UC-NRLF *B saa s?o THE BRITISH ACADEMY The Celtic Inscriptions of Gaul Additions and Corrections By Sir John Rhys Fellow of the Academy [From the Proceedings of the British Academy, Vol. F] London Published for the British Academy By Hem:y Frowde, Oxford University Press Amen Comer, E.G. Price Ten Shillings and Sixpence net BY THE SAME AUTHOR From the Proceedings of the British Academy^ VoL I Studies in Early Irish History Price V" Irom the Froceedings of the British Academy^ VoL II Celtae and Galli iPrice 4/-i The Celtic Inscriptions of France and Italy Price 7/6 From the Proceedings of the British Academy^ Vol, IV Notes on the Coligny Calendar Price 10/6 net THE CELTIC INSCRIPTIONS OF GAUL ADDITIONS AND CORRECTIONS By sir JOHN^HYS fellow of the academy Read Nov, 22, 1911. This paper is miscellaneous : it begins with a recently discovered group of inscriptions, which have not yet been edited with special regard to their philological characteristics, and it ends by returning to the Coligny Calendar. Our Celtic inscriptions are so scarce and therefore so enigmatic in their formulae, that there is little hope of extracting their meaning except by attacking them repeatedly, and the greater the number of scholars who do so the greater the hope of tenable results. I flatter myself that I have made some progress, but I claim the right of further revision, whether as the result of the spade exposing new finds to the light of day or of philologists making fruitful suggestions leading to better ways of treating the material already in hand. It is needless to say that this process of revising and re-revising, one scholar improving on another's conclusions, is not agreeable to a certain class of minds; but I do not believe that an Academy so young as ours counts many Fellows who have a blank wall of finality before their eyes. When we cease revising the results obtained by ourselves and others, we may take it that we have ceased learning. I have before me an offprint of a paper in the Revtie du Midi con- taining an account of an important discovery in June, 1909, of five Celtic inscriptions near Cavaillon, the ancient Cabellio, in the Depart- ment of Vaucluse. It is from the pen of M. Mazauric, the able and courteous keeper of the Archaeological Museums of Nimes, and to his notes he has appended a useful plate bearing date October 14, 1909. He describes the circumstances of the finding of the stones, showing that there were a score or more together, but only five V Ql 280049 2 PROCEEDINGS OF THE BRITISH ACADEMY with inscriptions on them, though four others appear to have once been inscribed with letters rendered illegible by the wear and tear of their surfaces : those I have not seen. He has come to the conclusion that they had all been used to mend a road after one of the frequent floods of the local stream called the Coulon. He thinks that as monuments they stood alongside an ancient Roman road near where it issued from the old Cabellio, and that they were not carried to the spot where they have been found (about four kilometres from that town) till the fourth or fifth century of our era, a conclusion which he appears to have drawn from his finding on two of the stones ' le monogramme du Christ grave a la fa9on des graffitV — the reference is to the second stone and the third in the following list. I should have said, that the five stones are now in the space enclosed in front of the old hospital, a building which has been bought, with great public spirit, by M. Michel Jouve and his brother and sister, in order to fit it up as a public museum. There, with their assistance, I examined the stones. Since then they have been good enough to superintend the photographing of the inscriptions. See photographs 1-5, which they have generously presented to me for the following notes : — 1. M. Mazauric gives the height of this stone as about 2 metres including the piece protruding at the bottom (O'" 21) as tenon for insertion into the pedestal : the side measure at the bottom he gives as O'" 46 and that at the top 0°^ 40. He restores it in his sketch and speaks of it thus: — 'Nous avons dit que cette superbe tombe etait complete. L'ensemble de la plinthe, de la stele et de la pyramide mesure environ 2™ 54 de haut.' The reading is : — €AOYICCA MArOYP€l riAoYA The last vowel but one of the second line is rounded, € not E ; but the one at the beginning is so damaged that nothing is left except the three extremities. Even harder to read is the first letter of the third line : it is impossible to be certain whether it was f or T. On the whole it seems to me somewhat more likely to have been the former than the latter. Lastly, the second line is so close to the first that the top of the f covers a portion of the lower part of the above it. Holder in his Altceltischer Sprachschatz cites the name Elvissa, together with Elvisiios, from Mariasaal in Carinthia : see the Berlin Corpus Inscr. Latinarum, III. 4909 ; also Elvisso, genitive Elvissonis, ibid., III. 5523. MArOYP€l (with MA ligatured) appears to be the genitive of Magureos, a name which would seem to mean * related THE CELTIC INSCRIPTIONS OF GAUL 3 to, or having to do with, Maguros\ possibly 'son of Maguros\ Magureos may perhaps be treated as a variant of Magurios, cited by Holder as Magurius (fem. Maguria), from Rosendorf in Carinthia C./.Z., III. 4962, Linz III. 6010. 128, Padua V. 2787, Trent V. 5034, Rome VI. 5750. The occurrence of these names as far east as Carinthia suggests questions which I am not prepared to discuss, but the fact should be noted. What is to be made of Giava or Giaua is very uncertain, and before facing that difficulty let us see how the epitaph is to be construed. Two ways occur to me ; (1) one of them is that which I suggested in my former paper on The Celtic Inscr. of France and Italy (p. 21) in the case of the column from L'lsle-sur-Sorgue, now in the Musee Calvet at Avignon. This I should now read — AAr€NNOPiri OYePeT€[l] MAP£[00]YI * To or for Adgennorix Mareus [son] of Vereteos,' after the model of Roman inscriptions like Devilliae Catulini JiL Titiolae, ' Of Devillia Titiola, daughter of Catulinus.' So in the present instance one might translate — 'Elvissa Giava, daughter of Magureos.' (2) The other rendering would be as follows : — ' Elvissa, the Giava of Magureos.' The only objection to this syntax is that the usual position of the genitive is not before but after the noun governing it; but in the early stages of Celtic when the case endings were intact, such a con- struction had probably not become the rule. Indeed, Celtic poetry still offers plenty of exceptions : for one Irish instance see De Mag, ' God's Slave,' on page 5 below. Setting this aside, we have ' Magu- reos's Giava', suggesting some such translation as Magureos's wife, sister, daughter-in-law, cousin, or niece. The word for daughter is excluded by the fact that the word for son or daughter seems never to have been inserted in such inscriptions. If we treat Giava as a word expressing family relationship the only hope of identifying that relationship must rest on the probable etymology of the word. The following occurs to me : — There is a common Welsh word giau, older spelling gieu, ' sinews, muscles,' singular gewyn (for giew-yn\ Corm^h. geiow, 'sinews,' singular geien : I have not detected any form of the word in Irish. If we look at ' sinew ' as meaning a kind of cord or string, we may compare Latin nurus (for ^sntisus) and German Schnur in the sense of ' string, or tie ', and Schnur ' a daughter-in-law ' : for the chief cognates see Walde's Lateinisches etymologisches Wdrterbuch, s. v. nurus, Giava may have meant any one of the relationships which I have suggested ; but it is natural to regard it as synonymous with such Indo-European words as Schnur, nurus, and Greek vvos. The render- 4 PROCEEDINGS OF THE BRITISH ACADEMY ing of the epitaph in that case would be ' Elvissa, daughter-in-law of Magureos '. In Stokes's Urkeltischer Sprachschatz, translated into German and edited (as the 2nd part of Fick's Vergleichendes Worterbuch der indo- germanischen Sprachen) by Bezzenberger, p. 117, a stem gjd is given, meaning ' a sinew, tendon, chord, string ' : it is illustrated by a Welsh gi, ' nervus^ plural gmu^ and further by the Sanskrit jya^ fem. ' a bowstring', Greek ^to?, 'a bow,' Lithuanian gija^ 'a thread or string.' I have never met with Welsh gi except in Dr. Pughe's dictionary, nor do I believe that^w was originally a plural; but the use of the isolating derivative ^eVzK'?/^ tended to give it the force of a plural as in other cases of the kind. Further, the accentuation of the cognate forms convinces me that the starting-point was not gja but rather guja or guia. That is to say our giava was gidua, but for an earlier guidua. This was lightened of the u just as giewyn in modem Welsh has dropped the i in becoming gewyn, which is the actual form. Otherwise gjd over against ;3tos would remain without a parallel : see Pedersen's Verglei- chende Grammatik, I. 108. In other words both y8t-os and gi-dva come from gui- meaning sinew, tendon, string, or bond of connexion, so that the derivative giava would mean a woman who stood in the position of being a bond of relationship between two families, such as a daughter-in-law. We may possibly have the word for consort, yoke-fellow, wife, formed with the aid of the preposition du (in English ' to ') imbedded in the names Dugius and Dugia, Dugiavus and Dugiava, much as if we met with Conjvgalis as a cognomen in Latin. They occur chiefly in Latin inscriptions in the neighbourhood of Brescia and Lake Garda in North Italy : see Holder, s.vv. Dugim,^ Dugiava. A good parallel is supplied by a Celtic compound Comiogia, which occurs in an inscription beginning with Enica Comiogia Nevi f(ilia) found in the neighbourhood of Saluzzo in Piedmont (C /. L., V. 7641). Comiogia might also be rendered by Corijugalis, as Com-jpg-^ exactly * He gives sometimes a spelling with c instead of g. This is due partly perhaps to the confusing of names of different origins hy his authorities and partly to misreading on their part, which is hard to avoid in the case of G and C in ancient inscriptions in the Latin alphabet. Of the feminine the Revue Celtique, III. 166, has the spelling Dugiavva, from Brescia, on the authority of Muratori, MCCLXXIII. 6. * The ancient combination og was liable in Brythonic to be modified into ou or ow, which in North Wales became successively ew, eu, au : thus iau (pronounced iau) is the book Welsh for ' yoke, jugum '. But in most of South Wales ou or ow was only changed into om, which to my hearing is sounded oi or something very near oi. Thus what is called in Mid- Wales and North Wales jaw is in the South iou : it was THE CELTIC INSCRIFPIONS OF GAUL 5 covers the con-iuff- of con-iug-s = coiijux, genitive con-jug-is. Putting aside the prefix com = Latin con, the rest of the word is formed like Greek ffyta which is applied to Hera, Juno Jugalis, as patroness of marriage, (rv-fvyto?, (rv-fi;yta, * joined, united, joining.' 