207Chapter 8
Behavior and the Bones
Osbjorn M. Pearson and Jane E. Buikstra
I. INTRODUCTION
The reconstruction of the behaviors and lifestyles of prehistoric peoples from their skeletal remains and archaeological contexts constitute primary goals of bioarchaeology. Today bioarchaeologists attempt to meet these goals through a combination of biomechanical analyses, studies of osteoarthritis and trauma, and other observations (Larsen, 1997; Bridges, 1992, 1994b, 1996; Ruff, 1992, 2000; Hawkey and Merbs, 1995). The effort to use such data to produce an impression of prehistoric lifeways has become increasingly visible over the last three decades, owing its popularity to the influential work of a host of earlier researchers. J. Lawrence Angel was one of the earliest advocates of what has become the current approach, as illustrated by his description of three 9000-year-old skeletons from Hotu Cave, Iran:
Femoral neck torsion, tibial head tilting, gluteal crest development, platymeria, platycnemia, and stressed extensor and rotator muscle insertions form a complex [cf. Wagner 1927 (1926): 115–117] called the bent-knee gait, often misinterpreted. This applies to the use of the legs flexibly, like a skier, and not a posture. Stress on the ilio-tibial band, iliac crest, and lower and upper lumbar areas (possible herniation of lowest nucleus pulposus in number 2) suggests further that the Hotu women may have done some standing and working with braced legs (as pulling on a fish net) as well as much climbing in rough country, carrying, and digging. The injuries to the thumb-wrist joints and little finger of number 3 suggest possible fighting but more plausibly hard manual work perhaps more specialized than digging for roots: flint chipping, plaiting baskets, net-making, or possibly midwifery or shaminism. The pelves of the two women show enough bone reaction at ligament attachments and insertion of the abdominal wall muscles (rectus and external oblique aponecosis [sic]) to hint that pregnancy may have been frequent and without rest period. (Angel, 1952:265)
208Angel’s work on other skeletal samples such as the Archaic period remains from Tranquillity, California, further exemplified this holistic approach to behavioral reconstruction and allowed Angel to paint a detailed portrait of at least some of these people’s activities and to advance informed speculation about others:
The Tranquillity people show other postural specializations in the frequency of flexion facets at the ankle (80 percent) and in retroversion of the tibia in two out of four cases. Together with the marked femoral pilaster and platymeria, these suggest active running in rough terrain. In five out of nine cases the olecranon fossa floor is perforated, a condition linked with elbow hyperextensibility. As expected, four out of these five cases are female. This may relate to the general “economy of bone” which the Tranquillity people show: the shafts of all long bones are flattened about to the degree seen in Old World Paleolithic and other hunting populations and often show a sinuosity and extra sharpness of muscle attachments which approach the bowing of sabre shin seen in actual malnutrition. (Angel, 1966a:3)
II. ROOTS
Much of the recent work by bioarchaeologists to reconstruct the activity patterns and lifestyles of prehistoric peoples has followed Angel’s lead, but with attempts to incorporate improved methods, new approaches, and a wider comparative framework of populations for which homologous data are available. It should not be forgotten, however, that Angel also stood upon the shoulders of giants, and the roots of behavioral reconstruction are to be found much earlier in time.
Functional and behavioral interpretations of skeletal remains ultimately arose from anatomy and those trained in it, whether in England, Germany, or the United States. By the late 1800s, European physicians and anatomists followed one of two traditions: a traditional one that emphasized typology and classification and a relatively new one that focused on the plasticity and adaptability of the body over a lifetime. The second approach became almost synonymous with the name Rudolf Virchow, whose profound influence led him to be regarded as the father of the medical study of pathology.
By the late 1800s, Virchow — and, by extension, nearly the entire German anatomical and medical establishment — placed great emphasis on the plasticity of the body, including the skeletal system, in response to external forces. At the time, German academia also led the world in technological innovation and engineering, and the exuberance and vigor of this field of inquiry also influenced German anatomists. The most visible product of this intellectual 209cross-fertilization was the work of Julius Wolff on the structure and development of trabecular bone, research that formed one of the bases of what 20th-century researchers came to regard as Wolff’s “law” (for a historical summary, see Martin et al., 1998). Wolff originally formulated his proposition as a means of understanding how trabecular bone adopted an architecture that allowed it to resist mechanical stresses with a minimum amount of material. This “law” formed a homologue to models that mechanical engineers of the time were developing for iron trellis systems that could bear great loads with a minimum of material (Martin et al., 1998). This emphasis on plasticity and adaptation, characteristic of the German anatomists from Virchow’s day onward, greatly influenced the work of a number of important figures in anthropology, including Franz Boas, Rudolph Martin, Franz Weidenreich, and, more recently, Friedrich Pauwels, Adolph Schultz, and Holger Preuschoft.
Most contemporary anatomists in other European nations, Great Britain, and America could read German and were at least aware of the German emphasis on functional adaptation. Some, including Sir Arthur Keith, adopted a perspective heavily influenced by functional considerations (Keith, 1940); most, however, remained committed to more traditional, typological approaches to anatomy and, by extension, the nascent science of anthropology. In America, Aleš Hrdlička embodied and greatly advanced the traditional, typological approach toward morphology. Trained as a physician in the Czech Republic, Hrdlička was clearly cognizant of contemporary German anatomical studies, but his approach to anthropology was to remain firmly typological (see Chapters 1–3). Hrdlička’s work on the shapes of the femur and tibia included a typological categorization of shapes of the shafts (Hrdlička, 1898, 1934a,b), but his later work also included a perspective on the development of distinctive shapes of femoral shafts (Hrdlička, 1934a,b), a study of comparative shapes of homologous primate femora (Hrdlička, 1934d), and a comprehensive treatment of femoral third trochanters and hypotrochanteric fossae (Hrdlička, 1934c, 1937b).
