UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A review of Lungshanoid sites using cluster analysis and multidimensional scaling 1977

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


UBC_1977_A8 L06.pdf [ 7.96MB ]
JSON: 1.0058401.json
JSON-LD: 1.0058401+ld.json
RDF/XML (Pretty): 1.0058401.xml
RDF/JSON: 1.0058401+rdf.json
Turtle: 1.0058401+rdf-turtle.txt
N-Triples: 1.0058401+rdf-ntriples.txt

Full Text

A REVIEW OF LUNGSHANOID SITES USING CLUSTER ANALYSIS AND MULTIDIMENSIONAL SCALING SHYH-CHARNG LO B.A., National Taiwan University, 1 9 7 0 A Thesis Submitted i n Partial Fulfillment of The Requirements for the Degree of Master of Arts i n The Dept. of Anthropology and Sociology We accept this thesis as conforming t o t h e required standard THE UNIVERSITY OF BRITISH COLUMBIA February, 1 9 7 7 Shyh-charng Lo, 1 9 7 7 In presenting this thesis i n partial fulfilment of the requirements for an advanced degree at the University of Br i t i s h Columbia, I agree that the Library shall make i t freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Depart- ment or by his representatives. I t i s understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of Br i t i s h Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1 W 5 Abstract Lungshanoid cultures are distributed i n approximately the same area as the Yangshao Culture and the Classic Lungshan Culture i n the north as well as along the south- east coast of China including Central and Southwest Taiwan. A l l these cultures represent a mixed or transitional culture between Yangshao and Lungshan. In the past decades, a great number of these sites have been found, excavated and classified into several cultures with different local names, such as the Ch'u-chia-ling culture, the Liang-chu culture, the T'an-shih-shan culture, the Ta-wen-k'ou culture and the Ch*ing-lien-kang culture. These sites date from the 2nd to 4th millennia B.C.. Cluster analysis i s employed to review the present classification of these Lungshanoid cultures. Nineteen sites scattered throughout Southeast China are chosen as the OTU or data units and 80 characters are isolated as the variables i n this Q mode test. Seven cluster emerged as a result of this s t a t i s t i c a l analysis. Cluster I, II, III and IV f i t into the traditional Ch*u-chia-ling culture, Liang-chu culture, T^an-shih-shan culture and Ta-wen-k^ou • • • 1 1 1 culture. However, the previous Chiang-pei type of the Ch*ing-lien-kang culture i s shared by cluster V and VIIi the Feng-pi-t*ou site and traditional Chiang-nan type of the Ch'ing-lien-kang culture are grouped into cluster VI. The configuration from the multidimensional scaling on the f i r s t and second vectors seems that the patterning of the sites on these two vectors agree with the clusters represented by the dendrogram. Especially, the vertical dimension can be seen as a shift i n pottery character and variation of implements. The probable meaning of these clusters, such as different time periods, different people, different langu- ages, or even different technologies, has also been briefly discussed. This study presents the f i r s t attempt at the application of clustering and scaling techniques to Chinese archaeological data. More detailed study of these site reports are necessary. i v TABLE OF CONTENTS Abstract i i Table of contents i v Acknowledgement ' v Chapter 1. Introduction 1 2. The Lungshanoid Cultures 6 3. The Di f f i c u l t y of Classification 16 , 4. General Idea of Cluster Analysis and Multidimensional Scaling 35 5. The Use of Cluster Analysis and Multidimensional Scaling 40 6. Discussion 51 7. Conclusion 6l Bibliography 66 Appendix A. Original Reports of 19 Sites 72 B. Attribute L i s t 73 C. Chinese Character for Proper Names and Technical Terms 77 V Acknowledgement My thanks are due to Professors Richard Pearson and Patricia Hitchins for their continuous encouragement and advice through the last two years, their many sugges- tions on the early drafts, and their kind and patient help in rendering the manuscript into readable English, and to Professor E. G. Pulleyblank for his suggestions and comments. I also wish to thank Professor Richard Matson, from whom I learned cluster analysis and multidimensional scaling. Without this knowledge the present study could not have been undertaken. S. C. Lo 1 Chapter 1 Introduction In the past decade, the importance of Southeast China in Par Eastern archaeology has increased markedly. One of the reasons for this interest is that more and more archaeological evidence has been discovered thereby making available more material with which to compare, analyze, and study the prehistory of the area. After the pollen analysis at Jih-yueh-Van was done by Tsukada Matsuo in 1964-65 (Tsukada 1 9 6 6 ) , and the excavation at Spirit Cave in Thailand by Chester F. Gorman in 1967-68 (Solheim 1969)» Carl 0. Sauer^s (1952) theory on a Southeast Asian Agricultural heartland began to attract renewed attention. For an illustration of this development we can refer to Chang's work ( 1 9 7 0 i 175-185)» in which he sets down the archaeological and botanical evidence for the beginnings of agriculture in East and Southeast Asia and, on this basis, argues for agricultural origins there. Thus, the increased importance i n i t i a l l y attributed to the whole of 2 Southeast Asia by Carl 0. Sauer has stimulated further archaeological research in the southeastern regions of China. In addition, a number of CI 4 dates for this area have been published during the last four years (LIA 1972a, 1972b, 197^)• Some of these are so early that archaeo- logists have had to shift their original theories and make adjustments for the new data. K.C. Chang, for instance, insisted several years ago that the neolithic culture in South and Southeast China was the result of the rapid extension of "Chung-yuan culture" of the North China Nuclear Area (Chang 1969). But now, he seems to have dramatically modified his view on this matter. For ins- tance, in a recent article (197̂ » 3^-38)t he proposes a cultural relationship among North China, South China and Southeast Asia along the following linesi (1) Various cultures developed in different parts of the Far East during the neolithic age. All of them developed their own culture to adapt to their different environments. Owing to contact, they were similar to some degree (p. 36), (2) Since a continued cultural development — from the Yangshao culture to the Lungshan culture and to the Shang 3 civilization — only occured in North China, the Yellow River Valley was the earliest nuclear area in the Far East (pp. 3 6 - 3 7 ) . ( 3 ) However, the Lungshanoid cultures in South China probably developed from Cord-marked pottery as another cultural category. Some aspects of this culture were derived from the Yangshao culture, while the Lungshanoid cultures made some cultural contributions to North China. The peoples of the Lungshanoid cultures seemed to be among the ancestors of the Malayopolynesians (p. 3 7 ) . (4) It is obvious that the new evidence in Thailand is extremely important in the prehistory of Indochina but "they don't have to be made to bear — ineffectively — upon the origin of distant China" (p. 3 7 ) • In his article N in Current Anthropology (1973)» Chang also clearly points out thati "the Pan-p'o dates of North China (ZK-38, ZK-121, ZK- 1 2 7 ) and Tainan date of Taiwan (SI-1229) show that by at least 4000 B.C. two ceramic cultures existed side by side in North and Southeast China. Neither can be said to represent the i n i t i a l form of i t s ceramic tradition, and neither can at this time be seen as derivative of the other" ( 1 9 7 3 * 5 5 ) . 4 He also notes in the same article "the possibility is certainly strengthened that the Lungshanoid horizon f i r s t "began to form in the Lower Yangtze Valley and its adjacent coastal areasj in any event, the Miao-ti-kou II Culture has now further diminished in stature as the ancestral culture of al l other Lungshanoid cultures. ... As a horizon, the Lungshanoid could s t i l l have come about, on the founda- tion of prior culture in the Lower Yangtze Valley (possibly a southern cord-marked pottery culture not dissimilar to the one represented by the Hsien-jen Cave remains in Wan-nien, Kiangsi), as the result of a strong and stimulating cultural impact from the Yangshao Culture" (1973« 5 2 7 ) . In other words, i t is quite obvious that an another "Nuclear Area" seems to have emerged in Southeast China. Mainland Chinese archaeologists, in spite of their own fixed ideology have stated that i t is "notable" that the radiocarbon dates of the Neolithic Culture in Southeast China are not later than those fdr the Chung-yuan (An 1 9 7 2 1 40, Wu 1 9 7 3 * 5 6 - 7 ) . This paper has been written primarily as a pilot study to test the traditional classification of Lungshanoid sites in this area by using cluster analysis and multidimensional scaling. TABLE 1 Some Important Radiocarbon Dates from China 5 Sample Archaeological Site Associated B.C. (573014-0 Culture half-life) North China Zk-38 Pan-p'o (Shensi) Yang-shao 4115*110 Zk-121 - - 3955*105 ZK-122 - - 3890+105 ZK-127 - - 3635*105 ZK-134 Hou-kang (Honan) - 3730*105 ZK-76 - - 3535*105 ZK-110 Miao-ti-kou (Honan) - 3 2 8 0 * 1 0 0 ZK-185 Ta-ho-ts'un (Honan) - 3 0 7 5 * 1 0 0 ZK-111 Miao-ti-kou (Honan) Miao-ti-kou II 2310195 ZK-126 Wang-wan (Honan) Lung-shan 2 0 0 0 * 9 5 ZK-133 Hou-kang (Honan) - I 9 6 0 * 9 0 Southeast China ZK-39 Hsien-jen Tung (Kiangsi) Lungshanoid(?) 