1 They are produced here in their order as printed at the beginning of the second volume of the Myvyrian Archaiology of Wales, and the series or versions are indicated as i, ii, iii. Version ii will be found printed in the third volume of the Cymmrodor, pp. 52–61, also in the Oxford Mabinogion, pp. 297–308, from the Red Book of Hergest of the fourteenth century. The letter (a, b, c) added is intended to indicate the order of the three parts of the Triad, for it is not the same in all the series. Let me here remark in a general way that the former fondness of the Welsh for Triads was not peculiar to them. The Irish also must have been at one time addicted to this grouping. Witness the Triad of Cleverest Countings, in the Book of the Dun Cow, fol. 58^a, and the Triad of the Blemishes of the Women of Ulster, ib. 43^b. ↑

2 As to the names Drystan (also Trystan) and Essyỻt, see the footnote on p. 480 above. ↑

3 This was meant to explain the unusual term gỽrdueichyat, also written gỽrdueichat, gỽrueichyat, and gwrddfeichiad. This last comes in the modern spelling of iii. 101, where this clause is not put in the middle of the Triad but at the end. ↑

4 The editor of this version seems to have supposed Pendaran to have been a place in Dyfed! But his ignorance leaves us no evidence that he had a different story before him. ↑

5 This word is found written in Mod. Welsh Annwfn, but it has been mostly superseded by the curtailed form Annwn, which appears twice in the Mabinogi of Math. These words have been studied by M. Gaidoz in Meyer and Stern’s Zeitschrift für Celtische Philologie, i. 29–34, where he equates Annwfn with the Breton anauon, which is a plural used collectively for the souls of the departed, the other world. His view, however, of these interesting words has since been mentioned in the same Zeitschrift, iii. 184–5, and opposed in the Annales de Bretagne, xi. 488. ↑

6 Edited by Professor Kuno Meyer (London, 1892): see for instance pp. 76–8. ↑

7 See Windisch’s Irische Texte, p. 256, and now the Irish Text Society’s Fled Bricrend, edited with a translation by George Henderson, pp. 8, 9. ↑

8 Windisch, ibid. pp. 99–105. ↑

9 See the Oxford Mabinogion, p. 196, and Guest’s trans., i. 302, where the Welsh words a golỽython o gic meluoch are rendered ‘and collops of the flesh of the wild boar,’ which can hardly be correct; for the mel in mel-uoch, or mel-foch in the modern spelling, is the equivalent of the Irish melg, ‘milk.’ So the word must refer either to a pig that had been fed on cows’ milk or else a sucking pig. The former is the more probable meaning, but one is not helped to decide by the fact, that the word is still sometimes used in books by writers who imagine that they have here the word mel, ‘honey,’ and that the compound means pigs whose flesh is as sweet as honey: see Dr. Pughe’s Dictionary, where melfoch is rendered ‘honey swine,’ whatever that may mean. ↑

10 Windisch’s Irische Texte, p. 133, where laith lemnacht = Welsh ỻaeth ỻefrith, ‘sweet milk.’ ↑

11 Coỻfrewi was probably, like Gwenfrewi, a woman’s name: this is a point of some importance when taken in connexion with what was said at p. 326 above as to Gwydion and Coỻ’s magic. ↑

12 This reminds one of Geoffrey of Monmouth’s Henvinus, whom he makes into dux Cornubiæ and father of Cunedagius or Cuneđa: see ii. 12, 15. Probably Geoffrey’s connecting such names as those of Cuneđa and Dyfnwal Moelmud (ii. 17) with Cornwall is due to the fact, that the name of the Dumnonia of the North had been forgotten long before that of the Dumnonia to be identified with Devon and Cornwall. ↑

13 See the Oxford Mabinogion, p. 104, and the Oxford Bruts, p. 292. ↑

14 See the Oxford Bruts, pp. 299, 317, 345–6, 348, 384. I learn from Prof. Anwyl that Casteỻ Penweđig is still remembered at Ỻanfihangel Genau’r Glyn as the old name of Casteỻ Gwaỻter in that parish. ↑

