Thompson’s own affirmation of the irreducible, independent reality of historical evidence, and of the various ways in which it can be interrogated, is in general a model of good sense. Some of the distinctions he draws—as between ‘value-bearing’ and ‘value-free’, or ‘lateral’ and ‘structural’, types of evidence—are perhaps less clear-cut than he suggests. But few writers, or reflective readers, of history would dissent from his description of the ‘historian’s workshop’ here. The difficulties really begin on the other side of his enumeration of the different kinds of questionnaire that can be employed in looking at primary evidence. This is sharply brought home when Thompson recommends the ‘reality rule’ of J.H. Hexter, that the historian should seek out ‘the most likely story that can be sustained by the relevant existing evidence’, as ‘helpful’—only to have to regret immediately afterwards that ‘it has been put to work by its author in increasingly unhelpful ways, in support of a prior assumption that any Marxist story must be unlikely’.9 But, of course, the banality of the formula is precisely the guarantee of its disutility: who is to determine what is relevant, or for that matter what constitutes a story? We are immediately referred back to the thornier problem of historical concepts. Thompson does not attempt to expound or justify the specific set of categories that defines historical materialism—an abstention with important consequences later in his essay. He suggests in passing, with perfect propriety, that ‘there are other legitimate ways of interrogating the evidence’10 than those which have formed the major patterns of inquiry for Marxist historians. Rather than dwell on the particular canons and procedures typical of Marxist historiography, he emphasizes the common ‘test of historical logic’11 to which they along with all others must submit. In a fine paragraph, he then represents the general verdict of the discipline thus: ‘The court has been sitting in judgement upon historical materialism for one hundred years, and it is continually being adjourned. The adjournment is in effect a tribute to the robustness of the tradition; in that long interval the cases against a hundred other interpretive systems have been upheld, and the culprits have disappeared “downstairs”. That the court has not yet found decisively in favour of historical materialism is not only because of the ideological parti pris of certain of the judges (although there is plenty of that) but also because of the provisional nature of the explanatory concepts, the actual silences (or absent mediations) within them, the primitive and unreconstructed character of some of the categories, and the inconclusive determinacy of the evidence.’12
The forms of appeal that the court of the historical discipline allows are dual: ‘evidential’ and ‘theoretical’. Evidence, as Thompson notes, he has already sufficiently discussed. What of theory? Here appeal must be to ‘the coherence, adequacy and consistency of the concepts, and to their congruence with the knowledge of adjacent disciplines’.13 Wherein, then, does the force or fallibility of Marxist historical concepts lie? Thompson does not address himself directly to this issue. Instead, he poses a wider question: what is the distinctive nature of historical concepts in general—Marxist or non-Marxist? His answer is that they are ‘expectations rather than rules’, for they possess a ‘particular flexibility’, ‘necessary generality and elasticity’, a ‘coefficient of mobility’14 due to to the quicksilver nature of the historical process itself. The ‘categories change as the object changes’.15 Once this is understood, it can be seen that while historical materialism is distinguished ‘by its stubborn consistency (alas, a stubbornness which has sometimes been doctrinaire) in elaborating such categories, and by its articulation of these within a conceptual totality’,16 for similar reasons it is also perpetually imperilled to a greater degree than non-Marxist historiography by the danger of a rigid and static conceptualization that is radically inappropriate to historical eventuation. ‘It is the misfortune of Marxist historians (it is certainly our special misfortune today) that certain of our concepts are common currency in a wider intellectual universe, are adopted in other disciplines, which impose their own logic upon them and reduce them to static, a historical categories. No historical category has been more misunderstood, tormented, transfixed, and dehistoricized than the category of social class … It is not, and never has been, the business of history to make up this kind of inelastic category.’17
Here, however, Thompson is under a misapprehension. His argument in effect amounts to a claim for a legitimate laxity of notions that would be the peculiar privilege of the historian. But the nature of the historical process warrants no such special licence. The fact that its object continually changes no more relieves the discipline of history of the duty of formulating clear and exact concepts for its comprehension than it does meteorology—a physical science whose data notoriously change rather more swiftly and mercurially than those of history itself. If the weather remains largely unpredictable (and uncontrollable), the meteorologist does not resign himself to professions of the inherent approximation of his study: he seeks to push back the limits of our knowledge by further scientific investigation, which will involve not less but more conceptualization, of wider ranges of evidence. So it is in every other science. History is no exception. Brecht once remarked that if human behaviour appears unpredictable, it is not because there are no determinations, but because there are too many.18 The historian’s necessary duty of attention to the particular event or the concrete custom is not to be discharged by bending or stretching general concepts around them. It can only be acquitted by reconstructing the complex manifold of their actual determinations, which will always demand further—more rigorous—conceptualization. Thompson tends to see concepts as models or diagrams of a reality that never quite behaves itself, in an alternation of the ‘abstract’ and the ‘particular’ which forgets this central injunction of Marx: ‘The concrete is concrete because it is a synthesis of many determinations, hence unity of the diverse … the abstract determinations lead towards a reproduction of the concrete by way of thought’.19 If categories are to be categories at all, they demand precise and unequivocal definition. To capture the processes of change which indeed characterize history, historical concepts have to be very carefully formulated and specified: but they will only be concepts if they fix some structure of invariance, however much internal variation such a structure may allow—in other words, however wide its morphology. Does this condition of intellectual cogency preclude an adequate grasp of any diachronic history? In no way. On the contrary, far from being especially liable to a schedule of unduly static concepts, as Thompson contends, Marxism preeminently possesses concepts that both theorize the possibilities and limits of historical change as such (contradiction), and explore the dynamic of particular processes of development themselves (the laws of motion of capital). Its repertory remains, of course, partial and provisional—in a sense mere overtures to the composition of a plenary history. The absences and insufficiencies of its explanatory instrumentarium to date are not in doubt: Althusser emphasizes them as much as Thompson. But they are reason, not for retreating from theoretical endeavour, but for advancing towards fuller analysis. In other