For Britain the fifth century was traumatic: the withdrawal of the last of the Roman garrison in the first decade of the century was followed by increased raiding by invaders who established settlements: Irish in the west, Angles and Saxons in the east. But not all of Britain had been part of the empire and not all Roman Britain had been governed or Romanised with the same intensity. The populations living in the non-Roman, or non-Romanised, areas could well have been larger than those of the Romanised parts: mountainous areas can support surprisingly large populations. In the early 19th century half of Scotland’s population was found in the Highlands. One arresting possibility is that much of the landscape in AD500 could have appeared as it did centuries later or even, to an extent, today. It is common ground among experts that most of the island was by no means as God had left it: the woodland cover had been largely cleared before the Romans came, often long before. But, more surprisingly, there is a good case for supposing that over wide areas the pattern of fields, minor roads and even boundaries was pre-Roman. The principal supporting argument is the relationship between ‘co-axial’ landscapes and Roman roads. In these widespread landscapes, the field-boundaries and the lines of minor roads are apparently determined by sets of approximately parallel lines, all lying in the same direction. It is as if areas covering many square miles had been laid out at some remote period, each at more or less a single go. These landscapes seem to antedate major Roman roads, which swing across them, doing little to determine their arrangement. Additional evidence for the detail of the countryside going back at least into the Roman period comes, for example, from Wiltshire: on the Marlborough Downs parish boundaries (very probably following ancient property boundaries) disregard the Dark Age Wansdyke, though this is still a great feature in the landscape.