This article is a revision of a paper presented at the 73rd Annual Meeting of the American Anthro-pological Association in Mexico City, 19–24 Nov. 1974. I gratefully acknowledge grants from the East Asian Institute of Columbia University which made possible the field research and preliminary analysis, and grants from the School of Arts and Science at Clarkson College of Technology which enabled the preparation and presentation of the paper and this article.

¹ For an excellent analysis of political machines, see Scott, James C., Comparative Political Corruption (Englewood Cliffs, N.J.: Prentice-Hall, 1972), chs. 6–9Google Scholar.

² However, in some areas of mixed ethno-linguistic settlement—such as Pingtung and Miaoli—factions comprise both Hokkien and Hakka leaders and supporters.

³ The results of five weeks' field research in Kaohsiung Municipality are briefly reported in my “Optimism in Taiwan's Port Centre,” Far Eastern Economic Review, LXXV, 6 (1972), pp. 28–29.

⁴ A pseudonym in honor of a goddess much beloved by residents of the field site and the people of Taiwan in general.

⁵ Taiwan's Mainlanders have concentrated in urban areas and near military bases.

⁶ In urban townships, the Chinese term for administrative village is li.

⁷ Capitalize d Roman letters represent administrative villages. Thus natural village BC consists of two administrative villages, B and C. Natural villages D1 and D2 have been combined into one administrative village, D.

⁸ Non-speakers of Chinese indicate that such translations as “relationships” and “connections” are misleading and confusing, as kuan-hsi are not equivalent to either English word. Thus, I shall use the Chinese term.

⁹ For the classic analysis of kan-ch'ing, see Fried, Morton H., Fabric of Chinese Society, New York: Praeger, 1953 (reprinted New York: Octagon, 1969)Google Scholar.

¹⁰ Generally, at higher levels, the locality kuan-hsi ties two persons who are residing away from their native place, in'a n administrative center. The general Chinese term for the locality kuan-hsi, t'ung-hsiang (Hok.: tong-hiong), can be literally translated as the ‘same native place’ kuan-hsi. (Hsiang, the same character as “rural township,” here means the less specific ‘native place’.) Within Matsu, almost all political leaders and voters continue to live in their native place, and the term “locality” is used (Mand.: ti-yü Hok.: te-hng [for which the Mand. romanization is ti-fang]). Regardless of residence or terminology, however, the kuan-hsi remains the same, a commonality of shared identification with native place. For discussion of locality at the central level in modern Taiwan, see my “Recent Leadership and Political Trends in Taiwan,” The China Quarterly, 45 (Jan/Mar 1971), pp. 136–43.

¹¹ The classmate kuan-hsi helped some younger candidates for Village Head and Township Assembly obtain votes from villagers of similar age. At the county level, several Matsu Township leaders who had graduated from a technical school helped a non-Matsu fellow alumnus obtain Matsu votes.

¹² This section summarizes a more extensive analysis of kuan-hsi in Matsu Township appearing in my “Local Politics in Rural Taiwan: A Field Study of Kuan-hsi, Face, and Faction in Matsu Township” (Ph.D. dissertation, Columbia Univ., 1975), ch. 3.

¹³ In Matsu Township, political office gave little power and few opportunities for profit; ibid., pp. 124–31.

¹⁴ For analysis of Matsu factions, ibid., ch. 4.

¹⁵ “Marketing and Social Structure in Rural China,” The Journal of Asian Studies, XXIV, 1–3. (1964–65). pp. 3–43, 195–228, 363–99.

¹⁶ Ibid., pp. 32flf.

¹⁷ “Marketing on the Changhua Plain, Taiwan,” in Economic Organization in Chinese Society, W. E. Willmott (ed.), (Stanford: Stanford Univ. Press, I97²). PP. 215–59; also his “Town and Country: Central-Place Theory and Chinese Marketing Systerns with Particular Reference to Southwestern Changhua Hsien, Taiwan” (Ph.D. dissertation, Cornell, 1973).

¹⁸ Map 2 suggests why the three natural villages combined into Administrative Village I have difficulty working together: Each natural village belongs to a different second-level marketing system. The small size of these villages, their isolation, and their lack of solidarity in administrative village affairs have given them very little voice in township affairs.

¹⁹ “Marketing” (n. 15 above), p. 36.

