It never helps historians to say too much about their working methods. For just as the conjuror's magic disappears if the audience knows how the trick is done, so the credibility of scholars can be sharply diminished if readers learn everything about how exactly their books came to be written. Only too often, such revelations dispel the impression of fluent, confident omniscience; instead, they suggest that histories are concocted by error-prone human beings who patch together the results of incomplete research in order to construct an account whose rhetorical power will, they hope, compensate for gaps in the argument and deficiencies in the evidence.

LRB 10 June 2010 | PDF Download