ABSTRACT
Ultimately the principles governing the direction of original research can be reduced to two. According to the rst, the historian takes one source or group of sources that falls within his or her general area of interest – say the records of a particular court or a body of diplomatic correspondence – and extracts whatever is of value, allowing the content of the source to determine the nature of the enquiry. Recalling his rst experience of the French Revolutionary archives, Richard Cobb describes the delights offered by a source-oriented approach:
The second, or problem-oriented, approach is the exact opposite. A specic historical question is formulated, usually prompted by a reading of the secondary authorities, and the relevant primary sources are then studied; the bearing that these sources may have on other issues is ignored, the researcher proceeding as directly as possible to the point where he or she can present some conclusions. Each method encounters snags. The source-oriented approach, although appropriate for a newly discovered source,
may yield only an incoherent jumble of data. The problemoriented approach sounds like common sense and probably corresponds to most people’s idea of research. But it is often difcult to tell in advance what sources are relevant. As will be shown later, the most improbable sources are sometimes found to be illuminating, while the obvious ones may lead the historian into too close an identication with the concerns of the organization that produced them. Moreover, for any topic in Western nineteenth or twentieth-century history, however circumscribed by time or place, the sources are so unwieldy that further selection can hardly be avoided, and with it the risk of leaving vital evidence untouched. In practice neither of these approaches is usually pursued to the complete exclusion of the other, but the balance struck between them varies a good deal. Some historians begin their careers with a narrowly dened project based on a limited range of sources; others are let loose on a major archive with only the vaguest of briefs. The former is on the whole the more common, because of the pressure to produce quick results that is imposed by the Ph.D. degree – the formal apprenticeship served by most academic historians. A great deal of research – probably the larger part – consists not in ferreting out new sources but in turning to well-known materials with new questions in mind. Yet too single-minded a preoccupation with a narrow set of issues may lead to evidence being taken out of context and misinterpreted – ‘source-mining’ as one critic has called it.3 It is vital, therefore, that the relationship between the historian and his or her sources is one of give and take. Many historians have had the experience of setting out with one set of questions, only to nd that the sources which they had supposed would furnish the answers instead directed their research on to quite a different path. Emmanuel Le Roy Ladurie rst turned to the land-tax registers of rural Languedoc with a view to documenting the birth of capitalism in that region; he found himself instead investigating its social structure in the broadest sense, and in particular the impact of demographic change:
At the very least there must be a readiness to modify the original objective in the light of the questions that arise directly from the sources. Without this exibility historians risk imposing on their evidence and failing to tap its full potential. The true master of the craft is someone whose sense of what questions can protably be asked has been sharpened by a lifetime’s exposure to the sources in all their variety. Mastery of all the sources must remain the ideal, however improbable its complete accomplishment may be.
