ii. 360, appears to have been astonished to find that Carn Cavall,

as she writes it, was no fabulous mound but an actual 'mountain in the district of Builth, to the south of Rhayader Gwy, and within sight of that town.' She went so far as to persuade one of her friends to visit the summit, and he begins his account of it to her with the words: 'Carn Cavall, or as it is generally pronounced Corn Cavall, is a lofty and rugged mountain.' On one of the cairns on the mountain he discovered what may have been the very stone to which the Mirabilia story refers; but the sketch with which he accompanied his communication cannot be said to be convincing, and he must have been drawing on his imagination when he spoke of this somewhat high hill as a lofty mountain. Moreover his account of its name only goes just far enough to be misleading: the name as pronounced in the neighbourhood of Rhayader is Corn Gafallt by Welsh-speaking people, and Corn Gavalt by monoglot Englishmen. So it is probable that at one time the pronunciation was Carn Gavall [133]. But to return to the incident recorded by Nennius, one has to remark that it does not occur in the Kulhwch; nor, seeing the position of the hill, can it have been visited by Arthur or his dog in the course of the Twrch Trwyth hunt as described by the redactor of the story in its present form. This suggests the reflection not only that the Twrch story is very old, but that it was put together by selecting certain incidents out of an indefinite number, which, taken all together, would probably have formed a network covering the whole of South Wales as far north as the boundary of the portion of Mid-Wales occupied by the Brythons before the Roman occupation. In other words, the Goidels of this country had stories current among them to explain the names of the places with which they were familiar; and it is known that was the case with the Goidels of Ireland. Witness the place-name legends known in Medieval Irish as Dindsenchas, with which the old literature of Ireland abounds. On what principle the narrator of the Kulhwch made his selection from the repertoire I cannot say; but one cannot help seeing that he takes little interest in the details, and that he shows still less insight into the etymological motif of the incidents which he mentions. However, this should be laid mainly to the charge, perhaps, of the early medieval redactor.

Among the reasons which have been suggested for the latter overlooking and effacing the play on the place-names, I have hinted that he did not always understand them, as they sometimes involved a language which may not have been his. This raises the question of translation: if the story was originally in Goidelic, what was the process by which it passed into Brythonic? Two answers suggest themselves, and the first comes to this: if the story was in writing, we may suppose a literary man to have sat down to translate it word for word from Goidelic to Brythonic, or else to adapt it in a looser fashion. In either case, one should suppose him a master of both languages, and capable of doing justice to the play on the place-names. But it is readily conceivable that the fact of his understanding both languages might lead him to miscalculate what was exactly necessary to enable a monoglot Brython to grasp his meaning clearly. Moreover, if the translator had ideas of his own as to style, he might object on principle to anything like an explanation of words being interpolated in the narrative. In short, one could see several loopholes through which a little confusion might force itself in, and prevent the monoglot reader or hearer of the translation from correctly grasping the story at all points as it was in the original. The other view, and the more natural one, as I think, is that we should postulate the interference of no special translator, but suppose the story, or rather a congeries of stories, to have been current among the natives of a certain part of South Wales, say the Loughor Valley, at a time when their language was still Goidelic, and that, as they gradually gave up Goidelic and adopted Brythonic, they retained their stories and translated the narrative, while they did not always translate the place-names occurring in that narrative. Thus, for instance, would arise the discrepancy between banw and Amanw, the latter of which to be Welsh should have been rendered y Banw, 'the Boar.' If this is approximately what took place, it is easy to conceive the possibility of many points of nicety being completely effaced in the course of such a rough process of transformation. In one or two small matters it happens that we can contrast the community as translator with the literary individual at work: I allude to the word Trwyth. That vocable was not translated, not metaphoned, if I may so term it, at all at the time: it passed, when it was still Treth-i, from Goidelic into Brythonic, and continued in use without a break; for the changes whereby Treth-i has become Trwyth have been such as other words have undergone in the course of ages, as already stated. On the other hand, the literary man who knew something of the two languages seems to have reasoned, that where a Goidelic th occurred between vowels, the correct etymological equivalent in Brythonic was t, subject to be mutated to d. So when he took the name over he metaphoned Treth-i into Tret-i, whence we have the Porcus Troit of Nennius, and Twrch Trwyd [134] in Welsh poetry: these Troit and Trwyd were the literary forms as contrasted with the popular Trwyth. Now, if my surmises as to Echel and Egel are near the truth, their history must be similar; that is to say, Echel would be the literary form and Ecel, Egel the popular one respectively of the Goidelic Ecell. A third parallel offers itself in the case of the personal name Arwyli, borne by one of Echel's companions: the Arwyl of that name has its etymological equivalent in the Arwystl- of Arwystli, the name of a district comprising the eastern slopes of Plinlimmon, and represented now by the Deanery of Arwystli. So Arwystli challenges comparison with the Irish Airgialla or Airgéill, anglicized Oriel, which denotes, roughly speaking, the modern counties of Armagh, Louth, and Monaghan. For here we have the same prefix ar placed in front of one and the same vocable, which in Welsh is gwystl, 'a hostage,' and in Irish giall, of the same meaning and origin. The reader will at once think of the same word in German as geisel, 'a hostage,' Old High German gisal. But the divergence of sound between Arwystl-i and Arwyl-i arises out of the difference of treatment of sl in Welsh and Irish. In the Brythonic district of Mid-Wales we have Arwystli with sl treated in the Brythonic way, while in Arwyli we have the combination treated in the Goidelic way, the result being left standing when the speakers of Goidelic in South Wales learnt Brythonic [135].

