It is a common observation that the social disciplines suffer from serious and deep-seated problems. Various authors have their lists (e.g., Blalock 1984) and I have my own ideas about what most needs attention, which I will discuss briefly here. The seminar on which this volume is based assuredly did not solve any of these problems, and we did not directly talk about some of them at all. Nevertheless, it is important to review these pervasive difficulties, both as a basis for discussing some results of the seminar and to encourage further work on these problems.
Adequate conceptualization of sociocultural phenomena is a task still in its infancy. It is quite unsporting, like shooting fish in a barrel, to ridicule social scientists for turgid, obscure, or pretentious jargon. How-ever, their prose often reflects a real effort to break free of the bonds of the categories, conceptions, and misconceptions of everyday plain English (or plain French, or whatever). These efforts have achieved limited success Solar; rbutrthsy arcsafrhcy. highest.importance.
Trying to think effectively about social phenomena calls to mind the fable of the blind men and the elephant. We are all groping. When someone “sees” a part of the beast we are trying to comprehend, it is only a partial and limited insight. So far, the different insights are incommensurate and we do not see how to put them together. A central point about the great successes of the physical sciences in the past few centuries is that people found good ways of "seeing" the phenomena they studied. It became reasonably clear what was important and also what was unimportant, which concepts of everyday thought had to be drastically redefined (e.g., energy) and which had to be discarded altogether (e.g., thinking of fire or heat as substances). To be sure, as Kuhn (1970) has emphasized, ways of seeing physical phenomena that were highly effective for a time eventually broke down and generated paradoxes that led to crises and revolutions, and to new pictures of the physical world. But the old pictures would not have generated crises if they had not been so clear and, up to a point, so effective, detailed, and accurate in their predictions. This kind of clarity and detail remains sadly lacking in the social disciplines. We have the specious clarity of the grand abstract schemes, clear because so bare of detailed and differentiated implications for different specific sets of circumstances. These schemes contrast with analyses of specific episodes that are detailed, subtle, and often persuasive, but hard to see how to generalize effectively. We do not yet have schemes that are simultaneously clear, able to handle rich content, and reasonably effective at prediction, or even able to persuade a majority of scholars that they offer satisfying understanding.
A debate exists between advocates of approaches modeled on the physical sciences and promoters of an “interpretive" approach, whose most prominent advocate is Clifford Geertz (Shankman 1984). Perhaps it is futile and misguided to try to achieve significant nomothetic insights into social phenomena. It must be admitted that efforts along these lines still leave much to be desired. However, too many social scientists have an impoverished view of the content, methods, and style of the “hard” sciences. It is not that there is something about a scientific approach that makes the richness of phenomena thin and dry, it is that too many people have an excessively thin and dry idea of what it is to do science. I believe we can make nomothetic generalizations that do justice to the idiographic richness of individual cases and that the best way to make progress is to continue to seek explanations that have a logical shape similar to explanations in the physical sciences. The last thing I advocate is that we should borrow specific concepts from, or seek close analogies with, physical or biological sciences. But I urge that we strive for models that achieve high outputs of usefully accurate predictions and postdictions of significant phenomena in return for relatively economical inputs of relevant data and parsimonious theory.
To make my arguments clearer, I will list several major problems in thinking and talking about social phenomena.
lf one examines a commonsense concept closely, it often turns out that "there isn’t any such thing." An acquaintance who launched a maior study of power reached the conclusion that there is no such thing as power. lt is like the episode in Lewis Carroll’s Through the Looking Glass in which Alice is in a toyshop whose shelves are obviously fully stocked, except that, exactly wherever one is looking directly, there isn’t anything. Our daily lives are replete with concepts (and words for these concepts) that are perfectly clear and unproblematic to all of us, except ` the persons who have studied them. A good example is the concept of “collapse of a civilization,” which, as the chapters in this volume show, is a far less simple idea than we have been accustomed to think.
What should be done about this? We cannot do everything at once, and it seems necessary to begin any investigation of social phenomena by taking for granted a number of terms that are, one hopes, peripheral or background concepts, and to concentrate on developing clear and nontrivial meanings for some limited number of key concepts. The strategy is to expand cumulatively the number of incisively defined concepts that are connected in a network of mutual relevance. To judge by experience in the physical sciences, there are some everyday concepts that are best done away with entirely. Often, however, a meaning is found that is related only remotely to everyday thought, but which plays a well-defined and useful role in technical thought. For example, quantum mechanics gives a picture of "matter” that is strikingly different from our everyday thought and experience. This has not led physicists to say that matter doesn’t exist. Instead, they say we have to think about matter in a way that, although it leads to highly accurate predictions, is disturbingly different from the way we grew up thinking about it.
A closely related problem is a tendency toward hairsplitting in some semantic domains, while casually using other terms that conceal semantic quagmires. It is hard and frustrating to clarify and make precise our thought and language, but it is vital work. We must beware of starting with a set of terms and then seeking meanings for them. We should try, instead, to identify genuinely different concepts and then assign names to each. Also, we should look for sets of phenomena that seem more than superficially similar, look for what seem to be recurrent kinds of relationships among them, and give clear names to each of these kinds of phenomena and relationships. Logical clarity in conceptual schemes is also important, but mere clarity is not enough if the schemes are too abstract to help when we try to understand specific instances in any depth. We may find the polysyllabic tongue-rwisters of the organic chemists ugly and unintelligible, but we do not ridicule them for the excellent reason that, to initiates, these chemical terms convey very clear and precise information. They are essential for investigations that are obviously highly effective.
While we focus on trying to develop clear, precise, relevant, and rich ideas about one topic, we necessarily use many other terms in an offhand and unexamined way, as if we all knew what they meant and as if we all understood exactly the same things by them. In fact, the amount of covert misunderstanding and miscommunication that goes on among students of human societies is hair-raising. The fallacy of arguing about words rather than ideas, of using different words for the same concept, is well known. But there is also the fallacy of failing to argue about different ideas because the differences are masked by shared use of the same words.
The main things to be done about these semantic problems are to be watchful of one another and to try to catch ourselves when we use words vaguely, or contradictorily, or with two or more confusingly different meanings. This is not to say that there is no proper use for ambiguity. On the contrary, we need some words whose meanings are broad, and we need others whose meanings are a bit fuzzy, with slightly indistinct semantic borders. However, and I intend no paradox, we must be precise about the broadness or fuzziness of these terms, and they must be complemented by other terms that are as nearly razor sharp and crystal clear as we can get them.
In later partsof this chapter, I have tried to put these exhortations into practice. I hope I have had some success, but readers will have missed the point if they do not think about how my terms and concepts could be improved.
Another persistent source of trouble is aping the shadow rather than the substance of fashionable terms that have proven genuinely useful in other fields. As long as the social sciences remain intellectually under-developed, they will be subject to “cargo cults” and messianic enthusiasms. The-only.remedy is to develop indigenous theory that is rich and successful enough to prevent our being easily seduced or bamboozled by clever, ambitious, and energetic persons spouting half-baked or half-digested versions of something that has become popular (often for good reasons) in some other discipline.
Our susceptibility to these fads is a major way of failing to communicate miscommunication. The fact that words such as entropy, information, homeostasis, energy, and force have very well defined and useful technical meanings in other fields means that it is easy to bootleg them into human studies without noticing that our use of them is, at best, extremely vague and is often only metaphorical. Some of us may be afraid to admit that we don’t quite know what they mean (the "emperor’s new clothes" syndrome), but I believe that more often the semantic murkiness is simply not noticed.
As an example of obscurity, consider the phrase, “This was due to the operation of the law of .... " The wording suggests that scientific laws commonly are agents that cause certain phenomena. However, laws are not agents; they are generalizations about relationships among entities that hold over some usefully wide range of circumstances. We may say, "This apple fell because of the operation of the Law of Gravity.” But we know that what is meant is that the apple moved toward the center of the earth because there is a physical force between any two masses, described by the Law of Gravity, and because motions of masses are affected by the forces acting on them in ways described by simple laws of motion. When social phenomena are said to be due to the operation of some law, it is far less clear that either author or reader appreciates that the explanation, if it is not simply fallacious, must be a treacherous shorthand for the effects of phenomena whose behavior is believed to be described by the law in question.
