\"Apples and Oranges? A Cross-Cultural Comparison of Burial Data from Northeast China.\" Journal of East Asian Archaeology 3(2001): 53–90.

APPLES AND ORANGES? A CROSS-CULTURAL COMPARISON OF BURIAL DATA FROM NORTHEAST CHINA BY

GIDEON SHELACH (The Hebrew University)

Abstract The analysis of burial data from two or more archaeological cultures is a common strategy in cross-cultural studies. Because burial data can be quantified (e.g., by number of burial goods, labor investment in man-hours, etc.), they lend themselves to statistical analysis. We must keep in mind, however, that such remains are those of ritual activity, and the statistical analysis performed here must be based on the reasonable assumption that each society is marked by sets of sumptuary rules. In the absence of an explicit theoretical consideration of this issue, anyone who looks at burial data cross-culturally therefore runs the risk of comparing “apples and oranges.” This paper offers a suggestion as to how this problem can be overcome and how meaningful cross-cultural comparisons of burial data can be carried out. Using burial data from the Lower Xiajiadian (ca. 2100–1600 bc) and Upper Xiajiadian (ca. 1100–600 bc) Culture periods, I attempt to compare the level and type of social complexity attained by different Bronze Age societies in northeast China.

Benefits and drawbacks of cross-cultural comparisons Comparison is an integral part of every analysis of mortuary data. Although this paper specifically addresses the more complex problem of cross-cultural comparison, we must acknowledge the fact that the analysis of mortuary data from a single culture, or even a single cemetery, is fundamentally comparative. For example, in contrast to habitation sites, where archaeologists uncover evidence that is likely to have been gradually accumulated, the excavation of cemeteries brings to light groups of more or less independent cases (i.e., graves). Any analysis of such data sets, regardless of how advanced or what research questions it addresses, compares these autonomous cases with one another. Even © Brill, Leiden 2002

JEAA 3, 3–4

54

gideon shelach

though this alone, I think, merits an explicit discussion of comparative methods, the need for such a discussion becomes even more apparent when data are taken from more than a single society. The first and most fundamental question to be asked is whether cross-cultural comparison of mortuary data is possible at all. In some respects this question revives the 30 year-old debate pitting the positivistic approach of Binford and Saxe against the views of the so-called “post-processual” archaeologists (Saxe 1970; Binford 1971; Hodder 1982; Shanks and Tilley 1982). While I recognize, as does the Saxe-Binford tradition of mortuary practices, the benefits of cross-cultural comparison, I disagree with the latter’s hallmark position that the search for universal laws is the aim of archaeological research. Indeed, I agree with Hodder that “a change to a less complex or less differentiated burial rite does not necessarily entail a change to a less complex society” (1982: 199). I nevertheless share with Binford and Saxe, as well as with more recent “processual” research on mortuary behavior, the belief that death rituals cannot be completely divorced from social, political, or economic reality (cf. Brown 1995; McHugh 1999). Accepting Hodder’s insight that the relationship between mortuary ritual and social phenomenon is not a straightforward one implies that we need first to examine the mortuary practices within each individual society before we can attempt to compare these practices cross-culturally. In this paper I attempt to translate this general understanding into operative research methods. Because burial data can be quantified (e.g., by number of burial goods, labor investment in man-hours, and so forth), they lend themselves to statistical analysis, so much so that we sometimes overlook the obvious fact that what we are analyzing are the remains of a ritual activity that may in fact mask reality. Explicitly or implicitly, we too readily assume that these mortuary practices were governed by rules in which the ceremonial treatment of the dead reflects the precise economic, social, or political position a person held during his or her life. Furthermore, we also assume that each society operated according to a single set of sumptuary rules. Although this second assumption does not always hold, in general this is a reasonable assumption, as long as one does not under-estimate the degree to which such “rules” can be adapted to particular circumstances. Such an assumption obviously cannot be applied uncritically to crosscultural comparison. We cannot expect that cultures distant from one another in time and/or space necessarily shared the same sumptuary rules. Studies of both past and living societies confirm that in some societies, mortuary ritual emphasizes and even exaggerates the social, political, or economic status of the deceased, while in others—for

