in Coxe's Monmouthsh., 1801: Gwent. iddei ben =Venedot. i'w ben; Hughes, 1822 S.W. aeth iddei dy; Dosp. Edeyrn, § 1260: Gwent. and Dimetian iddei; in the dialectal list, however, it is only said to be Gwentian. Cf. also Y Traeth., iii, p. 14.

In modern dialects: iddi fam, iddi mham (Aberdare; mh is S.W.). Iddei, iddeu is always used by Iolo Morganwg in Y Cymmr., iv, pp. 101-5 (Glamorgansh.); beth sydd genych chi iddu wed (= i'w ddywedyd) yn awr (Llanelli); idd i hercyd nhw shatre (Y Bedyddiwr, viii, p. 108, from Monmouthsh.; shatre tu ag atref; hercyd is S.W.), etc.

An accurate delimitation of the parts of South Wales in which iddei is used, would be of great interest for the localisation of many manuscripts, etc.

[65.] Iddei for i'w caused, in the South Welsh modern language, i with the article (i'r) to be supplanted by idd y, idd- being abstracted from idd-ei. In Traethawd ar Iawnlythyreniad . . . . yr iaith Gymmraeg, gan John Jones (Rhydychen, 1830, 8°), idd and odd are said to be regularly used in the S. W. vulgar language before the article: awn idd y tŷ, for i'r ty or i y ty (this latter form is only a construction of the author). D. S. Evans, Llythyraeth, mentions odd ei, odd eich, odd eu, for o'i, o'ch, o'u; and iddei, iddeu, iddein, iddeich, iddy, for i'w, i'n, i'ch, i'r. Rowlands, Gramm., 4p. 118: ydd y tŷ, odd eu tai. These forms really occur in the literary language of South Welsh periodicals: cf. Seren Gomer (Swansea, 1830), p. 80: gyda phob parch idd eich gohebydd ; xxiii, p. iv (written by the editor): idd ein gohebwyr; vol. v, p. 364 (1822): neu cyfieithiad W. S. ac y Dr. M. odd y Bibl Cymreig; ib. Dr. Pughe is said to write always idd yr; vol. xxvi, p. 271 (1843): some write am beth ag oedd yn bodoli, for am b. oedd yn b., odd y tŷ, odd ei ben, idd y dwfr, etc.; Y Bedyddiwr (Cardiff), 1851, p. 10: ac a adroddodd yr achos oddei ofwyad. These forms are not difficult of explanation. Iddei, iddeu, was separated into idd- and the poss. ei, hence idd with other possess. pronouns and with the article (idd ein, eich, y). Oddei, of course, follows iddei, as the full form of

the preposition is oc. Odd- in oddei was either wrongly abstracted from oddiwrth, oddiallan, etc. (: iwrth), or-and this is more probable—oc ei was transformed into odd ei after the model of idd ei, by reason of the meanings of the two prepositions (to and from) being strictly opposed to each other. Such contrast of meaning very often contributes to the mutual assimilation of two words or forms; a large number of examples of this kind of analogy were collected by Brugmann in his article on éví, èv, eis (Berichte der Sächs. Ges. der Wiss., 1883, pp. 181-195). Irish examples are sósar —sinser, tess—túaid, etc. (see Stokes in Bezzenberger's Beitr., ix, p. 92); cf. Welsh asswy and asseu-deheu.

[66.] In Liber Landavensis some instances of iddy seem to occur. This is not surprising, as the probable explanation of iddy, iddei, requires the assumption of the existence of the initial d of the preposition, which is indeed sometimes kept in L. Landav., a Gwentian text. I should put full confidence in these forms, if this text were more carefully edited and from the original manuscript, still in existence. In the great charter on pp. 113, 114: ac idythir hac idi dair (of eccluys Teliau). In the same charter: har-mefyl har sarhayt, etc., a guneel brennhin Morcanhuc hay gur hay guas (and his-) dy escop Teliau hac dy gur (to his―) hac dy guas (to his-); ha diguadef braut diam y cam a diconher dy escop Teliau ha dy (to his) gur ha dy guas. If these dy," to his", are to be relied upon, they are forms of very great importance, and, as far as I know, unique.

[67.] So we find o'e (o'i), i'w, o'e y, i'w y (i'w ei), y ei, eu, iddy (iddei), and perhaps dy, used to express "to his, her, their." In the following paragraph I will outline some guesses on the connection of these apparently disparate forms. The chief difficulty is the form of the poss. pronoun. *Do-í becomes *di (cf. Corn. dy, Bret. de, and probably Welsh dy in L. Landav.). To this, at the time when the initial d

