On the first page of his new book, the well-known British sociologist W.G. Runciman defines the classic and often ferocious debate he wants to deal with. It is, he writes, the debate between “those who affirm and those who deny that there is a fundamental difference in kind between the sciences of nature and the sciences of man.” This controversy, which “has continued without resolution for more than two hundred years,” he proposes to resolve. If the social sciences have not yet provided us with convincing theoretical knowledge or laws governing human behavior, is it because the matters they deal with are so complex as to prevent the discovery of such knowledge? Or is it because no one has found a suitable method for studying these materials? Or is it because the interests and values of social scientists have contaminated their work, with the result that instead of objective theory they have produced merely a series of tendentious, if occasionally brilliant, speculations?
At one extreme, among those concerned with this debate, are uncompromising positivists and behaviorists, and others, who claim that the social sciences should use methods of inquiry and patterns of explanation identical, for all practical purposes, with those in other sciences. Human beings, they acknowledge, are indeed more complex than stones or plants, but they are physical systems that are in principle predictable. Causal generalizations about them can be confirmed or disconfirmed by observations of their behavior in the same way that hypotheses about inanimate objects are. They hope that such causal generalizations will in turn be explained by more inclusive theories which will enable us to explain and predict such human conduct as the behavior of economic markets, kinship systems, crime and deviance, voting patterns, the causes of class conflicts and of war. It is true, they contend, that constructing and verifying these laws and theories is often limited by the frequent impossibility of conducting controlled experiments on human beings, and by the special problems that arise when self-conscious human beings study each other. But, they claim, these obstacles do not necessarily prohibit constructing a science of society.
At the other extreme stand the champions of “hermeneutic” and “interpretive” social theory, who claim that the methods of social science, if it is science, are not those of observing outward behavior, at least for much of its research. The explanations of social science differ in both form and content from the causal ones of natural science. The reason is that the business of social science is to study not behavior but actions that have meanings for those who perform them, and actions (like voting or engaging in political protests) cannot be grasped by causal generalizations but instead require a distinctive “method.” We will never understand why a man will respond in anger to being called an ass unless we understand the meaning of the insult when used in a particular “social context.” If we attempt to formulate a generalization that the utterance of the word “ass” invariably causes overt signs of anger (such as flaring nostrils) we would pass over the underlying reason for the anger; and in any case not everyone who is called an ass becomes angry. To grasp these “meanings” and their social contexts, they say, requires the exercise of a species of “understanding” akin to what we use when we try to make sense of a game or a document; it is not an inquiry into causes.
Runciman follows Max Weber and others who sought to formulate a philosophy of social science that would preserve the truth (and discard the falsity) in each of these positions. In particular, he seeks to establish both that a special form of “understanding” might indeed be pertinent to the practice of social science and that the methods of explanation and empirical confirmation of hypotheses are not very different from those in the natural sciences. Despite occasional dense patches of writing and somewhat excessive exhibitions of learning in which we are told about, among other things, the plight of untouchables in Tokugawa, Japan, iconological analysis of the paintings of Piero di Cosimo, and dining clubs in ancient Athens, the number and vividness of its illustrations and its many clever arguments about issues subsidiary to those treated by its main thesis make the book far more stimulating than a cursory inspection of its design (that of an old-fashioned treatise in the philosophy of science) would suggest.
Runciman’s central claim is that those who have debated the nature of the social sciences—in particular those who have asserted that whereas the natural sciences seek causal “explanations” of social facts, the social sciences seek to “understand” them—have failed sufficiently to distinguish between three kinds of “understanding,” what he calls reportage, explanation, and description. Once we do so, he says, we can accommodate both the contention of the positivists that the logic of explanation is the same in all the sciences and the claim of interpretative social theorists that social science must grasp the “meaning” of actions. Indeed, if we do so, he claims, the debate can “for all practical purposes” be regarded as closed.