Careful observation may be expected to add to the number of these instructive instances. It is, however, not to be supposed that all double forms of the names in these stories are to be explained in exactly the same way. Thus, for instance, corresponding to Lug, genitive Loga, we have the two forms Lleu and Llew, of which the former alone matches the Irish. But it is to be observed that Lleu remains in some verses [136] in the story of Math, whereas in the prose he appears to be called Llew. It is not improbable that the editing which introduced Llew dates comparatively late, and that it was done by a man who was not familiar with the Venedotian place-names of which Lleu formed part, namely, Dinlleu and Nantlleu, now Dinlle and Nantlle. Similarly the two brothers, Gofannon and Amaethon, as they are called in the Mabinogi of Math and in the Kulhwch story, are found also called Gofynyon and Amathaon. The former agrees with the Irish form Goibniu, genitive Goibnenn, whereas Gofannon does not. As to Amaethon or Amathaon the Irish counterpart has, unfortunately, not been identified. Gofannon and Amaethon have the appearance of being etymologically transparent in Welsh, and they have probably been remodelled by the hand of a literary redactor. There were also two forms of the name of Manawydan in Welsh; for by the side of that there was another, namely, Manawydan, liable to be shortened to Manawyd: both occur in old Welsh poetry [137]. But manawyd or mynawyd is the Welsh word for an awl, which is significant here, as the Mabinogi called after Manawydan makes him become a shoemaker on two occasions, whence the Triads style him one of the Three golden Shoemakers of the Isle of Prydain: see the Oxford Mabinogion, p. 308.

What has happened in the way of linguistic change in one of our stories, the Kulhwch, may have happened in others, say in the four branches of the Mabinogi, namely, Pwyll, prince of Dyved; Branwen, daughter of Llyr; Math, son of Mathonwy; and Manawydan, son of Llyr. Some time ago I endeavoured to show that the principal characters in the Mabinogi of Math, namely, the sons and daughters of Dôn, are to be identified as a group with the Tuatha Dé Danann, 'Tribes of the Goddess Danu or Donu,' of Irish legend. I called attention to the identity of our Welsh Dôn with the Irish Donu, genitive Donann, Gofynion or Gofannon with Goibniu, genitive Goibnenn, and of Lleu or Llew with Lug. Since then Professor Zimmer has gone further, and suggested that the Mabinogion are of Irish origin; but that I cannot quite admit. They are of Goidelic origin, but they do not come from the Irish or the Goidels of Ireland: they come rather, as I think, from this country's Goidels, who never migrated to the sister island, but remained here eventually to adopt Brythonic speech. There is no objection, however, so far as this argument is concerned, to their being regarded as this country's Goidels descended either from native Goidels or from early Goidelic invaders from Ireland, or else partly from the one origin and partly from the other. This last is perhaps the safest view to accept as a working hypothesis. Now Professor Zimmer fixes on that of Mathonwy, among other names, as probably the Welsh adaptation of some such an Irish name as the genitive Mathgamnai [138], now anglicized Mahony. This I am also prepared to accept in the sense that the Welsh form is a loan from a Goidelic one current some time or other in this country, and represented in Irish by Mathgamnai. The preservation of Goidelic th in Mathonwy stamps it as ranking with Trwyth, Egel, and Arwyli, as contrasted with a form etymologically more correct, of which we seem to have an echo in the Breton names Madganoe and Madgone [139].