15 See his note in Owen’s Pembrokeshire, p. 237, where he also notices Aber Tarogi, and the editor’s notes to p. 55. ↑

16 Mergaed for Mengwaed hardly requires any explanation; and as to Breat or rather Vreat, as it occurs in mutation, we have only to suppose the original carelessly written Vrēac for Vrēach, and we have the usual error of neglecting the stroke indicating the n, and the very common one of confounding c with t. This first-mentioned name should possibly be analysed into Mengw-aed or Menw-aed for an Irish Menb-aed, with the menb, ‘little,’ noticed at p. 510 below; in that case one might compare such compounds of Aed as Beo-aed and Lug-aed in the Martyrology of Gorman. Should this prove well founded the Mod. Welsh transcription of Menwaed should be Menwaeđ. I have had the use of other versions of the Triads from MSS. in the Peniarth collection; but they contribute nothing of any great importance as regards the proper names in the passages here in question. ↑

17 See the Oxford Mabinogion, pp. 41, 98, and Guest’s trans., iii. 313. ↑

18 See Geoffrey’s Historia Regum Britanniæ, vi. 19, viii. 1, 2; also Giraldus, Itinerarium Kambriæ, ii. 8 (p. 133). ↑

19 Itinerarium Kambriæ, ii. 9 (p. 136). ↑

20 Menw’s name is to be equated with the Irish word menb, ‘little, small,’ and connected with the Welsh derivative di-fenw-i, ‘belittling or reviling’: it will be seen that he takes the form of a bird, and his designation Menw fab Teirgwaeđ might perhaps be rendered ‘Little, son of Three-Cries.’ ↑

21 Identified by Professor Kuno Meyer in the Transactions of the Cymmrodorion Society, 1895–6, p. 73, with a place in Leinster called Sescenn Uairbeóil, ‘the Marsh of Uairbhél,’ where Uairbhél may possibly be a man’s name, but more likely that of a pass or gap described as Cold-mouth: compare the Slack or Sloc in the Isle of Man, called in Manx ‘the big Mouth of the Wind.’ The Irish name comes near in part to the Welsh Esgeir Oervel or Oerfel, which means ‘the mountain Spur of cold Weather.’ ↑

22 The word used in the text is ystyr, which now means ‘meaning or signification’; but it is there used in the sense of ‘history,’ or of the Latin ‘historia,’ from which it is probably borrowed. ↑

23 In the original his designation is Gwrhyr Gwalstawt Ieithoeđ, and the man so called is in the Kulhwch credited with the mastery of all languages, including those of certain birds and quadrupeds. Gwalstawt, found written also gwalstot, is the Anglo-Saxon word wealhstód, ‘an interpreter,’ borrowed. The name Gwrhyr is possibly identical with that of Ferghoir, borne by the Stentor of Fionn mac Cumhaill’s following. Ferghoir’s every shout is said to have been audible over three cantreds. Naturally one who was to parley with a savage host had good reason to cultivate a far-reaching voice, if he wished to be certain of returning to his friends. For more about it see the footnote at p. 489 of my Hibbert Lectures. ↑

24 The original has Pelumyawc, p. 138, and the name occurs in the (Red Book) Bruts, p. 355, as Pelunyawc, and p. 411, as Pelunea(wc) between the commots of Amgoed and Velfrey. The identification here suggested comes from Mr. Phillimore, who has seen that Peuliniawc must be a derivative from the name Paulinus, that is of the Paulinus, probably, who is mentioned in an ancient inscription at Ỻandysilio. There are other churches called after Tysilio, so this one used to be distinguished as Ỻandysilio yn Nyfed, that is, Ỻandysilio-in-Dyfed; but the pronunciation was much the same as if it had been written Ỻandysilio yn Yfed, meaning ‘Ỻandysilio a-drinking,’ ‘whereof arose a merrye jest,’ as George Owen tells us in his Pembrokeshire, p. 9. It is now sometimes called Ỻandysilio’r Gynffon, or ‘Ỻandysilio of the Tail,’ from the situation of a part of the parish on a strip, as it were a tail, of Carmarthenshire land running into Pembrokeshire. ↑