8 9 2 0*240 ZK-90 Ta-tun-tzu (Kiangsu) Ch ,ing-lien-kang3835*105 SI-1229 Kuei-jen (Taiwan) Cord-marked 3 6 3 9 * 6 0 ZK-55 Sung-tse (Kiangsu) Ch ,ing-lien-kang3395±105 ZK-49 Chtien-shan-yang (Chekiang) Liang-chu 2750+100 ZK-51 P'ao-ma-ling (Kiangsi) Lungshanoid 2 3 3 5*95 ZK-91 Huang-chien-shu (Honan) Ch'u-chia-ling 2270+95 ZK-124 Ch'u-chia-ling (Hupei) - 2195 * 1 0 0 ZK-125 - - 2 2 4 5 4 1 6 0 ZK-242 Chueh-mu-chiao (Chekiang) Liang-chu 1990+95 ZK-98 Tan-shih-shan (Fukien) Lungshanoid 1 1 4 0 * 9 0 Sample SI-1229 from Chang ( 1 9 7 3 p . 5 2 5 ) . and the rest of the samples selected from LIA (Laboratory of the Institute of Archaeology) 1 9 7 2 (p . 5 6-58), 1 9 7 4 (p.333-338). 6 Chapter 2 The Lungshanoid Cultures The great number of Neolithic sites i n China which have been discovered i n the last one or two decades have been classified into several different local cultures, such as the Ta-wen-k»ou culture, the Ch'u-chia-ling culture, and the Ch'ing-lien-kang culture (KKYCS 1 9 6 1 ) . A l l these cultures somehow share a similarity to Lungshan- l i k e sites i n the north; that i s , they represent a mixed or transitional culture between Yangshao and Lungshan. K.C. Chang (1959) terms the culture at these sites Lungshanoid. According to Chang (19681 144), these Lung- shanoid cultures are distributed i n approximately the same area as the Yangshao Culture and the Classic Lungshan Culture i n the north as well as along the southeast coast of China including Centralsand Southwest Taiwan. As a matter of fact, the name "Lungshanoid" i s the by-product of his earlier hypothesis of the rapid expansion of advanced village farmers from the North China Nuclear Area south- 7 eastward to new frontiers (19681 1 3 0 ) . Despite some revision of this hypothesis, the term is s t i l l quite meaningful today. In order to further understand the term Lungshanoid, Yangshao and Lungshan culture must be discussed f i r s t . Yangshao, the earliest well-established cultural stage of Northern China, is named after the site of Yang-shao- ts»un, in Mien-chih Hsien, Western Honan which, in 1921, was excavated by the Swedish geologist J.G. Andersson. The associated painted pottery has long been regarded as one of the very important diagnostic features of Yangshao culture. The distribution of this culture extends from Southern Shansi, Western Honan, and Central-Eastern Shensi to Eastern Kansu, Central Shansi, and Northern Honan (Chang 19681 8 8 - 8 9 ) . In 1928, seven years after the ex- cavation of Yang-shao-ts*un, the Lungshan culture was dis- covered by Wu Chin-ting at Ch*eng-tzu-yai, near the town of Lung-shan in the heart of Shantung province. Its thin, hard, lustrous black pottery was dramatically different from the painted red sherds at Yang-shao-ts^un. Only seve- ral years after the excavation at Ch^eng-tzu-yai in 1931» the Black Pottery culture (Lungshan) was found to have a 8 distribution not only in Eastern Honan and Shantung, but also in the area along the Pacific coast from Pohai Bay to Hang-chow Bay in Northern Chekiang (Chang 19681 122-124). Since the i n i t i a l discovery of these two different cultures, the interrelationship between the two of them has been energetically discussed. General speaking, Chinese archaeologists during the 1 9 3 0 t s regarded the Painted Pottery culture (Yangshao) of western Honan, Shansi, Shensi, and Kansu, and the Black Pottery culture (Lungshan) of Eastern Honan and Shantung, as a pair of opposing but parallel cultures of the late Neolithic period immediately preceding the rise of Shang civilization (Chang 19681 124, 1969* 3 - 4 ) . However, this -Two Culture" framework faced a dilemma when Lungshan-like gray pottery was discovered in the Weishui Valley of Shensi in 1943» In attempting to resolve this problem, the discoverer of Lungshan-like cultures, Prof. Shih Chang-ju, added a third culture — Gray Pottery Culture — to the Chinese Neolithic Cultures in his subsequent publication (Shih 1952: 65-75), On the other hand, two Japanese scholars, Mizuno Seiichi (1956) and Sekino Takashi ( 1 9 5 6 ) , were the fi r s t to suggest that i t was quite possible for the black pottery of the Lungshan culture to have developed from 9 the painted pottery of the Yangshao cultures. In other words, they would rather believe that these two cultures belonged to one single culture of two different time periods than contemporary cultures in two different geo- graphical areas. Nevertheless, both of them s t i l l accepted Prof. Shih's Gray Pottery Culture as a separate entity. Finally, in 1961 the authors of Hsin Chung-kuo t i K'ao Ku Shou Huo (The Archaeology of New China) stated that new discoveries in the last decade had forced a change in viewpoint* the hypothesis that the Lungshan cultures of the Chung-yuan were developed out of the Yang-shao culture had been generally accepted by archaeologists? no longer were they regarded as cultures of different origins (pp. 2 0 - 2 1 ) . K.C. Chang expanded on this hypothesis. Having carefully studied many aspects of Yang-shao and Lungshan culture such as settlement patterns, cultivation patterns, principal domestic animals, pottery techniques, burials, community patterns, art, cult and ceremonial practices, Chang formulated his "Lungshanoid" hypothesis which could contend with the problem of mixed cultures (or Lungshan- like cultures). He viewed Lungshanoid cultures as the results of expansion of advanced village farmers from the North China Nuclear Area to new frontiers. To him, 10 Yangshao cultures represented the establishment stage of village farming and the Lungshan cultures represented the formation stage of local cultures (1968* 128-131, 1969* 9 - 1 1 ) . A l l these Lungshanoid cultures have polished stone implements that include rectangular adzes, perforated knives, and sickles. The pottery at a l l these sites i s a mixture of red, gray and black wares, and i n decoration there i s a mixture of impressed, incised, and painted patterns. The pottery shapes, although different i n detail, share several basic forms 1 ting tripods with solid legs; tou with cut-out ring feet; kui-type jars, and the wide occurrence of l i d s (Fig. 1) (Chang 1968t 145-146). Also according to Chang (19681 1 2 8 - 1 2 9 ) * there were 15 items i n Lungshanoid cultures very different from typical Yang-shao cultural elements* (1 ) permanent settlement; relatively permanent occupation; (2) probable irri g a t i o n , use of f e r t i l i z e r and a fallow f i e l d system; ( 3 ) cattle and sheep i n sharply increased number i n addition to pigs and dogs; (4) far-reaching expansions into the eastern plains, Manchuria, Central and South China, indicating population pressure from 11 permanent settlement and greater productivity (?) (5 ) emergence of many regional styles; ( 6 ) more asymme- t r i c a l edges than symmetrical; more rectangular cross sections indicating extensive use of carpenters* tools (adzes, chisels, antler wedges); (7) semilunar and double-holed, or sickle-shaped stone knives and shell sickles, a find- ing which indicates more extensive use of havesting tools; (8 ) beginning of wheel-made pottery indicative of intensi- fied craft specialization; (9) scapulimancy which indicates intensified occupational specialization; ( 1 0 ) appearance of "hang-^u" village walls and weapons indicating^ a nece- ssity for f o r t i f i c a t i o n and means for offensive action; (ll)growing number of otherwise differentiated burials,which possibly indicates more r i g i d l y constituted classes, ( 1 2 ) concentration of jade artifacts at isolated spots i n one site indicating more intensive statusJ differentiation, (13) art not conspicuously associated with domestic crafts; possible association with theocratic crafts(?) (14) ceremo- nia l wares (eggshell forms and fine, well-made cups, f r u i t stands and shallow dishes), (15) evidence of an i n s t i - tutionalized ancestor cult; ceremonials far beyond the merely agricultural type, possibly associated with special- ized groups of people. 12 A great number of Lungshanoid sites have been found and excavated in the last two decades and also generally been classified into several cultures with local names, such as the Ch*u-chia-ling culture, the Liang-chu culture, the T'an-shih-shan culture, the Ta-wen-k'ou culture and the Ch,ing-lien-kang culture. Since the number of sites classified into the Ch'ing-lien^kang culture is so great — in Kiangsu province alone there are more than 65 sites — and as the study has been so intensive, Wu Shan-ching, a Chinese scholar in mainland, recently has suggested that the Ch'ing-lien-kang culture should be subdivided into a Chiang-pei (northern bank of Yang-tzu River) type and a Chiang-nan (southern bank of Yang-tzu River) type (Fig.2,3). He also suggested that the Chiang-pei type can be classified into four stages, the earliest being Ch'ing-lien-kang, followed by Liu-lin, Hua-t*ing, and Upper Ta-wen-k»ou. The Chiang-nan type can be divided into three stages, namely, Ma-chia-pin, Pei-yin-yan-ying and Sung-tse (Wu 1973* 4 5 - 5 5 ) . 