necessary to premise that the name of Banchorium has been omitted in the calculations, for reasons already alleged, and those of Octapitarum and Ad Vigesimum, as being foreign names. The name of Magna is omitted, partly on that account, and partly because the district in which it stands has been wholly Anglicised, and the place itself has assumed the English name of Kentchester. The general result is that, whereas in North Wales one-half of the ancient names of places are preserved, three-fourths remain in South Wales. But, if we subtract the additions of Richard of Cirencester, we have, in North Wales, only six out of fourteen names remaining in South Wales, thirteen out of seventeen. Again, of the names surviving in North Wales, the largest proportion are those of natural objects, which we should always expect to be the most permanent, and the remainder are those of towns or stations preserved in the appellations of the rivers on whose banks they stood. The most important conclusion of all is, that the names of the two races which inhabited North Wales, the Ordovices and Cangi, or Cangiani, are utterly lost, while those of the Demetæ and Silures, the inhabitants of the South, are preserved among us.

Now these considerations suggest the probability of a revolution of some kind among the inhabitants of Gwynedd, since the close, or, at all events, since the commencement, of the Roman domination in Britain. The nature or extent of such a revolution is a further question; all that can be said at present is, that it would

8 Banchorium and Deva are placed by Richard in the territory of the Carnabii.-De Situ Brit., i., c. 6, § 27.

seem to have involved a total or partial change of the population, and to have been at least so far complete, as to have obliterated a large proportion of the local names. And this probability is heightened, when we remember that we have to account for the introduction of a wholly new name into North Wales, I mean that of Gwynedd. The designation of Genania, although applied to this country, with some degree of hesitation, by Richard of Cirencester,9 can hardly be a latinized form of Gwynedd, the first two letters of which are invariably represented by V in Latin, as well by the later writers, who use the form Venedocia, and by the Romans themselves in writing other British names-as Venta for Gwent. It is also worthy of notice that, whereas Richard applies the name of Genania to a district much more extensive than any to which that of Gwynedd was ever applied, there is reason to think that Gwynedd was formerly used in a more limited sense than afterwards.

It is true that the probability does not amount to more than a presumption, and that we have to look for other evidence as well to confirm as to explain it. Such evidence is by no means wanting, although the documents on which it rests are obscure, and often contradictory. Nevertheless, there is quite enough to assure us that a change, of which it is not easy to measure either

9 "Ordovicia una cum Cangiorum Carnabiorumque regionibus, ni fama me fallit, nomine Genaniæ sub imperatoribus post Trajani principatum inclarescebat."-De Situ Brit., i., c. 6, § 25.

the extent or the degree, came over the population of Gwynedd, at some period subsequent to the commencement of the Roman dominion in Britain. The first notice we have of the event is to be found in the Triads, which, after enumerating the various races which had settled at different periods in our island, reckon among "the three invading tribes that came into the isle of Britain, and departed from it, the hosts of Ganfael Wyddel, who came to Gwynedd, and were there twenty-nine years, until they were driven into the sea by Caswallawn the son of Beli, the son of Manogan.” I call this the first notice of this event, because it is the earliest that occurs in the Triads, which are allowed to contain the earliest native authorities on ancient British history. Another Triad enumerates, among "the three dreadful pestilences of the isle of Britain, the pestilence from the carcases of the Gwyddyl, who were slain in Manuba, after they had oppressed the country of Gwynedd for twenty-nine years. It is evident that these documents relate to the same transaction, and we gather from them that North Wales, or some part of it, was under the dominion of a people called Gwyddyl, for twenty-nine years, who were finally expelled by Caswallawn, or Cassivellaunus, the opponent of Julius Cæsar. The name Gwyddel is to this day applied to the Irish, and is, etymologically, the same as Gael,3 the common name of the Irish, and Highlanders of Scotland.*

1 Trioedd Ynys Prydain. Myv. Arch., vol. ii., p. 58.

2 Ibid., p. 29.

3 The latter word is spelt Gaoidheal, the soft consonant being elided in pronounciation.

4 * It may be necessary to state distinctly the precise significations in

All that we are justified in concluding from the name is, that these occupants were a Gaelic race of some kind or other. In another Triad we meet with a curious allusion to a similar event, which must have occurred at a much later period. "The tribe of Caswallawn Law Hir put the fetters of their horses on their feet by two and two, in fighting with Serigi Wyddel, at Cerrig y Gwyddel, in Mon."5

