THE HISTORY OF THE HISTORY OF THE YI

STEVAN HARRELL

YI AS AN ETHNIC CATEGORY
The question "Who are the Yi?" was much more puzzling to me, a neophyte in Yi studies, than it seemed to have been to most Chinese writing on the subject either before or after 1949. The Chinese, in fact, be they scholars or ordinary southwestern peasants, seem to have always known who the Yi were or, before 1949, who the Lolo were. But to me the answer was not an entirely obvious one. There was, to begin with, considerable diversity within that group of approximately five and a half million people group defined as Yi by the Chinese People's Government. For example, I knew that they spoke languages which, while fairly closely related to each other, were by no stretch of the aural imagination mutually comprehensible. The Yiyu Jianzhi (Short Account of Yi Languages), gives figures of anywhere from 20% to 42% shared vocabulary between the Northern Dialect standard (Xide accent) and examples of the other five regional dialects of Yi. The fact that, after studying the Nuosu language of Liangshan (The Northern Dialect, in the official classification), I could in fact converse in that tongue, but could understand nothing of the Lipuo (Central Yi) language of north-central Yunnan, confirmed in practice what I had learned in theory. And when the Lipuo people told me they could understand Lisu (the language of a non-Yi ethnic group) pretty well, but could make no sense of Nuosu, I began to wonder about how the Chinese government structured its ethnic categories.
Cultural diversity was similarly puzzling. The Yi of Yunnan (outside the northwestern corner of the province, anyway), while showing considerable diversity among themselves, still seem to possess some traits in common, such as lowland agriculture, Han-style housing, and patrilocal marriage. But their society is structured on very different principles from that of the Nuosu, or Liangshan Yi, who have neolocal marriage, a highly developed patriclan system and a structure of social levels (called castes or classes or strata by various authorities) who are strictly ranked and stratum-endogamous. And the Yi of Guizhou and Northeastern Yunnan seem to have a still different structure. Some Yi have writing and some do not; those that do use similar scripts, but why are those that lack writing classed with Yi and not with Lisu or Hani, other minzu that are closely related in certain ways but have no writing system?
And then there are the disputed cases. The Sani of Lunan and Luliang counties, the folks who sell "beautiful bags" around the stone forest and in downtown Kunming, embroider their needlework with the designation Sani zu, or Sani minzu, not Yi zu. Some of them now claim that they are not in fact Yi (see Swain, this volume). And I had the opportunity myself to stay briefly with a group of people in Panzhihua City, southern Sichuan, who call themselves Shuitian zu, or "Rice-field people," though the government classifies them as a branch of Yi. They have no desire to be associated with the people they themselves consider Yi, that is the Nuosu of the surrounding hills, whom the Shuitian consider to be wild and uncouth barbarians (Harrell 1990).
Other foreign observers have also wondered about the boundaries of the Yi category, or drawn them differently from the way they have been officially drawn by the authorities. Lietard, for example, the most meticulous and unbiased of the early observers of the Yi, includes in his list of "Lolo tribes" the Lisu (1913:43-44). Mueller, in an authoritative summary published about the same time, includes the Woni of Puer and Simao, a group now officially classified as Hani (Mueller 1913:44). Mueller, perhaps with foresight, expresses the Westerner's same exasperation with the problem:
Doch hält es oft schwer zu entschieden, welche Stämme
noch als Lolo, welche als mehr selbständige Verwandte
dieses Volkes aufzufassen sind [ibid:40].
And Legendre, a doctor who spent years traveling here and there across Yi territory in Sichuan and Yunnan, is sure that the peoples in this category ought not to be classed together:
J'ai acquis cette fois la certitude qu'au Yunnan se trouvant
de nombreuses tribus qualifées "lolottes" et se disent telles,
qui n'ont presque rien de commun, physiquement et moralement,
avec celles de Leang-chan [Legendre 1913:392].0

Finally, there is the widely attested fact that the Yi themselves have never had a common name that encompasses all the people the Chinese have referred to as Yi or Lolo (Hsieh 1982:1; Vial 1898:24; Mueller 1913:39). Yi and Lolo are thus names applied by others, rather than names originally applied by the group itself; the category of Yi is one that has been constituted by outside observers.
So the problem presents itself clearly not as "Who are the Yi?" which is easily answerable by "Whoever the Nationalities Commission says they are," but rather "How did the Yi get an identity?" The quick answer to the question when phrased this way is, "Through the process of ethnic identification (minzu shibie), conducted in the 1950s, which employed Stalin's criteria of a nationality as having a common territory, common language, common economy, and common psychological makeup expressed in a common culture" (Lin Yaohua 1987; Jiang 1985). Again, the Yi were who the Nationalities Commission said they were, according to Stalin's four criteria. But it is clear from reading these retrospective accounts of the ethnic identification process that Stalin's criteria were not employed in any strict manner, but rather to confirm or legitimate distinctions for the most part already there in Chinese folk categories and in the work of scholars who wrote before the Liberation. So the question is unsolved: if the Yi already had an identity in 1956, how did they get it?
My answer is that they have acquired an identity, in scholarly circles at least, by having a history. It is generally accepted that groups of people consider themselves, or are considered by others to be, ethnic groups, if they see themselves or are seen as having common descent, and as acquiring certain common characteristics (ethnic markers) by virtue of common inheritance (Keyes 1976; Nagata 1981). Since the category Yi (along with its predecessor category, Lolo) was created not by the Yi people themselves but by the Chinese who administered, fought, and interacted with them, and by the Chinese and Western scholars who studied them, it is in the minds of the outsiders that all the Yi have always been assumed to be descended from a common ancestor. And in order to demonstrate this descent, the outsiders, Chinese and Western, have found it necessary to create the History of the Yi. Those who would create or defend the category (including scholars as well as the Nationalities Commission) must first write the History.
But of course the category Yi or Lolo has meant different things to participants in different civilizing projects. To the majority of Western authors, most of whom wrote about the Yi in the early decades of this century, the category was first and foremost a racial one, its history to be found in the migrations of peoples. To most Chinese scholars writing before the establishment of the People's Republic, when the Confucian civilizing project was still implicit even in the modernization and development schemes of various Republican and warlord governments, the category acquired meaning in the context of traditional historiography, of matching names of non-Chinese peoples found in standard histories to the non-Chinese existing in our own time. And to the scholars of the Chinese People's Republic, the category was important in the present as a Nationality that met Stalin's four criteria, and in the past as a group that was passing through the five universal stages of human history as defined by official Marxism. For them, the history was one of the development of productive forces and relations, and the Yi were thus defined and scaled as the first step of the Communist project. In each case, the History of the Yi served to show that the Yi category was a valid one, that it consisted of people with common descent, but because of the different ways people thought common descent ought to be manifested, the content of the history was different in each case. Let us examine the History of the History of the Yi as it was practiced by Westerners, by Chinese before the Revolution, and by Chinese after the revolution.

