14. Y gweinon cleiuon: pl. in both elements of the compound (not given in dictionaries) gwannglaf, which occurs in Buchedd Beuno Sant in this same MS. vol. Af i edrych fyn Tat y sydd yn wannglaf (Cambro-Brit. Saints, p. 14).

Kyd-doluryo, i.e, cyd-ddoluryo. Not in the dictionaries.

Dothef. A shorter form (unrecorded) of dyoddef.

14. Ymaythu ami. Ymaythu is not in the dictionaries, but it seems to be the infinitive of a verb related to ymaith, which is now used only as an adverb away, hence. Ymaith itself was also probably originally an infinitive. It is the Ir. imeacht ("s.f. walking, going".-O'Reilly), just as ymdaith is the Ir. imtheacht ("s.f. progress, departure", etc.-O'R.), and taith the Ir. teacht (do theacht, to come). Ymaythu ami, then, would be nearly the same as ymadael a mi.

Dros aunaythoch. Dros="in return for", a meaning not instanced by Zeuss. So on the next page (15), diolch ytha6 dros yua6r rod, etc.; and (152) dros vyygheredicrбyd.

15. Ymhoylyd (to turn one's self) is the colloquial form of ymchwelyd or ymchoelyd. But though a reflexive, this verb is commonly used as a simple transitive verb (see exx. in Pugh3). Ymhoylyd, 'mhoylu, are commonly used in Dimetian for "turning over": e.g., 'mhoylu teisen, to turn over a cake; 'mhoylu gwair, llafur, to turn over hay, corn, in harvesting. In Carmarthenshire it is also used of "ploughing": "Mhoylwr go lew w' i o grwt" ("I am a pretty good ploughman for a lad"), the writer once heard a Carmarthenshire youth modestly remark. 6rth yderuynnu (in putting him to death). not noticed in the dictionaries.

This use of terfynu (to end) is

Aatkyuorant. Atcyforio or adgyforio is an unregistered compound of ad (again, re-) and cyforio (to fill to the brim, to make to overflow). Cy-for seems to be from cy and mor, i.e., marg. the root of L. margo, etc. (Fick,3 iv. 187), and so would mean even with the brim".

Aymdinuant, if correct, is a form which the writer does not understand. Agherediyyon. Angharedig means, passively, "unloved"; and, actively, "unloving, unkind".

15. Ychwithe bod ynwell gennбch. A somewhat unusual construction, apparently an imitation of the Lat. Historic Infinitive so-called.

Gwaharthon, pl. of gwahardd (prohibition) of which Pugh3 gives only the pl. gwaharddoedd.

Ymroyssoch ych bod yn arglбythi. The use of the possessive pron. here before bod has some slight analogy to the peculiar Irish construction which uses the poss. pron. with the predicative noun in such sentences as, "He is a good man"-Ir. Tá se 'nn a dhuine mhaith (lit. he is in his good man).

16. Iaбnhau. An unregistered form=iawni or iawnu, to render right. Gwerthe, i.e., gwerthau, a pl. of gwerth, of which the dictionaries give no example.

Ll6e, i.e. llwau, pl. of llw; usual pl. Uwon.

162. Offryngau. The common forms are offrwm, offrymau. Kyflebaythe is doubtless the same as cyffelybiaethau (similitudes). dictionaries do not notice the present meaning-" examples”.

The

Trбy ybu haбs. Perhaps it should be trwy y rei y bu, etc.

17. Ymgalein (if the right form, the last four letters are indistinct in MS.) is doubtless the same as ymganlyn, to follow mutually.

Ambechr6yth is not in the dictionaries. It is a compound of am and pech, and may here bear its strict meaning-" mutual sinfulness".

Y benthic kyuan (the capital), lit. the entire loan.

Trбy li6 116, etc. (by means of oath, etc.); Cf. liw dydd, liw nos (by day, by night); Arm. liou ("licence, permission, congé”).

172. Setwyr. Qy. a corruption of "executor"?

6rth ymado, etc. It seems that something has dropped out here.

6rth gardode (in alms). None of the extracts in Zeuss under wrth (p. 682) exactly illustrate this meaning.

D6yllodrayth (deceit). An unrecorded abstract noun corresponding to the personal noun twyllawdr (deceiver) and the adj. twyllodrus (deceitful).

Ychwyhunein. Zeuss gives no example of this form of the second pers. plur., though he has ny hunein and ehunein (Gr. Celt. 2, p. 408).

18. Ymgeuethli6. The forms of this verb given in Pugh3 are ymgyfethl, to be striving together; and ymgyfethlu, to struggle mutually. Neither of these meanings suits this passage, which demands rather-" upbraid, expostulate with".

Rodi gwad (give denial). Welsh constantly uses the root form of the verb in this way as a subst., and especially after ar, exactly corresponding to the English forms with a-(shortened form of an=on), as a-fishing, etc. So we had (p. 122) ar duth (fr. tuthio, to trot), a-trot. Similarly, ar redeg, a-running; ar wib (fr. gwibio, to rove) e.g., myn'd heibio ar wib, to pass a-flying, on a flying visit; ar dan, ar daen (fr. tánu, tacnu, to spread), e.g., y mae y gwair ar dán, the hay is scattered, i.e, not in cocks or mows; y mae e' ar nydd (fr. nyddu, to twist, spin) i gyd, it is all a-twist ; ar dro, awry: ar dor (fr. tori, to cut), eg. (a provision dealer says to a customer), y mae gen 'i gosyn da ar dor ynawr, I have a good cheese a-cutting, in the course of being cut now.