burial DATA FROM NORTHEAST CHINA

55

example modern Western societies—differences among members of the population are de-emphasized at the time of death (cf. Pearson 1982; Cannon 1989). The symbolic expression of various other attributes at death—even universal categories such as gender or age group—may likewise vary among different societies. In the absence of an explicit theoretical consideration of this problem, therefore, the cross-cultural analysis of burial data runs the risk of comparing apples and oranges. I do not wish, however, to discourage any attempt at crossing geographical and temporal boundaries. Such efforts remain essential because it is only through cross-cultural comparison that we can hope to gain a scientific understanding of the nature and range of sociopolitical systems and developmental trajectories. As recent research on state formation suggests, “Through comparison we see exemplified what is unique in a civilization, why what works in one state does not work in another, and what is the more general shape of a civilization” (Baines and Yoffee 1998: 259). What is true of states (see Chang 1989; Baines and Yoffee 1998; Feinman and Marcus 1998) is also true of less complex societies (see Drennan 1995): only through cross-cultural comparison can we hope to differentiate between unique local attributes and more universally shared processes. Cross-cultural comparison is also important for enabling us to escape biases inherent in local historical or intellectual traditions. In the study of any culture area, the comparison of local and non-local trajectories provides fresh insights into sociopolitical processes and points to new explanatory models. Mortuary practices during the Lower and Upper Xiajiadian Culture periods This paper attempts to develop a common-sense approach for comparing burial data from the Lower Xiajiadian (ca. 2100–1600 ) and the Upper Xiajiadian (ca. 1100–600 bc bc) Cultures, two archaeological cultures of northeast China. Even though they share a similar spatial distribution centered on southeastern Inner Mongolia (Neimenggu ) (Figure 1) as well as a number of cultural attributes, the two cultures exhibit noticeable differences in their material . This has prompted a spirited debate regarding their respective levels of social complexity and their social, political, and economic organization. Indeed, the two cultures are so different as to make it difficult to compare since we cannot assume that they were guided by similar social rules. On the other hand, as I have pointed out elsewhere (Shelach 1999), we must try to understand the contacts between these two cultures if we are to explain social, political, and economic developments in Bronze Age northeast China.

56

gideon shelach

Figure 1. A map of northeast China and the location of sites mentioned in the text: 1. Dadianzi; 2. Nanshan’gen; 3. Pingquan Dongnangou.

The two cultures are centered on the region of Chifeng in Inner Mongolia and share a similar distribution that extends through present-day eastern Inner Mongolia, western Liaoning , and northeastern Hebei . The Lower Xiajiadian is distributed south of the Xilamulun River and north of the Yanshan mountain range, between the Yiwulüshan mountains to the east and the Yixun River to the west. The Upper Xiajiadian culture area is somewhat larger, with sites found north of the Xilamulun River and as far south as the Beijing –Tianjin area (Zhu 1987; Guo 1995; Shelach 1999). During both periods, the dead were buried in cemeteries located outside the habitation sites. Typical for the Lower Xiajiadian are earthen pit-graves, most of which contain a single supine skeleton in the extended position. While some graves are over 2.2 m long, others measure less than 1.7 m. The main difference, however, is in their depth. The mean depth is 1.4 m, with some of the pits much shallower than this and others over 7 m deep. In many graves, niches for burial offerings were excavated into one or more of the pit walls. Remains of wooden

burial DATA FROM NORTHEAST CHINA

57

Figure 2. Grave 726 at the Lower Xiajiadian Culture cemetery of Dadianzi (after Zhongguo 1996: 48).

structures—the so-called “coffins”—have been found inside some of the graves (Figure 2) (Aohan 1976; Neimenggu 1984; Zhongguo 1996). Most Lower Xiajiadian graves contain the remains of sacrificial animals, commonly pigs or dogs. Typical grave goods include ceramic vessels, stone and bone tools, and seashells, as well as different kinds of