One style in the study of human affairs is the formulation of grand, sweeping, general theories. Many people find them quite pleasing. Some of these theories are grandly wrong. Others, however, are not especially wrong, as far as they go, but they have remarkably little empirical content. They are very limited in their logical implications about just what should or should not be observed in specific circumstances. Such theories are resistant to falsification (it is hard to find evidence that is clearly counter to expectations), and they offer some people a satisfying sense of having explained much because they can be invoked as applicable in so many instances. But they have very weak predictive power and tend to be unfruitful; they fail to generate critical problems that could guide further research.
A contrasting style is the rich, meticulous, and often fascinating examination in depth of the detailed particularities of events in a certain time and place. Between these styles is a wide gulf. It is scarcely seeing the forest versus seeing the trees. Often users of the first style have trouble seeing the forests because of the continents, whereas followers of the second have trouble seeing the trees because of the twigs. We urgently need studies that lie between these extremes, that deal with particulars but do not assume that facts somehow speak for themselves. These studies should look beyond isolated examples at a reasonable number of instances in sufficient detail and with sufficient understanding so that they can arrive at insights that, though generalizable to a useful extent, have less than cosmic pretensions and can claim to explain fairly specific aspects of specific situations. This is, of course, exactly what we hoped to achieve in this volume.
Another problem in social studies is the diverse levels at which entities are conceived. Sometimes the objects of attention are individuals or house-holds, sometimes they are communities, sometimes they are institutions, and sometimes they are entire states. There are many levels, but we can speak roughly of a "micro” versus “macro” contrast. Some scholars tend to explain changes by the decisive acts of heroic individuals; others deny that specific individuals make any difference at all. It is platitudinous to say that neither extreme can possibly be the whole truth or even most of the truth. Merely saying this, however, does not get us very far. How can we effectively conceptualize explanations that accept the relevance of phenomena best described on diverse levels, some in terms of individual actors and some in terms of larger social units? lt simply won’t do to think that if one grossly inadequate model explains a little bit of the whole thing and other grossly inadequate ideas explain other bits we can just add them all up and, collectively, they will amount to a pretty good explanation. This "additive” strategy for improving theory, which some writers seem to favor, is painfully misguided. We should be looking for theory that integrates diverse things in some structured way. But how do we do this? How do we formulate theory that takes account of individuals and also takes account of “emergent” phenomena that refer to no specific individuals, but to things generated by interactions of many individuals?
It is helpful to keep in mind the difference between proximate (or immediate) causes and ultimate causes. We can always give an account in which the macrolevel outcome is viewed as the immediate consequence of the myriad microlevel details of what happened before. But can we give a more parsimonious account in which the microlevel details don’t matter, and the important parts of the macrolevel outcome are explained solely with reference to previous states of phenomena also described on the macrolevel? This doesn’t solve the problem, but I hope it formulates one aspect of it a bit more clearly.
Some people regard ideas, attitudes, and individual character (either of a few key actors or of large segments of society) as decisive for explaining sociocultural change. Opposed to this is the narrowly materialist view that ideas have no causal relevance, that changes in ideas are merely consequences of changes in material circumstances and do not play any independent role in causing further changes. I wish we could quickly dismiss both extreme idealism and extreme materialism. However, unhappily, examples of both continue to be common in scholarly publications (as Flannery 1971:400 notes).
It is important to distinguish two styles among materialists. Some simply tend to ignore mental phenomena. They explain material phenomena entirely in terms of other material phenomena and they just don’t talk about anything else. This approach is wholly inadequate for two reasons. First, mental phenomena are an exceedingly important category of social phenomena, and an account of a social situation is grossly unsatisfactory if it leaves them out or assumes that, if they are understandable at all, they are to be explained by some quite unrelated discipline. Second, it makes the unwarranted assumption that mental phenomena are not relevant for the explanation of material phenomena.
Many materialists, notably Marx, have known better than simply to ignore ideas altogether. Even such an extremist as Marvin Harris (1979) recognizes that people think and have feelings, that they are concerned about codes of behavior, that their actions are shaped by mental as well as physical needs, that they pursue strategies intended to achieve certain outcomes and avoid others, and that what people do is affected by what they believe. However, Harris argues that ideas rarely change spontaneously in any way that in turn causes significant changes in material circumstances. Causation flows overwhelmingly from the material to the mental. Changes in ideas that are, themselves, consequences of changes in the material realm may, in turn, have an influence on further material changes. However, these mental phenomena only mediate between one set of material circumstances and another set of material circumstances. Mental phenomena play an autonomous causal role only rarely and to a limited extent.
lt is unlikely that the role of mental phenomena is really as small in good explanations of sociocultural change as Harris argues. Nevertheless, we cannot simply dismiss this variety of materialism. In any case, 'material' circumstances are tremendously important for understanding sociocultural change. To underrate them is just as obtuse as it is to ignore mental phenomena.
There is, however, a serious error of method in I-Iarris’s brand of cultural materialism. He advocates treating as a last resort explanations that assign to mental phenomena some autonomous role in causation. Such explanations are to be entertained only if every imaginable explanation in purely material terms has been tried and found wanting (Harris 1979:56). The fallacy here is that what might be defended as a good strategy for discovery (although I would not defend it myself) is applied in the context of confrmtztion. If cultural materialist explanations are really better than other kinds (as evaluated by the accuracy of their predictions and the diversity and importance of the phenomena to which they apply), the best way to establish this is to compare them on an equal footing with other explanations, including explanations that allow mental phenomena to be causes as well as consequences.
I emphasize this because it brings us to what continues to be the key problem in understanding sociocultural change. Ideas can be both causes and consequences, but we cannot just “add” ideas and material circum-stances or tack on transfers of information as an extra feature of models that began by dealing with transfers of matter and energy. How, then, can we put ideas and material circumstances together in an effective and unified theory? The essays included in this volume should help to improve theory that tries to integrate the mental and the material. The interplay of materialist and idealist concepts was a recurrent theme in our seminar. There has been a surge of archaeological interest in this problem (e.g., Hodder 1982.), but Millon (personal communication) reminds me that we should familiarize ourselves with what earlier thinkers such as Max Weber (1978) had to say on the matter and with criticisms and elabo-rations of their ideas.
One borrowing from other disciplines that has been especially problematic in social studies is the concept of “evolution.” Attempts to apply this concept effectively to human societies have been extensively criticized by others, notably Yoffee (1979), and they receive further attention in the first chapter of this volume. I will add only that I think social evolutionists are especially prone to invoking the “operation” of the Law of This or the Principle of That as pseudo-explanations.
A second problematic borrowing is systems theory, especially “General Systems Theory.” This approach has also generated strong feelings, both positive and negative. Merrilee Salmon (1978) offers a levelheaded appraisal (cf. Doran 1970 and Spaulding 1973). It is hard to quarrel with the general notion that many of the entities we want to think about are systemlike. That is, they have reasonably distinct and differentiated parts (subsystems), and the parts tend to be interrelated. A change in one part is likely to have consequences for other parts, and the connections between parts form a more or less complex network with many reciprocal (two-way) relationships, rather than unidirectional, unbranching chains from prime causes to initial effects, secondary effects, tertiary effects, and so on. We must look at social phenomena in this broadly systemic way if we are to understand them. But we must also specify clearly and effectively the entities we are trying to think about, how we should think about their differentiated parts, and, especially, specify in some detail just how the different parts affect and are affected by one another. Above all, these specifications must be rich enough in logical implications about real-world instances to make it easy to imagine evidence that would falsify them and, if they survive repeated attempts at falsification, to carry us interestingly beyond what we already knew. One of the banes of "systems-talk” is the use of complex diagrams, full of boxes and arrows. Too often the arrows show only that a change in one part is somehow relevant for another part or, at best, a symbol is added to show whether a positive change in one part tends to cause a positive or a negative change in another part. Such diagrams have been described as “less than meets the eye." Several archaeologists who have attempted computer simulations have suggested that, in the current state of the art, one of the greatest benefits of such efforts is the salutary experience of being forced to think harder about exactly how the different parts affect one another (e.g., Aldenderfer 1981).