25 This Aber Towy appears to have been a town with a harbour in 1042, for we read in Brut y Tywysogion of a cruel engagement fought there between Gruffyđ ab Ỻewelyn and Howel ab Edwin, who, with Irish auxiliaries, tried to effect a landing. Not long ago a storm, carrying away the accumulation of sand, laid bare a good deal of the site. It is to be hoped that excavations will be made soon on the spot. ↑

26 See the Transactions of the Cymmrodorion, 1894–5, pp. 146–7. There are a good many clyns about South Wales, but our etymologists are careful to have them in most cases written glyn, ‘a glen.’ Our story, however, shows that the word came under the influence of glyn long ago, for it should be, when accented, clûn, corresponding to Irish cluain, ‘a meadow.’ We have it as clun in Clun Kein in the Black Book, p. 34^b, where I guess it to mean the place now called Cilcain, ‘Kilken’ in Flintshire, which is accented on the first syllabic; and we have had it in y Clun Hir, ‘the Long Meadow,’ mentioned above at p. 22. ↑

27 Cas Ỻychwr, ‘Loughor Castle,’ is supposed to involve in its Ỻychwr, Ỻwchwr, or Loughor, the name of the place in the Antoninus Itinerary, 484, 1, to wit Leucarum; but the guttural spirant ch between vowels in Ỻychwr argues a phonetic process which was Goidelic rather than Brythonic. ↑

28 Ỻwydawc Gouynnyat would seem to mean Ỻwydawc the Asker or Demander, and the epithet occurs also in the Kulhwch in the name Gaỻcoyt Gouynynat (Mabinogion, 106), to be read doubtless G. Gouynnyat, ‘G. who asks or demands’: possibly one should rather compare with Go-uynnyat the word tra-mynyat, ‘a wild boar’: see Williams’ Seint Greal, pp. 374, 381. However, the epithets in the Twrch Trwyth story do not count so far as concerns the place-names derived. ↑

29 Other instances of the like shortening occur in words like cefnder, ‘a cousin,’ for cefnderw, and arđel, ‘to own,’ for arđelw. As to Amman, it enters, also, into a group of Glamorganshire place-names: witness Aber Amman and Cwm Amman, near Aberdare. ↑

30 It should perhaps be looked for near Brechfa, where there is a Hafod Grugyn, and, as I am told, a Garth also which is, however, not further defined. For it appears that both Brechfa and Cayo, though now in Carmarthenshire, once belonged to Keredigion: see Owen’s Pembrokeshire, p. 216. But perhaps another spot should be considered: J. D. Rhys, the grammarian (p. 22 above), gives in the Peniarth MS. 118 a list of caers or castles called after giants, and among them is that of Grugyn in the parish, he says, of ‘Ỻan Hilar.’ I have, however, not been able to hear of any trace of the name there, though I should guess the spot to have been Pen y Casteỻ, called in English Castle Hill, the residence of Mr. Loxdale in the parish of Ỻanilar, near Aberystwyth. ↑

31 I have re-examined the passage, and I have no doubt that the editors were wrong in printing Gregyn: the manuscript has Grugyn, which comes in the last line of column 841. Now besides that the line is in part somewhat faint, the scribe has evidently omitted something from the original story, and I guess that the lacuna occurs in the first line of the next column after the words y ỻas, ‘was killed,’ which seem to end the story of Grugyn. ↑

32 Those who have discovered an independent Welsh appellative wy meaning water are not to be reasoned with. The Welsh wy only means an egg, while the meaning of Gwy as the name of the Wye has still to be discovered. ↑