58

gideon shelach

Figure 3. Painted ceramic vessels from the Dadianzi cemetery (after Zhongguo 1975a: 101).

stone beads. Much less common are small metal objects, oracle bones, and lacquer artifacts (Neimenggu 1984; Guo 1995; Zhongguo 1996). The most common ceramic vessel types found in Lower Xiajiadian graves are li hollow-leg tripod vessels, guan jars, ding solid-leg tripod vessels, and hu urns (Zhongguo 1996: 67). It has been suggested that these various container types may have made up the ritual vessel sets of the Lower Xiajiadian period, an idea that will not be further explored here. The most elaborate ceramics found in these graves are the polychrome vessels, which were painted red, white, and vermilion after firing (Figure 3) (Liu and Xu 1989). In contrast to unpainted vessels, which are also known from Lower Xiajiadian domestic sites, painted vessels are found almost exclusively in burial contexts. Their ritual function is also suggested by the fact that the colors used to decorate them are unstable and would have been easily rubbed off when handled. Like the graves of the Lower Xiajiadian Culture, most Upper Xiajiadian burials are earthen pit-graves that contain a single skeleton in the extended, supine position. However, aside from these basic similarities, the mortuary traditions of the two cultures are quite distinct. In contrast to Lower Xiajiadian graves, Upper Xiajiadian tombs have clear above-surface markers such as piles of stones or earth; the pits are shallower and usually lined with stone slabs. The stones inside most of

burial DATA FROM NORTHEAST CHINA

59

Figure 4. Some typical Upper Xiajiadian Culture graves (after Hebei 1977: 52–3).

the graves are arranged to create a narrow burial chamber (or “coffin”). Wooden structures were also built inside some of the larger graves (Figure 4) (Liaoning 1973; Hebei 1977; Zhangjiakou 1987; Zhu 1987). Bronze artifacts dominate the assemblages of Upper Xiajiadian graves. The most common include tools—knives, axes, and chisels—along with weapons such as daggers, socketed axes, and arrowheads (Figure 5). Also very common are ornaments: buttons, hooks, flat animal-shaped disks, and belt buckles (Figure 6). In the analysis of Upper Xiajiadian burials that follows, I divide these bronzes into two groups: small artifacts (which include different types of ornaments and arrowheads) and

60

gideon shelach

Figure 5. Upper Xiajiadian Culture tools and weapons (after Zhangjiakou 1987: 46).

large artifacts (which include tools and weapons).1 Vessels are a third and much less commonly found type of bronze artifact (Zhu 1987; Zhai 1994; Xiang and Li 1995). Many of these bronze artifacts are decorated with animal motifs or are themselves fashioned in the shape of animals. Stylistically, they are very different from the bronzes pro1 As will be seen, this arbitrary division, which is based on trying to valuate the artifacts in terms of labor investment, raw materials used, expertise needed, and prestige, is tested to see if it produces meaningful correlations.

burial DATA FROM NORTHEAST CHINA

61

Figure 6. Upper Xiajiadian Culture bronze ornaments (after Xiang and Li 1995:18) .

duced around the same time in Chinese states located to the southwest of this region (Linduff 1997). Other grave offerings include sacrificed animals, small stone and bone artifacts, shells, and ceramic vessels. The burial sample as a reflection of the entire society One particular concern in any analysis of mortuary data, but especially in a cross-cultural comparison, is the issue of how representative of the entire social system the burial sample actually is. Misrepresentation is a common problem that itself results from the selectivity of both ancient funerary practices and modern research strategies. In areas of central China, for example, excavation of cemeteries from the Western Zhou (ca. 1050–771 bc) and Spring and Autumn (770–481 ) periods has focused almost exclusively on the highest ranks of the bc nobility (see Falkenhausen 1999: 475). Thus, while the under-representation of commoners’ graves (and all the more so of the graves of

