Time and again, I have noticed a strange and stubborn insistence on assigning primacy to the ‘primary sources’ for the reconstruction of the past.