As Runciman expresses his position, “there is no special problem of explanation in the human sciences, but only a special problem of description,” by which he means not an account of external observations but something like acquiring a grasp of the “inner life” or “subjective knowledge” of the persons studied by social theory. “Properly defined,” he continues, “explanation and description can be distinguished both from each other and from either the reportage of facts or the advocacy of values.” Concerning evaluation, Runciman endorses the familiar thesis that although often intimately connected in practice, fact and value are in principle separable, so that social science can be “value free”—not in the sense that it has no distinctive values of its own (such as validity) but that neutrality among aesthetic, moral, and political values can be maintained in principle within social science, even though, he argues, it is not always desirable or necessary to do so.
But what exactly are these three kinds of “understanding” and how do they arise in the practice of social science? Runciman invites us to consider a wave of strikes by industrial workers. ” ‘Strike,’ ” he writes, “is already a term specific to societies in which labor is formally free, and it presupposes that within such societies concerted action is possible among groups of workers who have contracted to sell their labor to an employer, whether private or public, on specified terms, and who now agree among themselves to withdraw that labor.”
To apply the term “strike” to a sequence of behavior therefore implies that the inquirer has “understood” in an elementary sense what is occurring when workers leave their jobs. This is an instance of “reportage” of facts. But the theorist might also wish to go on to explain why what has been reported (or the aspect of what is reported that concerns him) has occurred. This second kind of understanding invariably involves the use of whatever relevant hypotheses, laws, and theories are at his disposal for him to derive a causal generalization that shows why the event occurred as it did. Thus he might seek to explain what set off the strike by reference to such factors as declining real wages, poor factory conditions, or the presence of political agitators, and whatever generalizations he may have about these factors.
But in addition, Runciman continues, the sociologist often should provide understanding of a third sort. He should “convey the nature of the institutions and practices by which the lives of the people whom he has been studying are governed in the way in which those institutions and practices were perceived and experienced by them.” He should, in Runciman’s somewhat unusual use of the word, “describe” the strike: he should show not merely that it occurred and why, but also “what it was like for those who were so motivated as to go on strike under the reported conditions to come to decide to do so. What did they think they were (‘really’) doing?” To know, for example, that they thought they were deciding to stop working
is not enough to convey to the reader an adequate sense of their sense of the context within which their actions took place and of the nature of the institutions and practices of the society, region, industry, community and plant as experienced by them. What were their ideas of their relations with their workmates, their supervisors and their union officials? What was, to them, the nature of the sense of grievance or discontent to which they were giving expression?
Runciman claims that these three kinds of understanding—reportage, explanation, and description—are distinct from each other. And they are different as well from the fourth activity of “evaluating”—asking, say, whether the strikes were legitimate or justified. But he does not argue that the four activities are always easy to distinguish, or are performed in the sequence I have just mentioned.
Reporting, Runciman says, is the stage of inquiry at which “facts” are stated; it
goes no further than the point where noises and gestures can first be labelled as actions because of the meaning of “them” which makes them so; and “meaning” here carries no weightier presuppositions than those necessary to distinguish clapping the hands from applauding a performance (sincerely or not), or dragging a stick across the sand from drawing a picture (artistically or not).
To specify an action, he continues, is “to account for the bodily movements which, together with both the agent’s intention and such features of the social context as may be relevant, make the action what it is.” In addition, we may use whatever language we wish in identifying the action, so long as our language, if conveyed to the actor, or to a rival social inquirer, “would be allowed by them to be entirely consistent with their own way of putting it.” But, he warns, an acceptable report must not “preempt” any subsequent explanation, description, or evaluation of the action. This occurs when we use certain loaded terms: for example, according to Runciman, the dictionary meaning of “revolution” specifies that it is a transformation brought about “from below.” Accordingly, if two observers are reporting an armed conflict, to apply the word “revolution” to it may be “to preempt” at least part of the explanation of what both observers agree to have occurred.” This is so because to apply the word “revolution” presupposes that the conflict was caused by a certain kind of event (say, political agitation among the lowest classes of society) and arose in a particular way.
The application of the word therefore has built into it an explanation and thus cannot serve as a “report.” On the other hand, “to say that a war is a ‘civil’ war is (merely) to report that the fighting is between members of the same society: it may be a revolutionary civil war or it may not,” and therefore the term “civil war” may be acceptable in a report.