Two major problems arise in many recent uses of systems concepts in studies of societies. First, the rather well developed systems vocabulary encourages us to think that we are much further along toward effective theory than we really are. It gives us a specious sense of clarity and understanding, because we have so many new labels to apply. However, many of these terms, such as “homeostasis,” “linearization,” and the like, have been either applied loosely or used, at least so far, as synonyms for more familiar expressions. They are useful to the extent that they call attention to things that might otherwise be overlooked. But they are no more than initial orienting concepts for beginning the hard thinking. However, people tend to define these systems ideas rather vaguely, then erect increasingly elaborate structures that are abstractions built out of abstractions. The-language flows most satisfyingly,·diagram follows diagram, and one has the sense that something is being talked about. There is far too strong a tendency to take the terms and concepts of systems theorists as relatively unproblematic building blocks and to think upward from them. Instead, we should take them apart, look at them more critically, and think down, as well as up, so that the systems concepts are better connected to rich bodies of actual data. The second major problem with systems thought is that it lends itself to begging a central question. It encourages us to assume that sociocultural entities are normally highly integrated (highly “systemic”) with well-developed mechanisms for self regulation. lt is remarkable that aggregates of individuals of the species Homo sapiens have been able to establish state societies at all, even problematically and for short times. Most writers about states assume that there was a time when humans had no states, and they feel compelled to account for the beginning of states. They rarely feel a comparable need to account for the persistence of states. But our psychological and social propensities evolved under the selective pressures prevailing in bands of rarely more than twenty-five to fifty persons, with population densities rarely higher than a few people in every hundred square kilometers. It seems extraordinary that, with nonindustrial technology, humans have created social systems incorporating ten thousand to a hundred thousand times as many individuals as the societies that shaped the evolution of ancestors biologically similar to us. This is an astonishing example of the plasticity of human behavior, of the cultural options that our biological substrate permits. My wonderment about this underlies all that I have to say about large-scale societies.
In fact, probably most states don't work very well. Many analysts, Marxists and others, have emphasized conflict, especially between classes in stratified societies. They have also argued that societies may experience contradictions – for example, between means of production and social relations of production. Sometimes the extent of conflict or the seriousness of the contradictions has been exaggerated and the amount of counter-balancing cooperation and compromise underrated. After all, cooperation and compromise are important in bands of twenty-five. Morton Fried (1967, 1978) and Elman Service (1975, 1978) have emphasized opposite sides of this issue. Both have seen some parts of the elephant (to return to my earlier simile). It is time to move beyond this particular debate.
The point I am more interested in emphasizing is that there are many sources of trouble for states besides the conflicting interests of diverse groups and the contradictions that arise when the social system and ideology are incompatible with the technology and organization of production. Other problems include technical difficulties in getting and properly analyzing important information, mistaken beliefs about the likely consequences of policies and actions, bungling, and everything implied by Parkinson’s (quite seriously intended) Law (Parkinson I957). I strongly suspect that the notion that any state has ever functioned really smoothly is just one of the myths of ruling elites. Even in their best days, nearly all states have probably been quite ramshackle contraptions, at best halfunderstood by the various people who made them, maintained them, coped with them, and struggled against them. Probably states avoid failure as long as most of them do, not because they function smoothly, with truly effective self-regulating devices, but because most of their people do not perceive rebellion as a realistically appealing option and feel that the best they can do is to put up with their state. It is instructive to read Goubert’s (1966) account of affairs in France during the reign of a king as strong and given to putting things in order as Louis XIV.
Many readers may feel that I have just repeated a number of truisms. Perhaps so, but when one gets absorbed in systems talk, it is much too easy to lose sight of these truisms. We get tricked into the assumption that states are normally well supplied with well-working mechanisms for self regulation until a "system pathology" comes along to upset things. I propose something closer to the opposite assumption: that large-scale human societies are normally not working quite the way they are supposed to, not wholly under anybody’s control, and at least somewhat misunderstood (in diverse ways) by everyone. This is not to say, of course, that all states are in equally bad shape. Obviously, at some times states are in much worse condition than at other times. Furthermore, what is “ good” for one segment of society may be “bad" for another segment, and vice versa.
Thus, it is surprising that states have come into existence at all, and the problem of how and why they fail is an aspect of the problem of · how and why they ever exist. By pointing out that humans do not “naturally” live in large societies and that we cannot smuggle in the assumption that states normally have highly effective homeostatic control mechanisms by labeling them systems, I want to call attention to the thought and effort required to create states and to keep them going. Eisenstadt also emphasizes this (Chapter 10, this volume). Statecraft is a human invention (or, rather, many related inventions) that commands respect, although ` possibly grudging respect. In certain contexts and in some scholarly traditions, people have known this for a long time. Most of the early great Chinese thinkers (Confucians, Legalists, Mohists, and others) were explicitly concerned with proper management of states. In this volume Hsu (Chapter 7) describes how it became accepted in Han China that good statecraft requires close attention by a large and highly schooled body of carefully selected officials. In classical Greece, politics were also a major object of philosophical discussion, and theories of statecraft were explicit.
I have the impression that China and Greece are exceptions, and that ` in other early civilizations there was much less reflection about statecraft or, at any rate, much less that was made public and recorded. Of course, even though officials tended to be more literate than most people in ancient societies, a good deal of practical development of techniques must have taken place and been transmitted without much systematic reflection or written discussion. Just as states differ in the material technology available to them, they must differ in the lore and techniques of statecraft at their disposal. Are there significant differences between Chinese and other early . states that can be related to the apparently much stronger tradition of self-conscious statecraft in China? I return to this question in connection with Bronson’s views (Chapter 8, this volume) about South Asia.
Phrases such as collapse of civilization and decline or fall of empires are examples of the troubles with words that I have discussed above. They have been used to mean a variety of things. Often they are used more or less interchangeably, or without much thought about differences in their meanings. We need to make these terms more precise.
One important distinction is that between the decline or deterioration of something and its actual termination. Judgments about moral or aesthetic decline are defensible if we take pains to make explicit the standards of quality and recognize that not all reasonable people need share these criteria. Nevertheless, I feel surer talking about decreases and increases in quantity or quality of more material phenomena, such as volume of long-distance trade, income per capita, social mobility, political centralization, security of life and livelihood, disparities in the distribution of income, and the like. We can speak fairly clearly about changes in primarily economic, political, technological, and demographic aspects of societies, and our discussions can often be phrased in terms of increase or decrease, elaboration or simplification, expansion or contraction. Admittedly, we often lack the data to infer much more than the direction of the change, and even this much can often be debated by reasonable persons. But at least we can understand fairly well what we mean by a change in one of these aspects of a society.
Contemporary views on what was happening, to the extent that they are available to us, are also relevant data. We need not, of course, agree with any of them. It is also important to remember that persons in different social positions may have had very different views, not all of which have been transmitted to us with equal clarity. Certain themes seem chronic, especially complaints that things are not as good as they used to be and dire predictions that they will get much worse. It is impossible to evaluate these without some idea of changes in their frequency and intensity and the extent to which, at any given time, there were others who took a brighter view of things.
In turning to phrases that refer to the termination of something, rather than to its decline, it is useful to distinguish two broad categories. One set refers to the end or transformation of a civilization, whereas the other refers to the political fragmentation of a large state or empire. This distinction is often blurred. One source of confusion is that civilization and society are often used nearly as synonyms, and society in turn is not sharply distinguished from state, so that it is not clear if the collapse of Rome means that the Roman great tradition was no longer being practiced; that Roman society was in a condition of acute disruption and, in important senses, had ceased to function; or that the Roman state had fragmented. Very often, of course, more than one of these is meant. Nevertheless, they are distinct, and we need to have distinct terms that mean precisely one thing or another thing, but not more than one thing.
We should clearly differentiate between state, society, and civilization and use the last term in a specifically cultural sense, to mean what Redfield and Singer (1954) call a “great tradition? To speak of the collapse of a civilization, then, should be to refer to the end of a great cultural tradition.