13 Liang-chu Culture Pig. 1 Artifacts of the Ch*u-chia-ling Culture, the Ch'ing-lien-kang Culture and the Liang-chu Culture, (all from Hsin Chung- kuo t i k»ao-ku Shou-huo, 1 9 6 l , p. 2 9 ) . i Fig. 2 Stages of Artifacts (Chiang-nan type), (From Wu 1973* 5 0 ) . F i g . 3 Stages of A r t i f a c t s (Chiang-pei type). (From Wu 1973* 50). M 16 Chapter 3 The Difficulty of Classification The position of classification in the study of archaeology has its undeniable importance, since archaeologists cannot avoid the step of classification whenever they deal with archaeological studies at any level. As a matter of fact, archaeology has been treated more as a discipline of science during the past decade. No archaeologist is pleased to be described as an "old archaeologists" who treats archaeology as art. In addition, R.C. Dunnell argues that "Classification is the systematic foundation of science" (1971» 8 7 ) . Thus, the importance of classification in archaeology is very obvious. However, no classificatory system is without some inherent shortcomings. First of a l l , the meaning of the term "classification" has not always been clearly defined when used by archaeologists. According to R.C. Dunnell ( 1 9 7 1 : 4 4 ) , classification is only a kind of arrangement which archaeologists usually treat as "classification", 17 and the other kind of arrangement is grouping (Fig. 4). ARRANGEMENT descriptive GROUPING GROUPS definitive IDENTIFICATION GROUPS MATCHED WITH CLASSES CLASSIFICATION CLASSES historical I 1 STATISTICAL NUMERICAL CLUSTERS TAXONOMY internal external ahistorical KEYS I TAXONOMY 1 PARADIGMS non-dimensional dimensional Figure 4. Kinds of arrangement. (From Dunnell 1971« 4 4 ) . Dunnell*s definitions for the&e&tertaslis/i;- "Classification will "be restricted to arrangement in the ideational realm and defined as the creation of units of meaning by stipulating redundancies (classes). Grouping will be used to denote arrange- ment in the phenomenological realm and defined as the creation of units of things (groups). Grouping and classification are articulated with one another 1 8 by means of identification, the process of using classes to assign phenomena to groups, essentially matching a system of classes with a body of phenomena to create groups which are analogous to classes" ( 1 9 7 1 * 4 4 ) . The reason Dunnell states for making such a distinction is also clear. He writes* "In the course of day to day living, a distinction between classes and groups is not necessary, for no new information is being conveyed within a singly cultural system and evaluation is not overtly con- ducted; however, for the purposes of scientific inquiry and the evaluation of its results, i t is necessary to make such a distinction. Without i t evaluation is impossible. The lack of such a dis- tinction in much of the archaeological literature has created a ;great deal of the confusion i n evidence and represents the transfer of a commonsense approach to scientific inquiry" ( 1 9 7 1 * 4 4 - 4 5 ) . To bridge the gap between the "old archaeologists" and the "new archaeologists" and to encourage archaeologists to be more explicit, Dunnell has very deliberately defined most of the terminology used in archaeology so that archaeologists can communicate efficiently. Besides clarifying the terminology from time to time throughout his book, he also points out the limitations and charac- 1 9 teristics of each different classification. Only when classificatory problems are fully resolved, will archaeo- logy be able to reach the realm of science. Nevertheless, the reason for this difficulty in classification (or arrangement) is also that archaeologists are easily led into subjectivity when making decisions on classification. Most archaeologists are aware of the problem. A.C. Spaulding1s "Statistical Techniques for the Discovery of Artifact Types" clearly illustrates this point. In this article, he states that i f artifact types really do exist, then they can be discovered by statistical methods ( 1 9 5 3 * 3 0 5 ) . In order to gain some knowledge of the use of classi- fication in Chinese archaeology, I have selected a number of examples dealing with both artifacts and larger cultural units. L i Chi^s Hsiao-t*un. vol. I l l , Fascicle I, part I, was the f i r s t work done. In this work, the author tried to use !a consistent format for organizing the material from one site. The primary scheme used by L i Chi ( 1 9 5 6 * 3 6 - 3 7 ) in his classification of pottery vessels from Hsiao-Vun is as followsi 2 0 1. The ceramics were coded using a three digit number. The shape of the bottom portion is described by the f i r s t digit. Vessels with pointed or rounded bottoms are set in the categories with the ordinal number from 000 to 0 9 9 t vessels with flat bottoms in that of 1 0 0 and 199» "the vessels with ringfeet with the ordinals between 200 and 2 9 9 , the vessels with tripods with the ordinals between 300 and 399» the vessels with four legs with the ordinals between 4 0 0 and 4 9 9 , "the covers of the vessels with the ordinals between 900 and 9 9 9 . 2. According to the shape of the highest part of the vessels within the categories divided above, the value of 1-99 (the second and third digit) is set out under the following orderi the vessels with bigger mouth and shallower depth are given a smaller value while those with smaller mouth and deeper body are given a larger value; in the meantime the angle of wall and bottom, and the construction of vessel*s l i p are also divided by several detailed criteria* the outward flaring rim is given a smaller value, while the inward curving rim is given a higher value. It is obvious that L i Chi has set very firm and objective criteria to classify and seriate the pottery vessels. 21 Dr. L i (1956J 37) comments concerning the illustrations of the pottery vessels which are arranged under the rules described above that "perhaps some of those extremely similar in shape are divided into two different groups but some of those quite different in shape are grouped into one cluster". For example, both forms of pottery of 5P and 1 5 N , which are similar in shape are divided into 005 and 015 groups (Fig. 5) (Li 1956t Corpus of Yin-hsu Pottery I), the same situation of 16G and 46D are also divided into 016 and 046 groups (Fig. 5»6) (Corpus of Yin-hsu Pottery I and II); while 107P and 107E or 107M quite different in shape are grouped into the 107 cluster (Fig. 7 ) (Corpus of Yin-hsu Pottery III), 279F and 279K into the 279 cluster (Fig. 8) (Corpus of Yin-hsu Pottery XI), 295D and 295G into the 295 group (Fig. 9 ) (Corpus of Yin-hsu Pottery XII), 309E and 309P or 309K or 309G into the 309 cluster (Fig. 10) (Corpus of Yin-hsu Pottery XIII), and so on. Nonetheless, K.C. Chang comments that L i Chi*s work is "too Scientific" (Chang 1 9 5 7 ) . It is true, since the criteria he uses ?are without much cultural significance, his classification of Hsiao-Vun Pottery is too close to natural science to be useful to a social science. Fig. 5 Examples of Typological Grouping from Hsiao-Vun. 5P and 15N are placed i n separate categories ^ (From L i , C. 1956) Fig. 6 Examples of Typological Grouping from Hsiao-Vun. *f l6G (see Fig. 5 ) and 46D are placed i n separate categories (From L i , C. 1956). rOTA IK 1078 1K l0B6 *• Fig. 7 Examples of Typological Grouping from Hsiao-t/'un. 107P and 107E or 107M are placed in one category (From L i , C. 1956). n+uryst HPKM IOO I BUM-* Y H 3 5 8 Fig. 8 Examples of Typological Grouping from Hsiao-t'un. 2 7 9 F and 279K are placed in one category (From L i , C . ^ 1956). *C« St4 2878 fc6 29IK ft 29IW fc<4> 294E ' HPKMI380 Fig. 9 Examples of Typological Grouping from Hsiao-t'un. 295D and 295G are placed in one category (From L i , C. 1956). 