WESTERN WRITERS AND THE HISTORY OF THE YI
Westerners were not the first to mention Yi history. There are the written accounts of the Yi themselves, in the forms of genealogies and legendary history (see Zheng 1947, Ma 1985, for example), and there are mentions of people thought to be the ancestors of the Yi in much of the traditional Chinese historiography, perhaps even from the pre-imperial Shu Jing (Book of History) on; certainly from the Shi Ji (Historical Records) of Sima Qian (2nd and 1st centuries A.D) and the two Han histories. In addition, there are several geographical works dating from various historical periods, such as the Huayang Guo Zhi and the Manshu (Book of Barbarians), that treat non-Chinese peoples in a somewhat systematic way. But none of these books concern themselves with the project of Yi history: defining a people by delineating its origin and development. The first to address themselves specifically to the project of Yi history were foreign scholars and travelers.
Foreigners had been coming into contact with Yi or Yi-type peoples for centuries; they may even be mentioned in Marco Polo's book. But it was toward the end of the 19th century that two events stimulated foreign interest in a project of Yi history. First and more generally, foreign travelers began crisscrossing China, and some of them took particular interest in the borderlands between China and Tibet, roughly the area where the Yi live. The English diplomat and explorer Edward Colbourne Baber traveled along the Anning and Jinsha River Valleys in 1877, ringing but not crossing Liangshan (Baber 1882), and several expeditions followed. These included, most noteworthily, those of the French physician A.-F. Legendre, who made and wrote about several trips around Sichuan and Yunnan in the first decade of this century (Legendre 1905, 1913).
The second development involved the French colonization of Indochina; they acquired Annam and Laos as protectorates in 1874 and outright in 1885 and became interested in expanding their presence in China through the "back door" of Yunnan. The railroad from Kunming to Hanoi was begun in 1895 and completed a few years later, and more importantly, French Catholic Missionaries, mostly from the Missions Etrangers de Paris, took Yunnan as their province, evangelizing both Chinese and minority peoples (see Swain, this volume). Certain of these missionaries compiled and/or published more-or-less full and detailed accounts of Yi customs, culture, language, and religion. Of the published accounts, by far the most complete and most scientific is that of Lietard (1911), concerning the Lolopo, a Central-Yi speaking group living between Dali and the Jinsha River; the account by Vial (1898) of the Sani (or Nyi or Gni) of Lunan and Luliang is also fairly complete. From Liangshan we have no formal missionary accounts, but several authors, including Lietard and Legendre (1906) derive much information from a certain Father Martin, resident for at least 18 years in Fulin (Hanyuan), outside Liangshan proper but definitely in Nuosu territory, who, according to Lietard, "semble avoir horreur de la publicité" (1913:18), and thus not the author of any published works. Finally, an English missionary, Samuel R. Clarke, lived in Guizhou for several decades, and published an account of the local Yi, among other peoples (Clarke 1911).
Western accounts of the history of the Yi continued into the 1930s and 1940s; notable from that period are the synthetic treatment by Feng and Shryock (1938), and the chapter on the Lolo in the massive Rassendynamik von Ostasien written by von Eickstedt in 1944 (von Eickstedt 1944:162-178). After the 1949 communist takeover of China, Western scholarship on the Yi seems to have disappeared, with the exception of Bradley's linguistic studies and a brief summary account by Dessaint, until the 1980s, when minorities in China have once again become reachable by Western researchers. Some of these, at least, have accepted without much question the now orthodox Chinese History of the Yi (see, for example Heberer 1984:209-210). Things may change in the next few years; Harrell (1989, 1990) for one seems intent on calling the whole enterprise into question with critical metahistorical speculations. But at present, the Western chapter of the History of the History of the Yi extends from the 1880s to the 1940s.
What then of the content of the History of the Yi as written by Westerners? The earliest missionary accounts, by those who in fact knew Yi peoples best, tend to rely on native legends and cultural traits in tracing the origins of the people. Vial, for example, deduces from linguistic affinities and Sani legends that the Yi probably came from "the region between Tibet and Burma (1898:2)", led by chiefs of families or clans, of which there must have been two, because of the nature of division of Eastern Yunnan and Northwestern Guizhou Yi society into Black, or landlord, and White, or subject, dialect and culture groups (tribus). There in Eastern Yunnan, he says, the Lolos had a kingdom, which endured until it was conquered by the First Emperor of the Qin. He says little of the later history. Lietard's account follows a similar course: he sees the Lolos as originating in the northwest of Yunnan, as attested to by the similarity in language, culture and customs with the Tibeto-Burman peoples (1913:58), and has having formed the population of the state of Dian in the second and first centuries B.C., a state whose king was, in all probability, "un Chinois naturalisé Lo-lo (1913:55)." The Cuan lords of Yunnan in the sixth and seventh centuries were likely also Lolo kings of Chinese origin. When the Nanzhao polity was formed, most of its subjects were Yi, but the king was from the Minjia (today called Bai).
In the accounts of these missionaries, very little about race or physical characteristics is used to explain Yi history. But in the accounts of scientists, natural and social, we find an explicitly racialist paradigm, one that derives more from the scientific impulse of generalization and systematization than from the intimate knowledge and the desire to describe accurately that seem to have motivated the missionaries. For these scientists, beginning with Legendre and in rudimentary form even with Baber, the History of the Yi is the history of a race, built on a series of unexamined assumptions, probably part of the "habitus" of Western culture, about the nature of ethnic or social groups. The first of these assumptions is that there are, inherent and inherited in groups of people, certain traits of personality, culture, even economy that persist over generations. The second is that these inherent and inherited cultural characteristics correlate more or less exactly with inherited physical characteristics. The third assumption is that there is such a thing as a pure race, a physical-cultural combination that originated who knows when, but far enough back in the past that we don't need to worry about it, and has persisted through the ages, at least in those places where the race has remained genetically pure. Corollary to this third unexamined assumption is a fourth, that mixed races are different from pure races; that when members of these pure races intermarry, not only physical features but cultural ones become blurred, mixed, undistinguishable. Hence there are no cultural characteristics that "belong" to these mixed races; in order to trace these people's origins, one must sort out the original, pure elements that went into the mixture.
