Genocidal violence leaves but a broken historical trace. The surviving records of Partition are marked by their fragmentariness. They move, in fits and starts, through jerks and breaks and silences – incoherent, stuttering, even incomprehensible – between the poles of testimony and rumour.

Testimony, Langer notes, is ‘a form of remembering’. Rumour, by contrast, is a form of doing – of making happen – by telling. The record of Partition clearly bears the mark of both. The importance of first-person testimony (for the judge, as for the historian) requires no underlining. ‘I was there …’; ‘I saw …’; ‘I can name …’; ‘I recognise …’; and (more than occasionally for the historian and the journalist, though perhaps less commonly for the judge) ‘I learnt from the most reliable witnesses …’ Testimony's method is that of particularising and individualising, specifying sites and bodies that carry the marks of particular events, making real in everyday, physical, nameable terms. Its difficulty in the limit case is that it needs to articulate an unparalleled, unthinkable history struggling to find a voice. How does the witness share ‘the particularity, the unshareability, and the incommunicability of pain in torture’? How can we speak for the dead, who are no longer present? How can we testify on behalf of the dead, if we are not dead? How can anyone who is not a Muselmann know what it is to be a Muselmann, as historians of the Holocaust have repeatedly said?