2. This is a fragment of a column measuring 0°^ 38 each side, with the present height of O'" 60. It reads : — BAAAYAO YIMAKKAPIO Yl The lower part of the last Y I has disappeared with the lost piece of the stone. But none of the letters are subject to doubt. The inscrip- tion makes the two names Ba\avhovL Max/captovt, and they would seem at first sight to be both of them datives of the 7i declension with ovl as in TpaaeXovL and Elvovl. But after reading Dr. O. A. Danielsson, Zu den Venetischen und Lepontischen Inschrifteri (Leipzig, O. Harras- sowitz), I am forced to regard the datives in -ovi as belonging to the o declension. The analogy of the other Cavaillon epitaphs would have led one, it is true, to expect a nominative here, but a dative -ovi sounds more natural than a nominative Balaudui, and probably represents an older stage of the more usual dative in w of the o declen- sion. So I transcribe the epitaph into Latin spelling as Balaudui Maccarivi, meaning 'To Balaudos (son) of Maccarivos \ With Balaudui compare Latumarui, p. 10 below. By Maccarivi one is reminded in the first place of Macarus and Macarius which Holder describes as partly Greek and partly Celtic ; but if we may trust the spelling we seem to have a vocable more written so also in Old Cornish : see the Phil. Soc. Transactions, 1860-1, p. 241. When, however, ou was immediately followed by i (or y) it was not changed into ou on account of the difficulty which we find in pronouncing u and i together. So though we have cyfieu-ad, ' a yoking or joining together ; conjunction ; con- jugation/ we have an archaic adjective with -in, which takes the form of cyfiewAn, 'equal, like, similar; even' (Silvan Evans's Geiriadur, s.v.), as it were under the same yoke as a means of colligating or joining together. This is often dis- regarded in book Welsh, as for instance in llysieu-yn, ' a herb,' from llysieu, llysiew-, ' herbs ' ; but the genuine form, still in use in colloquial Welsh, is llysiew-yn, or llysew-yn : compare giew-yn already mentioned. Among other instances in point may be mentioned Welsh meu-dwy, ' servus Dei/ with meu corresponding to Irish mug, mog, genitive moga, 'a slave,' whence Irish D6 mog (Stokes's Oengus, pp. 4, 10) for Mug DS, ' servus Dei ' ; and Lieu to Irish Lug, genitive Logo. The commoner form of Lieu, however, is Llew, while the old compound Lugu-ber makes Welsh Lleufer, 'a luminary, a light,' literally, 'a light-bringer ' : see Holder, s.v. Locuber. Take also Ogmios, probably pronounced in Gaulish Ogmiio-s, and making in Welsh Euvid, the name of one of the Sons of Don : see the Cymmrodor, XXI. 4-7, 62. 6 PROCEEDINGS OF THE BRITISH ACADEMY closely akin in the widely spread name of the potter Maccarus, in Maccius, Macco, Macconus, Maccus, Perhaps we may compare the Welsh inach^ ' a surety, a bail, a bond or pledge.' In that case the second name may be treated as a compound Mctcca-rivos, genitive Macca-rivi, and interpreted as the 'surety or pledge of Rivos' or perhaps as 'one pledged or dedicated to Rivos, the bondsman of Rivos ', the divinity meant being the god who figures in the month of August in the Coligny Calendar. The spelling with A;Ar, cc, leads me to think that the names of this group are not derived from the Greek word iidKap, ' beatus.' This is one of the stones on which M. Mazauric found the monogram : his note reads thus : — ' La face laterale gauche offre des traces de graffiti graves apres coup. L'un d'eux reproduit incontestablement le monogramme du Christ que Ton retrouve si frequemment sur les sarcophages des iv® et v® siecles, Faut-il voir la une sorte de christianisation de la pierre ? ' 3, This stone is 1"^30 in height and each of the sides 0°* 38 at the bottom and 0°^ 32 at the top, near which the inscription is placed, as follows : — KABIPOCOYI NAIAKOC That is Ka/3Lpos OvLvhuaKos ; the I at the end of the first line is a little bit faint and the P is damaged ; so is the A in the second line. The I before the A joins it. The syntax of this epitaph is clear enough. Vindiaco-s is an adjective qualifying Cahiros and meaning ' related to Vindios, connected with Vindios ', but in what way exactly we have no means of deciding. It might possibly mean in the sense of being son of Vindios or in some way associated with Vindios. Compare Anvalon- nacos, derived from the god's name Anvahs : see my Celtic Inscrip- tions, V, also other instances in xiii, xv, xvii. Vindios is derived from the adjective vindo-s, vindd, ' white,' Irish^nd, Welsh gwi/n, fem.gwen, ' white.' The name Ka^ipos is found elsewhere, to wit, on a stone dis- covered at Cologne, which reads Gato Cabiri f{ilio) civi Viromanduo (C. /. L., XIII. 8342 : see also 8341). The Veromandui gave their name to Vermandois, the district, roughly speaking, around the town of St. Quentin. Holder cites also a CABIRI AC VS derived from Cahiros and yielding the modern place-name Chabrac, near Tulle, in the Dep. of Correze, and a Merovingian coin of the seventh century reading CABIRIACO VIC from Belfort. He also cites Cabrtis from York, from Castel near Mainz, and from other places : it is probably a shortened form of Ka^tpoy, Cabirus, from which he derives a name Cabrianecum THE CELTIC INSCRIPTIONS OF GAUL 7 which derives immediately from Cahrian- : this reminds me of Chabran, the name of a street at Cavaillon. Thus we have traces of the exis- tence in the past of the name Kafiipos, Cabirus, in places so far apart in France as Cavaillon and St. Quentin, Tulle and Belfort. The meaning and derivation of the name offer considerable diffi- culties, which, as far as I can recall, have never been discussed. Celtic philology throws no certain light upon Ka/3tpoy, and one cannot help thinking that it is the singular of the Greek Kd^eipoi or Ka^Stpoi. The home of the cult of the KajStpoi was Samothrace and Lemnos. The question then is how the name spread in Gaul ; in this connexion it should be noted, that from an early date in the cult of the Cabiri in Samothrace and Lemnos they were popularly associated with the Dioscuri. In time that association led to their being identified with one another. On these points we need only consult Daremberg and Saglio's Diet, des antiquites grecgites et roinaines under the words 'Cabires' and 'Dioscures'. In the identifying or confounding of the Cabiri with the Dioscuri, the characteristics of the latter were probably the more conspicuous in the cult as practised in Gaul. This I take to be indicated by one of the Notre-Dame altars, now to be seen at the Musee de THotel de Cluny in Paris. I refer to that which has on its four sides representations of the horned god Cernunnos, Pollux, Castor, and Smertullo. Compare Dioscorus as a man's name at Nimes, for which see C, /. Z,., XII. 3508, also 4550. The cult of the Cabiri and Dioscuri probably reached Gaul through the same Roman channels as the cult of Cybele, Mithras, and other eastern divinities. Statues of the Dioscuri are by no means rare in France, and one of the last discovered was on Mont-Auxois : it has been described in the periodical Pro Alesia (Armand Colin, Paris), 1906, p. 40, plate ix. Lastly, there are near the foot of this stone certain symbols on the meaning of which M. Mazauric declines to pronounce an opinion. All that I remember about them is that I detected there a roughly sketched fish's head. He adds — * Je dirai simplement que ces signes sont certainement contemporains de I'inscription qui est au-dessus.' 4. This stone is 1°^ 22 high : the sides are uniform, measuring each Qm 42 rpj^g inscription consists of four lines at the top, which has been damaged, the upper portion of most of the letters in the first line having been carried away, together with the beginning of all the lines. The following are the details of the reading : — The first letter is a straggling M the first limb of which I could not find, but the photograph (No. 4) shows exactly where it was. Its last limb is joined by an I or rather the end of it is cut through vertically by that letter. We then read T I €C, and if you scrutinize the photo- 8 PROCEEDINGS OF THE BRITISH ACADEMY graph you will see by the side of the imperfect C a faint, well-formed I, which I did not succeed in detecting when looking at the stone; for the next thing I read was a shallow stop. Then comes what I misread as an incomplete A which the photograph shows to be another straggling M with its last limb produced upwards to make an I . We then have the greater part of a T with room for nothing more, the reading of the whole line being MITI6CI • MIT. Most of the second line is occupied by the letters MA TOY, which are quite clear in spite of bits of the surface having disappeared in the MA and between the Y and the A in the line below. The length of the top of the r will be noticed, the non-circular shape of the in contrast to the in the last line, and the sprawling appearance of the Y which is repeated in this same line. Now before the M I found a stop, and before that stop the photograph shows the top of a C together with a trace of its lower end. Behind the C there can have been no big letter, nothing larger than an I , and the photograph shows enough of the groove of that letter to indicate exactly where it was in the damaged part at the edge. The whole line thus reads IC • MA TOY. The third line begins with a T cut further from the edge than the initials of the other lines. It is followed by an I, to which a breakage gives something of the appearance of a triangle upside down. Then comes what I took to be another stop followed by an of a more unusual form than the one in the second line ; but it seems partly due to a bit of the surface being damaged low down towards the reader's left hand. As it is, it looks as if the inscriber had begun cutting the upper part of a big A. After the come NNA, of which the second N is badly formed, and the A has an accidental scratch reaching from its shoulder to the top of the Y above it. The fourth and last line begins with an imperfect K, the back of which is gone, leaving the arms sticking out : the lower arm ends in a fine line formed by the slipping of the work- man's tool. The letter is needlessly large and the next one is a fairly well shaped placed too far away from the K as if escaping from the latter's outstretched arms to shelter itself near the Y I which finish the epitaph. A breakage which affects the top of the Y spreads back to the circumference of the 0. Putting all this together the epitaph reads as follows : — MITieCI . MIT IC . MAroi' Tl . ON N A KOp You will have noticed that we have here two kinds of T : the first is the capital with its perpendicular groove worn shallow towards the THE CELTIC INSCRIPTIONS OF GAUL 9 bottom. The other tends to a minuscule form with the top stroke greatly produced backwards. The stops I should not have detected in the photograph; on the other hand until M. Jouve sent me the photograph I had not discovered the final I of Mtrteo-t or read correctly the following M I ; these failures prevented the possibility of my construing the epitaph. As it now stands the reading may be regarded as certain, although at the first glance it looked desperate. With the words separated it reads Mltl€(tl . Mltls • Mayovri . Oz^i/a- Kovi, which I propose to render thus : Mitiesis (and) Mitis (children) of Magutios and Onna. This suggests several questions, such as whether the two first names refer to two persons or one. Now Mitiesi, presumably for a fuller form Mitiesis, is clearly a derivative from the shorter Mitis. This makes it somewhat more natural to treat them as two names, those, let us say, of a sister and a brother (or sister), rather than to regard Mitis as an epithet or surname : it looks more like a proper name than an ordinary adjective. To what origin it must be assigned, it is hard to say. Holder cites Mitiacum (? Miciacum) and Mitiganna (? Meliganna), besides a larger variety beginning with met-, including Metela, Metilitis, Metillius which is, perhaps, to be explained as Meddill-ius. The Neo-Celtic words which suggest themselves are such as Irish air-rued, ' measure,' from an early metio-r of the same meaning and origin as Latin metior, ' I measure,' and Welsh medi, ' the act of reaping,' from an early meto of the same origin as Latin meto, 'I reap or mow': see Stokes-Bezz., pp. 203, 206. There are some personal names which may per- haps limit the etymological possibilities a little, such as Medieval Welsh My dan where the d may = t subjected to the usual lenition and so in