33 This name also occurs in a passage quoted in Jones’ Brecknock, ii. 501, from a Carte MS. which he treats as relating to the year 1234: the MS. is said to be at the Bodleian, though I have not succeeded in tracing it. But Jones gives Villa de Ystraddewi, and speaks of a chapel of St. John’s of Stradtewi, which must have been St. John’s Church, at Tretower, one of the ecclesiastical districts of Cwm Du: see also p. 497. The name is probably to be treated as Strad or Strat d’Ewe. ↑

34 A river may in Welsh be briefly called after anybody or anything. Thus in North Cardiganshire there is a stream called Einon, that is to say ‘Einion’s river,’ and the flat land on both sides of it is called Ystrad Einon, which looks as if one might translate it Einion’s Strath, but it means the Strath of Einion’s river, or of the stream called Einon, as one will at once see from the upper course of the water being called Blaen Einon, which can only mean the upper course of the Einon river. So here yw is in English ‘yew,’ but Ystrad Yw and Ỻygad Yw have to be rendered the Strath of the Yew burn and the Eye of the Yew burn respectively. It is moreover felt by the Welsh-speaking people of the district that yw is the plural of ywen, ‘a single yew,’ and as there is only one yew at the source somebody had the brilliant idea of making the name right by calling it Ywen, and this has got into the maps as Ewyn, as though it were the Welsh word for foam. Who began it I cannot say, but Theophilus Jones has it in his History of the County of Brecknock, published in 1809. Nevertheless the name is still Yw, not Ywen or Ewyn, in the Welsh of the district, though Lewis gives it as Ywen in his article on Ỻanvihangel-Cwm-Du. ↑

35 For exact information as to the Gaer, the Yw, and Ỻygad Yw, I am indebted chiefly to the courtesy of Lord Glanusk, the owner of that historic strath, and to the Rector of Ỻansantffread, who made a special visit to Ỻygad Yw for me; also to Mr. Francis Evans, of the Farmers’ Arms at the Bwlch, who would be glad to change the name Ỻygad Yw into Ỻygad dan yr Ywen, ‘the Source beneath the Yew-tree,’ partly on account of the position ‘of the spring emanating under the but of the yew tree,’ and partly because there is only a single yew there. Theophilus Jones complained a century ago that the Gaer in Ystrad Yw had not attracted the attention it deserved; and I have been greatly disappointed to find that the Cambrian Archæological Association has had nothing to say of it. At any rate, I have tried the Index of its proceedings and found only a single mention of it. The whole district is said to teem with antiquities, Celtic, Roman, and Norman. ↑

36 Theophilus Jones, in his Brecknockshire, ii. 502, describes Miarth or Myarth as a ‘very extensive’ camp, and proceeds as follows:—‘Another British camp of less extent is seen on a knoll on Pentir hill, westward of the Rhiangoỻ and the parish church of Cwmdu, above a wood called Coed y Gaer, and nearly opposite to the peak or summit called Cloch y Pibwr, or the piper’s call.’ This would probably be more accurately rendered the Piper’s Rock or Stone, with cloch treated as the Goidelic word for a stone rather than the Brythonic word for a bell: how many more clochs in our place-names are Goidelic? ↑

37 The Twrch would seem to have crossed somewhere opposite the mouth of the Wye, let us say not very far from Aust; but he escapes to Cornwall without anything happening to him, so we are left without any indication whether the story originally regarded Kernyw as including the Penrhyn Awstin of the Coỻ story given at p. 503. ↑

38 For this suggestion I am indebted to the Rev. Dr. Gaster in the Cymmrodorion’s Transactions for 1894–5, p. 34, and also for references in point to M. Cosquin’s Contes Populaires de la Lorraine, i. 134, 141, 152. Compare also such Gaelic stories as that of the Bodach Glas, translated by Mrs. Mackellar, in the Celtic Magazine, xii. 12–6, 57–64. ↑

39 In some native Welsh words we have an option between a prefix ym and am, an option arising out of the fact that originally it was neither ym nor am, but m̥, for an earlier m̥bi, of the same origin as Latin ambi and Greek ἀμφί, ‘around, about.’ The article, its meaning in the combination in banbh being forgotten, would fall under the influence of the analogy of the prefix, now am or ym, so far as the pronunciation was concerned. ↑

