kimkat0374k Lectures on Welsh Philology. 1877. John Rhys (1840-1915).
30-01-2018

● kimkat0001 Yr Hafan www.kimkat.org
● ● kimkat2001k Y Fynedfa Gymraeg www.kimkat.org/amryw/1_gwefan/gwefan_arweinlen_2001k.htm
● ● ● kimkat2194ke Cyfeirddalen yr Adran Ramadeg http://kimkat.org/amryw/1_gramadeg/gramadeg_cyfeirddalen_2194k.htm
● ● ● ● kimkat0369k  Lectures On Welsh Philology – Y Gyfeirddalen
www.kimkat.org/amryw/1_testunau/testunau-saesneg_welsh-philology_john-rhys_1877_0373k.htm
● ● ● ● ●  kimkat0374k Y tudalen hwn
...

 

0003g_delw_baneri_cymru_catalonia_050111
 (delw 0003)
 
 
 
 

Gwefan Cymru-Catalonia
La Web de Gal
·les i Catalunya
The Wales-Catalonia Website

 

Lectures on Welsh Philology. 1877.
John Rhys (1840-1915).



Y Llyfr Ymwelwyr / El Llibre de Visitants / The Guestbook:
http://pub5.bravenet.com/guestbook/391211408/


a-7000_kimkat1356k
Beth sy’n newydd yn y wefan hon?

 
---

6665_map_cymru_catalonia_llanffynhonwen_chirbury_070404
(delwedd 6665)

...
llythrennau cochion = testun heb ei gywiro
llythrennau duon = testun wedi ei gywiro

 

 


(delwedd B5991)

LECTURES ON WELSH PHILOLOGY.


BY JOHN RHYS, M.A.,
LATE FELLOW OF MERTON COLL., OXFORD,
PERPETUAL MEMBER OF THE PARIS PHILOLOGICAL SOCIETY.

LONDON: TάBNER & CO., LUDGATE HILL.
1877. [All rights reserved.]






 



(de
lwedd B6300) (tudalen 300)

300 LECTUEES ON WELSH PHILOLOGY. Ogam on the stone now occupying our attention is to be regarded as making one name Awwiboddib- or Awwi Boddib-, it must mean ' Nepotis Bodi-bevi.' The only thing which prevents me from reading the whole thus: Bewwlf] Awwi Boddi-blewwi], " B. nepotis Bodibevi," is the fact that it is not usual to begin with the right edge; but that is perhaps not a sufficient reason for not doing so here. This remarkable stone, then, commemorates either two or three distinct persons, who are shown, however, to have belonged to the same family by the name-element bev or bewm. 17. Brecknockshire. — A stone now standing near Sir Joseph Bailey's residence in Glan Usk Park, near Crickhowel, reads in Ogam: — ' " 1 1 1 /////x I I -n II I II I =- T u rpil... ...1 u n i which may be restored as meaning Turpilli [maqvi] Trilluni, seeing that the Latin reads Turpilli Ic Jacit Puveri Triluni Bunocati. 18. A stone preserved in Trallong Church in the neighbourhood of Brecon reads in Ogam: — Cuuace n n i wi + +^+H-7T-T^-T^^^^^^-^-++ I 1 w w e to 



 



(de
lwedd B6301) (tudalen 301)

LECTURE VI. 301 The Latin reads: — CVNOCENNI FILIVS CVNOCENI HIC lACIT, whence it would seem that Cunacennini is a kind of patronymic meaning C. filius C, and that Ilwmeto is an epithet. The broader end of the stone hears a cross enclosed, excepting the shaft, in a circle. 19. Glamorganshire. — On the roadside between Margam and Cynffig stands a stone which reads: — PYiTPEIVS CARANTORITS. The Ogam begins near the top on the right edge and reads: —  P[o]p e ... which appears to make Pope; but one cannot go further with any certainty of being right, as the original number of vowel notches terminating the name cannot now be determined; but they seem to have been between seven and ten, and it may be supposed that the name was Popei or Popeu. Both Popei and Punpeius are forms of the more usual Pompeius, and the explanation of them is to be sought in Latin, as was pointed out in the previous lecture. The character here guessed to 



 



(de
lwedd B6302) (tudalen 302)

302 LECTURES ON WELSH PHILOLOGY. mean p has not been met with elsewhere. The Ogam occupying the length of the right edge is too far gone to be deciphered; it seems, read downwards, to show the digits standing for — r — I — sm — qv — II —?z..., which, if read upwards, would make ...c — dd — n — mc — d — r.,.. On the whole I am inclined to think that all the Ogams formed one inscription continued round the top of the stone, where now, it is true, there is no trace of a letter. The stone now stands erect, but it has not always been so, if I am right in thinking that what is now the top has been worn smooth by the tread of feet. 20. The Eoman altar at Loughor, the Cas Llychwr of the "Welsh, and, according to some, the Leucarum of the Eomans, bears an Ogmic inscription which is, unfortunately, almost entirely illegible, excepting the last two groups' of digits, which make ic. Various guesses may be given, the two extremes of which would be Lekuric and Vehomagic, or, as I would put them, Lehuri C. and Vehomagi C. If the c stood for a word, the inscription was probably in Latin; but the altar shows no trace of any other letters than Ogams. 21. Devonshire. — A stone taken from Fardel, near Ivybridge, and deposited in the British Museum, has on it three different inscriptions, two in Eoman capitals more or less debased, and 



 



(de
lwedd B6303) (tudalen 303)

LECTURE TI. 303 one in Ogam, to which repeated reference has been made — it reads upwards on both edges: — nil III! 'I' ll III I I I! mil - Swaq qvuc i / I mil inn m" mi l n" i ii ii -^ M a qv i Qv i c i The Koman letters on the face bounded by these edges read: — FANONI MAQVIRINI. The third inscription is on another face, and consists of the name Sagranui in letters which are considerably later than the foregoing ones, the r, especially, being of the early Kymric type and the n formed like an h. 22. One of the three tombstones at Tavistock was brought thither from a place not very far off called Buckland Monachorum: it reads in Eoman capitals: — DOBVNNI FABEI FIl[l?]i ENABARRI. This explains the only portion of the Ogmic inscription still legible: -rnrr-^n-^ ///// ///// n aba r r 23. Cornwall. — A stone on Worthyvale farm, in 



 



(de
lwedd B6304) (tudalen 304)

304 LECTURES ON WELSH PHILOLOGY. the neighbourhood of Camelford, shows traces of an Ogmic inscription ending in 1 1 1 1 1 , i: the preceding letter is rather doubtful, but it may be an r. The other inscription is in debased- Eoman capitals with one or two Kymric letters intermixed, especially s and m: — LATINI 10 lACIT FILIUS MA...ABII, Let us now return to the Ogam alphabet and try to force it to tell its own history. In one of the Irish alphabets, which have evidently been based on it, the letters had the following names, which I copy from O'Donovan's Irish Grammar, p. xxxii.: — B ieith, the birch. M muin, the vine. 1 luis, the mountain ash. g gort, ivy. f fearn, the alder. ng ngedal, the reed. s sail, the willow. st or z straif, the sloe-tree. n nion, the ash. r ruis, the elder. H huath, the hawthorn. A ailm, the fir-tree. d duir, the oak. o onn, furze. t tinne (unknown). u ur, heath. c coll, hazel. e eadhadh, the aspen. q queirt, the apple-tree. i idkadh, the yew. This is the Bethluisnion alphabet, so called from its first letters: in another the letters are called after Biblical names, of which the first two are Bobel and Loth, whence it is called the Bobelloth alphabet. Consider now for a moment the cha- 



 



(de
lwedd B6305) (tudalen 305)

LECTURE VI. 305 racter of the four groups into which Irish tradition was wont to divide the letters: — i- i II III nil m il 3, 7 // /// //// m B, 1, w, s, n. M, g, lig, z, r. 2. I II III nil Hill 4. +^- 1 1 1 nil I W+ Ch, d, t, o, qv. A, o, u, e, i. It is highly improbable that this grouping can be as old as the alphabet itself; for it is not much of an attempt to classify the sounds indicated, while it is a classification of the symbols used. The sort of arrangement which it presupposes was, I conjecture, the following or some other one nearly resembling it: — I I ' / II 11.," // III III ' " / / / UN - a, b, chjin, o, 1, d, g, u, w, t, Dg, e, MM "" ////m i l m il '"" ///// s, c, z, I, n, qv, r. This conjecture is, I must tell you in passing, the most important of a good many which I am going to submit to you in this and the next lecture, and with it would fall most of my conclusions with respect to , the origin of Ogmic writing. If this is borne in mind, it will be needless for me to repeat it as we proceed. ' If you look again at the different kinds of digits, the question may occur to you, why the long ones are not allowed to cross the edge of the stone written upon at right angles. Now it is not im- u 



 



(de
lwedd B6306) (tudalen 306)

306 LECTURES ON WELSH PHILOLOGY, probable that, at one time, tbe vowels were of the description here suggested and not mere notches.* It is so at ^ny rate in one class of Irish Ogams, which are not, it is true, attested by the oldest monuments: still it may be that this peculiarity they show comes down from much earlier times. In them a would be not + but |, which would render it necessary to write m -f, and so with the other four. All this points to the conclusion that the oblique group is of later date than the other three, and the order last given may be allowed to give way to the following: — I , 1 II ,11 I II ,,, III nil ,,,, nil ' I " II "I III I'" III! a, b, ch, o, 1, d, u, w, t, e, s, o, m il Hil l '"" III III nil mil 1, n, qv, m, g, ng, z, r. There are other reasons for supposing the oblique group merely supplementary to the others: thus /// for ng dates probably after -f-j; g, and is formed from it by adding a score; but it must have been settled before ///// was hit upon for r, otherwise nobody would have thought of representing by means of the most cumbrous symbol in the alphabet the consonant which of all others is the one most frequently used in Welsh; and it is hardly otherwise in the case of the other Celtic * It is right the reader should know that the Ogams for the vowels in this volume are represented as much longer than, in strict proportion to the consonantal digits, they should be.^ 



 



(de
lwedd B6307) (tudalen 307)

LECTURE VI. 307 tongues. Hence it follows that ng, z, r, only got to be written j-j-j-, -j-fjj, ///// by way of addition to, or readjustment of the alphabet as previously used. Further, as the Ogam in one of the orders it admits of begins with + («), j (b), which may be treated as the equivalents in it of aleph, beth, or a, /3, we may go further and assume -'- {ch) to be, for some reason or other, the Ogmic equivalent of gimmel or y: this is confirmed by the fact of g appearing as -fj- in the later group, which suggests the same sort of relation between -•- and jj- as between the Latin letters C and G. Now, treating +, ■]-, -•-, as the historical equivalents of aleph, beth, gimmel, the Ogmic alphabet may be said to have coincided with the Semitic alphabet in its first three letters, excepting that the Irish grouping does not enable us to decide which of the six sequences — a, b, ch: a, ch, b: b, a, ch: b, ch, a: ch, a, b: ch, b, a — was the one adopted in the Ogmic system. Is this coincidence, it may be asked, purely accidental, or does it tend to prove that the framers of the Ogam were acquainted with some one or more alphabets of Phoenician origin? The answer to this question is to be sought in the number of combinations, as mathematicians term it, which the letters of the Ogam alphabet admit of when taken three and three together. But as the long group does not appear to have belonged to the 



 



(de
lwedd B6308) (tudalen 308)

308 LECTURES ON WELSH PHILOLOGY. alphabet in its earliest form, we can only calculate on the remaining fifteen letters. Now the number of permutations which fifteen letters admit of when taken by threes is 2730, which, divided by six, gives us the number of combinations as 455; that is, the chances against the coincidence being accidental are 454 to 1. But, to be on the safe side, let us discard -LU-Li, qn^ as being possibly a later addition to complete the scheme. The letters then are fourteen, which, taken by threes, admit of 364 combinations; and this reduces the chances to 363 to 1. But some writers appear to believe that it is, somehow, natural for alphabets to begin as the Semitic ones are found to do. Now these last begin with aleph, a consonant which a European would probably not have honoured with a place in an alphabet at all. If, however, it is our European a that nature intended to take the lead, the Shemites failed to obey the promptings of nature on this point: the same applies with still more force to the Irish, when they put together the Bethluisnion alphabet, and the Teutons, whose Kunic alphabets are found to begin withy, m, th, a, r, k, although the symbols for them were borrowed from the Latin alphabet, which did begin with A. Thus the facts within our reach seem to warrant our leaving out of the reckoning the alleged naturalness in question, so that, when it is found that the 



 



(de
lwedd B6309) (tudalen 309)

LECTURE VI. 309 chances are over 300 to 1 against the coincidence being accidental, it is highly probable that the framers -of the Ogam alphabet were acquainted with the Phoenician or some one deriyed from it. This being so, it is also probable that the sequence of the first three letters in the Ogam was no other than a, 6, ch, as in the trial alphabet mentioned above: — I , I I I ,, II III ,,, III nil ,,,, Mil I I II II III III i"i nil a, b, ch, 0, 1, d, u, w, t, e, s, c, M il l inn ' "" I II lli ll ll mil i, n, qv, m, g, ng, ^ A little further scrutiny of this last arrangement leads one to observe the apparently artificial quartering of the vowels in places 1, 4, 7, 10, 13. So, to get at the sequence which preceded this, we should, among other things, have to expel the vowel from its present position, which would admit the d to advance and the m to return from the supplementary group to the place which it probably occupied before it was relegated there. We should then have the following: — I I " III III ' " ' I II a, b, ch, d, 1, m, u, w, t, e, I II iini II 1 1 III" II 1 1 1 a, e, i, n, qv. Thus we seem to get a glimpse into the history of the changes which the Ogam alphabet has under- 



 



(de
lwedd B6310) (tudalen 310)

310 LECTUEES ON WELSH PHILOLOftT. gone, at the same time that, by restoring d to what was probably its old place, we nearly triple our former estimate of the probabilities of the case, the chances now being (without taking the sequence I m into account) exactly 1000 to 1 in favour of the supposition that the Ogam alphabet is connected with the Phoenician. So far as we have gone, the connection seems to amount to this: — 1. The framers of the Ogam alphabet did not take up all the Phoenician letters, but only about 14 or 15 of them. 2. These they took in their order in the Phoenician alphabet. 3. They translated the Semitic characters into straight lines, probably because they found them easier to cut on wood, which, it may be presumed, was the material which they mostly used to write upon, but chiefly, perhaps, because they may have already been in the habit of cutting scores resembling Ogmic digits on wood, horn, or bone. Such scoring, considered as mere scoring or carving, and without reference to its meaning, has been traced so far back in Europe as the quaternary period and the end of the mammoth age: a specimen from the sepulchral cave of Aurignac is described by M. FranQois Lenormant in the second edition of his Essai sur la Propagation de V Alphabet PMnicien dans VAncien Monde (Paris, 1875), i. 7, 8. So far no attempt has here been 



 



(de
lwedd B6311) (tudalen 311)

LECTURE VI. 311 made to show with which of the Phoenician alphabets, that is the Phoenician alphabet properly so called, or some one of those of Greece or Italy which have been traced to it, the Ogam is connected. History and geography do not encourage ■one to expect to find any immediate connection between the Ogam and the alphabets of Greece: the ordinary Roman alphabet hardly suits, as it has only the one symbol v for u and?», not to mention other reasons which might be adduced: similarly we might go on excluding the Etruscan and Runic alphabets. For the present, then, we shall rest content with the bare fact, that the Ogam is in a manner derived from the Phoenician alphabet, without proceeding to attempt to trace the connection between them step by step. The rest of this lecture will, accordingly, be devoted to a brief mention of some of the Goidelo-Kymric traditions bearing on the origin of writing among the Celts. The allusions in Irish literature to the Ogam are various and numerous, and a succinct account of the grammatical treatises, which deal with it, will be found in the following paragraph quoted from an abstract of a paper read before the Royal Irish Academy in 1848 by Prof. Graves, now Bishop of Limerick: — " The Book of Leinster, a MS. of the middle of the 12th century, contains 



 



(de
lwedd B6312) (tudalen 312)

312 LECTURE,S ON WELSH PHILOLOGY. a passage in which it [the key to the Ogam] ia briefly given. The Book of Ballymote, written about the year 1370, contains an elaborate tract, which furnishes us with the keys to the ordinary Ogham, and a vast variety of ciphers, all formed on the same principle. The Book of Lecan (written in the year 1417) contains a copy of the Uraicept, a grammatical tract, perhaps, as old as the 9th century, in which are many passages relating to the Ogham alphabet, and all agreeing, as regards the powers of the characters, with what is laid down in the treatise on Oghams in the Book of Ballymote. Dr. O'Connor, indeed, speaks of a manuscript book of Oghams written in the 11th century, and once in the possession of Sir James Ware. Mr. Graves has ascertained that this is merely a fragment of the above-mentioned Ogham tract. It is now preserved in the library of the British Museum, and does not appear to have b,een written earlier than the 15th or 16 th century." Some valuable extracts from, and fac-similes of the Ballymote tract have lately been published by Mr. G. M. Atkinson in the Journal of the Kilkenny Archceological Society (vol. iii. pp. 202-236), to which we shall have occasion to refer more than once. There, in answer to the question, " By whom and from whence are the veins and beams in the Ogaim tree named? " 



 



(de
lwedd B6313) (tudalen 313)

LECTURE TI. 313 we have the curious reply: — " Per alios. It came from the school of Phenius, a man of Sidon, viz., schools of philosophy under Phenius throughout the world, teaching the tongues (he thus employed), in numher 25." But, to pass by the other traditions respecting this early Fenian, we come to Ogma, who is said to have been the inventor of the Ogam, and from whom it is called Ogam, also Ogum, and, in later Irish, Ogham with a silent gh. Ogma is described as the son of Elathan of the race of the Tuatha de Danann, whence it is clear that he is as mythical a personage as Irish legend could well make him. And from his being called, as appears from Mr. Atkinson's paper, Ogma the Sun-faced, it seems probable that he was of solar origin. Ogma being much skilled in dialects and in poetry, it was he, we are told, who invented the Ogam to provide signs for secret speech only known to the learned, and designed to be kept from the vulgar and poor of the nation. . For not only was a system of writing called Ogam, but also a dialect, or mode of speech, bears that name. Of this O'MoUoy, cited in the preface to O'Donovan's Irish Grammar, p. xlviii., says: " Obscurum loquendi modum, vulgo Ogham, antiquariis Hiberniae satis notum, quo nimirum loquebantur syllabizando voculas appellationibus litterarum, dipthongorum, et triphthongorum ipsis dumtaxat notis." O'Dono- 



 



(de
lwedd B6314) (tudalen 314)

314 LECTURES ON WELSH PHILOLOGY. van further quotes an entry in the Annals of Olonmacnoise to the following effect, as translated, in 1627, by Connell Mageoghegan:— " a.d. 1328. Morish O'Gibelan, Master of art, one exceeding well learned in the new and old laws, civille and cannon, a cunning and skillfull philosopher, an excellent poet in Irish, an eloquent and exact speaker of the speech, which in Irish is called Ogham, and one that was well seen in many other good sciences: he was a canon and singer at Twayme, Olfyn, Aghaconary, Killalye, Enaghdown, and Clonfert; he was official and common judge of these dioceses; ended his life this year." To pass by, for the present, the motive attributed to Ogma in his invention, we seem to find him here in the character of the man of letters, and this is quite in harmony with the only trace of his footsteps which has been discovered on Kymric ground, namely, in the Welsh derivative ofydd, which probably stands for an earlier omUS = ogmi^, and seems to have formerly meant a man of science and letters; now it is defined to be an Eisteddfodic graduate who is neither bard nor druid, and translated into ovate. Thus, perhaps, it would be no overhasty generalising to infer that with the insular Celts Ogma's province was language as literature, as the record of the past and the repository of knowledge. The Gauls, on 



 



(de
lwedd B6315) (tudalen 315)

LECTURE VI. 315 the other hand, looked at their Ogmius, according to Lucian's account, from the point of view of language as the means of persuasion; for they represented him as an extremely old man drawing after him a crowd of willing followers by means of tiny chains connecting their ears with the tip of his tongue. Otherwise, be it observed, he seems to have had the ordinary attributes of Hercules, whence it would seem that he, like his Goidelic namesake, was of solar origin. It is probable, therefore, that his influence over the crowds who rejoiced to follow him was in the first instance due, not to his oratorical skill, the sweetness of his voice, or his power of persuasion, but to the contents of his words, to the wisdom he had to impart, and the wonderful experiences he could relate. . How could it be otherwise in the case of one — to borrow the words applied in the Odyssey to the sun — "^0? TravT e<j)opa km iravr eiraKovei? The Irish were perhaps alone in attributing to him the origin of letters and the cultivation of a dialect not understood by the people: at any rate Welsh tradition would seem to point in quite another direction. But it is hardly necessary to state that, owing to the Ogam having got out of use in the West of Britain as early as the 8th or 9th century. 



 



(de
lwedd B6316) (tudalen 316)

316 LECTURES ON "WELSH PHILOLOGY. the allusions to it in Welsh literature are exceedingly faint and nebulous. It may possibly be proved that those about to be here mentioned do not in any way refer to the Ogam; but the point I wish to insist upon is that they agree with Irish tradition in placing the origin of writing — whether Ogmic or other — before the Christian era. In the lolo MSS. (pp. 20'3-206), there are a few paragraphs on the Welsh alphabet from manuscripts supposed to be traceable to the possession of Llewelyn Sion, a Glamorganshire bard and collector of antiquities, who died in the year 1616. Certainly there seems to be no reason to think that they are, in the shape in which we find them, of an earlier date; but that does not prove them not to contain a slender element of ancient tradition beneath the incrustations of later times, and in spite of their evident reference, in the first instance, to the bardic alphabet called Coelbren y Beir.dd, which may be briefly characterised as the form the Eoman alphabet took when carved on wood by the Welsh in the 15th century: see Stephens's essay on the subject in the^rc^ Cambrensis for 1872, pp. 181-210. One of these paragraphs runs thus: " In the time of Owain ap Maxen Wledig the race of the Cymry recovered their privileges and crown: they took to their original, mother-tongue instead of the Latin, which had well-nigh overrun the Isle of Britain, 



 



(de
lwedd B6317) (tudalen 317)

LECTURE VI. 31 and in Welsh they kept the history, records, an classifications of country and nation, restoring t memory the ancient Oymraeg, their original wore and idioms. Owing, however, to their forgettin and misunderstanding the old orthography of tl ten primary letters they fell into error, and thi arose a disagreement as to [the spelling of] severs ancient words." The writer goes on to give ir stances which show that the latter part of tl passage is a mere corollary to the preceding par and applicable to nothing earlier than the numeroi foibles of Welsh orthography in the Middle Age Another of the paragraphs alluded to is to the fo lowing effect: " Before the time of Beli the Gres ap Manogan there were but ten letters, and the were called the ten awgrym, namely, a, p, c, ( t, i, 1, r, 0, s: afterwards m and n were discoverec and afterwards four others, so that now being sij teen they were established with the publicity an sanction of state and nation. After the coming ( the faith in Christ two other letters were adde( namely, u and ^, and in the time of King Arthi there were fixed twenty primary letters, as at pr( sent, by the advice of Taliesin Benbeirdd, Urie Rheged's domestic bard. It was according \ the alphabet of the eighteen that was arrange OIU, that is, the unutterable name of God: b< fore that system it was 010 according to the si2 



 



(de
lwedd B6318) (tudalen 318)

318 LECTURES ON WELSH PHILOLOGY. teen. Of principal awgrymau there are not to the present day more than twenty letters or twenty awgrymy The writer dwells on the repeated additions made to the alphabet, and the numbers he gives at successive stages are 10, 12, 16, 18, 20, which are clearly not all to be taken au pied de la lettre; for national sanction is not mentioned by him till we come to the alphabet of 16; and to what Aryan alphabet could 10 and 12 apply? He has supplied us with the key to his blundering in the word awgrym (now 'a hint or suggestion,' plural awgrymau), which is simply the 0. English word awgrim, augrim, algrim, borrowed. Now the Craft of Algrim was arithmetic (on the history of the word, see Max Miiller's Lectures^ ii. p. 300, 301), and it is clear that he has set off his account of the alphabet by a strange attempt to base it on the decimal system of numeration. It is not to be forgotten that Llewelyn Sion had probably heard of the algebraists and arithmeticians Vieta, Harriot, Wright, and Napier. Perhaps it is in the same direction we should look for the explanation of the mystic 010. In another version the arithmetical and alphabetical elements are kept somewhat more apart, the former showing an inveterate tendency to secrecy, which is not so evident in the 



 



(de
lwedd B6319) (tudalen 319)