still existed, at least in some positions (sandhi), as in L. Land., di was prefixed, giving di-dy, i-dy, later iddy, and, later, in the accented form of the poss. pronoun, iddei, iddeu. Cf. Ir. chucum, chucut, etc.? Dó-ei, do-i, gave o'ei, o'e, o'i, which existed till the 16th century, if not later. The reason of its disappearance is probably the exclusive use of o'i, o'e, for "from his", after oc (oc ei) fell out of use in North Wales. In South Wales iddei prevailed, and oddei for oc ei followed its analogy. The explanation of i'w offers by far the greatest difficulties. Rhys, Revue Celt., vi, p. 57 et seq., explained bwy in bwy gilydd, etc. (or mor bwy gilydd, etc., often in Middle Welsh texts), from po-i (i being the poss. pronoun). The colloquial form is bw. In the Book of Tal., Skene, pp. 138, 154: y ren rбy digonsei, rбy digones; rбy goreu, p. 158; náy kymrбy, p. 147; rwy golles, Myv. Arch., 2p. 160a (Cynddelw), etc.; cf. Rhys, Revue Celt., vi, p. 50 et seq.: rwy-, nwy-, are rho-, no-, and the infixed poss. pronoun i. The conditions of accent, etc., under which in pwy, rwy-, nwy- wy, sprang from *oi are not known to me; but these examples enable me to assume *do-i (to his) becoming under certain circumstances *dwy, *wy (d lost). There exists ryбgoreu, B. of Herg., Skene, p. 233, confirmed by Rhys, l. c. If this form is to be trusted, and if others similar to it exist, it would be possible to assume yw (written y'w) to come from *wy (do-i). But I prefer the following explanation: the loss of the initial d is certainly due to the influence of the different final sounds of previous words and their greater or lesser syntactic connection with *do. If, therefore, di exists in so late a MS. as L. Landav., this is no argument against assuming a much earlier loss of d in this preposition in certain positions. To this *wy, as in i-dy *dy, y was prefixed, giving *dywy, *ywy. The usual forms, y'w, i'w, are to be explained like bw from bwy, -ws (3rd pers. sing. of the s. pret.) from -wys. By the way, I think this South Welsh -wys to be formed after the analogy

of -wyt (part. pret. pass.), since as -es is, and -at -et -it (-wyt) coincided in vowel. -Wyt itself is not clear at all; it seems to be a wrong abstraction from the part. pret. pass. of the verb substantive, containing, besides the suffix, a part of the stem of the verb. The question now rises as to oe y, yw y, and i'w ei. It is a very seductive supposition, to take yw y in MS. Cleop. B. 5 (see § 62), for the y-wy which I thought to be the source of y'w, like -wys, so often written besides -WS. Is the modern i'w, ei, eu the successor of an older i'w-y, like iddei, iddeu, and the older iddy? Or was i'w too obscure, and ei, eu reintroduced to enforce the possessive meaning? And is this perhaps also the case with yw y in Cleop. B. 5? Oe y almost points in this direction, if it is to be relied upon at all, of which I am not wholly convinced. I am not able to decide this question, but this does not tend to disprove the other assumptions. In conclusion, I would draw attention to py (or mor by gilydd), occurring besides pwy (bwy, bw); bwy (bw): by, correspond exactly to *wy (y'w y, y'w): dy (L. Land., Corn., Bret.; i-dy in South Wales).

As to the explanation of i'w by Rhys, Rev. Celt., vi, if it is to be preferred to that proposed above, I would rather explain thev in the poss. 3rd pers. sing. by an analogical transfer from the teu (*tevos) of the 2nd pers. sing., as was the case with meu (1st sing.), and Ir. mo (see Kuhn's Zeitschrift, xxvii, p. 401, note 1).

D. THE RELATIVE PRONOUNS.

[68.] The "relative" pronoun a in its double function as simple relative and as so-called verbal particle (in constructions where its original demonstrative sense has faded away -see Zimmer, Kelt. Studien, ii, p. 59) is supplanted by y in South Welsh, more especially Gwentian texts, since the 16th

century—at least, I am not aware of any earlier instances. As to the explanation, see perhaps Rhys, Lectures, 2p. 147. Y, said to be the form of the oblique cases of a, has exceeded its proper domain of use; the reasons and the history of this analogical transgression are obscure to me.

Cf. Llyfr Gweddi Gyffredin, 1586, pref.: "a being relative or a voyce expletive for y, not used to them of South Wales"; and "y being rel, or a kind of expl. for a, vnto the North Welsh readers." J. D. Rhys, Gramm. : mi a garabh, peth a gaffer, and mi y g., p. y g. Davies, Gramm., p. 182 (1809): Dimetian mi y garaf (also in Richards, Gramm., 1753, p. 60). Williams, in Dosp. Edeyrn, § 823 and § 1116: Dimet. y for a (probably meant to denote South Wales in general, as always in Davies), and in the dialectal lists: Gwentian mi y garaf.

[69.] Sal., N. Test., uses y very often for a; e.g., pa beth y dderbyneist ac y glyweist;-am y pethe y ddywedesit yddynt can y bugelydd, f. 83b; ir apostolon y ddetholesei ef, f. 170a; y gwyr y ddodesoch yn carchar, f. 1776; am y pethae hyn y ddywedwn, f. 178a; yno y gesodesont (yd anvonesont) wyr y ddywedent; ni y (=ei) clywsam ef yn llavaru . . . ., f. 179a; etc. In Huet's Gweledigaeth Ieuan it occurs even more frequently. Cf. yr hon (sc. gweledigaeth) y rroedd dyw yddo ef yw ddangos yddy wasnaethwyr yrrein y orvydd yn vyan ddyfod y ben; ac y ddangosoedd gan y angel yddy wasnaethwr Ioan (ac ef y ddanvonoedd); yr hwn y dystolaethoedd o eir dyw ac o dyst. I. Chr. . . ac o pob peth ar y weloedd ef, f. 373a; ac ef y ddayth ac y gymerth y llyfr, f. 379a; a' phwy y ddychyn sefyll, f. 380b; (= a phwy a ddichon sefyll, ed. 1873); happys ywr neb y ddarlleyo ar rrei y wrandawant geyryey y bryffydolaeth hon, f. 3736, etc.

Also frequent in Addit. MS. 14,921, 16th cent. (John Maundeville's Travels, Gwentian dialect); Addit. MS. 15,038, f. 78b: am y wneythochi ero = Addit. MS. 14,973: am a wnaethochi erof; Llyfr Achau (Breconshire, 1602), Edward Kaer yn Arfon Yr Ayle y Bryodes