But it is important to note that when Runciman speaks of reported “facts” he does not mean either unchallengeable “brute data” free of theoretical presuppositions or data accessible only to those who have directly experienced them. He wishes to steer the social inquirer “between the Scylla of positivistic empiricism and the Charybdis of phenomenological hermeneutics.” Reports need to be couched, not in language purged of all theory, but in language whose application can be agreed upon by rival observers and inquirers. Thus if “civil war” itself comes to be regarded as contentious or “preemptive” of some explanation in a given case then the rival inquirers can retreat to some other word or expression if they wish to report the armed conflict, e.g., “fighting between groups of persons.”
Runciman’s discussion of “reportage,” however, raises the question whether neutral language of this variety is always available in conducting inquiries into social life, and he does not adequately address the claims of some philosophers of science and some interpretative social theorists who, as we shall see, have denied that such language is even available in much significant social scientific research, and who would therefore question the usefulness in practice of a methodological distinction between what Runciman calls “reportage” and what he calls “explanation.”
Whether the social theorist wishes to explain why the western Roman Empire fell, or (to use one of Runciman’s other examples) why French musicians of the nineteenth century didn’t compose more symphonies, or some other event or action or process, the logic of explanation, for Runciman, is always the same as in any other science. That is, the event or action or process is explained if it can be shown to be an instance of causal laws and generalizations. He also notes that theories in the social sciences are generally not like those deductive concatenations of laws that occur in well-developed sciences, partly because of the different part he claims laws play in a science like sociology: “it is not that laws play no role in it, but that those which do are not laws of it;…sociology is not a producer but a consumer of laws.” Thus, if a sociologist wishes to explain a rise in juvenile crime he does not need to discover a distinctively sociological law (let alone juvenile crime laws); in general, he will try to explain the phenomena in question by borrowing laws from other sciences such as psychology or economics or genetics.
Further, while Runciman does claim that theories in social science should be testable in principle, “there is no rule of method which forbids social scientists to reculer pour mieux sauter,” and the apparent untestability of some versions of such theories as Marxism and evolutionary theory need not “be interpreted as a symptom of their inherently unscientific status.” Even if they are untestable we may learn from them, and our efforts to revise them may contribute to scientific progress. As Runciman notes of scientific theories, “let us by all means discriminate between those parts which are testable and those which are not; but let us at the same time beware of dismissing those which are not as incapable of contributing in any way at all to the explanation of human behavior.”
Runciman recommends a number of relaxations in the demands made on social scientists that are consistent with these remarks. Thus he claims that since social scientists cannot test their hypotheses by manipulating human variables or reproducing human phenomena at will, researchers should “look for hypotheses of which history has happened to furnish a chance for quasi-experimental test.” For example, we may wish to explain the commonly acknowledged fact that the western Roman Empire fell when its frontiers ceased to be defensible by the forces available to protect them. Why couldn’t enough men be mustered to man these borders against barbarian incursions? One might try to explain this fact by a variety of reasons, such as poor performance among Roman generals, or the debilitating effect on the empire of frequent disputes about succession to the throne, or the depravity of the late Romans, or the logistic difficulties of moving troops rapidly enough to ward off attacks by nomadic tribes.
But suppose that it is claimed that the borders became indefensible for a different reason. The western Roman Empire depended heavily on slaves to till fields and to quarry mines, but before its fall the empire became increasingly unable to maintain or replenish its stock of slaves, whether because it could not breed enough of them domestically or because it could not purchase more of them across its borders. As a result, the hypothesis might continue, the domestic economy suffered; men who might otherwise have served in the army were compelled to stay at home, and as a consequence the frontiers were left insufficiently manned.
Runciman claims that we can “quasi-experimentally” test this hypothesis that the empire fell because of its dependency on slavery. For suppose, he says, we can discover another “overexpanded” empire similar to the Roman Empire in important relevant respects (such as level of technological development) which both succumbed to barbarian pressures (as in the Roman case) and yet whose “work-force was only negligibly servile.” In that case, he continues, the hypothesis that Rome fell because of its dependency on slavery would be challenged, because an instance of an empire in important respects like Rome which fell in a similar way can be produced in which slavery played no appreciable role of this kind. Moreover, Runciman claims to have found such an empire. “It is a fortunate accident of history,” he writes, “that there is one to hand in the instance of the later Han Empire in China, of whose agricultural population only a few percent, at most, were slaves but which progressively succumbed to barbarian pressure from the North, culminating in the sack of Lo-Yang in 311.”