27 g g i O W) CD • +> d O m +» CD 1 a O o cd •H c CO •H w -d B CD o o u a rH ft • 1 CD t. •pi P. o u OS o as. o •H ON bD O • O H NO O u UN O. O ON >» iH EH At ON • <M O o o CN at to •CJ •H (1> S Hi rH CO ft O § ON X o M cn ' o H t»0 •H 28 On the other hand, two years before Dr. Li*s publi- cation, Prof. Sung Wen-hsun's early work in the classi- fication of the stone implements from the Yuan-shan shell mound was somewhat successful in overcoming the defici- encies noted above. The stone implements of the Yuan-shan shell mound which he classified were the specimens in the Dept. of Archaeology and Anthropology, National Taiwan University, collected before 1950 but without any strati- graphic control. This collection from the Yuan-shan shell mound is the largest one in Taiwan. The whole frame of his deliberated hierarchical classification or taxonomy is as follows (Sung 1954-55)» I. Edged stone tool 1. End-edged tool A. Large thin shovel-shape tool a. Large polished shovel b. Large polished shovel with neck c. Large chipped shovel d. Large horned shovel B. Large flat axe-shape tool and hoe-shape tool a. Large convex hoe b. Large spoon-shape hoe c. Large flat hoe with neck 29 C. Various large adze-shape tools a. Large polished adze b. Large chipped adze c. Large columnar adze D. Medium and small axe-shape, shovel-shape, hoe-shape tool a. Polished stone shovel with neck b. Partly polished stone hoe with neck c. Chipped stone axe d. Rough polished columnar stone axe e. Small rough polished shovel f. Small polished axe E. Medium and small adze and chisel a. Columnar adze b. Columnar chisel c. Plat chisel d. Thin chisel e. Chipped columnar chisel f. Resharpened chisels F. Stepped celt- a. Stepped columnar adze b. Stepped columnar chisel c. Stepped flat chisel G. Shouldered celt 30 2. Lateral-edged tool a. Shouldered lateral-edged tool b. Irregular shape lateral-edged tool 3. Double-edged itool a. Double-edged sounding stone 4. All-edged tool A. Spearhead B. Arrowhead a. Stone arrowhead with flat bottom. b. Stone arrowhead with concave and perforated bottom c. Stone arrowhead with flat and perforated bottom d. Stone arrowhead with convex and perforated bottom C. Razor shape all-edged tool II. Stone implement without edge 1. Hammer-shaped tool 2. Ball-shaped tool 3. Sinker 4. Lid-shaped tool 5. Tools of no definite shape 6. Other a. Perforated slender tool b. Rectangular columnar tool c. Quadrilateral flat tool d. Rectangular thin tool 31 It is obvious that his hierachy was built with "function" and shape, such as the presence or absence of the utilized edge and its varying position, as the prime criteria. Consequently, his classification with functional, or "use" criteria was more strongly connected to human activity than that of Dr. Li's work. Prof. Sung did not take the next step of many current archaeologists, namely grouping the artifacts into tool kits or activity sets. The artifaotural classifications which we have discussed so far are based on a few criteria such as vessel shape, position of cutting edges, or size. These are "dimensions" in Dunnell*s terminology (1971*71)» It i s not very satisfactory to use this Jype of cl a s s i f i - cation for larger units such as cultural phases of culture- types. The classification of the Lungshanoid cultures for instance, is not the same sort of thing as the class- fication of stone implements of pottery vessels. It should be more complex since various aspects are involved,in the concept of culture. During the last two or three decades, these problems have very rarely been discussed. Quite often, the newly excavated sites have been classified into already established local Lungshanoid cultures. 32 To determine the concept of culture and to determine the methodology used in the classificatory systems pro- posed for these Lungshanoid cultures are two of my original research goals. The methodology used in the traditional classifica- tions of Lungshanoid sites has never been clearly dis- cussed as yet. It appears that there are no unquestionable or specific criteria to check how each site f i t s into the general classification; sites are viewed only in a very general way during the process of classification. Different cultures were named after the original sites in the various regions of South China. Consequently, the sites geographically close to each other were easily classified into the same group or culture phase. Although they perhaps will deny that they have been overly arbitrary in their decision-making, the archaeologists responsible for this classification have used a rather high degree of subjectivity. Not only the Ta-wen-k^u culture, the Ch*u-chia-ling culture, the Ch'ing-lien-kang culture are so treated, but also the most recent classification of the Gb^ing-lien-kang culture into Chiang-pei, Chiang-nan and their various stages was done with unclear:,criteria or methodology (Wu 1 9 7 3 ) • In other words, although the 33 process of classification employed here looks like that of the Midwestern taxonomic method developed by W.C. Mckern (1939)* the diagnostic traits used in this parti- cular Lungshanoid classification are fairJ^Qbsisure. For instances, basket weaving implements, wooden oars, wooden pestles and mortars, and a high percentage of black ware seem to be regarded as the diagnostic traits for the Liang-chu culture, while eggshell painted pottery and spindle whorls of painted pottery seem to serve as the diagnostic traits for the Ch'u-chia-ling culture. But most of the time the vague phrase "from the point of view of the cultural characteristic" (tit JnUfi ftttyis widely employed. However, the basic concepts of culture underlying these classifications seemingly are the total traits of the artifacts. In other words, they use a nor- mative approach and treat culture as a body of shared ideas, values and beliefs — the "norms" of a human group as in Flannery's description of culture history. This is one of the differing views of culture summarized by Flannery ( 1 9 6 7 : 1 0 3 ) . Since the time that culture was defined as man's extrasomatic adaptation to his total sociological and 34 ecological environment (White 1959* 15) i the necessity of a different kind of classification has become more urgent. Because artifacts are considered to be the product of human ac t i v i t i e s , only a class i f i c a t i o n which reflects the use of artifacts and the pattern of human behavior behind their manufacture i s meaningful. In other words, an ideal cl a s s i f i c a t i o n should be neither too subjective nor too objective to reflect the artifacts* cultural or social meaning. Since i t should deal with the objects i n cultural terms, cluster analysis seems genuinely helpful i n attempting to attain this aim of classification. The main reason for this l i e s i n the fact that the process of cluster analysis involves two funda- mental stepsi subjectively selecting out as many as possible of the features which have cultural significance i n many aspects of culture, and objectively viewing the features with equal weighting during the process of s t a t i s t i c computing. 35 Chapter 4 General Idea of Cluster Analysis and Multidimensional Scaling Cluster analysis is a kind of statistical clustering. According to R.C. .Bunnell's (1971« 95) classification, i t belongs to the category of non-elassificatory arrangement (Fig. 4 ) . Cluster analysis is also a kind of hierachical grouping technique. This techniqueswas f i r s t developed in biological taxonomy by Robert R. Sokal and Peter H. A. Sneath ( 1 9 6 3 ) . Basically they tried to use numeric values to describe the characters and then to/ calculate a taxonomic distance between organisms. Not long after this idea deve- loped in biology, i t was borrowed by archaeologists to analyse their numerous data. With the help of computers, many archaeologists, such as Matson and True (1970, 1 9 7 4 ) , or Binford (1966) and Hi l l (1970) have successfully used this technique to attempt to illuminate some difficult aspects of archaeology — the social structure and human behavior, for instance. No wonder Michael R. Anderberg stated that "Cluster Analysis may be used to reveal 36 structure and relations in the data. It is a tool of discovery" ( 1 9 7 3 * 4 ) . According to Sokal and Sneath ( 1 9 6 3 * 1 2 0 ) , and M.R. Anderberg ( 1 9 7 3 * 1 1 ) * "data units" are the logical fundamental units in a large majority of individual or- ganisms and can be looked at in terms of "subject", "observation", "case", "element", "object", or "event". Consequently, in archaeological applications, stone a r t i - facts, pottery type, design element, or even the archaeo- logical site can frequently be used as data units or OTU (operational taxonomic units). The other basic term in the use of cluster analysis is "variable". The distance of the difference or the degree of the resemblance among the data units must be consistently described in terms of their characteristics, attributes, class memberships, traits, and other such properties. Collectively, these descriptors are called "variables". The resemblance of data units completely depends on the values of a l l the variables. One very important and basic axiom in this technique of analysis is that every character (variable) is treated as having equal weight. Thus, i t is possible to describe the data unit by calculating the value of the 37 presence and the absence of each variable in that unit. According to the coded value of the presence and absence of each variable in every single data unit, we can decide the degree of resemblance of any pair of data units. The degree of resemblance of any pair of data units can be described by a concrete figure — a similarity coe- fficient. According to Sokal and Sneath (1963* 129-130), and Anderberg (1973* 89), there are many formulas to calculate different similarity coefficients. After figuring out the similarity coefficients, not only are we able to do cluster analysis with them, but one can also do multi- dimensional scaling using these figures. The value of these coefficients is always between 0.0 and 1.0. The higher the value of the similarity coefficients the more resemblance between the pair of data units. In addition to the similarity coefficient, the degree of the resemblance among the data units also can be described by the distance of the difference among the data units. The distance of the difference of data units is calculated by the formula of 1.0-coefficient. Accordingly, the lower the value of the distance the higher the degree of the resemblance among data units. 