Examining the accounts of Legendre, Mueller, von Eickstedt, and Feng and Shryock will give us a feel for how these racialist assumptions shaped the history of the Yi as seen by foreign scientists. Legendre was the first to do "physical anthropological" studies on the Yi; when he traveled to Hanyuan he took cephalic indices of some people introduced to him by the modest Father Martin. From these and from measurements in Yunnan, he determined that the current Lolo population in fact came from three strains. There was an autochthonous strain, represented by Tai and Mon-khmer peoples, that was short, dark, brachycephalic, "une race Polynesienne tres bas dans l'échelle humaine (1905:399)": the Negrito. Mixed with this was the Chinese, the Mongoloid type, who had entered Sichuan and Yunnan early in the historical era. And then later, there came the Lolos: the original Loloish strain, which was related to Caucasoid peoples of central Asia, was "un type supérieur (1905:477)" tall, high-nosed, dolicocephalic, with broad shoulders, erect carriage, and perfect proportions.
In Legendre's account, the Lolos of his day in Sichuan and Yunnan were in fact very different because they represented very different mixtures of these races. In Liangshan, where the Lolos remained unconquered and unmixed with the other races, they retained not only their racial purity but their original cultural characteristics; they were warriors, bold and sometimes treacherous, ready to counter their invaders to the point where the Chinese, to whom Legendre concedes superior intelligence, had to overcome the Lolos slowly and peacefully (ibid:478-79). In Yunnan, however, the people the Chinese call Lolos are hardly worthy of the same designation (see above, p. 4). These people were mostly Mongoloid with some apparent mixture of Negrito or autochthonous blood, with none of the fierceness or love of vendetta found among the Lolos in Liangshan. They were peaceable, resigned, capable of submitting to all yokes (1913:391). From this Legendre derives the conclusion that there is in fact no unity to the Lolo, that the latter is in fact a category imposed by the Chinese, one with no justification in the scientific facts (1913:392). As a Westerner, he doubts the category; as a man of the early 20th century, he does so on the grounds of racist assumptions about the relationship between what he calls "physical and moral" characteristics.
Mueller, in a synthetic article written around the same time, is willing to give more credence to Chinese sources, at least of the traditional historical kind, and is less extreme than Legendre in rejecting the connection between northern and southern Yi. Even he admits that physically and culturally, the Lolos of Liangshan and those of Yunnan have little in common, though linguistically, he says, there is more justification for placing them all in a single category (1939:48). Since he is unwilling to give up the category Lolo altogether, he reconciles the physical and cultural differences by attributing to the Yi of Yunnan great mixtures with other peoples, both in (Thai) blood and in culture, and keeping those of Liangshan as a relatively pure type (ibid:50-51). These, he says, must have come from the north or northwest; by the Western Han the historical sources point to people in Yunnan who bind up the hair on their forehead into a kind of horn; from the similarity to today's Yi in Liangshan, these must have been Yi, as were the later commoners of the "Thai-Reich" of Nanzhao from the 9th to the 11th centuries (ibid: 48-49). Mueller thus manages to retain both the unity of the Yi category and the assumption that race and culture are united, even if this is hard to see when mixing of one sort or the other has occurred.
Von Eickstedt, whose data for his comprehensive work include visits to Liangshan and other areas of China, presents a similar picture, but with the further wrinkle that the difference between Black and White (aristocrat and commoner) castes in Liangshan is also attributable to racial differences. His historical scheme goes like this: The Lolo and the Miao-Yao peoples are originally part of the same group, Europid (sic.) peoples of Central Asia who were driven from their homelands by droughts in the first millennium B.C., and who then settled in the high plateaux of northeastern Tibet. From there, however, they were compelled by population pressure to move again into the still largely empty lands to the south (1944:174-5). This meant, of course, that they first settled Liangshan, and it is in Liangshan that the aristocratic Black Lolos preserve the true nature of that race:
Sie bilden zweifellos den eigentlichen Kern des Lolotums,
gelten als kühn, räuberisch, gastfrei und offen, kriegs-
gewöhnt und kriegsbereit, und sind voller Hass und Miss-
trauen gegen die Chinesen, während sie gegen Europäer
keinerlei Abneigung zeigen [1944:168].

Es herrschen blutige Fehdern, Mord, Diebstahl und schranken-
lose Habgier, aber auch Liebe zum edlen Zuchttier und Achtung
vor die Frau [ibid.].

Among the White Lolos, who are ordinarily the descendants of captured Han slaves, as well as among the Lolos of Yunnan, however, little of that lawless martial tradition remains; they are basically a Mongoloid type, or at best a mixture of these with true Lolos, and retain little or nothing of the Black Lolo system of social classes or their former pastoral economy.
Such attention to racial factors may not seem surprising in a German writing during the Nazi period, or even in a Frenchman or German writing before World War I. But Feng and Shryock's account of 1938, written by an American and a Chinese in the time of Boas and Herskovits, seems more surprising: although it, like Mueller's account, relies on culture as well as race, it gives the impression of an underlying assumption that the two naturally go together. In the first place, they use data from Legendre and Ting to argue that Black and White Lolos were racially different: the blacks were a conquering group of a single racial stock, while the Whites were subjugated peoples from a mixture of different stocks. Due to pressure from Chinese, blacks in Yunnan in recent centuries have been decimated or driven northward, and remain only in a few areas (presumably, but not explicitly, the northeast of the province) (Feng and Shyrock 1938:107-8). Exactly how this happened is hard to know, since it is difficult or impossible to match the names of former peoples found in Chinese historical sources with the ethnic groups that must have existed at the time: "Because of the presence of other tribes, it is often hard to tell whether a mentioned tribe was Lolo or not (ibid.:108)," in other words, there were Lolos all through history, but we can't tell whether particular sources were talking about them or somebody else. The Cuan peoples of the Northern and Southern Dynasties and early Tang period were, according to Feng and Shryock, of mixed ethnic stock: while the Eastern Cuan were definitely Lolo, the Western Cuan were probably composed of several different ethnic groups. The only conclusion the authors can draw was that Cuan was a political, rather than an ethnic category; the conclusion I draw from that is that for the authors, an ethnic group is a racial and/or cultural, but not a political group (ibid.:117). At the very moment Feng and Shryock were working out this analysis (and not far from Yi territory), Edmund Leach was studying the Kachin.
