‘Our own time demands an unblinking encounter with civil war,’ writes David Armitage, a professor of history at Harvard. ‘The three hundred years between 1648 and 1945 constituted an era of war between states; the last sixty years appear to be an age of war within states. Indeed, this is the most striking change in patterns of human conflict for centuries … Since 1989, barely 5 per cent of the world’s wars have taken place between states.’

Armitage’s goal, in this wide-ranging and informative book, is to examine the history of the idea of civil war as it has developed from the Romans until the present time. His motives for engaging in this conceptual investigation are not simply analytical. He also wants to challenge the Roman view that civil war, which the Romans regarded with horror, is a mode of human conflict that cannot be eliminated: ‘We should be cautious about assuming civil war is an inevitable part of our makeup – a feature, not a bug, in the software that makes us human. For that would be to doom us to suffer civil war ad infinitum, never to reach Kant’s promise of perpetual peace.’ The method Armitage adopts to undermine this distressing conclusion is a variety of genealogical inquiry, first clearly adumbrated by Nietzsche, in which concepts are not assumed to have any fixed or definitive core but are recognised as being continually changing in their meanings.

Conceptual genealogy of this kind has, of course, been widely deployed by scholars: Michel Foucault, for example, claimed that ‘sexuality’ is a historically specific construction rather than a description of immutable human behaviour. Armitage adds to the genealogical method the claim, made by the philosopher W B Gallie in