148 NOTES ON ANCIENT DEFENSIVE EARTHWORKS.
soldiers probably garrisoned soma of them for short periods
before the district had thoroughly accepted Roman rule, or
during local disturbances afterwards. Though the Romans
were by no means indifferent to strong natural positions for
their camps, they did not value them as places of refuge, but
as means of keeping order in a district ; as centres for repelling
pirates from their shores, not as places of security from their
attacks. Positions like those of Reculver, Richborough,
Chichester or Colchester were strong enough for them, if
suitable in other ways. Of course, where a primitive hill-fort
was advantageously situated, from the Roman point of view, it
was utilized, as in the cases of Dover and Old Sarum : but they
were generally ill-placed for Roman purposes. Though the
Romans occupied permanently the site of Dover Castle, their
nearest station westward was not where the so-called "Caesar's
Camp" stands on the lofty Chalk escarpment north of Folke-
stone, but on low ground close to what was then sea at Lymne,
west of Hythe.
On crossing the Thames into Essex we find ourselves in a
district in which there are no strong sites for hill-forts comparable
to those on the Chalk in Kent and Sussex. For the surface of
Norfolk, Suffolk and Essex consists almost entirely (as I have
already remarked) of strata much softer and less coherent than
the Chalk. Such comparatively high ground as exists has a
much gentler slope ; and the impression of flatness given by the
Eastern Counties arises from the general absence of distinct hills,
not from the presence of an unusual expanse of ground but
slightly above the level of the sea. Except in parts of Southern
Essex, the Chalk and the overlying Tertiary formations are
almost everywhere covered by the Boulder Clay and by other
superficial beds, and most of the present surface consists of a
plateau which occasionally rises to a height of more than 300 feet
above the sea, but is usually below that level, the Chalk and the
Tertiary beds appearing chiefly in the valleys of the various
streams intersecting the flat-topped, drift-capped plateau. Such
being the nature of the country, it is evident that in order to
make a thoroughly eligible site therein for an ancient defensive
work, it was usually necessary to elevate it artificially, ,so as to
secure some advantage in height over the highest adjacent
ground. Of course, a Norman architect would feel no need for
an artificial mound for the site of his keep, inasmuch as that
structure itself would give the required elevation. And he would