If the later History of the Yi, as presented by Feng and Shryock, is clear in its outlines if not in its details, their origin and early history are still matters for speculation. They dismiss theories based on language, because of insufficient knowledge, and concentrate on the two characteristics they think most salient: race and culture. They discuss at length racial theories propounded by Legendre, Ding, and others, all of which connect the Lolo to Iranian or other central Asian, Caucasoid types. In the end, though, they find these theories inconclusive (ibid.:126). They then go on to consider cultural hypotheses, noting that there are affinities to Mongols and other northeastern Asian peoples in the use of felt, the lack of pottery, and the possible division into Black nobles and White commoners. But contrary evidence comes from the fact that the Lolo, while they keep herds, do not use milk; thus they must not have originally been a pastoral people, and the origins in northern Asia thus seem questionable (ibid.:126-127). In the end, Feng and Shryock, unlike many analysts, do not put forward a pet theory of their own. But they do hold to the assumptions about the importance of race, and about the enduring nature of a group, with its own racial and cultural characteristics, that endure across the ages.

CHINESE HISTORY OF THE YI BEFORE 1949
As mentioned above, traditional Chinese scholars did not engage in the History of the Yi project in any systematic way. When Chinese scholars did begin to take up the project, many of them did so as part of an international ethnological or ethnohistorical community, so that their scholarship did not evolve entirely independently of their Western colleagues'. But there are definite differences of emphasis which reflect the fact that whatever methodological influence the Chinese scholars may have received from the West, they were still engaging in a project that derived from an assumption that was purely Chinese: the assumption that Lolo or Yi was a real category. For the Chinese scholars, such an assumption is not explicit; rather it is part of their conception of the world from the beginning, something that is not examined and does not have to be. Thus we do not find, as among the Westerners, any speculation that the Lolos might not be a single people after all; rather the assumption is that the category is real and the task is to find out how the people got where they are today.
In addition to this paramount assumption of a real category, there are other characteristics that distinguish the Chinese approach from the Western. Closely connected to the first assumption is the premise that the categories with which we deal are very old; certainly by the time historical records appear, the Lolos were already in existence. Another important assumption is that names hold the most important key to unlocking the secret of the Lolos' origins and history. If one can correlate the names of peoples living at different epochs, as described in traditional historiographic works, with the peoples known to exist in our time, one has solved the puzzle: in a sense one has traced the Yi through their history, which is a branch of History with a capital H, which stands for Han. A final important assumption, not nearly so prominent in Western works, is that cultural differences among various Yi groups are due to differential exposure to a higher, more advanced Han culture. In the service of this eminently Chinese project, scholars of the Republican period used many of the same tools used by their Western counterparts: they paid attention to race, to character traits, and to cultural practices. But the fundamental assumption remains that of the real category, which can be traced through history. Let us examine briefly some of the tracery.
One of the earliest accounts was that of Lu Simian, in his Zhongguo Minzu Shi, published in 1933 (Lu 1933). This work is divided into chapters, each chapter covering a category of people named in the early histories, whose descendants can presumably be traced until today. For example, describing the history of the peoples in the West and Southwest of China, there are chapters on the Miao and the Qiang. The Yi are treated in the chapter on the Pu, which begins "The Pu are also called the Pu, and in addition called the Pu, and today are known as Lolo. They are also one of the large races of the southwest (237)." The Pu, a name that goes as far back as the Shu Jing and the Zuo Zhuan, were a group found in Yunnan by the time of the Han, but who had earlier been spread out over Henan, Hubei, Hunan, Guizhou, Yunnan, and Sichuan. They were forced out of their eastern territory, but later formed the Yelang kingdom in Yunnan. After that, they appear again as the rulers of the Eastern and Western Cuan, whose black man and white man inhabitants, respectively, correspond to the Black Lolos and White Lolos of today (241).
This account is in some ways the most purely Chinese of the sources I examine here: it begins with a name from ancient historical sources, assumes that the group named existed even earlier, and traces the group through to the present. It says little about cultural traits; its main interest is in the history of the assumed category.
Another explanation, by Jiang Yingliang, also relies on historical documents, but comes to completely different conclusions. According to Jiang, who visited Liangshan in 1941, but did not publish his results until seven years later, the Yi ( ) of Liangshan have an even longer history. The earliest mention of these people is in historical accounts of the Zhou Dynasty, which even before it conquered the Shang already was having trouble with peoples, variously referred to as rong and man, who belonged to the Qiang groups. From the earliest Zhou rulers to You Wang, the last king of the Western Zhou, there were frequent attacks from the Qiang, eventually forcing the Zhou to move their capital eastward from Changan to Luoyang. During the Spring and Autumn period (771-481), the Chinese kingdoms of Qin, Han, and Zhao all attacked and defeated various groups of Qiang, who were forced to move westward. After this, they dispersed into various areas in the Sichuan-Gansu border region. The Baima people, for example, as described in the Later Han History, showed such traits as felt manufacture and legal protection of women, traits they hold in common with the Yi of Liangshan today. From here the Qiang/Yi people moved further south, until by the end of the Northern and Southern Dynasties they occupied most of the territory east of the Tibetan Massif, south of the Chengdu Plain, and north of modern Kunming.
The continuity of these peoples, according to Jiang, was broken up in the Sui and Tang Dyansties, when they were dispersed, some of them becoming the various peoples of the Tibeto-Burman family found today, some assimilating to Han language and culture. Only those remaining in Liangshan, because of their lack of contact with the outside and their independence from government rule, have preserved their primitive social conditions (Jiang 1948:20-22).
Jiang's account, unlike Lu's, is interested in cultural traits and linguistic relatedness, but it basically follows the same, Chinese strategy of assuming the cultural unit in question is very old (in this case, at least three thousand years), equating its identity with that of non-Han peoples described in traditional Chinese historiography, and tracing cultural differences at least partly to differential contact with advanced Han civilization.