In contrast, the collapse or fall of a state or an empire is a concept with obvious political reference. However, the coming apart of a large political system into a number of smaller, politically autonomous units is more accurately described as political fragmentation. I urge that we avoid using collapse or fall as synonyms for political fragmentation.
In this scheme the breakdown of a society is not synonymous with either the collapse of a civilization or the fragmentation of a state. Social, political, and cultural troubles tend to exacerbate one another and are often systemically interrelated. However, it is possible to have acute social problems (if not a complete social breakdown) without political fragmentation. It is also probably possible that some kinds of large states can break apart without incurring severe social troubles.
As Yoffee observes in Chapter 1, the complete termination or even the rapid drastic transformation of a civilization has been a rare event, at least so far. Political fragmentation is more common. For example, the great traditions of Mesopotamia and Egypt lasted for about twenty-five centuries, and the Chinese has endured nearly as long. But each has seen several cycles of political fragmentation followed by reconstitution of empires. And, although the Roman empire assuredly came apart politically, Bowersock (Chapter 6) argues that the civilization of Classical Antiquity did not perish, but was transformed in Late Antiquity and the early Middle Ages.
We asked seminar participants to focus mainly on political fragmentation, partly because this wouid provide us with more instances. Another reason, as is suggested by the Roman example, is that it can be difficult to decide when it is best to say that a given civilization is no longer practiced by any society and when it is better to think of the civilization as surviving, but in a drastically altered form. This is an important problem, but we feared that if we encouraged participants to emphasize it we would be led too far from questions that interested us more. We hoped that if we concentrated on political fragmentation we would have a theme large enough to be important but specific enough to enable us to make substantial progress within the scope of the seminar. We suggested that the central topic should be successes, failures, and problems of sratecraft. We asked participants especially to discuss specific episodes in which large, nonindustrialized states fragmented and were succeeded by smaller states that did not immediately reunite politically.
At the same time, we did not want to discourage participants from discussing anything but political matters in the narrowest sense. In order to understand how and why large states fail it is necessary to understand the connections between politics and all sorts of economic, social, technological, and environmental phenomena, as well as the connections between politics, ideologies, and the modes by which people believed, thought, and felt. Our principal goal was to gain a better understanding of how all these things influence one another. However, we did not want especially to look at instances of termination or transformation of great traditions; we wanted mainly to examine instances of political fragmentation. We hoped that participants would bring up whatever seemed relevant for understanding these instances. This, naturally, involved us in discussions of great traditions, among other things.
Political fragmentation is only one of the kinds of troubles that beset governments. When we were planning the seminar, a colleague at Brandeis suggested that one of our instances might be the fall of the French ancien regime in 1789. That was a time of political and social crisis. Many holders of high office and/or high status lost their positions and often their lives. The old political system was abruptly ended and replaced by a series of very different systems. There was much more than a rebellion, that is, the attempt forcibly to replace one set of officeholders with other people. In 1789 there was a true revolution: the introduction of a new system of government, with new ideologies and new bases for claims to legitimate political authority. However, this revolution was not followed by any significant political fragmentation. In fact, later French governments allowed less regional autonomy than had existed under the ancien régime. What fell in 1789 was not a state, but a system of govemment.
In the first century B.C. the Roman republic suffered a long series of political crises, not resolved until Augustus established the imperial system of government. He took pains to preserve many republican forms, but the changes he made amounted to a transformation ofthe Roman political system, and for all practical purposes the Roman republic ceased to exist. But as in the French example, the fall of a system of government was not followed by political fragmentation. As these instances show, the collapse of empires, in the sense of political fragmentation, should not be confused with the fall of specific administrations or even the breakdown of systems of government.
All these vicissitudes of political systems are important kinds of sociocultural change, well worth study. However, we were not being arbitrary by concentrating on fragmentation. Political troubles that do not lead to fragmentation, especially troubles that lead instead to changes that facilitate greater political centralization, are often connected with expansion, increase, and/or elaboration in technology, population size, production and exchange, and the like. I have taken pains to make a distinction between political fragmentation and cultural, social, economic, or environmental troubles, but the fact remains that such troubles are often concomitant with fragmentation.
Some seminar participants stuck rather closely to the theme of political fragmentation, whereas others paid more attention to civilizational aspects. Thus the seminar turned out to be less tightly focused than we had expected. However, the rise, transformation, decline, and fall of great traditions, although less easily dealt with, are of deeper significance than the persistence or disintegration of states. It is just as well that they insisted on finding their way into our discussions.
ln Chapter 10, Kaufman sketches general kinds of troubles that occurred repeatedly in many different states. He does not describe any specific state but provides a very good synthesis of states in general. Kaufman sketches a typical trajectory. We should try to usefully disaggregate this ideal trajectory. Can we identify distinct types of states that have different kinds of troubles or, at any rate, different troubles to different degrees? Can we develop theory connecting kinds of states with kinds of troubles, or does every instance require its own particularistic explanation?
Before I turn to these questions I will outline the main recurrent kinds of trouble as they appear to me. Many of my ideas on this subject are ` ·similar to ·Kaufman's, although I put them in different words and with somewhat different emphases. ·
All institutions, and especially states, must have resources at their disposal in order to persist: goods, services, and (where it exists) money. Most resources are perishable or expended in the process of being used, so they must constantly be supplied afresh. More durable resources and facilities (such as buildings, fortifications, roads, and hydraulic works) require expenditures for their maintenance. In other words, states need enough income to do what they have to do. It is hardly surprising that this is one of the problems they often face, but I do not think we pay enough attention to how many states, how large a part of the time, find this a major problem. Occasionally, as by rapid expansion through easy conquest of rich neighbors, a state acquires income that vastly exceeds its immediate needs. But such periods are invariably brief. In Chapter 4 Culbert asks what happens when the sources of easy income dry up. Sometimes the state itself disappears, or at least fragments. But often political integration of a large region persists long after there is no more easy income from expansion, as in both the western and eastern Roman empires, many Chinese dynasties, and the Teotihuacan state. In other cases, such as the Old and Middle Kingdoms of Egypt, income from conquests was perhaps never significant. How did large states balance their costs and expenses when they could not simply capture resources from others?
In cases in which states did succeed in establishing a long-term (or potentially long-term) balance between costs and expenses, the inverse question arises. Why were they not able to do so indefinitely? The histories of empires that persisted much longer than their times of rapid growth can be read as struggles over the generation, control, and use of scarce resources, and of the consequences of these struggles. The systematic comparative study of such struggles is largely a task for the future. Here I will attempt only an outline of such an analysis.
The fundamental problem faced by states is put by Micawber’s Principle: income must be greater than or equal to expenditures. There can be accounting deficits, but the state must, somehow, maintain long-terrn access to sufficient quantities of goods and services. It is not the absolute amount of income or expenditures that matters, nor is it necessary that income exceed expenditures by very much. What is critical is that (since reserves were always finite and mostly perishable) expenditures cannot exceed income for very long without disastrous consequences. Because the processes that reduce income are different from those that increase costs, it is useful to discuss them separately.
If a rapidly expanding state depends heavily on captured income it may be that, all along, it is "living on capital,” that is, making expenditures at a rate that could not possibly be sustained for long by the conquered territories. I am not certain that there are any really clear examples of this. If there are, it seems such a state would have to find ways of quickly reducing its expenditures or else collapse.
Often, however, a large state controls territory that, in terms of environment and available technology, can produce a large income over a long time. At least five kinds of things can reduce the income available to the central authority. Kaufman has a similar list but we differ in the relative importance we assign to different factors.
The first on my list is increasing avoidance of taxes. Tax obligations imposed by a state are probably rarely equal even in theory or equally effectively enforced for everyone. However, inequalities may increase over time. This can occur through reduction of taxes or total exemption for certain institutions, individuals, or social categories – for example, religious foundations in ancient Egypt and many other places, gentry in Han China, nobility in prerevolutionary France. It can also occur through widespread fraud or simply defaulting on taxes, as may have happened increasingly in the later western Roman empire. In either case, three consequences are likely. First, the central authority has less revenue at its disposal. Second, those whose tax burden is lightened can use a higher proportion of their income for their own purposes (including transmission to heirs or successors). This strengthens them and makes them more independent of the central authority at the same time that the latter’s resources are declining. Those who are already relatively wealthy tend to be the most successful in reducing the taxes they pay, and this further increases socioeconomic differences within the state. Third, since it is extremely unlikely that the central authority will cut its costs in proportion to the reduction in the revenue obtained from some elements of the society, it is highly likely that tax burdens will be increased for the remainder, who tend to be the less wealthy. This causes further hardships for them.