38 After the correlation coefficients are figured out, the next step is to group a l l the data units into clusters depending on these coefficients or distances. This step is also open to many methods. There are single linkage (nearest neighbor), complete linkage (farthest neighbor), simple average, group average (unweighted pair group), - Lance-Williams flexible, and Ward's method (Matson and True l»9?4i 5 5 - 6 1 ) . However, not a l l of them are commonly used in the field of archaeology. Each method has different characteristics. Whatever method is going to be used, a half matrix of coefficients or distance should be prepared f i r s t . The fi r s t linkage of any method should be the smallest distance or the highest coefficient. Then the nearest neighbor should be chosen to make a new half matrix, i f the single linkage method is used; while the farthest neighbor should be chosen in cases of complete linkage. The main procedure is to use these two methods alternately until a l l data units are linked together. The average method differs from the above only in using the average instead of the nearest or farthest neighbor. After the points of linkage are chosen, a dendrogram for these linkage clusters can be drawn on the basis of these linkage points. 39 On the other hand, basing the analysis on a matrix of correlation coefficients, the distance between two points (data units) also can be scaled in several dimen- sions. The distances among data units can be easily expounded on a pair of coordinates (two dimensions). But, when there are more than 3 characters (variables), the scaling of the distance Ibetween two data units becomes more and more complicated. In Euclidean hyperspace, the distance between two points (i.e. data units) is determined by a l l their dimensions (i.e. characters). Thus, the distance between 2 points in four dimensional space is defined as dJ=(Wv-W^)+(Xi-X^)%(Y1-Y^)+(Z^-Z^jL. In an n-space, a hyperspace of n dimensions, therefore, the maxi- mum distance will be n (based on characters with maximal values of unity). These multidimensional distances are scaled independently by each pair of dimensions. Sometimes the distribution of data units in each separate configura- tion based on two dimensions is helpful in interpreting certain phenomena in archaeology. Neither similarity coefficients nor multidimensional scaling can be calculated by hand efficiently. Fortunately, nowadays the well-developed computer can execute these complicated processes in a few seconds. 40 Chapter 5 The Use of Cluster Analysis and Multidimensional Scaling The use of these techniques in Chinese archaeology has not teen attempted "before. I believe these methods contribute a more objective but still,culturally signifi- cant approach to the procedure of classification. From the general description of these techniques above, i t is quite obvious that the variables (or features) are probably the most important links in the process of analysis, because the resemblance of data units completely depends on the values of the variables, the cluster of data units also depends on the value of the a l l of the variables in each data unit. Thus, the choice of variables to test the resemblance and to make the cluster is the crucial factor. This procedure of choosing variables i s , in my opinion, the concrete expression of the values of social science, since we have to make a somewhat subjective decision in selecting variables which have cultural and/or social meaning. After choosing the variables, a l l the remaining 41 procedures are completely objective and fixed within the realm of natural science. A pilot study of these techniques in Chinese archaeo- logy on 19 Lungshanoid sites chosen for their relative importance is presented here (Pig. 1 1 ) . These sites and their traditional classification are listed below (KKYCS 1961, K.C. Chang 1968, Wu 1973* 4 5 - 6 1 ) . I. The Ch^u-chia-ling Culture1 No. 13 (Ch'u-chia-ling) No. 14 (Shih-chia-ho) II. The Liang-chu Culture* No. 10 (Ch'ien-shan-yang) No. 11 (Shui-t'ien-pan) No. 12 (Liang-chu) No. 18 (Lao-ho-shan) III. The T'an-shih-shan Culturet No. 16 (T'an-shih-shan) IV. The Ta-wen-k*ou Culture* No. 15 (Ta-wen-k'ou) No. 17 (Kang-shang-ts'un) V. The Ch'ing-lien-kang Culture* Va. Chiang-pei type 4-2 Fig. 11 Location of 19 Lungshanoid Sites I t Erh-chien-ts'un 2, Tai-kang-shih 3$ Ch'ing-lien-kang 4, Ta-tun-tzu 5, Liu-lin 6, Hua-ting 7, Ma-chia-pin 8, Pei-yin-yan-ying 9» Sung-tse 10, Ch'ien-shan-yang 11, Shui-t»ien-pan 12, Liang-chu 13, Ch'u-chia-ling 14, Shih-chia-ho 15, Ta-wen-k*ou 16, T'an-shih-shan 17, Kang-shang-ts»un 18, Lao-ho-shan 19, Feng-pi-t'ou 43 No. 1, (Erh-chien-ts^n) No. 3» (Ch'ing-lien-kang) No. 4, (Ta-tun-tzu) No. 5. (Liu-lin) No. 6, (Hua-t»ing) Vb. Chiang-nan type No. 2, (Tai-kang-shih) No. 7, (Ma-chia-pin) No. 8, (Pei-yin-yan-ying) No. 9 t (Sung-tse) VI. The Feng-pi-t'ou site (Site No. 1 9 ) . The study was carried out in the following manner. Eighty characters were, selected for the 1 9 selected sites which are scattered throughout Southeast China. These characters vwere scored as; .present or /absent for each archaeological site. The characters are mainly pottery and l i t h i c traits; no floral, faunal, or locational data were used. The main reason for this is that the degree of description in each individual site report is different. Some of them are too detailed for our purposes while others are too general. However, by using 80 characters, I felt that the sample would be sufficiently large to minimize any distortion of site relationships that might 44 follow from including possibly interdependent characters (Appendix B). The use of a combined program written by R.G. Matson, Dept. of Anthropology and Sociology, the University of British Columbia, (for the main program), J.J. Wood, Dept. of Anthropology, Northern Arizona University, (for the sub- routine of hierachical grouping), and R.J. Sampson, (for the subroutine to print dendrogram), permits the quick and accurate calculation of Jaccard's coefficient of distance among each pair of sites and the efficient grouping of similarity. The results of the cluster analysis (simple average) appear in the form of a dendrogram (Fig. 1 2 ) . "There is no necessary implication of •genetic* relation- ship in the dendrogram; i t should be interpreted merely as an indicator of taxonomic distance, not as a family tree" (Matson & True 1 9 7 0 i 1 2 0 2 ) . The dendrogram exhibits the following patterns* (cluster I) sites 14, 13; (cluster II) sites 18, 12, 11, 10; (cluster III) site 16; (cluster IV) sites 17, 15; (cluster V) sites 6, 5, 4; (cluster VI) sites 19, 8, 9t 7, 2; and (cluster VII) sites 3 and 1. Cluster I, II, III and IV show exactly the same results as that of traditional classifications. In other words, 45 VA o CA NO o CA oo VA CA VA O 00 CO •4- VA O O o VA o ^ - ON *̂ ON CA VA 00 o CM VA IN. ON o\ VA O VA O CA CA CM CM CM i H • • • • • • • o O o O o O O VII 1 3 VI IV -Hi - l l 7 9 8 19 4 5 6 15 17 16 10 11 12 18 13 14 Pig. 12 Dendrogram of simple average linkage cluster analysis on a matrix of Jaccard,s distances. 4 6 cluster I fi t s into the Ch'u-chia-ling culture, cluster II fi t s into the Liang-chu culture, cluster III f i t s into the T»an-shih-shan culture and cluster IV f i t s into the Ta-wen-k*ou culture. Nevertheless, the traditional Chiang- pei type of the Ch^ing-lien-kang culture is shared "by cluster V and VII in the dendrogram of the cluster analysis. On the other hand, the Feng-pi-t»ou site in Southeast Taiwan and the traditional Chiang-nan type of the Ch'ing-lien-kang culture are ̂ grouped into a single cluster (VI). The multidimensional scaling of these sites has also been carried out. This complicated process also depends on the manipulation of a computer. The computer program written for these multidimensional scalings is based on Torgerson's method ( 1 9 5 2 , 1 9 5 8 ) . "£» • .r.aliaî u®, a "In this technique, a matrix of the products of the distance from the centroid or origin of the configura- tion is calculated from a distance matrix. This product matrix then canlbe solved for the eigenvectors. The resulting factor matrix is the solution configuration, with the fi r s t factor being the most important axis, the second the next most important, and so on" (Matson & True 1 9 7 4 : 6 4 ) . The solution configurations do not quite f i t the traditional classification. The percentage of trace of the f i r s t 4 47 factors (4 dimensions) are as follows* 22.63$, 17.33$, 10.94$, 8.94$, (Fig. 13, 14). Figure 13 is a plot of the configuration on the most important vectors (first and second dimensions). It seems that the patterning of the sites on these two vectors agree with the clusters represented by the dendrogram. The vertical dimension, in particular, (vector 1) can be seen as a shift in pottery character and variation of implements. It is quite clear that cluster II (sites 10, 11, 12, 18), in which the pottery is predominantly black ware and various kinds of impression or incision, is distributed in the top of the configuration. Cluster V (sites 4, 5, 6 ) , on the other hand, in which the pottery character is dominated by red ware appears at the bottom of the configuration. In^addition, cluster II is rich in wooden implements but lacks bone implements; cluster V is completely the reverse. Although the Cl4 date of 3835 B.C. (LIA 1974* 334) for point 4 (Ta-tun-tsu site) is the earliest date among a l l the Lungshanoid sites, i t seems indefensible to interpret Vector 1 as the temporal dimen- sion since point 16 (the T'an-shih-shan site) is dated 1140 B.C. (LIA 1974(5)* 337) and point 10 (Ch»ien-shan- yang site) is dated to 2750 100 B.C. (LIA 1972(5)* 57). 