Another account from about the same time, by Zeng Chaolun (Zeng 1945), follows this basic pattern but includes more factors in its explanation and gives us an even clearer picture of the distinctive Chinese approach to the History of the Yi. Zeng's original named base are the Nan Yi (Southern Yi ), as mentioned by the poet Sima Xiangru of the Han dynasty. Unlike the Miao, originally a lowland people forced out by the Chinese, the Yi have always occupied the entire Liangshan region (94). Where they came from is not certain, but racial and linguistic factors seem to point to affinity with the Tibetans, since the Yi language, like those of Tibet and Japan, has an "inverted word order."
The differences between various groups of Yi are also explicable by history, according to Zeng. The majority of the Yi in Yunnan, for example, are so much Sinified that their original customs and habits have almost completely disappeared (95). [In Kunming, they haven't used them for years.] It is only in Liangshan, where the Yi have lived undisturbed and unacculturated for thousands of years, that one can find the pure, unadulterated Yi culture. There the Yi are divided into tribes, by which Zeng seems to mean something like clans, each of which has its own character traits: some friendly, some hostile, some strong and some weak, some treacherous, ruthless, or fierce, some more gentle (96-97). It is also in Liangshan where the "social class" division into Black aristocrats and White commoners is found; although we cannot prove that this existed among the other branches of the Yi, Zeng thinks that it did, and was abandoned under Chinese influence (95). The system, as it is found in Liangshan, seems to be a relic of "the outmoded feudal system," and is something that ought to be abolished. At the same time, Zeng sees the slave system as providing more security for the Wazi (captured Han slaves) than the tenancy system provides for tenants in the Han areas (112).
Zeng's account, like Lu's and Jiang's, traces the Yi from a known historical people, and although he does not correlate the Yi at every step of their history with some people mentioned in the books (his main purpose, after all, is ethnographic rather than historical), we can still see the basic assumptions of an ancient people with a name and an essence, and of the influence of "superior" Han culture as the explanation for any internal contradictions or differences among Yi groups. The criticism of the "outmoded Feudal system" is also characteristic of Chinese accounts; scholarship on Chinese minorities in this period is often rather applied in nature; it assumes not only Han superiority but superiority of modern institutions over traditional ones.
Such a vision of Yi history as a branch of History seems to have been all pervasive in pre-1949 Chinese ethnology and other scholarship. For example, Wei Huilin (1947), in an article basically concerned with reform and community development in Yi areas, prefaces his remarks by a recitation of the historical names of the Yi people, starting with the Lu and Luo of the Spring and Autumn period, and running through the Xi, Kunming, and Qiongdu of the Han, the Nanman attacked by Zhuge Liang in the Three Kingdoms, the two Cuan of the Southern Dynasties, the Nanzhao and Wuman of the Tang and Song, the Wuyi and Luolouman of the Yuan and Ming, and the Luoluo of the Qing. Many other examples, each deriving the modern Yi from a slightly different succession of ancient groups, are mentioned in Hsieh 1982. And while the particulars of these hypotheses differ, as does the evidence they use, they all hold to the position of an ethnic group that has existed over the long haul from the times of earliest historiography to the present.

THE REVOLUTION IN THE HISTORY OF THE YI AFTER 1949
The History of the History of the Yi since 1949 is much longer than either of the earlier two episodes. There are two reasons for this. First, there has been much more official and scholarly attention paid to minorities generally in the People's Republic than was paid to them earlier; the national policies of jural equality as well as integration into a "United, Multi-ethnic State" (Tongyi Duominzu Guojia) certainly have a lot to do with this. The result is the second reason, that simply a lot more is known about the Yi and their history than was ever known before. In particular, more Yi-language manuscripts and stone inscriptions have been found, and scholars have begun to take some of these things seriously, things that were previously dismissed as unreliable legends. Also, there has been much more archaeology, and as with the trends in Chinese history, archaeology and legends have been correlated into a much more detailed and coherent picture of the past. If a neophyte (even more neo than I) asked me for a recommendation on what to read about Yi history, I would certainly recommend a P.R.C. account over either an older Chinese one or one from a Western pen.
But P.R.C. history is, of course, no more "objective" than earlier history from China or history written by Westerners. It too has its purposes and its assumptions. One of the purposes of this modern History of the Yi is carried over from its Republican predecessor: to document the integrity of the category and show how it has persisted through time in spite of the changes that the Yi have gone through. But added to this is another element, which is at the same time a purpose and an assumption. This is the construction of the History of the Yi, like the history of everything else, according to the five stages of history laid out by Soviet historiography, the stages corresponding to the primitive, slave, feudal, capitalist, and socialist modes of production. Added to this is the breakdown of the primitive stage into matrilineal and patrilineal phases, following the plan laid out by Lewis Henry Morgan (1877) and distilled by Frederick Engels (1883). Adopting the five-stage model of history and the Morgan model of the primitive stage makes possible a much more complete and detailed History of the Yi. This is primarily because the five stages of history are defined in terms of their forces and relations of production, and the sub-stages of the primitive stage are defined in terms of the particular social structures. So source materials that were formerly ambiguous or difficult to interpret can be made clear; details can be filled in. Most importantly, cultural traits, only hinted and guessed at in previous kinds of history, can now be supplied by inference from what we know must have existed when the Yi were at a particular stage. The result is a History of the Yi that for the first time contains fairly detailed reconstructions of the culture and social structure of the Yi in premodern times.
Other assumptions in the modern History of the Yi carry over from the Chinese model of the prerevolutionary period. The existence of the category, basically unchanged from very early times, is of course still assumed, and the formulation of the history acts to justify the boundaries of the category, just as the boundaries of the category define the scope of the history. This is reminiscent of the dialog between history and legitimacy traced by Sangren (1988) in his analysis of temples in Taiwan: as with the Yi as an ethnic category, the present defines the scope of the history, and the content of the history legitimates the claims made in the present.
Another carry-over assumption is that internal cultural, linguistic, or even mode-of-production differences between different branches of the category are to be explained not by any kind of original differences, but by the effects of different kinds of environmental and acculturational influences. Those branches of the Yi that had more contact with the Han culture (now defined not as inherently superior, but rather as farther along in the five-stage developmental scale) are likely themselves to be more developed. Those whose natural environments were more favorable are likely to have developed further on their own. And any branches that have any characteristics that are more like Han society than the corresponding characteristics of other branches are assumed to have undergone acculturation toward Han culture. In other words, what is Yi is assumed to be whatever is most different from Han; other variations are assumed to be the result of Han influence.