Myd-naw, Myd-no (' lolo MSS.', pp. 102, 109, 139) made in Modern Welsh into Bydno, the name of a small tributary of the Wye, near Pant Mawr above Llangurig, in Montgomeryshire. It is possible that Mydno equates with the Irish name Mid-gnu, Mid-nu (Book of Leinster, f 369^, 372^). Stokes gives in his Oengus a 'mac Midna mic Meite ' (p. 100). The difficulty as to the Irish names with mid- is that some of them doubtless involve the word for the ancient drink mid, *mead.' Perhaps this does not apply to the name Mithigen (Book of Leinster, f 317«^ 340*), which Prof. Kuno Meyer's 'Raw- linson B. 502' gives as Mithigen and Mithigedn (129^ 15P), but the name is obscure. The termination in Mitiesis reminds one of the form of such river names as Atesis and Tamesis (also Tamesa), but we cannot be sure that the pronunciation was not Mitiessis, which might be associated with names collected by Holder to illustrate the affix 10 PROCEEDINGS OF THE BRITISH ACADEMY -ess- as in Antessius, Revessio, and especially Lucressisy by the side of which he has Lucressa. We now come to Maguti : the nominative to which it is to be referred was probably Magutios, with the same magu as in Magurei, already mentioned: Holder gives such instances (s. v. -nt-io-) as Ammutiits, CillutiitSf and Taranutius, and he has parallel formations under -itio-, -atio-. Similar derivatives occur also in Irish, as for instance in the Ogmic genitive Lugutti, written also Lugudi, a name found in Adamnan's Life of St. Columha (Reeves, 74, 236) as Lugudius : the base is the Lugu- of the god Lugus's name, in Irish Lug, genitive Logo (later Logo). Another is Curcitti, derived from an ancient form of the name Core, genitive Cuirc, In later Irish the derivative was spelt Corcthe : see the Book of Leinster, fol. 325^. The English rendering which I have just given has shown you that I treat -kovl, -qtii, as a Celtic equivalent of Latin -que, ' and,' also that I regard Onna here as a genitive. More precisely I should write it Onna, for an older Onnds, to be compared with such old Latin genitives as vidSfJhrtu/ndSfJhmilids, and Greek x^P«?) O^as. This genitive survives in Irish in the form mnd ( = mnd) ' mulieris ', and inna, na, * of the ' (fem.). This d{s) appears to have been the early termination : see Brugmann^^, II. ii. i. 152, 284, 361, who goes so far as to suggest that the usual e of Irish genitives feminine like tuaithe, ' of a tribe,' has only been adopted from another declension (ibid. 154). The con- junction -Kovi or -qui seems to me to have belonged elsewhere to the language of the inscriber of the rejected version of the Todi bilingual. He was the better cutter of letters on stone, but his Latin was faulty, and, among other blunders, he perpetrated statuitqui for statuitque, I am inclined to explain his doing so on the supposition that the con- junction was -qui, not -que, in his own language : see my Celtic Inscrip- tions, p. 71. In an inscription in a Gaulish dialect on a vessel found at Ornavasso, on the Toce, near Lago Maggiore in North Italy, our -qui = Latin -que is represented by -pe. The vessel reads Latu- marui \ Sapsutaipe \ uinom \ na^om, meaning ' For Latumaros and Sapsuta^ . . . wine'. See Danielsson, loc. cit., pp. 17-19. His reasoning convinces me that the inscriptions known as Lepontic are Celtic, and I hope to find an opportunity to see them. I have said nothing as to the etymology and meaning of Onna for the sufficient reason that they elude me. Possibly, however, we have this vocable in the Ov€- of Ov€p€(TT- on a stone from St. Saturnin- * The dative Sapsutai recalls Eo-Kcy^ni, kiovviai, and B\avhooviKovviai {Celtic Inscr., pp. 20, 76). THE CELTIC INSCRIPTIONS OF GAUL 11 d'Apt. The other element here may be derived from the name which gives the genitive RE ST I on an Aries stone to be seen in the Musee Calvet at Avignon : see my Celtw Inscriptions, No. viii ; also No. xvi% where M. Romyeu's MS. is mentioned as recording an inscrip- tion reading ON GOYOnO AIOYI • BPATOY (not BRATOY). The stone came from the ruins of Glanum and should be at St. Remy, but it cannot be found. The Cavaillon inscription to be mentioned next, with and C ornamented with a central point, suggests the legend ONOOYOnO AIOYI • BPATOY[A€ KANT€N], 'Onovopos (gave) firstfruits to the goddess according to her command.' Here the TT may mean p or else fl or Tl.^ That, however, must remain uncertain at present. 5. This stone measures 1°* 44 high, and the sides are uniform, measuring each 0°^ 44. The reading is as follows : — MICCO YKOC GIAOY KNOG That is Mlo-o-ovkos ^lXovkvos, which should mean Missucos, son of Silus. Holder quotes Silus^ fem. Sila, from Latin authors and inscriptions. The first syllable of Missucos is probably to be equated with that of Missilltis, cited by Holder, together with other spellings Medsillibs, Meddillus^ and a feminine Messilla, Here Missucos and ^ With ri the name would be Ovo-ovoyio with a second element which could be identified with the iiogi of the compound Vogi-toutus in an inscription near Greifenberg in Carinthiaj reading according to the Corpus, III. 4724 : — ' Atestati Bricconis f(ilio) patri Devvae Atiougon. f(iHae) matri C. Antest(ius) C. f(iHus) Lutumarus M. Antest(ius) C. f(ilius) Vogitoutus Antestia C. f(iHa) Banona ' ; and another (No. 4908) reading 'Daphino G(ai) Juli Vogitouti l(iberto).' The former is remarkable among other things for containing as a woman's name Dewae, dative of Devva, that is Dewa or Deva, the feminine of devos, the Cehic word for god, which occurs in the Coligny Calendar (p. 92) as a dative DEVO. In other words Atiougo's daughter was called Dewa or Goddess ; but there was another Celtic word for goddess which we have in the vocative as devvi in one of the Rom defixiones, and we have it here in M. Romyeu's copy as AIOYI, that is diui, and the inscription falls readily into the Bratude group as Ovoovoyio Aiovi . j3paTou[5e KavTfv] I see my Celtic Inscriptions, pp. 33, 84. Vogi-tout- and Opo-ovoyi- remind one somewhat of such Greek names as 'Exe-noXis, 'Ex(-^r}fjLos, with exf for f^x^) ^^^ 'Aarv-oxos, 'Hvi-oxos with oxos for foxos. ^ The dd appears to represent a lisping pronunciation common in Gaul of ss where apparently that was derived from ns, and there seems to have been an intermediate spelling with ds, for besides Holder's Medsillus and Messila we have such instances as Ressatus or Bedsatus, Ressi-marus or Redso-marus cited by him from Carinthia (C.I.L., III. 4727). As a variant of the spelling with ds we may regard da; in the nominative midx, 'month/ in the Coligny 12 PROCEEDINGS OF THE BRITISH ACADEMY Missillus or MeddilliLS are probably reduced forms of some such com- pounds as Messi-gnatos or Meddi-gnatos : Holder gives the deriva- tive Meddignatius from Brambach''s Corpits Inscr, Rhenanarum^ 1336. Possibly the element missu may be of the same origin as Latin mens- in mensuSj^ mensor, menstira, mensa : see Walde's Dictionary. ^** We must not leave Cavaillon without mentioning the inscrip- tion discovered a few years ago by M. Michel Jouve on the right bank of the river Durance. I reached Cavaillon from Avignon in the after- noon ; M. Jouve took me at once over various archaeological sites near the town, and we climbed to the hill of Saint-Jacques, which lies between the town and the Durance, and presents a highly precipi- tous side to the river. On the top are the remains of a very ancient Calendar ; it is of the same origin as Latin mensis, and is represented in the Neo-Celtic languages by Irish mi, genitive mis, Welsh mis, Cornish mis, Breton miz. Such forms as Veliocasses and VeliocaBis seem to imply a word cansa or canso- which in the Teutonic languages yielded the Gothic word hansa, ' a band or cohort,' German hanse, A. -Saxon hos, 'a society or guild.' In the Neo- Celtic languages I trace it in the Cornish casgoord glossing satellites, ' atten- dants, escort, train, retinue,' in later Cornish cosgor, cosgar, ^a retinue.' Casgoord occurs in the Bodleian MS. 572, fol. 43^ : see Zeuss, Gram. Celtica^^, p. 1062, and Stokes, Transactions of the Philological Society, 1860-1, p. 243. The 00 in casgoord probably means that the word was accented casgord : the Breton form is given in the Catholicon as coscor, ' famille, mesgnie,' Latin ^ familia '. Dr. Davies in his dictionary gives the Welsh as cosgordd and gosgordd, a corrupt form, which is the only one used in Modern Welsh. But cosgordd is not actually the oldest spelling, for the MS. of the Nennian Genealogies has in pedigree xii Eleuther, cas cord, maur, that is, E. cascord maur, ' E. of the great retinue ' : see the Cymmrodor, IX. 175. Casgoord, cascord analyses itself into cas-gord from an early compound cansa-corio- with the corij)- which we have in the Irish caire or coire, 'a band or troop,' in Gothic harjis, ^ a. host, legion,' A. -Saxon here, German heer, ' an army.' So Welsh cas-cord or cos-gord would mean the host or band of the village or community, just as trefgord = treha-corio-, ' the people of the tref' or homestead ' : see Owen's Ancient Laws and Institutes of Wales, I. 258, where it is spelt trefcort. We have cord not only in cordlan, ' the village pen or fold,' now corlan, mostly ' a sheep pen ', but possibly also in the nickname Idawc Cord Prydein, which would in that case mean ^ Idawc of the retinue of Picts ' : see ^ Rhonabwy's Dream' in the Oxford Mahinogion, p. 147. ^ This mens- is represented in the Irish mess, ^judicium' (Zeuss**, 787), Mod. Irish meas, ' estimation, award, appraisement ' (O'Donovan) ; there is also a com- pound coimhmheas meaning ^a comparison, contest'. Both occur in a passage in Keating's History of Ireland (Dinneen, vol. II, p. 198), where one reads that, after Conall Cernach had slain Meisceadhra in single combat, his rivals Laogaire and Ciichulainn ceased to contest the champion's morsel with him. The words are : — do leig Laoghaire is Cii Chulainn da gcoimhmheas r4 Conall, ar n-a mheas nach deama ceachtar diobh fMn a chomhmor soin do ghniomh goile na gaisgid riamh, ^ Laoghaire and Cuchulainn ceased measuring themselves with Conall, as they reckoned that neither of them had ever performed so great a deed of bravery or valour.' THE CELTIC INSCRIPTIONS OF GAUL 13 stronghold, still showing a piece of walling made up of big stones overlooking a deep hollow leading down to the backwaters of the Durance. The land on which the ancient fortress stood has been purchased by M. Jouve in order to prevent the destruction of the old wall there. From this high ground we could see over against us, in the distance across the river, the site of the ancient town of Glanum, while higher up the river he pointed out to me the promontory of Orgon, and the neighbourhood where the Vebrumaros inscription was found. With some trouble we descended into the hollow which I have mentioned, at the mouth of which M. Jouve thought early traders exposed their wares for sale to the inhabitants of the hill fort. Finally, we came to a halt on a sloping rock near a deep pool which had probably been at one time part of the channel of the Durance : the river now flows on the other side of its bed some distance away. On the sloping part of the rock M. Jouve called my attention to a number of footholds cut in it. I tried some of them and found them quite safe. Evidently it had been a landing-place when the volume of the river flowed on the Saint-Jacques side. A little below the places cut for the foot, where the rock became more nearly horizontal, was the spot where M. Jouve found the inscription. This was when a great overflow of the Durance had swept away all the soil which had accumulated over the footholds and the inscription. The inscription has been read by M. Maruejol and his friends as follows (Celtic Inscriptions, p. 23) : — OYEAPOY (DHKIKOC I did not feel sure as to the first Y : at any rate, there is room for it. My suggestion (after seeing the cast at the Musee Calvet) that one should read A instead of A is not corroborated in the present condition of the inscription. Lastly, at the end of the second line I failed to trace the letters OC, and to some extent I had to take the on trust. For unfortunately we arrived on the spot rather too late in the day : the rays of the sun had already left the part of the rock for which we wanted a good light. ^ The first difliculty we have to face is the uncertainty whether we are to read A or A in the first line. Now if we take the latter ^ On a visit paid by M. Jouve and M. Maruejol to the inscription soon after its discovery there was no lack of strong sunlight, as is amply proved by a photo- graph taken on the occasion. I have to thank the former gentleman for a copy (Photo. 