though imperfectly, the facts which our legal and philological arguments
do not yield.

I need not here insert a sketch of Roman Britain. But I may call
attention to three of its features which are not seldom overlooked. In
the first place, it is necessary to distinguish the two halves of the
province, the one the northern and western uplands occupied only by
troops, and the other the eastern and southern lowlands which contained
nothing but purely civilian life.[1] The two are marked off, not in law
but in practical fact, almost as fully as if one had been _domi_ and the
other _militiae_. We shall not seek for traces of Romanization in the
military area. There neither towns existed nor villas. Northwards, no
town or country-house has been found beyond the neighbourhood of
Aldborough (Isurium), some fifteen miles north-west of York. Westwards,
on the Welsh frontier, the most advanced town was at Wroxeter
(Viroconium), near Shrewsbury, and the furthest country-house an
isolated dwelling at Llantwit, in Glamorgan.[2] In the south-west the
last house was near Lyme Regis, the last town at Exeter.[3] These are
the limits of the Romanized area. Outside of them, the population cannot
have acquired much Roman character, nor can it have been numerous enough
to form more than a subsidiary factor in our problem. But within these
limits were towns and villages and country-houses and farms, a large
population, and a developed and orderly life.

[Footnote 1: For further details see the Victoria County Histories of
_Northamptonshire_, i. 159, and _Derbyshire_, i. 191. To save frequent
references, I may say here that much of the evidence for the following
paragraphs is to be found in my articles on Romano-British remains
printed in the volumes of this History. I am indebted to its publishers
for leave to reproduce several illustrations from its pages. For others