48 •12 • 10 18 .11 >14 • 13 u o #19 * 8 * 2 +1 *7 AL5 W5 cluster VII + * 4 cluster VI * « Yectorj y cluster V Vi Pig. 1 3 Plot of site distributions duster IV * obtained by multidimensional c l u s t e r 1 1 1 * scaling (Vectors 1 and 2 ) . cluster II © cluster I © 49 •18 A15 0 1 3 <?L4 +L •12 A17 ©11 tf6 *|4 M5 ? + 3 *2 #8 * 9 * 1 9 o 10 u o +> o cluster VII + cluster VI cluster V ^ cluster IV A cluster III & cluster II o cluster I © % 6 Vector 3 Fig. 14 Plot of site distributions obtained by multidimensional scaling (Vector 3 and 4). 50 The isolation of site 16 i n the dendrogram i s shown when plotted on vector 3 and 4 i n Figure 14. 51 Chapter 6 Discussion What do the clusters presented here mean? No doubt they are the result of the research method — they may provide a new classification or R.C. Dunnell*s (1971* 87- 110) non-classificatory arrangement of the Lungshanoid sites. Thus, the seven clusters can stand for seven new categories of the 19 Lungshanoid sites. Since both the method and the results of this classification are different from those of the previous studies, some additional comments will be made. The different methods of classification have been discussed above, thus the discussion here should emphasize the results of cluster analysis including its similarities and differences with traditional classification. Cluster I, II, III and IV f i t into the Ch'u-chia-ling culture, the Liang-chu culture, the T'an-shih-shan culture and the Ta- wen-lc'ou culture. This means that traditional cla s s i f i - cation of those sites is maintained in the cluster analysis 52 and vice versa. Although the remaining clusters are different from the traditional ones, they s t i l l reveal some degree of similarity to traditional classification. However, there must be some importance to the fact that the Feng-pi-t»ou site is put into the same cluster with the traditional Chiang-nan type of the Chfing-lien-kang culture and the traditional Chiang-pei type of Ch ,ing-lien- kang culture is divided into two clusters. The position of Feng-pi-t*ou and the Chiang-nan types of the Ch'ing- lien-kang culture in the same cluster has at least two possible explanationsi (1) The similarity in their cultural traits was ignored because of the geographical distance and political isolation of Chinese archaeologists on the mainland and in Taiwan. (2) Some mistakes may be present in the process of the cluster analysis carried out here, probably in the original selection of attributes. The reasons for the traditional Chiang-pei type of the Cl^ing- lien-kang culture being divided into two clusters can also be interpreted in two ways* (1) The difference or distance of cluster VII and cluster V in cultural traits really exists. (2) Some mistakes have been made during the process of doing the cluster analysis, probably in attribute selection. However, because of the close f i t of clusters 53 I, II, III, and IV with those of the traditional c l a s s i f i - cation, the selection of attributes appears to have sub- stantial validity. As a matter of fact, the meaning of the clusters can be further discussed with additional implications. The following questions can be asked« Do the clusters stand for different time periods, different peoples, different languages, different races, or even different technologies? Some of these alternatives are quite obvious, while some are not. After the publication of C14 dates in 1 9 7 2 and 1 9 7 4 , the different and/or somewhat overlapping time periods of the clusters in the Lungshanoid horizon were made clear. Since the T'an-shih-shan site is dated at 1140+90 B.C. (LIA 1 9 7 4 ) , cluster III should be considered the latest of those clusters presented in this paper with C14 dates. The earliest clusters should be cluster V and VI, because the Ta-tun-tzu site of cluster V is dated to 3835*105 B.C. (LIA 1 9 7 4 ) and the Sung-tse site of cluster VI is dated to 3395*105 B.C. (LIA 1 9 7 2 ) . Cluster II, in which the rice husks of Ch'ien-shan-yang site are dated to 2750±100 B.C. (LIA 1 9 7 2 ) , should be considered later than cluster V and VI but earlier than cluster I and III (Table 2 ) . 54 TABLE 2 The Grouping Clusters With Some CI 4 Dates Cluster No. of Site Name of Site B.C.(5730*40 half-life I 13 Ch'u-chia-ling 2195*100 14 Shih-chia-ho II 10 Ch•i en-shan-yang 2750*100 11 Shui-1•i en-pan 12 Liang-chu 18 Lao-ho-shan III 16 T•an-shih-shan 1140+90 IV 15 Ta-wen-k'ou 17 Kang-shang-ts•un V 4 Ta-tun-tzu 3835*105 5 Liu-lin j 6 Hua-ting VI 2 Tai-kang-shih 7 Ma-chia-pin 8 Pei-yin-yan-ying 9 Sung-tse 3395+105 19 Peng-pi-t1ou 1460*80 VII 1 -3 Erh-chien-ts•un 6h» ing-lien-kang 55 The second question is also very important but fairly difficult to answer. Have the different clusters anything to do with different ethnic groups and their associated languages? In other words, were the different clusters produced by discrete groups of people speaking different languages? The answer to this question might shed light on the meaning of the clusters presented here. In view of the different dialects spoken in this area today, i t is easy to accept the idea that many languages were present during the Lungshanoid horizon. Consequently, i t quite possible that the different clusters may represent people speaking different languages. Sometimes in one cluster even more than one language may have been used. In ancient Chinese history, especially during the Spring and Autumn Annals and Warring States Periods, the cultures of Wu, Yueh and Ch»u have for a long time been noted as distinctive from the Chting-yuan cultures. It is very probable that cluster I was related to the ancestors of the Ch'u people. According to the research done by Prof. Wen Chung-i (1967» 1-21, 1 6 7 ) , most of the Chfu people were indigenous, except for some of the ruling class who immigrated from north China. Prof. Wen considers these to be a branch of the "Southern People", perhaps one of the "Indonesian" groups 1 56 in earlier times. Cluster II (the Liang-chu culture) was largely ancestral to the subsequent Yueh culture and cluster IV (the Ta-wen-k*ou culture) which geographi- cally overlapped the area of the Shantung Lung-shan culture falls into the area occupied by the so-called Eastern Yi people according to historical texts (Chang 19681 1 5 9 ) . According to Prof. E.G. Pulleyblank, (personal communica- tion), Eastern Yi, Wu, and Yueh may have been Mon-khmer language speaking people, and Ch'u may have been one of the Miao-Yao speakers. To deal with ethnic groupings in this prehistoric context raises the discussion to a more difficult level. First of a l l , the definition of "different ethnic groups" is not an easy one to make. If this definition rests on linguistic considerations, the discussion will be the same as above. If the definition is based on social structure, i t will be extremely difficult to see the relevance between the cluster and the ethnic groups. However, when we focus on the subsistence pattern, there are some hints for us to assume the possibility that some different clusters may represent different ethnic groups. Since natural resources, such as the flora and 57 the fauna, or even the soil type and climate, vary in different geographic areas (see'Fig. 15» 1 6 ) , i t i s obvious that the people of the different clusters employed somewhat different technologies to pursue their different subsistence patterns. Cluster II, which is especially rich in rice husks, basket-weaving implements, wooden oars, wooden pestles and mortars is a good example of technological adaptation to local environment. In addition, i t also clearly illustrates that this cluster has its own long history of development of rice cultivation with the adaptation to the swampy lowlands. In fact, the moist and swampy lowlands led to the preservation of these implements which