Like the Roman dress, the Roman language was largely diffused even among those Spaniards who had not Italian burgess-rights, and the government favoured the _de facto_ Romanising of the land.[32] When Augustus died the Roman language and habits prevailed in Andalusia, Granada, Murcia, Valencia, Catalonia, Aragon; and a good part of this is to be accounted for not by colonising but by Romanising. By the ordinance of Vespasian previously mentioned the native language was restricted _de jure_ to private intercourse. That it held its ground in this, is proved by its existence at the present day; what is now confined to the mountains, which neither the Goths nor the Arabs ever occupied, must in the Roman period certainly have extended over a great part of Spain, especially the north-west. Nevertheless Romanising certainly set in very much earlier and more strongly in Spain than in Africa; monuments with native writing from the imperial period can be pointed to in Africa in fair number, hardly at all in Spain; and the Berber language at present still prevails over half of north Africa, the Iberian only in the narrow valleys of the Basques. It could not be otherwise, partly because in Spain Roman civilisation emerged much earlier and much more vigorously than in Africa, partly because the natives had not in the former as in the latter the free tribes to fall back upon. [Sidenote: The Spanish community.] The native communal constitution of the Iberians was not perceptibly to our view different from the Gallic. From the first Spain, like the Celtic country on either side of the Alps, was broken up into cantonal districts; the Vaccaei and the Cantabri were hardly in any essential respect distinguished from the Cenomani of the Transpadana and the Remi of Belgica. The fact that on the Spanish coins struck in the earlier epoch of the Roman rule it is predominantly not the towns that are named, but the cantons,--not Tarraco but the Cassetani, not Saguntum but the Arsenses--shows, still more clearly than the history of the wars of the time, that in Spain too there once subsisted larger cantonal unions. But the conquering Romans did not treat these unions everywhere in like fashion. The Transalpine cantons remained even under Roman rule political commonwealths; the Spanish were, like the Cisalpine, simply geographical conceptions. As the district of the Cenomani is nothing but a collective expression for the territories of Brixia, Bergomum, and so forth, so the Asturians consist of twenty-two politically independent communities, which to all appearance do not legally concern each other more than the towns of Brixia and Bergomum.[33] Of these communities the Tarraconensian province numbered in the Augustan age 293, in the middle of the second century 275. Here, therefore, the old canton-unions were broken up. This course was hardly determined by the consideration that the compactness of the Vettones