LECTURE VI. 319 case of tlie latter: " la tine principal times of the race of the Cymry the letters were called ystorrynau [supposed to mean cuttings; but if cuttings, ^hj jaot fractions Pi: after the time of Beli ap Manogan they were called letters, and before that there were only the ten primary ystorryn, which had been a secret from everlasting with the bards of the Isle of Britain for the preservation of record of country and nation. But Beli the Great made them sixteen, and subject to that arrangement he made them public, causing that thenceforth -there should never be secrecy with regard to the knowledge of the letters, subject to the arrangement which he had made touching them, while he left the ten ystorryn under secrecy. After the coming of the faith in Christ the letters were made eighteen, and afterwards twenty, and so they were retained to the time of Geraint Fardd Glas, who fixed them at twenty-four." The next extract is from a document on Bardism cited by Mr. D. Silvan Evans in Skene's Four Ancient Books of Wales (ii, 324): he assigns it to the end of the 15th century, and gives references which will here be utilised. The passage in point is not very lucid, but it seems to mean this: " The three elemeiits of a letter are /|\, since it is in the presence of one or other of the three 



 



(de
lwedd B6320) (tudalen 320)

320 LECTUKES ON WELSH PHILOLOGY. a letter consists; they are three beams of light, and it is of them are formed the sixteen ogyrvens, that is, the sixteen letters. Belonging to another art also there are seven score and seven ogyrvens, which are no other than the symbols of the seven score and seven Welsh parent-words, whence every other word." The /|\ would be a correct analysis of the letters of nations who habitually wrote on slips of wood, as the nature of that material would compel one to avoid the use of curves and horizontal lines: thus it would apply to Ogams and Eunes as well as to the Coelbren y Beirdd, which the writer decidedly had in view. The three beams of light was an after-thought, or a bit of another tradition; but what mostly interests me in this extract is the word ogyrven. The sixteen ogyrvens are evidently the same as the sixteen letters of the previous extracts; but the seven score and seven seem to refer to some theory of root-words, and their number was not, as might be expected, very definite; for, to go still further back, in a passage in the Book of Taliessin, a manuscript of the 14th century, they are given as exactly seven score (Skene, ii. 132, 325):— "^eith vgein ogyruen , Yssyd yn awen" i.e., there are in awen [muse, poetry] seven score 



 



(de
lwedd B6321) (tudalen 321)

LECTUKE VI. 32 Ogyrvens. The two kinds of Ogyrvens woul seem to match the Ogam alphabet and the Ogai dialect of Irish tradition, but what is more remart able is that Ogyrven is the name of a person, an a person not a whit less mythical than Ogmj He is variously called Ogyrven, Ogynven, Ogyrfai and (with the prefixed g of late Welsh) Gogyrfai as in a popular rhyme referring to bis daughte Gwenbwyfar, Arthur's wife: — " Gwenliwyfar f erch Ogyrfan gawr, Drwg yn feohan, gwaetli yn fawr." Gwinevere, giant Ogyrvan's daughter, Naughty young, more naughty after. He is better known in Welsh poetry in connec tion witb Ceridwen, the lady who owned tl cauldron of sciences (jpair gwybodau), and whos inspiring aid Welsh poets are still supposed t invoke: thus in two of the poems in the Blac Book of Carmarthen, a manuscript of the 12t century, we meet with a formula of invocation i which she is called (Skene ii. 6, 6) Ogyrve amhad, which is supposed to mean " Ogyrven offspring." They are also associated in severs poems in tbe Book of Taliessin (Skene ii. 15' 156), and in one of the instances Ceridwen cauldron is called Ogyrven's: — " Ban pan doeth o peir \ig[ When up the Muses three Ogyrwen awen teir:" ) ' ( From Ogyrven's cauldron can X 



 



(de
lwedd B6322) (tudalen 322)

322 LECTURES ON WELSH PHILOLOGr. However, Mr. Silvan Evans translates it " High when came from the cauldron the three awens of Qogyrwen." The difference is immaterial here, as he calls attention to a poem of Cynddelw's where Ceridwen and Ogyrven are associated by the poet — he flourished in the 12th century — who calls himself a " bard of the bards of Ogyruen," with, probably, the same meaning as though he had said " of Ceridwen:" see the Mi/v. Arch, of Wales, p. 167 of Gee's edition (Denbigh, 1870). To project this on the solar myth theory, Gwenhwyfar and Ceridwen are dawn-goddesses, and their father Ogyrven must be the personification of night and darkness; and this is confirmed by the etymology of the word Ogyrven, which would have been in 0. . Welsh probably Ocrmen, divisible into Ocr-men. The first element ocr seems to have been meant in the Luxembourg Folio, where atrocia is" glossed arotrion, which appears to be a clerical error for arocrion, if that indeed be not the correct reading. Now, just as Welsh ac, oc, ' and, with,' stand with respect to such words as Greek ayxa>, Latin angustus, German eng, so ocr, ogr, stand to the words which Fick, in his dictionary^ (p. 9), derives from anghra, such as Zend angra, ' evil,' anra, ' evil, bad: ' for a few parallels see the Eevtie Celtique, 



 



(de
lwedd B6323) (tudalen 323)

LECTURE VI. 323 ii. 190, The other part occurs also in tynghedfen, a word which is used as a synonym of the simpler tynghed, 'fate, destiny.* The former was probably at one time meant to express the personification more clearly than the latter, though it does so no longer. The men (mutated fen or ven) in question can hardly be of a different origin from the English verb to mean and its congeners, among which may be mentioned Greek fievos, Sanskrit manas, ' courage, sense,' manyus, ' courage, zeal, anger, rage,' Zend mainyu, ' spirit, sky.' This last qualified by anra, ' evil, bad,' makes in the nominative anro mainyus (Justi), ' the evil spirit par excellence, Ahriman, or the devil of the Persians and the great adversary of Ormuzd.' Thus our Ogyrven seems to be almost the literal counterpart of Ahriman, and might be rendered the evil spirit: Ogyrwen, if not a mere phonetic variation, would be he of the evil smile, while Ogyrfan shows the same element fan (for man) as in Cadfan, on an early inscribed stone Catamanus. In both it is probably of the same origin and meaning as the English word man, so that Ogyrfan would have meant the evil man, and even now we call the devil y gwr drwg, ' the bad man.' His attributes are, unfortunately, so weather-worn that Welsh literature hardly enables us to make them out, which is, perhaps, partly 



 



(de
lwedd B6324) (tudalen 324)

324 LECTURES ON WELSH PHILOLOGY. due to His having been dethroned by the devil of the Bible, and partly to his connection with Ceridwen and Gwenhwyfar. But a clue to them appears to be oifered us in another form of his name: in Gee's Myv. Arch, of Wales, p. 396, it is Ocurvran, that is in later spelling Ogyrfran, which would mean the evil crow, and suggests a community of origin with the Irish Badb: see Mr. Hennessy's article on the latter in the Revue Celtique, i. 32-57. The Badb is described as having the form of a crow and as a bird of ill omen, confounding armies, impelling to slaughter, and revelling among the slain. This will serve as a provisional key to the meaning of a reference to Ogyrven in one of the poems in the Black Book already alluded to: the lines are very obscure and run thus (Skene, ii. 6): " Ry hait itaut. rycheidv y naut. rao caut gelin. Ey chedwis detyf. ry chynis gretyw. rac llety w ogyrven.'' The meaning is by no means clear, but " rac caut gelin^'' which cannot but mean " against the insult of an enemy," suggests that its parallel in the following line, rac lletyw ogyrven, must be "against a sinister fate," or something nearly approaching it, as indicated by the adjective lletyw, now written lleddf. Similarly we are enabled to guess what Cynddelw meant {Myv. Arch, of Wales, p. 154) when he praises a certain 



 



(de
lwedd B6325) (tudalen 325)

LECTUEB VI. 32 man as being " a hero of the valour of Ogyrfan, gwron gnryd Ogyrfan, where Ogyrfan seems 1 mean war and slaughter, probably personified. In support of this view of Ogyrfen, we hav( besides tynhedfen, a third compound, namely Aei fen, which, as aer is battle, war, must mean spirit or divinity concerned with war: it is, accorc ing to Dr. Davies's Welsh-Latin Dictionary, foun used in the feminine and applied to the riv( Dee, which need not surprise you, as the De Deva, probably means ' the goddess,' and as tl river is still called in Welsh Dyfrdroy, ' the wat( of the divinity: ' Giraldus calls it Deverdoeu, tl full spelling of which would now be Dyfrdwyw i Dyfrdroyf, whereby he upsets the popular et; mology, which explains the word as meaning tl water of two {rivers). On river-names of th class see M. Pictet's paper in the Revue Celtiqu ii. 1-9. However, the word occurs also in tl sense of war or battle generally, as in Englynion Gdrugiau {lolo MSS. 263), where we read: — " Goruc Arthen ap Arth Hen Rhag ffwyr esgar ac asgen, Llafn ynghad ynghadr aerfen; " i.e., Arthur ap Arth Hen against foeman's attai and injury made the blade (for use) in battle, stout war. But why should the origin of letters have bei 



 



(de
lwedd B6326) (tudalen 326)

326 LECTURES ON WELSH PHILOLOGY. connected with Ogyrven, whose character was from the first that of a dark and concealing heing? One might answer that it was for the same reason which made the Irish attribute the motive of secrecy to Ogma, though that ill agreed with his solar origin: both versions, it may be, merely reflect the feeling with which the ignorant many would regard the language, whether written or spoken, of the learned few. On them the impression of mystery and awe produced by the sight of certain characters cut on wood may easily be conceived to have led them to call them the un\gogyrven ar bymtheg, that is, as though we called them ' the sixteen devils.' Later, however, a solar patch was, so to say, sometimes sewn on the tradition, in the shape of a reference to the three sunbeams /|\, which still hold their place as a sacred symbol or talisman at the head ' of Eisteddfodic announcements. But perhaps the question as to the relation in which Ogyrven stood to letters is best disposed of by asking another, namely. How it is that there exist even now people who think that knowledge and science are of the devil? In former times this was, no doubt, very much more commonly the case than it is now. The cryptic view taken of writing by the ignorant, and incorporated in the Irish tradition touch- 



 



(de
lwedd B6327) (tudalen 327)

LECTURE VI. 32 ing the Ogam, has sometimes led Irish archaeolc gists into the error of thinking that the Ogam wa really a cryptic contrivance. It is true that in i1 last days it may have fallen into the hands ( pedants, but it still remains to be shown that eve a single Ogmio monument of respectable antiquit in Ireland can in any sense whatever be said to I of a cryptic nature. It is, of course, but naturs that writers, who have no wish or no time to stud the laws of phonetic decay, should find in earl Irish names merely disguised forms of the: modern continuators. Their view is also suppose to derive support from a passage in Comae's Gloi sary, which explains the Irish word fd as " wooden rod '• used by the Gael for measurin corpses and graves, and this rod was," we ai told, " always in the burial-places of the heather and to take it in his hand was a horror to ever one, and whatever was abominable (adetche) f them, they used to put in ogham upon it {^i6ke&' Three Irish Glossaries, p. Iv.). Here it ha been supposed that we have an allusion to cryptic fashion of recording the sins of a decease person; but it is difficult to see anything crypti in the whole proceeding, unless it be the act ( leaving the/"^ in the burial-place, which, in thE case, may have been meant to suggest, in a del; 



 



(de
lwedd B6328) (tudalen 328)

328 LECTURES ON WELSH PHILOLOGY. cate manner implying no ignoring of tlie faults and shortcomings of the departed, that thenceforth his name would have the full benefit of the maxim: " De mortuis nil nisi bonum."  ( 329 ) 



 



(de
lwedd B6329) (tudalen 329)

LECTURE VII. " Nous nous sommes efforcfi jnsqu'3. present de reconstituer les etapes successives qui couduisirent depuis la premidre origiue de I'art d'6crire jusqu'^ rinvention dSfiuitlTe de I'alphabet. Nous avons vu combien cette graude et f6conde inTention, qui aiuena recriture a son dernier degre de perfection et en fit un instrument completement digne de la pensee humaine, fut lente El se produire, combien p^ni- blement elle se dggagea, par une marche graduelle, de I'ideograpbisme originaire. Nous avons vu comment pour y parvenir il avait fallu la combinaison des efforts successifs et des gSnies varies d'un peuple philosopbe, les Egvptiens, qui sut con9evoir la decomposition de la syllabe et de I'abstraction de la consonne, puis d'un peuple pratique et marchand, les Pheniciens, qui rejeta tout Element id€ographique et reduisit le phonetisme, demeur6 seul, k I'emploi d'une figure unique pour representor chaque articulation. Mais aussi cette invention, qui demeurera I'etemelle gloire des fils de Chanaan, ne fut faite qu' une seul fois dans le monde et sur un seul point de carte, et, une fois accomplie, elle rayonna partout de proche en proche." — Pbakjois Lenoemant. This lecture will be devoted mainly to conjectures, and tlie facts adduced, it may as well be admitted at the outset, will be few and far between. Of the latter, the principal one is the Phoenician alphabet, for which, however, we have to use the Hebrew version, as giving us the order of the letters, and also their names in a form which cannot be materially different from that which they had in Phoenician. The other leading fact is the Ogam system as attested by the oldest monuments extant in Wales and Ireland. Given 



 



(de
lwedd B6330) (tudalen 330)

1 2 aleph beth a b a b a b  1 2  3 4 gimel daleth ch d ch d oh d  3 4  5 he — — —  6 waw — — —  7 zain — — —  8 cheth — — —  9 ieth — — —  10 11 12 yod caph lamed  1  1  1  5  13 mem m m m  6  14 nun  n u  7  15 saxaech  — —  16 ain  u u  8  17 18 pe tsade  P s  9 10  19 20 koph resh  c r  r  11 12  21 shin, sin  s  13  22 taw  t t  14 



 



(de
lwedd B6331) (tudalen 331)

LECTURE VII. 331 It appears accordingly that the Semitic letters from 4 to 12 were altogether discarded, and that we have now to set out from mem: consequently one cannot help referring n, c, r, t in the Ogam, to nun, koph, resh, taw respectively. Further, as he, maw, yod had been passed over, the only remaining letter which could be treated as a vowel was ain, which the Greeks made into o. It looks as though this was treated at first as u in the Ogam and written -|-[-j-, that character having probably only acquired later the value of w in order to differentiate it from +++• If this is right, then samech is to be regarded as thrown out, for the Ogam leaves it no room between ^ and -'-'-'■. The result so far as we have gone is shown in column iii.: still we have only 11 letters for the 22 of the Phoenician alphabet, while the Ogmic scheme offers room for 15, so we take in the remaining ones which have not been excluded, and the result is column iv., which, arranged Ogmically, gives us the following trial alphabet: — 1- I I ' II II " III III '" 'III nil "" mil mi l a, b, ch, d, 1, m, n, u, p, s, c, i, B, t. Here, it will be observed, we have two sibilants, namely, from tsade and sin respectively: in trying to make these square with the details of our hypothesis, one is led to conclude that the latter was 



 



(de
lwedd B6332) (tudalen 332)

332 LECTUEES ON WELSH PHILOLOGY. set apart for z: the alphabet will then stand thus: — 2- 1 I ' 'h i " i iii ii ' " i i iii ii i ""in i ii iiii a, b, oh, d, 1, m, n, u, p, B, c, t, i, t. The next point to be noticed is that this shows only two vowels, a and w; even so it had the advantage in this respect over the Semitic alphabets, which had none. Now if the Ogam is connected with the Phoenician alphabet the values of ff , +++, ' WW, +++++, would seem to have been at first d, n, s, z, while their only attested values are found to be 0, u, e, i respectively. It follows that the consonants must have been ousted by the vowels; but as this does not appear to have been done at once or methodically, one must infer that at one time the symbols in question had two values each, the one consonantal and the other vocal: accordingly -H- had the values of d and o. This I would write shortly do, without, however, giving the Ogam +1 the value of the syllable do, but the separate values of d and o; and so with the others, thus: — I I I I I ni l mil do, nu. Be, zi. That the vowel values are here of later date than the consonantal ones, is also probable from the regular intervals at which they occur in the arrangements suggested and presupposed by the 



 



(de
lwedd B6333) (tudalen 333)

LECTURE VII. 333 grouping of the Irisli Ogam, wliich has already been referred to in connection with its leading letters b, h, m, a, and the permutations they admit of. But how did the vowels get into these positions, and how were the consonants dislodged? We seem to have a clue to the answer in the case of nu, which one cannot help regarding as suggested by the letter-name nun: similarly zi, for si, is to be referred to the name sin. The case of ++, do, looks as if the spelling daletk of the Hebrew name of the fourth letter did not exactly give the pronunciation, which the first Ogmists learned to give the word as they heard it. Was the latter more nearly doleth, which approaches, I am told, the Arabic pronunciation of the word as used for the letter and for door at the present day, or are we to assume rather that they translated the word into their own language, that is into an Aryan equivalent beginning with do, such as would, for instance, be Welsh dor, and drws (for dams'), Irish dorus, all with dor for dvor, 0. English dor, ' door '? Lastly, the vowel e was probably associated at first with the name pe or resk; but sooner or later the analogy of +, ++, +++, f|-l4+, would naturally lead to the use of fH-F or se with the values of s and e, and perhaps even to the modification of its name into a form more nearly approaching sede than tsade. Of course, if one could 



 



(de
lwedd B6334) (tudalen 334)

334 LECTURES ON WELSH PHILOLOGY. assume that the Phoenician pronunciation of the word had e and not a in its first syllable, a shorter path to the same result would lie open. In case it should appear more satisfactory to bring on the scene a deas ex Tnachina and to suppose a systematic modification of the alphabet by a grammarian, it is to be observed that such a modification must have been confined to giving some or all of the Ogams new names instead of the Semitic ones. The former in the cases in question would have to be regarded as either beginning with, or consisting of the syllables do, nu, se, zi, or else od, un, es, iz, or some of both sets. For our present purposes, however, the ambiguities of the Ogam at this stage may be represented as follows: — 3- I I ' I I II " III II I ' I I "" IIHIll l l l a, b, ch, do, 1, m, nu, u, p, Be, c, r, zi, t. The answer to the other question as to how d, n, s, z were dislodged, will offer itself as we go on: the next step in advance which seems to have been taken appears to have been the filling of the cadre of the Ogam by the addition of a symbol for qv, thus: — 4. I I ' I I II I ' I I I I I I III nil n i l a, b, oh, do, 1, m, nu, u, p, se, c, "" II i i r -^ I-, Zl, .t, qv. The further working of the same sense of system 



 



(de
lwedd B6335) (tudalen 335)

LECTURE VII. 335 seems to have sooner or later occasioned c and r to change places, so that c and qv should stand side by side: — f\ 1 I I I II III III 1 1 1 1 1 1 1 1 0. 1 I -" II II " 111 111 iiii nil a, b, ch, do, 1, m, nu, u, p, se, r, c,  "" IIII! zi, t, qv.  So far the ambiguities in our versions of the Ogam alphabet have been left standing. Now the symbols in places 1, 4, 7, 10, 13, have throughout retained the vowel values here attributed to them, while the consonantal values of those in 4, 7, 10, 13, are unknown to the Ogam system, as attested by our monuments. Hence the simplification was effected by providing other symbols for the four consonants in question. Let us begin with ++, do, and see how matters will then look. If one leaves ++ to represent o, how is d to be written? Three courses suggest themselves: d may be written ^ and a new symbol invented for m; it may be written jj-, which would necessitate a new symbol for I; or lastly, a new symbol may be provided for d without disturbing any other letter. The last would seem to recommend itself in point of simplicity, but it has against it the circumstance that m is, as a matter of fact. 



 



(de
lwedd B6336) (tudalen 336)

336 LECTURES ON WELSH PHILOLOGY. represented by / in the oblique group: the course adopted then was as follows: — 6-^1 ' II II " Ml II I '" I'll n il "" a, b, ch, 0, 1, d, nu, u, p, Be, r, c, ^^ m il ' "" / zi, t, qv, m. Now the foundation had been laid of a new group: the first addition was a symbol for g, "which had been left unprovided for when ch took the place oi gimel: — 7 I , I II ,. II III ... Ill nil Mil /. I I ' II II III III III! MM a, b, ch, 0, 1, d, nu, a, p, se, r, c, + 11 1 ' IIMI - ' "" /// ZI, t, qv, m, g. The next addition was, naturally enough, to provide for ng: — 8- I I ' II II " III III '" I 'll MM "" a, b, ch, o, 1, d, nu, u, p, se, r, c, + IIII mi l '"" ////// zi, t, qv, m, g, ng. The next step was to dispose of zi: this was done by relegating z to the new group: — Q I I I I M ,1 1 m MM mi y. +■■ I II .11 III 111-'" iin-im a, b, oh, o, - 1, d, nu, u, p, se, r, c, 'I'll mil '"" ////////// i, t> qT, m,g, ng, z. The case of se seems to have been dealt with 



 



(de
lwedd B6337) (tudalen 337)

LECTURE VII. 337 differently, s being written jjjj, and r relegated to the new group: — 10. I I ' II II 11 111 I II I I ' M i l ,111 nii a, b, ch, o, 1, d, nu, u, p, e, s, ■ c, mil iii i i '"" I II III nil Hill 1, t, qv, m, g, ng, z, r. The symbol for p was found to be useless as such, owing to that sound not being used in the languages of the Celtic nations: its place was utilised for t, whereby d and t were brought near one another: — n. I I I II II II III I II II I M i l 11 ,1 I'll a, b, ch, o, 1, d, nu, u, t, e, s, o, mil M ill ' "" III III nil IIIII i, — qv, m, g, ng, z, r. The way was now open for nu to be disposed of, so the consonant was placed in the place vacated by t: nu was allowed to stand so long, probably, because -j-p]- was available for u: — J2. I I I M ,1 " III 1,1 "I MM 11,1 "" a, b,cli, 0, 1, d, u, u, t, e, s, o, m il M i l l ' ^ ^ ^' J -H- lll nil IIIII i, n, qr, m, g, ng, z, r. The anomaly of having two symbols for u in the alphabet was disposed of by setting jjj apart for m, Latin v. Otherwise the Celts have never shown themselves anxious to distinguish in writ- T 



 



(de
lwedd B6338) (tudalen 338)

338 LECTURES ON WELSH PHILOLOGY. ing between the semi-vowels and the corresponding vowels. After this final touch the Ogam alphabet stood as follows, from which we set out: — 13. +-H-^f-Tr^^4 l II I I N Mil nil i l M a, b, oh, 0,1 1, d, u, w, t, e, s, c, mil mil ' "" hll HI nil mil 1, n, qv, m, g, Dg, z, r. Let US here pause to look around us and try to ascertain whether they are not mistaken who regard the Ogmic alphabet as an isolated phenomenon in Europe. We fail in the direction of Greece and Eome, so let us look nearer home, to the Teutonic nations, especially as there is reason to believe that the last word has not yet been said on the history of the Eunic alphabets, which they formerly used. Fortunately for one who is not at home in Scandinavian languages and antiquities, an important work has lately been published on the origin and development of Kune-writing in the North, by Dr. Wimmer, a Danish scholar who is well known in the philological world, and who has opportunities of personally examining the most important Eunic monuments of the North (JRuneskriftens Oprindelse og Udvikling i Norden of Ludv. F. A. Wimmer: Copenhagen, 1874). 



 



(de
lwedd B6339) (tudalen 339)

LECTURE VII. 339 Kunic monuments may be roughly said to have been found in all countries inhabited by nations of Teutonic descent, but the oldest of those monuments cannot be regarded as dating before 200 A.D. There are two chief varieties of the Runic alphabet, one consisting of 16 letters and the other of 24. Dr. Wimmer undertakes to show that the former is derived from the latter, which is arranged into three groups, as follows: — 1. f, u, Ip, a, r, k, g, w — 8. 2. h, n, i, y, eu, p, z, s — 8. 3. t, b, e, m, 1, ng, o, d — 8. The Eunes representing most of these letters turn out to be the capitals of the Roman alphabet of 23 letters, borrowed from the Romans during the Empire not long after the time of Julius Osesar. The others are later additions formed by modifying some of the earlier ones; and they are the Runes for y, w, y, eu, ng, d. Thus for the form of the remaining 18 Runes one can account by the direct means of the Roman alphabet, while it leaves their arrangement a question which Dr. Wimmer, like those who have written before him, cannot answer. This, then, is our next great fact, namely, that the Teutons must, in all probability, have had a prae-Roman alphabet of 18 letters, 



 



(de
lwedd B6340) (tudalen 340)

340 LECTURES ON WELSH PHILOLOGY. which at the time when they were induced to adopt the Eoman characters instead of their own stood as follows: — 1. f, u, p, a, r, k — 6. 2. h, n, i, p, z, s — 6. 3. t, b, e, m, 1, o — 6. The fact of the Eunic alphabet or the Futhark, as it is called from its first letters, being from the first arranged into groups, appears to be a distinct indication that it is the outcome of some such a system of writing as the Ogam. So I venture to proceed to show how it can be connected with the alphabet which has served as a key to the history of the changes which the Ogam may have undergone at the hands of the Celts. But before beginning to do so, it is to be noticed that the Celtic 6, cA, d have to be translated into_/, h, J) in order to comply with the usual way of transcribing the Futhark: and for its earlier history the change here implied is very little more than this, as will be made clear later. Our first three alphabets as given in the foregoing series will accordingly stand thus: — i. +-T-i-H-n-^w-TTT-^-tw-rrrr ^^ i ' ' 1 1 1 1 1 1 1 a, f, h, Jj, 1, m, n, u, p, s, k, r, a, t. a, f, h, J), 1, m, n, u, p, s, k, r, z, t. 