Earnest Hooton exerted a strong influence on the development of bioarchaeology and functional interpretations of human remains, in part due to his detailed descriptions of the remains from Pecos Pueblo (Hooton, 1930; see also Chapters 1, 2, and 4). Hooton had trained in anthropology in England, where he was exposed to a broad range of research methods, including the new German focus on somatic plasticity, the early developments in biometry, and statistical descriptions of populations pioneered by Francis Galton and Karl Pearson, as well as the classic, typological approaches to morphology that still dominated British anatomy and were to form the basis of much, but not all, of Hooton’s work.
The first work in North America on behavioral interpretations of human remains preceded the more influential, later work on the topic by Hrdlička, Hooton, and others. Some of the earliest investigators realized what has become 210a dominant paradigm today: a comprehensive bioarchaeological approach to inferring behavior — individual or group — requires consideration of both archaeological contexts and human remains. The 19th-century Hemenway Expedition discussed earlier (see Chapters 1 and 5) serves as an early North American example. One of Cushing’s goals, influenced by his prior ethnological and archaeological experiences, was to study grave accompaniments in order to know the sex, the condition of life, and other facts about the individual. As mentioned previously, he believed this information would lead to “vivid, even historic knowledge of the people” interred at Los Muertos (Hinsley and Wilcox, 2002:200). In complementary fashion, Washington Matthews and colleagues (1893) were quite eager to infer behavior through the study of human bones. Noting that neither septal apertures of the humeri nor platycnemia occur in children, these authors argued that both conditions arose due to specific activities. For example, they inferred that grinding corn led to the development of septal apertures among women. They also took issue with Manouvrier’s (1888) deduction that platycnemia necessarily developed through hyperactivity of the tibialis posterior muscle and was necessarily or even frequently associated with hunting lifestyles on rough terrain (see also Kennedy, 1989; Ruff, 2000). They argued instead that behavioral interpretations of platycnemia should be based on a more broadly based consideration of biomechanical principles.
When the tibialis posticus assumes the inverse action, the tibia becomes a lever of the second class, with the fulcrum at the ankle joint, the power at the insertion of the muscle, and the weight (which in ordinary cases is but the weight of the body and the clothing) at the knee joint. There are three ways (besides frequency of impulse) in which the distance through which the lever moves, as in climbing hills; second, by diminishing the time in which it moves, as in running and jumping; third, by increasing the weight, as in lifting and carrying heavy loads. Largely to the third way we are inclined to attribute the prevalence of platycnemia among various American races, including the Saladoans. (Matthews et al., 1893:224)
Following such 19th- and earlier 20th-century scholarship, research that focused on functional and behavioral interpretation of human remains experienced a great acceleration from the 1970s onward. This increased interest has its roots in Washburn’s (1951, 1953) “New Physical Anthropology” and in the holistic conception of anthropology imparted by Hooton upon his students. With respect to bioarchaeology, the contributions of J. Lawrence Angel, Sherwood Washburn, and T. Dale Stewart loom large. At the close of the 20th century, interpretations of prehistoric people’s patterns of activity have been based on four primary forms of data: cross-sectional geometry, osteoarthritis and trauma, and muscle markings in addition to an assortment of other traces of behavior left on bones or teeth.
211III. INTERPRETING PREHISTORIC PATTERNS OF ACTIVITY
A. CROSS-SECTIONAL GEOMETRY
1. History and Application
Following a period of near invisibility, midcentury, biomechanical approaches once more assumed significance in late 20th-century interpretations (Bridges, 1985, 1989a; Bridges et al., 2000; Larsen, 1995; Larsen et al., 1995, 1996; Ruff, 1991, 1992, 1994, 1999, 2000; Ruff and Hayes, 1983a,b; Ruff et al., 1984; see also Chapter 13). The cross-sectional geometry of long bones of an animal are commonly upheld as one of the best indicators of the mechanical forces that the animal had adapted to resist in life, and thus a reasonable reflection of habitual activities (Ruff, 2000). Stimulated by the structural analysis of platycnemia (Lovejoy et al., 1976),1 researchers investigated topics such as mobility patterns and sexual division of labor across time and space in a variety of archaeological skeletal samples, basing their inferences on bone shape.
The thickness of limb bones of animals has long been of interest in functional morphology, from Galileo’s observations of allometric changes in animal limb bones to the present (Preuschoft, 1971; Wainright et al., 1976; Alexander, 1977; Pauwels, 1980; McMahon and Bonner, 1983; Currey, 1984; Schmidt-Nielsen, 1984; Currey and Alexander, 1985; Martin and Burr, 1989). Anthropological interest in the relationships between bone cross-sectional geometry and function largely grew out of the broader fields of biomechanics and functional anatomy as reflected in the work of Pauwels (1980). Early applications of beam mechanics to model the strength of human long bones were made by Pauwels (1980), Endo and Kimura (1970), Kimura (1974), Lovejoy et al. (1976), and Lovejoy and Trinkaus (1980), among others. The development of technology to digitize the cross sections of long bones and of computer programs such as SLICE (Nagurka and Hayes, 1980) that could rapidly calculate second moments of area from bone sections allowed the proliferation of studies of cross-sectional geometry during the 1980s and 1990s.
For bioarchaeologists, Ruff and Hayes’ (1983a,b) study of the cross-sectional geometry of the Pecos Pueblo femora and tibiae proved to be an influential landmark. The study was quickly followed by investigations of changes in limb bone 212cross-sectional geometry that had accompanied the shift to agriculture on the Georgia Coast (Ruff et al., 1984). This study corroborated Larsen’s (1981) earlier findings that a decline in femoral strength, a decrease in the development of the femoral pilaster, and an overall decrease in size accompanied the transition to agriculture in the same region. In the late 1980s and early 1990s, the notion that hunter–gatherers were taller, healthier, and led more physically demanding lives than later horticultural or more intensive agricultural populations became a widely accepted paradigm (e.g., Cohen and Armelagos, 1984; Larsen, 1982; Ruff et al., 1993). It is significant, therefore, that Bridges (1989a) described an instance from northern Alabama in which the transition to Mississippian agriculture failed to produce the expected pattern and instead found that the Mississippian males had stronger legs than their Archaic predecessors and that Mississippian females had both stronger legs and considerably stronger humeri than their Archaic counterparts, a change that was accompanied by a decrease in upper limb asymmetry.
