the actual. It did not take long after the initial innovation for further innovations on the neck-banding theme to follow. Making narrower coils, overlapping coils, applying indents, and various combinations thereof represent major variations on the theme that occurred during the ninth and tenth centuries. Once these major innovations had occurred, a series of small, incremental changes, such as degree of overlap and spacing of indentations, could result in further, short distance exploration of this new area of design space. Thus, the florescence of innovation in neck banding during the ninth and tenth centuries may represent a simple combinatorial expansion of actual design space, which was unrelated to any external conditions. These new variants on the neck banding theme could have arisen through a blind innovation process requiring no prior knowledge of the consequences of the innovations. In situations with conflicting design constraints, as is the case with ceramic cooking pots (see Chapter 7), and limited engineering knowledge, almost certainly the case for Pueblo potters circa AD 900, blind variation can be equally or perhaps more effective than rational calculation at finding solutions to design problems.

I doubt that any one of these possible explanations can account fully for the increased variation in neck banding during the ninth and tenth centuries. Instead, I suspect that we can explain different aspects of this variation in distinct ways. For example, we may be able to explain the application of corrugations over a greater amount of the vessel surface by their performance benefits when used on cooking pots. Innovations that increased the cost of making utility wares without providing some performance benefit may have been selected because conditions arose that favored costly signals. Other innovations may have been selectively neutral. These may have arisen because of an increase in blind innovations caused by environmental stress or exploration of new areas of design space. These neutral variants would have been sorted by drift. The apparently stochastic fluctuations in the relative frequencies of different kinds of neck banding shown in Figure 13 suggest that most of these trait combinations were selectively neutral. However, certain attributes that make up these trait combinations do show directional change indicative of selection. These attributes include the frequency and degree of coil overlapping and the extent of the vessels covered with exposed coils. However, selection appears to have become much more important at the scale of attribute combinations after the tenth century with the development and spread of full-body corrugation.

Adoption of Full-Body Indented Corrugation

Within a very brief period, the tremendous variation that had characterized neck-banded assemblages during the tenth century was reduced to the almost exclusive use of a single corrugation technique by the early eleventh century. This technique involved the substantial overlapping and systematic indentation of exposed coils. Using this technique increased the strength of the joins between adjacent coils and, consequently, made more feasible the use of corrugation over more of the vessel surface. Shortly after the adoption of these techniques during the late tenth century, the first full-body corrugated vessels appear in the record.

Full-body corrugation spread extremely rapidly over a large area of the Colorado Plateau. Full-body corrugated vessels appear to replace neck-banded utility wares in the southern San Juan basin, Mesa Verde, Kayenta, Sinagua, and Virgin Branch regions nearly synchronously during the early to middle eleventh century. Some of this spread of corrugation coincides with the reoccupation of areas that were abandoned during the tenth century. The spread of full-body corrugation beyond these areas took longer, and its replacement of earlier technologies was apparently less abrupt and complete in some areas such as the Rio Grande region to the east and the Mogollon and Sinagua regions to the south. In some areas, such as the Gallina region of north-central New Mexico, it appears that corrugation was never adopted to any significant degree. These differences in the spread of full-body corrugation may indicate variation among areas of the Southwest in patterns of population expansion, interaction, and, possibly, economic pursuits during the eleventh century.

Extending the use of exposed coils to the lower body and basal portions of cooking jars results in a more durable vessel by reducing the amount of thermal stress generated during use. In the Mesa Verde region, this improved durability of cooking pots appears to be reflected in a reduced accumulation rate of utility wares. This may signal a decrease in the labor investment required for utility ware production, but the increased cost of fully corrugated vessels may have offset the savings engendered by improved durability. The adoption of full-body corrugation also marks the first time for which we have secure evidence of large scale, probably economic, production and distribution of utility wares in the northern Southwest. This appears in the massive import of corrugated vessels into Chaco Canyon from the Chuska Valley. In some areas, increased investment in production of white ware vessels, which were used for storage and serving, accompanies the adoption of full-body corrugation.

These changes in the environment in which utility ware vessels were manufactured and used would have made full-body corrugation a benefit to the people possessing the technology. The improved durability of full-body corrugated vessels when used for cooking may have made moist cooking of food more efficient than it had been previously, further boosting the caloric return of this method of food processing. Substantial growth of Puebloan populations during the eleventh century may have been supported, in part, by this improved efficiency of food processing and shorter birth spacing made possible by the early weaning of infants. Improvements in climate and intensified agricultural practices also contributed to the population growth seen during this period of Pueblo expansion. However, the rapid adoption of full-body, indented corrugation to the almost complete exclusion of earlier neck-banded varieties over a large area of the northern Southwest cannot be explained by increased fecundity alone. As with some of the changes in neck banding, there were too few biological generations during the period of adoption for the impacts on fertility to account for the widespread use of full-body corrugation.