 



(de
lwedd B6341) (tudalen 341)

LECTURE VII. 341 iii. +.^-L++.^-il.+++.^ill.^+.^.iiil.+^.^^^ a> f, ^, fo. 1. m, nu, u, p, se, k, r, zi, t. The systematising tendency confined the vowels to one kind of characters, and -p|-|ceased to be used for u:— iv. +-T-L++-^-li-+f|-^-iii.+^.^.lLU.+^^.^^ a, f,li,]jo, 1, m, nu, — , p, se, k, r, zi, t. This allowed r to move one place forward and to enter another class: — ■ V. +-^i-++-^ii-+H-^-Li^-++++-TnT-^-+++++-nm a, f, h, j)o, 1, m, nu, r, p, se, k, — , zi, t. Now it was possible to separate the two values of ■mn thus: — vi. +-^i-^-^-ii-+++-^-iii-+^-^-iiii-++m-ymT a, f > ^, Jjo, 1, m, h", '•> p. se, k, z, i, t. The next step seems to have been the invention of a new symbol for t: let us suppose it to have been an oblique score: — vii. i-pl-ii-^li-ni-pp^-LU1 1 1 1 ,1 1 1 -Ull.fH^.y! a, f , h, Jio, 1, m, nu, r, p, se, k, z, i, t. This naturally became the commencement of a new group: the fitst addition was a character for 6, which had previously been expressed by the same means asy.— viii. I I ' II II " IH"||| '" III! a, f, h, yo, 1, m, nu, r, p, se, nil "" mil /// k, z, 1, t, b. 



 



(de
lwedd B6342) (tudalen 342)

342 LECTURES ON WELSH PHILOLOGY. The next step taken seems to have been to separate the values of ])<?. This was done by writing Jj either jj or ^, and that hesitation rendered it necessary to have new symbols for I and m: — ix. I I I II II II III III I" nil a, f, h, o, J>, }), nu, r, p, se, nTT-^-+4+H^ // /// //// k, z, i, t, b, m, 1. Why m should precede I in the new group I cannot say, and it should be borne in mind that the Runic alphabets are by no means uniform as to the sequence of m and I: Dr. Wimmer (pp. 190-196) thinks, it is true, that the sequence was at first invariably m I, but I am not quite convinced by his reasoning that that o{ I m may not be equally old. Eventually ^ ceased to be used for J), and became available for the consonantal power of nu: — X. + I I II II II III II I III nil a, f, h, o, J), n, u, r, p, se, n i l ' " 1 1 II nil mil k, z, 1, t, b, m, 1. Now a new symbol was invented for s, which should stand by the side of that for the nearly-related sound of z: — xi. I I I I I II II I ll -Ill II I I II! 1, f, t, 0, J), n, u, r, p, e, Ti ll " " mi l ' "" //////.//// k, z, i, B, t, b, m, 1. Here we have an alphabet, which I would call a 



 



(de
lwedd B6343) (tudalen 343)

LECTUBE VII. 343 Teutonic Ogam, consisting of four kinds of digits admitting of being grouped as follows: — xii. 1. a, 0, u, e, i — 5. 2. f, }), r, k -4. 3. h, n, p, z, s — 3. 4. t, b, m, 1 — 4. And tbis is, in fact, precisely the order of tbe consonants in tbe tbree groups of tbe pra3-Roman alphabet of tbe Teutons as proved by tbe Futbark; and we migbt stop bere. For tbe dispersion of tbe vowels among tbe consonants in tbe latter creates no difficulty wbicb we are bound to account for. It probably only marks another step in advance, when the Teutons gave up writing their Ogam on two conterminous planes, and took to tbe laths or planed rods of historical times, wbicb make it hopeless now to find an early specimen, and with regard to wbicb Dr. Wimmer quotes the words of Venantius Fortunatus in tbe 6th century: — " Barbara fraxineis pingatur runa tabellis, Quodque papyrus agit, virgula plana valet." It may be supposed that it was found inconvenient to distinguish four kinds of digits on one surface, and that this led to one of them being given up. On what principle the vowels were distributed in the other groups it is not easy to see; but the 



 



(de
lwedd B6344) (tudalen 344)

344 LECTURES ON WELSH PHILOLOGY. broad vowels a, u, are placed in the i^-group, the narrow vowel i in the ZT-group, and in the remaining one the transition vowels e and o, which were once supposed not to have existed in the early stages of the Teutonic languages; but that theory is now exploded:. — xiii. 1. f, u, ]j, a, r, k — 6. 2. h, n, i, p, z, s — 6, 3. t, b, e, m, 1, o — 6. These were the letters for which the Teutons adopted the Eoman characters; a single instance will suffice to show how additions were made to this Futhark. The Eune for k was the Latin C, reduced into straight lines, thus <: two of these placed thus x were invented to represent y, and appended to the J'-group by the side of the Eune for k: somewhat similarly was formed the Eune for ng, which was placed in the T-group. The number of the Eunes in the ^ET-group was kept on a level with the other two by the invention of one for y (as in Mod, English ye. Old Eng. ge), the place of which was settled by its affinity for the vowel i: — xiv. 1. f, u, ]), a, r, k, g — 7. 2. h, n, i, y, p, z, s— 7. 3. t, b, e, m, 1, ng, o — 7. 



 



(de
lwedd B6345) (tudalen 345)

LECTUEE VII. 345 Then Eunes for w and d seem to have heen added to the first and third groups respectively: — XV. 1. f, u, y, a, T, k, g,w — 8. 2. h, n, i, J, p, z, s —7. 3, t, b, e, m, 1, ng, o, d — 8. To make the second group of the same number of Eunes as the other two, and of the same number of vowels in particular, the doubtful expedient was resorted to of inserting a diphthong in it: — xvi. 1. f, u, \), a, r, k, g, w — 8. 2. h, n, i, y, eu, ]>, z, s — 8. 3. t, b, e, m, 1, ng, o, d — 8. It is to be observed with respect to the shorter Futhark of sixteen letters which Dr. Wimmer derives from the longer one, that, while it has dropped three of the eighteen original Eunes and modified the values of some of the others, it includes only one of the six post-Eoman ones; so that it may still perhaps be questioned whether the other five ever got all into general use. But this and many other points, on which I should like to have dwelt, do not affect the order in which the Eunes are grouped, and by means of which the prse-Eoman alphabet of the Teutons seems to prove itself to be of the same origin as the Ogam of the Celts. 



 



(de
lwedd B6346) (tudalen 346)

346 LECTURES ON WELSH PHILOLOGY. Here is the place to call atteation to the direction of the writiDg: the Ogam is, as a rule, written from right to left, and as to the Eunes, Dr. Wimmer concludes "that they were originally so written too, but that, as they very readily lent themselves to the contrary direction, the latter also was at times adopted with the former, giving rise to ^ovarpo^Bov writing of the ordinary kind. There was, however, a simpler Bustrophedon which he calls snake-twisted (slangedrejet), in the course of which the person writing turned the object he wrote upon round, or, where that was not feasible, as in the case of a large stone, shifted his own position: the writing would then run thus: — A, b, c, d, e, f, g, ^ C— I. •oig 'd 'o 'u 'm \ ^\ This you will have noticed was one of the ordinary methods pursued by the writers of the Ogmic monuments of Wales. In the case of the Eunes, Dr. Wimmer admits that it is common enough on the later monuments, whereas- he has found it only on one from the older Iron Age, and then in conjunction with the common or inverted Bustrophedon. Nevertheless, if Eune-writing is but a continuation of the Ogmic system, it can only be 



 



(de
lwedd B6347) (tudalen 347)

LECTURE VII. 347 an accident that it has not been more frequently met with on the older monuments. The inverted Bustrophedon is to be met with in some of the oldest Greek inscriptions, and occasionally in Etruscan ones, whereas the simpler one is rarely detected in Greece or Italy, and its appearance in Wales. and Teutonic countries is a point in favour of the view that the Runes and the Ogam are connected with one another. Why both were written mostly from left to right, while the Phoenicians wrote from right to left is a question which I am not prepared to meet; but the answer is perhaps to be sought in the fact, if such I am right in thinking it to be, that when cutting a series of scores or notches on a piece of wood, one is able to work with more ease and neatness by beginning at the end nearest one's self than at the other. Assuming that it has been shown to be probable that the Ogam and the prae-Runic alphabet of the Teutons are connected, one may ask how they may be connected? that is, are we to regard one as derived from the other, or both as independently derived from the Phoenician alphabet, whether directly or indirectly? Clearly one has no business to try the latter alternative, unless the other turn out inadmissible: then our first business is to try to ascertain whether the Teutonic alphabet is derived from the Celtic one or vice versa. Not 



 



(de
lwedd B6348) (tudalen 348)

348 LECTURES ON WELSH PHILOLOGY, to depart from the order we have hitherto followed, we shall in the first place suppose the Celtic entitled to precedence. In the absence of historical data the question must be settled on phonological ground. We have a ready test in the Ogmic ch: how is it that, while betk and daleth yielded Ogmic h and d, gimel on the other hand yielded ch, and not g? To this the Celtic languages can give no answer, but the Teutonic ones can, which compels us to suppose the Celts to have had their Ogam alphabet from the Teutons, and derives confirmation from the fact that the sound of ^ or _/ remained withoujt being provided for, at least by a strictly Ogmic symbol. This leads me to consider very briefly some points in the phonology of the Teutonic languages, which, I feel assured, you will consider no hardship, seeing that the English we are at this moment using is one of them, and that it is nearly related to our own Celtic vernacular. When it is said with regard, for instance, to the words irrepdv and feather that the y of the latter is the p of the former subjected to provection, this assigns only the limits of the change: at any rate one of the latest writers . on the subject would place between p and Teutonic / the intermediate steps of b and v: I allude to Mr. Henry Sweet in his History of English Sounds (pp. 76-81), and in an 



 



(de
lwedd B6349) (tudalen 349)

LECTDKE Vn.  .349  appendix to his edition of King Alfred's West-Saxon Version of Gregory's Pastoral Care (pp. 496—504). The conclusions he draws in the latter may be tabulated thus: —  Aryan Parent-speech. Teutonic. Stage i. Stage ii. T D DH d t dd dh (th). d. P B BH b v(f). P- b. K G GH . gg kh, h -. k. g-  If this is nearly correct, as I suppose it to be, one would have to suppose the Teutons to have got their Ogam at a date corresponding to the first Teutonic stage in this scheme, that is after they had reduced Aryan t into d, but before the latter had been reduced to dh (= th in this), whence later th (as in thin). Here it will be observed that the guttural surd was subjected to more changes than the corresponding dental and labial. *' The explanation must be sought," Mr. Sweet thinks, "in an important phonetic law: general weakening tendencies attack the strongest articula- 



 



(de
lwedd B6350) (tudalen 350)

350 LECTUEES ON WELSH PHILOLOGY. tions first. Accordingly we find that while original d and b [our Teutonic stage i.] have only passed through one stage of weakening, original initial g has passed through no less than three: gh, kh, and h, in the last reaching the extreme of phonetic decrepitude " (Appendix, p. 502). That is, the changes in question would stand somewhat as follows if we regard only their chronological order: — Phoenician . . h(eth), g(imel), d(aleth). Teutonic 1 . . b, g, d. ,, 2 . . b, gh, d. ,, 3 . . b, kh, d. „ 4 . . V, h, dh. From this it appears that Teutonic phonology fully meets the difficulty which presented itself in our former supposition, and that we have, therefore, to abide by the other, namely, that the Celts got their Ogam from the Teutons, and the latter directly or indirectly from the Phoenicians. Now we are in a position to bring our supposed Teutonic Ogams into more complete harmony with the history of phonetic decay and change in the languages of that name. The first would be more correctly written thus: — I. I I II I II I I I II III III n i l nil "" a. b, g, d, 1, m, li, u, p, B, k, r,  H-m-  nrr t.  -LECTURE VII. 351 In No. II. we should have to recognise the change of g into gh^ thus: — II. I I I II I I II I II i ii-iH I II I! a, b, gh, d, 1, m, n, u, p, 8, k, r. mil  rmr z, t. In the next we have to suppose a further change of gh into kk or ch: — III. 1 I I II II I' I I I I II III nil ni l a, b, kh, do, ], m, nu, u, p, se, k, "" IM II mr /, Zl, t. This is now the stage in which the Teutonic alphabet must have been when the Celts became acquainted with it and borrowed it, if, as I believe, we are right in thinking them to have done so. Alphabets IV., V., VI., VII. will now stand thus: — IV. +^^- 11 II II III I I I '" MM i m- a, b, kh, do, 1, m, nu, — , p, ae, k, III! inii ll'll-jiill r, zi, t. V. I I ' II II " I I I I I I I'l nil n i l a, b, kh, do, 1, m, nu, r, p, se, k, "" mil ii iii  ZI, t.  VI. -I-T-M+- II II III I I I III M i l nil a, b, kh, do, 1, m, nu, r, p, se, k. 



 



(de
lwedd B6351) (tudalen 351)

1 1 " n i l i, t. 



 



(de
lwedd B6352) (tudalen 352)

352 LEOTUKES ON WELSH PHILOLOGY. ^"- n ,' ' " N " '" IN '" "" II" a, 0, kn, do, 1, m, nu, r, p, se, k, I ' ll m i l -/ z, i, t. At this stage one finds reasons to conclude that b had been reduced to v (as in vat), but not so universally as to make a character for b unnecessary: on the contrary alphabet No. VIII. provides for it: — VIII. I I I I I I , I I II I I I I I I I nil n i l a, V, kh, do, 1, m, nu, r, p, se, k, i i ' i mil / // z, i, t, b. Alphabets IX., X., XL, XII., and XIII. will then run thas: — IX. I I I I I II Ill III n i l n i l a, V, kh, o, d, d, rni, r, p, se, k, "" iNii I II iii -m z, 1, t, b, m, 1. X. +-^^ 11 I I II III III ' " nil nil a, V, kh, u, d, n, u, r, p, se, k, "" I'M I II III nil z, 1, t, b, m, 1. XL ^T^+i- n II 11 1 III '" II " mi "" a, V, kh, 0, d, n, u, r, p, e, k, z, I ' '" I II III nil i, B, t, b, m, 1. XII. 1. a, 0, u, e, i — 5. 2. V, d, r, k —4. 3. kh, n, p, z, 6 — 5. 4. t, b, m, 1 — 4. 



 



(de
lwedd B6353) (tudalen 353)

LECTURE VII. 353 XIII. 1. V, u, d, a, r, k — 6. 2. kh, n, i, p, z, s — 6. 3. t, b, e, m, 1, o — 6. Now we have come somewhere near ths time when the Teutons translated their Ogmic digits into the letters of the Eoman alphabet; and it is found among other things that kk had been so far modified in sound, that is as an initial, and especially perhaps as the initial of its own name, as to allow of its being represented by Latin H, whence the Rune for it. D got to be represented by the Latin Z>, whence the Rune p, which is merely D with the perpendicular prolonged; and Dr. Wimmer thinks he recognises in the Rune for the sonant sibilant the Z of the Roman alphabet. It is not very clear why F was chosen to stand for j: was it that F represented the Latin consonant which most nearly approached Teutonic V, or was it that even then the latter, as an initial, had begun to assume the sound ofy as in English and German at the present day? The foregoing alphabet will- now stand thus: — XIV. 1. f, u, J), a, r, k— 6. 2. h, n, i, p, z, s — 6. 3. t, b, e, m, 1, o — 6. At this stage it is probable that the S'-Rune stood not only for k but also for cA and y, until at 



 



(de
lwedd B6354) (tudalen 354)

354 LECTUKES ON WELSH PHILOLOGY. length the last-mentioned consonant got to he thought of as more nearly related to k, and a symbol for it invented from the ^-Rune as in alphabet XIV.:— XY. 1. f, u, J), a, r, k, g— 7 2. h, n, i, y, p, z, . s— 7. 3. t, b, e, m, 1, ng, o — 7. The last addition of importance to the Futhark was a Eune for d, which was formed by joining together two J)-E.unes. The necessity for this arose from the fact that the sound represented by ]) underwent, more or less generally, a change from d into _dh (liable under certain circumstances to be further modified into th in some of the Teutonic languages). Not only were these the last changes to which the Futhark bears testimony, but it seems doubtful whether they have ever been gone through by some of the languages in question. Mr. Sweet, however, is inclined to think otherwise: his words are — "At first sight we are tempted to assume retention of an older pronunciation, at least in the case of Dutch and German, where the d appears in the earliest documents, but the non-occurrence of an analogous h for the actual w or _/ makes it almost certain that the d in Dutch and German, like the corresponding stop of the Scandinavian languages 



 



(de
lwedd B6355) (tudalen 355)

LECTUEE VII. 355 has arisen from earlier dh " (App. p. 499). The Fathark, then, in its complete state is the following, which has already been more than once mentioned: — XVI. 1. f, n, ],, a, r, k, g, w— 8. 2. h, n, i, y, en, p, z, s — 8. 3. t, b, e, m, 1, ng, o, d — 8. It is right, however, to state that some Futharks lack some of the additional Eunes alluded to, while others have several more than have here been mentioned; moreover, while the latter are placed at the end, there is, as might be expected, some -difference as to where the former are inserted in the Futharks containing them. Thus on a knife found in the Thames in 1857, and guessed to date about the year 700, the order is as follows: — 1. f, u, J), a, r, k, g, w— 8. 2. h, n, i, y, eu, p, z, s — 8. 3. t, b, e, ng, d, 1, m, o — 8. It will here be observed that the Eunes for ng and d have been inserted next each other after e, but without inverting their order, in the third group, which is otherwise highly interesting as giving us the variant sequence I, m. Before proceeding further a word may not be here out of place as to the number of changes crowded into our conjectured history of the Ogam, 



 



(de
lwedd B6356) (tudalen 356)

356 LECTURES ON WELSH PHILOLOGY. whether Celtic or Teutonic. In the first place, then, that crowding is more apparent than real, as the Ogam seems to have been many centuries in use before the oldest specimens known to us were produced. On the other hand it is not to be overlooked, that an alphabet like the Ogam, which is composed of scores and groups of scores would naturally change much faster than if it were not so, as a change in respect of one symbol would naturally induce other changes, which need not take place in an alphabet consisting of symbols the individuality of which depends on their difference of form. Now I shall have to say something on the difficult question of the names of these letters; but I can only call your attention to a few of the leading facts, passing by many points which I cannot profess to deal with. Any one, however, who wishes to make a special study of this subject will have to consult Mr. George Stephens's massive work on The Old Northern Runic Monuments of Scandinavia and England (London and Copenhagen, 1866-67). Perhaps I could not here do better than place side by side a certain number of the alphabets in point for your inspection. The names in column i. are from an alphabet contained in an old English manuscript ( Cotton. Otho. B. 10) now lost: it has been hesitatingly assigned to the 9th century by Mr. 



 



(de
lwedd B6357) (tudalen 357)

LECTURE vn. 357    f^ 9 9   ,£>   •5 » ,S -2 3 9 s -3 So S a S s ° -o 3 g S   1 1 =• 'I '1"^ " s,'E, ■ '1=3 1 g.^ = - ■ &i g ^    H o   i^ a^   3 .JO, .:;: •::: ,  S^-3 rfi,>o2j« se JH.2 aS" BS    s '. f-* '. Oil;«343:: e— ";dc3   _, fl cio JS: *2 t4 "Sots t, itj: d h ^   1 i . I i r.-s g life g-'i'U Jll. =^ I |l   >M MS 3 2 ,a.-i3 <8 h,J^35»!*S^ c.« p,N n*3 « 0) Srf o  



 



(de
lwedd B6358) (tudalen 358)

358 LBCTUEES ON WELSH PHILOLOGY. Stephens, whose No. 5 it forms: a copy of it is also given in fac-simile by Dr. Wimmer, p. 79. Column ii. is taken from an alphabet in a Vienna MS. {Codex Salisb. 140) which Grimm supposed to be a transcript from an English original brought to Germany towards the end of the 8th century: the transcript is considered as dating from the end of the 9th century or the beginning of the 10th by Dr. Wimmer, who gives a fac-simile of it by the side of the one just mentioned. Column iii. is from the so-called Abecedarium Nordmannicum of a St. Gall manuscript of the 9th century: it forms Stephens's No. 6, and is given in fac-simile by Wimmer, p. 191. Column iv. is copied from Stephens's No. 46, and comes from a Yienna manuscript ( Cod. 64): it appears to be of High German origin. Column v. is from Wimmer's names of the letters of the shorter Futhark as he finds it used in the later Iron Age in the North, p. 153. Column vi. is the same, as given in the Book of Ballymote, an Irish MS. of the 14th century, extracts from which have been published, with tracings of the original, by Mr. G. M. Atkinson in the Journal of the Kilkenny Archaeological Association for 1874, pp. 205-236: ar for ur is due possibly to a clerical error, and the abbreviated name of the 5- Rune is perhaps to be read bergann. Column vii. is from the alphabet attri- 



 



(de
lwedd B6359) (tudalen 359)

LECTURE VII. 359 buted to Nemnivus in a manuscript of "Welsh origin, now in the Bodleian, and dating from the 9th century. Stephens's No. 53 seems to be a copy of it, though not a very exact one. The account given in the original of the history of this alphabet is more curious than correct: " Nemniuus istas reperit literas uituperante quidam [sic] scolasticus saxonici generis quia brittones non haberent rudimentum at ipse subito ex machinatione mentis suae formauit eas ut uituperationem et hebetitudinem deieceret gentis suae." Then follow the Runes, which Nemnivus cannot have invented; so that nothing remains to be attributed to his inventiveness excepting perhaps some of the "Welsh names of the letters, and that only in a very qualified sense. Columns viii. and ix. are taken from the extracts already referred to as made by Mr. Atkinson from the Book of Ballymote. The names here given to the letters are those of trees and shrubs; and column ix. does not materially differ from the letter-names already cited from O'Donovan's Irish Grammar, excepting that the spelling in the former is older. Beginning with the first six or Teutonic columns, we have feoh, ^orn, os, rod, ceriy hcegl, nyd, peorfS, eolhx, sigel, tir, man, lagu, occupying positions where some traces of the Semitic names might be expected. It is, however, clear at a glance that 



 



(de
lwedd B6360) (tudalen 360)

360 LECTUEES ON WELSH PHILOLOGY. we have here to do with several which are beyond all doubt Teutonic. Thus hcegl and its congeners are the Teutonic words for hail, chosen probably with a view to their suggesting the two sounds of the Ogam ^, namely kh (or K) and g. '0. Norse sol means sun, and 0. English sigel or sygel appears to have had the same signification. Eolhx or ilcs was, according to Dr. Wimmer, p. 119, in an earlier stage elhyaz, elMz (Scandinavian elMr, owing to the change of z to r), containing the Z-sound as its final, because ■ it did not occur initially: compare the case of ing. The name, however, led to confusion and misunderstandings as to the value of the Rune, which I need not enumerate. Lagu in 0. English meant law and lake, with the latter of which the 0. Norse logr appears to agree; but in the St. Gall Abecedarium we have the Rune called lagu the leohtu, which is duly rendered in Nemnivus' alphabet by louber, i.e., lleufer, ' a light, a luminary.' Neither have the extant names of the old ^l-Rune anything to do with the Semitic name of aleph, as they are supposed to go back to an earlier Teutonic form, ansuz, which, becoming in the course of phonetic decay ans, os, &c., led to various modifications of the old Rune: one of these had the name aac, ac, ' oak,' another asc, asch, ' ash.' In passing it may be mentioned that somewhat similar changes 



 



(de
lwedd B6361) (tudalen 361)

LECTURE VII. 361 occurred in connection with the (5-Eune, and that in the Scandinavian languages Ger, Yer, or Ydr, the name of the y-Rune was, in consequence of another process of phonetic decay, reduced to dr, which supplied the North with another yl-Eune. The reason why the name of the y-Rune is mostly given as beginning with g is the same why ye and yes are in 0. English written ge and ges, which cannot be here dwelt upon. Now there remain to be traced to Semitic origin the Rune-names feoh, Qorn, rad, cen, nyd, pear's, tir, man, namely to beth, doleth (for daleth), resh, koph, nun, pe, taw, mem. Now, supposing the Teutons to have adopted these names with their knowledge of letters, directly or indirectly, from their Semitic teachers, they would, in compliance with a law which obtained in Teutonic at a very early date, curtail them (see Schleicher's Compendium,^ pp. 338-340) into be, dol, re, co, nu, me, leaving pe intact, and probably treating tarn as tau. Later they would seem to have completed these syllables into words with definite meanings, apart from their being names of the letters. Thus be, passing into ve, fe, was extended into feoh, fech, whence also feu and other shortened forms, all of which are phases of the word which in Mod. H. German is written vieh, ' a beast.' i<!e was made into some -such a word as 



 



(de
lwedd B6362) (tudalen 362)

362 LECTURES ON WELSH PHILOLOGY. rceda: rad and rat with the vowel a owe that vowel only to the intimate connection between a and ee in Teutonic declensions: compare the case of man, to be noticed shortly. Other forms of the Rune-name not given in the table are r^, rehir, rehrt. One finds a trace of the name ko (from kopK) in kaun, chaon, con, and chon, some of which have in some alphabets been appropriated by q: besides cen, it is found that chen and che are given, suggested perhaps by ce, the Latin name for c; but it is far more likely that the vowel e was selected to indicate that the consonant had a palatal sound, and to distinguish it from the corresponding velar sound, for which it is said an English Rune called kalk was used: see Moller's Palatalreihe (Leipsic, 1875), pp. 18, 27. Nu (from nun) is more regularly represented in nyd, naut, naud, not, 'need.' Pe is lengthened into peor^S, peord: pert, perd, peoih also occur, but as to perc and perch they seem to be provections of here or beorc, the name of b, for which accordingly other names, such as birith and the like, were provided. Tare treated as tau appears to have naturally led into the Teutonic forms corresponding to Greek Zev'i, Vedic Dyu, represented in English by Tues-day for Tiwes-dmg: the 0. Norse name of the same divinity in the Edda is given as Tyr, genitive Tys, accusative Ty; the 0. H. 