62

gideon shelach

those strata of the population who were not buried in lineage cemeteries) should at the very least be acknowledged in any analysis of mortuary data from this period, special attention should be paid to any comparison with data sets that do include non-elite graves (see Falkenhausen, this issue). In general, we should be aware of sample biases and the limitations the data impose. With regard to burial data from northeast China, the relatively ample data available for the Lower Xiajiadian Culture contrasts with the much more spotty funerary record of the Upper Xiajiadian Culture. Funerary behavior during the Lower Xiajiadian period is well represented by the almost completely excavated cemetery of Dadianzi , located in Aohan Banner , Inner Mongolia (Figure 1). Data from the excavation of 716 well-preserved graves at Dadianzi suggest that the cemetery was used by all sectors of the community. 2 In contrast, the 81 Upper Xiajiadian graves that make up the sample analyzed here belong to no less than 15 different sites scattered over a large region. Although many more graves from this period have been excavated, the published reports provide sufficiently detailed information only for these 81 burials. As in many other parts of the world, these publications tend to focus on the richest graves, with burials containing few or no objects often ignored or mentioned only in passing.3 As a consequence, we can expect the Upper Xiajiadian grave sample to be biased towards larger and richer graves.4 Although, as pointed out above, such biases should not prevent us from attempting to analyze and compare the funerary data recovered from the two cultures, the differences between the two data sets should be acknowledged whenever the results of our interpretation may be affected. Direct comparisons and basic statistics In China, as in many other places, the comparison of mortuary customs typically relies on the measurement and recording of quantifiable features. For example, it is possible to state that on average the graves of 2 A total of 804 graves were excavated at this cemetery. My omission of badly preserved graves for which the information is not complete should not affect the validity of the sample. 3 See for example the report on 11 graves excavated at Pingquan Dongnangou in Hebei Province. Although the contents and measurements of the graves are not fully described, it is clear that only four of these graves contained grave goods (Hebei 1977). 4 We need to be aware of this bias, and it is discussed further below. On the other hand, the accidental nature of the discovery of most graves leads me to regard them as a relatively unbiased sample of graves belonging to the “archaeological culture,” although they were not associated with a single polity.

burial DATA FROM NORTHEAST CHINA

63

one mortuary tradition are larger and more elaborately furnished than those of another. While this type of analysis is not without its merits, I would argue that a much more productive research avenue is the systematic analysis of correlations and tendencies within each individual sample (e.g., the correlation between the age of the deceased and the richness of the grave), followed by the cross-cultural comparison of such tendencies, rather than the comparison of raw values themselves. Comparisons of mortuary traditions that rely on raw measurements and numbers are potentially misleading. Thus, when two traditions are governed by different sumptuary rules, differences in attributes such as grave size and number of grave goods do not necessarily reflect real socio-political differences. On the other hand, we cannot completely overlook the socio-political implications of certain straightforward observations as to grave elaboration, size, and wealth, since these do help reveal elite’s ability to command the labor force needed in the construction of their graves and in the production of their grave goods. Although such observations cannot by themselves prove that one society was more highly stratified than another, they can provide the starting point for a more advanced analysis. What we would usually like to compare is the amount of labor invested in the construction of the grave itself and in the production of the artifacts placed in it (Tainter 1978). This is not always easy to do. Environmental conditions, the level of technical expertise, and construction methods all help determine the amount of labor expended in the building of the graves. The same is true of the labor needed to produce different types of grave goods: for example, does the labor invested in the production of one small bronze tool equal that required to make one large ceramic vessel? While some of these concerns can be addressed through a better understanding of the production process (through experimentation, for example), one cannot entirely escape the uncertainty introduced by the many variables involved. Because of the similar geographical distribution of the Upper and Lower Xiajiadian Cultures, and because a number of behavioral and technological attributes were shared by both cultures, I believe that it is permissible to assume that a similar amount of effort would have been required to excavate and prepare graves in each culture. Tables 1 and 2 suggest that in comparison to Upper Xiajiadian graves, Lower Xiajiadian graves were on average larger and therefore more labor intensive. On the other hand, the extensive use of stone slabs in graves of the Upper Xiajiadian period must have increased the cost of constructing these graves. For neither period is the total number of grave goods a good indication of labor investment since most objects are beads and other small artifacts that are relatively cheap to produce. Moreover, because beads