182. Poenedig; here active in force.

Ochuanu. Also an unregistered form of the same force as ochfaü, found on

P. 42.

Oir, i.e., ohir, fr. qohir, delay.

THE CELTIC LANGUAGES IN RELATION

TO OTHER ARYAN TONGUES.

BY REV. JOHN DAVIES.

As the paper on the Celtic languages, published in the last number of the Cymmrodor, has been made the subject of some unfavourable comments, I wish to offer a reply to them (which shall be as brief as possible), that the disputed points may be, at least, more clearly understood.

The main argument of the paper has not been assailed. No reasons have been given, for instance, why the Welsh words corn and llin should be classed as borrowed words, while the Germ. horn and lein are treated as unborrowed. But I am asked if I hold that words common to the Sanskrit and Celtic languages may not, in any case, have been borrowed by the latter from the Latin? This I have not undertaken to show. I contend only that such words ought to be regarded as derived from a common Aryan stock, unless the contrary can be proved historically or otherwise. I have been referred to the eminent German philologist, Windisch, and I accept the reference. He notices a derivation of the Irish caille, a veil, from the Lat. pallium; and, after pointing out that the connection is not probable, he adds, "Why may not the Ir. caille be a genuine Irish word?" (Warum soll ir. caille nicht ein echt irisches wort sein?-Kuhn's Beiträge, etc., viii, 18.) I ask the same question with regard to the words which I have discussed. Why must they be necessarily treated as borrowed words because they bear a resem

blance, often a very remote one, to Latin forins? Is it sound philology, for instance, to assume that the W. cum and the Bret. comb must be borrowed words, and to connect them with the Lat. concava as their source?

The main object of Windisch's paper is, however, to prove that at some undefined period the letter p vanished from the Celtic languages, and that when it re-appeared, at a later time, it was used only in borrowed words, or as the representative of an older k for kv or qv. I read his paper when it appeared in Kuhn's Beiträge, but was not convinced by his arguments. It is certainly true that a primitive p has disappeared from many Celtic words, and that in their modern form this letter often represents an older k or kv, but it does not follow that an Indo-Germanic p has not been retained in any genuine Celtic words. Mr. Whitley Stokes maintains that it has been retained in some instances. I have read the paper in the Revue Celtique (vol. ii, p. 337), in which the writer controverts the opinion of Mr. Stokes; but I fail to see that the Ir. Gael. pailt (plenteous) can be disposed of by suggesting that it may be borrowed from the Eng. word plenty, or that if the root pak may be assumed for the Indo-European mother tongue, yet "for the ItaloCeltic branch one must postulate quaqv (kak)". This is assuming as true what has not been proved. The German philologist Fick holds a contrary opinion. I quote from the Verg. Wört. der Indogerm. Sprachen, 3rd ed., 1874, "pak, kochen, reifen, . . . . lat. coquo, sup. coctum, kochen (für poquo durch eine Art Assimilation, wie quinque für pinque. s. pankan)." "Corn. peber, pistor, popei, pistrinum; cymrisch popuryes, pistrix; ksl. peka, kochen. vgl. skr. pac', kochen, pac'a, kochend" (i, 133; ii, 155). Professor Curtius thinks it is doubtful whether kak or pak is the original form, but he adds, " auf die Form pak gehen deutlich die sanskritischen und slawischen Formen zurück, ebenso die ältere

Präsensform Téoow=πekjw” (Grundzuge,2 409). Professor Ascoli suggests that both forms may have existed simultaneously from the period of original unity (Corsi di Glottologia, p. 78). Professor Fr. Müller maintains that the Sans. panćan (five, Lith. penki, W. pump) is connected with Sans. pankti (series), and was primarily pancant, standing in a row, i.e., the five fingers (Beiträge, ii, 398). It must then be the primitive form.

The results of these different theories may be seen in the varying explanations of the Lat. pars, W. parth. Ascoli connects a primitive part with Sans. pat, to cleave (Corsi, etc., p. 80). Fick infers an Aryan par as the source of the Lat. pars, and refers to the Sans. par (prī) to spend (Wört.3 1,664). In the Revue Celtique (vol. ii, 333) the W. parth is assumed to be derived from spart, for squart, and to be connected with an Aryan skard, to break, and Sans. khad. I will not here discuss the question whether Ascoli or Fick has chosen the best Sans. relative, but this may be said, that they have referred to living words, and that the assumed root, squart, is wholly imaginary. It may be noted that the letter is supposed to have fallen from the Sans. words pat and khad, and yet a suggestion of this kind on my part, in another instance, has been treated as an impossibility. Few philological changes are more common (compare E. speak and Germ. sprechen). Professor Pictet has compared the W. pallu and other words with corresponding forms in Irish. For this he has been assailed by Windisch and Ebel (Beitr. viii, 25; iii, 278), but until it has been proved that pallu is only a modern form, his conclusions cannot justly, I think, be condemned as unsound. Windisch states that the Ir. stem alla, in di-all (declinatio) is from a root palla, which he connects with the Lith. pulu (to fall), and Old N., falla (to fall, to fail; cadere, deficere, Egillson.—(Beitr. viii, 2). This is identical with the W. pallu (deficere, Davies).