Bridges (1989a) pointed to a variety of other studies (Pickering, 1984; Goodman et al., 1984; Lallo, 1973; Hamilton, 1982) that had suggested that bone size, muscle marks, or arthritis incidence — all of which tended to be treated at the time as nearly equivalent indicators of activity — provided additional evidence that changes in subsistence with agricultural intensification had required increasing amounts of labor and activity rather than the reverse. Bridges (1991a) soon reported that the comparison between frequencies of osteoarthritis in Archaic and Mississippian people from northern Alabama produced the opposite pattern of what the cross-sectional geometry indicated: the foragers had more osteoarthritis in their joints. For the time, Bridges showed a great sensitivity to such contradictions (see also Bridges, 1989b, 1990, 1991b, 1992, 1994b, 1996). Toward the end of her career, Bridges (1997) began to test the relationships between various traits taken to be indicators of activity, a research direction that foreshadowed one of the current forefronts of research and to which we return at the end of this chapter.
Additional studies of cross-sectional geometry of the long bones of prehistoric populations continued to appear at a rapid pace in the late 1980s and early 1990s. Prominent examples include Brock and Ruff’s (1988) study of changes in cross-sectional geometry in the American Southwest; Robbins and co-workers’ (1989) study of Late Woodland limb bones from Delaware; Fresia and colleagues’ (1990) documentation of the decline in the bilateral asymmetry of the humerus on the Georgia Coast; Larsen and colleagues’ (1995) report on the rugged skeletons from Stillwater marsh and other Great Basin sites (see also Ruff, 1999); Ruff’s (1994) description of extraordinary development of the femoral pilaster in femora from the southern Plains; Ledger and co-workers’ (2000) analysis of the limbs of 18th-century slaves from Cape Town, South Africa; and Stock and Pfeiffer’s (2001) documentation of substantial variation in limb bone structure between two groups of hunter–gatherers, Andaman Islanders and Precontact Khoisan from the Cape 213of South Africa. By no means is this list exhaustive and it reflects the visibility that cross-sectional geometry has achieved as the most highly regarded measure of activity patterns. In addition to work on recent populations, a large number of studies were devoted to the cross-sectional geometry of Upper Paleolithic people, Neanderthals, and still earlier hominins (Senut, 1985; Grine et al., 1995; Churchill et al., 1996; Holliday, 1997a; Pearson and Grine, 1996, 1997; Churchill and Formicola, 1997; Ruff et al., 1994, 1999; Ruff, 1995; Trinkaus and Ruff, 1999a,b; Trinkaus et al., 1991, 1994, 1999; Pearson, 2000; Holt, 2003).
Data for such studies were initially digitized from photographs of sectioned bones or from CT scans (Ruff and Leo, 1986), but Runestad et al. (1993) developed a method of molding the external contour of a bone, taking biplanar, orthogonal X-ray films of the bone and using the endosteal surface visible in the X-rays to approximate the endosteal contour of the section. The contour mould and X-ray method has been subsequently used in a large number of other studies (Churchill, 1996; Churchill and Formicola, 1997; Holliday, 1997a,b; Holt, 2003).
Likewise, the biomechanics of primate and human mandibles have been reproduced utilizing a beam model, with the cross section of the corpus acting as the beam section (Hylander, 1988; Daegling, 1989; Daegling and Grine, 1991; Dobson and Trinkaus, 2002). Furthermore, some studies of mandibular cross-sectional geometry have been able to compare their results to experimentally determined strains acting on the mandible (Hylander and Johnson, 1994; Chen and Chen, 1998; Daegling and Hylander, 1998, 2000; Daegling and Hotzman, 2003). Given the amount of data available for the bony structure of the mandible, the direction and magnitudes of the muscles that act upon it, and the amount of bite force that can be generated, it has also been possible to construct finite-element models of how human and primate mandibles and crania deform during mastication (Korioth et al., 1992; Richmond et al., 2005; Strait et al., 2005; Ross et al., 2005).
2. Criticisms of Cross-Sectional Geometry
During the late 20th century and into the 21st century, skeletal biologists began to question certain fundamental assumptions of biomechanical approaches to behavioral reconstructions. Concerns were expressed concerning uncritical acceptance of fundamental, 19th-century assumptions (Wolff’s “law”) and failure to recognize recent research in bone biology, especially “mechanobiology.” Tendencies to interpret nonsignificant results and to dismiss confounding variables were also cited (Bice, 2003). Lovejoy and colleagues expressed this concern:
A common assumption that has long pervaded interpretations of the hominid postcranium is that the distribution of bone, in both its cortical and cancellous forms, can be viewed as an uncomplicated “record” of the bone’s loading history. However, during the past decade, highly aggressive research protocols, together with their 214continual reintegration into novel theoretical approaches, have cast strong doubt on this presumption. It can no longer be used as a perfunctory basis for the direct interpretation of skeletal form. Too many data have accumulated which negate so simplistic an approach. (Lovejoy et al., 2002:97)
A variety of experimental studies have found that bones are not actually loaded or bent in the directions that anthropologists initially expected they would be and that the axis of bending does not pass through the centroid of area of the section as analytical programs such as SLICE (Nagurka and Hayes, 1980) assume it does (Gross et al., 1992; Demes et al., 1998, 2001; Lieberman et al., 2004). Both kinds of exceptions to expected functional patterns in bones constitute sobering findings for those who wish to use cross-sectional geometry in their reconstructions of the lives of prehistoric people. Jurmain (1999) also questioned the current utility of studies of the cross-sectional geometry of prehistoric people’s long bones to shed light on their patterns of activities because very few clinical studies have actually documented the effects of specific activities on the cross-sectional geometry of human long bones. Without such data from living subjects, interpretations from studies of ancient bones will likely remain only interesting speculation, regardless of whether such inferences seem plausible or not.