6), which shows not the inscription but the area cleared of earth by the flood of the Durance ; there was too strong a glare for the inscription to be reproduced, but it shows M. Maruejol with his finger on the first letter. 14 PROCEEDINGS OF THE BRITISH ACADEMY there is little hope of discovering a name that would fit, whereas if we take the former we seem to arrive at a nominative Ovcbpov for an earlier Ovehpovs or IJedru-s of the same origin perhaps as Ptolemy's Ovehpa or Uedra, a name which survives in English as Wear, borne by a river in the North of England ; not to mention that Wearmouth is supposed to be the Caer Weir of the Book of Taliessin : see Skene's Four Ancient Books of Wales, II. 200, 201, and Rhys's Celtic Britain^^\ p. 119. In Welsh the d before r might become i as in cadeir from cat[h^dra, ' a chair ' {Celtic Folklore, p. 282). On these lines the Welsh equivalent of (Jedrus would be Gweir (in lenition Weir), modern Gwair ( Wair), as in Llwyn Gwair, ' Gwair's Grove,' near Nevern in Cardiganshire, and in Ynys Weir, understood to be the Welsh name of Lundy Island in the Bristol Channel. Gwair was a mythic hero figuring in Welsh poetry : see my Celtic Folklore, p. 678, Celtic Heathendom, p. 248, the Oxford Mabinogion, p. 306. Holder under -wc?ro-7i would connect with Ptolemy's Vir-vedrum in North Britain the O. Slavonic vedru, * clear, bright, cheerful.' The other vocable ^r]KiKos begins with the sound of <^ or f, which was always rare in early Celtic, as it is derived from the combination sp'h or sp. Both in Goidelic and Brythonic it seems to have been reduced into f, which Brythonic has retained (written ff), while Goidelic has further changed initial f into s, as for instance in the Latin loanwords srian, ' a bridle,' Welsh ffrwyn, the former probably from frenum and the latter from the plural frena. So with suist, ' a flail,' Welsh ffust, both from Latin fustk} Now (f)r]KLK09 appears to be an adjective in -ko-s serving as an epithet to the name in the first line. It should mean ' resembling, having to do with, or related in some way to, what was indicated by the previous part of the word, /)ov6ios, which has been ^ For some more instances see the Cymmrodor, vol. XXI, pp. 54, 65, where Irish 8ust should be corrected into suist. The consonantal changes here in ques- tion are treated otherwise in Thurneysen's Handbuch, pp. 80, 137, 521. THE CELTIC INSCRIPTIONS OF GAUL 15 mostly supposed to be the Somme ; but C. Miiller in his edition of Ptolemy (I. 219) argues for a smaller river called the Bresle, which reaches the sea not far from the little town of Eu in the Department of the Seine-Inferieure. He prints the name with a 8, but Holder, s. V. Frudis, cites Gliick as correcting it into 4>pourios, which was probably the genuine form. It is given in the genitive case, so that in the nominative it would have been 4>povTis or Friitis, practically the same word as old Welsh ^rw^, ' a stream,' whence in the Book of Llan Ddv, Csjnfrut, 'crooked stream,' Guen/rw^, 'White Stream,' and similar compound names still common in Wales, the modern spelling being ffrwd^ CsLmffrwd, Gwenffrwd^ &c. : so in Breton Jrut, frot, in modern spelling yroi^c?. The kindred Irish word is sruth, genitive srotho, srotha, a masculine of the u declension, while the Welsh ffrwd is a feminine, which probably belonged to the i declension like ^povTis. Still it is, etymologically speaking, impossible to sever srtith a^nd ffrwd} At first I regarded the rock inscription as not Celtic, but I hope that I have now made it probable that the previous guess was wrong. Speaking more precisely I should say that the language of it may be taken to have been the same as that of the five Cavaillon tombstones ; and this, with its test particle -kovi, I regard as the same as that of the votive inscriptions of the district distinguished by the ^paTovbc formula : see my Celtic Inscriptions, pp. 78-81. The language of the whole group, as thus expanded, is probably to be regarded as the kind of Celtic in which the Coligny Calendar was drawn up. For want of 1 I mention this as it is usual to refer these words to a root sreu, whence Greek pfcoj ' I flow,' pvTos, ' flowing/ Sanskrit sruta, ' flowing,' Lithuanian srove, ' a river current/ Old Bulgarian struja, ' flumen.' Similarly the kindred words in the Teutonic languages mostly begin with sir as in English stream, German Strom. It is seen that there was some freedom of choice in this instance as between ^r and str\ but neither will directly fit ffrwd. For to make sreu, srut-j the immediate antecedent to the Celtic words as is usually done, would require one to suppose the s to have become/ in Welsh, a change which nobody, as far as I know, has ever detected in that language. What ffrwd postulates is an initial combination spr or spr : that is frutis, ffrwd and sruth, starting from sp'rut-j which was reduced to frut-. This last was retained in Brythonic, while in Goidelic it underwent the later change which made the f into s. We seem to have a parallel in the German sprudel, ^a bubbling well, a hot spring,' sprudeln, 'to bubble, to gush, to flow.' Whether we are dealing here with a single root sreu, liable to be euphonized into streu and spreu, or with several parallel roots, is of no special importance for our argument, which warrants our tracing the river name Ffraw in Anglesey, and Asser's Frauu, the Welsh name of the Dorset river Froom, to the same origin as the English word stream, German strom, Lettish straume. See Fick^^, IIL 502, and Stevenson's Assers Life of K, Alfred, pp. 37, 248, 249. 16 PROCEEDINGS OF THE BRITISH ACADEMY a name I have sometimes called it Celtican, but I have no objection to its being called Ligurian, provided the term be used to mean the earliest Celtic speech in use in ancient Gaul. ^"^^ Before leaving the Department of Vaucluse I wish to discuss No. xi in my Celtic Inscriptions. It comes from the neighbourhood of Apt, and is in the Calvet Museum. It has always been a great puzzle to me ; but I pored over it in August, 1910, with the result that I have, I think, made some progress, involving my giving up some of my previous guesses. Thus my reading of the last word as VALE will not stand : the Editor of the Corpus was perfectly right in reading the broken letter as A and not A. In this inscription the sigma is written C, a form possibly derived from Z rather than from the minuscule Greek C. Moreover, the first letter I was now able to make out in the first line was a rather small C, and after it I seemed to read OYI. Then comes what I ventured to consider a form of Latin R, but I think now that it was meant for a Greek K, and in this I am confirmed by its similarity to the kappa in BAANAOOYIKOYNIAI on the Gargas palimpsest stone in the court outside. The whole name will then read KAIPNITOYC, Clirnitus. In that case we have the K again in the second line, namely, following NA, so that a part at least of that line would read NAKNOC, but I am puzzled what to read before the N A : the symmetry of the inscription seems to require some lettering there. I thought once that I could faintly trace the leading features of AANAKNOC or perhaps of MANAKNOC with AN ligatured. What I have now made out would approximately read thus : — COYI . KAIPNITOYC [AAJNAKNOC lAAE Treated thus, the inscription ceases to be an exceptional mixture of Greek and Latin letters, as no distinctive Latin letters appear in this version. The nominative seems to have been an obscure KkipviTovs, with an epithet or let us say a patronymic in -a-Kvo9 parallel to -l-kvos and -ov-Kvoi as in Avovodtlkvos (pp. 37-9 below) and Cl\ovkvo9 already given. If we were to read Ovva into the second line we should have OvvaKvos, possibly meaning 'son of a mother Ovva''; but though metronymic names might pass muster in ancient Ireland, I cannot tell whether this would apply to Gaul. In any case we need not consider that question here, as Celtic, like Latin and Greek, is found to have had proper names in a which were not confined to women. See Stokes's Celtic Declension, pp. 17, 92, and compare Thurneysen's Handbuch, p. 176. Among the Irish instances suggested Plate THE CAVAILLON IXSCRl'p'rib'Sife' ' Photo. 1 Photo. 2 ; -'Plate II THE CAVAILLON INSCfllPTfONS - ' .a • » • • • Tlate III «• • 4 THE CAVAILLON INS^RlPfmN^* ...:'..:•::.••. Photo. 5 Photo. 6 THE CELTIC INSCRIPTIONS OF GAUL 17 by Stokes occurs a genitive Lane, Laine. The nominative should accordingly be Lan (for an early Land), whence the name of an old tribe, Macu Lane, in the south of Ireland (Stokes's Feilire, Dec. 6, and p. clxxx). By accident that name would seem to fit here, where AavaKvos would mean ' son of Lana \ It will serve well enough to indicate how the second line should probably be treated, though I need hardly say that I am not convinced that I have hit on the right reading or that our late colleague was right in regarding I^n as a man and not a woman. As regards I A A E I can say nothing, but only note the fact that its final syllable recalls ^parovbe. Possibly, therefore, it may have meant thankfully (merito), willingly (libens), or else dutifully (j)ie). There then remains -o-ovt, which looks like the ending of a name in the dative case, such as Elvovl and rpaa-eXovi. The nominative corresponding should end in -a-ovs, that is . . . a-ovSi better . . . a-os. As the inscription does not suggest ^parovbe or Kavrev, I am inclined to regard it as an epitaph, a view which is perhaps favoured by the fact that the stone is part of a rude pillar. In that case I should construe it thus : — ' To . . . . sos : Clirnitus .... nacnos dutifully (put it up).' II The Archaeological Museum at Nimes contains various inscriptions which I saw in 1909 and 1910, though I have not described them. Some of them reached the Museum subsequently to the writing, in 1905, and to the publishing, in 1906, of my paper on the Celtic Inscriptions of France and Italy ; and some which were there even then, I must have overlooked. There are also errors in my account of some of those which I noticed in that paper. Such are my reasons, good and bad, for reverting here to inscriptions at Nimes. 1. The first to be mentioned is on a rude block of reddish stone found in the ancient necropolis discovered in the Saint-Baudile quarter of Nimes : it is described by M. Mazauric in his account of the acquisitions by the Museum in the years 1906 and 1907. See his Musees archeo- logiques de Nimes: Recherches et Acquisitions (Chastanier, Nimes, 1908), p. 16. The stone is 0°^ 75 high by O"" Ti wide : the lower part is rough and untrimmed, while the upper part was rounded ex- cept where on both sides a level area was provided for the writing. As you face it the side near your left hand may be called No. 1, V g2 18 PROCEEDINGS OF THE BRITISH ACADEMY and that opposite your right hand No. 2. They read thus ^ respec^ tively : — 1. 