 



(de
lwedd B6363) (tudalen 363)

LECTUKE vir. 363 German forms are Ziu or Zio, genitive Ziwes. In most of the alphabets where dceg, dag, is provected into toe, the T-Eune becomes Ziu. How tir and ti stand with respect to Tyr and Ziu is not clear. Me extended into men would lead into the declension of man, which would then naturally- become its name, as will be seen from the foUow- insr: —  0. English. 0. Norse. Singular.  Nom. man, mon. mannr, maSr. Gen. mannes. . manns. Bat. men. manni. Ace. man. mann. Voc. man.  Inst. men.  Plural.  Nom. men. menn, mennr, me8r. Gen. mannk manna. Dat. mannum. . monnum. Ace. men. menn. Voc. men.  Inst, mannum.  The presence of n also in wen, uyn, the name of the 1^-Rune, would seem to indicate that the lengthening of the Rune-names into significant words belongs mostly to a time after the Teutons had adopted the characters of the Eoman alphabet. The thorny case has been reserved to the last: the name of the Rune in question occurs variously as \orn, dorn, "pur, pars, purs, doro, and derhu. 



 



(de
lwedd B6364) (tudalen 364)

364 LEOTUEES ON WELSH PHILOLOGY. The shortest form to be inferred from these appears to be dor or Jjor, which, not being do or J)0, agrees well with the supposition that we have to set out fronj doleth. On the other hand, I have no reason to give for the change of I into r. Join to this the difficulty as to the vowel, and it must be admitted that the history of the names of this Rune is far from satisfactorily made out. This does not, however, materially affect the foregoing theory: for as far as regards the supposition that the Ogam ++ acquired the two values of d and by reason of its name one might, had one adopted a different arrangement, argue backwards from '^orn, ^ors, instead of the other way from daleth. Let us now turn to the Welsh and Irish columns of the table. The Welsh words cusil, guichr, hull, iechuit, traus may, for anything one can now say to the contrary, be the ones which suggested themselves to Nemnivus on the spur of the moment: braut, rat, parth, muin, louber are also Welsh words, but a glance at the Teutonic and Irish names of the corresponding letters makes it highly improbable that the choice made of them was altogether accidental. Dexu, nihn, surg, egui are obscure; but dexu reminds one of derhu in Stephens's alphabet 47, nihn of Irish nin, and egui of eh and eho: ieil was borrowed probably 



 



(de
lwedd B6365) (tudalen 365)

LECTURE Til. 365 from a Teutonic source, and so undoubtedly was jich. Not only were the writers of the medieval tracts on Irish Ogam well acquainted with Eunic alphabets, but most of the points of similarity between the Celtic names, whether Welsh or Irish, and the Teutonic ones point to the direct influence of the Runes, more especially after the coming of the Northmen and their settlement in Ireland. This circumstance greatly diminishes the value of the evidence afforded by the Celtic alphabets cited. In two or three instances, however, we seem to detect in them traces of an earlier tradition coming down possibly from the time when the Celts adopted the Ogam from the Teutons. To this category I would refer "Welsh alar and Irish ailm, as reflecting, hardly by mere accident, the first syllable of aleph. Similarly Irish dwr, also duir and dair, 'oak,' are remarkable for their agreement with Teutonic thur^ thor-n, thor-s: possibly dexu is a clerical error for deru, now derw, 'oak,' As to beitk, beithi, bethi, 'birch,' it may be that we have here only a translation of beorc, ' birch,' or else forms of much older standing, being the Irish extensions of the Semitic beth, borrowed from the Teutons before they had discarded the final consonant of the word. However this may be, the position of beith at the head of the Irish alphabet was probably what led to the 



 



(de
lwedd B6366) (tudalen 366)

366 LECTURES ON WELSH PHILOLOGT. unlucky freak of giving all the other letters the names of trees and shrubs. The reason why the names of the three letters in question should have escaped the later influence of the Runic alphabets, would be that the Runes originally corresponding to them had in the meantime changed their values, that for a having become o or o, and that for b having acquired the value v, _/, and that for d the value of dh, th. Not to pursue this subject of the names of the Runes further, it may be said that some of them appear to favour the view that the latter are descended from the Phoenician alphabet, which is, however, only a portion of the theory which I have endeavoured to set forth in this and the previous lecture. Its chief points are the following: — The Ogam alphabet is of a double origin, forming a sort of compromise between the East and the West. The characters used, if considered merely as writing and without reference to their meaning, are European and traceable to the quaternary period: the same may probably be said of the direction of the writing from left to right. The order of the letters, on the other hand, and some of their names, admit of being traced to a Phoenician origin. The Celts appear to have got their Ogams from 



 



(de
lwedd B6367) (tudalen 367)

LECTURE VII. 367 the Teutons, who seem to have used an alphabet of that description before they adopted the characters of the Eoman alphabet. Here I stop, leaving unanswered such questions as the following, which the foregoing conjectures naturally suggest: — Were the Teutons the original framers of the Ogam alphabet, or did they merely adopt it from another nation in more direct communication with the East? "Was it based on some prehistoric version of the Phoenician alphabet in use in Italy or Greece, among Slavonians or Scythians, the latter of whom Eustathius mentions as in the habit of writing on small boards or wooden tablets (o-av/Ses)? Could the Teutons have come in direct contact with the Phoenicians on the coast of Thrace, or on the Danube? Had they a trade-route connecting Germany and the Baltic with the Euxine or the Bosphorus? It is enough for our present purpose to find that there is no reason to think it impossible for a knowledge of letters of Phcenician origin to have reached Germany in very early times; and even the mythical history of the Greek alphabet brings Cadmus not only into Greece, but also into Thrace and in contact with the lUyrians. There can be no objection to these attempts to divine the history of Ogmic writing being ended where they were begun, namely, with the mention 



 



(de
lwedd B6368) (tudalen 368)

368 LECTURES ON WELSH PHILOLOGY. of a few points which seem to favour the conjecture that the Irish adopted it from the Kymry. In the last lecture it was suggested , that if we might venture to follow the supposed westward course of civilisation and culture, we might assume the Ogam to have made its way from Britain to Ireland: in support of this application of that generalisation, we may appeal to the analogous case of the introduction into Ireland of the Kymric way of forming the letters of the Eoman alphabet, whether as debased capitals or as the still further modified characters which have ever since been used in writing and printing Irish: nay, I would go further, as will presently be seen, and suggest that it was the very same men who taught the Irish to cut Eoman letters on stone who also taught them to do so with the Ogam, whether they were previously acquainted with the use of it on slips of wood or not. An early specimen of the more modified form of the Roman letters or, as I would term them, early Kymric letters, occurs on a stone at Inchaguile in the county of Galway, which reads in mixed capitals and Kymric minuscules Lie Luguaedon Macci Menueh; and we meet with slightly debased capitals on the Killeen Cormac stone reading iwbne drwidbs, with NE conjoint and the S reversed. The view here advocated is supported also, as far as it goes, by 



 



(de
lwedd B6369) (tudalen 369)

LECTURE vir. 369 the fact that the Ogmic method of writing fell into disuse and ohlivion much earlier in Wales than in Ireland. The same thing -would also follow from the supposition that the Celts did not invent the Ogam but adopted it from the Teutons, who may be thought to have more readily come in contact with the Celts of Britain than with those of the sister isle, whether directly, or indirectly through the Gauls of the Continent. Of Irish epitaphs in Ogam those where we meet with full case-endings form, in all probability, the oldest class. One of these is the Killeen Cormac stone, reading Uwanos Ami Enacattos, and in Latin Juvene Druvides in Roman capitals as already stated. Here the presence of the two inscriptions strongly reminds one of those of Wales, not to mention the fact pointed out on another occasion, that the Latinity is such as might have been learned in Wales. Altogether one is tempted to attribute the whole to some Irish ecclesiastic who had studied in South Wales, or at home under an Irish teacher who had derived his ideas of Latin from some such a source. In any case it dates, no doubt, after the introduction of Christianity into Ireland. Perhaps the most interesting stone in Ireland 2 A 



 



(de
lwedd B6370) (tudalen 370)

370 LECTURES ON WELSH PHILOLOGY. is that on Brandon Mountain, said to read on one of its angles, lim^y^.^+H+-/-+fH+^-++W ///// ///// I l-TTTT i.e, Qwimitirros, in which Irish archaeologists rightly recognise the genitive of a word which meant priest. Later it appears as cruimter, hut most commonly cruimther and cruimhther, genitive cruimtkir: it is repeatedly written crubthir in the Latin life of St. Oybi published in the Cambro-British Saints, pp. 183-187. An interesting article occurs on cruimther in Cormac's ■ Glossary, which is rendered thus by Mr. Stokes: — " Cruimther, i.e. the Gaelic of presbyter. In Welsh it is premter: prem ' worm ' in the Welsh id cruim in the Gaelic. Cruimther, then^ is not a correct change of presbyter: but it is a correct change oi premter. The Britons, then, who were in attendance on Patrick when preaching were they who made the change, and it is primter that they changed; and accordingly the literati of the Britons explained it, i.e. as the. worm is bare, sic decet presbyterum, who is bare of sin and quite naked of the world, &c., secundum eum qui dixit ego sum vermis, &c." The literati of the Britons are proved by the allusion to prem, now pryf, ' worm,' to have been men of considerable etymological resource, but their attempt to connect 



 



(de
lwedd B6371) (tudalen 371)

LECTURE VII. 371 premier with it must be declared a failure, the word being in fact merely the form taken in Welsh by the Jja,tin prcebitor, ' giver, supplier, purveyor.' The following hexameters quoted by Ducange under prcebendarius are to the point: — ■" Praebitor est, qui dat pisebendas: suscipiens has PrsBbendariua est, sicut legista docet nos." And prcebendarius was otherwise called provendarius — " qui provendam sen praehendam percipit," whence the Cornish proundeir, pr outer, ' a priest or parson.' If we look at the Latin praebitor it is probable that the O. Welsh form, here given as premier Oind. primter, would have been, more correctly reproduced, premitr, or, with the irrational vowel expressed, premitir or premiter, which had it not become obsolete would now be prefydr or possibly prefydwr — the equivalence of m and b has already been instanced in the case of the bilingual stone at Pool Park near Euthin. From premitr the Irish would appear to have formed Qvrimiterr, and the modification of i into u in cruimtker and crubthir must be due to the influence of the v in qv; compare the case of 0. Irish coic, ' five.' Thus the genitive Qvrimitirr-os, later cruimthir is an equivalent of the Latin prcebitor-is, whence it would seem that the genitive ending of imparisyllabic nouns in Irish was as corresponding to 



 



(de
lwedd B6372) (tudalen 372)

372 LECTURES ON WELSH PHILOLOOr. Latin -is, G-reek -o?, Sanskrit -as, which is also the way Mr. Stokes would explain Uwanos as the equivalent, on the Killeen Cormac stone, of ivvene[s] for Juvenis. But what interests one most is the qv which we find here for Latin p; and this raises the question as to who effected the suhstittltion — was it the Irish or was it the Welsh? If the latter they must have done so when they had as yet no p in their language, and when qv was the nearest approach they could make to it: in that case the Irish adopted the initial as they heard the word from Patrick or his followers, and in Welsh itself the qv here, as everywhere else, would in the course of phonetic decay be modified back again later into p. But the substitution of qv for p is a greater change than the facts of the case seem to warrant us in supposing — the usual assumption that the Irish substituted c for p ignores them altogether and is out of the question. By qv I mean the combination written qu in German, that is a velar k followed by a w pronounced by means of the lips and without the assistance of the teeth, which, on the other hand, take part in the pronunciation of English v, Welsh _/; accordingly, as Early Welsh qv has yielded p, and as the language may be supposed to have proceeded in this instance, as elsewhere, gradually and not by leaps or ' strides, I would assume the steps to have been 



 



(de
lwedd B6373) (tudalen 373)

LECTURE vir. 373 successively §•», joo, p. Now supposing the Kymry to have borrowed Latin words with jo at a time when their qv had become jow, a combination which may be heard in such French words as puis, and when they had no other jo in their language, nothing would seem more natural than that they should unconsciously substitute their pv for Latin p and make such a word as prcebitor into pvrebitr or pwemitr: when the Irish came to adopt the latter from their Celtic neighbours, they, as not being used to the sound of p, would probably be forced to change pv into qv, which is a much smaller change than the substitution of qv for p. This seems to have been also the history of the words — r 0. Ir. clum, Welsh pluf, ' plumage,' 0. Ir. corcur, Welsh porphor, ' purple,' Ir. caisc, Welsh pasc, ' Easter,' Ir. eland, Welsh plant, ' children,' from Jjatinpluma, purpura, pascka, and planfa, to which one might possibly add Irish j^dckell from an early form of the Welsh gmyddhwyll, ' chess or draughts: ' the passage of these words and of pnsbitor through or from Welsh into Irish I should assign, roughly speaking, to the time between 450 and 650. Both on account of the labialising of qv and of borrowing proper names and other words from Latin, which involved p the Kymry had occasion for a special symbol for p in Ogam: we have met with two such, and one of them was borrowed by the 



 



(de
lwedd B6374) (tudalen 374)

374 LECTURES ON WELSH PHILOLOGTi Irish to represent the later Irish p produced by the provection of h, as in poi already alluded to as occurring in an epitaph reading Broinioonas poi netat Trenalugos. But this appears to have been the result of acquaintance with the last addition to the Kyniric Ogam rather than a matter of orthographical necessity as poi continued to be written also boi '^fuit or qui fuit:'''' for instances of p and h in the Irish verb ' to be ' see the Grammatica Celtica,^^^. 491—501. After the Irish had developed the sound oi p in their own language in the way alluded to, there was, of course, no reason why they should modify it when they came to borrow ecclesiastical terms involving it from Latin in the 8th and 9th centuries: such is, probably, the origin of the majority of the words which show p in later Irish. The stone on which Qvrimitirros occurs is inscribed with a cross; the same is the case with the one reading Tria maqva Mailagni, and probably with many more, but I have no adequate information on this point. So, taking all things together, I should be inclined to ascribe the earliest Irish monuments in Ogam to the 6th or the latter part of the 5th century, and there seems to be no reason why the Ogmic method of writing may not haive been first introduced into Ireland by Kymric missionaries or by Irish ieccle- 



 



(de
lwedd B6375) (tudalen 375)

LECTTTKE VII. 375 siastics who had been educated in Wales. There is, however, a notion abroad that the Ogam was essentially pagan, but in reality it was no more so than the Roman alphabet: the only distinction we find made between them was simply this — when Latin was written, the characters used were the letters of the Roman alphabet more or less modified, but when Early "Welsh was to be written the Ogam was resorted to. Change the scene to the sister isle and one would expect to find the monuments of that country consisting of Latin in Roman letters and Early Irish in Ogam: it turns out to be so, excepting, of course, that the former are very few in number, as knowledge of Latin was probably rare as yet in Ireland — the case must have been somewhat different later when that country no longer received missionaries from other nations, but sent her own sons forth in that capacity to all parts of the west of Europe. The correctness,' however, of the view here suggested must, to some extent, depend on the answer which Irish history and archaeology can give to the question, whether there are traces of any religious establishments of Kymric origin in the south of Ireland, from which as centres the practice of writing epitaphs in Ogam might have extended itself to those parts of the island where Ogmic monuments have been found. 



 



(de
lwedd B6376) (tudalen 376)

376 LECTUEES ON WELSH PHILOLOGY. This leads to a short notice of a somewhat different theory, based on the distribution of Ogam-inscribed stones in Ireland: I allude to the following words in a paper read before the Royal Irish Academy, Nov. 30, 1867, and entitled " An Account of the Ogham Chamber at Drumloghan, County of Waterford, by Richard R. Brash, M.R.I.A." (Dublin, 1868), pp. 14, 15:— "The great majority, then, of our Ogham monuments are found in the province of Munster, and principally in the counties of Kerry, Cork, and Waterford, embracing a large extent of the south and west coast, from Tralee Bay, in Kerry, to Waterford harbour. As near as I can ascertain, the following numbers of monuments have been found:— in Kerry, 75; Cork, 42; Waterford, 26; Limerick, 1; Clare, 1. These are all in the province of Munster. All the rest of Ireland supplies but 10; of these 5 are in the county of Kilkenny, still a southern county; the others are divided as follows: — 1 in Wicklow, 1 in Meath, 2 in Roscommon [where the remaining one is we are not told]; so that for the purposes of our argument it may be fairly assumed that the three southern counties named above form the Ogham district. Again, it is worthy of remark that the majority of these monuments are found on the seaboard of the above-named counties — very many of 



 



(de
lwedd B6377) (tudalen 377)

LECTUKE VII. 377 them on the strands. The Drumloghan find is within three or four miles of the sea, as are many others of the Waterfor-d and Kerry Oghams; those found in the county of Cork are more inland." Though the late Mr. Brash's conclusions were seldom such as I could accept, he seems to have been thoroughly acquainted with the Ogam district, and I have no reason to doubt the substantial accuracy of his figures, or to suppose that subsequent finds have materially modified the ratios between them. His inference from them was that the Ogam was not invented in Ireland, but introduced by a maritime people, who landed on the southern or south-western coast of the island: he would identify them with the Milesians of Irish legend, and suppose them, accordingly, to have come from Spain, and originally from Egypt. This last piece of extravagance, which he was willing to accept, needs no discussion, but I would not go so far as to say that Ireland was never invaded from Spain, or that the Milesians went forth from Britain, but I would suggest that the Ogam-writing invaders of Ireland, if such are to be postulated, for which, I must confess, I see no necessity, are far more likely to have set sail from our own shores, say from Pembrokeshire, which is the leading Ogam county this side of St. George's Channel, than from Spain. Supposing such an 



 



(de
lwedd B6378) (tudalen 378)

378 LECTURES ON WELSH PHILOLOGY. emigration to have happened in the 5th or 6th century, one would naturally look for the primary cause of it in the westward pressure exercised by the English.



 



(de
lwedd B6379) (tudalen 379)

 (379 )  APPEINTDIX.  A.— OUR EARLY INSCRIPTIONS. Bepoee giving a list of our Early Inscriptions, a ■word is necessary as to the nature of Aryan nomenclature. The subject has been lucidly treated by Dr. Fick in his recent work on Greek personal names {Die griechischen Personennamen, Gottingen, 1874). The materials which he has there brought together clearly show that originally every Aryan name of man or woman took the form of a compound of two single words, and that this, more or le.ss modified, has come down to historical times among the various Aryan nations of Asia and Europe, excepting in Italy and Lithuania. As instances may be mentioned such names as the Sanskrit Gandrardja, from catidra ' shining ' and rdja ' king,' or the Greek @i6diasog, from ^eo's ' god ' and 33goi/ ' gift.' The number of words used in this way does not appear to have been at any time very great, but in many cases each pair yielded two names, as in the following: Sanskrit Deva-gruta, Cruta-deva, Greek Qio-bta^og, Awgo-^sos, Servian Milo-drag, Drago-mil, 0. German Hari-herht, Berht-hari, Early Welsh Barn-vend-i, Vendu-har-i, Mod. Welsh Cyn-fael, Mael-gwn. From the older class of full names most Aryan nations also formed eventually a number of shorter ones by omitting one of the constituent parts, the remaning one being used by itself, either. with or 



 



(de
lwedd B6380) (tudalen 380)

380 LECTTJEBS ON WELSH PHILOLOGT. ■without a special termination, as in the case of Sanskrit Datta from Beva-datta, Civa-daita, or the like, and Greek N/xlas, N;z/a5, 'Slx.uv, together with a good many more, from N/xo'-/ia^os, N/xo'-ffrjaros, or a sijnilar full name. By way of classifying the contents of the following inscriptions, it may be premised that they contain about 160 different names, several of which occur more than once. About 30 are either incompletely read, or, for other reasons, difiScult to classify; the remaining ones are partly Celtic and partly Latin, in the proportion of about three of the former to one of the latter. The Celtic ones are of two kinds, namely, those which belong to the Aryan system of names and those which do not. The latter are comparatively few, and may have originally been epithets or qualifying words appended to the full names: (a) some nine or ten of them seem to be quasi-compounds, such as Mucoi-breci and Maqyi-treni, while (fi) about half a dozen are adjectives formed from common nouns by means of the affix ac or oe, Mod. Welsh awg, og, such as Bodvoci, Der^aci, Lovernaci, Senacus, Tegernacus, Tunccetace. The former may be classified as follows: — 1. Considerably more than one half of them are compounds made up of two simple words, and of these last (a) the greater number are of four syllables, such as Barri-vendi, Netta-sagru; others have been reduced to three syllables by the loss of the connecting vowel, as in the case of Clotuali for Glutovcdi. (j8) A few beginning with prefixes such as so- or do-, as in So-lini, J)o-hunni, may be regarded as having never been more than three syllables long, while, on the other hand, we have no certain instance of a compound of more than four syllables in length, excepting (y) those involving tigirn or tegern, as Gato-tigirni and Tegerno-mali: it is doubtful whether the e in Camelorigi be not an irrational vowel, which would reduce the name to four syllables. (3) To these must be added two derivatives



 



(de
lwedd B6381) (tudalen 381)

 APPENDIX. 381 from full names, namely, Cunacenniwi and Qvenvendani, which imply Cunacenni and Qvenvendi respectively. 2. Names of the type of Gwyn 'white,' Arth 'a bear,' are not unusual in Welsh; but to one looking at the meaning of such words it is seldom apparent why they got to be used as proper names, while the analogy of the nomenclature of other Aryan nations makes it certain that they mostly came to be so used, not so much by virtue of their fitness in point of signification, as by way of abbreviation of full names: thus Gwyn, for instance, stood originally for some such a form as Gwyndaf or Penwyn, and Arth for Arthgen, Arihfael, or the like. Our early inscriptions yield us the following instance^ in point: Bandus, Bladi, Broho, Comne, Cavo, Daari, Magli, Meli, Porius, Qvid, Tren, Valci, Vetta, to which may be added Rialo-brani as probably involving the Goidelo-Kymric name Bran qualified by an adjective: compare English names like Littlejohn. 3. The shorter forms are more usually met with in our Early Inscriptions with special affixes appended to them. The most common of these is -agn-i, as in Broccagni, Corbagni, Gurcagni, Gurcagnus, Ercagni, Maglagni, Ulcagni, Ulcagnus, to which must be added one in -egn-i, namely, Gunegni. Besides these we have two in -mi-i, Fanoni and Vendoni (twice); two in -uc-i, Fannuci and Swaqqvuci; two in -ic-i, Berici and Torrid; two in -iv-i or -iw-i, namely, Ercilivi and Nogtivis, to which it is right to add the name ending in -urivi on the stone lost at Llandeilo: this termination may, as was seen in the form Cunacenniwi, be used in the case of a full name. The remaining derivatives are few and various. Besides the foregoing names we have about twenty epithets or qualifying words attached to the former in our inscriptions. Of these about two thirds are of Welsh origin, while the rest is Latin. Altogether they are far 



 



(de
lwedd B6382) (tudalen 382)

382 LECTURES ON WELSH PHILOLOGY. more miscellaneous than the names they accompany: one of them, Ordous, seems to refer to the tribe of the person commemorated; one, Gocci, ' red,' is an ordinary adjective, and Tovisaci, 'having the lead, a leader or prince,' is a noun of adjectival origin; next comes Maqvirird, which may be a quasi-compound. Then we have two adjectives compounded of a noun and an adjective, as in many other instances in Welsh — I allude to Anate-mori, ' soul-great,' and Ei-metiaco, ' cere-Jiastatus.' Lastly, passing by Seniargii and Kedomavi as obscure, we come to Bwrgo-cavi, Duno-cati, Il-wweto, and Monedo-rigi, which may be guessed to mean ' city-guardian,' ' town-warrior,' ' much-speaking ' and ' mountain-king ' respectively. It is almost needless to state that Early Welsh names hardly contain anything that does not find its continuation or counterpart in those of later periods in the history of the language; but to do justice to this would, to judge only from the materials I have already collected, probably require a larger volume than this. It may, however, be here pointed out that the printed books containing the greatest number of Welsh names are the following: Liber Landavensis, the - Camhro-British Saints, the Myvyrian Archaiology of Wales, Annales Camhrice, Brut y Tywysogion, the lolo Manuscripts, and t/ie Mahinogion. The best collection of Breton names known to me is that in the indexes to De Courson's edition of the Cartulaire de I'Abbaye de Redon (Paris, 1863): a number of Cornish names occur in the manumissions in the Bodmin Gospels, published by Mr. Stokes in the Revue Celtique, i. pp. 332-345. For Irish names I have used the indexes to the Annals of the Kingdom of Ireland by the Four Masters (Dublin, 1856), the Martyrology of Donegal (Dublin, 1864), Reeves' Adamnan's Life of St. Columha (Dublin, 1857), and other books. Lastly, my references to Teu-



 



(de
lwedd B6383) (tudalen 383)

APPENDIX. 383 tonic names are based on the first volume of Dr. Ernst Fbrstemann's AUdeutsches iVaJwenSwc^ (Nordhausen, 1856). A word now as to the formulae of the inscriptions. As a rule they are of the simplest kind — occasionally, for instance, the whole inscription consists of only one proper name, but more frequently it is followed by that of the father of the person commemorated, making A. films B., or in the genitive A. fili B. with corpus or sepulcrum to be supplied by the reader, to which one may add that any personal name used may have an epithet or defining word attached to it. In other instances we have hie jaeit, but the adverb, written also ie, may be omitted, while, on the other hand, we once meet with jam ic jadt and with hie in tumulo jacit or in hoc tumulo iacit. And as to iacit ioTJacet, it is to be noticed that it is the form regularly used, there being only one certain instance in which iacet is known. The substitution of -it for -et in this word may possibly be altogether due to Welsh influence, as -it seems to have been formerly the prevalent Welsh ending of the third person singular of the present indicative of the active voice, while -et was probably more usually associated with the' imperative or potential mood. However, it is right to add that Frohner, in his preface to his very handy little book entitled Inscriptiones Terras Coctce Vasorum Intra Alpes, Tissam, Tamesin Bepertce (Gottingen, 1858), cites, p. xxvi., the forms habit, valiat, hahiant, porregerit { = por- 7-igeret), cessissit, a,nd potuissit. Lastly a t is prefixed to every epitaph which happens to be accompanied- by a cross of any kind on any part of the stDne inscribed, as well as when the stone itself l^s been fashioned into the form of a cross, which is seldom the case. In a few instances the monogram of Christ forms the heading, which is here indicated by prefixing the Greek letters, XPI, which it implies. 