With regard to the problems created by the fact that bones can be loaded in directions we might not predict and that the neutral axis of bending may not pass through the centroid of area of a section (as we generally assume), Lieberman and colleagues (2004) found that the section modulus of a bone is likely to contain more error than other variables such as the torsional second moment of area (J). The section modulus (Z) of a cross section of a beam or bone is defined as the section’s second moment of area divided by the perpendicular distance from the bending axis to outermost point of bone mass in the section (Martin et al., 1998). The section modulus is currently (Ruff, 2000) considered the most useful—and most biomechanically meaningful (Martin et al., 1998)—cross-sectional property to analyze in skeletal material. In light of Lieberman and co-workers’ (2004) finding, anthropologists might be well advised to emphasize analyses of J standardized for body size, which was popular from the early 1990s until 2000 (Ruff et al., 1993, 1994; Larsen and Ruff, 1991; Larsen et al., 1995).
A final, recent development in the study of cross-sectional geometry that affects its utility in making behavioral inferences about prehistoric populations is the growing realization that bones may not model in response to exercise or habitual activity in the same way across the life span. Ruff and co-workers (1994) pioneered some of the recent interest in the ontogeny of cross-sectional geometry with a model of the ontogeny of femoral cross sections that hypothesized that activity during childhood would produce extra subperiosetal apposition and decrease the rate of endosteal resorption, whereas strenuous exercise during adulthood could produce endosteal stenosis but only a modest amount of additional subperiosteal deposition. A variety of recent studies have suggested that activity 215during childhood, especially during the adolescent growth period, appears to exert a more substantial influence on the size and shape of adult bones than exercise later in life (Kannus et al., 1995; Khan et al., 2000; Kontulainen et al., 2001; for a review, see Pearson and Lieberman, 2004). The implications of these findings have yet to be fully explored by bioarchaeologists, but the great variety of subsistence practices employed in prehistory offer a promising area of inquiry for studies of the ontogeny of bone shape and strength.
B. OSTEOARTHRITIS AND TRAUMA
1. Arthritis
While reports of arthritic change appeared during the 19th century, emphasis was frequently on describing the most extreme cases. For example, Langdon (1881:249) discussed the fusion of all thoracic and lumbar vertebrae in ancient remains from the Madisonville, Ohio, cemetery site, attributing the condition to arthritis deformans. Similarly, Whitney (1886:444) described remains of an older man recovered from a stone box grave near Brentwood, Tennessee: “both elbow joints are roughened and irregular and the surface in spots looks like ivory. His joints must have grated like a rusty hinge when he attempted to move them, and the stiffness and restricted motion must have been the same as is seen in the rheumatic cripple of to-day.” Thus, 19th-century observers emphasized description, diagnosis, and the degree to which behavior had been limited by the arthritic condition, not on the behaviors that might have caused the condition.
Comparative, population-based descriptive studies are found throughout the 20th century, e.g., Hrdlička (1914a), Stewart (1947, 1966), and Jurmain (1977a,b, 1980, 1990, 1991). Stewart’s research included age-related patterning as well as population comparisons for Native American skeletal series. He reported extreme arthritis in the lumbar vertebrae of Inuit peoples when compared to Pueblo Indians. Extending this comparison to the knee, hip, elbow, and other joints, Jurmain (1977a) described more severe arthritic changes along with earlier onset for Alaskan Eskimos. Ortner (1968), concentrating upon the elbow, also concluded that arthritic change in Inuit remains was more extreme than that observed in Peruvians.
Angel’s (1966a) study of 35 Archaic period skeletons from the Tranquillity site, mentioned earlier, provides a vivid example of one of the best of the early studies of the pattern of arthritic degeneration to draw inferences about living conditions and labor. Angel wrote:
There is plenty of evidence that they lead strenuous lives. All four preserved vertebral columns show fully developed hypertrophic arthritis in cervical and lumbar regions and one shows a healed fracture at waist level plus herniation of the disk nucleus into the body of the fourth lumbar vertebra. This degree of wear and tear of the disks and 216ligaments at the age of 25–40 is typical of hardworking populations (Gejvall, 1960, ch. VIII; Nathan, 1962; Stewart, 1958a) and one or two decades ahead of our vertebral column aging. (Angel, 1966a:3)
Several additional points of interest are illustrated by the preceding quotation: Angel’s profound knowledge of human anatomy; his close, collegial association with T. Dale Stewart, whose deft studies of vertebral anomalies and pathologies influenced both Angel and the subsequent adoption of the entire field of paleopathology; and Angel’s familiarity with contemporary studies in Europe.
Angel’s work on the Tranquillity remains became an oft-cited landmark study that proposed an explicit link between osteoarthritis (OA) and specific behaviors (see also Chapter 11). In this report, Angel noted “6 of 13 people have arthritis in the elbow joint, usually including eburnation after friction removal of cartilage over the capitulum” (Angel, 1966a:3). Consideration of possible causes for the high frequency of this pathology led Angel to the idea that throwing darts from a spear-thrower (or atlatl, to use the Aztec word) might be the cause. He wrote:
The spear thrower, of course, puts extra stress on the arm muscles and elbow. Hence it seems logical to describe this special pathological change as “atlatl elbow.” Laughlin (1963), Stewart, Merbs, and others have noted it among the Alaskan Eskimo and Aleut. It is less frequent in female skeletons. But it does occur in two out of four Tranquillity females even though the arthritic lipping is slight. Possibly seed-grinding has some effect. It is equally likely that a genetic weakness or avascularity of the joint plays a part in small and isolated populations. This is given point by the frequency of a similar elbow avascular necrosis in baseball playing Japanese, as opposed to Westerners (Nagura, 1960). (Angel, 1966a:3)
Angel noted that other throwing actions should also cause shoulder and clavicular stresses, not observed in this sample. Angel’s term, “atlatl elbow,” for the condition proved to be influential in many subsequent studies (e.g., Jurmain, 1977a; Bridges, 1990). Angel attributed the pathology in females to seed grinding using a mano and metate, inspiring Merbs (1980) to coin the term “metate elbow” for it. Recognizing that genetic factors can also influence patterning, Angel’s research combined both focused observations within joints and considerations of overall patterning.