Another name which I am inclined to regard as brought in from Goidelic is that of Gilvaethwy, son of Dôn: it would seem to involve some such a word as the Irish gilla, 'a youth, an attendant or servant,' and some form of the Goidelic name Maughteus or Mochta, so that the name Gilla-mochtai meant the attendant of Mochta. This last vocable appears in Irish as the name of several saints, but previously it was probably that of some pagan god of the Goidels, and its meaning was most likely the same as that of the Irish participial mochta, which Stokes explains as 'magnified, glorified': see his Calendar of Oengus, p. ccxiv, and compare the name Mael-mochta. Adamnan, in his Vita S. Columbæ, writes the name Maucteus in the following passage, pref. ii. p. 6:--

Nam quidam proselytus Brito, homo sanctus, sancti Patricii episcopi discipulus, Maucteus nomine, ita de nostro prophetizavit Patrono, sicuti nobis ab antiquis traditum expertis compertum habetur.

This saint, who is said to have prophesied of St. Columba and died in the year 534, is described in his Life (Aug. 19) as ortus ex Britannia [140], which, coupled with Adamnan's Brito, probably refers him to Wales; but it is remarkable that nevertheless he bore the very un-Brythonic name of Mochta or Mauchta [141].

To return to the Mabinogion: I have long been inclined to identify Llwyd, son of Kilcoed, with the Irish Liath, son of Celtchar, of Cualu in the present county of Wicklow. Liath, whose name means 'grey,' is described as the comeliest youth of noble rank among the fairies of Erin; and the only time the Welsh Llwyd, whose name also means 'grey,' appears in the Mabinogion he is ascribed, not the comeliest figure, it is true, or the greatest personal beauty, but the most imposing disguise of a bishop attended by his suite: he was a great magician. The name of his father, Kil-coet, seems to me merely an inexact popular rendering of Celtchar, the name of Liath's father: at any rate one fails here to detect the touch of the skilled translator or literary redactor. [142] But the Mabinogi of Manawydan, in which Llwyd figures, is also the one in which Pryderi king of Dyfed's wife is called Kicua or Cigfa, a name which has no claim to be regarded as Brythonic. It occurs early, however, in the legendary history of Ireland: the Four Masters, under the year A.M. 2520, mention a Ciocbha as wife of a son of Parthalon; and the name seems to be related to that of a man called Cioccal, A.M. 2530. Lastly, Manawydan, from whom the Mabinogi takes its name, is called mab Llyr, 'son of Llyr,' in Welsh, and Manannán mac Lir in Irish. Similarly with his brother Brân, and his sister Branwen, except that she has not been identified in Irish story. But in Irish literature the genitive Lir, as in mac Lir, 'son of Ler,' is so common, and the nominative so rare, that Lir came to be treated in late Irish as the nominative too; but a genitive of the form Lir suggests a nominative-accusative Ler, and as a matter of fact it occurs, for instance, in the couplet:--

Fer co n-ilur gnim dar ler Labraid Luath Lam ar Claideb [143].

A man of many feats beyond sea, Labraid swift of Hand on Sword is he.

So it seems probable that the Welsh Llyr [144] is no other word than the Goidelic genitive Lir, retained in use with its pronunciation modified according to the habits of the Welsh language; and in that case [145] it forms comprehensive evidence, that the stories about the Llyr family in Welsh legend were Goidelic before they put on a Brythonic garb.