0.010 43.600 2.250 2.93 1.3

total no. of artifacts 0.000 863.000 7.465 45.935 6.14

no. of stone artifacts 0.000 856.000 4.669 44.780 9.57

no. of ceramic vessels 0.000 12.000 2.050 1.562 0.76

no. of painted ceramic vessels 0.000 10.000 0.598 1.173 1.96 0.000 15.000 0.559 1.095 1.96

no. of ‡ animals

no. of metal objects 0.000 11.000 0.070 0.535 7.64

* Tables 1, 3, 5, 7, 8, 9 and Figures 7, 8, 11 are based on data collected from Zhongguo 1996. † Here and in other tables, the size of Lower Xiajiadian Culture graves is calculated only for those 713 graves for which this data is available. ‡ Here and in other places where animals are discussed, the minimum number of animals in each burial is provided. Although this analysis does not distinguish between different animal species, it is worth remembering that the domesticated pig and dog were the most common sacrificial animals during the Lower Xiajiadian period. A more detailed analysis of animal bones recovered from the Dadianzi cemetery has been carried out by Rowan Flad (1998 and paper in this volume).

Minimum Maximum Mean Standard-Deviation Coefficient of Variability (CV)

grave †size ( m3 )

Table 1. General statistics for Lower Xiajiadian Culture graves (sample includes all 716 well-preserved graves from the site of Dadianzi).*

64 gideon shelach

0.014 20.064 1.765 3.132 1.775

0.000 1272.000 99.852 194.409 1.95

total no. of artifacts

no. of ceramic vessels 0.000 2.000 0.652 0.564 0.98

no. of bronze artifacts 0.000 513.000 47.901 91.479 1.91

no. of large bronze artifacts 0.000 91.000 3.630 13.100 3.4

no. of bronze vessels 0.000 28.000 0.580 3.677 6.43

no. of ornaments and small objects 0.000 1268.000 91.704 188.678 2.06

* Tables 2, 4, 6, 10 and Figures 9, 10, 12 are based on data collected from the following: Liaoning 1973; Zhongguo 1974; Zhongguo 1975b; Hebei 1977; Zhongguo 1981; Jianping 1983; Zhongguo 1984; Ningcheng 1985; Zhangjiakou 1987; Xiang and Li 1995. † Here and in other tables, the size of Upper Xiajiadian Culture graves is calculated only for those 69 graves for which this data is available.

Minimum Maximum Mean Standard-Deviation Coefficient of Variability (CV)

grave †size ( m 3)

Table 2. General statistics for Upper Xiajiadian Culture graves (81 graves from various sites).*