40 Possibly the benwic was thrown in to correct the reckoning when the redactor discovered, as he thought, that he had one too many to account for: it has been pointed out that he had forgotten that one had been killed in Ireland. ↑

41 It is just possible, however, that in an older version it was named, and that the place was no other than the rock just above Ystrad Yw, called Craig Lwyd or, as it is said to be pronounced, Craig Ỻwyd. If so, Ỻwyd would seem to have been substituted for the dissyllable Ỻwydog: compare the same person called Ỻwyt and Ỻwydeu in the Mabinogion, pp. 57, 110, 136. ↑

42 The name is well known in that of Ỻanrhaiadr yn Mochnant, ‘Ỻanrhaiadr in Mochnant,’ in the north of Montgomeryshire. ↑

43 Between Colwyn Bay and Ỻandudno Junction, on the Chester and Holyhead line of railway. ↑

44 I have discussed some of the traces of the Goidels in Wales in the Arch. Camb. for 1895, pp. 18–39, 264–302; 1899, pp. 160–7. ↑

45 In fact the genitive Grúcind occurs in the Book of Leinster, fo. 359^a. ↑

46 The sort of question one would like to ask in that district is, whether there is a spot there called Beđ y Rhyswyr, Carn y Rhyswyr, or the like. The word rhyswr is found applied to Arthur himself in the Life of Gruffyđ ab Cynan, as the equivalent probably of the Latin Arthur Miles (p. 538 below): see the Myvyrian Archaiology, ii. 590. Similarly the soldiers or champions of Christ are called rysỽyr crist in the Welsh Life of St. David: see the Elucidarium and other Tracts (in the Anecdota Oxoniensia), p. 118. ↑

47 Rudvyw Rys would be in Modern Welsh Rhuđfyw Rys, and probably means Rhuđfyw the Champion or Fighter, as Rhys is likely to have been synonymous with rhyswr. The corresponding Irish name was Russ or Ross, genitive Rossa, and it appears to come from the same origin as Irish ross, ‘a headland, a forest,’ Welsh rhos, ‘moorland, uncultivated ground.’ The original meaning was presumably ‘exposed or open and untilled land’; and Stokes supposes the word to stand for an early (p)ro-sto- with sto of the same origin as Latin sto, ‘I stand,’ and as the English word stand itself. In that case Ros, genitive Rossa, Welsh Rhys, would mean one who stands out to fight, a προστάτης, so to say. But not only are these words of a different declension implying a nominative Ro-stus, but the Welsh one must have been once accented Ro-stús on the ending which is now lost, otherwise there is no accounting for the change of the remaining vowel into y. Other instances postulating an early Welsh accentuation of the same kind are very probably ỻyg, ‘a fieldmouse,’ Irish luch, ‘a mouse’; pryd, ‘form,’ Irish cruth; pryf, ‘a worm,’ Irish cruim; so also with ych, ‘an ox,’ and nyth, ‘a nest,’ Irish nett, genitive nitt, derived by Stokes from nizdo-, which, however, must have been oxytone, like the corresponding Sanskrit nīdhá. There is one very interesting compound of rhys, namely the saint’s name Rhwydrys, as it were Rēdo-rostus to be compared with Gaulish Eporēdo-rīx, which is found in Irish analysed into rí Eochraidhi, designating the fairy king who was father to Étáin: see Windisch’s Irische Texte, p. 119. Bledrws, Bledrus, as contrasted with Bledrys, Bledris, postulate Goidelic accentuation, while one has to treat Bledruis as a compromise between Bledrws and Bledris, unless it be due to misreading a Bledruif (Book of Ỻan Dâv, pp. 185, 221–2, and Arch. Camb. for 1875, p. 370). The Goidelic accent at an early date moved to first syllables, hence cruth (with its vowel influenced by the u̯ of a stem qu̯r̥t) under the stress accent, became, when unstressed, cridh (from a simplified stem cr̥t) as in Noicride (also Nóicrothach, Windisch, ibid., pp. 259, 261, 266) and Luicridh (Four Masters, A.D. 748), Luccraid, genitive Luccraide (Book of Leinster, 359^f), Luguqurit- in Ogam. ↑