But Runciman’s generous relaxation of the demand that ought to be made on social scientists in testing their hypotheses may be entirely too generous. For is it true that the two empires are sufficiently similar to serve as genuinely contrasting cases in which the influence of slavery on the Roman Empire can be tested? Do the two cases have enough relevant traits in common except slavery? The answers are in considerable doubt.* Runciman admits that the contrast between the two empires is “of course far from furnishing a strictly controlled comparison,” but he suggests that comparisons of this kind are nevertheless as much as we can expect when examining hypotheses about such complex issues. Where doubts of this kind can arise, however, it is entirely unclear what probative weight, if any, to assign to the “test.”
Runciman’s most original observations concern “description”—understanding not what something is, or what caused it, but what it was like—a topic he notes is more familiar to students of literature and biography than to social scientists. A description, as we have seen, “successfully conveys to the reader what ‘their’ experience was like to them,” and the practical test of such a description is whether the person who claims to have arrived at such an understanding “can talk back to the persons whose experiences are those described in such a way as to elicit their agreement that that is indeed what the event, process or state of affairs was like for them.” A description, he continues, must not be oversimple, or ahistorical, or exaggerated, or suppressive, or ethnocentric. It must not “mystify” its subject matter in the sense that the inquirer does “not merely fail to see what ‘they’ see but…supposes that they see it in a way which is pre-emptively dictated by a descriptive theory of his own.” If it concerns a group of persons, a description must also be “representative” in that those whose thoughts and actions are described “would agree that the divergent points of view of distinguishable groups or categories within their institution or society have been taken into account.”
But to be good, a description must also be “authentic,” by which Runciman means not that it must somehow cause us to recreate the experience described but that it must result from the exercise of “the active and sustained avoidance of what Dr. Johnson called ‘cant,’ George Orwell called ‘humbug’ and I shall call, in the idiom of my own period and milieu, ‘bullshit.’ ” He suggests that there is “authenticity” in Max Weber’s account of “the spirit of a bureaucratic age as one of a ‘mechanized petrifaction’ in which the specialists have lost all sense of vocation,” or in the celebrated passage in Marx “in which he seeks to convey what it is like for the worker to feel himself alienated and diminished by the sense of being, as it were, ’emptied’ into his product in such a way that his labor becomes an object existing independently of himself and standing in opposition to him.”
Runciman’s treatment of “description” is incomplete in two ways. First, he fails to clarify what precisely is wanted when we are asked to give a description of what it is like for us to be or to do something. Is there such a thing as a single, correct description of what it is like for us to be performing some action or having some experience? Does any one of us know what it is like to be us now? If these questions can be answered positively (as some have thought they can), then a description (in Runciman’s sense) could be true or false, because there would be something for them to be true or false of. But Runciman argues that descriptions can be neither true nor false, not because there is no fact of the matter to be correctly identified, but seemingly because different descriptions can be of “equal authenticity” even when both are given in answer to the same question.
As a result, he continues,
the test of a good description is not whether it ‘fits the facts,’ for the researcher’s problem in framing it lies in choosing between different reports all of which are in accord with his own or some other trusted observer’s observations and none of which would be rejected as inaccurate by those whose behavior is being described.
But do these considerations really show that descriptions cannot have a truth-value, or do they rather point to the existence of perhaps insuperable obstacles in decisively determining what that truth-value is in particular cases? And why might not the fact that different, but equally “authentic,” descriptions can be given in answer to the same question be construed as evidence that differently placed persons have had different experiences or have been led to interpret or misinterpret them differently by their special circumstances? Whatever the answer to these philosophical questions may be, Runciman does not promote our understanding of them.