burial DATA FROM NORTHEAST CHINA 65

66

gideon shelach

did not usually function as single artifacts but rather as parts of larger objects (such as necklaces), the total numbers of beads is not a very meaningful criterion. More useful to the analysis are the larger artifacts, especially the ceramic vessels of the Lower Xiajiadian period and the large bronzes of the Upper Xiajiadian, for which the respective mean numbers per grave are similar.5 Likewise, the average number per grave of painted ceramic vessels during the Lower Xiajiadian is similar to the average number of bronze vessels in Upper Xiajiadian graves. Of course, the production of Upper Xiajiadian bronzes was far more labor consuming than that of Lower Xiajiadian ceramics. Labor investment comparisons With the above assumptions, I estimate that the mortuary practices of the Lower and Upper Xiajiadian periods on average consumed more or less the same amount of labor. However, in the study of social and economic organization, averages interest us less than the way values are distributed among the different social strata. The histograms and box-and-dot plots shown in Figures 7 and 8 present these patterns visually. It can be seen that while the plots for most variables tend to display normal distributions, a few of the cases have elevated values well to the right of the curves, a tendency that is particularly apparent in the Upper Xiajiadian. The same phenomenon can be numerically addressed through a comparison between the highest and lowest ends of the social or economic spectrum. A customary way of dealing with this issue is to present examples of graves attributed to the upper stratum of society. Two such examples are grave no. 726 at the Lower Xiajiadian cemetery of Dadianzi and grave no. 101 at the Upper Xiajiadian site of Nanshan’gen . Grave no. 726 is 4 m long, 1.4 m wide, and 7.8 m deep. The remains of a wooden coffin were found in the pit, which also had niches dug into three of its walls (Figure 2). The niches contained all eleven of the grave’s ceramic vessels, nine of which were painted. In addition, the grave contained the bones of 3 sacrificial animals (2 pigs and 1 dog), a stone axe, 206 small stone artifacts, 255 shells, and the remains of 3 lacquered vessels (Zhongguo 1996). Grave no. 101 at Nanshan’gen is 3.8 m long, 2.3–1.8 m wide and 2.4 m deep. The pit, which contained stone walls, was marked by a pile of stones over it. Also found were the remains of a wooden coffin that contained 508 bronze artifacts, 64 of these large (including 18 vessels and 20 5 The significance of these types of artifacts, rather than the total number of all grave goods, is further demonstrated later on by the results of the linear regression analysis and the factor analysis.

burial DATA FROM NORTHEAST CHINA

67

Figure 7. Histograms and box-and-dot plots of Lower Xiajiadian Culture burial variables.

weapons). Grave no. 101 also yielded 7 stone artifacts and 3 gold ornaments (Liaoning 1973). This comparison suggests that although in both cultures much work was invested in the construction and furnishing of the richest graves, the investment would have been greater in the Upper Xiajiadian Culture. A more systematic way of addressing this issue is to compare, for each culture, the amount of labor invested in graves of the upper and lower strata of society. Tables 3 and 4 compare the largest and richest

68

gideon shelach

Figure 8. Histograms and box-and-dot plots of Upper Xiajiadian Culture burial variables.

10% of graves to the smallest and poorest 20% within each culture. Ignoring the total numbers of artifacts (which as noted earlier are not a good indication of the amount of labor invested in the graves), a comparison of the figures presented in Tables 3 and 4 suggests that the gap in the amount of labor invested in the graves of the upper and lower strata of society was larger during the Upper Xiajiadian period

69

burial DATA FROM NORTHEAST CHINA

Table 3. Comparison between large/rich and small/poor graves during the Lower Xiajiadian Culture period (average values are shown). grave size total no. of ( m 3) artifacts 10% Largest/Richest 6.760 Graves 20% Smallest/Poorest 0.665 Graves Ratio of Largest/Richest 10.165 to Smallest/Poorest

no. of ceramic vessels

no. of no. of painted animals ceramic vessels 2.160 1.760

30.600

5.020

0.239

0.014

0.007

0.049

128.033

358.571

308.571

35.918

Table 4. Comparison between large/rich and small/poor graves during the Upper Xiajiadian Culture period (average values are shown). grave size total no. of total no. no. of no. of ( m 3) artifacts of bronzes large bronze bronze vessels artifacts 8.551 187.250 168.625 24.250 5.875

10% Largest/Richest Graves 20% Smallest/Poorest 0.307 Graves Ratio of Largest/Richest 27.853 to Smallest/Poorest

16.571

0.429

11.230

393.065

0.000

0.000

Infinitely Infinitely high high

than it was during the Lower Xiajiadian period.6 This is especially clear in the case of the large bronze artifacts and bronze vessels, with no specimens of either of these artifact categories recovered from the poorest Upper Xiajiadian graves. During the Lower Xiajiadian period, in contrast, no artifact type appears to have been exclusively possessed by the elite. Another way of addressing the same issue is to consider the variability of artifact classes seen in the graves of each culture. Here we take a step away from the actual numbers in favor of looking at their distribution patterns. My assumption is that a greater degree of variability may reflect more pronounced social, political, or economic stratification. In 6 Since, as mentioned earlier, we are probably missing the poorest graves of the Upper Xiajiadian period, actual differences between rich and poor graves during this period may have been even greater.