One important consequence of overtaxing the less powerful is that (whether or not land or anything else is “private property” in our sense) unless the units collectively responsible for taxes are quite large (certainly larger than extended families) their capacity to weather short-term crises such as crop failures or inopportune deaths of key members will be significantly reduced. Few nonindustrialized states seem to have been effective sources of credit to the majority of the people they ruled. This means that households without reserves to tide themselves over brief but severe crises must often seek loans from local individuals or institutions with more resources, who can exploit the situation to extend their own control over resources at the expense of those who controlled fewer resources to begin with. The details of how this happens will vary according to differences in rules about the transmission and alienation of rights over resources, but I strongly suspect that the general process is nearly universal. For example, some scholars are puzzled by evidence for a significant number of households without rights to land in late pre-conquest Aztec society, in view of rules that made communal (calpulli) rights to land theoretically inalienable. I would have been puzzled by the opposite – by evidence that, in a society as stratified as that of the later Aztecs, aristocrats had failed to find a way to dispossess some of the weaker commoner households.
I realize that, in spite of references to some examples, I, like Kaufman, am sketching a generic trajectory, rather than an account of what actually happened in any specific case. Future studies should test the applicability of my model to specific instances and reject, modify, or elaborate it as seems best. For example, many states have gone through periods of reorganization, often provoked by crises, in which the procedures for obtaining revenue were reformed. These efforts were usually only partially successful. Nevertheless, they may have substantially prolonged the lives of many states. Systematic study and comparison of various reform efforts, of different methods used and different degrees of success, should be very worthwhile.
A second kind of trouble is a tendency for a higher proportion of government revenues to accrue to holders of intermediate-level offices, especially those in the provinces, and never become available to the central authority. The effect is much the same as for the problem previously discussed: the central authority is weakened while other relatively powerful elements in the state are strengthened. Furthermore, both central and regional authorities are trying to get revenue from the same population, and it is likely that there will be an increase in the overall tax burden.
The third way a state can lose income is by declining productivity of the sources of income. In principle, this could occur as a result of long-term environmental changes, but in most instances of political fragmentation there is little or no evidence that such changes were important. (They should be distinguished from brief environmental crises, in which the key point is not the existence of the crisis, but the ability of the state and society to react effectively to it.) In some instances politically motivated overexploitation seems to have led to environmental damage that seriously impaired the government’s sources of income. In Chapter 3, Yoffee describes how in Mesopotamia the necessities of empire, or what were perceived as necessities, seem to have been exceptionally destructive to regions outside the core area. Centralized governments repeatedly made demands that could be met only by damaging capacity for future production. An implication is that Mesopotamia was never amenable to Geertzian agricultural involution. Did heads of empires never understand this, did they understand it but succumb to temptations to maximize present profits at the expense of the future, or did they feel subject to necessities that left them no choice?
Politically motivated environmental damage may also have been important in the Southern Maya Lowlands, although the evidence from that region is not yet clear. Kaufman gives a good deal of emphasis to this factor. It was important in some cases, but so far I see no good evidence for it in many other instances.
A variant of this kind of trouble occurs when the damage is not to the physical environment, but to households or other productive institutions. Extraction of taxes (in the form of goods, services, or both) by the government may take so much from these institutions that they are left without the resources or the will to be as productive as before. Stated as an abstract principle, without quantification, this point seems undeniable. What is unclear – seemingly as unclear in the United States in the 1980's as in any ancient empire – is how much impact a given degree and kind of taxation on which persons and institutions, in which circumstances, will have on their productivity.
The fourth way in which income can be lost is through loss of political control over territories, either because they have been captured by outsiders or because political fragmentation is already underway. Kaufman mentions a fifth way. Lucrative foreign trade may decline, often for reasons beyond the state’s control. The collapse of the Old Assyrian state described by Yoffee in Chapter 3 is an excellent example.
By an increase in costs I mean an increase in real costs, aside from apparent increases due to inflation. One reason for such an increase is that the state may actually be required to do more. Sometimes other states or barbarians pose increasingly serious military threats, although, as Hsu points out in Chapter 7, we should never take this for granted. It needs to be demonstrated independently in each case. A second possibility is that the state needs to do more, or at least is called upon to try to do more, to offset environmental problems. These problems may themselves be of the state’s making, as I pointed out above. Third, if population increases, the state may be called upon to provide services for more people. (My view about population increase [Cowgill 1975] is not that it rarely happens, but that we should not assume that there is always a natural tendency toward increase.) Finally, modern states are sometimes required to do more because their inhabitants develop new expectations about state services. I do not know if this was a significant factor for any nonindustrialized state, but it is a possibility that should be borne in mind.
The other major kind of reason for rising real costs is less effective use of revenue due to bureaucratic proliferation and/or increasing corruption, rigidity, incompetence, extravagance, and (perhaps) inefficiency. These are often lumped, but we should think about each separately. By bureaucratic proliferation I mean an increase in the number of offices (and officeholders) required to do much the same tasks as before. The tendency of bureaucracies to proliferate is perhaps the best example of a unidirectional trend furnished by any class of social phenomena. As Parkinson (1957) points out, one of the major effects is for the officeholders to make more work for one another. Thus, they may be incorruptible, highly competent, adaptable within limits, and keenly interested in managing the state’s affairs thriftily and efficiently. That is, there may be no increase in corruption, rigidity, incompetence, extravagance, or willful inefficiency, yet less gets accomplished with a given amount of revenue. Similarly, grafters may be highly competent, resources may be used efficiently but toward extravagant ends, and so forth. Rigidity, proliferation, corruption, incompetence, and extravagance are distinct concepts. I am not sure if inefficiency should be defined as something distinct from all these or used as just a general term for their combined effects.
Bronson and Bowersock both point out that we should not assume that the prevalence of corrupt or incompetent officeholders always increases over time. This is true, but we should also not think that the existence of a few flagrant examples in the early stages of empires settles the matter. There may be increases in one or more of the factors I listed above. It is hard to get good quantitative data on amounts of corruption, inefficiency, incompetence, and extravagance. Bureaucratic proliferation should be relatively easy to quantify, because it is reflected in official tables of organization and other documents. We also need to study how all these factors work together systematically to affect adversely the ratio of income to accomplishments in specific instances.
Eisenstadt (1969) uses accountability to refer to the idea that a ruler, even if he is politically supreme, must nevertheless conform to some set of moral principles in order to maintain his legitimacy. I use the term with reference to holders of subordinate offices. Subordinates are accountable to the extent that those above them can effectively detect and penalize poor performance or nonperformance of whatever superordinates want subordinates to do. Rulers, especially “strong” rulers, usually try to make their subordinates highly accountable. Subordinates often try to reduce their accountability. Struggles between these two opposing motives loom large in the histories of nearly all states. They have much to do with the collapse of states, partly because the struggles interfere with achieving other political objectives and partly because the logical extreme of reduced accountability for provincial officials is total autonomy – that is, political fragmentation.
No study of political collapse can ignore the tension between accountability and autonomy. However, to suggest that this opposition “explains” what happens to states is roughly like "explaining” what happens in a chess game by saying that white wants to checkmate black’s king and black wants to checkmate white’s king. We have to try to understand how the struggle between accountability and autonomy gets carried out in specific circumstances. What resources and maneuvers are available to rulers, and what are available to their subordinates? The struggle over control of government income (or potential government income), discussed above, is vital, but it is not the whole story. Rules of the political system are also important – for example, whether the state is in theory absolutist, whether the rulers are accountable (in Eisenstadt’s sense) to other aristocrats, or whether (as in the United States) authority is parceled out through deliberate “checks and balances.” It is important to know whether there are other elements of society, such as wealthy merchants or numerous (but rarely wealthy) peasants, that different players can try to use as allies or can try to pit against one another. Ideology can also be important, especially if ideological differences tend to separate rulers and their supporters from holders of subordinate offices.