The greater efficiency of full-body corrugation may have been one factor that led people to choose this technique over the less efficient neck banding. In addition to this greater efficiency, the longer use-life of full-body corrugated vessels may have compensated for, or perhaps enabled the increased investment in white ware production that took place in some areas. Thus, with the trade of pottery on the rise, as it was in at least some areas, individual potters and pottery making groups may have benefited from the adoption of full-body corrugation of their utility wares. By concentrating utility ware production on a technology that extended the use-life of vessels, fewer vessels would have needed to be produced. This would have allowed them to shift more of their production time to other kinds of pottery. Modern psychological studies indicate that people do have the cognitive abilities necessary for making these kinds of cost-benefit calculations (Cosmides 1989; Cosmides and Tooby 1992). Consequently, this kind of decision-making algorithm may have been responsible for the rapid selection for full-body corrugation.

However, discontinuities in the spread of full-body corrugation suggest that this explanation may not yield a complete understanding of the adoption problem. Although the adoption of full-body corrugation was extremely rapid over a large area, it occurred later and was apparently less abrupt in other areas such as the Rio Grande valley, the southern edge of the Colorado Plateau, and farther south in the Mogollon region. If full-body corrugation was so clearly beneficial, why was the technology not adopted more rapidly in these other areas as well? One possibility is that the conditions that made full-body corrugation beneficial-routine use of moist cooking and, possibly, extensive trade in pottery or investment in producing painted pottery-were not present in these other areas. Further research on the use of utility wares and pottery exchange in these peripheral areas is needed to evaluate this possible explanation for the delayed adoption of full-body corrugation in some areas.

Another possible explanation is that patterns of interaction among different regions of the Southwest structured the spread or replication of knowledge regarding the manufacture and benefits of full-body corrugation. The area of rapid and complete adoption of full-body corrugation may have been involved in frequent interactions, which facilitated the replication of this variety of corrugation. If interaction between this core area for corrugation and adjacent areas was not as frequent or was constrained in some way, this could have inhibited the replication of full-body corrugation beyond the core area.

The apparent development of full-body corrugation in the southern San Juan basin during the late tenth and early eleventh centuries coincided with two events that may have affected the patterns of large-scale interaction in the American Southwest. First, the substantial growth of Pueblo population during the eleventh century discussed above also involved migrations of people into new areas and areas that had been abandoned during the late ninth and early tenth centuries. These migrations may have been sufficient to encourage the replication of full-body corrugation in the areas receiving the migrant populations, and inhibit the spread of corrugation beyond the area of the migrations if migration was the primary mechanisms of dispersal.

The second event involved the emergence of the Chaco regional system. Although the exact nature of this system remains obscure, a variety of evidence suggests that Chaco Canyon may have been the center of the first truly large-scale, integrated sociopolitical system in Pueblo prehistory (Lekson 1996, 1999; Lekson et al. 1988). If the establishment of the Chaco regional system took place through the replication of a suite or system of political, religious, economic, and technological traits, as some have argued, full-body corrugation may have been just one part of this larger replicating entity. Perhaps the simultaneous development of full-body corrugation and the Chaco regional system, likely a historical accident, led people to ascribe some meaning to corrugation in relation to the Chaco system that went beyond its utilitarian benefits. In evolutionary terms, full-body corrugation may have benefited from the successful replication of the larger scale Chaco system. Beyond the area covered by this regional system, full-body corrugation spread more slowly because it lacked the additional semantic replicator that it had within the Chaco system. The use of indented corrugation on some bowl exteriors among the Anasazi during the Chacoan period, but not in the later Pueblo III period, may have occurred because of this added semantic component to replication.

Current evidence suggests that both migrations and the development of a large-scale sociocultural system affected the spread of full-body corrugation. To the east and south, the area of rapid adoption of full-body corrugation corresponds with the apparent edge of the Chaco system. An abrupt boundary to the spread of corrugation exists to the east, which roughly corresponds with the edge of the San Juan basin. Although innovations in painted designs appear to have moved easily across this boundary (Wilson 1995), full-body corrugation did not. This suggests that the spread of corrugation farther to the east may have been limited by something other than simple interaction frequency. One possibility is that the meaning of corrugated wares differed from the painted pottery and this distinction obstructed the spread of corrugation. Another possibility is that the populations living in the Rio Grande valley and other areas east of the San Juan basin had not yet adopted moist cooking as their primary means of food preparation. If true, full-body corrugation may not have offered the benefits that it did in areas where moist cooking was common. Full-body corrugation was eventually adopted in the Rio Grande area during the twelfth or early thirteenth century. This adoption, however, was not as complete as in other areas to the west, and earlier plain, incised, and neck-banded styles continued to be used in the Rio Grande area after the appearance of full-body corrugation.