70

gideon shelach

a highly variable sample, many cases differ significantly from the average, the standard deviation is larger, and the distribution curve is flatter. The Coefficient of Variability (CV), which equals the ratio of the standard deviation to the mean, is a common way of expressing this idea. As illustrated in Tables 1 and 2, the CV values of mortuary variables for both periods vary significantly. However, a comparison of those variables we identify as more closely linked to social structure—grave size for both periods, the number of ceramic vessels and painted ceramic vessels for the Lower Xiajiadian, and the number of large bronzes and bronze vessels for the Upper Xiajiadian period—suggests a greater degree of variability during the Upper Xiajiadian period. The fact that some of the CV values for the Lower Xiajiadian period are high may be the result of the structure of the data (e.g., it may be due to the number of stone artifacts and the total number of artifacts) or of the rarity of certain artifact categories (e.g., metal objects), where a single case can distort the results. A common sense evaluation of the CV results therefore appears to lend support to the conclusions drawn from Tables 3 and 4. Although knowledge of the degree of variability among graves for single variables is certainly helpful, we are also interested in determining the extent to which different variables co-vary. In a highly stratified society, for example, we would expect to detect a strong correlation linking different dimensions of labor investment, such as that associated with the construction of the grave and the production of grave goods. A strong correlation among such variables is a good indication of the presence of sumptuary rules that express the social prestige of the deceased in a systematic way.7 A careful consideration of the strength of correlations among variables can also provide important insights into the nature of social stratification. The Pearson correlation matrix is a useful way of detecting and interpreting the strength of correlations among all variables in a data set. One general observation regarding Table 5 is that for the Lower Xiajiadian period most variables of mortuary activity are positively correlated with one another. In fact, only the spindle whorl category is negatively correlated with some of the other variables. A strong correlation exists among grave size, the number of ceramics, and the number of painted ceramics (values of 0.7, 0.722 and 0.634). As pointed out 7 I make this claim based on an assumption that the existence of clear sumptuary rules would dictate how much energy to invest in a grave and what types of objects to place in it, and this would result in correlations. However, since sumptuary rules could reflect different aspects of the deceased’s social, economic, and political position, more than one correlation cluster could exist, and we should try to observe and explain all of these.

1.000 0.312 0.176 0.223

0.096 0.327

0.047 -0.043

1.000 0.187 0.294 0.700 0.722

0.570 0.326

0.335 0.099

Metal Artifacts

0.004 -0.029

0.130 0.998

1.000 0.190 0.231

Stone Artifacts

0.393 0.157

0.524 0.231

1.000 0.634

C eramic Vessels

0.273 0.074

0.545 0.263

1.000

Painted C eramic

0.272 0.030

1.000 0.156 0.020 -0.019

1.000 1.000 0.176

Sacrificial All Artifacts Age at Animals (total no.) Death

1.000

Spindle Whorls

* 1 and –1 represent respectively the largest possible positive and negative correlation values. A value of 0 indicates the absence of any correlation between two variables.

Size Metal Artifacts Stone Artifacts Ceramic Vessels Painted Ceramic Vessels Sacrificial Animals All Artifacts (total no.) Age at Death Spindle Whorls