As I said, I cannot offer anything like a systematic analysis along the lines I have outlined. I have tried, however, to establish a basis for further discussion. I will also offer some suggestions about possible differences in the kinds of troubles most characteristic of different kinds of states.
States vary greatly in their territorial extent. The relevance of this obvious point, especially for large, nonindustrialized states, is that sheer difficulty in communication can make it much harder for a ruler to keep physically distant subordinates accountable. It is not because of the amount of information that needs to be processed, but the time it takes for messages and responses to messages to get from one place to another. It is harder ` to check up on distant subordinates, and it is also necessary to give them greater legitimate powers of discretion, because they must frequently react to local events without taking time to get instructions from the political center.
The difficulties imposed by geographic distance are part of the problems that Hsu refers to in his discussion of “regionalism." Kaufman mentions techniques to keep subordinates more accountable. These include frequent rotation from one region to another, not sending an official to a region where he has family or property, keeping close relatives of regional officeholders in the capital (where they are, in effect, hostages), and using special agents of the ruler to check on what is going on in the provinces. Different rulers have employed some or all of these techniques to varying degrees. Nevertheless, officials have often succeeded in reducing their accountability, above all by gaining lifetime tenure in a specific region, acquiring increased control over land and other regional resources, and getting control over selection of their successors (which means that they can keep their offices as a sort of family property).
Besides differences in communication difficulties, states vary in how they administer territories, as many other writers have recognized. Some territories are simply subjugated. Local rulers may be left in place if they are obedient, or they may be replaced by governors. In either case, the central authority does little on the local level except claim a proportion of taxes, impose penalties if demands (taxes, and perhaps people for military or other service) are not adequately met, suppress fighting with other regions within the state, and offer protection against foreign invaders. In other cases territories are incorporated, by which I mean that the central authority takes part in local administration and provides substantial goods and services, especially management of public works and reserves of food and other resources.
I am not entirely happy with the terms subjugation and incorporation, but other terms that have been proposed, such as indirect versus direct rule or hegemonic versus territorial empires, seem to me to have at least as many problems or not to refer to quite the same distinction.
A large state may handle some regions by subjugation and others by incorporation, or use one or the other of these forms exclusively. Subjugation and incorporation are ideal types. It is not clear to me whether most real-world instances are plainly one or the other, or whether many instances fall somewhere between these types. I also am not sure if subjugation and incorporation, as I have defined them, are most usefully seen as values of a single variable. It may well be better to disaggregate them into several variables.
At any rate, it seems that a subjugated region might easily gain its autonomy simply because local dignitaries assert (or reassert) their lack of accountability to the central authority. This might happen with little impact on the general population. In contrast, incorporated regions are more dependent on the effective functioning of the central authority. If the central authority has trouble doing what is locally expected of it (for any of the reasons outlined above), the effect is likely to be felt by the general population, as well as by regional ofhceholders. In systems talk, a state that operates by subjugation is more nearly decomposable, whereas the systemic linkages are much stronger when rule is by incorporation. A simple-minded prediction is that subjugation states should tend to be fragile, short-lived, and suffer relatively moderate social and economic consequences when they break up, whereas incorporation states will tend to last longer but suffer relatively severe social and economic troubles when they do break down. However, it is doubtful that these simple predictions fit all the evidence. Can all the short-lived empires of Mesopotamia be considered subjugation states? The Teotihuacan state was long-lived but it is unlikely that it ruled more than its core area by incorporation. Nevertheless, the contrast between subjugation and incorporation may prove useful. At any rate, it suggests one axis of differentiation within the broad and quite heterogeneous category of nonindustrialized state.
A distinction roughly similar to what I have called incorporation and subjugation has been made by many scholars. Nevertheless, this and many of the other matters I have discussed above were not emphasized during the seminar, and the preceding chapters do not systematically present data that bears on all the questions I have raised. They do, however, provide much that is highly relevant to these questions, as well as to a diversity of other important topics.
Political troubles almost surely contributed to the Maya collapse, but it was probably not an instance of political fragmentation, since Late Classic southern lowland Maya society was probably never unified politically. However, the Maya collapse may be one of the few instances of the end of a great tradition, if one regards the southern lowlands tradition as a distinct civilization rather than a variant of a more generalized Maya civilization. It was certainly the end of one major line in a family of closely related Maya traditions. Other lines survived, with more or less drastic changes, in the highlands and in the northern lowlands,‘but the demographic collapse in the southern lowlands sems severe enough to have, just by itself, terminated the southern lowlands variant of the Maya great tradition. Whatever other factors may have also worked for or against its continuance, there probably simply weren’t enough people left to sustain it any longer.
Culbert (Chapter 4) also stresses ecological instability. This deserves some elaboration. As I understand Geertz’s (1963) argument in Agricultural Involution, some environments can sustain certain techniques of agricultural intensification almost indefinitely. There is a limit to output because (without some drastic innovation in technology) a point is reached at which an excessively large increase in labor per hectare is needed to achieve a very slight increase in output per hectare. However, the situation is "stable” in the sense that there is no limit to the intensity of land use that the environment can indefinitely sustain. The point along the curve of agricultural output per hectare as a function of labor input per hectare that a given farming household will actually occupy is determined by all the complex factors (costs of labor, household labor supply, land rents, taxes, prices of crops, household demand for income, political and social pressures, etc.) that determine an acceptable, desired, or needed ratio of labor input to crop output. The only limit to intensification, however, is that, although in principle a household can intensify its labor input per hectare indefinitely by concentrating all its labor on smaller and smaller patches of ground, there is an ill-dehned limit beyond which the household (not the environment) cannot possibly sustain a further worsening of the ratio of labor input to produce output.
Some environments offer a contrasting situation. Up to a point, higher labor inputs per hectare result in higher outputs per hectare, as in the previous case. Beyond that point, however, the environment is damaged in ways that sharply reduce the amount that can be produced per hectare. When Culbert and others speak of the instability of agricultural intensification in the Maya lowlands, or fragility of the environment, I believe they refer to this kind of relation to the environment.
Culbert argues convincingly that a subsistence crisis may have contributed to the Maya collapse. However, although I strongly agree that the ultimate population reduction was very great, according to Culbert’s evidence it took a century or so for the depopulation to run its course. This implies that something other than a single nearly instantaneous demographic disaster was involved. If survivors of whatever caused cessation of the erection of monumental structures and hieroglyphic inscriptions at Tikal and other sites could hang on for a few generations, why did they not hang on indefinitely? Were they forced out by continuing problems of the environment or by political or other troubles, or did they choose to leave because the central southern lowlands had finally become less attractivé than other regions available to them?
This leads to a related problem: the duration of depopulation in the southern lowlands. In the twentieth century, for a variety of social and political reasons, population growth in the southern Maya lowlands has been very rapid, partly through migration and partly through natural increase. Persistently sparse population from the sixteenth through the nineteenth centuries can be understood as one aspect of the general Meso-american demographic disaster in the wake of the Spanish conquest. But why did the southern lowlands, especially the Petén, fill up so slowly between the tenth and the sixteenth centuries? Even a moderate amount of in-migration, plus a very moderate local rate of natural increase, would have led to more people than were there at the time of Spanish contact. It is hard to believe that, whatever environmental disaster may have occurred in the ninth and tenth centuries, it took many centuries for the land to recover to the point at which long-fallow swidden, at least, would have been again feasible over large areas. Perhaps environmental recovery was much slower than I suppose. Pertinent evidence on this point is badly needed. I suspect that a more important cause of the slowness of repopulation may have been economic and political factors that limited the appeal of the southern lowlands for Postclassic Mesoamericans. As Culbert and others have suggested, new patterns and routes of trade may have impeded recovery of the southern lowlands.
Millon (Chapter 5) distinguishes the final dramatic and violent destruction of the city from the events that made its destruction possible. Hsu makes a similar distinction between the Yellow Turban rebellion and other immediate agents or symptoms of Han political fragmentation, and the phenomena that made these troubles possible. In both cases to look behind the immediate causes does not mean that we must follow an endless chain of causality into the remote past. It is rather a matter of looking for deeper or more general processes, and if there is no absolutely given end to this effort, there are are least strategic points at which to pause.