As noted by Jurmain (1999), the use of OA to infer behavior became less popular during the final decade of the 20th century. This may in part be attributable to critical evaluations, such as those of Bridges (1992:80): “while arthritis is undoubtedly related in part to forces placed on the joint, it is not a straightforward indicator of the level or type of normal activities.” Jurmain (1990, 1991, 1999) concurs, concluding that the most productive approaches appear to be those that investigate patterning in multiple joints, using the total available skeletal sample as the database (see also Rothschild, 1995; Waldron, 1994). Other matters of concern are nonstandard data-recording protocols and the absence of statistical testing (Bridges, 1992).
217Jurmain’s (1999) review of the clinical and epidemiological literature on living people showed that such studies provide only ambiguous and contradictory evidence for the link between activity and OA. Instead, injury to joints, which may or may not be a predictable consequence of certain activities, emerges as the most important risk factor for the development of OA later in life. As Jurmain (1999) notes, it is clear from clinical studies that many joints are able to sustain vigorous, long-term loading from distance running and other activities without developing osteoarthritis (Hoffman, 1993; Panush and Lane, 1994; Lane et al., 1993). Jurmain (1977b) described distinct patterns of age of onset of OA in various joints in different populations, and today it appears that developmental age and activity interact in complex ways to produce OA:
Some joints (elbow and hip particularly, as compared to other joints) appear to be under differential risk, given the age of the onset of mechanical loading; early injury and/or modification of joint mechanics can produce OA changes later in life. (Jurmain, 1999:105)
Likewise, a variety of studies suggest that different joints may develop OA in response to dissimilar stimuli: “the knee appears to be most pone to activities involving repetitive bending, while the hip and spine appear to be more at risk as the result of heavy lifting” (Jurmain, 1999:105). If reinforced by future findings, such results will mean that the observation of OA in different locations in a skeleton may reveal different types of information about the physical activities of the person rather than providing a gauge of overall levels of activity. The clinical literature is replete with contradictory and complex findings about the associations between osteoarthritis and activity, however, and Jurmain’s words of caution are worth repeating:
The association of OA with specific activities is not clearly supported in contemporary contexts by either the occupational or sports literature. Further, the implications for and limitations on osteological interpretations are obvious. (Jurmain, 1999:105)
Part of the difficulty in applying clinical studies of risk factors and OA to bioarchaeological studies arises from the fact that clinicians usually define OA in a different way than anthropologists. In clinical settings, erosion of the cartilage in joints, damage to subchondral bone, and narrowing of the joint capsule are used to diagnose OA, whereas many osteologists’ definitions have included the development of osteophytes around the joint capsule, a phenomenon that is not of clinical relevance unless the osteophytes interfere with the joint’s function (Jurmain, 1999). Osteologists should score the two phenomena separately (Buikstra and Ubelaker, 1994). It remains likely that some — and perhaps much — of the OA that anthropologists have attributed to “activity” is in fact due to activities across the life span, but some is almost certainly due to injuries, and more research and more caution in drawing conclusions from traces of OA are both clearly warranted.
2182. Trauma, Including Spondylolysis
During the 19th century, studies of paleopathology in ancient Native American remains typically included “injuries” as one of three categories, with the others being “anomalies” and “diseases” (Matthews et al., 1893; Whitney, 1886). Some, such as Whitney (1886:436), distinguished between fractures and dislocations, although fractures were the most commonly described injury. In contrast to researchers’ early preoccupation with the impact of arthritis upon activity, fractures, especially cranial fractures, were both described in exquisite detail and attributed causally to aggressive behaviors. Observed within the Madisonville sample, for example, was a partially healed, extensive fracture that retained a depression “just above the ear which nicely fits one of the round-headed stone hammers found in the cemetery” (Langdon, 1881:252). From a small sample drawn from across North America, Whitney (1886:439) felt that in three examples of cranial fractures “there was a strong presumption in favor of their being due to intentional violence. The seat, the left side of the head, especially favors this view, as it presupposes that the persons who gave the blows were right-handed.” Arrow wounds were also reported (Langdon, 1881).
Twentieth-century bioarchaeological inquiry continued to report evidence of both intentional and accidental trauma (see also Ortner and Powell, 2006). While descriptive reports occurred throughout this period, comprehensive comparative studies appeared relatively late in the century, most postdating Lovejoy and Heiple’s (1981) influential attempt to establish age-specific fracture patterns in the late prehistoric Libben site (Ohio) skeletal sample.
A 2001 summary of the history of violence by Walker described considerable variation across time and space in the Americas, as also reported by Ortner and Powell (2006). Walker and Lambert’s extensive, contextualized studies of trauma, for example, identified increased violence during the Middle Period for the Santa Barbara Channel islands, a time of resource stress (Lambert, 1994, 1997; Walker, 1996). Late prehistory saw little violence in some locations, while chronic warfare apparently caused the death of at least one-third of the adults interred at the Norris Farms (Illinois) Oneota site (~AD 1300; Milner, 1995; Milner et al., 1991). The roughly contemporaneous Crow Creek Massacre site (South Dakota) provides ample evidence of traumatic death and violent, postmortem treatment of nearly 500 individuals, including scalping and dismemberment (Willey, 1990; Willey and Emerson, 1993; Zimmerman et al., 1981). Ortner and Powell (2006) emphasize that scalping clearly predates European contact, documented as early as Middle Archaic times (Mensforth, 2001; Smith, 1995, 1997).