In the Historia Britonum, attributed to Nennius, we meet with another account of the expulsion of the Gael. He informs us that Cunedda and his eight sons came from the north, from a province known as Manau Guotodin, and expelled the Scots from Gwynedd, Dyfed, which the terms "Celtic," "Gaelic," &c., are used; especially as some confusion exists in people's minds on the subject. The common name of Celtic is applied to all and each of the members of a family of nations, distinguished by certain phenomena of language and organization. This is the ethnological use of the term, and is the result of a generalization from existing facts. It must carefully be distinguished from the historical use of the term, as applied to a race whom the Greeks and Romans found in various parts of western Europe. Whether the historical Celts were Celtic in our use of the word, i. e., whether they possessed the distinctive marks of language and organization, it is one of the problems of ethnology to determine. Now this Celtic family is found to divide itself into two branches, one of which, at present occupying the Highlands, Hebrides, Man, and a great part of Ireland, in a tolerably pure state, is called Gaelic. The other, in possession of Wales, Cornwall, and Brittany, is here, as elsewhere, for convenience, denominated Cymraic. A closer connexion is found to subsist between the Bretons and Cornish, than between either of those people and the Welsh. These facts are stated here, to avoid needless verbal discussion; although they must be familiar to the majority of my readers. Those who wish to see the subject of Celtic ethnology clearly drawn out, will do well to read Dr. Prichard's "Essay on the Eastern origin of the Celtic Languages;" and a memoir, by M. Adolphe Pictet, "De l'affinité des Langues Celtiques avec le Sanscrit.”

5 Myv. Arch., p. 62.

and from the districts of Gower and Kidwelly. Their expulsion is placed about the close of the fourth century, and, although the date of their immigration is not stated, we are left to infer that it was synchronical with the occupation of Dalriada and Man by their countrymen." To the testimony of Nennius we may add that of Rhyddmarch, the author of the life of St. David, as a writer whose date we are able to fix. He speaks of the Saint being persecuted, in his hallowed retreat at Menevia, by a certain Scottish tyrant, by name Boia, who had built himself a strong castle, overlooking the Rosy Vale, in which St. David had establised himself with his companions. The name of this regulus is preserved in Clegyr Foia, a precipitous volcanic rock, surmounted by

8

6 "Novissime venit Damhoctor, et ibi habitavit cum omni genere suo usque hodie in Brittanniam. Istorith, Istorini filius, tenuit Dalrieta cum suis; Builc autem cum suis tenuit Euboniam insulam, et alias circiter; filii autem Liethan obtinuerunt in regione Demetorum et in aliis regionibus, id est, Guir et Cetgueli, donec expulsi sunt a Cuneda et a filiis ejus ab omnibus Brittanicis regionibus."-Hist. Brit., § 14. "Mailcunus magnus rex apud Brittones regnabat, id est, in regione Guenedotæ, quia atavus illius, id est, Cunedag, cum filiis suis, quorum numerus octo erat, venerat prius de parte sinistrali, id est, de regione quæ vocatur Manau Guotodin, centum quadraginta sex annis antequam Mailcum regnaret, et Scottos cum ingentissima clade expulerunt ab istis regionibus, et nusquam reversi sunt ad habitandum."-Ibid., § 62.

7 Nennius, on the authority of the "peritissimi Scottorum," places the migration of the Scots from Ireland to Dalriada, in the sixth century B.C., that is to say, in the present case, in a period anterior to history. Mr. Skene, in his ingenious Essay on the Highlanders, dates the last occupation of Dalriada, A.D. 503, and appears to consider the earlier migrations as fabulous.—Vol. i., pp. 15–20.

8 Ricemarus in Vita Sti Davidis apud Whart. Angl. Sacr. II. Giraldus omits the words "Scottus quidam," which are supplied by Wharton in the margin. Rhyddmarch lived in the eleventh century.