represented peoples• activities in that time and space. Cluster I, rich in rice husks and water fowl, appears to be an example of adaptation to a swampy lowland environment as well. Nevertheless, the dendrogram of cluster linkages (Fig. 12) indicates three major clusters: VI and VII in one clus- ter, III, IV and V in the second cluster, I and II in the third cluster. On the basis of this third cluster in which the two smaller clusters are rich in rice husks, basket- weaving, wooden implements, and water fowl, i t could be 58 Fig. 1 5 . Soil regions of China (Ma, Y.C., 1 9 5 7 a ) . VI, Korichnevyi soil and brown earth. VII. Yellow-korichnevyi soil . VIII. Red and yellow earth. IX. Red earth. X. Mountain steppe soil and mountain dark-korichnevyi soil. XI. Mountain brown earth and mountain korichnevyi s o i l . (From Wang 196l« 1 7 . ) . I MONTANECONIFEROUSFOREST PREDOMINATED BY SPRJCE AND nn ' ' / " S ^ * I I MONTANE CONIFEROUS FOREST PREDOMINATED ^ • ET LARCH ~ ~ | 1 MONTANE CONIFEROUS FOREST PREDOMINATED BY SPRUCE • j t MONTANE CONIFEROUS FOREST PREDOMINATED BY FIR | i MIXED NORTHERN HARDWOOD FOREST PREDOMINATED BY BIRCH 3 | t MIXED NORTHERN HARDWOOD FOREST OF MAPLE. BASSWOPD AND BIRCH | 7. DECIDUOUS BROAD-LEAVED FOREST PREDOMINATED BY DECIDUOUS OAKS glfjS I MIXED MESOPHYTIC FOREST pllll 9L EVERGREEN BROAD-LEAVED FOREST OF EVERGREEN OAKS, SCHIMA AND LAURELS L IX EVERGREEN BROAO-LEAVED FOREST Or EVERGREEN OAKS, SCHIMA AND LAURELS. r i ~ ~}. *"TH f , m S "ASSONIANA IN SECONDARY STANDS i"| IL EVESCF-EEN BROAD-LEAVED FOREST OF EVERGREEN OAKS, SCHIMA ANO LAURELS, I'il'li'MI WITH PINUS YUNNANENSIS IN SECONDARY STANDS KB tX. THE MAIN TYPES OF NATURAL VEGETATION OF CHINA MONGOLIA. KOREA AND JAPAN »««c W W ; vtc r. tvi | Fig. 16 The Main Types of Natural Vegetation of China and Adjacent Regions (From Wang 1 9 6 l t 1 1 ) . 60 concluded that the three major clusters are the result of environmental adaptation. However, this hypothesis is weakened "by the fact that the other two clusters do not seem to reflect similar environmental adaptation, but are based on a wide variety of different characteristics. The relationship between these characteristics appeared to be too complicated for interpretation at this time. It is also very interesting that clusters I and II are clearly distributed in the upper half of Vector 1, and cluster III, IV and V are distributed in one end of Vector 2, while cluster VI and VII are distributed in the other (Fig. 1 3 ) . The multidimensional scaling confirms the cluster analysis in broad outline. It is obvious that this pilot study employing cluster analysis and multidi- mensional scaling brings us some new and difficult phenomena to explain. However, from another point of view, the com- plexity and/or multidimensional nature of the Lungshanoid cultures is also revealed by this method. 61 Chapter 7 Conclusion In view of the complexity and multidimensional nature of Lungshanoid cultures, I feel i t is meaningful to discuss further the position of the Lungshanoid cultures within Chinese Neolithic culture as a whole, especially in the relationship with the Yangshao culture in the Chung-yuan region and the Lungshan culture in the eastern part of North China. During the long debate on the relationship between the Yangshao culture and the Lungshan culture in the past two decades, the Lungshanoid cultures have gra- dually assumed greater importance. In chapter 2, I men- tioned that the debate almost reached a final conclusion when Hsin Chung-kuo t i K'ao-ku Shou Huo (The archaeology of New China) and Prof. K.C. Chang*s The Archaeology of Ancient China (1963 and 1968) were published. This con- clusion which suggested that Lungshan culture developed directly from Yangshao culture was strongly affected by the excavation of the Miao-ti-kou site, since the cultural 62 stratigraphy presented there shows that the Yangshao cultural elements were stratigraphically below those of the Lungshan cultural elements (KKYCS 1959). Besides the evidence of stratigraphy, the cultural features jiave also been examined carefully. Thus, Prof. K.C. Chang suggests (1968* 135)» "The •transitional• nature of the Miao-ti-kou II pottery is of particular significance; i t has caused many scholars to embrace the view that the Honan Lung- shan pottery could have been derived from the Yang- shao." Now, since the publication of C14 dates in 1972 and 1974, the theory ' that Miao-ti-kou II is the origin of a l l the Lungshan cultures has totally been disproved. Particularly,the date for Miao-ti-kou I of-3280*100 B.C. and for Miao-ti-kou II of 2310*95 B.C. show the impossi- bi l i t y of their continuity or the "transitional" nature; the dates of Ta-tun-tzu of 3835*105 B.C. and Sung-tse of 3395*105 B.C. reference point out a new direction from which to approach this issue. As a matter of fact, long before the CI 4 dates were published, some scholars such as Dr. L i Chi in Taiwan and Prof. Su Ping-ch fi on the mainland, had pointed out the im- possibility of that theory. The reason given by Dr. L i 63 (1963* 1-12) is based on the re-examination of the report on Miao-ti-kou. He pointed outi ( 1 ) the fundamental difference both in the method of firing and the general shapes of the ceramics of Miao-ti-kou I and II, ( 2 ) the people of Miao-ti-kou I culture seem to be much more sedentary than the Miao-ti-kou II people, ( 3 ) scapulimancy which perhaps was more important than black pottery in the Lungshan culture features was not found in Miao-ti-kou II. Dr. Li hints that the origin of the Lungshan culture should be in the east. Although Prof. Su (1965» 51-82) in his i article, "Some problems concerning the Yangshao culture", did not definitely point out the transitional nature of Miao-ti-kou II, he strongly suggested that the Ch'ing- lien-kang culture or the lower Ta-tun-tzu culture could be contemporary with the early Yangshao culture (Su 1 9 6 5 i 7 7 ) . In the same article, he also mentions that the later Yang- shao culture of the Chung-yuan was strongly influenced by the Ch'ing-lien-kang — Ta-wen-k'ou culture and the Ch*u- chia-ling culture (p. 7 9 ) . In other words, the Lungshanoid cultures with their dynamic characteristics are treated as an antecedent culture of the Lungshan cultures, which in Kiangsu is approximately as early as the Yangshao cul- ture at the Pan-p'o site. 64 Generally speaking, the main purpose of this attempt to employ cluster analysis and multidimensional scaling, is to test traditional classifications of Lungshanoid sites by a more objective means while s t i l l maintaining somewhat subjective aspects assumed to have social or cultural mean- ing. In other words, using a great number of characters (variables) as the determinants in the process of classifi- cation with the computer is viewed as more objective than the use of only a few, but the process of choosing charac- ters i t s e l f is subjective. A method including objective and subjective aspects in the procedure is optimal. Since a l l cluster analysis and multidimensional scaling are executed by the computer, the only thing left for the archaeologist to do is the choosing of characters as the variables of OTU (operational taxonomic units) or data units. This step is not only very important but also very difficult, because the variables we choose should have as much cultural and/or social significance as possible. When one wants to deal with the published material on Chinese archaeology using this technique, one faces a serious problem in terms of the nature of these site reports — the different degree of description in each individual site report. Some of them are to detailed for our purposes while others are too 65 general making i t difficult for the researcher to isolate the variables for the comparison of each site. Especially when the presence and absence of variables show similar patterns at every site, with differences only in frequen- cies, i t becomes extremely difficult to select a sufficient number of appropriate characters. It is also difficult to determine when variation within a variable is sufficiently significant to warrant the establishment of a distince variable. However, this research represents only the i n i t i a l step of statistical analysis of Lungshanoid sites in China. More detailed re-examination and study of these site reports are necessary. Nevertheless, this study does make clear the possible contributions to Chinese archaeology of further analysis using computer techniques. - END - 66 Bibliography An, Chih-min 1972 "Lueh lun wo-kuo hsin-shih-ch'i shih-tai wen-hua t i nien-tai wen-t'i," (Discussion on the chronological problems of our country's neolithic culture), K'ao-ku 1972, vol. 6, pp. 35-44. Anderberg, Michael R. 1973 Cluster Analysis for Applications, New York and London, Academic Press. Binford, Lewis R. and Sally R. 1966 "A preliminary analysis of functional variability in the Mousterian of Levallois Facies", American Anthropologist, vol. 68, pp. 238-295. Chang, K. C. 1957 "Li Chit Hsiao-t'un t'ao-ch'i shang chi", (Li Chii A classified and descriptive account with a corpus of a l l the main types of the pottery of the Yin and Pre-Yin period). Bulletin of the Dept. of Archaeology and Anthropology. National Taiwan University, vol. IX-X, p. 153. 