 



(de
lwedd B6384) (tudalen 384)

384 LECTURES ON WELSH PHILOLOGr. ANGLESEY. 1. Hie Jacit Maccudecceti (Penrhos Lligwy). This name may be treated as Maccu Decceti or Maccu-Decceti: as to maccu see Appendix B., and as to Decceti and Decheti see pages 174, 175, 176 180, 181, 203, 274, 277. 2 origi Sic Jacit (Llanbabo). I have not seen the stone, and this is all I can guess with any approach to certainty from the fac-similes of the inscription in Hiibner's collection. The name intended is evidently of the same formation as Gamelorigi and the like. 3. Culidori Jacit Et Orvvite Mulier Secundi (Llangefni). Here mulier would seem to mean uxor. Culidori is a name I cannot trace later, but Orvvite, on which see pages 210, 211, may be the early form of Erwyd in Ponterwyd ■ in Cardiganshire, which druid-mad charlatans are sometimes pleased to transform into Pont-derwydd. The formula of this epitaph stands alone. 4. Hie Beatu^s] Saturniiius Se\_pultus\ [J]aeit . Et Sua Sa^ncta] Conjux . Palx] (Llansadwrn). The stone has been damaged so that the inscription is incomplete: Hubner makes svasa into Suasa, but such a name is quite unknown to me, and as the line is incomplete I have ventured to suggest SIM sancta as the full reading, but this is only to await a better guess. 5. Gatamanus Rex Sapientisimus Opinatisimus Omnium, Eegum (Llangadwaladr). Gatamanus occurs later as Gatman, Gadfan, and as to King Cadfan and his name see pages 168, 169, 212, 323. CARNAEVONSHIEE. 6. Meli Medici Fill Martini Jacit (Llangian). This would seem to mean Gorpus Meli Medici Hie Jacit: Mel occurs as the name of a disciple of St. Patrick, and first bishop of Ardagh [Four Masters, under the year 487); it



 



(de
lwedd B6385) (tudalen 385)

APPENDIX. 385 also enters into the composition of several proper names of men, such as Melldeyrn. With medici here compare fahri on one of the stones at Tavistocji. 7. Veradus Pbr Hie Jacit (Cefn Amwlch). There is nothing to prevent our regarding Veradus as the same name as Guroc in the Liber Landav., p. 170, if it be not of Latin origin. 8. Senacus Prsb Sic Jacii Cum Multitudnem Fratrum Prespiter . . . (Cefn Amwlch). Multitudine is shortened one syllable, and ended in a silent m (see p. 208). Senacus would seem to consist of sen-, whence our hen ' old,' Ir. sean, with the affix ac attached to it, and to be exactly equivalent to the Irish name Seanach; however, it is unusual to attach the affix ac to an adjective, and but for the Irish name one might explain Senacus as meaning Senacus from the sen- possibly implied by our Mod. Welsh hoen ' vigour, liveliness.' 9. Jovenali Fili Eterni Hie Jacit (on the farm of Ty Corniog in the parish of Llannor). The first name is better known as Juvenal, and appears in the Liber Landavensis (pp. 166, 259) as Jouanaul, a form which it assumed, instead of the Jouenaid to be expected, probably under the influence of the 0. Welsh Jouan ' John,' with which it may have been popularly associated. The other has survived in the name of Llanedern or Edern, still borne by a neighbouring village. 10. Yendesetli (buried in the same place with the last mentioned). The name survives as Gwennoedyl (Gambro-Brit. SS., pp. 267, 268), Gwynhoedl {lolo MSS., p. 141), Gwynoedl, Ghvynodl {Myv. Arch. pp. 741, 426): the last of these is borne by the neighbouring church of Llangwynodl, now commonly curtailed into Llangvmodl or Llangvmadl, the founder of which is supposed to have lived in the 6th century: see the passages alluded to in B 




 


(de
lwedd B6386)

(tudalen 386)

386 LECTURES ON WiSLSH PHILOLOGr. the Gamhro-Brit. SS. There can, I think, be little doubt that the stone bearing the foregoing inscription was meant for him. 11. Alhortus Eimetiaco Hie Jacet (JAs,TX&&\h.ai2icri). This is the only instance perhaps, we have of jacet in its correct Latin form. Alhortus is read also Ahortus; see pages 205, 279. If the correct reading is Alhortus, it is probably to to be analysed into Alhrort-, of which the syllable alh has been mentioned page 279; the other, ort, may be the same which occurs in the form Orth as a man's name in Lewis Morris's Celtic Remains, p. 176': it may be of the same origin as the Latin portare. As to Eimetiaco, see pages 179, 207, 215, 225, 279, and Appendix C. 12. Fill Lovernii Anatemori (Llanfaglan). Here corpus or sepulcrum is to be supplied, but even then it is not easy to say how it is to be construed: it can hardly mean Anatemori Fili Lovernii, for the arrangement of the words is against that view, and Anatemori looks more like an epithet than a leading name; nor can I accept Hiibner's reading it upwards — it stands thus:  FILI LOVEENII ANATEMORI  So I am inclined to regard it as being FilinLovernii Anatemori, which, but for the inscriber's wish to show off his Latin, would most likely have been left Maqvi-Lovernii Anatemori: compare Maqvitreni in Ogam, and Maqyeragi in Ogam and in capitals. As to other points connected with this epitaph, see pages 209, 212, 216. 13. XPI. Carausius Hie Jacit In Hoc Congeries Lapidum (Penmachno). I cannot explain the bad Latin of this inscription as far as concerns gender, but with the 



 



(de
lwedd B6387) (tudalen 387)

APPENDIX. 387 s of congeries compare tlie case of Nogtivis, p. 208;. nor can I find any trace of the name Caratisius in later Wekh. 14. Cantiori Hie Jacit Venedotis Give Fuit \C'\ons6brino Ma[g\U Magistrati (Penmaclino). Such genitives as magistrati were usual as early as the time of the Gracchi, nor does the inscription contain a single fault which is not justifiable on Latin ground: see pages 168, 179, 213, 215. The c of consohrino is tolerably certain, and so is the g of Magli, which appears later a,s Mael, and enters into the composition of other names: the Irish form is mdl, said to mean ' a noble, a prince, a king,' and not the maol or mael of such Irish names as Maolpadraig or Maelpadraig, 'the tonsured servant of P.,' which is more likely to be the formal equivalent of our moel ' bald, without hair, without horns.' As to Cantiori, I would regard it as a nominative standing for an earlier Cantiorix, and would treat the whole as meaning — Cantiorix Hie Jacet: Venedotius Civis Fuit, Gonsohrinus Magli Magistratus, which is tolerably simple Latin, whatever may be said of its elegance. But I should add that Professor Hiibner construes it thus: Cantiori. Hie iacit, Venedotis cive(s) fuit, \e^nsohrino{s) Ma[g]li magistrati. 15 oria Ic Jacit (Penmachno). This is a part of an inscription probably commemorating a woman. 16. iSanct. . . . Filius Sacer[dotis . . . .] (Tyddyn Holland, near Llandudno). The stone is described in a book entitled " The History and Antiquities of the Town of Aberconwy and its Neighbourhood, with Notices of the Natural History of the District, by the Eev. Robert Williams, B.A., Christ Church, Oxford, Curate of Llangernyw" (Denbigh, 1835). At page 137 it is said that the following inscription was copied from the stone in question in the year 1731: 



 



(de
lwedd B6388) (tudalen 388)

388 LECTURES ON WELSH PHILOLOGY. SANCT ANVS SACEI ISIS This convinces me that the epitaph was then as incomplete as it is now, for I feel confident that what was then read ANTS is the FiLivs of the above reading, which is, it is true, far from certain. But since then the difficulties of the inscription have been greatly increased by the fact that the late occupant of the cottage close by undertook to deepen some of the letters for the benefit of English tourists. As it now stands, the ct of the old copy is a big D reversed. I have failed to read isis, or to satisfy myself that the line of which it formed a part was ic iacit. If it formed an epithet to the father's name it would be useless to attempt to guess the original. The reaiiing of the rest of the inscription was probably either Sanctus Filius Sacerdotis or Sancti Filius Sacerdotis with Sancti for Sanctis = Sanctius: one of these perhaps is implied in the O. Welsh name Saith (Liber Landav., p. 200), and probably also in Sant, the legendary name of St. David's father. Or else it may have been Sanctanus or Sanctagnus Filius Sacerdotis; for, that Sanctagnus or Sanctagni occurred as a name used at one time by Kymric Christians is rendered probable by the Welsh derived form Seithen-in, and by a passage in the preface to Saneldris Irish hymn in the Liber Hymnorum which is thus rendered by Mr. Stokes: " Bishop Sancton made this hymn, and when he was going from Clonard westward to Mat6c's Island be made it. And he was a brother of Mat6c's, and both of them were of Britain, and Mat6c came into Ireland before Bishop Sancton." According to another account they were grandsons of Muireadhach Mulndearg, king of Ulidia, who is 



 


(de
lwedd B6389)

(tudalen 389)

APPENDIX. 389 stated to have died in the year 479 {Four Masters, ii. p. 1190): Matbc is most decidedly an early form of our Welsh Madog. With Sanctanus compare Justanus, the name of a bishop of St. Patrick's creation in Ireland. DENBIGHSHIKE. 17. Broliomagli Jam Ic Jacit Et Uxor Ejus Caune (Voelas Hall, near Bettws y Coed). Brohomagli and Caune are nominatives: see pages 177, 179, 181, 203, 204, 223, 276. 18. Vinnemagli Fili Senemagli (Gwytherin). The second name occurs also as Senomagli: it should in later Welsh be Henfael, but it does not seem to occur, while Vinnemagli duly appears as Gwenfael, in the lolo MSS., p. 144, for an intermediate Vennemagli. 19. Saumilini Tovisad {in capitals) | (Pool Park, near S — helino \To\wisaci {in Ogam) ) Euthin). The difficulties of this inscription have been noticed on p. 290: Tovisad is undoubtedly the early form of our tywysog ' a prince, a leader,' but as it is left untranslated, it is likely to have been here regarded more as an epithet than an indication of the man's rank. FLINTSHIRE. 20. Hie Jacit Mulier Bona Nobili (Downing, brought from Caerwys). Here mulier bona may possibly have been meant as an equivalent for the Welsh gvireigdda ' good wife,' and Nohili, for Nohilis, was, I am inclined to think, her husband's name: if it is to be treated merely as the ordinary adjective nohilis, the epitaph has no parallel on Kymric ground. MONTGOMERYSHIRE. 21. Hie \In\ Tumulo Jacit E\e\stece Filia Paternini Ani XIII In Pa (Llanerfyl). The inscription is not alto- 



 


(de
lwedd B6390)

(tudalen 390)

390 LEOTUEES ON WELSH PHILOLOGY. gether legible, and it is impossible to guess with certainty the second letter of the first name; but it was probably E or 0. Ani stands probably for Annis, but the age look? rather like an ixiii. MEEIONETHSHIEE. 22. Oavo Seniargii (Llanfor). Others may prefer dividing it into Gavoseni Argii: it is written like one word, though it can hardly be one. As to Cavo, see pages 215, 223: Seniargii is difficult to explain beyond the fact that it probably stands for Senja-argii and not Senaargii, as the latter would have yielded not Seniargii, but Senargii; it is further possible that -ii is the antecedent of our modern termination ydd in personal nouns such as cynydd ' a huntsman,' from cvm ' dogs,' dilledydd ' a tailor,' from dillad ' clothes:' the same perhaps applies in the case of Lovernii . see pages 209, 215, 216, 223. Lastly, it should be mentioned that what I have here supposed to be // should possibly be read H, which sometimes in Roman inscriptions resembled ||j but it is hardly probable. 23. Porius Hie In Tv/mulo Jacit Homo Ghristianus Fuit (Llech Idris, near Trawsfynydd). The first two syllables of the adjective are represented by the Greek abbreviation xpi: it is to be noticed that Porius stands over jacit at the end of the second line, so that it is not improbable that it is to be read after tumulo or jacit — in the former case we should have a sort of a rude couplet running thus: — Hie In Tumulo Porius Jadt; Homo ChristianViS Fuit. The name Porius survives as^ Pir in Mainour Pir (Liber Landav., p. 117), now Manorbeer, in Pembrokeshire, Pii'-o (pp. 14, 17), later Pyr. 




 



(de
lwedd B6391) (tudalen 391)

APPENDIX. 391 24. Ccelexti Monedorigi (Llanaber, brought from a farm in the neighbourhood). On Ccelexti for Ocelestis, see pages 207, 208. Monedorigi seems to be an epithet composed of monedo-, now mynydd ' mountain ' (compare the Sc. Gaelic monadh ' moor, heath '), and rlgi, for ngis, the genitive of what would have been in the nominative rl, for an earlier rlx, now rhi ' king, lord; ' so Monedorigi probably meant ' mountain-king.' 25. Pascent. This is said to have been on a stone which once existed at Towyn: the inscription is probably incomplete, and the name meant was most likely written Fascenti with a horizontal I. 26. Rec Jacet Salvianus Bwrsocavi Filius Gupetian. This is reported to be the reading of a stone which was found at Caer Gai, near Llanuwchllyn: it has long since been lost, but the inscription may be conjectured to have been Hie Jacit Salvianus Burgocavi Filius Oupetiani, also with a horizontal I, which the antiquaries of former days did not always copy, as they did not know what to make of it. Salvianus and Cupitianus are Roman names which are otherwise known in Britain: see Hubner's Inscr. Lat. Brit, Nos. 986 and 887. Burgocavi, which is here a nominative, evidently involves the name Gav-o, which we have on the Llanfor stone in the same neighbourhood: the common element in Gavo Seniargii and Burgocavi very possibly implies the blood-relationship of the two men meant, and it is natural to conclude that Gaer Gai, which translated into an older form must have been Gastra Gavi or Gavi Gastra, bears the name of some person of the same family, perhaps of this very Burgocavi mentioned in the lost inscription. S is very frequently misread for G in our Early Inscriptions, and the name here in question was probably Burgocavi, in which we sholild in that case have the Welsh equivalent of hurgh, borough: horg was 



 



(de
lwedd B6392) (tudalen 392)

392 LECTDEES GN WELSH PHILOLOGy. used by Cormac and others as an Irish word for town. So Burgoeav-i would mean ' he who watches over, provides for, or takes care of the town.' As to the origin of Cavo and Caune, suggested at page 223, see Curtius' " Outlines of Greek Etymology," No. 64, and compare the names Ayii/iOxouv, 'iw'Troxotav, and the like. CAEDIGANSHIRE. 27. t Bandus Jacit (Silian). The first letters of this inscription are bisected by the shaft of a small cross horizontally placed before the epitaph. 28. Corhalengi Jacit Ordous (Cae'r Felin Wynt, near Penbryn). As to Ordous and the case of Corhalengi, see pages 177, 207, 212. Corhalengi appears to be composed of corba, of the same origin as the Irish corb, which Cormac mentions as meaning ' a chariot: ' the Welsh words related are corf, corof, corfan, carfan, whence corba in Corhalengi may have meant ' a beam,' ' a frame-work,' or ' a chariot.' The other element in Corhalengi. msLy be of the same origin as the Irish lingim " salio; " but I am rather inclined to regard it as the Celtic equivalent of Latin longus, English long, and this would harmonise with Evolengi should that turn out to mean ' long-lived.' 29. Velvor Filia Broho (Llandyssul). It is not evident whether the inscription is complete or not, but I am now inclined to think it is. Broho we have already met with in Brohomagli .- see pages 177, 181, 203, 204, 276. As to Velvor, it is to be divided into Vel-vor, of which vel- stands for val-, as in Clotuali and Cunovali, and represents a pree-Celtip valpa, Gothic vulfs, English wolf, but why it has e is not clear. Nay, in Forstemann's AUdeutsches Nameribuch we have the exact Teutonic equivalent of our Velvor in the feminine Wolf war from Salzburg. Eormally the vor of Velvor is best explained by supposing it to be 



 



(de
lwedd B6393) (tudalen 393)

APPENDIX. 393 the antecedent of our gwr 'man,' plural gwyr; but gwr is now only mascullDe, but that it was once feminine -or common as to gender is possible — compare dyn ' a man,' which, was habitually used in the feminine by the Welsh poets of the Middle Ages. 30. Trenacatus Ic Jack Filius Maglagni, and in Ogam Trenaccatlo (Llanfechan or Llanvaughan, near Llanybyther). The syllable tren is represented in later Welsh by tren ' impetuous, strenuous, furious: ' the other element is now cad ' battle, war,' and Trenacatus means ' impetuous in battle.' Of Trenaccatlo I can only make Tren ac Gatlo, ' Tren and Catlo,' which would now be Tren a Chadlo. Gatlo stands for Catu-lo with the same catu as in Trenacatus and Catotigemi; the meaning of lo is not so easy to guess, but it may possibly be the Early Welsh equivalent of Latin lupus, ' a wolf,' though the derived forms show not o but ov, ou in Loverni, Lovernaci, and the Breton louarn ' a fox.' Accordingly Catlo would mean ' the wolf of battle: ' other names to be compared are Cynllo and Trillo. Maglagni survives as Maelan in Garthmaelan, the name of a place in Merioneth. For some account of related forms see the remarks on the Llanfaglan stone, Carnarvonshire, and the Merthyr stone, Carmarthenshire; see also pages 212, 290. 31. Potenina Muliier (Goodrich Court, near Ross, whither the stone was taken from Tregaron). The rest of the inscription is gone, the above being on a fragment of the original stone. Above the first n there is a small hollow, which if not a mere fray in the stone may mean that one is to read nt, and to regard the name intended as Potentina and not Potenina. The name Potentinus occurs in a Roman inspription at Caerleon, and Potentina is mentioned in Becker's collection, JDie romischen Inschriften und Steinsculpturen des .Museums der Stadt 



 



(de
lwedd B6394) (tudalen 394)

394 LECTURES ON WELSH PHILOLOGY. BEECKNOCKSHIEE. 32. Rugnia — o \FiYi Vendoni (Devynock). The first name has been read Pugniacio, but I read it evgniatto, EVGNiAvio or EYGNIAVO, making into V, in the last-mentioned guess, what others have read as a sort of open followed by /. The first part of the name is no doubt represented by the later Run, Rhun, and JRugniavio or Rugniavo might be explained as belonging to that class of names which end in jaw or jo, such as Oeidjo, Peibjo, and also Teilo, which is the regular Southwalian continuator of the O. Welsh Teljau, Teiljmi. In the Liber Landavensis (pp. 31, 86, 96) it occurs also written Teliau-us and Teliau-i, which come pretty near our Rugniavo; but as this is a genitive, the nominative must have been either Rvgnjus or Rugnjaus, and so in the case of Teilo probably, and all names of the kind. The two first letters of Fili were on a part of the stone which has been cut off, but I do not think that there is a letter wanting at the beginning of the first name, which, as it now stands, begins with a good R, and there is no excuse for reading it P. The name Vendoni occurs also on one of the Clydai stones, and seems to be continued in the Welsh feminine Gwenonwy. 33 Filius Victorini (Scethrog, near Brecon). The first name is hopelessly gone, owing to the stone having been used as a roller: I have guessed it to be Nemni, whence Nemnivus. 34. Dervaci Filius Justi Je Jacit (on an old Roman road in the neighbourhood of Ystradfellte). If Dervaci be a Latinised form for jDervacis, Dervaciiis, then the name may be analysed into an adjective formed by means of the affix -<Sc, from derv-, now derw, ' oak.' A Justus, traced by some to Wales, assisted St. Patrick in Ireland. 35. Turpilli Ic Jacit Puveri Triluni Dunocati, and in Ogam Turpil\li Tri\lluni (Glan Usk Park, 



 



(de
lwedd B6395) (tudalen 395)

APPENDIX. . 395 near Crickhowel). The Ogam notches for the first i are gone, and the first I in Trilluni is somewhat doubtful. Some think there are Ogams on the top of the stone after Trilluni, but I can make nothing of them. On most of the peculiarities of this inscription see pages 21, 167, 175, 176, 177, 178, 182, 210, 211, 220, 300. Triluni no doubt stands for Trilluni, the first element in it being the Welsh numeral for ' three,' which must have the I doubled after it, as Trilluni would be the representative of an earlier Tris-luni. The name may, therefore, be explained as Triformis by identifying lun with our modern word llun, 'shape, form; ' but this can hardly be said to be confirmed by Lunar\c\hi on another stone. 36. t Gunocenni Filius Citnoceni Hie Jacit, and in Ogam Cunacenniwi Ilwweto (Trallong, near Brecon). On this epitaph see pages 30, 162, 172, 173, 177, 178, 211, 212, 296, 300, 301; and as to Ilwweto, see pages 210, 300. It may be added that the word is probably to be analysed into Il-wwefo, whereof il is identical with ill in Illtud, Illteym, and, probably, with el in the 0. Welsh names Eljud, Elhearn, and the like — Illtud also occurs as Eltvtiis: in Irish it is always il, which is an U-stem, meaning ' much,' and of the same origin as Greek woXu?, Ger. viel. The o of wweto would seem to be the ending of the genitive, for an earlier -os, and the whole appears to be identical with Fifho, the genitive of the O. Ir. name Feth (Stokes's Goidelica,^ pp. 84, §5). Fetk and wweto come perhaps from the same source as gwed in the Welsh verb dy-wed-yd, ' to say, to speak: ' if so, Ilvrweto might be explained as meaning ' much-speaking,' or possibly ' much-spoken-of: ' compare TIoXu^riTtjs, UoXuptj/jLoc, and the like. 37. Adiune (Ystradgynlais). This is probably a fragment, but Adiune seems to be a nominative feminine: see page 217. 