48 These operations cannot have been the first of the kind in the district, as a writer in the Archæologia Cambrensis for 1862, pp. 159–60, in extracting a note from the Proceedings of the Society of Antiquaries (series II, vol. i. p. 10) relative to the discovery of the canoe, adds a statement based on the same volume, p. 161, to the effect that ‘within half a mile of Ỻyn Ỻydaw there are the remains of a British town, not marked in the ordnance map, comprising the foundations of numerous circular dwellings. In some of them quantities of the refuse of copper smeltings were found. This town should be visited and examined with care by some of the members of our Association.’ This was written not far short of forty years ago; but I am not aware that the Association has done anything positive as yet in this matter. ↑

49 According to Jenkins’ Beđ Gelert, p. 300, the canoe was subsequently sold for a substantial price, and nobody seems to know what has eventually become of it. It is to be hoped this is not correct. ↑

50 See Holder’s Alt-celtischer Sprachschatz, s. v. Litavia. ↑

51 For these notes I am indebted to Williams’ Dictionary of Eminent Welshmen, and to Rees’ Welsh Saints, pp. 187, 191; for our Paulinus is not yet recognized in the Dictionary of Christian Biography. His day was Nov. 22. ↑

52 There are two other inscriptions in South Wales which contain the name Paulinus, one on a stone found in the neighbourhood of Port Talbot in Glamorgan, reading Hic iacit Cantusus Pater Paulinus, which seems to imply that Paulinus set up the stone to the memory of a son of his named Cantusus. The other, found on the site of the extinct church of Ỻanwrthwl, near Dolau Cothi in Carmarthenshire, is a remarkable one in a kind of hexameter to the following effect:—

Whether we have one or two or three Paulini in these inscriptions I cannot say. Welsh writers, however, have made the name sometimes into Pawl Hên, ‘Paul the Aged,’ but, so far as I can see, without rhyme or reason. ↑

53 Since I chanced on this inscription my friend Professor Lindsay of St. Andrews has called my attention to Plautus’ Asinaria, 499 (II. iv. 92), where one reads, Periphanes Rhodo mercator dives, ‘Periphanes a wealthy merchant of Rhodes’; he finds also Æsculapius Epidauro (Arnobius, 278. 18), and elsewhere Nepos Philippis and Priscus Vienna. ↑

54 See Stokes’ Patrick, pp. 16, 412. ↑

55 This will give the reader some idea of the pre-Norman orthography of Welsh, with l for the sound of ỻ and b for that of v. ↑

56 The softening of Cafaỻ to Gafaỻ could not take place after the masculine corn, ‘a horn’; but it was just right after the feminine carn, ‘a cairn.’ So here corn is doubtless a colloquial corruption; and so is probably the t at the end, for as ỻt has frequently been reduced to ỻ, as in cyfaiỻ, ‘a friend,’ from the older cyfaiỻt, in Medieval Irish comalta, ‘a foster brother or sister,’ the language has sometimes reversed the process, as when one hears hoỻt for hoỻ, ‘all,’ or reads fferyỻt, ‘alchemist, chemist,’ for fferyỻ from Vergilius. The Nennian orthography does not much trouble itself to distinguish between l and ỻ, and even when Carn Cabal was written the pronunciation was probably Carn Gavaỻ, the mutation being ignored in the spelling, which frequently happens in the case even of Welsh people who never fail to mutate their consonants in speaking. Lastly, though it was a dog that was called Cafaỻ, it is remarkable that the word has exactly the form taken by caballus in Welsh: for cafaỻ, as meaning some sort of a horse, see Silvan Evans’ Geiriadur. ↑

57 An instance or two of Trwyd will be found in a note by Silvan Evans in Skene’s Four Ancient Books of Wales, ii. 393. ↑

