the graffiti on potsherds or tiles that are known to me as found in towns or country-houses are equally Roman. Larger inscriptions, cut on stone, have also been found in country-houses. On the whole the general result is clear. Latin was employed freely in the towns of Britain, not only on serious occasions or by the upper classes, but by servants and work-people for the most accidental purposes. It was also used, at least by the upper classes, in the country. Plainly there did not exist in the towns that linguistic gulf between upper class and lower class which can be seen to-day in many cities of eastern Europe, where the employers speak one language and the employed another. On the other hand, it is possible that a different division existed, one which is perhaps in general rarer, but which can, or could, be paralleled in some Slavonic districts of Austria-Hungary. That is, the townsfolk of all ranks and the upper class in the country may have spoken Latin, while the peasantry may have used Celtic. No actual evidence has been discovered to prove this. We may, however, suggest that it is not, in itself, an impossible or even an improbable linguistic division of Roman Britain, even though the province did not contain any such racial differences as those of German, Pole, Ruthene and Rouman which lend so much interest to Austrian towns like Czernowitz.

[Footnote 1: Proc. Soc. Antiq. London, xxiii. 108; Eph. ix. 1290.]

[Illustration: FIG. 6. FRAGMENT OF INSCRIBED TILE FROM PLAXTOL AND RECONSTRUCTION OF THE INSCRIPTION FROM VARIOUS FRAGMENTS. (The letters were impressed by a wooden cylinder with incised lettering, which was rolled over the tile while still soft. In the reconstruction CAB in line 2 and IT in line 3 are included twice, to show the method of repetition.)]

It remains to cite the literary evidence, distinct if not abundant, as to the employment of Latin in Britain. Agricola, as is well known, encouraged the use of it, with the result (says Tacitus) that the Britons, who had hitherto hated and refused the foreign tongue, became eager to speak it fluently. About the same time Plutarch, in his tract on the cessation of oracles, mentions one Demetrius of Tarsus, grammarian, who had been teaching in Britain (A.D. 80), and mentions him as nothing at all out of the ordinary course.[1] Forty years later, Juvenal alludes casually to British lawyers taught by Gaulish schoolmasters. It is plain that by the second century Latin must have been spreading widely in the province. We need not feel puzzled about the way in which the Callevan workman of perhaps the third or fourth century learnt his Latin.

[Footnote 1: See Dessau, Hermes, xlvi. 156.]

At this point we might wish to introduce the arguments deducible from philology. We might ask whether the phonetics or the vocabulary of the later Celtic and English languages reveal any traces of the influence of Latin, as a spoken tongue, or give negative testimony to its absence. Unfortunately, the inquiry seems almost hopeless. The facts are obscure and open to dispute, and the conclusions to be drawn from them are quite uncertain. Dogmatic assertions proceeding from this or that philologist are common enough. Trustworthy results are correspondingly scarce. One instance may be cited in illustration. It has been argued that the name 'Kent' is derived from the Celtic 'Cantion', and not from the Latin 'Cantium', because, according to the rules of Vulgar Latin, 'Cantium' would have been pronounced 'Cantsium' in the fifth century, when the Saxons may be supposed to have learnt the name. That is, Celtic was spoken in Kent about 450. Yet it is doubtful whether Latin 'ti' had really come to be pronounced 'tsi' in Britain so early as A.D. 450. And it is plainly possible that the Saxons may have learnt the name long years before the reputed date of Hengist and Horsa. The Kentish coast was armed against them and the organization of the 'Saxon Shore' established about A.D. 300. Their knowledge of the place-name may be at least as old. No other difficulty seems to hinder the derivation of 'Kent' from the form 'Cantium', and the whole argument based on the name thus collapses. It is impossible here to go through the whole list of cases which have been supposed to be parallel in their origin to 'Kent', nor should I, with a scanty knowledge of the subject, be justified in such an attempt. I have selected this particular example because it has been emphasized by a recent writer.[1]

[Footnote 1: Vinogradoff, Growth of the Manor, p. 102. I am indebted to Mr. W.H. Stevenson for help in relation to these philological points.]

CHAPTER IV

ROMANIZATION IN MATERIAL CIVILIZATION

From language we pass to material civilization. Here is a far wider field of evidence, provided by buildings, private or public, their equipment and furniture, and the arts and small artistic or decorative objects. On the whole this evidence is clear and consistent. The material civilization of the province, the external fabric of its life, was Roman, in Britain as elsewhere in the west. Native elements succumbed almost wholesale to the conquering foreign influence. In regard to public buildings this is natural enough. Before the Claudian conquest the Britons can hardly have possessed large structures in stone, and the provision of them necessarily came in with the Romans. The fora, basilicas, and public baths, such as have been discovered at Silchester, Caerwent and elsewhere, follow Roman models and resemble similar buildings in other provinces. The temples show something more of a local pattern (Fig. 7), which occurs also in northern Gaul and on the Rhine, but this pattern seems merely a variation of a classical type.[1] The characteristics of the private houses are more complicated. Their ground-plans show us types which, like the temples just mentioned, recur in northern Gaul as well as Britain, but which differ even more than the temples from the similar buildings in Italy, or indeed in the Mediterranean provinces of the Empire. The houses of Italy and of the south generally were constructed to look inwards upon open impluvia, colonnaded courts and garden plots, and, as befitted a hot climate, they had few outer windows. Moreover, they could be easily built side by side so as to form, as at Pompeii, the continuous streets of a town. The houses of Britain and northern Gaul looked outwards on to the surrounding country. Their rooms were generally arranged in straight rows along a corridor or cloister. Sometimes they had only one row of rooms (Corridor House, Fig. 8); sometimes they enclosed two or three sides of a large open yard (Courtyard House, Fig. 9); a third type somewhat resembles a yard with rooms at each end of it. In any case they were singularly ill-suited to stand side by side in a town street. When we find them grouped together in a town, as at Silchester and Caerwent—the only two examples of Roman towns in Britain of which we have real knowledge—they are dotted about more like the cottages in an English village than anything that recalls a real town (Fig. 10).

[Footnote 1: British examples have been noted at Silchester and Caerwent, and in many scattered sites in rural districts. For Gaulish instances, see Léon de Vesly, Les Fana de région Normande (Rouen, 1909); for Germany, Bonner Jahrbücher, 1876, p. 57, Hettner, Drei Tempelbezirke im Trevirerlande (Trier, 1901), and Trierer Jahresberichte, iii. 49-66. The English writers who have published accounts of these structures have tended to ignore their special character.]

[Illustration: FIG. 7. GROUND-PLANS OF ROMANO-BRITISH TEMPLES. CAERWENT
AND SILCHESTER.]

[Illustration: FIG. 8. GROUND-PLAN OF A SMALL CORRIDOR HOUSE FROM
FRILFORD, BERKSHIRE.

(From plan