Nor, secondly, apart from cataloging a number of different forms of inauthenticity, mystification, and misapprehension, does he provide much guidance in the practical matter of how we might arrive at consensus in judging descriptions to be acceptable. For example, when different appraisals of the same descriptions are made, how is this controversy to be resolved, especially if, as Runciman claims, the descriptions themselves are neither true nor false?
Runciman does provide, however, a careful and extended discussion of the related question of how descriptions may be conveyed to the reader. Here the main problem is not merely to translate the words people use into a different language, but “to bridge the divide between the reader’s experience and the experiences of the people whom the researcher wishes to describe to him, and this can only be done by appealing at some point and in some form to what the reader is presumed already to understand.” Novels and films and photographs, Runciman says, can also serve as means of conveying descriptions. Novels “are sociology to the extent that their authors make them so.” Similes can also accomplish this “bridging” of experience, he claims, as when one might say that “eating meat, for a vegetarian, is like eating human flesh for a meat eater.” It may be objected, however, that “bridging” can only occur if the simile in question accurately conveys the experiences of “vegetarians,” and this raises once more the question of whether there is some common experience of this kind. It seems not to be so, however, since one who refrains from eating meat for reasons of health is unlikely to give the same description of what his experience is as one who does not eat meat out of respect for the “rights” of animals.
As we have seen, Runciman claims that if his distinctions are accepted and his arguments are correct, the classical debate over the nature of the social sciences “can be regarded as closed.” But regarded as closed by whom? Is Runciman’s use of his distinctions likely to close the controversy to the satisfaction of those who have participated in the debate? To take but one example out of many possible ones to illustrate why I think it does not, consider whether his claims are likely to settle one of the points of disagreement between “interpretive” social theorists and those who oppose them. According to at least one influential formulation of the interpretative position, the “meanings” that partially constitute human actions are typically elements in a configuration of categories and rules that are embedded in behavior. They are part of “social reality” or a “form of life” discoverable less by empirical observation than by “making sense” of conceptual connections. These rules and conventions “constitute” this social reality in the way that the rules of chess “constitute” chess: they define the possibilities and limits of a way of behaving and must be “grasped” if we are even to identify a significant range of actions that occur in it. To understand a handshake requires that we “see” or “make sense” of the connection between the clasp of hands and such socially recognized concepts as courtesy or forgiveness, but also that we know what can and cannot be accomplished with a handshake, for example that with it we bring a business transaction to a close, but not, say, draw money from a bank or pay a bill.
In extreme versions of this view, not only must social inquirers “grasp” through “interpretations” the conceptual linkages between particular actions and ways of life; they are also unable to test these interpretations except by other interpretations. They cannot escape this “hermeneutic” circle of interpretation because “social realities” are held to be generally “incommensurable.” No “neutral” language can be devised in which “objective” empirical tests of competing interpretations can be couched. We can, for example, describe the properties of virtually any material body anywhere in the language of Newtonian physics, but because “forms of life” are not comparable, there can be no neutral language of this kind in which to “translate” or describe the social realities of Indonesian peasants, Swiss mountain climbers, and British factory workers. Accordingly, neither the grand structures of social scientific laws and theories envisaged by nineteenth-century positivists nor even the social science described by Max Weber (in which “interpretation” plays the limited role not of helping us verify our hypotheses, but of helping us arrive at our hypotheses) is “possible.”
Those sympathetic to Runciman’s position would no doubt claim that this extreme view is flawed, that it grossly exaggerates the degree to which social science is devoted to the discovery of deep conceptual connections in the world views of human beings, and that many actions and events in social life can be identified without the discovery of such connections. They would further argue that its central notion of “social reality” is vague and that its suggestion that we cannot “really” understand one item of social life without understanding its place in the whole of social life is unfounded. Moreover, they would claim that its prescription that there is no way out of the circle of interpretation is just a metaphysical dogma and misleadingly implies that once one has “entered” or understood another form of life one can never stand back from it and test one’s understanding of it against observations that do not presuppose it.