As to the Mabinogion generally, one may say that they are devoted to the fortunes chiefly of three powerful houses or groups, the children of Dôn, the children of Llyr, and Pwyll's family. This last is brought into contact with the Llyr group, which takes practically the position of superiority. Pwyll's family belonged chiefly to Dyfed; but the power and influence of the sons of Llyr had a far wider range: we find them in Anglesey, at Harlech, in Gwales or the Isle of Grasholm off Pembrokeshire, at Aber Henvelen somewhere south of the Severn Sea, and in Ireland. But the expedition to Ireland under Brân, usually called Bendigeituran, 'Brân [146] the Blessed,' proved so disastrous that the Llyr group, as a whole, disappears, making way for the children of Dôn. These last came into collision with Pwyll's son, Pryderi, in whose country Manawydan, son of Llyr, had ended his days. Pryderi, in consequence of Gwydion's deceit (pp. 69, 501, 525), makes war on Math and the children of Dôn: he falls in it, and his army gives hostages to Math. Thus after the disappearance of the sons of Llyr, the children of Dôn are found in power in their stead in North Wales [147], and that state of things corresponds closely enough to the relation between the Tuatha Dé Danann and the Lir family in Irish legend. There Lir and his family are reckoned in the number of the Tuatha Dé Danann, but within that community Lir was so powerful that it was considered but natural that he should resent a rival candidate being elected king in preference to him. So the Tuatha Dé took pains to conciliate Lir, as did also their king, who gave his daughter to Lir to wife, and when she died he gave him another of his daughters [148]; and with the treatment of her stepchildren by that deceased wife's sister begins one of the three Sorrowful Tales of Erin, known to English readers as the Fate of the Children of Lir. But the reader should observe the relative position: the Tuatha Dé remain in power, while the children of Lir belong to the past, which is also the sequence in the Mabinogion. Possibly this is not to be considered as having any significance, but it is to be borne in mind that the Lir-Llyr group is strikingly elemental in its patronymic Lir, Llyr. The nominative, as already stated, was ler, 'sea,' and so Cormac renders mac Lir by filius maris. How far we may venture to consider the sea to have been personified in this context, and how early, it is impossible to say. In any case it is deserving of notice that one group of Goidels to this day do not say mac Lir, 'son of Lir,' filium maris, but always 'son of the lir': I allude to the Gaels of the Isle of Man, in whose language Manannán mac Lir is always Mannanan mac y Lir, or as they spell it, Lear; that is to say 'Mannanan, son of the ler.' Manxmen have been used to consider Manannan their eponymous hero, and first king of their island: they call him more familiarly Mannanan beg mac y Lear, 'Little Mannanan, son of the ler'. This we may, though no Manxman of the present day attaches any meaning to the word lir or lear, interpreted as 'Little Mannanan, son of the Sea.' The wanderings at large of the children of Lir before being eclipsed by the Danann-Dôn group, remind one of the story of the labours of Hercules, where it relates that hero's adventures on his return from robbing Geryon of his cattle. Pomponius Mela, ii. 5 (p. 50), makes Hercules on that journey fight in the neighbourhood of Aries with two sons of Poseidon or Neptune, whom he calls (in the accusative) Albiona and Bergyon. To us, with our more adequate knowledge of geography, the locality and the men cannot appear the most congruous, but there can hardly be any mistake as to the two personal names being echoes of those of Albion and Iverion, Britain and Ireland.

The whole cycle of the Mabinogion must have appeared strange to the story-teller and the poet of medieval Wales, and far removed from the world in which they lived. We have possibly a trace of this feeling in the epithet hên, 'old, ancient,' given to Math in a poem in the Red Book of Hergest, where we meet with the line [149]:--

Gan uath hen gan gouannon.

With Math the ancient, with Gofannon.

Similarly in the confused list of heroes which the story-teller of the Kulhwch (Mabinogion, p. 108) was able to put together, we seem to have Gofannon, Math's relative, referred to under the designation of Gouynyon Hen, 'Gofynion the Ancient.' To these might be added others, such as Gwrbothu Hên, mentioned above, p. 531, and from another source Lleu Hen [150], 'Llew the Ancient.' So strange, probably, and so obscure did some of the contents of the stories themselves seem to the story-tellers, that they may be now and then suspected of having effaced some of the features which it would have interested us to find preserved. This state of things brings back to my mind words of Matthew Arnold's, to which I had the pleasure of listening more years ago than I care to remember. He was lecturing at Oxford on Celtic literature, and observing 'how evidently the mediæval story-teller is pillaging an antiquity of which he does not fully possess the secret; he is like a peasant,' Matthew Arnold went on to say, 'building his hut on the site of Halicarnassus or Ephesus; he builds, but what he builds is full of materials of which he knows not the history, or knows by a glimmering tradition merely--stones "not of this building," but of an older architecture, greater, cunninger, more majestical. In the mediæval stories of no Latin or Teutonic people does this strike one as in those of the Welsh.' This becomes intelligible only on the theory of the stories having been in Goidelic before they put on a Welsh dress.