Size

Table 5. Pearson correlation matrix for Lower Xiajiadian Culture graves.*

burial DATA FROM NORTHEAST CHINA 71

72

gideon shelach

above, these variables seem to encompass most of the labor investment in Lower Xiajiadian graves. A strong correlation among these variables indicates that large graves also tend to have more labor-intensive grave goods in them, suggesting a sumptuary system with a clear grading of labor investment. The fact that the total number of artifacts found in each grave is not well correlated with grave size or the number of ceramic vessels illustrates a point made earlier, namely that the total number of grave goods is in many cases inflated by the large number of stone beads or other small stone artifacts, all objects that were probably very cheap to produce and therefore not very meaningful in either an economic or a social sense. This claim is supported by the very strong correlation that exists between the total number of artifacts and the number of stone artifacts (value of 0.998). It is interesting to note in Table 5 the relative weakness of the correlation between age at death and the major labor investment variables— grave size, number of ceramic vessels, and number of painted ceramic vessels (values of 0.335, 0.393 and 0.273 respectively). In societies where status reflects a person’s life-time achievements, we would expect to find a strong correlation between age and status—the longer a person lives, the longer he or she has to achieve high status and to accumulate the attendant benefits and material. The weak correlation during the Lower Xiajiadian between age at death and the major labor investment variables suggests a society in which status may have been ascribed at birth and was less dependent on personal achievements and age. If status were mainly ascribed, this would not mean that achievement played no role at all in its determination, since there exists some degree of correlation between age at death and labor investment (see the above values). Further support of this can be seen in the fact that among the largest and richest 10% of graves, the average age at death is 37.8 and the age at death of 17 out of 71 skeletons found in these graves is 50 or older. In contrast, the average age at death of skeletons in the smallest and poorest 20% graves is 18.5, with only 9 out of 142 of the tomb occupants dying at the age of 50 or more. This suggests that achieved status was not totally ignored but that, at least with regard to burial treatment, it was much less important than status ascribed to the person through birthrights (for a different analysis, see Flad, this volume). Patterns observed in the Upper Xiajiadian correlation matrix are in many respects similar to those of the Lower Xiajiadian matrix (Table 6). Most of the mortuary variables for this later period are positively correlated with one another, the one exception being the negative correlation between the number of ceramic vessels and most other variables. A strong correlation—stronger even than that for the Lower Xiajiadian period—exists among three categories associated with high labor investment, namely grave size, the number of large bronze artifacts,

1.000

0.989 0.971 0.845 0.333 0.420

0.381 -0.199

0.794 0.747 0.747 0.608 0.667

0.257 -0.055

Large Bronze Artifacts

1.000 0.802

Bronze Artifacts

0.375 -0.196

1.000 0.968 0.820 0.322 0.406

Bronze Vessels

0.405 -0.190

1.000 0.721 0.310 0.393

B ronze Tools

0.277 -0.202

1.000 0.284 0.368

Bronze Weapons

0.128 0.044

1.000 0.995

Small Artifacts

0.174 0.020

1.000

All Artifacts (total no.)

1.000 0.086

Stone Artifacts

1.000

C eramic Vessels

* 1 and -1 represent respectively the largest possible positive and negative correlation values. A value of 0 indicates the absence of any correlation between two variables.

Size 1.000 Bronze Artifacts 0.821 Large Bronze 0.842 Artifacts Bronze Vessels 0.825 Bronze Tools 0.761 Bronze Weapons 0.855 Small Artifacts 0.608 All Artifacts 0.676 (total no.) Stone Artifacts 0.261 Ceramic Vessels -0.170

Size

Table 6. Pearson correlation matrix for Upper Xiajiadian Culture graves.*

burial DATA FROM NORTHEAST CHINA 73

74

gideon shelach

and the total number of bronze artifacts placed in each grave (values of 0.842, 0.821 and 0.802). Although the total number of artifacts is strongly correlated with grave size (value of 0.676), the total number of artifacts is more strongly correlated with the number of small artifacts such as beads and other ornaments (value of 0.995). Such small artifacts inflate the total number of grave goods in much the same way as the small stone artifacts of the Lower Xiajiadian period. Stone artifacts and ceramics are not well correlated with other variables, and their distribution among graves may better reflect other social categories such as gender, professional association, etc. Unfortunately, because most publications on the Upper Xiajiadian period lack information on age at death, it is not possible to address the issue of whether status was achieved or ascribed. The significance and strength of the correlations brought to light by the two Pearson correlation matrices can be further explored using linear regression analysis. A plot of the size of Lower Xiajiadian graves against the number of painted pottery vessels found in them reveals a clear linear correlation (Figure 9). The value of R2 (or the proportion of the total variability explained by the regression) is 0.514.8 This result is highly significant; the t-test value for the significance of the slope is 27.438 (P