Finally, race is out; language and culture are in as the primordial characteristics that originally define a group and constitute its basic essence.
The interaction between the definition of the category in the present and the formulation of the history is a constant and seamless one, but for the purposes of this analysis it is easier to look at it from one side at a time. I will thus first describe briefly how the category Yi was formulated in the process of minzu shibie, or ethnic identification, using history as an important guide to identification. I will then describe at somewhat more length how, given the category that was now official and buttressed by historical evidence, the more complete History of the Yi has been written by different authors.
Long shrouded in silence, the process by which the ethnic identification process was actually done has now become the subject of a few articles (Lin Yaohua 1987; Jiang Yongxing 1985). The announced method was to ask first for social groups that thought themselves to be minzu to make application to the authorities; in Yunnan 260 groups submitted their names. After this, teams of specialists in culture, led by Lin Yaohua, and in language, led by Fu Maoji, would investigate the validity of the claims according to the standard of Stalin's four criteria (Lin 1987:1). But in fact the investigators ended up using the criteria very flexibly. In particular, they found that the third criterion, a common economic base, was not a characteristic of very many minzu in China. We can see this with the Yi; according to official histories, Liangshan was still in the slave stage of society in 1950, while Yi areas in Yunnan and Guizhou had passed into feudalism for hundreds of years. But, argued the identification teams, the criteria delineated by Stalin were in fact based on what happens in capitalist society; "the common economy of areas where many minzu lived together linked together different minzu, but it did not eliminate their respective ethnic characteristics; the result was that the existence of a common economy was not obvious in any minzu area (Lin 1987:2)." What this says, it seems to me, is that the identification teams already had their categories in mind; if Stalin's criteria went against the pre-existing categories, it was the pre-existing categories that took precedence.
There were, of course, problematic cases, small groups that claimed independent minzu status, and whose claims had to be investigated. Several of these were people who spoke Yi languages; Lin gives the example of two groups that reported themselves as Tujia (no relation to the officially recognized Tujia minzu of Guizhou, Hunan, and Hubei) and as Menghua. About 170,000 people reported that they were Tujia; another 40,000 claimed to be Menghua. Investigation teams first determined that the Menghua and Tujia were the same; the Menghua were Tujia who had migrated south from Menghua county, and 76% of a sample of 1000 vocabulary items were the same. Then, ethnographic investigations disclosed that this Tujia group (including the previous Menghua) had "retained many common Yi [cultural] features, such as clan-elder systems, surname exogamy, levirate, remnants of cremation, ancestral spirit platforms, polytheism, and magical arts. In addition, they could intermarry with Yi. For this reason, the Tujia and Menghua were determined to be a branch (zhixi) of the Yi (Lin 1987:3-4).
This description of the actual process reveals the presence of several of the abovementioned assumptions about ethnic groups. The Chinese investigators basically already knew what Yi meant; not fitting Stalin's criteria was no serious impediment to declaring them a minzu. In addition, when there were problematical cases, it was language and culture traits that were examined in order to determine where these people fit. Common vocabulary items indicate a genetic relationship between languages, that is, a common linguistic history. And common cultural traits are always spoken of as "retained" (baoliu). That is, there is the presumption that Yi all had common traits in a primordial past, and that any group that has any of these traits has retained them in the face of acculturative pressure (rather than borrowing them, for example), as well as the converse proposition that progress comes from contact with the outside rather than from the people's own internal development.
All along, there has been a stated principle that the process of ethnic identification should not be carried out by fiat, but that the wishes of the people themselves ought to be taken into account. But as admitted in Jiang Yongxing's 1985 retrospective on thirty years of identification work, this principle was not always followed. Jiang states, in fact, that one of the reasons why there are still so many unsolved ethnic identification problems in the Southwest is that the identification work has consistently overemphasized historical kinship (Lishi shang de xueyuan lianxi) and underemphasized the wishes of the ethnic peoples (minzu yiyuan) (Jiang Yongxing 1985:309-315). This critique, I think, strikes at the heart of the matter; the wishes of the ethnic peoples cannot always be made to coincide with the pre-existent Chinese category, while history can usually be formulated in such a way that the category remains intact (see McKhann and Diamond, this volume). This seems to be the reason why most remaining problems of ethnic identification involve groups that want to break away from the larger (Han or minority) minzu in which they have been classified, rather than independent groups who want to be amalgamated with a larger minzu (Jiang Yongxing 1985:304).
The process of ethnic identification was thus little threat to the category of Yi (which by common consent of representatives of many branches, was the name adopted for the whole group); history had been used in the formulation of the category, as had linguistic and cultural commonalities, both inferred as demonstrating a common history. This did not mean, however, that the definitive History of the Yi had been written. Of course there was and still is a lot more to find out, in the sense of strictly empirical facts. What the completion of ethnic identification did mean for history was that any serious historical treatment would have to contribute to the demonstration that History was something common to the whole category. In addition, history would have to show how the category or branches thereof passed through the five universal stages.
All this has resulted in a richer, more complete, more detailed history, and one in which differences among branches of the category must be explained in terms of differential experience rather than any kind of primordial or essential separateness. This does not mean that there is a single, official, orthodox history, without any disputes. There is, on the contrary, still no agreement about where the Yi originated. What is agreed on is that wherever they originated, they all originated there and have diverged afterwards.
To illustrate how the History of the Yi looks under modern conditions, that is, assumed unity of the category, progress through the five stages, and explanation of differences in terms of differential environmental conditions and varying degrees of Han influence, I will recount the story as told by one historian in some detail. Since I have only read one full-scale history, Ma Changshou's Yizu Gudai Shi (Ancient History of the Yi), I will use it as the example, keeping in mind that some of Ma's positions, particularly regarding the origin of the Yi, are still quite controversial.
The first topic treated in Ma's history is the nature of primitive society among the Yi. The only sources available for investigating this question are Yi traditions, which are available in two forms: books of stories of the creation of the world, and recitations of genealogies at festivals, war councils, etc. From these sources, we can tell that the first, or matrilineal, stage of Yi primitive society was longer than the patrilineal stage; this is because many Yi genealogies recite twelve "dynasties" at the beginning of the world; of these, only the last two show the characteristic Yi pattern of "father-son linked names" (fu zi lian ming), in which a man's second name becomes his son's first name. That the Yi were matrilineal in late prehistoric times is also confirmed by the legend of Asu Awo (a Han transcription of a Yi name) looking for his father. Briefly, this is the tale of a man who was born in the old days, when people, "after birth knew their mother but not their father." Asu Awo went looking for his father, but was not successful until a spirit told him to perform ancestor worship. After this, paternal grandfathers and grandsons, as well as fathers and sons, knew each other. This is a story of the transition from matrilineal to patrilineal descent in the later stages of primitive society (Ma 1985:1-2).