2. AAre AAr^N NOYIA OOYA eAEBP The third line of No. 1, which is situated at the bottom of the levelled space, is very uncertain, especially the € in both instances. On face No. 2 the levelled space does not extend so far down^ so it has only two lines of writing. The inscriber here failed to make the letter € to scale: he began it too small, so it stands almost above the line. Otherwise the letters of this line are by no means badly shaped. The reading of the second line is uncertain except that of the first and that of a big A ending the line and beginning A€A€. I looked for an I before the A, but I could not trace it or find room for it. No. 1 seems to divide itself thus : AbyevovL 8e8e ^p , meaning possibly : ' Adgenui gave according to order ' to such and such a divinity the kind of offering termed kuvtcv or KavT€va, ' first- fruits,'' in other inscriptions of that part of Gaul. I shall have something to say presently concerning /3parov8e, otherwise this is the construction to be expected after the analogy of the other inscrip- tions of the kind collected on pp. 78, 79 of my previous paper; but these two inscriptions were never finished. Moreover AbyevovL was a form of the dative like Elvovi, TpaoreKovif and . . . ctovl (pp. 5, 10, 17 above). I look at all the writing on this stone as the outcome of a mason's practising his hand or merely amusing himself by ignorantly imitating inscriptions which he had seen. In No. 2 he equated with his AbyevovL another form, Abyevoov, which we know to have been a dative of the u declension. It could not be nominative : witness Tapavoov in the Vebrumaros inscription (Celtic Inscr., vii). It is highly improbable that he had ever seen either AbycvovL or Abyevoov as anything but a dative : at any rate the forms given by Holder are Adgennus and Adgennim, Adgennia, all from Nimes, and Aby€vvo-pi$ from LTsle-sur-Sorgue in the Department of Vaucluse (Celt. Inscr., x); not to mention Adgennonius from the neighbourhood of Novara in N. Italy. Now Adgennus occurring in a Latin inscription may represent a Celtic * The reading is very difficult, and M. Mazauric differs from me in detecting an N in No. 1 at the end of line 1 (A Ar€N), and in No. 2 another N beginning line 2(N00YA). He may be right in both, and also in not suggesting a third line at all. THE CELTIC INSCRIPTIONS OF GAUL 19 Adgenno-s or Adgennu-s^ and it is right to say that considerable latitude in the matter of declension appears to have been allowed in the case of Celtic proper names. Witness the instances collected in the Corpus, XIII, part iii, p. 119, where we find cited such nominatives as Biicco, Biicciis, Buccio and BiLCcius, Cot\t\oy Cot\t\i(^^ Cot\t\i09 and others. The etymological spelling requires nn, and the name Ad-gennu-s or Ad-genno-s seems to have meant 'gainer, winner, one who makes acquisitions ' : compare Con-genno-, p. 29 below. 2. From Montmirat (Gard) comes a fragment of a Gallo-Roman altar, presented to the Museum in 1907. M. Mazuric, loc. cit., pp. 71-3, gives a careful description of the locality and the anti- quities found there. The fragment in question bears the letters BPATOYT . . . . The last letter is incomplete owing to a breakage: I tried to read BPATOYA in the hope that it represented an original jSparovbe, but it is impossible, and the T seems certain. Besides, it is hard to see how that vocable could have begun the inscription. The probable explanation is that it is part of a proper name B P AT Y-T . . . . where the second element began with the consonant t, say of -toutios ; but it is right to state that the only compound of Bratu which Holder has found is the place-name Bratuspantium. 3. In the same year, 1907, M. Mazauric, while on a walk to the part of Nimes called Saint-Cesaire, detected an inscribed stone in a ditch near the Cafe de Font Jaisse. It is now in the Museum, where it attracted my attention. It is a very rough stone, and, according to him {Recherclies et Acquisitions, p. 73), it measures O"" 72 as its greatest length by O-" 32 high by 0"^ 45 thick. The face bearing the inscription had not been very carefully levelled, and the letters are tall and irregular, reading as below : — PITOY M. Maruejol, who has carefully studied it, thinks that Ritu is the dative of Ritum, the Celtic riton meaning a ford, and that it was the name of a local divinity, in fact that of the spring called Jaisse, and he adduces among others that of Ritona from Montaren near Uzes in the Department of Gard, I must confess to some difficulty in applying a word meaning a ford to a well or spring, and I am more inclined to treat it as a man''s name, say that of the mason who placed the stone in its position. In that case I should regard Ptrou as a nominative for Ptrovy of the u declension, and equate it with Ritits on pottery bearing the stamp RITVS F{ecit) found in various towns in the Valley of the Rhine (C. /. L., XIII. iii. 10010. 1643). Holder cites also RUT VS. F. F. from Le Chatelet, near Charleroi, in W PROCEEDINGS OF THE BRITISH ACADEMY Belgium. The name may have meant 'Runner', being possibly derived from the root ret, 'to run,' as in Irish rithim, 'I run,' Welsh rhed-eg, for early retica, ' the act of running.' It is doubtless a simplification of compounds like Ritu-maros and Riturgalos: see Holder, s. vv. Ritumara and Ritukalos, 4. Celtic also, in all probability, is the inscription in this Museum from Uzes (Gard), reading as follows : — CENIKIOC •.• ABPCO The stone appears to have been found at a well, and it is broken off close to the omega ; so close in fact that the right side of that letter is imperfect. How much exactly there was of this line it is impossible to say ; not to mention that the stone may have been deeper, pro- viding room for a line or two more beneath the one partly remaining. The names Senicius or Senecius, and Senicia or Senecia, appear according to Holder to have been common in Carinthia, Carniola, and North Italy, not to mention the related form Senecio in N. Italy, Nimes, Vienne, and Mainz. They are derived from a stem senec, which we also have in the Latin senea? (= senec-s) and senec-tus. The still simpler form in Celtic is represented in Irish by sen-, ' old,' Welsh hen- and hen of the same meaning, Greek e^oj, ' old.' The epithet may have been ABPOJNIOC, ABPCjl)NIKNOC, or the like. The ^ in Ahr- may be either original or stand for an earlier /x, as in CO BR EXT- (in the Coligny Calendar) for comrecht, Welsh cyfreith, cyfraiih, 'law.' That is probably the case here, as our a/3pa)- seems to be related to the Irish adjective amhra, which Dinneen explains as ' good, great, noble, prosperous, lucky ', while in Stokes's Oengus both amrae and ad-amrae are rendered ' wondrous '. Com- pare the following gloss cited from the Wiirzburg Codex 15^ in the Grammatica Celtica^\ p. 916, imforlinged mor n-amri de (' factum inde multum miraculi'). Amrae, Amra, genitive amri, belong to an adjective amr-io-, amr-id, and we have the stem as amar-, amir-, in the name Amar-gen, Amirgen-us {Thesaurus Palaeohibemicus, II. 262, 316), later spelling, Amhairghen, 'Wonder-child,' in Welsh Ahr-gen, which occurs in the Book of Llan Ddv, where we have also Emrdil (for earlier Amr-dil), Ebrdil (fem.), later Efrdil, Euyrdil, Erdil (?), pp. 76, 78-80, 192, 348, 358, 359, 364. 5. The next inscription I wish to mention is that which I read at first Marm/co . . Kovvovj^p .... I think I can now do it greater justice than I did in 1905, and in any case I must correct my statement {Celt. Inscr,, p. 38) that it was found at Nimes ; by what error I arrived at that conclusion I cannot now discover. In any case there is no THE CELTIC INSCRIPITONS OF GAUL 21 doubt that it is the identical stone given in C. /. L., XII, p. 832, as reading . . . AAHAO . . . | . . . KOAAOYPT The editor gives a reference to Allmer's Revue Epigraphique, II. 82, 83, whence he copied this reading. Allmer's account of the stone is that it was found near Collorgues, in the neighbourhood of Uzes, in clearing and digging some land about the year 1869, and that it was imme- diately broken into fragments, of which the one with the above lettering was saved by a well-known antiquary, M. Lombard- Dumas, who deposited it at a place of his at Garrigues. When and how it found its way into the museum at Nimes I do not know, but it was sub- sequently to Allmer's writing about it. He describes it as measuring in height 0°^ 16 and in length at the bottom 0°* 30 and at the top 0*" 25 ; but it is best to produce his description in his own words : — 'Fragment detache d'une pierre de moUasse lacustre, presentant primitivement la moitie superieure d'un spheroide a sommet tronque et creuse en la forme d'une profonde cuvette circulaire.' The lettering was on the circumference of the dish at the top, but the surface is so worn away that the upper ends of the letters of the first line are gone. I read the first letter as A (not A, which AUmer preferred), but it may be the latter half of an M, and what he gives as TT I took to be Tl or fl. The A following is fairly clear, but its end curved outwards, which helped to suggest a C. In fact, on the cast accompanying the original, somebody has indicated in paint or pencil a tall, neat, open C by using the curve of the lower part of the A and producing it above on a surface which is gone in the original. It is needless to say that Allmer's evidence completely disposes of it, for the next trace of writing after the A he took to be a small 0. There he was mistaken ; for in the first place one does not expect the small to be placed on the level of the bottom of the other letters, but nearer the top, where in this instance the original surface is broken off ; and in the next place I find, on careful examination, that it cannot be an at all. It can only be the lower half of a B or rather less than the half, as the breakage leaves the top of the lower half open. On the other hand, it has the interior angle at the foot of the B intact. Next, the A B suggests an to complete the dative plural of some designation of the Mother Goddesses, such as A ndounn ABO or NamausiJcABO, In the second line I read N N for r'r', and not A A, and the last letter but one is B not P, while the last letter is P not f ; the top is formed into a little triangle often characteristic of that letter. The lower part of the perpendicular limb of the letter is somewhat indistinct, but AUmer would seem to have detected it, since he read 22 PROCEEDINGS OF THE BRITISH ACADEMY the whole as f. The possible readings, with conjectural extensions, may be represented as follows : — MarpejSo ^A^IABo KONNOY BParovSe Kovvov is probably a nominative of the u declension for earlier KovvovSf or Kovvvovs, and we seem to have a derivative in the Cunur of CVNVANOS, the name of a prince of the Arverni, cited by Holder from coins of that ancient people; not to mention a related form CON NO, which he produces from coins of the Lemo vices, and treats as of the o declension. In that case we may have it perpetuated in the Irish Conn, genitive Cuinn, ' Quin ' ; but as he gives Conno no final s in any of his instances, the name may have been of the consonantal declension making a genitive Connonos, not Conni. Having detected a dative and a nominative, we may look at the inscription as a whole ; and the first point to be noticed is that it probably never had a verb : the dative sufficed to convey the sense that Connus had given to the Mother Goddesses of the locality some- thing it was not necessary to name, to wit, the hollowed stone on which the dedication was written. This stone dish, which the vandals who found it forthwith broke into pieces, was a vessel required possibly in the cult of the Mother Goddesses. What their exact name was it has been found impossible to say: the dative may have been Latiabo or Lagiabo, Matiabo or Magiabo, not to mention Lapabo or Mapabo, Of these guesses I should prefer Matiabo, from the adjective matis, 'good.'