Both their design (Fig. 19) and their gorgeous
colouring are Celtic in spirit; they occur not seldom in Britain; on the
Continent only four instances have been recorded.[2] Here certainly
Roman Britain is more Celtic than Gallia Belgica or the Rhine Valley.
Yet a complete survey of the brooches used in Roman Britain would show a
large number of types which were equally common in Britain and on the
Continent. Exceptions are always more interesting than rules--even in
grammar. But the exceptions pass and the rules remain. The Castor ware
and the Gorgon's head are exceptions. The rule stands that the material
civilization of Britain was Roman. Except the Gorgon, every worked or
sculptured stone at Bath follows the classical conventions. Except the
Castor and New Forest pottery, all the better earthenware in use in
Britain obeys the same law. The kind that was most generally employed for
all but the meaner purposes, was not Castor but Samian or _terra
sigillata_.[3] This ware is singularly characteristic of
Roman-provincial art. As I have said above, it is copied wholesale from
Italian originals. It is purely imitative and conventional; it reveals
none of that delight in ornament, that spontaneousness in devising
decoration and in working out artistic patterns which can clearly be
traced in Late Celtic work. It is simply classical, in an inferior
degree.

[Footnote 1: Michaelis, Loeschke and others assume an early intercourse
between the Mosel basin and eastern Europe, and thereby explain both a
statue in Pergamene style which was found at Metz and appears to have
been carved there and also the Neumagen sculptures. As all these pieces
were pretty certainly produced in Roman times, the early intercourse
seems an inadequate cause. Moreover, Pergamene work, while rare in
Italy, occurs in Aquitania and Africa, and may have been popular in the
provinces.]

[Footnote 2: I have given a list in _Archaeologia Aeliana_, 1909, p.
420, to which four English and one foreign example have now to be added.
See also Curle, _Newstead_, p. 319, and R.A. Smith, _Proc. Soc. Ant.
Lond._, xxii. 61.]

[Footnote 3: I may record here a protest against the attempts made from
time to time to dispossess the term 'Samian'. Nothing better has been
suggested in its stead, and the word itself has the merit of perfect
lucidity. Of the various substitutes suggested, 'Pseudo-Arretine' is
clumsy, 'Terra Sigillata' is at least as incorrect, and 'Gaulish' cov