But the question at hand is not whether the interpretative view is correct but whether Runciman’s use of his distinctions and his argument that “there is no special problem of explanation in the human sciences, but only a special problem of description” could dispel controversies between exponents of rival conceptions of social science. Thus, for example, would interpretative social theorists who hold the position just sketched be likely to accept Runciman’s suggestion that once they recognize his distinctions between three senses of “understanding” they will also recognize that their central concern has been with “description” rather than with reportage and explanation? How would they regard his view that their extreme conclusions about the character of social science need not follow once this concern is so formulated because “description” can be accomodated within a conception of social science that nevertheless holds its methods of explanation to be the same as those used in other sciences?
Thus Runciman’s suggestion that there is no special problem of explanation merely invites the response that he has made contentious use of his distinctions to misleadingly insulate the interpretative theorist’s concerns from the alleged “objectivity” of reportage and explanation. The interpretative theorist would not need to challenge the general distinction between, say, understanding what a prayer is, what causes praying to occur, and what it is like to pray. But he would argue that his concern with the problems of “making sense” of conceptual linkages in the “world view” of actors is with specifying conditions necessary to, and inseparable from, the identification, not of what acts are like, but of what they are. Thus the anthropologist is not inquiring into what it was like for the native to pray but what the native’s prayer was, what it consisted in, what concepts and meanings “constituted” it.
To call this activity of the anthropologist “description” in Runciman’s sense suggests that actions like praying can be independently indentified and that then we can go on to ask what it was “like” for different persons to do it. But the interpretative theorist would argue that the “it” in question cannot be identified in this way, but only through inspection of the connections between the behavior—the gestures, movements, noises—in question and the rules and concepts of the native’s form of life. Accordingly, he would say that his interpretative concerns have just as much a claim to be called reportage as “description.” This would not, however, be “reportage” in Runciman’s sense according to which a valid report cannot “preempt” any subsequent explanation or evaluation; for the report of an action like a prayer, according to the interpretative theorist, generally incorporates into it the explanations and evaluations ingredient in the “native’s” point of view. He might further argue that because no neutral language of reportage can be devised to speak of different forms of life and because the way in which one grasps conceptual connections (and tests one’s grasp of such connections) is entirely different from that used in formulating and testing causal generalizations, Runciman’s view that there is only a special problem of description in the human sciences, not one of explanation or reportage, is unfounded: there is a special problem of reportage and explanation as well, and as a result the methods of social science and its standards of validity do indeed differ radically from those of other sciences.
Since this is just another way of raising the old problem of the “nature” of the social sciences, one might suspect that Runciman’s use of his distinctions has merely renamed and rearranged the components of this old question in such a way that rather than disappear it can rearise in a new formulation. If this is so, of course, then not only is the debate not “closed,” but Runciman’s methodological maxims fail to pass his own pragmatic criterion of success: to tell practicing scientists “what conditions their research must fulfill if its results are to be accepted as uncontroversial by fellow-practitioners,” and to enable them “better to succeed in what they set out to do.”
This outcome is unsurprising because the debate in question rests on a number of intransigent philosophical questions—such as whether reports of actions can be couched in a “neutral” language or whether “descriptions” can be true or false—that are not likely to be resolved soon to the satisfaction of rival theoretical schools in philosophy or outside it. Indeed, no one would take the debate seriously unless he took these problems seriously, and it is therefore unlikely that progress can be made in resolving the debate without an accompanying advance in our understanding of these questions.
Runciman apparently thinks that he can “close” the debate without seriously addressing them. He claims that he writes from the “viewpoint” of the “methodologist” who can “bypass” philosophical disputes that inevitably arise with reflection upon the practice and content of social science. As a methodologist, he continues, not only will he “not presume to teach practising sociologists their job” or tell them “how to conduct their research,” his arguments will not “advance by a single step the resolution of any of the disputes among philosophers of science on which its arguments touch.”
Runciman is surely correct to stress that science is a social institution whose activities are controlled by conventions and criteria, whether tacit or openly acknowledged, which are accepted by its “practitioners,” and that working scientists need not try to satisfy the abstract doubts of “lecture-room skeptics” who propound such insoluble puzzles as whether groups are as metaphysically “real” as persons. Such puzzles, it might be added, can hardly be said to have caused any considerable delay in the blossoming of social science. But this policy of “bypassing” philosophy will not help resolve a debate that can scarcely be settled without a larger understanding of the philosophical problems that animate it. In any case, Runciman does not really succeed in bypassing philosophy, for while he may profess to circumvent a number of philosophical questions (such as the meaning of “cause”), a number of his “methodological” recommendations are simply disguised (and often undefended) declarations about some of these questions.