^ The whole would then mean 'To the Good Mothers Connus (gives this) at their bidding \ But Marpe^o, though the word lends a certain symmetry to the inscription, does not seem essential to the sense here suggested ; for instead of calling them Good Mothers it may have sufficed to call them the * Good ' ones in the feminine, as if in Greek the feminine ayaOals had been applied to them. ■^ Dr. Stokes in his Fick volume gives the old Celtic word for ' good ' the two forms matis and mato-s (p. 199)^ and to the words connected with mati-fs would helong our adjective in -la^o, while others are derived from mato-s. The former is exemplified in the Irish word maith, ' good,' in Manx mie, while the Scotch Gaelic is math, not to he traced to matis. On the other hand the Coligny Calendar has only matuSy which seems to be of the u declension. All this would seem to imply no less than three declensions in the case of the one adjective, but I take it that matos is not original, but produced by the encroachment of the o declen- sion on the more restricted u declension, as when Latin magistrntus was given a genitive magistrati. The Welsh word was mat, now mad, with regard to which we only know for certain that it cannot be traced to matis, but to matos or matus. THE CELTIC INSCRIPTIONS OF GAUL 23 Nothing has been said as to the probable use made of the deep stone dish ; but if we could establish a parallel between the Mother Goddesses and the Fairy Godmothers of the Good People, the Fairy Tales of Wales might shed some light on the question. For one of the common requirements of the Welsh Fairies when they entered people's houses at night, when the inmates were asleep, was to have vessels provided for them full of clean water for washing and dressing their infants. If we may apply this to the case of the Good Mothers, the stone dish was a sort of font in nightly use for the washing of the babies, which an ancient altar at Cirencester associates with them, as I learn from our colleague Professor Haverfield. I may add that in Glamorgan, and other counties of South Wales, Fairy changelings are called Bendith y Mamau, ' the Mothers' Blessing.' In support of the view that the Mother Goddesses became the Bonnes Dames, the Dames Blanches, and similar Fairies of Medieval France, I may cite the authority of the late M. Florian Vallentin, in a paper contributed to the fourth volume of the Revue Celtique, pp. 27-36. Popular belief represented them as inhabiting rocks, grottoes, foun- tains, and the ruins of ancient castles, and engaging in the protection of the weak against the oppression of the strong, and, by their apparitions, frightening nocturnal spoilers and murderers from their evil purposes, while some of them crowned with a mysterious aureole the beginnings of great families and foretold their destinies. They sometimes take the poetic form of Melusine, of the Dame Blanche of the Avenels, of the Banshee of the Fitz-Geralds, and the like (p. 29). To come back to the idea of ablutions, M. Vallentin states that the memory of the Mothers is perpetuated in a legend attached to a curious monument of nature situated a very short distance to the north of the Church of Saint-Romain-en-Gal, near Vienne. It is called Puits des Fees or Fort des Fees, and he quotes from a previous writer ^ the following description of it : ' Sur un petit rocher qui regarde le Rhone aupres de Saint-Romain sont trois creux ronds que la nature seule a formes, quoiqu'il semble d'abord que Tart y ait travaille apres elle. On dit qu'ils etaient autrefois frequentes par les fees ; qu'ils etaient remplis d'eau quand il leur plaisait et qu'elles y venaient prendre souvent le plaisir du bain ; car on feint que toutes ces fees n'avaient pas de plus char- mante volupte que celle-la ' (p. 33). 6. The Collias inscription in the Nimes Museum was guessed in my Celtk Inscriptions, pp. 39-41, to read Ekivvos Viovixavios Avboovv- * Chorier, Recherches sur les antiquites de la ville de Vienne, p. 183. 24 PROCEEDINGS OF THE BRITISH ACADEMY va^o hebe ^parovbe Kavrev and to mean 'Ecinnos son of Riumanos gave firstfruits to the Andounnas by their decree'; but I fear that I must give up my reading of the first name. I thought I had found a ligature standing for IN or Nl which as I was led to believe occurred in an inscription at Dijon (C /. L., XIII. 5465), to wit, in a name which has been read DASILLInWSO/^ . VX, the Editor remarking that he did not know whether the nondescript character was N or N I or I X I or something else. In August, 1909, I had the curiosity to go to the Dijon Museum and look at the stone, I found the character in question to be more like MA ligatured, though it was not exactly that either : in any case it was not what I wanted. The inscription has been otherwise badly read : I made it DABILEIil/JSO. The A and B are ligatured, and the next letter is either I or L, and there may have been VX at the end. The first name (written with B for V) is to be doubtless correlated as Daviky with the Davilos cited by Holder from a potter's stamp D AVI LI M(anw) found at Vienne (C./.L. XII. 5686. 301). To return to the Collias inscription, I seize on this opportunity of offering a solution of the difficulty experienced in the interpretation of /3/oarov6e as a Celtic word ending with the preposition de, ' from,' made into a postposition. The late Celtic scholar, M. d'Arbois de Jubainville, was, I fancy, influenced chiefly by this when he undertook to argue that the group of inscriptions containing the hratude formula were not Celtic. I tried to deal with his arguments in my paper on the Celtic Inscriptions of France and Italy, pp. 19, 79-81, and suggested that possibly hratude had nothing at all to do with de^ ' from,' whether Celtic or Latin, but that it had the meaning of ex imperio, ex iussu, by virtue of its being in the oblique case of a de- rivative bratud formed from hratu. This analysis has failed to lead me to any result, but since then a study of the formation of certain adverbial phrases in Irish has supplied me with a clue. Turning to the Adverhia Hihemica in the Grammatica Celtica^\ pp. 608, 609» I find that the first set of instances consists of the definite article prefixed to the neuter of an adjective, and the case in which the words are is called the dative : Zeuss says dative or ablative, but even that is too narrow, as the case meant covers also the senses of the locative and the instrumental. It would be somewhat less mis- leading to call it a dative-ablative case ; but for the sake of brevity Irish grammarians are in the habit of calling it simply dative. The first instance which Zeuss gives is in biucCy ' paulo, paulatim,' from becc, 'little, small.' The exact equivalent in Welsh is ijnf achy the adverbial rendering of bach, 'little, small.' The next kind of THE CELTIC INSCRIPTIONS OF GAUL 25 adverbial phrase in what is probably the same dative case is made in Irish to end in -id (sometimes written -Uh), From among the instances given under this head the following may be chosen : — Ind oindid^ gloss- ing ' semel ', or (spelt ind oeridaid) glossing ' singiilatim ', where ind is the definite article in the dative, and oinde or oenda is an adjective derived from oin, Sen, 'one/ Welsh un^ Latin unu-s, for an older oino-Sf ' one ' ; and ind aicnetid^ gl. ' naturaliter ', from aicneta, ' natu- ralis,' derived from aicned, ' natural We may also take participles ending in -the or -tha, such as that in ni in tuasailcthid^ gl. ' non abso- lute' (tuaslaicim, * I let loose or make loose '). To these and the like Thurneysen (Handbnch, p. 229) adds some instances of nouns, such as in diglaid (glossing ' ulciscenter '), from digal^ O.Welsh digal, now dial^ ' vengeance "* ; and ind dirmith (gl. ' summatim \ otherwise explained by the word ' breviter '). The noun here is drem, dram^ ' number,' for an early ad-rlma. It is a feminine of the a declension, nominative drem or dram^ genitive dirme, dative drim, accusative drim-n. My notion is that the final d represents a declensional element -de or -di appended to the dative in the instances in question. However, we cannot get that to fit with the Irish declension as we have it : one has to go back to the early Celtic form of that declen- sion. It will be convenient here to follow Stokes in his Celtic Declen- sion, p. 102, where he gives the declension of reda, ' a chariot,' in what he has called Proto-Celtic. On that I proceed to project such cases as we want of the Irish word dram, direamh, as follows : — Stokes's Proto-Celtic. Early Goidelic. Historical Irish. Nom. reda ad-rlma aram, arem, direamh. Gen. redes ad-rlmes airme, airmhe. Dat. rede (redl ?) ad-rlme (ad-rimi ?) arim, arimh, aireamh. Let us now append to the early dative ad-riml or ad-rime the syllable de ; then adrime-de, subject to the rules as to Irish desinences, yields us drimid, diiuiid, which is practically what we have in the adverbial expression i7id airmith, ' summatim, breviter.' It may be asked why I select -de, but it would take too long to discuss vowel harmony in * The Welsh equivalent both in meaning and derivation is eirifj as in an-eirif, •^numberless.' In eirif ihe first i stands for an earlier d, as in cadeir, cadair, ' a chair,' adapted from cathedra : see p. 14 above. The simpler words Irish rim and Welsh rhif, for older rim, also mean ' number ', but they differ in gender, the Irish rim being feminine, while the Welsh rhif is masculine like the Teutonic cognates, such as the A. -Saxon rim and O. H. German rim. It looks as if Irish rim, originally masculine, had been made feminine under the influence of drem. On the other hand drem in its modern form of aireamh is given as masculine by Dinneen ; the gender of Welsh eirif has not been ascertained. 26 PROCEEDINGS OF THE BRITISH ACADEMY Irish : let it suffice to say that a broad vowel would not fit, so we are left to choose between e and t, and I have taken the former, that is I have supposed the element added to the dative to have been -de} When similar treatment is applied to ^paTovbe we have perhaps to regard ftpaTov, bratu, as a dative, and not as the stem of a word of the u declension. We have such a dative in an inscription containing Magalu, ' to or for Magalos,' on a vase found in the neighbourhood of Bourges, and now preserved in the National Museum in the Chateau de Saint-Germain (Celtic Inscriptions^ xxxii). But the dative in u belongs mostly to words of the o declension ; so here the nominative implied would probably be brato-s, not bratu-s ; but what can brato-s have been ? Comparative philology supplies an answer. The Latin words grates, gratia, grdtus are found to have had as their initial a consonant which in Celtic would have been gu liable to be simplified into b. Thus Latin grdtu-s would have corresponding to it in Celtic brdto-s : see Walde's Dictionary, s.v. grates, and compare Thurneysen, p. 190. Now comes the question of the meaning of bratu-de. Stokes in his Celtic Declension, p. 63, quoted from De WaPs De Moeder- godinnen the following inscription : Matronis Afliabm M. Marius Marcellus pro se et suis ex imperio ipsarum. He drew from this and similar cases the conclusion that ex imperio would be the equivalent of ^paTovbe. But ex imperio or ex in.ssu is by no means the most usual phrase in Latin ex-votos ; and it is hard to believe that in none of the Celtic ones in question was the faithful donor allowed to let his co-religionists understand that he was acting of his own free will in the matter of offerings to the divinities whom he worshipped. Thus one may say that the ex imperio interpretation is not to be