 



(de
lwedd B6396) (tudalen 396)

396 LECTUEES ON WELSH PHILOLOGY. 38. Hie Jacit .... (Ystradgynlais). This is also a fragment, but it is distinct from the preceding one. 39 «... curi In Hoc Tumulo \Jacii\ (Abercar, on the way from Merthyr Tydfil to Brecon). There is no reason to think that this inscription is incomplete, but the stone has been built into the wall of an outhouse at Abercar. 40. Tir . . . Fili\u,s Catiri. This belonged to the same neighbourhood but has been destroyed or lost; the first name is said to have been read Tiheriiis; and Catiri, also given as Catai, is otherwise unknown to me. GLAMOEGANSHIEE. 41. Vendumctffli Hie Jacit (Llanillteym, near Llandafif). This inscription is in early Kymric letters. Vendu in Vendumagli is identical with the first part of Vendoni, of Vendubari, of Vendesetli (otherwise Yennisetli), and of Viniiemoffli, which is, in fact, the same name as Vendumagli in spite of the difference of spelling. 42. Tegernactis Filius Marti Hie Jacit (in a field near Capel Brithdir). The letters are very rudely cut, and the G marks the transition from G to 3, being of the same form almost as an inverted Z or an angulated *S^, and identical with the g on the Inchaguile stone in the county of Galway: some of the other letters are Eoman capitals, but the early Kymric character prevails on both stones. Tegernacus is now Teyrnog, Irish Tighearnach, Anglicised Tierney; Marti is probably the genitive of Martins. I have found no other trace of it in Welsh nomenclature. 43. t Bodvoei Hie Jacit Filius Catotigirni Pron/epus Eternali Vedomavi (on a mountain near Margam). Some of these forms have been discussed pages 31, 92, 207, 212, 213, 223, 224. The name Bodvoei is said to occur as Boduacus on a 



 



(de
lwedd B6397) (tudalen 397)

APPENDIX. 397 stone dug up at Nismes in France (see the Arch. Gamhrensis for 1859, p. 289). In that case I should treat Bodvoci as a modification of Bodvaci, and analyse it, like Dervaci, into Bodv-dc- or Bodv-oc-, with bodv- of the same origin as in the Gaulish Boduo-gnatus and the \C'\aihv,'bodvae of a Gaulish inscription, in which Mr. Hennessy recognises the Badh-catha or war goddess of Irish mythology (see his interesting paper in the Rev. Celtique, i. 32-55), which we meet with as a man's name, Boducat, in the Cambro-Brit. SS., pp. 105, 123; we trace bodv- also in the name which is variously written Blbodtiffo, Elbodg, and Elhodu in the Amiales Gambnae, pp. 10, 11. 44. Punpeius Garantorius, and in Ogam Pope (Cynffig, near Margam). The first name does not seem to appear elsewhere on Welsh ground, but Garantorius may possibly be identical with the Gerentir-i, Gerennhir, Gerenhir, Gherenhir of the Liber Landav.; pp. 175, 191, 202, 203, 228, 230. As to other points connected with this inscription, see pages 21, 206, 207, 215, 301 of this ■volume. 45. Macaritini Fili Beri\d\ (The Gnoll, near Neath, whither it was brought from the parish of Llangadog). There is some doubt as to the last letters of the father's name: both that and the son's are otherwise unknown to me. Macaritini stands probably for Maceratini, and is a derivative from the name given as Macerati by Desjardins in his Notice sur les Monuments Epigraphique de Bavai et du MusSe de Douai (Paris, 1873), p. 136. 46. PavXi . . . Eili Ma ... (a fragment at Merthyr Mawr, near Bridgend). These names may have been in full Paulini and Maqveragi or the like. 47. . . . ic, in Ogam on the Loughor altar: the rest is not to be made out with any certainty: see page 302. 



 



(de
lwedd B6398) (tudalen 398)

398 LECTURES ON WELSH PHILOLOGY. CAEMAETHENSHIEE. 48. Hie Jacit Ulcagnus Fius Senomagli (Llanfihangel ar Arth). The first name occurs also in Cornwall, and in an Irish inscription as Vlccagni: the nearest form which survives in Wales is perhaps Ylched in Llechylched, in Anglesey, and -wlch in the name Ammwlch, for Amb-ulc-, in Cefn Ammwlch, in Carnarvonshire, and possibly in Llanamwlch, near Brecon: see pages 205, 206. 49. Qvenvendard Fill Barcuni (Parcau, near Whitland): see pages 22, 23, 170, 254, 281. 50. Gurcagni Fill Andagelli (Gelli DywyU, near Newcastle Emlyn). Gurcagni survives in the form Circan in the lAher Landav., p. 153, and on Irish ground both Corcan, probably for Corcdn, and the shorter Gore are to be met with as personal names: they may possibly, if standing for score, be of the same origin as scale, " servus," in Teutonic names. Andagelli in its first element reminds one of the Gaulish forms Andecumhorius, Anderovdus, Andecamulum, &c. The other element gell- seems to meet us in Gellan {Liber Landav., pp. 138, 146, 193, 195), and it may perhaps be of the same origin as the verb gallu, ' to be able; ' but nothing certain could be said of the composition of the word as long as no modern form of it is known. 51. £arrivendi Filius Venduhari Hie Jacit, and in Ogam . . . Magyi M . . . (Llandawke, near Laugharne): see pages 171, 212, 298. These names are in Irish Bairrfhinn and Fimibharr, of which the former is in Welsh Berwyn, and the latter would be Gwynfar, but I am not aware that it occurs: the meaning of the former is ' white-topped or white-headed.' 52. Mavoh .... Fili Lunar\c\hi Cocci (Llanboidy). The first name is incomplete, owing to the end of the stone having been broken ofi^, and it is possible that 



 



(de
lwedd B6399) (tudalen 399)

APPENDIX. 399 Lunarchi had no c. As to the former, it may have been in iull Mavo-heni, for an earlier Mavo-seni dating before Welsh s began to be changed into h: see pages 223, 224, 278. 53. . . . turn . . . This is all that is legible of another inscription at Llanboidy: the stone now stands erect in the churchyard, but it must have long lain in a very different position, as it is worn smooth, the foregoing being the only legible portion of an epitaph which probably contained the formiila Hie In Tumulo Jacit. 54. t Bladi Fili Bodiheve, and in Ogam Awwi Boddih .... and Bevrw . . . (stone found at Llanwinio, taken to Middleton Hall, near Llanarthney): see pages 217, 218, 299. The reading of Bladi is doubtful, but if it should turn out to be Bladi, this would probably be found to be of the same origin as hlavd in Anhlaud in the Camhro-Brit. SS., p. 158. In Davies' dictionary hlawdd is quoted as meaning " agilis, celer, gnavus, expeditus, impiger, properus,'' and the compounds aerfiawdd, cadflawdd, cynjiawdd, gorflawdd, trablawdd are mentioned. Bladi cannot, I think, be identified with blaidd ' a wolf.' 55. Gaturugi FUi Lovernaci (Merthyr, near Carmarthen). The i of Gaturugi is horizontally placed, and rather faint, but I think it is there. The name analyses itself into catvr, identical with the cato- of Catotigerni, Mod. Welsh cad ' battle, war,' Irish cath, the other element, rug4, is not easy to identify, but it may be presumed to be the same which we find in a longer form in Eugniavo, and if it be of the same origin as our modern rhu-o ' to roar,' Latin rugire, Caturugi would mean he who roars in battle; but the older meaning of the root rug seems to be to break, in that case the name would mean he who breaks tlie battle. Lovernaci is of course of the same origin as Lovernii, and both come from a shorter loverrin, which, though lost in Mod. Welsh, occurs in Cornish as 



 



(de
lwedd B6400) (tudalen 400)

400 LECTURES ON WELSH PHILOLOGY. lowern ' a fox,' Bretoa louarn, the same, Irish Loam, Anglicised in Scotland into Lome. Traces of it occur in the lAher Landav., pp. 135, 166, 251, in Gruc Leuyrn, Louern, and Crucou Leuim, and several localities in Wales are still known by the name of Llywemog, which would be formally identical with Lovernac-i, but meaning probably ' abounding in foxes,' whereas as a man's name it is more likely to have meant ' foxy, or like a fox.' Lovern- possibly stands for Za[p]-arK-, from the same origin as Latin lupus ' a wolf; ' the simple form perhaps occurs as lo in Catlo and Cynllo, which last can be matched by a Conlouern from the Liber Laiidav., p. 146: see also the remarks on Lovernii in No. 12. Others connect lovern- with Laverna. 56. Gorbagni Filius Al . . . (Pantdeuddwr, near Abergwili). The second name begins with A, followed, I think, by an L, which suggested to me the name Alhorti. Gorhagni is a name which also occurs in an Irish inscription, and I would identify it with Garfan in Llancarfan and Nantcarfan, in the Liber Landavensis Nant Garban and Vallis Carbani. As to the change of vowel, compare corfan, ' a metrical foot,' with car/an as in car/an gwehydd, ' a weaver's beam,' car/an gwely, ' a bedstead,' carfan o wair, " hay laid in rows," which I copy from Pughe's dictionary, where one meets with the followingquotation fromSalesbury: "Eisteynt yn garfanau ofesur cantoedd, a deg adeugeiniau;" " they sat down in rows of the number of hundreds and of fifties." Hence it would seem that corfan and carfan are desynonymised forms of the same word. See also the remarks on Gorhalengi in No. 28. 5f. + Gunegni (Traws Mawr, near Carmarthen). This name is singular in its being Gunegni and not Cunagni, which is the form analogy suggests; but it should perhaps be regarded as offering us an early instance of a modulated 



 



(de
lwedd B6401) (tudalen 401)

APPENDIX. 401 into e by the influence of i in the following syllable, a change well-known later in Welsh. In that case Cunegni would be a variant of Cunagni, which is to be regarded as the early form of the name which appears subsequently as Gonan, Ginan and Cynanl 58. Severini Fili Severi (Traws Mawr). Severi occurs in Cornwall also. 59. Regin . . . Filius Nu[v]inti (Cynwyl Caio). The first name is now incomplete, but so much of it as can be read corresponds to the later name Regin, Rein; the v in Nuvinti is also a matter of guessing, as it has disappeared from the stone, and the name is otherwise unknown to me, unless we have it in Ednywain. €0. Talo\ri\ Adven\ti] Maqv[eragi] Filiu[s] (Dolau Cothy). The parentheses enclose letters which are no longer on the stone, but were formerly read on it. Whether Adventi should not have been read Adventid = Adventicis = Adventicius, which in late Latin meant advena, it is now impossible to tell, nor can one say that Adventi = Adventis = Adventius is out of the question: further, it is difficult to decide whether it is nominative or genitive, and, consequently, whether it or Magyeragi is to be regarded as the epithet or surname. So, though I should treat Talorl as a nominative standing for an earlier Talo-rix, I have to leave it an open question whether the epitaph means Talorix Filius Adventi Maqveragi or Talorix Adventis Filius Maqveragi. As to Magyeragi, Dr. Haigh thinks that he has found it also in Ogam on a stone at St. Florence in Pembrokeshire. 61. Servatur Fidcei Patrieque Semper Amator Hie Paulinus Jacit Cu[lt]or Pie[nfi]sim[us ^qui] (Dolau Cothy). This Paulinus is supposed to have attended the synod of Llanddewi Brefi some time before the year 569: see Haddan and Stubbs' Gouncils and Ecclesiastical 2 



 



(de
lwedd B6402) (tudalen 402)

402 LECTURES ON WELSH PHTLOLOGT. Documents, i. p. 164. As to the peculiarities of the spelling, see pages 215, 216. This epitaph forms a kind of a distich: — Servatur Fidaei Patriequ? Semper Amator, Hie Paulinus Jacit Cultor Pientisimus Mqui. 62. Vennisetli Fill Ercagni (Llansaint, near Kidwelly). As to Vennisetli, which is the same name as Vendesetli, see No. 10. Ercagn-i occurs as Erchnn in the Liher Landav., pp. 146, 191, and a farm in the neighbourhood of Aberystwyth is still called Rhoserchan; we have also early forms nearly related to Ercagni in Ercilivi and Ercilinci on the Tregoney stone in Cornwall. Irish has the stUl simplei form Ere, and in Welsh erch, erckyll, means 'terrible, formidable, dismal.' 63 Jacet Cureagnus urivi Filius. This is an inscription which Edward Llwyd, in a lettei published in the Ar. Cam., for 1858, p. 345, gives as beinj at Llandilo, but nothing is known of it now — his Jacet is not likely to have been so written on the stone. 64. Decabarbalom Filius Brocagni, and in Ogaw Deccaibanwalbdis: the stone is said to have been a' Capel Mair, near Llandyssul, but it appears to be de stroyed, and the foregoing cannot be an accurate copy of it Brocagni, more correctly written, would have beei Broccagni: it is the early form of the well-known nami Brychan, and is in Irish Broccdn. See pages 181, 291. PEMEEOKESHIEE. 65. Solini Filius VendQui (Clydai). The first name i to be detected possibly in the Liber Landav. pp. 190 193, in the form Hilin, which would in that case be Hylk or Hylyn if it occurred: this would exclude the possibilit; of the name being the Koman S6linus. It would b 



 



(de
lwedd B6403) (tudalen 403)

APPENDIX. 403 interesting as giving us the early form so- of our prefix hy-. See also page 171 and No. 32 in this list. 66. Etterni Fill Victor, and in Ogam Ettern\{\ .... 7[ic\tor (Clydai). See pages 182, 293. 67. t Dob . . . . i \F\ilms Evolengi, and in Ogam DoM ...t..Ci.s.. (Dugoed, near Clydai). The final i of EvoUngi is horizontally placed in the bosom of the G, and is so faint that some maintain that there is no such a letter on the stone. If leng- means, as has already been suggested, ' long,' then Evo-leng-i may mean ' long-lived or he of the long life,' as there is no obstacle to our supposing evo- to stand for evo- and to be the Early Welsh equivalent for Latin cevum and its congeners: the Irish form is eva in Evacattos, and f roni the Continent we have Evotalis given by Frohner, p. 42, as found at Reinzabern. See also pages 206, 212, 244, 293, 294. 68. t Trenegussi Fili Macutreni Sic Jacit, and in Ogam Trlnagusu Maqyi Maqvitreni (Cilgerran). As to Trenagusu or Trenegussi, the syllable trerir is represented in later Welsh by tren 'impetuous, strenuous, furious,' and the other element appears in O. Welsh as gust in Gingust, Irish Congus, Chirgust, Ir. Fergus, JJngust, Ir. Oingus, Anglicised Angus: in Irish there are a good nM,ny more of these compounds, and they all mal?e their .genitives in o{s), as in Fergus, gen. Fergusso or Fergosso., and an inscription offers the genitive Cunagussos, whence it may be inferred with certainty that the Goidelo-Kymric form implied in these names was gustus, genitive gustos, formally identical with the Latin gustus of the f7-declension. But as the Welsh retains the st without reducing it to ss or s it is likely that the nominative was shorn of its termination at an early date: thus while a nominative Trenaausius became Trenagust, the genitive- Trenagustos 



 



(de
lwedd B6404) (tudalen 404)

404 LECTUEES ON WELSH PHILOLOGY. or Trenagustwos became by assimilation Trenagmsos, Trenagusso, Trenagussu: the retention of the t was favoured also by the accent falling in Welsh on gust which we know must have been the case with Ghiorgust, as it has passed through Gwrgwst, Gwrwst into Grwst as in Llanrwst. The use of macu and magy-i as synonymous in this inscription is to be noticed. See pages 30, 180, 211, 212, 293, and Appendix B. 69. t Nettasagru Maqvi Mucoibreci (Bridell). This is in Ogam only, and in Mucoibreci, which may be treated as Mucoi-Breci or Mucoi Breci, Bred is very uncertain: see page 292. Mucoi is the genitive of the word which in the Cilgerran inscription appears as macu and elsewhere as macco: as to the variation of the stem vowel see Appendix B. Nettasagru is to be analysed into Nettasagru, of which netta occurs several times in Irish Ogam and is rendered " propugnator " by Mr. Stokes. It probably stands for iienta of the same origin as the O. H. Ger. ginindan ' to take courage,' Gothic ana-nanthjan ' to take courage, to venture,' 0. Eng. ne^an, ' to go on boldly, to venture, to dare.' The other element sagr- comes down into later Welsh in the verb haer-u ' to affirm,' and Haer, a woman's name, in the lolo MSS., p. 21, and Lewis Morris's Celtic Remains, p. 237. The Irish form is sdr 'very' (sdr mhaith ' exceeding good '), saraghadk ' conquest, victory ' [ag saraghadh ' exceeding ')— I quote from Edward Llwyd: to these may be added Sdraid, the name of a lady who figures in Irish legend — the genitive of the corresponding masculine may be recognised in the Sagarettos of an Irish Ogmic inscription. Among related words in other languages may be instanced Sanskrit sah 'to hold, to restrain, to resist, to overpower,' Greek 'ixi^, ^X"?"^' o^uiog, but the most interesting are the Teutonic forms, 



 



(de
lwedd B6405) (tudalen 405)

APPENDIX. 405 among wHch may be mentioned Gothic sigis, German sieff ' victory ': our sagr- takes the form sigl- or sigil- in Teutonic names, so our Haer is matched by a fem. Sigila, and the Irish Sdraid letter for letter by the fem. Sihilinda, Sigilind: see Forstemann'a book, columns 1087, 1095. As applied to men sagr- and its equivalents probably meant powerful, firm, victorious, but as applied to women they, no doubt, meant firm, resisting, chaste, which afibrds us an interesting glimpse into early Celto-Teutonic morals. Both sig- and nand- enter extensively into the composition' of Teutonic names, but the nearest instance to our Nettasagru which Forstemann gives is Siginand. 70. Sagrani Fill • Gunotami, and in Ogam Sagramni Maqvi Cunatami (St. Dogmael's, near Cardigan). As to Sagramni it is not easy to say how it should be analysed; at first sight it seems to be a sort of middle participle from the early form of the verb haer-u, but analogy is in favour of the view that it is a compound; but of what elements 1 It may be Sag-ramn-i or Sagr-amriri: in the former case we should have sag- (whence the sagr-, sager-, already discussed), and ramn which is not very easy to explain. In the other case we should have sagr- and amn-, which might possibly be a derivative from the root am ' to attack, assail, injure ' (see Fick's dictionary,^ i. p. 19): the whole might then mean ' a powerful assailant.' Teutonic names show an element resembling the latter part of Sagramni in such names as Imino, Emino, Emeno, Tmnus, Ymnedrudis, Imnegisil, Imnachar, &c. (Forstemann, 777, 779). Gunatami or Gunotami is in Mod. Welsh Gyndaf, and is composed of cwre- and tamr: the former of these is a common element in proper names, and occurs as cune in Cuneglas-e and is explained by Gildas as -« X — « <£ 1nni/^ " 'Fho /^fTiaf. aTTllaV»l.a invn. la nOt Of SUch 



 



(de
lwedd B6406) (tudalen 406)

406 LECTUEES ON WELSH PHILOLOGY. extensive use, but it occurs besides in Eudaf, Gawrdaf, Gwyndaf, Maeldaf, and more than one river in Wales is called Taf — whether it is to be referred to the root tarn or stahh is not clear (Fick's dictionary, i. 593, 821). As to other points connected with this inscription, see pages 29, 182, 183, 184, 212, 293. 71. Yitaliani Mmereto, and in Ogam Witaliani (Cwm Gloyn, near Nevern). See pages 167, 176, 179, 215, 288, 294. 72. t Tunccetace Uxsor Daari Sie Jacit (Trefarchog or St. Nicholas). The name is to be analysed into ,Tunccet-ac-e and would be now Tynghedawg or Tynghedog, tunccet being now tynghed ' fate ': thus it was probably the exact equivalent in meaning of the Latin name Fortunata. As to other points connected with this epitaph see pages 206, 216,217, 244. To the remark on the doubling of the a in Daari, p. 216, add the following instances from the Continent, mentioned by Frohner, p. xxvii: — Craaniani from Eiegel, Maiaanus from Luxembourg, and JRicaamaariu from Paris; also Vaaro, from Bingen,' cited by Becker, p. 78. And, lastly, with Tunccetace compare temppistataem for tempestatem, instanced by Frohner, p. xxix. 73. Uvali Fill Dencui Cuniovende Mater Ejus (Spittal, near Haverfordwest). If evo- in Evolengi means cevum, then Evali may possibly have had the meaning of Etemalis or Vitalis in other inscriptions. Dencui is obscure: it may be either a compound Den-cui or Denc-ui, or else a derivative, in which case we should probably compare Dinui and Sagranui: it is to be remarked that the reading of it is not certain. The vend- of Cunio-vende we have already met with, but cuni-o is obscure: it would seem to be derived from cun- as in Gurw,tami. 74. Clotorigi Fili Paulini Marini Latio (LlandysUio). 



 



(de
lwedd B6407) (tudalen 407)

APPENDIX. 407 Owing to the face of the stone having begun to peel off, I am not certain whether the first name should be read CLOTOKiGi or CLTjTOEiGi. Later the name became Glotri and Clodri: the corresponding Teutonic forms in Forstemann's list are Chloderich, and Hlodericus. As to what follows Paulini, it is hard to know what to make of it: various ways of explaining it occur to me, but none of them is satisfactory. On the whole I would suggest that MAEiNiLATio should be divided into Mdrini, an epithet to Paulini, and Latio, which would then have to mean ' from Latium,' or ' from Litau, i.e., Armorica '; for Latium and 0. Welsh lAtav,, now Llydaw, used to be confounded — witness the Ovid gloss di Litau, ' to Llydaw,' intended to explain the Latin Latio: the same thing happened also in the case of Leiha, the 0. Irish equivalent of Llydaw. 75. Euolenggi Fili Litogeni Hie Jacit (LlandysUio). The letters are mixed Roman and Kymric, but there is no excuse for reading the first name Euolenus: Evolengi has already been mentioned: see page 399. As to Litogeni, it is no doubt of the same origin as the Gaulish forms Litugena, Litugenius, and partly as Litumara = O. Welsh " litimaur frequens." See pages 183, 253. 76. Camelorigi Fili Fannud (Cheriton, near Pembroke). The second element of Camelorigi requires no further explanation, but the other is more obscure. The e may be the irrational vowel which is omitted in Nettasagru as compared with the Irish Sagareftos: in that case the name might be written Camlorigi, which would conform better with the analogy of the other Early Welsh names. It is possibly of the same origin as the first part of the probably Gaulish name Camalodunum: the root is ham ' to vault, to bend, to envelop,' from which are derived xd/zivoi, xa/ioga, vliLnsoc. xdaapos; Lat,, camurus, camera; Ger., himmel} 



 



(de
lwedd B6408) (tudalen 408)

408 LECTURES ON WELSH PHILOLOGY. ' sky, heaven, canopy, roof of a carriage; ' and the Teutonic names containing the same element are Himildrvd, HimUger, Eimilvad, Berhthimil, while one might at first sight be tempted to equate Eamalri, the name of the King of France's steward mentioned in the Saxon Chronicle under the year 1123, but this is perhaps not to be thought of, as it is also written Amauri, and the h supposed to be inorganic: see Forstemann, 77, 687. As to the affix mc in Fannuci see page 282. The fann- we meet with in this name is probably of the same origin as the German verb ' spannen, to be stretched, to be in suspense,' Eng. span. 77. t Maqveragi in Ogam (St. Florence): see page 296. 78. t Magolite Bar . . . cene in Ogam (in the chapel on Caldy): see page 297. DBVONSHIEE. 79. Bohwnni Fahri Filli Enaharri, and in Ogam .... ndbarr .... (Tavistoclj:). I am not perfectly certain as to the second I in filli; see page 303, and No. 88 in this list. The first name seems to be the same as that of the tribe whom Ptolemy calls Ao^oimoi in South Wales. Enaharri contains one element, harr-, which has already been noticed: the other ena seems to be the same as the ene of Eneuin of a much later inscription nowin the chapel at Goodrich Court. It stands probably for enna = an earlier enda, which in Irish occurs as a man's name, Enna, Enda: in modern Welsh names it is of course reduced into en- as in Enfail, Enddwyn and the like, with which may be compared in the Teutonic languages, Enda, Indgar, Indulf, or else Ando, Andegar, Andarich, and the like. 80. Sahini Fili Maccodecheti (Tavistock, brought thither from Buckland Monachorum). Sahini, which would be the 



 



(de
lwedd B6409) (tudalen 409)

APPENDIX. 409 genitive of SaMnus, a name well-known to epigraphists, is perhaps to be read Sarini; for owing to a hole cut in the stone it is now impossible to say which is right. The form Sabini would be identical with Hefln in the lolo MSS., p. 108. As to Maccodecheii, see pages 163, 174-177, 180, 181, 203, 274, 277, and Appendix B. 81. (a) Fanoni Maqyirini: (b) Sagranui, and (c) in Ogam Swaqqvud Maqyi Qvici (British Museum, brought from Fardel, near Ivybridge). Fanoni stands probably for Fannoni, of the same origin as Fannud. The meaning of rin in Maqyirini is not evident, nor is one certain as to the formation of the name Sagranui: the n is written like an H, but it is not probable that it is to be read so; moreover, the H and the V are so placed as to suggest a conjoint character for MN or NN: they are not quite joined. Thus thepossible readings are Sagranui, Sagranui, Sagramni, Sagrahui. This inscription is not on the same face as that of Fanoni, nor in so early letters: see also pages 282, 303. 82. Valci Fili F . . . . aius (Bowden, near Totness). I have not yet had time to visit the locality of this stone, and I take the above from Hiibner's book — I have faUed to guess the reading of the rest, though the epitaph seems to be complete; the fac-simile seems to come originally from Gough's Camden. The first name would seem to be identical with our gwahh in Gwalchmm. CORNWALL. 83. Latini Ic Jacit Filius Ma:: arii, and in Ogam, traces of an inscription ending in i (Worthyvale, near Camelford). The father's name is partly illegible, and the final i is horizontally placed and of an unusual length: see also page 209. As Latini — there is no excuse for read- in™ '*■ ^'»''''»«' — ia poinijnatiga i t Tirnba.b1v .sta nd.s for 



