§ 2. In the oldest forms of Goidelic found in the ogam inscriptions, Primitive Keltic qu̯ from Aryan qu̯ remains; but in the oldest British it had already become p, and it is p in Gaulish. Traces of a Keltic qu̯ language in Gaul are seen in names like Sequani; and in some recently discovered inscriptions further evidence of the survival of such a language is believed to have been found. As the change of qu̯ to p is the earliest soundchange known which is not common to the whole branch, it seems reasonable to classify the Keltic languages as above § 1 i (7).
The more usual classification adopted in recent years is that in which the Keltic languages are grouped into "insular" and "continental". But this is a negation of all classification; it is as if we were to group together English and Icelandic as insular Germanic! Thurneysen now calls it a "geographic" classification (Gr. 1), which is equivalent to saying that it is no classification at all. It arose out of the view put forward in Rhys's LWPh.² (1879) pp. 16 ff. that the language of the ogam inscriptions in Wales is an old form of Welsh. Thurneysen, KR. (1884) pp. 7 ff., adopts this view; dismisses Rhys's later view, CB. (1884) p. 215, that the ogams are Irish; and concludes that, as the ogams have qu̯, the change qu̯ > p in British is much later than the same change in Gaulish. Of course, if the ogams are Welsh, there was no difference in the 5th cent. between Welsh and Irish, and both differed from Gaulish, which alone had p. Hence the classification into insular and continental. But the assumption on which it is based is groundless; no one now holds that the ogams are Welsh.
If it is denied that a systematic classification of the groups is possible, it would be better to take them separately than to adopt a classification which implies a close relationship between Goidelic and British. But there seems no sufficient reason for separating British from Gaulish. It is now admitted that Brit. p from qu̯ is ancient; and it is extremely improbable that this p developed independently of Gaulish p. Tacitus, Agricola xi, tells us that the speech of the Britons differed little from that of the Gauls. The Gaulish forms Πεννο-ουινδος, Vindomag(os), ambact(os), Voretovir(os) are identical with the British forms which we have to postulate as the originals of the Welsh penwyn 'white-headed', gwynfa 'paradise', amaeth 'serf', gwaredwr 'saviour'. It is for those who would separate British and Gaulish to prove that Tacitus was wrong.
For the continental qu̯ dialect or group of dialects various names have been suggested, as Sequanian (Nicholson), Pictavian, Celtican (Rhys), Ligurian (Jullian). The language of the Coligny calendar contains both qu and p; but whether the latter is secondary, or borrowed from Gaulish, or represents Aryan p, cannot yet be decided, since independent evidence as to meaning is lacking. The presence of Ar. p, if proved, would constitute these dialects a class apart.
