Cities are an integral part of human experience but they emerged quite late in the history of Homo sapiens. It is therefore more extraordinary than we might care to admit that the great majority of humans are expected to be city dwellers by 2100. Greg Woolf is a lively and learned guide to ancient cities. He mainly focuses on the Mediterranean from the third millennium BC onwards, from the cities created by Bronze Age empires, such as Knossos, Mycenae and Troy, through the Etruscan and Greek city-states to the cities of the Roman Empire, the high point of which marked the only time in the history of the Mediterranean that its shores have lain under the rule of a single power.

The book is curiously constructed, with early chapters full of Dawkins-style reflections on evolution, ranging right across the globe. The point Woolf is making is that cities were not an invention of ancient Egypt or Babylonia but could be found in separate civilisations across the planet, from China and India, with their great riverine cities, to the Mississippi, where in around AD 1100 Native Americans built Cahokia. What is absent is a clear definition of the term ‘city’. In the Mediterranean, the great majority of cities, as Woolf states again and again, were small, with just a few thousand inhabitants, and yet there were hundreds of them. Such cities could also be found in the inland provinces of the Roman Empire as far afield as Britain, the Balkans and the Maghreb. ‘Megalopoleis’, enormous cities containing hundreds of thousands of inhabitants on the scale of Rome, Alexandria and Constantinople, were a rarity.

Woolf does not consider carefully enough the very varied origins of cities. A typology of cities would have helped shed light on this important subject, which was discussed by the eminent Belgian historian Henri Pirenne in the early 20th century, looking at medieval cities. Some medieval cities emerged (or re-emerged