Additional traces of human behavior have been described from other portions of the body. The nonmasticatory use of teeth as tools received scholarly attention during the early 20th century (Leigh, 1925a), an interest that has been maintained 219since that time (Milner and Larsen, 1991; Larsen, 1997). Transversely oriented occlusal grooves were noted in anterior teeth of several hunter–forager groups from Texas, the Great Basin, California, and British Columbia (Bement, 1994; Larsen, 1985; Schulz, 1977). The Great Basin samples were studied through scanning electron microscopy, which revealed multiple fine scratches following the main axis of the groove. It has been suggested (Larsen, 1985, 1997) that some form of flexible material, such as sinew or plant fibers, was passed repeatedly over the teeth. Notching and lingual surface wear associated with extramasticatory functions have also been reported for groups from Texas (Hartnady and Rose, 1991), Tennessee (Blakely and Beck, 1984), and the Georgia Coast (Larsen, 1982).
As noted in the first section of this chapter, 19th-century scholars such as Wyman (1875) considered cannibalism a likely explanation for the archaeological recovery of fragmented human bone that had been treated in the same manner as faunal remains. This subject again assumed marked visibility through the work of White (1992) and Turner (1983; Turner and Turner, 1999), who focused on evidence from the Greater Southwest. Both scholars concentrated on developing detailed protocols for identifying evidence of cannibalism, including evidence of burning, cut marks, pot polish (smoothed surfaces due to boiling), and fragmentation patterns. While alternative explanations were proposed, including the destruction of social deviants such as witches (Darling, 1993, 1998; Dongoske et al., 2000; Ogilvie and Hilton, 1993; Martin, 2000), the recovery of human myoglobin from a human coprolite recovered archaeologically from the Cowboy Wash site (Colorado) demonstrated that at least one person consumed human flesh in the ancient Southwest, ca. AD 1150 (Marlar et al., 2000).
Merbs’ (1983, 1995, 1996a) work on degenerative changes among the Inuit was both influential and showed the potential of careful study of trauma to elucidate patterns of prehistoric activity. Merbs’ work included a careful consideration of spondylolysis, including sacral spondylolysis (Merbs, 1996a), and, ultimately, a rigorous exploration of the etiology of spondylolysis (Merbs, 1996b). This work allowed Merbs to make interesting interpretations:
Sacral spondylolysis was a relatively common phenomenon in Alaskan and Canadian males during late adolescence and early adulthood but . . . the condition would correct itself, leaving a permanent record only in those unlucky enough to die young. Although the unusually vigorous activity patterns of these males appear to have been a major cause of the stress fracturing that produced the spondylolysis, specific (but largely unspecified) anatomical variations and delayed vertebral maturation may also have been significant contributors. (Merbs, 1996a:365)
An explicit focus on the reconstruction of habitual behavior led Merbs (1969, 1983) to also investigate osteoarthritis, along with osteophytosis, compression fractures, spondylolysis, and anterior tooth loss. Working with historic period Canadian Inuit (Sadlerimiut) remains from Southampton Island, 220Northwest Territories, Merbs formulated explicit behavioral expectations based on ethnohistoric accounts. These expectations guided his behavioral reconstructions, an approach that received widespread recognition and approval, (e.g., Bridges, 1992; Jurmain, 1999).
C. Musculoskeletal Stress Markers
We close this review with a discussion of musculoskeletal stress markers (MSMs), which are also commonly called enthesopathies, entheses, or, more coloquially, muscle markings. Along with cross-sectional geometry, many anthropologists consider MSMs and OA in joints and the vertebral column to be indicators of activity patterns and a reflection of skeletal responses to its mechanical environment (Jurmain, 1977a, 1980; Kennedy, 1989; Larsen, 1995; Hawkey and Merbs, 1995). The expression of both OA and MSMs tends to become more common and more pronounced with age (Jurmain, 1977a, 1980, 1999; Dutour, 1992; Hawkey and Merbs, 1995; Wilczak, 1998; Wilczak and Kennedy, 1998; Weiss, 2003a,b, 2004).
The early history of observation of MSMs by no means achieved the degree of precision and specificity that researchers have sought to achieve since the early 1990s, but less formalized or systematic observations of muscle markings constituted part of the examination of skeletal remains from the early days of American physical anthropology. Hrdlička (1937b), for example, penned a detailed account of structural variants associated with insertion of the gluteus maximus and offered a comprehensive summary of the etiologies that had hitherto been proposed for the development of those features. Perhaps the most influential observer and interpreter of the significance of muscle markings was J. Lawrence Angel. From his early work (e.g., Angel, 1946a), Angel displayed an acute sensitivity to what variations in both overall skeletal morphology and areas of tendon attachment might reveal about prehistoric lifeways and activity patterns. An early example of this sensibility may be found in the following passage from his description of the three Epipaleolithic indviduals from Hotu Cave, Iran:
The upper surfaces of the tibiae are tilted more than usual and the laterally compressed shafts of the shinbones have a diamond-shape cross section. Fibulae are deeply fluted. The femora are distinctly platymeric or thickened transversely in the upper shaft as if to take stress from strong abductor and lateral rotator muscles. The deep gluteal fossae adjacent to marked crests, the strong adductor tubercles, the stressed origin areas for gastrocnemius, and on the tibiae the increased origin area for deep muscles supporting the arches of the feet confirm the suggestion that muscles involved in rough-country travel were well-developed. (Angel, 1952:259)
Likewise, Charles Snow (1974) paid careful attention to the development of muscle markings in his description of pre-contact Hawaiian skeletons from 221Mokapu, Oahu. Snow’s text paints an evocative portrait, as for example his summary of femoral muscle markings:
Almost all of these bones show well-developed pilastering of the linea aspera. This buttressing, reinforced bony ridge was strong evidence for well-developed flexor and extensor muscles. . . . The bone relief of the trochanteric region was bold and showed extensive muscular areas. Likewise, in the popliteal region at the back of the knee, the adductor tubercle was very well developed. (Snow, 1974:47)
Snow was fortunate to have detailed accounts of the daily habits, work, recreation, and other physical activities of Hawaiians from the period of contact that he could use to draw links between behaviors and the osteological traces of heavy musculature that he observed. Such close attention to ethnographic accounts of labor and activity patterns have informed some of the best analyses of other ostensible markers of activity, including Ruff and Hayes’ (1983a,b) analyses of the cross-sectional geometry of the Pecos Pueblo limb bones, Merbs’ (1983, 1996a) work on trauma and degenerative disease among the Inuit, and Bridges’ (1989a) account of changes in the cross-sectional geometry of limb bones in Indians from Northern Alabama during the transition from foraging to agriculture. However, while Bridges’ (1989a) work illustrates the judicious use of ethnographic accounts, it also illustrates another problem with interpreting patterns of prehistoric activities: everything about the activities of the ancient foragers in Alabama must be inferred and thus are not “known.” This problem becomes exacerbated in progressively more ancient societies and may be particularly problematic in Paleolithic societies, which experienced living conditions, including surprisingly low population densities (Stiner et al., 1999, 2000), that may not have a close historical analog.