 



(de
lwedd B6410) (tudalen 410)

410 LECTURES ON WELSH PHILOLOGY. Latinis = Latinius — see page 178; several instances of the simpler form Latinus are cited by Frohner, p. 50, and one of tliem seems to come from London. 84. Vlcagni Fili Severi (Nanscow, near Wadebridge). These names we have already met with: one is Celtic and the other Latin. 85. Vailathi Fili Vroclia:: i (Wilton, near Cardynham). The reading of the one name is not very certain, and the origin of both is obscure; in any case the inscription must be a comparatively late one, as proved both by the spelling and the style of the letters, 86. Annicuri (Lanivet, near Bodmin). I have not seen it, and I cannot explain the name, but the first part anni is probably to be regarded as identical with anda- in Andagelli: the rest coincides with the portion read of a name on the Abercar stone, Brecknockshire: see No. 39. 87. t Brustagni Hie Jacit Cunomori Filiws (The Long Stone, near Fowey). The first name has been read Cirusius, but what has been taken to be CI is an inverted D; moreover, the ius of Cirusius does not account for all the traces of letters on that part of the stone, but my -agni is rather a guess than a reading. Brustagni would be the early form of our Drystan; compare also the Pictish Brostan, Brosten, Brust, and other related forms. Cunomori is composed of eun-, already noticed, and mor- probably the prototype of our adjective mawj ' great ': the name is now Cynfor in Wales. 88. Bonemimori Filli Tribuni (Kialton, near St. Columt Minor). The name which here occurs as Bonemimori is tc be met with in a variety of forms, I am told, on th« Continent: filli stands no doubt ioifilji or fillji, with whicl may be compared fiUia, Julliacus, Julliani mentioned bj



 



(de
lwedd B6411) (tudalen 411)

 APPEND IS. 411 FroKner, p. xxix; also Turpilli on the Glan Usk Park stone. Filli seems to be the spelling also on one of the Tavistock stones. The father's name seems to be the Latin trihunus used as a proper name. 89. Conetoci Fili Tegernomali (St. Cubert). The lettering though clear, is rude and inclines to early Kymric, especially the G which has the form of a J" being intermediate between the Capel Brithdir specimen and the ordinary Kymric 3. Conetoci stands possibly for an earlier Cunatad or Cunotaci, but whether that would be a derivative with the suffix dc or oc, or a compound Guno-tdc4 is not evident. In the former case Conetoci fpould imply a noun conet, possibly of the same origin as connet in the Gaulish name Gonconneto-dumnus, but more likely of the same as our con in gogonedd or gogoniant ' glory/ whence Conetoci might mean gloriosua or the Uke. Compare Tunccet-ae-e, O. Welsh Marget-jud, and the Gaulish Orgeto-rix. As to Tegernomali, see pages 31, 213: it means 'king-like or lord-like ': the only other name of the same formation in Welsh which occurs to me is Jonafal [Brut y Tywys. p. 28, Myvyr. Arch. p. 659, 692): counpaie dihafal ' without a like, unrivalled,' and Breton Hiaval. The author of a life of St. Samson, who is supposed to have written in the earlier part of the 7th century, addresses his preface " ad Tigerinomalum Episcopum," where we have Tegemomalum, spelled with an irrational i: the epitaph in question is also in all probability to be ascribed to the 7th century. 90. Nonnita Erdlivi Jiicati Tris Fili Erdlind (Tregoney). Nonnita was a woman, though she and her two brothers are here termed " tris ( = tres) fili: " it was the name of St. David's mother, and has come down in Eglwys Nunyd, that is, in Welsh spelling Eglwys Nynyd, the name of an extinct church near Margam: otherwise in Welsh tradition it usually takes the shorter form Non or 



 



(de
lwedd B6412) (tudalen 412)

412 LECTUEES ON WELSH PHILOLOGT Nonn. Ercilivi and Ercilinci are of the same origin as Ercagni already noticed: on the Trallong stone we -find Gunocenni Filius Cunoceni called in Ogam Cunacennivd, and the present instance is a fair parallel, Hrcil{mci) Filius being made into Ercilivi, or, as it would be in Ogam, Erciliwi. On the termination inc, now ing, see the Arch. Camhrensis for 1872, page 302. Ricati probably means king of battle: compare the Teutonic Rihlmd (Forstemann, 1047); however it does not appear that the Welsh name is a compound, for were that the case we should expect to find it assuming the form of Eigocati or the liie: so it remains that it should be treated as consisting of a nominative ri (for an earlier rix) and the genitive of the stem catu, which would have been in Early Welsh catu or cato, making the whole word into Eicato or Eicatu, which, dealt with in the same way as Trenagusu, made into Trenegussi in the Latin version, would yield Eicati and retain at the same time quite as much of the appearance of a nominative as the Ercilivi immediately preceding it. 91. t Vitali Fill Torrid (St. Clement's, near Truro). This inscription is preceded by a group of very much smaller letters which seem to make Isnioc, which has never been explained. Vitali, for Vitalis, is a Latin name which occurs in inscriptions of the time of the Eoman occupation: see also page 176. It is not improbable that Torrid, on the other hand, is Celtic; as we have the name Twrrog which would have been in Early Welsh Turrac- or Torrac-. To this may be added from the Lichfield Gospel a compound name Turgint with gint as in Bledgint, now Bleddyn, which probably meant wolf-child, as gint, seems to be our formal equivalent to Latin gens, gentis, Lithuanian gimtis, ' race,' gentis, ' a relative,' Ger. kind, ' a child.' But I would not be certain that our torr in Torrid is the equivalent of tlm lfindin<T oioment 



 



(de
lwedd B6413) (tudalen 413)

APPENDIX. 413 in the Teutonic names, Thurismund, Tlmrismod, Thorismutk, Thundnd (Forstemann, 1200, 1201). 92. OMuali' Morhatti (Phillack, near Hayle). Morhatti, the composition of which is not very transparent, is found in the Bodmin Manumissions in various forms, such as Morhatho, Morhaiiho. The other name is easily explained: it is made up of clot, now clod ' praise, fame ' and ual, that is wal, which stands for a prae-Celtic valpa English wolf: so that Clotual- is exactly matched in Fbrstemann's list by Ghlodulf, Clodulf, Hlvdolf, Mod. H. Ger. LvdoVph: compare the case of Velvor in No. 29. 93. [ In Pa\ce Mul\ier\ Requievit n . . . . Gunaide Hie [/m] Tumulo Jaeit Viscit Annas Xxxiii (Ka,yle). The reading of this epitaph is, I fear, hopeless: as to Gunaide see pages 217, 222. 94. Qvenatauci Ic Dinui Filius (Gulval, near Penzance). Qvenatauci stands probably for Qvennatanci: see pages 211, 212, 224, 281. The name Dinui is obscure, and I cannot find a trace of it elsewhere. 95. Rialohrani Gunovali Fili (Lanyon, pronounced Lannine, near Penzance). Gunovali consists of elements which we have already noticed: it is in Mod. Welsh Gynwal, and in Irish Gonnell, see page 85. The exact Teutonic equivalent occurs as Sunulf or Sunolf (Forstemann, 762): similar instances are Gatgual (Liber Landav. p. 132) = Ir. Gathal, Gurguol (p. 157), Bvdgual (p. 263), Tutgual, Tudwal = Ir. Tuaihal, better known in its Anglo-Irish dress" as Toole; these are duly represented in Forstemann's list by Hathovulf, Waraulf, Botolf, and Theudulf respectively. The other name, Rialohrani, consists of hran, ' a crow,' which occurs as a proper name among both the Welsh and the Irish: in Rialohrani it would seem to be qualified by an adjective rial-o-, which I should take to Ta.t3.-n. friendly -J. — ,'«j,7^ frnm t.iip 1-nnt, nri ' to love. to em'ov.' whence 



 



(de
lwedd B6414) (tudalen 414)

414 LECTURES ON "WELSH PHILOLOGY. the English word /riered On Breton ground, however, it occurs as an independent name in the form of Siol. 96. XPL Senilus Ic Jacit (St. Just in Penwith). The inscriber left out the letters ni from Senilus and inserted them afterwards above the line. Whether the name is Celtic or Latin is not easy to decide; but the termination us would seem to imply that we have here to do with a form totally distinct from the Latin Senilis. SCOTLAND. 97 Hie Meinor Jacet Princ .... Dumnoceni Hie Jacet In Tumulo Duo I'm . . Liberali . . . (near Yarrow Kirk, Selkirkshire). The letters appear to be very far gone, and the reading of them, as here guessed from Hiibner's book, to be of very little value, but we seem to find in them one Celtic name and one or two Latin ones. Dumnoceni begins with the same element as Dumnorix or Duhrtorix, and the Mod. Welsh Dyfnwal, Irish Domhnal, Anglo-Irish Donnell: the syllable cen, if it is not to be read gen, stands probably for cenn, as in Cunocenni. 98. In Oc Tumulo Jacit Vetta F[ilid] Victi (In Kirkliston Parish, between 7 and 8 miles from Edinburgh). I have never seen the stone, and I am not convinced that the father's name is complete as it now stands. Scotch antiquaries usually treat Vetta as a man's name, and complete the word following into filiw, but for no better reason, it would seem, than that they think they detect in Vetta the name of a warrior of the Hengist and Horsa family. But to me the inscription appears to differ in no particular from those of Wales and Cornwall; but even if a Teutonic Vetta were meant, analogy would lead one to expect his name not to appear exactly in that form in this inscription. However, the genitive masculine correspond- 



 



(de
lwedd B6415) (tudalen 415)

APPENDIX. 415 ing to our Veita is cited by Frohner as Vetti: in one instance it comes from Xanten, and in the other from Stettfeld. Supposing, however, that the doubling of the t in Vefta is inorganic, the name would naturally connect itself with the 0. Irish Feth and Ilrwweto in No. 36.  -^.-^MAGGU, MUGOI, MAQVI, MAGW7. Lest difficulties should seem to be intentionally slurred over, some remarks will here be made on the word maccu and others related to it. The inscriptions most nearly concerned are the following: — . No. 1. Hie Jacit Maccu-Decceti (Anglesey), No. 80. Sabini Fili Macco-Decheti (Devonshire). No. 68. (a) Trenegussi Fili Macu-Treni ") (Pem- (6) Trenagusu Maqvi Maqvi-Treni V broke- No. 69. Nettasagru Maqvi Mucoi-Breci J shire). Irish inscriptions offer us not only mucoi, but also m/iiccoi and moco, and later Irish mocu and maccu, whence it appears that we may regard mucoi, in No. 68, as the genitive of a form which the Kyfnry wrote indifferently macco, maccu, maeu, or perhaps mucco, muccu, mticu. Moreover, Mucoi-breci does not seem to be a compound, and the same may be said of Maccu-Decceti in No. 1; but Macco-Decheti and MoyCtirTreni have been treated as though they were compounds, and their first element left without being changed into the genitive, as it strictly should have been. Irish enables us to analyse these forms into their constituent parts: these are muc- or mace-, which we have in Welsh in mag-u, formerly mac-u ' to nurse, to rear, to bring up,' as in the proverb ' Gas gwr ni charo'r wlad a'i i Htintnfnl '" ^ — -«tV./^ ^/^■.>r.I, Tir>f f>io lanri that rears 



 



(de
lwedd B6416) (tudalen 416)

416 LECTURES ON WELSH PHILOLOGY, him," and a word which in Irish appears as o ' a grandson, or descendant,' genitive ui — everybody is familiar with it as the prefixed with a misleading apostrophe to Irish names, as in O'Gonnell, O'Donovan, O'Mooney, and the like. The nearest related Welsh word still in use is w-yr ' a grandson,' but both have lost an initial p, and are of a common origin with the Latin jnier ' a boy.' Mr. Stokes, in the first volume of Kuhn's Beitrcege, takes the meaning of maecu or mocu to be grandson or descendant: he mentions the following instances, p. 345: — " De periculo Sancti Colmani Episcopi Mocusailni " (Adamnan's Life of St. Columha, p. 29); " Silnanum filium Nemanidon Mocusogin" (i6., p. 108); "Sancti Columbani Episcopi mocu Loigse animam" (ih., p. 210), but there lay, he says, six generations between this Columbanus and Loigis; " De Erco fure Mocudruidi " {ib., p. 77) — we meet elsewhere with Maccudruad; " Brendenus Mocualti " {ib., 220); " Quies Cormaic abbatis cluana Macconois" {Annals of Ulster, A.D., 751)— the abbey is stiU called Clonmacnoise; "Dubthach Macculugir" (Tirech. 13), which he finds transformed in the Liber Hymnorum into " Dubtach mc. huilugair," i.e., " D. filius nepotis Lugari" — the same would seem ,to have been the fate of maccu generally in later Irish. In his Goideliea, Mr. Stokes mentions two other instances, namely, MuircM Maccumacktheni, p. 84; also, p. 62, a Macculasrius in a Latin hymn for Lasridn, whence, he suggests, that maccu, may be equivalent to the diminutival ending -dn. Since the printing of the books alluded to, Mr. Stokes has communicated to me some further notes on maccu. Among other things, he finds that in Irish it had the force of " gens, genus," as, for instance, in the words " ad insolas maccu-chor" {Booh of Armagh, 9, a. 2) j moreover, that maccu or m.ocu had this meanins; is proved, he thinks, by 



 



(de
lwedd B6417) (tudalen 417)

APPENDIX. 417 its interchange with corca and dal, as in Mocu-Dalon = CoTca-Dallan (Adamnan's Life of St. Columha, p. 220, in Moeu-runtir = Bal-Ruinntir {ih., p. 47), in Mocu-Sailni = Dal-Sailne (ih., p. 29), in Mocu-themne = Corcu-temne {Book of Armagh, 13. b. 2), and Gorcii-teimnt (ib., 14. a. 1), and in the fact that the phrase " de genere Euntir " appears as a translation of Mocv^Runtir. Such instances as Colmani episcopi Mocusailni, and Goltimhani episcopi mocit, Loigse, he regards as references to the Irish tribal bishops, which should be rendered C. episcopi gentis Sailni, and C. episcopi g&ntis Loigse. Judging from our inscriptions, we have no reason to think that the Kymry used maccu in a collective- sense, and the meaning which seems to be suggested by the origin of the word and its uses is ' reared offspring,' or, perhaps, more strictly, ' offspring in the course of being reared,' that is in the singular, let us say, a child, a boy, or a young man who has not done growing, and ultimately a young man without any further restriction of meaning. This is confirmed by the fact that the same person seems to be called Maeu-Treni and Maqvi-Treni in No. 68 — in any case, the distinction between maccu and maqv-i cannot have been so considerable that they could not, under certain circumstances, both be applied to the same person. But we have other means of fixing the meaning of maccu; for the genitive mucoi, in its form of maccoi, has come down bodily into Mod. Welsh as macwy, the signification of which will be evident from the following examples: — " Myned a wnaeth i'r maes a dau faccwy gydag ef," ' he went to the field accompanied by two young men,' quoted in Dr. Davies' dictionary from Historia Owein ah Urien; in the next quotation from Cynddpw in the Myvyrian Archaiology (Gee's edition), p. 183a, the word is 



 



(de
lwedd B6418) (tudalen 418)

418 LECTURES ON WELSH PHILOLOGY. " Kan diffyrth Trindaut tri maccuy o dan Tri meib glan glein ovuy." A third instance, interesting also as being in the dual number, may be added from the Mabinogion iii. 265 — " deu vackwy -winenon ieueinc yn gware gwydbwyll," 'two auburn-haired young men playing at chess.' The word was eventually degraded to mean an attendant or a groom: compare the Greek iraibiov yielding us the French and English page. Eeturning to the phonology of the words in question, we may notice that the oi of Early Welsh could but yield wy or oe in Mod. Welsh; and as to the retention of the case vowel compare such instances as olew ' oil,' and pydm ' a pit,' from oleum and puteus. This was secured by the accent being on the ultima, which is proved to have been its former position by the fact that the word is now macwy and not magwy. Then as to the interchange of a, o, and M, in the first part of these words, one is driven to compare them with the Welsh ae or ag, formerly also oc 'and, with,' agos, cyfagos 'near, neighbouring,' Irish agus 'and,' O. Ir, ocus, occus, and comocus 'near.' It is tolerably certain that these words come from the same origin as Greek ay-^cv, &yx' 'near, nigh, close by,' Lat. angustm, Ger. eng ' narrow,' all from a lengthened form of the root agR, namely angh. Thus it appears that in our Celtic forms the mute preceded by the nasal underwent provection into c or cc — other instances of the feame kind have been briefly mentioned by me in the. Eeime Geltique, ii. 190-192, — and the nasal imparted to the vowel its obscure timbre: perhaps one should rather say that the vowel was nasalised, and came to be rendered by a, o, or u, while both Irish and Welsh ultimately restored it to a clear a. By a parity of reasoning the first part of our maccu should be referred to a root mangh, but is there such a 



 



(de
lwedd B6419) (tudalen 419)

APPENDIX. 419 root! There is; but Fick gives it ^ only as a lengthened- form of magh, ■whence he derives, among others, the following words: Sanst. mahant ' great,' Greek (Ji-rj^o; ' a means, expedient, remedy,' Lith. magt)ju ' I help,' O. Bulg. mogcH^ ' I can, am able,' Gothic magan, Eng. may. The meaning which he ascribes to it oscillates between the ideas of growing and causing to grow, of being able and making able. It is to the same origin that one has to refer our map, mob 'son,' Early Welsh' mxiqy-i, the nominative corresponding to which must once have been magyas. For Irish inscriptions show not only the common forms, maqv-i, but also maqqv-i and moqv-i, where the hesitation as to the vowel points to the same cause as in maccu and mucoi. Thus tnaqvas, genitive 7naqv-i, analyses itself into maq-vors, that is mac-va-s or mac-was: compare ebol ' a colt,' formerly epawl, a derivative from ep-, the Welsh representative of 0. Irish ech ' a horse,' and Lat. equus, O. Lat. eqvas, for ec-vo-s, as may be seen from the corresponding Sanskrit, which is ag-va-s: the Greeks had both /T-rof and 'ikxos. On the use of the affix va in the Aryan languages see Schleicher's Compendium, § 218: in Welsh, excepting where the v preceded by c, as in these two instances, has yielded qv, p, b, it is now represented by w as in erw ' an acre ' (compare Lat. arvum), malw-od ' snails,' carw ' a stag: ' compare Lat. cervus. Besides the foregoing forms which are to be referred to the longer root marigh, we have also one from the shorter magh, namely, meu in meudwy ' a hermit,' for meu-dwyw = " servus Dei," in Irish cele-dS or Guldee. Meu- stands for mag-: see page 13. The Cornish was maw ' a boy, a lad, a servant,' Breton maovrez ' a woman,' Ir. mugh ' a servant,' 0. Ir. mv^, gerdtive moga, Gothic magus ' a boy, " " 1— H! aid. 



 



(de
lwedd B6420) (tudalen 420)

420 LECTURES ON WELSH PHILOLOGY.  C— SOME WELSH NAMES OF METALS AND AKTICLES MADE OF METAL. The words I here propose briefly to discuss are the following; — alcam ' tin,' arj'an ' silver, money,' aur ' gold,' hath ' a stamp, coin, money,' efi/dd ' copper,' elli/n ' a razor,' grut- in the proper name Grutneu, haiarn ' iron,' together with ei- in Eimetiaco, mwn, mwnai ' ore,' plwm ' lead,' pres 'brass, coppers, pence,' ystaen 'tin.' It is evident. at a glance that these are not all of native origin, some being the result of borrowing from Latin, and some from English. i. L The first to strike one as borrowed from Latin is plwm, ' lead,' horn. plumhimi: there are in N. Cardiganshire lead mines which are popularly supposed to have been worked by the Eomans. The Irish appear to have retained a native term of the same origin as English lead or lode, in the Irish gloss luaidhe " plumbum." See Stokes' Irish Glosses, p. 83. 2. In the next place there can be no doubt that our aur ' gold ' is the Welsh form of aurum. For were aur simply cognate with aurum, which, in all probability stands foi ausum, it should be now not aur, but some such a form as au or u. 3. As to arian, that is, arjan ' silver, money,' formerlj arjant, Breton arc'hant or arc'hand, Cornish archans, Irish arffaf, later airged, the case is not so easy to decide I am inclined to think all these forms to have beei borrowed from Latin. 4. It is much the same with ystaen, a dissyUabh accented on the a; as now used, it is neither more noi less both in form and meaning than the English wore «tom, but Dr. Davies in his dictionary sives stannum, ai 



 



(de
lwedd B6421) (tudalen 421)

APPENDIX. 421 its only Latin equivalent, whUe he explains ystaenio as "maculare, m^culis conspergere." The Breton is stean, Cornish stean, and Mr. Stokes gives the Mod. Irish as sdan, whUe Edward Llwyd writes stdn. None of these is such as to convince one that it is not to be traced to the Latin stannum, or what is supposed to be the older form of the word, namely stagnum. 5. To these may be added our hath or math commonly used in the sense of ' a kind, species, the like of '; formerly it meant also ' money, coin, treasure,' as in the lolo MSS. p. 194, and this is the meaning which prevails in the longer forms, iathu ' to coin or stamp money,' hathodyn ' a medal,' and haihol ' coined or stamped.' These words come, no doubt, from the same source as the French hattre ' to beat,' as in battre monnaie ' to coin money.' The French verb is traced by Diez {Etym. Worterbuch der romanischen Sprachen; Bonn, 1869) through an intermediate latere to the classical batuere, ' to strike, beat, hit,' at the same time that he quotes instances of the former with ti, of which one at least dates from the 6th century: Ducange gives battare, battere, and battire, together with baptidere and baptire, as in baptire monetam = nummos cudere; but it would be useless to question or define the connection between these forms and batuere without examining the texts in which they are said to occur; but it may here be pointed out that the Welsh words are best accounted for by battare, the participle of which, battdtus, is implied in our bathod-yn ' a medal.' The old meaning of bath or math, namely that of a stamp or mark made by beating, is betrayed by the preposition stUl sometimes used after it, as in math ar ddyn ' a kind or stamp of man,' literally ' a stamp on man.' But as the connotation of the word has been forgotten, it is becoming ii._ j„-i .: — 4.„ —;4.„ ~^^ti, „ A^-,,^ -nrViinii foiij pg exactly 



 



(de
lwedd B6422) (tudalen 422)

422 LECTUBES ON WELSH PHILOLOGY. witli the English ' a kind or stamp of man.' Bath and math are further interesting as being in a state of incipient desynonymisation: thus one may say math o anifail ' a kind of animal,' but not bath o anifail, and anifail o'i hath hi would be ' an animal like her,' while anifail o'i math hi would mean, if it occurred, 'an animal of her species or genus,' with a more explicit reference to classification. Math in virtue probably of its meaning ' coin, money, treasure,' has treated Welsh mythology to several proper names — 'Compare the Greek -aXoZros ' wealth, riches,' and Pluto or Plutus, the name of the god who guarded the treasures of the earth. Thus we have a Math ah Mathonwy with his headquarters near the lake of Geirionydd, in Carnarvonshire, in a wild district by no means ill chosen for a Cambrian Pluto: unfortunately, but, perhaps, accidentally, the Mahinogion make no allusion to the guardianship of the treasures of the subterranean world d,s one of the duties incumbent on the weird king of Caerdathl. But it is remarkable that one of the leading personages in the Welsh myth which comes nearest to the well-known story of the rape of Proserpine bears the name of Matholwch, and in some other respects recalls the classic Pluto, while one or two of the incidents mentioned in the tale fall into striking agreement with a part of the account of Gudrun: see Cox's Tales of ■ Teutonic Lands (London, 1872), pp. 190-201, and the story of Branwen Verch Llyr in Lady Charlotte Guest's Mahinogion, iii. pp. 81-140. ii. 1. To return to the question of our names for tin, it is to be noticed that the word now in common use among the Welsh is none other than the English one. In the Bible, however, and other books it is called alcam, to which Pughe tried to give the more easily explained form of alcan. But there_is_no i 



 



(de
lwedd B6423) (tudalen 423)

APPENDIX. 423 fact that it must be the outcome of a comparatively recent borrowing from English: witness the use made of the word alehymy by Milton in the lines — " Toward the four winds four speedy cherubim Put to their mouths the sounding alehymy, By herald's voice explained; the hollow abyss Heard far and wide, and all the host of hell With deafening shout returned them loud acclaim. " 2. To the foregoing may be added the word mwnai, which Dr. Davies explains as moneta, nummus: the word undoubtedly comes from the English money in its older form of moneie, which is the Latin moneta introduced through the medium of French: however the Welsh word no longer means money but ore or metal, and so did the shortened form mwn even in Davies' time as the only meaning he gives it is quodvis metalhim fossile, which it still retains. It is also frequently pronounced and written mwyn: at any rate there is no satisfactory evidence that this is an instance of confounding two different words. 3. Lastly must be mentioned pres, ' brass, pence,' which seems to be a loan-word of older standing in the language, as it comes from the 0. English hrees, ires, now brass; the change of the initial consonant occurs in other words borrowed from English, not to mention Fluellen's^foocZ and prains, which are probably too late to help us here. iii. 1 . — Passing on to the remaining words, which are of Welsh origin, one may begin with efydd ' copper,' O. Welsh emid, " aes," in the Capella glosses. The Irish equivalent umae with u for an earlier a, as in ubhal, ' an apple,' Welsh afal, is, as Mr. Stokes kindly informs me, either a masculine or a neuter of the Ja- declension. Consequently it is probable that the d of emid had the J „* n,„ ^^ nf nnr Ttindem efvdd and represented an  k^. 