²⁰ The number of households visited in each natural village ranged from 5 to 27 and averaged 14.52, with a median of 13 households per village. The number of marriages recorded per village ranged from 28 to 271 and averaged 120, with a median of 107 per village. The average number of marriages recorded per household was 8.26. The surname of each household was recorded to verify proper surname distribution for each village and to aid in uncovering twice-recorded data. An effort was made to sample various sections of each village and to obtain data primarily from households of “apparent” average wealth. Very poor and wealthy households were sampled in lesser numbers.

²¹ Crissman, “Town” (n. 17 above), ch. 6.

²² Ibid., p. 286.

²³ Ibid., p. 287. See this article's Appendix for additional findings of the Matsu affinal kinship survey.

²⁴ For a more detailed analysis of Matsu temples and their political roles, see my “Local” (n. 12 above), pp. 192–202.

²⁸ Most of the thirty-odd church members belonged to the same Village N2 lineage. My hypothesis that the church would cause factionalism in Village N2 proved incorrect; the affinal kinship survey showed extensive Christian-'Buddhist” in-termarriage, despite the explicit Christian preference for all-Christian marriages. A short field study revealed that Christian sects do form the basis of factions, however, in several aboriginal areas of Taiwan.

²⁶ Before 1920, Villages Di and D2 were grouped with what now are Matsu Township villages, but in 1920 they became part of West Township. Village D became part of Matsu Township in 1959.

²⁷ Pao (‘fort’) was the most common term, but the terminology for this unit varied according to locale. Other terms used in Taiwan included li, fang, and even hsiang. In modern Hokkien terminology, chuang means a natural village; however, during the Ch'ing it meant an administrative village (even though chuang more often consisted of only one natural village than is the case with ts'un today). After 1920, the Japanese used chuang to mean town- ship in exactly the same sense hsiang is utilized at present.

²⁸ The best and most detailed works are T'ai-wan sheng t'ung-chih [hereafter TWSTC] (A Comprehensive History of Taiwan Province), chüan yi (I), t'u-ti chih, chiang yü p'ien (Taichung: T'ai-wan sheng wen-hsien wei-yuan hui, 1970); and Hung Min-lin, Ch'en Han-kuang, & Liao Han-ch'en (eds.), T'ai-wan pao-t'u chi (Atlas of Pao in Taiwan), Taichung: T'ai-wan sheng wen-hsien wei-yüan-hui, 1969. Both works have excellent tables showing administrative evolution at the township and village level, but TWSTC takes the modern township as the basic unit while Hung et al. takes the pao as the basic unit. Although the tables in Hung et al. are less accurate and detailed than those in TWSTC, the Hung volume has the advantage of extensive, detailed maps dating from the early Japanese period (which prove the greater accuracy of the TWSTC tables). Of several other works consulted on sub-county government and politics in Taiwan during the Ch'ing, by far the most useful proved to be Tai Yen-hui, “Ch'ing-ta i T'ai-wan hsiang-chuang chih chien-li chi ch'i tsu-chih” (The Establishment and Organization of Townships and Villages in Taiwan during the Ch'ing Dynasty), Tai-wan ching-chishih chiu cbi (Taiwan Economic History, Compilation No. 9), No. 76 in T'ai-wan yen-chiu ts'ung-k'an series (Taipei: T'ai-wan yin-hang, 1963), pp. 56–85.

²⁹ The Chinese character for this pao, lacking the earth radical, differs from the pao (fort) of the Ch'ing and early Japanese periods.

³⁰ Oriental Despotism: A Comparative Study of Total Power, New Haven: Yale Univ. Press, 1957Google Scholar.

³¹ Kinship and Community in Two Chinese Villages (Stanford: Stanford Univ. Press, 1972)Google Scholar, ch. 2; also “The Sociology of Irrigation,” in Economic Organization in Chinese Society, W. E. Willmott (ed.), (Stanford: Stanford Univ. Press, 1972), pp. 193–213.

³² My own observation about the apolitical nature of Taiwan's irrigation associations received confirmation from several specialists with wide experience in irrigation throughout Asia, including Taiwan. They noted the apolitical nature of irrigation administration in all of the observed countries is probably because farmers believe irrigation to be too important to be left to the vagaries of politics. (Discussion during “Workshop on Rural Local Govern-merit in Mainland China and Taiwan' sponsored by the Rural Development Committee of the Center for Int'l Studies, Cornell, 19–20 Apr 1974.)

³³ Family form, class structure, and land tenure and reform are discussed at greater length in my “Local” (n. 12 above), pp. 220–27.