Returning to the present, other studies of MSMs have made use of accounts of labor conditions to enrich interpretations of the pattern of observed muscle markings. In an influential article on the life stresses of slavery, Kelley and Angel (1987) combined observations of muscle markings, patterns of arthritis, and historical information about living conditions and diet to interpret the pattern of morphology in skeletal remains. Their study constitutes an early, systematized attempt to quantify and compare the development of muscle insertions of enslaved ironworkers from Catoctin Furnace, Virgina. The resultant picture of life and activity patterns could be painted with broad strokes:
Our best evidence for occupation and related pathology is from the Catoctin site. The muscle crests we compare are the deltoid, pectoral, teres, and supinator (see Figs. 3–5). The former are, of course, involved in the lifting of heavy objects. Their development in teenagers or young adults females indicates heavy work of a type not common to twentieth-century females. In combination with shoulder or vertebral breakdown, including separated L5 arch, and schmorl herniation, the picture of hard, heavy labor is substantiated. (Kelley and Angel, 1987:207)
222While reports of osteoarthritis in behavioral reconstructions declined in the late 1990s, attention turned to MSMs. Use of the atlatl and similar behaviors were inferred by Kennedy (1983) to be related to hypertrophy of the ulnar crest to which the supinator muscle attaches. Studying the relative development of such “enthesopathies” (tendinous insertions or ligamentous attachments) became increasingly popular for behavioral inferences during the 1990s (Jurmain, 1999).2 Hawkey and Merbs (1995:325) caution that such markers are ideal “for a study of activity-induced changes in a population” only in large, well-preserved skeletal series, preferably those dating to a relatively narrow time span where cultural and genetic isolation and a limited number of specialized, known activities exist. Another concern in the use of MSMs for behavioral inferences is that there is little scientific evidence that directly links enthesopathies to specific activities (Jurmain, 1999; Robb, 1994; Ruff, 2000).
Spurred by the examples presented by Angel (1952, 1966a; Angel and Kelley, 1986; Angel et al., 1987; Kelley and Angel, 1987), Kennedy (1983, 1984, 1989), and others (e.g., Dutour, 1986, 1992) of the power of MSMs to provide grist for the mill of interpretation of prehistoric lifeways, work on more rigorous methods for quantifying and comparing MSMs began in earnest in the late 1980s and continued vigorously through the 1990s. Hawkey and Merbs (1995) produced an influential study of MSMs in Hudson Bay Inuit from two time periods, the “Early Thule (Classic Period)” and “Later Thule (Transitional/Historic).” Based in part on Hawkey’s master’s thesis (Hawkey, 1988), Hawkey and Merbs’ study of this population has become a landmark in the study of MSMs. The methodology they used to quantify MSMs has been widely adopted — with and without modifications — by many subsequent studies (e.g., Steen and Lane, 1998; Weiss, 2004). Key aspects of the method include assessing each muscle origin or insertion site for three features: robusticity markers, stress lesions, and ossification exostosis. Each is scored along an ordinal scale with photographs and descriptions to guide the researcher in making allocations (Hawkey and Merbs, 1995). In this protocol, “robusticity” generally refers to the overall size and prominence3 of the origin or insertion area, “stress lesions” usually refer to resorptive pitting in an attachment site, and “ossification exostoses” denote small spurs of ossified ligaments or aponeuroses protruding from the attachment site. These three features were then combined into an overall ranked score of expression that placed the least weight on “robusticity” and the most weight on the degree of development of “stress lesions” (Hawkey and Merbs, 1995).
Although Hawkey and Merbs’ (1995) methodology proved highly influential, a large variety of other methods, many of them less clearly or precisely defined, 223have also been proposed for the quantification of MSMs. Among the best-defined methods are those of Wilczak (1998), who quantified attachment areas by digitizing chalk outlines of insertion areas, and of Robb (1998), who advocated a system of seriation of MSMs from least to most pronounced. Most studies find more pronounced muscle marks in males than in females, even when controlling for age.4 The different methodologies for scoring MSMs have also produced some interesting, conflicting results that suggest that additional work is needed to clarify how closely they correspond and under what circumstances they will tend to produce differing results. For example, Hawkey and Merbs (1995:326) reported very little correlation with age among adults, noting that “[a]lthough a gradual increase in attachment robusticity was noted from young to middle to adult, the differences were not significant statistically, and all adult samples were pooled.” Using the same methodology for scoring MSMs, Elizabeth Weiss (2003b, 2004) found significant correlations with age and bone length in both the humerus and the lower limb. Likewise, Wilczak (1998) reported a complex set of correlations between insertion size and age.