58 For more about these names and kindred ones, see a note of mine in the Arch. Cambrensis, 1898, pp. 61–3. ↑

59 See my Hibbert Lectures, pp. 398–401. ↑

60 See the Black Book of Carmarthen in Evans’ facsimile, p. 47^b; Thomas Stephens’ Gododin, p. 146; Dent’s Malory, preface, p. xxvi; and Skene’s Four Ancient Books of Wales, ii. 51, 63, 155. ↑

61 See the Göttingische gelehrte Anzeigen for 1890, p. 512. ↑

62 See De Courson’s Cartulaire de l’abbaye de Redon, pp. 163, 186. ↑

63 See Reeves’ note to the passage just cited in his edition of Adamnan’s Vita, pp. 6, 7. ↑

64 Here possibly one might mention likewise Gilmin Troetu or Troedđu, ‘Gilmin of the Black Foot,’ the legendary ancestor (p. 444) of the Wynns of Glyn Ỻifon, in Carnarvonshire. So the name might be a shortening of some such a combination as Gilla-min, ‘the attendant of Min or Men,’ a name we have also in Mocu-Min, ‘Min’s Kin,’ a family or sept so called more than once by Adamnan. Perhaps one would also be right in regarding as of similar origin the name of Gilberd or Gilbert, son of Cadgyffro, who is mentioned in the Kulhwch, and in the Black Book, fo. 14^b: at any rate I am not convinced that the name is to be identified with the Gillebert of the Normans, unless that was itself derived from Celtic. But there is a discrepancy between Gilmin, Gilbert, with unmutated m and b, and Gilvaethwy with its mutation consonant v. In all three, however, Gil, had it been Welsh, would probably have appeared as Giỻ, as indicated by the name Giỻa in the Kulhwch (Oxford Mabinogion, p. 110), in which we seem to have the later form of the old name Gildas. Compare such Irish instances as Fiachna and Cera, which seem to imply stems originally ending in -asa-s (masculine) and -asā (feminine); and see the Journal of the Royal Society of Antiquaries of Ireland, 1899, P. 402. ↑

65 An article in the Rennes Dindṡenchas is devoted to Liath: see the Rev. Celtique, xvi. 78–9. As to Celtchar, genitive Celtchair, the name would seem to have meant ‘him who is fond of concealment.’ The Mabinogi form of the Welsh name is Ỻwyt uab kil coet, which literally meant ‘Ỻ. son of (him of) the Retreat of the Wood.’ But in the Twrch Trwyth story, under a slightly different form of designation, we appear to have the same person as Ỻwydeu mab kelcoet and Ỻwydeu mab kel coet, which would seem to mean ‘Ỻ. son of (him of) the Hidden Wood.’ It looks as if the bilingual story-teller of the language transition had not been able to give up the cel of Celtchar at the same time that he rendered celt by coet, ‘wood or trees,’ as if identifying it with cailt: witness the Medieval Irish caill, ‘a wood or forest,’ dative plural cailtib, derivative adjective caillteamhuil, ‘silvester’; and see Windisch’s Irische Texte, p. 410, s. v. caill. ↑

66 Windisch’s Irische Texte, p. 217, and the Book of the Dun Cow, fo. 47^b. ↑

67 There has been a good deal of confusion as to the name Ỻyr: thus for instance, the Welsh translations of Geoffrey of Monmouth make the Leir of his Latin into Ỻyr, and the personage intended is represented as the father of three daughters named Gonerilla, Regan, and Cordeilla or Cordelia. But Cordelia is probably the Creurđilad of the Black Book, p. 49^b, and the Creiđylat of the Kulhwch story (the Oxford Mabinogion, pp. 113, 134), and her father was Ỻûd Ỻawereint (= Irish Nuada Airgetlám) and not Ỻyr. Then as to the Leir of Geoffrey’s Latin, that name looks as if given its form on the strength of the legr- of Legraceaster, the Anglo-Saxon name of the town now called Leicester, of which William of Malmesbury (Gesta Pontificum, § 176) says, Legrecestra est civitas antiqua in Mediterraneis Anglis, a Legra fluvio præterfluente sic vocata. Mr. Stevenson regards Legra as an old name of the Soar, and as surviving in that of the village of Leire, spelled Legre in Domesday. It seems to point back to a Legere or Ligere, which recalls Liger, ‘the Loire.’ ↑