Yoffee points out that the very practices that made it possible for rulers of some Mesopotamian city·states to build empires contributed to the collapse of those empires. Millon also suggests that certain practices that were highly useful early in the development of the Teotihuacan state later contributed to its downfall. In both cases the arguments are plausible, but the Mesopotamian empires usually fell within a few generations, whereas Teotihuacan held on nearly ten times as long, for some seven centuries. Do we have any idea what explains the difference?
Bowersock (Chapter 6) shows how skeptical we must be about popular suppositions. The eastern (Byzantine) part of the Roman empire lasted into the fifteenth century, and Bowersock argues that, at least until the Arab invasions of the seventh century, the western empire itself did not collapse or even, in any clear sense, decline. He argues further that a clearcut Roman climax, a high point that could mark the onset of decline, is itself hard to define.
Bowersock points out that different parts of a state can undergo quite different experiences. There can be clear evidence for decline or destruction of specific cities or districts at the same time that other areas are flourishing as never before. I add that this can also be true for different social sectors. The worst of times for some may be the best of times for others.
If local decline is detectable, then, by the same token, local prosperity is identifiable. The implication is not that overall tendencies in large states are unknowable, but that evidence about a particular region cannot be considered generally applicable to an entire empire.
Bowersock’s challenges to comfortable common knowledge are salutary. An "instance of collapse" about which so much is known proves to be much more complex, puzzling, and even paradoxical than the models we put forward, often with considerable confidence and satisfaction, to explain instances for which the data are far more sparse.
Nevertheless, I am reluctant to think that the western Roman empire had little in common with other instances discussed in this volume. Clearly the transition to the so-called Dark Ages was a transformation far more than a collapse of the Greco-Roman great tradition. However, political fragmentation is another matter. By the fifth century the western empire had broken up into a number of independent states. Peter Brown 1971:113-19) writes that
Two generations . . . [after about A.D. 400], the western empire had disap-peared . . . The failure of the western emperors to defend themselves against the pressure of barbarian attacks after 400, and, when attacked, to win back lost territories, can be largely explained in terms of the basic economic and social weaknesses of western society . . .And, a few pages later,
The reasons for the collapse of the imperial government, in the west, are far from simple. Questions of morale came into play, as well as economic and social factors. Perhaps the most basic reason for the failure of the imperial government, in the years between 380 and 410, was that the two main groups in the Latin world – the senatorial aristocracy and the Catholic Church – dissociated themselves from the fate of the Roman army that defended them. . Both groups. unwittingly sapped the strength of the army and of the imperial administration; and, having hamstrung their protectors, they found, somewhat to their surprise, that they could do without them . . . The disappearance of the western empire, therefore, was the price for the survival of the Senate and the Catholic Church.
The barbarian invasions did not destroy west Roman society, but they drastically altered the scale of life in the western provinces. The imperial government . . . lost so much land and taxes that it remained bankrupt to the time of its extinction in 476. The senators lost the income of their scattered estates. They were able to make good some of their losses by rack-renting and chicanery in the areas where their power was strongest . . . Communications suffered . . . In western Europe, the fifth century was a time of narrowing horizons, of the strengthening of local roots, and the consolidating of old loyalties. (Brown 1971:116)
Again, “the politics of Roman courtiers at the new barbarian courts were local politics. The idea of a united western empire was increasingly ignored” (Brown 1971:12.9).
To be sure, in the sixth century much of the west was reconquered by the eastern empire, and it remained under Byzantine control for centuries (Brown 1971: 1 3 2.-3 4). But by the seventh century, the classical tradition rapidly disappeared in the west, according to Brown (1971:176), who characterizes the change as a “simplification of culture” (Brown 1971: 174).
Brown, like Bowersock, stresses the survival, or transformation rather than collapse, of many aspects of Antique culture. Moreover, there was no single, brief, decisive crisis, but a complex series of political, social, economic, ideological, religious, and ethnic changes, lasting over centuries. Nevertheless, the political system of the Roman west did come apart. Recapture of parts of it by the eastern empire was not the same as reconstitution of a political system from within.
Questions of socioeconomic decline do not seem settled. There was certainly no simple monotonous downward trend in all aspects in all parts of the later western empire. Gunderson's (1976) evidence that in many places land prices did not fall cannot be lightly dismissed. Neither should it be taken as the last word on the subject. For example, the failure of nominally higher rates of taxation to depress land prices cannot be interpreted unless the extent of tax evasion is also considered. The government’s willingness to negotiate tax exemptions in return for lump-sum payments suggests that it had become difficult to collect all the taxes owed by large landholders. A full assessment of economic well-being must also consider commodity prices and the state of industry as well as land prices. Gunderson suggests that a surge of urbanization will accompany the diffusion new manufacturing-techiques followed naturally by a subsequent decline in urban populations as new techniques are assimilated and industry spreads to the countryside. I do not think evidence from other societies supports this idea.
My own guess is that land prices may have stayed high because, for those who could afford the prices at all, large landholdings were becoming more and more necessary for maintaining reasonably secure high status, while theoretical tax rates were becoming less related to the amounts the imperial government could actually extract from wealthy individuals.
Gunderson also argues that the labor shortage of Late Antiquity need not imply a decline in population. Evidence for a population decline is ambiguous. Much of the earlier literature on Roman population is demographically naive and unsatisfactory (e.g., Boak 1955). Other population studies (e.g., Finley 1969: 153-61) are much better. I think no one could seriously argue that population in the west increased more than slightly between the second and the fifth centuries. I suspect that there was actually a general decline in numbers, at least in central Italy, though there may have been increases in some more distant parts of the empire.
It is incongruous that questions of population remain so controversial for such a “historic" society as the Roman empire, while archaeologists often discuss prehistoric population levels and trends with great confidence. There are probably faults on both sides. Anyone who thinks seriously about problems in deriving good population estimates from archaeological data has to recognize the wide range of uncertainty in such estimates. The archaeologists who actually do the work often acknowledge these uncertainties (as Culbert does in this volume), but readers tend to lose sight of them. Even so, we may have a better idea of population trends in the southern Maya lowlands than in the Roman empire. This may be more a result of different kinds of archaeological fieldwork than of greater caution on the part of students of Classical Antiquity. Serious application of modern techniques of settlement archaeology to substantial parts of the Roman world (now underway in some places) should make important contributions to our knowledge of ancient population trends.
This leads to a final point. The abundance of historical materials for Rome is undeniably an immense advantage, yet I think that only a few Classical archaeologists are beginning to make good use of recent developments in anthropological archaeology. An example is the 1983 book by Hodges and Whitehouse, which challenges some views about Late Antiquity and exemplifies approaches that may shed a great deal of light on large questions that have not been resolved by the methods hitherto employed.
In emphasizing ideological differences between the Confucian bureaucracy and the inner court of the Han emperors, Hsu (Chapter 7) takes a strongly idealist position. During the earlier Han, the bureaucrats elaborated the doctrine that, according to Confucian principles, they were persons of a special kind, with both the right and the duty, by virtue of their training, to criticize the conduct of the government in general and of the emperor in particular. In the terms of Eisenstadt and jaspers, their efforts made China into an “Axial Age" society.
I agree that ideology made a difference, but I find it hard to say how much difference. How much longer would the Han dynasty have lasted if the bureaucracy had had no coherent ideological basis for their conflicts with the inner court? Many of the Confucian bureaucrats used their high offices in order to acquire wealth in land, and as Hsu says (p.185), at least the most influential “were able to weld successions of leadership lasting several generations and actually constituting a form of hereditary nobility.” Thus, although the high level of ideological consciousness among the literati undoubtedly exacerbated the conflict and perhaps hastened the end, could it have been more important that bureaucrats were increasingly independent of the emperors for their wealth and positions?