The next question is where and when this happened. The key figure in all this is a man called Zhongmuyu in Han, or Zzemuvyvy in Yi, who is acknowledged by all Yi and some Hani to be their common ancestor. If we figure the numbers of generations in genealogies, we come up with about 80, which at 25 years per generation comes to the former Han (10). With regard to the place, Zzemuvyvy is referred to in Yi manuscripts from as far afield as Liangshan and parts of Yunnan as living in places that can tentatively be located near Kunming; in Guizhou manuscripts, his abode can be located in Northeastern Yunnan, near present-day Zhaotong. Therefore we can posit an early migration of the Yi from central Yunnan toward the northeast (5-7).
After Zzemuvyvy in many Yi manuscripts come the six ancestors (liu zu). Since their names do not link with Zzemuvyvy, the chances are that they were more than one generation after him; anyway legend has it that the eldest pair were the founders of clans who settled in the south (near Kunming again), the next pair founded clans that settled in the north (near Zhaotong), and the youngest two established clans that settled in the East (in northwestern Guizhou) (9). At the time of the formation of the clans, the Yi were still not divided into classes; that is, they were still in the stage of primitive society. But especially the eldest two clans, who settled in the south, frequently fought with other tribes; the result was that they began to enslave war captives, which we know from Engels was the origin of the slave system, the second stage of history and the first to involve class divisions (12-13).
By the Eastern Han period, then, the Yi had populated most of eastern and northeastern Yunnan, as well as northwestern Guizhou. It remained for them to enter Liangshan. Those who did so were members of one of the middle pair of clans, called Heng (Yi, He) in Guizhou and called Guhou in Liangshan. These clans, it will be recalled, were originally settled in Northeastern Yunnan; some of them remained there and became the historical Pu. Others either stayed in the area, moved to Yongning in the extreme south of Sichuan, or combined with the other of this pair of clans to form what would become most of the aristocratic Black Yi clans of Liangshan today (14-15). If it sounds complicated, it is. The point is, it all fits together: Yi traditions, Chinese history, the five stages and, most importantly, all the branches of the Yi come into one coherent account.
When the clans moved from the east into Liangshan, they subjugated the natives; this probably happened between the end of the Han and the first few decades of the Tang. The subjugated natives, referred to as Puren, Puren, Tulao, and other names that may correspond to modern Xifan and Naxi, as well as another group who are now extinct, all became the commoner subjects of the Black Yi clans; they are thus ancestral to the White Yi of present-day Liangshan society (19-22). Not only the Yi category, but the Naxi and Xifan, can be traced through history in this way.
The next question Ma takes up is the formation of the slave mode of production. This probably happened first in Yunnan: we can see from the Dian bronzes of the early Han period that there was a slave polity existing in central Yunnan at that time. Many of the scenes show masters or mistresses and slaves, the latter sewing, spinning, working in the fields, or in one very striking casting, about ready to be the victim of a human sacrifice. That this was slave labor is beyond doubt, since lots of people are working together, and we know that labor is more individualized under the later feudal system. These slaves all appear from their costumes to be of different minzu; from the specific nature of their clothes or hairstyles at least four of the eight identifiable minzu are probably Yi. The slave system was thus established in all the Yi areas at least by the Three Kingdoms period(39-49).
Generally characterized, the period of the Eastern Han and Three Kingdoms was characterized by competition between the slave system of the Yi and the feudal system of the Han; from the end of the Three Kingdoms to the beginning of the Tang, a period in which the Chinese polity was weak in the Southwest and had very little influence on Yi society, was the period of consolidation of the slave system, culminating in the establishment of the Nanzhao kingdom, itself a polity based on the slave mode of production (54-5).
This period of the development of slave society was the period of dominance of the Cuan. Cuan was originally the surname of one of the biggest slaveowning families in Yunnan; later it came to be applied to the people under this and other families' rule. Most historical sources speak of the Wuman (black barbarians) of Eastern Cuan and the Baiman (white barbarians) of Western Cuan; the question debated by historians is the connection between these and the later Black and White social strata in Yi society. According to Ma (this is still a disputed issue), there is no direct correspondence. The Baiman, living in the Western Cuan area (roughly between Kunming and Dali) were influenced early on by Han society, especially after the opening up of the trade route from Chengdu through the Anning River valley after Zhuge Liang's southern expedition. In the fertile plains of this area, they developed irrigated agriculture and a flourishing economy, and also adopted many features of Han culture and society. In the Eastern Cuan area, by contrast (in eastern and northeastern Yunnan), development proceeded more slowly because of a less favorable natural endowment and less contact with progressive (feudal) forces; here the slave system became consolidated. It is clear that the eastern Cuan were the Yi; the Western Cuan seem to have been of several sorts, of which at least one was the ancestors of the modern Bai people (69-73).
The Nanzhao kingdom, established in Yunnan in the 8th century, was the inevitable result of the development of the Yi slave polity to a certain degree. Nanzhao was a multi-ethnic polity; it is still under dispute who the rulers of the first, or Meng, dyansty were; the subsequent Qi, Yang, and Duan ruling lineages were definitely Bai. The social system, although possessed of some feudal influences already, was basically a slave polity, in which the ruler gave land to lords of different ranks, and the lords in turn, according to the Man Shu (Book of Barbarians)
After the harvest is over, the barbarian official (man
guan)
divides the grain according to the population of
the tenant households; what is left over is transported
to the official [quoted in Ma 1985:80].
Ma points out that this demonstrates that there was no fixed amount that was due to each household, but rather the slavelord gave the slaves just enough so they wouldn't starve. This was reminiscent of Liangshan before the Democratic Reforms, when the slaves were in an equally oppressed position.