68 I say in that case, as this is not quite conclusive; for Welsh has an appellative ỻyr, ‘mare, æquor,’ which may be a generalizing of Ỻyr; or else it may represent an early lerio-s from lero-s (see p. 549 below), and our Ỻyr may possibly be this and not the Irish genitive Lir retained as Ỻyr. That, however, seems to me improbable on the whole. ↑

69 Here it is relevant to direct the reader’s attention to Nutt’s Legend of the Holy Grail, p. 28, where, in giving an abstract of the Petit saint Graal, he speaks of the Brân of that romance, in French Bron, nominative Brons, as having the keeping of the Grail and dwelling ‘in these isles of Ireland.’ ↑

70 The Dôn and Ỻyr groups are not brought into conflict or even placed in contact with one another; and the reason seems to be that the story-teller wanted to introduce the sons of Beli as supreme in Britain after the death of Brân. Beli and his sons are also represented in Maxen’s Dream as ruling over Britain when the Roman conqueror arrives. What is to be made of Beli may be learnt from The Welsh People, pp. 41–3. ↑

71 These things one learns about Lir from the story mentioned in the text as the ‘Fate of the Children of Lir,’ as to which it is right, however, to say that no ancient manuscript version is known: see M. d’Arbois dc Jubainville’s Essai d’un Catalogue de la Litérature épique de l’Irlande, p. 8. ↑

72 See Skene’s Four Ancient Books of Wales, ii. 303, also 108–9, where the fragment of the poem as given in the Book of Taliessin is printed. The line here quoted has been rendered in vol. i. 286, ‘With Matheu and Govannon,’ which places the old pagan Gofannon in rather unexpected company. A few lines later in the poem mention is made of a Kaer Gofannon: where was that? Skene, in a note on it (ii. 452), says that ‘In an old list of the churches of Linlithgow, printed by Theiner, appears Vicaria de Gumanyn. The place meant is probably Dalmeny, on the Firth of Forth, formerly called Dumanyn.’ This is interesting only as showing that Gumanyn is probably to be construed Dumanyn, and that Dalmeny represents an ancient Dún Manann in a neighbourhood where one already has Clach Manann, ‘the stone of Manau,’ and Sliabh Manann, ‘Mountain of Manau’ now respectively Clackmannan and Slamannan, in what Nennius calls Manau Guotodin. ↑

73 This occurred unrecognized and, therefore, unaltered by the scribe of the Nennian Pedigree no. xvi in the Cymmrodor, ix. 176, as he found it written in an old spelling, Louhen. map. Guid gen. map. Caratauc. map. Cinbelin, where Caradog is made father of Gwydion; for in Guid-gen we seem to have the compound name which suggested Gwydion. This agrees with the fact that the Mabinogi of Math treats Gwydion as the father of Ỻew Ỻawgyffes; but the pedigree itself seems to have been strangely put together. ↑

74 See Bertrand’s Religion des Gaulois, pp. 314–9, 343–5, and especially the plates. ↑

75 The Oxford Mabinogion, pp. 40–3; Guest’s Mabinogion, iii. 124–8. ↑

76 See Louis Leger’s Cyrille et Méthode (Paris, 1868), p. 22. ↑

77 See Pertz, Monumenta Germaniæ Historia Scriptorum, xii. 794. The whole passage is worth quoting; it runs thus: Erat autem simulacrum triceps, quod in uno corpore tria capita habens Triglaus vocabatur; quod solum accipiens, ipsa capitella sibi cohærentia, corpore comminuto, secum inde quasi pro tropheo asportavit, et postea Romam pro argumento conversionis illorum transmisit. ↑

78 See The Welsh People, pp. 56–7. ↑