To the south, full-body, indented corrugation spread rapidly to the boundary between the Anasazi and Mogollon culture areas where the spread slowed abruptly. It was not until after the collapse of the Chaco system in the twelfth century that full-body corrugation continued to spread south into the Mogollon region. Corrugation appears to have spread rapidly after about AD 1150 along with masonry pueblo-style architecture and black-on-white painted pottery. Southward migration during the twelfth and thirteenth centuries may have again played an important role in the further spread of corrugation, but this has not yet been demonstrated (Lekson 1993, 1996b; Reid 1989; Reid et al. 1996). In the Mogollon region, plain, neck-banded and incised utility wares continued to be used after the introduction of full-body corrugation, and indented corrugation occurs frequently on bowls and jars. If corrugated vessels were commonly used for tasks other than cooking in the Mogollon region, as appears to have been the case, then corrugation may have spread to this area for reasons other than its benefits as a cooking pot. Perhaps some meaning attributed to corrugation played a role in its replication through the Mogollon area. The nature of this meaning and its relationship to the collapse of Chaco are intriguing subjects for speculation.

To the west and north of the San Juan basin, full-body corrugation appears to have spread rapidly during the eleventh century well beyond the area commonly recognized as participating in the Chaco system. Perhaps migrations associated with the Pueblo II expansion of agricultural populations in the northern Southwest accounts for some of this rapid spread of corrugation. However, migration does not appear to explain the spread of corrugation into the Fremont areas of Utah and Nevada. Among the Fremont, other utility ware styles continued to be used along with corrugation and indented corrugation appears on a wider variety of vessel forms than on the Colorado Plateau. We still know so little about the production and use contexts of corrugation in these areas that formulating adequate explanations poses a formidable challenge.

Return to Plain Utility Wares

After its adoption, full-body corrugation remained the dominant Puebloan utility ware for another 400 years. During this time, corrugation persisted through several dramatic changes in Pueblo society including large-scale population movements, some involving the abandonment of entire regions, and momentous changes in climate, social organization, and ideology. Although not explicitly a focus of my empirical research, we do know that the techniques used to manufacture full-body corrugated vessels did not stagnate during this period of dominance. Innovations came mainly in the areas of vessel morphology and the elaboration of patterns created by the placement of indentations and appliques. In addition, small amounts of plain gray vessels continued to be made throughout the period of dominance by corrugation. However, many of these plain vessels were actually produced by smearing or obliterating corrugations.

In the late fourteenth or early fifteenth centuries, the long-standing yet sporadically employed technique of smearing corrugations became much more common and widespread; eventually replacing indented corrugation. By the middle to late fifteenth century, the smeared or blind corrugation was itself replaced by entirely plain vessels. Unfortunately, we still know very little about this change back to plain-surfaced utility wares. However, existing evidence does indicate that changes in the processing of corn occurred at about the same time as the shift back to plain cooking pots. The adoption of flat bread making in the fifteenth century may have significantly reduced the amount of corn processed by moist cooking in pots. If so, then the additional costs of producing corrugated cooking pots may not have been adequately returned through the benefits of control over boiling and improved vessel durability.

Because we still know so little about the return to plain utility wares, I will not try to explain this change as fully as I have attempted to do with the other three aspects of the corrugation problem. Instead, I want to address an aspect of the shift back to plain vessels that has perplexed me and other pottery specialists working in the Southwest. This is the question of why Pueblo potters went to the trouble of making corrugated pottery only to smear and obliterate the coils after they had applied them. From a purely economic standpoint, this extra step is difficult to explain. If one is going to make plain-surfaced pottery, it would be less costly to use large coils or slabs as they had done before the advent of corrugation.

However, smearing corrugated coils as a part of a transition back to plain pottery may be understandable from an evolutionary perspective in which the different ways to make utility ware vessels are the replicators. During the period of experimenting with neck banding techniques, each different technique was a viable replicator. Hence, coil size, amount of overlapping (if any), indentation, incising, etc. all varied more or less independently of one another resulting in a wide variety of combinations. After the adoption of full-body indented corrugation, only a single replicator remained from among the diverse array present in earlier assemblages. Not only did the diversity of techniques decline precipitously, but the scale of the replicator may also have changed. Instead of each different technique replicating independently, a suite of techniques, including substantial overlapping and systematic indenting over the entire exterior surface, replicated as a single unit. Consequently, when the environment of utility wares changed during the fourteenth and fifteenth centuries, which may have eliminated the benefits of full-body corrugation, potters could not draw on earlier techniques because they were no longer viable replicators. To get back to a plain-surfaced utility ware, the potters had to start from a corrugated vessel first. By the time the Spanish arrived in the Southwest during the sixteenth and seventeenth centuries, smearing corrugations to make a plain utility ware pot was no longer practiced. Sometime during the fifteenth or early sixteenth centuries, innovations must have occurred that replaced corrugation resulting in a less costly utility ware technology.

Return to the Table of Contents

Go to Chapter 9