 



(de
lwedd B6424) (tudalen 424)

424 LECTURES ON WELSH PHILOLOGY. an earlier semi-vowel j: for other instances see the Rev. Celtique, ii. 115-118. The base would then have been emija or rather amija, the a being modulated later into e through the influence of the i following in the next syllable. Further we have found m standing for an earlier h, and, supposing this to be an instance in point, amija may be restored to the form ahija. We have also analogy for thinking abija to represent an earlier abisja, and supposing the b here, as frequently, to stand for an Aryan gv, we substitute for abisja an earlier form agvisja: assuming this last to be also a word inherited by the Teutons, one gets almost exactly the Gothic aqvizi, genitive aqyizjos, English axe. I said almost exactly, for aqvizi is feminine, while efi/dd is masculine, but the 0. Welsh plural emedou " aera " in the Ovid glosses would seem to come from a singular emed, which could hardly fail to be feminine like the Gothic equivalent. This equation can scarcely be of more interest to the glottologist than the student of early civilisation and culture. 2. The word ellt/n, ' a razor,' and its congeners somewhat reverse the relative positions which have just been assigned Celts and Teutons. Ellyn is proved to stand for eltinn or rather altinn by the Breton adten, earlier autenn, Irish altan, all from a simple alt, which occurs in Breton as aot, aod, ah " rivage de la mer, plage, bord de I'eau," Cornish als " littus," where we should say glan y mor ' seashore,' or min y mor ' the edge of the sea.' In Welsh the same word is allt, also gallt, which is sometimes given as meaning a cliff, but it does not so much mean that or the edge of a hill, — for it need not have an edge, brow, or cliff, — as the whole ascent of any rising ground, which may, therefore, be compared to the side of a blade, such, for example, as that of a razor, regarded as forming an inclined plane; and this may have been originally the idea conveyed by 



 

 



(de
lwedd B6425) (tudalen 425)

APPENDIX. 425 the Irish alt, which Mr. Stokes translates 'a cliff or height.' From alt were formed a masculine altinn whence Welsh dlyn, O. Cornish elinn [read eUmn\ " novacula," and a feminine altenn whence the Breton autenn, adten ' a razor.' As to alt itself, it probably stands for a base alda or, let us say, aid-: for other instances of the provection of sonants into surds see the Jiev. Celtique, ii. 332-335. Now we seem to detect aid-, but with r instead of I, in the Greek word a.p8ig " the point of anything, as for instance of an arrow," in the 0. Norse ertj'a " to goad, to spur on,'' and in the Mod. H. German erz ' ore, brass: ' see Fick's dictionary,^ i. 498. 3. It has already been pointed out that our aur is a borrowed word, but the name Grudneu, which occurs in an inscription of the O. Welsh period as Grutne, with its final u cut off by the marginal ornamentation on the stone, seems to put us on the track of a native word for the precious metal. The Greek word is x^uaog, which Curtius, in his Outlines of Greek Etymology, No. 202, regards as derived from a base ghartja, while gold and its Teutonic congeners, together with the 0. Bulg. zlato ' gold,' imply a • simpler base, gharta. Now the corresponding process to that whereby ghartja yielded ■x^gvaoc, and gharta the English gold and 0. Bulg. zlato would result in giving gharta or ghartja the form grut, grud, in Welsh; so that we are at liberty to equate Grutneu, Grudneu with the Greek name Xouffoms'jj, in all respects excepting that of gender: even this reserve is not to be made in the case of Grudyen (Mabinogion iii. 98), for Grut-gen, and the Greek Xjuffoysujjs. Besides these we have Grudlwyn {Mah. ii. 211); and in the Myvyrian Archaiology Grudneu (p. 389) is also called Grudnew (p. 404^, Gruduei ("p. 397^, Grudner (p. 412), of which the two last may be real names distinct from Orud- ^^■„ n-nA Yi^f -moroW TniHf.a.tpn rp.fl.dinffs of it. Before leav- 



 



(de
lwedd B6426) (tudalen 426)

426 LECTURES ON WELSH PHILOLOGY. ing these forms it is right to mention that the steps from gharta or ghartja to grud would be gliart-, gorU, grot-, grut, grud. The same is the history, for instance, of the Med. Welsh drut ' a hero,' now drud, plural drudjon, as in the name of the Denbighshire village of Gerrig y Drudion, i.e., the stones of the heroes, which it is the habit of the people who are diruid-mad to write Gerrig y Dmidion. Now drUt, dried comes from dharta, whence also the Sanskrit dhrta, formed from the verb dhar ' to hold, to bear, to support, to make firm, &c.' It would perhaps be more in keeping with Celtic analogy to set out from ghardta oi gharMja and dhar&ta: compare Welsh Haw, Ir. Idmh, from pldma for an earlier pdlama, Greek •jraXd/ji.ri, ' the palm oi the hand, the hand,' O. Eng. folm, folme, the same. 4. Before attempting the history of the word haiarn, ' iron,' it will be necessary to analyse the epithet Eimetiaco on the Llanaelhaiarn stone, which I propose to divide into Eirmetiac-o, whereof the o is the ending of the Latin nominative for -as = -us. Now metiac- probably means as a matter of pronunciation metjac, which would later have. ■ according to rule, to become metjauc, meitjauc, meidjawg. meidjog, liable also to begin with 6 instead of m, as no rule has hitherto been discovered as to the interchange of those consonants. The word, however, only survives as a feminine in the names of certain plants, of which three kinds are distinguished by the adjectives, rhudd ' red,' llwyd ' grey, glas ' blue.' One finds the following synonyms in Dr. John Davies's Welsh^Latin dictionary (London, 1632), and Hugh Davies'! Welsh Botanology (London, 181.3): a. T feidiog rudd = [ranuticulus] "flammula" (J D.), = "polygonum amphihium, amphibious persicaria" (H D.). These are not the same plants. Those meant by Dr Davies are of the tribe of the rwmnculus_Qi_raimiiculu 



 



(de
lwedd B6427) (tudalen 427)

APPENDIX. 427 flarnimda, called in English the lesser spearwort, by reason of the spear-shaped appearance of the radical leaves of the plant. Those alluded to by Hugh Davies agree better in colbur with the Welsh description, and are also said to be generally of an acuminate or speary character. ^: Y feidiog Iwyd = "y ganwraidd Iwyd, llysiau leuan, llysiau llwyd, Artemisia" (J. D.) = " artemisia vulgaris, mugwort " (H. D.). Y ganwraidd, ' the hundred-root,' is given by H. Davies simply as a synonym for yfeidiog: llysiau llwyd and llysiau leuan are the same, and are called in English St. John's wort. The commonest of these plants, artemisia vulgaris, or mugwort, looks at a distance very spiry and acuminate, and the shape of its leaves recalls the sharpness suggested by a spear or lance; and I find that some species of St. John's wort also have lance-like leaves and a spiry or acuminate growth. 7. Yfeidiog las = " mantell Fair, muntell y corr, palf y Hew, Ghimilla, hedera terrestris, pes leonis, patta leonis, stellaria" (J. D.) = " gleehoma Aederacea, gill, ground ivy" (H. D.). Jlere we meet with hopeless confusion, plants so different as the alchemilla, gleehoma, and stellaria being classed together; but it is perhaps to be accounted for by the overlapping of the characteristic suggested by the term y ganwraidd, and that intended to be conveyed by its synonym y feidiog. But none of the plants alluded to under this head, excepting the stellaria, suggests the idea of a spear or lance, which we find in the case of the other two sets. The stellaria, or stitch wort, is called tafod yr edn ' bird's tongue ' by H. Davies, its leaves being remarkably like a bird's tongue both in form and rigidity, and singularly sharp and lance-like in appearance: this is proved by a specimen which lies before me -j: il-. -J. 77 — .•„ /,„?«.< «« tr,-^ tit'^/.Ti +nrrof.'her with Other 



 



(de
lwedd B6428) (tudalen 428)

428 LECTUEES ON WELSH PHILOLOGY. specimens, as well as plates, and a careful description of all the plants here in question, I am indebted to the kindness of Mr. Drane, fellow of the Linnean Society. Thus, it seems, that we are at liberty to conclude that all the plants which were originally called y feidiog owed that name to their leaves or growth reminding one of a spear: so h&djog, meidjog, or metjac- may be treated as an adjective formed with the termination awg, og, E. Welsh ac, which, to judge from the use generally made of it, ■would give the word the meaning of ' having a spear or lance, armed with the spear: ' so we might render it into Latin by hastatus, and regard y feidiog as meaning (herha) liastata; similarly Beidauc rut, i.e., BeMjawc Evdd, the name of a son of Emyr Llydaw in Englynion y Beddau (Skene, ii. 31, 32), would be Hastatus JRufus, or Hastatus the Bed. The word for spear or lance which metjac- may be supposed to imply must have been, at least the base of it, meti, metja, or possibly matja, if the influence of the J may be supposed to have occasioned the a to become e; its origin would probably be the same as that of the Welsh verb medru ' to shoot or hit a mark ' {Mabinogion, n. 212), now used only in the secondary senses of kennen and komien, savoir and pouvoir, as that of the Gaulish mataris ' a kind of spear or pike,' and as the Lithuanian metu ' I cast or throw, O. Prussian metis (Fick) ' a cast or throw.' There is, however, it should be noticed in passing, another group of words to which it might possibly be referred, namely, that represented in Welsh by medi ' to reap,' Latin metere, Eng. muth. In the former case, to which I give the preference, the weapon meant would be one for hurling or thrusting, and in the latter one for cutting; it is, however, not necessary to decide between them as far as concerns the qualifying syllable ei in Eimetiaco, which may naturally be supposed to specify the material. And if that is so 



 



(de
lwedd B6429) (tudalen 429)

APPENDIX. 429 there can be no mistaking the word — ^it is our early equivalent for Latin ces, genitive CBris, and Alhortus Eimetiaco would in other words be Alhortus ^re-hastatus. The same d seems to occur in the name Eiudon on a stone at Golden Grove, near Llandilo, which dates no earlier than the 0. Welsh period, and the question arises how it is this ei had not by that time yielded the usual diphthong oe or wy. The reason is probably to be sought in the fact that it was originally not ei, but e plus the semi-vowel j; and this leads one back to consider the cognate forms. The Latin appears as a monosyllable in ces, but not so in Hen- or a/tem- in Ahenobarbus, ahenus aenus, aheneus, aeneus, in which dJien- or aen stands for ahes-n- as may be seen from the Umbrian ahesnes (Corssen i. 103, 652). ^s and ahes- represent an Aryan original ayas, which appears in Sanskrit as dyas ' metal, iron,' and in Gothic as aiz, proved by its z (for s) to^have been once a dissyllable accented on its penultimate: see Kuhn's Zeitschrift, xxui. 126. But a word which in Gothic was aiz must according to rule appear in 0. English as wr or dr, Mod. Eng. ore. Our parallel to these is the ei in question, and in the fact of its not passing in Welsh into oi, whence m {wy) or oe, we have a proof of its representing an early form ej tor aja or ajas. Analogous instances offer themselves in ei ' his,' ei ' her,' and heidd, now haidd, 'barley,' for forms which in Sanskrit are asya, asyds, and sasya respectively. But the Goidelo-Kymric Celts dropped the medial s so early, that for our purpose one may set out from aja, ajas and saja or sajja, modified in Welsh into eja, ejas, and seja: to haidd may be added hlaidd, ' a wolf,' which enters into Welsh names, and appears in the genitive as JBlai in Irish, where also perhaps Blddn = our Bleiddan: the base would v„ 7.7«o»v, trnyr\ -., rnnf irtyrng. whence Sa,nskrit gras ' to 



 



(de
lwedd B6430) (tudalen 430)

430 LECTURES ON WELSH I'HILOLOGT. take into one's mouth, to seize with the teeth, to devou One is also reminded of such Greek formations as riXin and a\riSeia, from nXie-Jo-i and dM^ie-ja, by the WeL derivatives in aidd or eiddj-, e.g., hen ' old,' henaic ' oldish,' heneiddjo ' to grow old,' per, peraidd ' swee pereiddjo ' to make sweet,' gwlad ' the country,' gwladau ' countrified,' llew ' a lion,' llewaidd ' like a lion,' gwei ' the look of a thing,' gweddaidd ' looking well,' that: ' seemly or decent.' 5. How aj/as has been shortened has just been show but it appears slightly dififerent in some of its derivativ( namely, in the Latin ahenus, aJieneus, for ahesnus ahesnei in the Gothic eis-arn, ' iron,' Ger. eis-en, ' ferrum,' eis-er ' ferreus,' 0. Eng. is-en, ir-en, also is-ern, and an enigmal irsem, Mod. Eng. iron, dialectically ire. These forms rn are represented in the Celtic languages by Irish ia and Welsh haiarn or haearn, ' iron.' Here it is interesti to observe that as the Bronze Age preceded the Iron A{ the idea of iron is not found conveyed by the shorl European forms ces, aiz, cer, ore: that comes in only wi the derivatives eisen, eisarn, isern, to which one may a Welsh haiarn and Irish iarn. In eisarn, eisern, isern, t simple form ayas has been contracted into eis-, is-: so the^ common language of the Celts, probably before th separation, whence (1) the Gaulish is-amo- in the pla name Isarnodor-i, which must have meant the ' In door,' while (2) the Goidelo-Kymric Celts dropping th( reduced eisarn- either into ejarn-, which had to beco: in Irish earn, iarn, in consequence of the elision o usual in that language, or else into iarn-, which had become in Welsh eiarn, haiarn or haearn. But what we to make of the h in the latter? This, if orgai should be matched in Irish by an s, whence it would, first sight, seem that the two words cannot be connect 



 



(de
lwedd B6431) (tudalen 431)

APPENDIX. 431 a view, however, which one could not entertain without the strongest reasons to back it. It has, accordingly, been suggested that haiarn, stands for aiham with an h representing the s of eisarn-. But that seems to be inadmissible, as vowel-flanked s probably disappeared in the Goidelo-Kymric period, and that not by way of h, but of z, for which the Ogam alphabet provided a symbol. My conjecture is that haiarn does stand for aiham, but with an h evolved by the stress-accent, and that, when later the accent moved to the first syllable, the h followed it, excepting in some parts of S. Wales, where the word is now ham, which was arrived at possibly by discarding the unaccented syllable of aiham . compare such cases as that of dihdreb ' a proverb,' diarhebol ' proverbial.' It is right, before dismissing the word haiarn, to say that it is also found written haearn, hayam, and hauarn, while in O. Welsh names it occurs as haern and heam as in Haemgen {Lib, Landav., p. 197) and Biuheam (lb., pp. 166, 169, 175). The 0. Breton form is hoiarn, which, through an intei^mediate houiarn (with oui = ui in the Italian word cui) has yielded the Mod. Bret, houam; similarly the Cornish became hoem. These curious forms seem to show that Breton and Cornish continued to change e, ei, di, into oi, ui, later than the Welsh, and all taken together throw light on, and receive light from, the history of a class of words of which the following may be taken as instances: — a. Glaiar, claear, clauar 'lukewarm,' Mod. Bret, hlouar and, according to Llwyd, Icloyar, with which it is usual to compare the Greek;^X*ajos, but that is hardly admissible, unless the latter be the representative of an earlier exXiagig. ^. JDaear, dayar, daiar, and poetically daer 'earth,' Mod. Bret, dollar, Com. doer: the original form may have been d(h)iar-, d(h)ipar-, or d(h)isar-, or else d(h)eiar-, &c. y. 



 



(de
lwedd B6432) (tudalen 432)

432 LECTUKES ON WELSH PHILOLOGY. Gaeaf, gayaf, gauaf, 0. Welsh (in the Lichfield Codes gaem ' winter,' Mod. Bret, goal" or goO'v, but in the dialec of Vannes gouicC^, Corn, goyf, O. Irish gaim, dugaim^iu "ad hiemandum" (Stokes' Irish Glosses, p. 166), Lai hiems, Greek %£//Hiuii. The root of all these forms is ghiair, which, treated as ghjam and reduced to gam, is the origii of our gafr, ' a goat; ' the first meaning of that word bein, probably ' one winter old: ' the same is the history o ^Ifiagog, feminine %//ia/ga ' a goat,' and of O. Norse gymb ' a one year old lamb: ' see Curtius' Greek Etymology, Nc 195. b. Graean, graian 'gravel, sand,' Mod. Bret, grcma; may possibly belong here, but the nearly related word gr points in another direction, e. Haiach, haeach, hayach hayachen, haechen " fere " (Davies), " an instant, instantlj almost, most " (Pughe), are also words the history of whic] is obscure. But not so (Q traian, traean ' a third part, Irish trian (E. Llwyd), which are undoubtedly of the sam origin as tri ' three,' or rather derived from it.  ( 433 )



 



(de
lwedd B6433) (tudalen 433)

 ADDITIONS AND CORRECTIONS.  p. 22. Y Penvyyn — I was not aware at th.^ time that Peawyn occurs as a genuine proper name, that is, without the article: several instances are to be met with in The Record of Carnarvon. P. 23. Not only qv has passed in Welsh into p, h, but tv also, as is proved by the masculine termination ep, now eh, which enters into the affix ineh, as in rhviyddineb, " ease,'' from rhwydd, " easy,'' and into the affix tep, now deb, as in purdeh, " purity," horn, pur, " pure," and undeb, " unity," from un " one." In Old Irish undeb was 6enfu, genitive 6entad or 6entath; this affix has several forms in Irish, which, together with the Welsh equivalent, postulate an earlier -ndaiva. Compare the Sanskrit affix tva in Schleicher's Compendium, § 227, and as to Welsh t, d answering Irish U, t, d, ' it may, I think, be regarded as a rule, that when ggf, dd, bb (whether produced by provection or the assimilation of a nasal) become cc (c), tt (t), pp (p), reducible in Modern Irish to ff, d, b, the" corresponding consonants in Welsh are c, t, p reducible also to g, d, b. Take, for instance, Welsh ac, ag, " and, with," agos, " near,'' Irish ag, " with," agus, " and," from angh-; Welsh map, mab, " son,'' Irish mace, mac, from mangh-; Welsh gwraig, " woman, wife," plural gwragedd, Irish /race, from the same origin probably as Latin virgo; and Welsh cret, now cred, " belief, faith," Old Irish creifem, " faith," Scotch Gaelic 2 B 



 



(de
lwedd B6434) (tudalen 434)

434 LECTURES ON WELSH PHILOLOGY. creid, "believe," from the same origin as Sanskrit gradda- dhdmi, " fidem pono: " see pages 72 and 435. P. 41. Where Welsh reduces c, t, p into g, d, h and Irish into ch, th, ph, I am inclined to think that both languages reduced them first to Cj, t^, p^, which were further modified into g, d, h in Welsh and ch, th, ph in Irish. P. 46. To the instances of analogous cases in other languages mentioned on pages 46 and 47 might be added the case of Danish, as to which Herr Sievers says, p. 126, that its initial consonants are pronounced very forcibly and strongly aspirated, while the same consonants, as medials and finals after a vowel, are allowed to become spirants of very little force or even to be altogether lost. Surd mutes, when initial, are frequently aspirated in Modem Welsh, and this must also be the explanation of the ch in chrotta and the th in Thaph and the like: see pages 118, 232. P. 48. As to nn for nd, the change is now proved to have taken place rather early in the Early Welsh period by the discovery of the Llansaint stone with its Vennisetli, which is identical with a somewhat earlier Vendesetli on one of the Llannor stones: so Vendumagli, which is in all probability later than either, can only have been the old spelling of what was then pronounced Vennumagli, a name identical in fact with the Vinnemagli of the Gwytherin stone: this last form is remarkable as the only instance known of the retention of the i of vind- which elsewhere appears as vend^ or venn-. P. 66. Another way of looking at Welsh ith for ct is suggested by an elaborate article, in the Memoires de la Society, de Linguistique de Paris, iii. pp. 106-123, bearing the title " Kemarques sur la phon^tique romane — i parasite et les consonnes mouill^es en frangais: " the same appears even more clearly in the second volume, pp. 482, 483, of Dr. Johannes Schmidt's work entitled Zur Geschichte des



 



(de
lwedd B6435) (tudalen 435)

 ADDITIONS AND CORRECTIONS. 435 indogermanischen Vocalismus (Weimar, 1875). There he mentions a German dialect in which hnecht, recht, wechseln, hexe become knaicht, raicht, waickseln, haicks: the i he ascribes to the influence of the guttural becoming palatal and imparting its i element to the vowel proceeding. This applied to the Welsh instances would lead one to suppose that noct- before yielding our noeth had to pass through nocht-, noichth, noith-; similarly (see page 209) peis, pais, from pexa, and air, aer, from agr-, would imply as intermediate forms peixa and aigr-. This view would comprehend also such cases as that of the i of doi, now doe, or more fully as still used in South Wales y ddoe " the day, i.e. jesteiday; " the Breton is deat^h. Same page, line 15, for " certainly " read " possibly: " the n alone is doubtful. P. 68. The principle attempted to be established on pages 67, 68, and 69 is fully recognised, I find, by Sievers, p. 134 of the work already alluded to. P. 72. An excellent account of graddhd, (fee, by M. Darmesteter, wUl be found in the Memoires de la Societe de Linguistique de Paris, iii. pp. 52-55, where he shows that ^raddadhdmi consists of grad, an indeclinable and obsolete word for heart, and dadhdmi, " I set or place," so that the compound means " I set my heart," both in the way of confidence or trust, and of desire or appetite: similarly the Latin credo, from which the Celtic forms cannot be derived, as some have thought, is to be analysed into cred-do, with cred- of the same origin as cor, cord-is, English heart. Modern Irish croidhe, Welsh craidd,^ both of which postulate as their earlier form crad-ja of the same formation as the Greek x^adlij. P. 91. The existence of several kinds of a in the parent-speech has recently been proved in Curtius' Studien, ix. — oa^ An>T 



 



(de
lwedd B6436) (tudalen 436)

436 LECTURES ON WELSH PHILOLOGY. P. 92, five lines from the bottom, for " Early " read " Old." P. 102. That the ch in buwch stands for an s is still very doubtful: compare hwch " a sow,'' which as a river-name is Hwch in Wales, Suck in Ireland, and Sow in England. The next Article on Duw had perhaps better be cancelled on account of the Old Welsh diu, genitive doiu or duiv, in Cormac's Glossary, and so in Chiasduiu, of which I take Guasduin and Guasdinu in the Liber Landav., p. 267, to be misreadings, later it became Gwasdwy, which is printed GwasMuy in The Record of Carnarvon, where we have also Gwassanfreit and Gwasmyhangel: compare Gwas Grist and Gwas Teilo, which occur elsewhere as men's names, also meudwy, " a hermit," lit. " God's servant," for meu-dwyw, and Giraldus' Deverdoeu, now Dyfrdwy, " the Dee," see p. 325. Further, dwyw- occurs in dwywol, an archaic form of dwyfol " divine," and in Breton doue is God. P. 109. To the instances of the reduction of diphthongs in accented syllables add the following in unaccented final ones: Gynfal, Deinjol, and Gwynodl for earlier Gynfael, JDeinjoel, and Gwynhoedl, which prove that the accent has here retreated from the last syllable to the penultimate. In the same category one may include such words as gde, " a leech," for geleu (compare Sansk. jaMka, " a blood-leech," of the same origin as jala " water "), hore, " morning," for horeu; and all such plurals as pethe and peilia, the two prevalent pronunciations of petheu or pethau, "things," in colloquial Welsh, and so in other cases. ' I^^or is one to exclude the innumerable modern instances which come under the head of what Herr Sievers has happily termed Eeciprocal Assimilation and briefly described, pp. 136, 137. This takes place, for example, when natives of South Wales reduce such words as enaid, "soul," and



 



(de
l