applied too frequently, even if one had no other to offer. That, however, is not the case, for if we may treat brdto-s as the etymological equivalent of Latin grdtus, we may assume that the meanings were approximately the same. In that case (3paTovb€ might be rendered approximately by the Latin adverb grate, ' with pleasure, agreeably, willingly, thankfully, gratefully.' In other words it was an expression of thanks, with much the same ^ After writing this I happened to have my attention drawn to the Gram. Celtica, p. 231, where one reads : ^Abl. adj» propriae formae (cf. gallicum ^parovde) I indoracdid (gl. dorice)/ &c. But here no hint of a suspicion occurs that the declensional element which helped to constitute the propria forma was etymologically one and the same in both. At the last moment I notice that Holder in his third volume, col. 926, quotes R. v. Planta as equating ^parov- with Latin merito. This also, should it prove tenable, would suit my view as to the -5e. THE CELTIC INSCRIPTIONS OF GAUL 27 force as the Latin formula ' votum solvit libens merito ', except that ^paTovb€ involved no express reference to a vow. ^^^ Before leaving Nimes I wish to mention one or two Latin inscrip- tions of considerable Celtic interest. The first of them comes from an oppidum on the mountain of La Baume, near Belvezet (Gard) : see M. Mazauric's report for 1908, p. 41, and also for 1906, p. 34. He gives it as reading : — TERTIVS. TIN CORICIS F. SE GOMANNAE V. S. L. M. The discovery of the name Tincorix interests me as helping us to the analysis of that of Tincommius, son of Commius ; he is well known by his British coins, and his name is doubtless a shortening, for euphony's sake, of TincO'Commius, The goddess Segomanna's name is to be added to Holder's Seyo/xartKos, occurring in a Nimes inscription which he has pronounced suspect. ^^^ An inscription found at Nimes in 1906 is given by M. Mazauric, p. 27, as reading : — D . M MESSINAE MESSINI FILIAE TASGIA . TITVLLA POSVIT Here Tasgia is to be placed by the side of Holder's masculine Tasgitis. The spelling with g possibly indicates that the s was pronounced soft, that it was in fact z, and Holder seems right in deriving tasgo- from an earlier tazgo-, to which is related the Irish Taidgg, Tadc^ later Tadhg, one of the commonest personal names in Ireland, at any rate until it began to have Thaddaeus, Timothy^ and other New Testament names substituted for it. 7. While staying at Nimes in August, 1910, we made an excursion to Montpellier, at the invitation of our friend Professor Babut, in order among other things to see the inscribed stone from Substantion. That place is about three kilometres to the east of Montpellier on the other side of the river Lez. My friend reminded me that it is a station mentioned in the Itinerary from Bordeaux to Jerusalem. The stone was discovered in 1840, and the portion of it found inscribed consists of two edges of a square table forming part of the top of a pillar, which he describes as ' un chapiteau dorique ', He adds that ' Les caracteres sont graves sur le tailloir comme I'inscrip- tion nimoise MATPEBO N AMAY^IKABO '; see C.I.L., XII, 28 PROCEEDINGS OF THE BRITISH ACADEMY p. 383, where the Editor terms the part of the stone inscribed 'Capitulei marmorei abacus'. The sides of this square measure about 2 feet 4 inches, but only two sides are visible, for it has been cemented into a corner of a court of the University. It is believed, however, that the other two sides had been so damaged that no. writing was left on them when the fragment was put into its present position. What can be now read stands as follows : — lA ! INOYCIA The lower ends of several letters occur before the A ; the n^xt before it was probably I, or else one of the letters F, T, or P. After the A comes U, which might be an I joining the beginning of an A. The letters on the second side are all quite legible as far as they go, but the stone breaks off after the A. However, the C looks as if it had a point in its centre, a feature by no means unusual in other inscriptions, but M. Babut was decidedly of opinion that it is only a little excoriation of the stone, and the Editor of the Inscriptions de Langiiedoc, NimeSf No. 107, appears not to reckon it a part of the lettering ; see Holder, s. v., . . . . inoucif where a Latin c has no business, if one may trust the A, Y, A, which argue Greek values for all the letters. But whether one is to divide the words between the I and the A is not certain, for one cannot rely on the absence of a point there. Holder pronounces for A[EAE] which would suggest the ^parovbe formula, and prevent our construing the whole as an epitaph. This is strongly corroborated by the position of the lettering and the general similarity of the whole to the Nimes inscription in honour of the Nemausian Mothers. Possibly ivovai was not the whole of the name. Holder has collected some names ending in -tiso-, -tissa- (At(t)iissay Bergtissa, &c.), and -ussio-y -ussia (Atussia, Cantnssius^ Cintussia^ &c.), but perhaps the ones most nearly in point are Ber- gussa^ and the Bergusia to be mentioned presently as the name of a goddess (p. 34 below). The dative of Bergus(s)a and Bergusia might be Bergusi in both alike. So vice versa, . . . lvovo-l might be the dative of the name of a goddess . . . ivovaa. The whole inscription might accordingly be : ' So-and-so to the goddess . . . inusa gratefully gave firstfruits.' This is a mere conjecture, but whatever the inscription was as a whole, the care with which the lettering was cut, and the place where it was cut, indicate that it was regarded as an important document. So much the greater the pity that it has been practically lost to Celtic epigraphy. 8. Before leaving the inscriptions of the South of France I may here mention one which I have not seen, to wit, the one given in C /. L.y, THE CELTIC INSCRIPTIONS OF GAUL 29 XII. 5793, as belonging to the vicinity of AUeins (Bouches-du- Rhone). It reads as follows : KOrrENN OAITANO C KAP0IAITA NIOC This would mean ' Congennolitanos son of Carthilitanos ', and litano- is represented exactly in Irish by leathan and in Welsh by llydan, ' wide, large."* But what did Congenno- mean ? I can make nothing of it, unless it meant acquisition or possession, from the same origin as the GendUl-i of the Stainton bilingual in Pembrokeshire, and its Welsh derivative name Gennillin. Among the kindred words are Latin ^vekendo^ ' I lay hold of, seize, grasp, catch, take,' Greek yavhavdi, exabov, x^iaoiiaL^ ' I seize hold of,' English ' get, heget^ forget \ The reduction of nd to nn, n, has parallels in such forms as Esanekoti^ AnoJcopokios, for Exandecotti, Andocomhogios in the Celtic inscription at Novara in N. Italy. The whole compound may, accordingly, have meant * one who is large as to his possessions, one who acquires far and wide \ The genn in Congenno- is probably the same as in Adi-geniji)- already discussed at p. 18 above ; the predominant spelling doubles the n in that name likewise. The other name, Kap^tAtray-, is still more obscure ; but possibly the stands for a lisped j, and KapOi goes with the carsi- of such names as Carsius and Carsia, Carsicios, Carsidius^ and Carsidia quoted by Holder. But these are too obscure to be of any present help, and it seems preferable here to give 6 its normal sound of th. It is well known that in Brythonic rf, re, rp are represented by rth, rch, rff or rph, while Goidelic retains the older consonants. The difference forms a far- reaching distinction between Goidelic and Brythonic. The modifica- tion would probably be mostly ignored for centuries in the ortho- graphy as a corruption. So there is very little chance of discovering when it came in ; but it is possible that it took place both in Gaulish and Brythonic ; that is to say, there was a tendency to this pronunciation even before Brythonic separated from Gaulish. In that case we could equate Kapdi- with the first syllable of Kaprapos in the celebrated Nimes ex-voto to the Marpe/So 'NafxavaLKa^o {Celtk Inscr., p. 34), and with the cart-i of one of the Rom defixions (ibid., p. 95), also with the cart-i (for carthi) in Cartimandus, Cartimandua, a name familiar to the readers of Tacitus. The second element in them has been interpreted by M. d'Arbois de Jubainville (Noms Gaulois, pp. 127, 128) as 'one who thinks, reflects, meditates, minds', the origin being the same in fact as that of the Greek 30 PROCEEDINGS OF THE BRITISH ACADEMY \erh fxavddvoD, ifxadov, ^ I have learnt"'; one might doubtless add the English word mind, both noun and verb. What the meaning of Kapdi, carti^ may have been is not known, but my conjecture is that it signified ' strong, powerful, vigorous \ as this would seem to fit the cases mentioned without any violence to the context where there is any. Cartimandus would accordingly mean ' one who thinks power- fully, one possessed of a strong or vigorous mind '. Similarly KapOi- \LTavos should mean 'one who is powerful or active far and wide'., Ill 1. The number of inscriptions in Celtic found on Mont-Auxois and brought to Alise-Sainte-Reine near the Alisia of ancient Gaul, has materially grown since my paper on the Celtic Inscriptions of France and Italy was written. In fact at that time there was only one such, and my notes on that require to be revised in the light of an ex-voto in Latin to be mentioned presently. I read the former now as follows : — MARTIALIS - DANNOTALI I EVRV V VCVETE - SOSIN CELICNONo- ETIC GOBEDBI " DVGllONTllO VCVETIN ^IN ... ALISllA^ In that paper (p. 4) I had committed the mistake of placing the last leaf but one at the beginning of the last line but one : it really belongs to the last line of all. The gap I suggested filling by inserting DV, but I am now in doubt, since the D, had it been there, would still probably show a small portion of the bottom opposite the reader's left hand. I cannot suggest at present how the lacuna is to be filled. The Editor of the C. I. L, (vol. XIII. 2880) has suggested a leaf. In that case, the line originally read, ^IN^ALlSllAc iigim, have had wide acceptance, as we trace to it the Welsh mechdeyrn, mychdeyrn^ 'monarch, ruler,' Cornish mychtern, * a lord, a sovereign,' and Breton machtiem : the Latin plural is entered in De Courson's index to his ' Cartulaire de I'Abbaye de Redon ' thus : ' Machtierni, Tiarni, Tyranni,' and •' Machtierni vel principes plebium et parochiarum '. On these lines it would be reasonable to interpret avvoticnos as meaning ' the son of an avvotis, one of the rank of an avvotis ', a word which, as we said, was likely to have signified lord or head. This derives support from the fact that the monument is of unusual pretensions for one in a Celtic, language. (3) C€Cla, C€Cly, Sesia, Sesis. CECIA would be the feminine of C€CIOC, which is cited by Holder from Rome as Sessiios in Latin (C /. Z., XV. 420). He quotes Sessis from Ennodius, Bishop of Pavia, who lived from 473 till 521 ; but he only uses it as the name of a tributary stream of the Po, now called Sesia, near Vercelli in Piedmont : another form of the latter name Holder gives as Sesites, (4) kAAMAKi, /cAAMAKt, Clamaci, Calmaci. Holder cites Villam Clamiciacum as the old name of Clamecy in the Department of Nievre. Clamici comes very near the Clamaci which is favoured by the reading of the second line of this inscription. If instead of KXa we read Kak we should have KAAMAK I, genitive of KAAMAKIOS, which Holder finds imbedded in Calmaci-acus, whence the modern name of Chaumuzy, borne by a place near Rheims. Calmac- would be of the same origin as Irish calmu, * valiant, brave,' from an early 46 PROCEEDINGS OF THE BRITISH ACADEMY calmio-, which Brythonic, treating as calmiip-, has made into celfyd, ' ingenious, artistic,"* Breton Jcalvez, ' a carpenter ' : compare Welsh cel/icyn, ' a small article of furniture.' We have possibly to fall back on a shorter genitive K A AM At or K A AM At like Eccaios and the other instances collected by Holder under -aio-s, to which may be added the genitive Oxtai occurring in one of the Moritasgus inscrip- tions to be mentioned presently. It looks, however, as if a longer genitive like KAAMAKI would fit the space occupied by the name somewhat better. (5) TAP MA, Garma. The probable explanation of this name is that it is a levelling of Gorma, of the same origin as Modern Irish ^orm, 'blue."