Hsu says little about taxation, revenues, and the funding of the Han government. In his position paper prepared before the seminar met he indicated that, during the earlier part of the Western Han, suppression of the descendants of “meritorious generals” and other quasi-feudal aristocrats left over from before the Han resulted, by the reign of Emperor Wu (141-87 B.C.), in a very high degree of centralization, with all officers of government highly accountable to the emperor and with an effective and relatively equitable tax system. Fairbank, Reiscliauer, and Craig (1973:70-81) say that thereafter the general trend (with some partial reversals) was toward greater avoidance of taxes by the wealthy (including Confucian bureaucrats), heavier tax burdens on those who could not avoid taxation, and increasingly severe problems in financing govemment operations. However, during our seminar Hsu questioned whether financial problems were especially severe for the Han government. He has subsequently emphasized (personal communication) that fiscal conditions in the reign of Emperor Wu were even worse than during the last part of the Han dynasty. Nevertheless, there was no sign of collapse at the time of Emperor Wu. This underlines the hazards of relying too much on a single source such as Fairbank, Reischauer, and Craig (1973). It also shows we should not take it for granted that (as I suggested earlier in this chapter) fiscal conditions always tend to get worse as empires get older. It seems there is no simple correlation between solvency and ef-fectiveness of centralized political control.
Bronson (Chapter 8) argues that we underestimate the role of “barbarians” in the fragmentation of states and overestimate the tendency for corruption and incompetence to get worse as empires get older. These are provocative ideas that deserve consideration, but more data are needed before either can be accepted as fully established.
Periods of political unification have been atypical and relatively brief in South Asia. Bronson suggests this may be due to vulnerability to incursions by barbarians from the north. This is a debatable conclusion. For example, the Mughals, themselves invaders from the north, imposed an unusually high degree of political unification. Their empire was short-lived, but I do not think troubles on its northern frontier had much to do with its demise. European traders and their armies played a role, but I doubt if they were decisive either. I suspect the relative infrequency and short duration of large states in South Asia is better explained by features of the South Asian environment, by a political "style” that was part of the South Asian great tradition, or by some combination of both. Bronson considers these possibilities but rejects them. I do not think they should be ruled out.
The contrast between South Asia and modern China is interesting. After the fall of the Ch’ing dynasty in the early 1900s, China experienced several decades during which effective rule was largely by regional “war-lords.” Reestablishment of a strong, centralized government was a wholly internal achievement, carried out in spite of foreign interventions and invasions. In some ways it matters profoundly that the ideology of the new government is Marxist-Leninist rather than Confucian. However, I strongly suspect that even if Marxist-Leninist ideology did not exist China would be reunified by now. In South Asia, on the other hand, a major effect of nineteenth-century British imperialism was to impose much more political unification than had existed before. Immediate fragmentation into Pakistan, India, and Sri Lanka was concomitant with British withdrawal, and a few years later east Pakistan split off as Bangladesh. Even within some of these four states, separatist tendencies (regional, ethnic, and religious) are very strong and troublesome.
In spite of whatever differences modern technology makes, I suspect these twentieth-century contrasts reflect some very enduring differences between China and South Asia. The natural environments of central and southern India are perhaps not hospitable to nonindustrialized empires. The troubles of empires (or would-be empires) in the plains of the Indus and Ganges probably had more to do with their neighbors to the south than with those in the north. Marvin Harris (1979:107- 8) argues that political and social differences between China and South Asia can largely be explained by environmental differences. Harris probably has part of the answer. However, I suspect that there is much that is not explained by either external barbarians or internal environments, that Chinese ideologies and techniques of statecraft have facilitated political centralization, whereas those of South Asia have favored smaller states and have not been so useful to persons who have tried to build or maintain empires. This is scarcely an original idea. I regret that I cannot discuss it further here.
Both Bronson and Bowersock point out that corruption and scandalous behavior can easily be found in the early days of empires. Some of the most flagrant rogues and self-seekers were in at the beginnings of empires, and blithering incompetents can be found there also. Possibly their proportions tend to increase as empires get older, but quantitative data on this point will be difficult to obtain. More significant is that a flourishing empire, especially one making huge profits through successful wars (a theme Bronson stresses), can tolerate relatively high costs due to corruption and incompetence. Still more important, and not emphasized by Bronson, is the ability of strong central authorities to intimidate, coopt, manipulate, or liquidate scoundrels, and to get rid of at least some of the worst incompetents. As middle and lower levels of bureaucracies entrench themselves, it becomes virtually impossible to weed out the lazy and incompetent. Also, as holders of higher offices acquire independent power bases and a large say in who succeeds them in office (so that recruitment is more nearly by ascription than achievement), the central authorities must find it harder to place competent and loyal persons in key positions, and harder to control action at cross-purposes to their own. It is not the existence of scandalous behavior, rule-breaking, and bungling that is crucial: it is the impact that all of these actions have.
Boundaries received much attention during the seminar, and Eisenstadt, especially, discusses them in this volume (Chapter 10). They are important, especially because boundary-maintaining mechanisms are part of what systems need to ensure their continued reproduction. However, what goes on inside and outside the boundaries is also important. When boundaries disappear or lose their significance, it is probably because differences between what is happening inside and what is happening outside have become unimportant, rather than because boundary-maintaining mechanisms per se have failed. More specifically, as an empire moves toward political fragmentation, regions become more autonomous, ‘ and it makes less difference whether a region is nominally part of the empire or not. This seems, for example, a useful way of looking at events around the western Mediterranean in Late Antiquity.
Costs of Stability
A theme throughout this chapter is that states do not simply persist of their own accord. Political systems must constantly confront problems, and resources and human skills must be employed unceasingly in maintaining the state. However, concentration on this fact makes it easy to assume that political fragmentation is always a bad thing, to be avoided if possible. There are costs as well as benefits in maintaining a political system indefinitely. Fragmentation may offer new opportunities and facilitate change. Can a large state be stabilized without inhibiting innovation and stifling cultural creativity? Excessive instability is bad, but too much stability may not be good either. We should try to provide a balanced assessment of the costs as well as the benefits of the techniques by which states, modern as well as ancient, have coped with challenges to their continued existence.
I shall mention only a few of the topics for further research suggested by the chapters in this volume. Why did some empires last so much longer than others? How closely is their duration connected to degrees and kinds of integration, economic and social as well as political? How do empires solve (or not solve) the problem of adequate income when fast wealth through easy conquest of rich neighbors is no longer possible? How do empires respond to crises? Why are some times of trouble fatal, whereas others are not? What orderly relations, if any, hold between fiscal troubles and developmental cycles of empires? Are there trends over time in the incidence of scoundrels or incompetents in governments? Is the incidence of either fiscal troubles or misbehavior really just as high early as late? lf so, are there structural reasons why the effects of such sources of trouble are sometimes less serious, or is it simply that part of the time an empire is so successful that it can tolerate a startling amount of systemic malfunctioning?
Struggles between heads and important subordinates over accountability and autonomy seem universal, but can we identify important variants? Were certain strategies in these struggles given much greater emphasis in some instances than in others? If so, are the differences explainable by differences in environments, technologies, and relations of production, or were differences in ideas about rules, techniques, and purposes of political activity also important?
Finally, what of the role of ideology in imperial expansion? Conrad and Demarest (1984) argue that ideologies gave the Incas and Aztecs decisive edges over their competitors. Ideology was also important in the explosive Islamic conquests of the seventh century. Other peoples, however, such as the Romans, were very successful empire builders on a much lower ideological plane, motivated less by a sense of mission than by quite pragmatic appetites for power and wealth. Was strong dependence on ideology one reason some empires were short-lived, since there may be little else to hold things together if ideological fervor wanes?
Underlying all these questions is the insistence that, if ideas are not merely epiphenomenal, we need greatly improved concepts, especially hypotheses that have withstood testing against substantial bodies of evidence, about causal connections between ideas and material phenomena.
Parts of this chapter somewhat resemble the position paper I wrote in advance of the seminar, but I have made many changes in response to comments by the other seminar participants, and by jane Kepp and an anonymous reader. Norm Yoffee tried to get me to prune back a number of my more florid digressions. I resisted many of these suggestions, but the chapter is better than it would have been without all the comments on earlier versions.
: Norman Yoffee; George L. Cowgill, eds., The Collapse of Ancient States and Civilizations, ch. 11., 1988 University of Arizona Press