Thus, for example, when he claims that descriptions cannot be true or false, he is in effect, to use his own language, “preempting” at least part of a solution to the philosophical question of whether there is such a thing as a determinate subjective experience about which propositions may be true or false. It is therefore no surprise that the debate over whether social science needs to “understand” the meaning of human actions in some special sense can rearise (as we have seen) at this and other points of Runciman’s argument which he has left unprotected because of his decision to bypass philosophy. Runciman appears to wish at once to provide reasons for ending the debate and also to step aside from the task of resolving (or dissolving) the questions on which it largely rests. The strategy cannot be followed profitably, and the consequence of following it is that Runciman achieves neither objective.
Nor is Runciman’s positive conception of “methodology” without difficulty. If it provides neither elucidation of the philosophical problems of social science nor what he calls “technique,” what precisely is it supposed to do, and why should we follow the maxims and recommendations that comprise it? Runciman claims, for example, that “there is no logical, but only a contingent, relation between moral, political and aesthetic values and the values of science” and that a methodologist can always reconstruct a work of social science so that “the value judgments are sifted out and the explanatory and/or descriptive content is left behind.” But since he is not resolving philosophical issues, he will not provide a sustained discussion of the complicated philosophical question of the relationship between fact and value; he accordingly offers us little more than an unjustified declaration about the issue. On the other hand, since he will not provide a “technique,” the declaration is of little help to the practicing sociologist who, if he is troubled by problems of evaluation, is likely to be less concerned with the “logical independence” of the values of science and other values than with questions of how they are contingently related and how to distinguish evaluations from factual reports and explanations as he goes about his inquiry.
Runciman might admit all of this and yet claim that he is concerned only to specify the “conditions” that “must” be fulfilled by social science research if “its results are to be accepted as uncontroversial” by social scientists. But what is the justification for these “conditions?” Why “must” these conditions and not others be fulfilled? And how is Runciman in a position to say in advance what can, or cannot, be accepted as “uncontroversial” by social scientists? In the end, not only does Runciman’s book strangely sidestep the debate to which it is ostensibly addressed, but the “methodology” it provides is too thin to be of appreciable use. Its maxims appear to be rules from nowhere.
November 8, 1984
For example, it is not clear that Runciman is correct is claiming that the Han Empire “progressively succumbed to barbarian pressure from the North, culminating in the sack of Lo-Yang in 311.” Joseph Alsop has kindly drawn my attention to sources which imply that Runciman’s claim is incorrect, in particular to two books by the late Arthur F. Wright, Buddhism in Chinese History (Stanford University Press, 1959) and The Sui Dynasty (Knopf, 1978). According to Wright, Lo-yang was indeed sacked in 311 AD by a Hun confederacy and the sack “marks a turning point in Chinese history comparable, as Arthur Waley has suggested, to the sack of Rome by the Goths in 410.” But Wright also claims that the Han Empire had dissolved some ninety years before the sack of Lo-yang, so that the latter could hardly have been the “culminating” event in the empire’s fall. Wright states that the later, or eastern, Han Empire began to decay as early as the second half of the second century AD; he dates its end at 220 AD, which Alsop claims is “universally accepted as marking the end of the Han.” In that year, according to Wright, the Han were displaced by the Ts’ao family (headed by the military adventurer Ts’ao Ts’ao), which in turn founded what Wright explicitly calls the “successor dynasty” to the Han, the Ts’ao-Wei (220–265). In 265, he continues, another dynasty altogether, the Chin, seized power from the Ts’ao, and was soon thereafter caught in “a many-sided struggle for the succession known as the War of the Eight Princes (290–306).” By 311, the city of Lo-yang was no longer a Han city, but in the hands of the western Chin. Its sack, therefore, played no role in the fall of the Han Empire. ↩