During the Nanzhao period and the subsequent Dali kingdom (973-1283), there were a lot of migrations of peoples, many of them part of royal schemes to settle or defend various areas, others the result of people fleeing the control of the central authorities. In general, during the Dali kingdom there was more and more development of the productive forces, more and more influence of feudalism (91-94). But the transformation was not complete until the conquest by the Mongols in 1283, when Qubilai issued an order prohibiting the capture of slaves in northeast Yunnan, and when the Mongol prince known as Saiquchi in Chinese distributed land and tools to conquered Yi peoples, showing that the tenants, who had no land or tools previously, were still working under the slave mode of production.
The Yuan represented the real watershed for the Yi in Yunnan. The area was once again, for the first time since the later Han, a fully integrated province of China, and Han migration into the area, as well as economic development in terms of trade, mining, and intensified agriculture, brought about the complete triumph of the feudal system there (96-97). In Liangshan, on the other hand, the slave system as we know persisted until 1956, though it, too, was not devoid of feudal influences, especially in the Qing dynasty around the peripheries of the area. Here the slave system persisted under the rule of tusi (native officials) during the last three dynasties. In Liangshan, unlike the situation in Yunnan, land was poor and trade routes inconvenient, so the productive forces were unable to develop. The result was that the Ming and Qing governments retained big slaveowning clans as tusi, and did not try to alter the social system in the area (106-108).
But even in Liangshan, there was influence from Han society. During the Ming and Qing periods, there was a gradual encroachment of Han peasants and feudal land relations, especially in the peripheral areas around the Anning Valley. The result was that the big tusi became more and more agents of the central feudal regime, and less and less representatives of the slaveowning class. This in turn led to revolts in the late Ming, and the replacement of tusi in the core areas of Liangshan with independent rule by local slaveowners (110-112). In more peripheral areas, however, the feudalization of society continued, so that prominent Black Yi often had two kinds of dependents at the same time: Yi slaves and Han tenants (116-118).
As mentioned above, Ma Changshou's account, written in 1959 but not published until 1985, is not without its controversial points: in particular the ethnic identity of the Nanzhao rulers is still a matter for lively discussion (see Qi 1987), and the Yunnan origin of the Yi is disputed by those who think they came from the northwest. With regard to the latter issue, a recent article by Chen Tianjun (1985) demonstrates even more clearly than Ma Changshou's book the power of the five-stage and Morganian historical schemes. According to Chen, the origin of the Yi goes back further, to the San Miao of classical History, who were always fighting against the Xia dynasty. These San Miao most likely lived in the middle reaches of the Yangtze River, whence some of them were driven westward into modern Qinghai, where they amalgamated with local peoples to become the Qiang. The San Miao were already patrilineal when they fought the Xia, which means that the matrilineal-patrilineal transition recorded in the Yi genealogies and in the story of Asu Awo looking for his father probably happened between five and four thousand years ago in that area (Chen Tianjun 1987:109).
The Qiang, in turn, according to Chen, were driven out of their homeland in the Northwest by pressure from the Qin in the fourth and third centuries B.C. (see Jiang Yingliang's 1948 account, above); they fled south along the Tibetan marches to the later homeland of the Yi in Yunnan. Their patrilineal clan society lasted until this time, and the Yi show up again as the Kunming, a tribal people harassed and eventually enslaved by the Dian kingdom, which was nominated by the Pu. Under the influence of the Han dyansty, with its opening up of salt and mining industries in Yunnan, the Yi passed from the primitive to the slave stage (112-113).
In this account, we can see the slave-feudal transition occurring at three different times in the three different areas of Yi settlement. In Yunnan, it came earliest, at the time Nanzhao was taken over by Dali, while it persisted in Guizhou and northeast Yunnan until the Ming, and in Liangshan, in modified form at least, to the 1950s (114-117).
Both in Ma Changshou's account and in Chen Tianjun's, the five stages of history are a unifying force. They not only make sense of what would otherwise be ambiguous statements in old historical records; they also explain why, in modern times, people as diverse as the Lipuo, Sani, Menghua, and Nuosu can be shown to be nothing but varieties of a single category Yi, distinguished from each other by nothing but unequal rates of development through those stages ordained by the objective laws of historical development. We even find the Yi following the basic patterns of development within the feudal stage: a recent article by You Zhong shows how, in areas of heavy Han influence, the overthrow of Tusi in the late Ming and the replacement by officials appointed from the center represent the transition between the earlier substage of manorial feudalism and the later stage of the landlord economy, which was the status of most of Chinese society in the late centuries of the Imperial Era (You 1987:190-192).
DISCUSSION
The History of the Yi thus has its own History; in the hands of different kinds of writers the Yi have acquired very different kinds of Histories. I am not here to judge the varieties against one another. I think that in each case, the historians have known what they were going to write before they even did the research for their History. Westerners were interested in races; Chinese before 1949 in correlating the Yi History with the Capital H-History; and Chinese after 1949 in using objective and universal historical laws to lend legitimacy to an ethnic classification (see Litzinger, this volume, for a comparable case among the Yao). And in each case, they did a fairly good job: that is, they produced internally consistent Histories that explain what they set out to explain: in each case, who the Yi are and how they got to be that way.
Again, though, the outsider finds himself a bit disturbed by this. Is a relatively ideologically neutral History of the Yi possible? Could one write such a history with an agenda that was neither racialist nor nominalist nor orthodox Marxian? What would an observer who studied Yi society anew from an ethnographic perspective, visiting communities and interviewing local people in Liangshan, Guizhou, and many parts of Yunnan, and then reading the classical sources and the Yi traditional manuscripts, come up with? Would such an observer find a unified group, a Yi category that was internally consistent? A series of related peoples with different, though perhaps related, histories and only remotely similar societies? A native history and a history created by outsiders, or several histories created by outsiders? It is difficult to predict, since it is difficult if not impossible to write the History of the Yi over again from scratch. Since 1956 the category Yi has come to exist not only in the minds of Han scholars and administrators, where it has always been, but also in the administrative and budgetary plans of the Nationalities Commission and the provincial governments, and increasingly in the minds of the various kinds of Yi, who learn their own history and culture not only through native ceremonies and recitations of genealogies (which, as we see, can easily be shown to support the orthodox, unified interpretations) but also through the curriculum of the schools. They too have come to be Chinese, and as such their history has become part of Chinese history. And Chinese history, now as in the Imperial Era, belongs to and is defined by the ruling orthodoxy. Under this orthodoxy, though there may be disagreements about specific points, minzu have already been defined, and the Yi are a single minzu. Any future historian of the Yi must take this into account.