D. CRITICISMS OF ENTHESOPATHIES, OSTEOPHYTES, AND OSTEOARTHRITIS
Since 1995, there has been a large increase in publications on MSMs and an even larger number of presentations on muscle markings at the annual meetings of the American Association of Physical Anthropologists (e.g., Munson Chapman, 1997; Steen and Lane, 1998; Churchill and Morris, 1998; Peterson, 1998; Lovell and Dublenko, 1999; Molnar, 2003; Pany et al., 2003; Toyne, 2003). Despite the surge in interest in MSMs, there are very few clinical studies that have actually linked MSMs, and their degree of development, with specific activities (Jurmain, 1999), largely because the osteophytes interpreted as MSMs by osteologists generally do not cause discomfort to living people and are thus not of clinical significance. However, a few researchers are now focusing on the problem of how activities in life correlate with the development of pits and osteophytes in MSM development (Zumwalt et al., 2000; Zumwalt, 2004).
Many questions about MSMs still remain to be answered. Do repetitive activities or overuse injuries cause MSMs? Do occasional, high-stress activities produce MSMs and are such infrequent, high-magnitude strains more likely to produce MSMs than more repetitive but lower-strain activities? Are there individual differences in the risk of developing MSMs after performing specific activities? Are there population-level differences in the probability of developing 224rugged muscle insertion sites in response to performing specific amounts of given activities? Are there age effects so that activities performed at a young or old age have dissimilar probabilities of influencing the expression of MSMs? Until more is known about the etiology of MSMs, interpretations of what they show about prehistoric activities will necessarily remain speculative, however logical that speculation may seem.
So far, there have been very few ontogenetic studies of the development of MSMs from childhood into adulthood. The literature contains more ontogenetic studies of OA, and these show low frequencies in early adulthood followed by increasing frequencies later (Jurmain, 1999). Jurmain (1999) has urged anthropologists to pay special attention to the age of onset of OA in specific joints in comparisons between sexes and populations. A problem with studying age of onset arises from the fact that in clinical studies, injury to a joint, particularly injury in childhood, repeatedly emerges as a major risk factor for OA later in adulthood (Micheli and Klein, 1991; Jurmain, 1999). Most people survive from mid-childhood to early adulthood (Wood et al., 2002). As a result, most osteological series contain very few skeletons of juveniles older than about 5 years of age, making it very difficult to accurately assess the probability of injury to joints. The upshot for osteologists is that the best way to solve the problem of etiology of OA will be via more clinical research on living people. Studies of archaeological populations may also prove invaluable, but it is doubtful that they will ever be able to match the diagnostic ability of clinical studies in which many more factors such as body mass, actual activity patterns, diet, history of injuries, and the like can be accurately measured and taken into account.
IV. CONCLUSIONS
The reconstruction of prehistoric lifeways and activity patterns from skeletal remains has been one goal of physical anthropologists from the very origin of the discipline in the United States. Key early influences on American physical anthropologists and anatomists primarily included contemporary British, French, and German anatomists, who often worked under differing research paradigms, yet were also generally mutually aware of each other’s work. In particular, under the direction of Virchow from the 1850s onward, the “German school” of anatomy placed great emphasis on the plasticity of tissues, including muscle and bone, to environmental factors, including work and activity. Many other anatomists, importantly including Hrdlička, remained firmly rooted in the older tradition of typology, which today has much less importance in the functional interpretation of skeletons than the paradigm championed by Virchow and his students.
The approach to reconstructing behavior espoused by most modern bioarchaeologists perhaps owes its origin to the combination of the holistic approach to 225skeletal anatomy fostered by J. Lawrence Angel, a student of Earnest Hooton, and the development and subsequent surge in popularity of studies of the cross-sectional geometry of bones. Modern methods include cross-sectional geometry, patterns of osteoarthritis, musculoskeletal stress markers, trauma, and other observations. Virtually all of these data sets are problematic, as thoughtfully critiqued by Jurmain (1999), and considerable work remains to clarify which of these features provide the best indications of activity and how the various forms of data are interrelated.
Bridges’ (1989a, 1991b, 1997) work highlighted the fact that cross-sectional geometry, osteoarthritis, and the development of muscle markings might not be closely correlated and might, in fact, not be interchangeable indicators of activity. Rather, her work suggested that these aspects of skeletal morphology might arise from differing influences. More work on skeletal and living populations is clearly needed to elucidate how the various forms of data are intercorrelated as well as what activities are responsible for the development in life of the features that we can observe in skeletal populations. Encouragingly, some workers have already taken additional steps in this direction, including Churchill’s (1996) factor analysis of upper limbs, which included measures of cross-sectional geometry, muscle lever and load arms, and other dimensions in Neanderthals, early modern humans, and a series of recent comparative populations. Likewise, E. Weiss’ (2004) work on the intercorrelations among MSMs scored via the Hawkey-Merbs’ method, cross-sectional geometry, body size, sex, and age stands as a very useful study of how these properties are interrelated. More studies of living people and the factors that we assume have generated the patterns of cross-sectional geometry, osteoarthritis, and MSMs in prehistoric populations are badly needed. Physical anthropologists have cause to feel optimistic at this juncture: all of these studies are feasible and will undoubtedly serve to enrich our understanding of the lives of our ancestors.
In sum, although bioarchaeological studies of behavior have failed to establish signatures for specific activities, group-level inferences have compared and contrasted groups with different lifeways. Sexual division of labor has also been addressed, as have topics such as cannibalism and the extramasticatory use of teeth as tools. While the goal of behavioral reconstruction is central to 21st-century bioarchaeology, researchers must also pay attention to the need for rigor in their studies and not fall into the “activity-only myopia” decried by Jurmain (1999).226