When saying that the Mabinogion and some of the stories contained in the Kulhwch, such as the Hunting of Twrch Trwyth, were Goidelic before they became Brythonic, I wish to be understood to use the word Goidelic in a qualified sense. For till the Brythons came, the Goidels were, I take it, the ruling race in most of the southern half of Britain, with the natives as their subjects, except in so far as that statement has to be limited by the fact, that we do not know how far they and the natives had been amalgamating together. In any case, the hostile advent of another race, the Brythons, would probably tend to hasten the process of amalgamation. That being so, the stories which I have loosely called Goidelic may have been largely aboriginal in point of origin, and by that I mean native, pre-Celtic and non-Aryan. It comes to this, then: we cannot say for certain whose creation Brân, for instance, should be considered to have been--that of Goidels or of non-Aryan natives. He sat, as the Mabinogi of Branwen describes him, on the rock of Harlech, a figure too colossal for any house to contain or any ship to carry. This would seem to challenge comparison with Cernunnos, the squatting god of ancient Gaul, around whom the other gods appear as mere striplings, as proved by the monumental representations in point. In these [151] he sometimes appears antlered like a stag; sometimes he is provided either with three normal heads or with one head furnished with three faces; and sometimes he is reduced to a head provided with no body, which reminds one of Brân, who, when he had been rid of his body in consequence of a poisoned wound inflicted on him in his foot in the slaughter of the Meal-bag Pavilion, was reduced to the Urdawl Ben, 'Venerable or Dignified Head,' mentioned in the Mabinogi of Branwen [152]. The Mabinogi goes on to relate how Brân's companions began to enjoy, subject to certain conditions, his 'Venerable Head's' society, which involved banquets of a fabulous duration and of a nature not readily to be surpassed by those around the Holy Grail. In fact here we have beyond all doubt one of the heathen originals of which the Grail is a Christian version. But the multiplicity of faces or heads of the Gaulish divinity find their analogues in a direction hitherto unnoticed as far as I know, namely, among the Letto-Slavic peoples of the Baltic sea-board. Thus the image of Svatovit in the island of Rügen is said to have had four faces [153]; and the life of Otto of Bamberg relates [154] how that high-handed evangelist proceeded to convert the ancient Prussians to Christianity. Among other things we are told how he found at Stettin an idol called Triglaus, a word referring to the three heads for which the god was remarkable. The saint took possession of the image and hewed away the body, reserving for himself the three heads, which are represented adhering together, forming one piece. This he sent as a trophy to Rome, and in Rome it may be still. Were it perchance to be found, it might be expected to show a close resemblance to the tricephal of the Gaulish altar found at Beaune in Burgundy.

Before closing this chapter a word may be permitted as to the Goidelic element in the history of Wales: it will come again before the reader in a later chapter, but what has already been advanced or implied concerning it may here be recapitulated as follows:--

It has been suggested that the hereditary dislike of the Brython for the Goidel argues their having formerly lived in close proximity to one another: see p. 473 above.

The tradition that the cave treasures of the Snowdon district belong by right to the Goidels, means that they were formerly supposed to have hidden them away when hard pressed by the Brythons: see pp. 471-2 above.

The sundry instances of a pair of names for a single person or place, one Goidelic (Brythonicized) still in use, and the other Brythonic (suggested by the Goidelic one), literary mostly and obsolete, go to prove that the Goidels were not expelled, but allowed to remain to adopt Brythonic speech.

Evidence of the indebtedness of story-tellers in Wales to their brethren of the same profession in Ireland is comparatively scarce; and almost in every instance of recent research establishing a connexion between topics or incidents in the Arthurian romances and the native literature of Ireland, the direct contact may be assumed to have been with the folklore and legend of the Goidelic inhabitants of Wales, whether before or after their change of language.

Probably the folklore and mythology of the Goidels of Wales and of Ireland were in the mass much the same, though in some instances they reach us in different stages of development: thus in such a case as that of Dôn and Danu (genitive Danann) the Welsh allusions in point refer to Dôn at a conspicuously earlier stage of her rôle than that represented by the Irish literature touching the Tuatha Dé Danann [155].

The common point of view from which our ancestors liked to look at the scenery around them is well illustrated by the fondness of the Goidel, in Wales and Ireland alike, for incidents to explain his place-names. He required the topography--indeed he requires it still, and hence the activity of the local etymologist--to connote story or history: he must have something that will impart the cold light of physical nature, river and lake, moor and mountain, a warmer tint, a dash of the pathetic element, a touch of the human, borrowed from the light and shade of the world of imagination and fancy in which he lives and dreams.