1959 "Hua-nan yuan ku min-tsu wen-hua-shih t ' i kang", (An outline of ancient people's culture historyin south China) Bulletin of the Institute of Ethnology"Academia Sinica, no. 7. 67 1968 The Archaeology of Ancient China. New Haven and LondonJ Yale Univ. Press. 1969 "Hsin-shih-ch'i shih-tai Chung-yuan wen- hua t i k*uo chang", (The extension of nuclear area culture in neolithic), The Bulletin of the Institute of History and Philology, Aca- demia Sinica, vol. XLI, part 2, pp. 317-349. 1970 "The beginnings of agriculture in the Far East" Antiquity, vol. XLIV, no. 175, pp. 175- 185. 1973 "Radiocarbon dates from China: Some i n i t i a l interpretations", Current Anthropology, vol.14, no. 5, pp. 525-528. 1974 "Comments on the interrelationship of North China, South China, and Southeast Asia in ancient times", Journal of the Hong Kong Archaeological Society, vol. V, pp. 34-38. Davis, J. C. 1973 Statistics and Data Analysis in Geology. New York: John Wiley & Sons. Dunnell, Robert C. 1971 Systematics in Prehistory, London: Collier- Macmillan. Flannery, Kent V. 1972 "Culture history v. cultural process: A debate in American archaeology", Contemporary Arch- aeology, Mark P. Leone (ed,), Southern Illinois University Press, pp. 102-107. 68 H i l l , James N. 1970 "Broken K Pueblo: Prehistoric social organiza- tion in the American southwest", University of Arizona Anthropological Papers, no. 18. KKYCS, "K'ao-ku Yen-chiu Suo" (the Institute of Archaeology, Academia Sinica) 1959 "Miao-ti-kou and San-li-Ch'iaoJ*, Archaeological Excavations at the Yellow River Reservoirs Report, no. 2, Peking: Science Press. 1961 Hsin Chung-Kuo Ti K'ao Ku Shou Huo (Archaeo- logy in New China), Peking: Wen-wu Press. L i , Chi 1956 "Hsiao-t'un" Archaeologia Sinica no. 2, vol. 3, fascicle 1, part 1. Taipei: Institute of His- tory & Philology, Academia Sinica. 1963 "Hei-t&ao wen-hua tsai Chung-kuo shang-ku-shih so chan t i ti-wei," (Position of the Black Pottery culture in the history of ancient China), Bulletin of the Dept. of Archaeology and Anth- ropology National Taiwan University, no. 21- 22, pp. 1-12. LIA (Laboratory of the Institute of Archaeology, Academia Sinica) 1972a "Report on C14 Dates (l)"K'ao-ku vol. 1. pp. 52-56. 1972b "Report on C14 Dates (2)" K'ao-ku vol. 5. pp. 56-58. 1974 "Report on C14 Dates ( 3 ) " K'ao-ku vol. 5. pp. 333-338. 69 Matson, R.G. and True, D.L. 1970 "Cluster analysis and multidimensional scaling of archaeological sites in Northern Chile," Science, vol. 169, pp. 1201-1203. 1974 "Site relationships at Quebrada Tarapaca, Chile: A comparison of clustering and scaling techniques", American Antiquity, vol. 39» no. 1, pp. 51-74. McKern, W.C. 1939 "The midwestern taxonomic method as an aid to archaeological culture study", American Antiquity, vol. 4, no. 4, pp. 301-313. Mizuno, Seiichi 1956 "Prehistoric Chinas Yang-shao and Pu-chao- chai", Proceedings of the Eighth Pacific Science Congress and the Fourth Far Eastern Prehistory Association, vol. I, fascicle 1. Sauer, Carl 0. 1952 Agricultural Origins and Dispersals. New York: American Geographic Society. Sekino, Takeshi 1956 "On the black and grey pottery of ancient China", Proceedings of the Eighth Pacific Sci- ence Congress and the Fourth Far Eastern Pre- history Association, vol. I, fascicle 1. Sokal, Robert R. and Peter H. Sneath 1963 Principles of Numerical Taxonomy, San Francisco: W. H. Freeman. 70 Solheim5II, W.G. 1969 "Reworking Southeast Asian prehistory", Paideuma, Band XV, pp. 125-139. Spaulding, A.C. 1953 "Statistical techniques for $heIdiscovery~df artifact types", American Antiquity, vol. 18, no. 4, pp. 305-313. Su, Ping-ch»i 1965 "Kuan yu Yangshao wen-kuo t i jo-kan wen-fi", (Some problems concering the Yangshao culture), K'ao-ku Hsueh-pao 1965, vol. 1, pp. 51-82. Peking: Wen-wu Press. Sung, Wen-hsun 1954-55 "Pen hsi chiu tsang Yuan-shan shih ch*i", (Stone implements from Yuan-shan shell mound collected "before 1950) (Parts 1-3), Bulletin of the Dept. of Archaeology and Anthropology, National Taiwan University, no. 4, pp. 28-38, nO. 5, pp. 44-58, no. 6, pp. 34-45. Tsukada, M. 1966 "Late pleistocene vegetation and climate in Taiwan (Formosa)", Proc. Nat. Acad. Sci., vol. 55. PP. 543-548. Wang, Chi-wu 1961 "The forests of China", Maria Moors Cabot Foundation Publication Series, no. 5. Harvard University. 71 Wen, Chung-i 1967 "Ch'u wen-hua t i yen-chiu," (Aspects of the culture of Ch'u), Monograph of the Institute of Ethnology, no. 12, Taipei: Institute of Ethnology, Academia Sinica. White, Leslie A. 1959 The Evolution of Culture. New York: McGraw- H i l l . Wu, Shan-ching 1973 "Lueh lun Ch'ing-lien-kang wen hua," (Discussion on Ch^ng-lien-kang culture), Wen-wu, 1973 vol. 6, pp. 45-61. 72 APPENDIX A Original Reports of 19 Sites Site No. 1. K'ao-ku. 1962(3). 2. K'ao-ku. 1962(3). 3. K'ao-ku,Hsueh-pao. 1955(9) and K'ao-ku T'ung-hsun. 1958(10). 4. K'ao-ku Hsueh-pao. 1964(2). 5. K»ao-ku Hsueh-pao. 196l(l) and 1965(2). 6. Wen-wu Tsan-k&ao T'zu-liao 1 9 5 6 ( 7 ) . 7. K'ao-ku. 1961(7). 8. K'ao-ku Hsueh-pao. 1958(1). 9. K * ao-ku Hsueh-pao. 1962(2). 10. K'ao-ku Hsueh-pao. 1960(2). 11. K'ao-ku Hsueh-pao. 1960(2). 12. Chekiang Neolithic Culture Illustration. 1958. 13. K'ao-ku T'ung-hsun. 1956(3). 14. K*ao-ku T'ung-hsun. 1956(3). 15« Wen-wu. 1959(10). 16 . K'ao-ku Hsueh-pao. 1955(10). 17. Wen-wu. 1959(10). 18. K'ao-ku Hsueh-pao. 1958(2). 19. Chang, K.C. Fengpitou,^Tapenkeng. and the Prehistory of Taiwan. 1969. APPENDIX B Attribute List 1. predominance of sandy red ware 2. sandy red ware 3. predominance of sandy grey ware 4. sandy grey ware 5. predominance of fine red ware 6. fine red ware 7. predominance of black ware 8. black ware 9. predominance of grey ware 10. grey ware 11. painted pottery 12. black pottery with red painting 13* eggshell pottery 14. eggshell painted pottery 15. spindle of painted pottery 16. pottery spindle ^17. pottery ball 18. pottery pestle and/or mortar 19. pottery paddle and/or pad 20. pottery bracelet and/or ring 21. pottery net sinker 74 22. white pottery 23. ting 24. tou (without the cut-out holes) 25. tou (with the cut-out holes) 26. jar 27. pot 28. bowl 29. basin 30. dish 31. beaker 32. cistern (crock) 33. I i 34. kui 35. tsun 36.1 fu 37. l i p spouted water vessel 38. pointed bottom water vessel 39. long neck bottle (or jar) 40. cord impression 41. mat impression 42. check impression 43. basket impression 44. small dots punctuated 45. incised by comb and/or fingernail 46. engraving 47. applique decoration 48. net impression 49. shell impression 50. stone axe (cylinder, without hole) 51. stone axe ( f l a t , with hole) 52. stone adze 53. stone chisel 54. stone knife 55. stone knife with holes 56. stone pestle and/or mortar 57. grindstone (or whetstone) 58. stone spindle 59. stone net sinker 60. stone plough 61. stone b a l l 62. stone arrowhead 63. semilunar stone knife 64. stone axe with shoulder 65. stone adze with step (step adze) 66. stone hoe 67. shoe-shaped stone knife 68. ring of jade or agate 69. oyster shell implement 70. basket weaving (implement) 71. wooden oar 72. wooden pestle and/or mortar 73. bone needle 74. bone ornament 75. bone arrowhead 76. bone awl 77. bone knife 78. bone harpoon 79. bone chisel 80. head east and/or south (burials) 77 APPENDIX C CHINESE CHARACTERS FOR PROPER NAMES AND TECHNICAL TERMS An Chih-min Q -JxL Chekiang j i _ Ch'eng-tzu-yai \ j£ Chiang-pei 3-i *t, Chiang-nan >*. $$) Ch'ien-shan-yang Chfing-lien-kang -| ̂  ^ Ch'u-chia-ling yg £ ^ Ch'u ^ Chueh-mu-chiao -fl ^ ^ Chung-yuan K\ ̂  % Erh-chien-ts'un J=- ^ j " Feng-pi-t»ou / > | L ] | P u k i e n it Hang-chow Bay jjj^ -j-lj hang-Vu ^ Honan ^ Hou-kang )% (£) Hsiao-t»un ^ Hsien-jen Tung /j/L A. Ms) Hsin Chung-kuo t i £fj ij> ig) $.|) K»ao-ku Shou Huo ^ & H£ ?8 Hua-t'ing- ^ t , /y Huang-chien-shu ^ ^ Jih-yueh-t»an ^ Kang-shang-ts•un /g) -t * f Kansu "fr K»ao-ku ^ £ K» ao-ku Hsueh-pao ^ 4 ^ K'ao-ku T'ung-hsun ^ £ Kiangsi ̂ i . Kiangsu JJ- Kuei-jen jf|» ^ kui Lao-ho-shan 4* l i ft Li Chi Liang-chu Liu-lin Lung-shan Ma-chia-pin 78 Miao-ti-kou Miao-Yao % fft& Mien-chih Hsien ^ |£ Mizuno Seiichi ^ gj,^^. Pan-p'o -^L P • ao-ma-ling j g ^ Pei-yin-yan-ying jig ^ Pohai Bay ^ *| Sekino Takeshi ff[ Shang Shansi zl» r& Shantung d* Shensi Shih-chia-ho Shih Chang-ju ^ j | -#a Shui-t'ien-pan ^jc # Ping-chti jij Su Sung-tse Sung Wen-hsun ^2 Ta-ho-ts'un ^ >>j ^.j" Ta-tun-tzu ^ :J- Ta-wen-k'ou X >iL a T'ai-kang-shih J-^ ^ Tainan p T*an-shih-shan ^ iL> tou £ Tsukada Matsuo J rg^ # i ^ . tsun Wan-nien Wang-wan Wei-shui , ̂ Wen Chung-i ^ ^ — Wen-wu Wen -wu Tsan-k'ao ^ ^ | i ̂ . T fzu-liao Wu ^ Wu Chin-ting % £ ^ Wu Shan-ching % ^ % Yin-hsu Yuan-shan j^j JU Yang-shao-ts'un 4<j> -̂ f£ ^ Yang-tzu River , Yi £ Yueh


Citation Scheme:


Usage Statistics

Country Views Downloads
United States 1 0
China 1 0
City Views Downloads
Unknown 1 0
Beijing 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}


Share to:


Related Items