The great potential of comparative history for understanding individual instances of history has so far been imperfectly realized. This is largely because historians do not know how to do comparative history. In comparing two phenomena, they have started with one, and asked whether the other is a second instance of it. The familiar Model A, in all its fulness of familiar detail, is applied to a new Test Case B. Since all events are at some point unique, Test Case B eventually fails to correspond. A proposed case of feudalism, for example, will fail to match the prototype (typically, the social and legal structures of mediaeval France - and for that matter, why not mediaeval Germany?) because it did not have subinfeudation, or because the people did not make chansons at Courts of Love, or tilt with lances on fields of honor, or something such. This is extremely silly, but it is also extremely common.
The fallacy involves several separate errors. It may help to list them.(1) One error is assuming that the familiar instance is also the normal instance, and that it can validly be enshrined at the outset of the investigation as a touchstone. It is quite possible that the European experience (the usual source of these prototypes) is atypical in human history, and that some other strand of development represents the most likely, or most natural, or most fully realized, course of events. This we will never know unless we pick our models in the first place from a wider range of data, and assess their common elements more judiciously. Arthur Waley noted that China sometimes presents in fairly complete form phenomena which are missing or distorted in the European case. Edwin Pulleyblank made a similar point in his Cambridge inaugural lecture. It is not to the credit of Sinology, or of history at large, that it has failed to follow up these hints.
(2) Another error is the failure to isolate significant elements from the totality of each situation being considered; the failure to generalize. Given that no historical instance is ever fully replicated, the only question worth asking is whether certain elements or aspects of those instances recur, and if so, whether their recurrence has any analytical interest. This question cannot be fruitfully asked until the significant elements have been identified. Those elements cannot be accurately identified unless more than one instance is examined. And the right way to examine those instances is to compare them ab initio. Starting with one instance as the standard, even without predispositions concerning that instance, leads nowhere.
(3) Another error is failing to allow for different contexts, especially of cases separated in real time. The Merovingian feudal arrangements had contractual elements. The Jou feudal ones did not. It helps to notice that the law of contract had not yet arisen during the Jou period, whereas in mediaeval western Europe (which is probably a freak case for this reason), we are observing a rudimentary political form as it takes shape in the context of an already developed legal form. The question is not which is the true instance of whatever it is they are both instances of, but each is shaped by the time and technology frame in which it emerges.
(4) A kindred error is failing to allow for differences of scale. The same structure may not articulate the same way if applied to a territory twice as large. It may not articulate at all if the territory is ten times as large (the Mencians were already aware of this in the 03rd century). You cannot make a little elephant, or a large mouse; those designs only work with a certain relationship between size and gravity. We are wrong if we expect them to do otherwise. We are wise if we set ourselves to discover the limits of scale within which a given tendency can realize in a certain way. Given the lack of a historical precedent, the founding fathers of the American republic asked whether a "republic" on a large scale could even exist. That was a sensible question. Comparative history has lost much sophistication since 1776.
(5) An especially stultifying error is to import developmental schemas into the analysis of particular cases. Separate moments of history have to be independently described before there is any chance of building a theory which is based on, and at some level of abstraction comprehends, all those moments. The initial focus has to be descriptive, not evolutionary. Only at a later stage may an evolutionary tendency validly emerge. The evolutionary fallacy usually takes the form of the assertion (not always openly made) that situation X always gives rise to situation Y - and if a given X did not do so, then it was not truly an X after all. Historians need to get comfortable with the idea of incipient, aborted, and exceptional developments. Not all multi-state systems are a chemical mix whose normal final condition is a unified state. Not all historical "reactions" go to what a chemist would call "completion." The Chinese states ended up unified. So did the Indian ones. The Greek ones did not. Why not? Now we perhaps have an askable question.
(6) And as if the above traps are not enough, there is always cultural chauvinism. "That alien X is not like my Y because my Y is unique and incomparable." Or PC delicacy: "We know only ourselves, and we must never make judgements about any other culture, especially judgements of inadequacy or incompleteness, and must instead assert that all cultures are equally complete and equally satisfactory." Good Grief.
The comparative question, in its F-word form, was once posed with great clarity, and the instigator was no historian but a physicist. This was J Robert Oppenheimer, then the Director of the Institute for Advanced Study. That question, as is recounted in the Foreword to Rushton Coulborn (ed): Feudalism in History, led to the creation of an ACLS Committee on Uniformities in History (excellent name), and to the conference reported its findings in the Coulborn book (not so excellent). The instinct was sound: to survey cases at large and ask what it is that recurs in different times and places. The execution was pathetic. The essays in the volume do not attempt to isolate recurring elements. They revert to the idea of one sample being the touchstone for the others. Another fallacy of many of those essays was the intrusion of a developmental criterion. The sequence fallacy would argue that feudalism is not truly feudal unless it leads to the next stage in some imagined schema of necessary historical development. The superstition that history passes through predetermined stages is common to ancient Greece and ancient China, and that superstition is thus itself attractive material for comparative study. But it needs to be banished from the presumptions of the serious historian.
Obviously, history is not getting anywhere under its present management. We therefore have a modest proposal. It is that history, with its sister science, philology; and its associated science, archaeology, should be relocated in the Department of Physics, where it can usefully maintain contact with the essential skills of astronomy, botany, chemistry, engineering, and statistics; that it should be given lab coat and calipers, and put to work under the chastening guidance of some latter-day Oppenheimer.
Diary 26 March 2001: We have just attended an AAS panel where a bright young scholar proved that Jou China was not feudal - by showing that it was not Merovingian.
We have already acknowledged that comparative history is dead. But we deplore having its corpse dug up and publicly dishonored in this way. We give fair warning: The next person to commit this particular outrage within our knowledge will be mentioned by name in this space.
18 Jan 2006 / Contact The Project / Exit to Comparative History Page