Call me old-fashioned. I don't mind. Call me a sentimental Leftie. I don't mind that either. I do mind, though, when I am told in a book-length study that claims to be a serious work of history that its focus is upon urban society and its manifestations as a text rather than a state of affairs. I object, too, when a debatable proposition is advanced as an unchallengeable axiom. Mayne claims that his book is a defence of the post-structural ‘linguistic left’ against the charge of historians that it has ‘muddied historical analysis by introducing a spurious disconnection between language and culture on the one hand, and the material world and practical politics on the other’ (p. 4). But is it such a defence? Am I foolish to take exception to claims which raise fundamental questions about how we know the world? Mayne, if I have understood him properly – and his language is not always easy to follow – is a proponent of a particular epistemological standpoint which underpins a theory of language as a self-subsisting social entity from which there is no referent in the extralinguistic world. Language as a significatory rather than a representational phenomenon is thus assigned a very considerable autonomy in relation to other social phenomena and one which is much greater than historians and social scientists had previously allowed. The alleged superiority of texts over sources is derived from a tradition of philosophical enquiry that stands outside and at odds with the empirical character of historical and social science.