The growing trend for historians to rely on oral evidence is not without its problems. It is naive to suppose that someone's testimony represents a pure distillation of past experience, for in an interview each party is affected by the other. It is the historian who selects the informant and indicates the area of interest; and even if he or she asks no questions and merely listens, the presence of an outsider affects the atmosphere in which the informant recalls the past and talks about it.

The end product is conditioned both by the historian's social position vis-å-vis the informant, and by the terms in which he or she has learnt to analyse the past and which may well be communicated to the informant. In other words, historians must accept responsibility for their share in creating new evidence.

But the difficulties are far from over when the historian is removed from the scene.

For not even the informant is in direct touch with the past. His or her memories may be contaminated by what has been absorbed from other sources (especially the media); they may be overlaid by nostalgia ('times were good then') or distorted by a sense of grievance about deprivation in childhood, which only took root in later life. To anyone listening, the feelings and attitudes are often what lends conviction to the testimony, yet they may be the emotional residue of later events rather than the period in question.