* Stokes in his Fick volume, p. 114, has two words gorm, one from gormo-s, ' warm, roth,' and the other from gorsmo-s, * dunkel,"* but they seem to me to be one and the same word with a curious variation of meaning. The Welsh is gwrm, ' dun, dusky, of a dark brown colour ' ; but I take it that neither the Irish nor the Welsh was originally the name for any special colour. This is corroborated by a gloss of O'Davoren — * Gorm .i. urdairc ' : see Stokes's Three Irish Glossaries, p. 94. Urdairc means ' splendid, illustrious, glorious \ and the lady whose name has here been guessed to have been Sesia or Sesis may have had an epithet with some such signification. (6) BIPAKOTCOY is the dative of BIPAKOTOYC, J5iraco^w5, of the u declension, a compound which analyses itself into Bira-cotu-s, The qualifying element bira is of the same origin as Irish bir, biur. Mod. Ir. bior, genitive bero, bera, ' a spit, a lance point, a spike,' Welsh ber, ' a spear, a lance, or pike ; a spit for roasting meat ; a skewer or stake.' The related Latin form is veru, 'a spit or stake,' which suggests that both the Irish and the Welsh were originally neuter of the u declension, like Latin veru, a fact attested in the case of the Irish word, though both are now masculine : see Hogan's Irish Neuter Substantives in the ' Todd Lecture Series ', IV. 206. The other element is Cotu-s of the u declension, and possibly to be identified with Caesar's Cotus which is treated as of the o declen-^ sion ; but this does not prove that it was not of the u declension in the Celtic of the Aedui. That it was so is suggested by the probably related form Cotu-atus, the name of a leader of the Carnutes, also mentioned by Caesar : see Meusel's edition of the De Bello Gallico, vii. 3, 39. The Indo-European root seems to have been kuth, whence in Teutonic hud, related possibly to Greek Kevdia, * I hide ' : see Kluge, s.v. Hutte. The Welsh is represented by cot, cod, and cwd, * a bag or sack.' So Bira- cotu-s may be inferred to have meant ' a hiding or heeding with a spear ', more generally speaking of ' one who THE CELTIC INSCRIPTIONS OF GAUL 47 is a protection effected by means of the spear \ Names of the u declen- sion, be it noticed, may be masculine or feminine : this is probably masculine. (7) TIC .... ANN CO. This represents the dative singular mascu- line of a name ending in the nominative in ANNOC. The second element might be AANNO-C which is suggested to me by the danno- of Danno-talis, p. 30 above, but it might just as well be BANNO-C: in fact this will be found to be the more probable of the two. Among the related forms cited by Holder may be mentioned Banna, and the place-names Bannaciacus, Bannaventa, Bannolus, Bannovallum, and others. The meaning may possibly be akin to that of Irish hann^ ' IsiW,'' Jbr-bann, ' command,' from banno-, ' a ban, inter- dict, prohibition': see Stokes in Fick's book, II, p. 159, and d'Arbois de Jubainville, Les Premiers Habitants de PEurope, II". 335. The first element in the compound reads T I followed by traces of a letter which seems to have been C. The whole of it may have been TIC A of the same origin as the Tisios, Tisia, Tisiacus and the like cited by Holder, but hardly, perhaps, to be severed from others that he gives such as Tessius (Teddins), Tessilhis (Teddillus), Tessilla, f. Tessig- nius {Teddicniu^). These suggest that tess- represents an early tens- to be traced in the Latin tensu-s, tensu-m, tensd, ' stretched out, dra^vn tight, strained, tense,' Gothic -fansan, O.H. German dinsan, ' to draw or pull,' Lithuanian te'^s-ti, ' to draw or stretch.' Should these sur- mises prove well founded, Tia-a-pavvo-s may have meant ' one who strictly observed the prohibitions affecting him, one who carefully avoided the ill-luck of violating his gessa or taboos ' : a man's gessa play an important part in old Irish tales. However, the exact signi- fication of these proper names is of no great consequence here ; one is more troubled by the inability to decide whether the space required some such spelling as Tia-a-a^avvo), TKna^avvoi, or the like, rather than TLo-a^ai'Vd), to which I give the preference. (8) KOBPITOYAOaY is the dative of Ko^/)tTov\ov-9, Cobritulu-s of the u declension. The first element is Ko^pi-, in O. Irish cobir, Modern Ir. cobhair, cabhair, fem. 'help, assistance, support ' : see the Gram. Celtica, p. 781, and Stokes's Fick volume, p. 169. When the first element in such a name as that in question here means protection or safety one naturally looks to the second element to convey the idea of bearer or bringer, as in the case of such Greek names as OvrjaC-cpopo-^, Ekinbo- (f>opQ-St Kap7ro-c^opo-9 and the like. So the tovXov-s of the Celtic com- pound may be taken as meaning bringer or bearer, and the compound Cobri'tulu-s may be rendered ' Aid-bringer '. The vocable tul- of the same origin as Latin tid (as perfect oifero) occurs in several Irish per- 48 PROCEEDINGS OF THE BRITISH ACADEMY sonal names, such as the following, to be met with in the pedigrees in the Book of Juemster—Tulach, fo. 368« ; Tuladrain (genitive), f. 357^ Tolaing (genitive), f. 317^ ; Tola, f. 350^, gen. Tola, f. 350^ and Tolai, f. 358^ ; Tuli latha, f. 355^ ; Tolaid ^ (gen.), f. 331^, 332^ Among these we seem to have an inverted compound parallel to Cobri-tulu-s in Tolang, gen. Tolaing, in the tribal designation Hui Mceldin Tolaing, 'the descendants of Maelan, the bringer of protection or safety,' where aing is to be interpreted by means of the verb angim, aingim, ' I protect,' which Thurneysen would refer to some such theme as aneg- (Handbuch, p. 331). Lastly, I should guess Kol3pLTovXovs to have been a woman's name. Compare Cartimandus, already mentioned (p. 29), which appears to have been the Celtic name of the Brigantian queen. A woman's name in -us must, however, have struck a Roman as some- what strange, and she seems to have been provided with an alternative one which sounded feminine enough. This was Cartimandua, derived from Cartimandii-s. See Mr. Furneaux's edition of the Annals of Tacitus, xii. 36 (II, p. 259, note). (9) Now should come a word ending with B as an abbreviation of the dative plural of an ancient vocable meaning children. The search for it brings me back to the question of the small fragments with letters or portions of letters on them. Commandant Esperandieu sketches three of them in his article in the journal Pro Alesia (Nos. 3 and 4), p. 45. For the student's convenience and with the Com- mandant's kind permission, these are embodied in photograph 9f. The first of them is used by M. de Ricci to finish TAAOC in the first line ; and to explain how he does it I give his sketch of the whole inscription in photograph 9 d, which shows also, in the left margin, how he reads the small fragment, which is second in the Commandant's description. Lastly, besides these three bits we seem to have one at least figuring above the KNOC fragment in photograph 9c. It appears also in a sketch of the whole front of the monument given me by M.Louis Matru- chot, the editor of Pro Alesia. That sketch is here reproduced as photograph 9 e. The sketches by M. de Ricci and M. Matruchot were not intended for publication, but they are so instructive at different points that I have taken the liberty of putting them into a per- manent form. I have not seen any of the small fragments in question, * According to Stokes and Strachan's Thesaurus, II. 267, this is read Tolit in the Book of Armagh (fo. 13^ 2) by Gwynn, but they have printed ' Cuil Tolat '. In the Book of the Dun (fo. 52^) it is ' Cuil Tulad (? Talad'), and Hogan's Onoma- sticon has, besides the other forms, those of ^ C. Tlialaith ' and ' C. Toladh ', after the Four Masters who also have Cuil Toladh (a. m. 3303). The genuine forms were possibly Tolatj genitive Tol{a)it for early Tulanto-Sj gen. Tulantu THE CELTIC INSCRIPTIONS OF GAUL 49 and M. Matruchot states that he does not know what became of them. Even if we could handle the small pieces it would probably be hard to fit them into their places, but the two with letters belonging to two different lines can only go into the name beginning with T I C in the third line, and to the abbreviated word ending with B in the line underneath. Let us first take the reading approximately adopted by the Commandant and M. de Ricci of the bit ^ with three letters A A TICA r^ » or „ , and add it to TIC. We then have ^a . In the next BA BA B;^ place let us add on from photographs 9c,de, what M. Matruchot reads as a P with what I should treat as the lower portion of a B standing above it : the trial group would then look as follows : — TICABANNCa) B^P B: A This, however, admits of simplification, since the letter between B and P cannot be a consonant ; so A is eliminated in favour of A, and the result stands thus : — TICABANNOl) BAP B: It remains for us to fill the gap in the lower line. As what we are looking for is the dative plural of a word meaning offspring, children, the BAP at which we have automatically arrived reminds us of the Aryan root bher represented in Latin hyfero, ' I bear,' and fors, forte, ' chance,' English bear and birth, Irish beraim, ' I bear,' and breith, 'birth,' Welsh ci/mer, *take,' for com-ber: I refer to Stokes's articles on bero and its derivatives (Pick, ii, pp. 169, 170). One of them is in fact just what is wanted here, namely, brti-s, 'tragen, Geburt.' In Irish it yields breith, brith, the verbal noun corresponding to beraim,. ' I bear,' and meaning the act of bearing, birth ; in Welsh bryd in edfryd, edryd (for ate-brti-s), ' bearing back,' in the sense of * restor- ing', and cymryd (from com-brit), *the act of accepting, in the sense of bearing away with you, of taking to yourself.' The letters, there- fore, to be inserted seem to be T I , which receive some confirmation in the bit which shows a part of the perpendicular of such a letter as T or I : see M. de Ricci's sketch and photograph 9/*. We arrive thus at a form BAPTIB, and extend the abbreviation into BAPTIBOor BAPTIBOC, corresponding to what might be in Latin fortibiLS from fors, ' chance,' though differing widely in sense. The * The Commandant refers it to lines 2 and 8, but I take that to be a slip : it can only belong to lines 3 and 4. V G 4» 50 PROCEEDINGS OF THE BRITISH ACADEMY spelling ar representing a stage in the pronunciation of what philo- logists write r, cannot be reckoned a difficulty, as we are ignorant of the exact nature and history of the sound to which they apply that symbol suggested by Sanskrit analogy. With regard to the signification we may here mention, that in O.Norse this word was hurd-r and that it signified not only birth but also embryo and what is born, as does also the etymological equivalent, hirth^ in English: see the New English Dictionary, s.v. We are not without evidence of the analogous transition of sense from the abstract to the concrete in the case of the Celtic equivalents. Ed-fry d in its reduced form of ed-ryd (plural edrydau) meant not only the act of restoring, but also, in direct reference to bearing and birth, that which is born ; for it is found used by the poets to signify stock, family, or as Dr. Davies puts it, 'genus, prosapia, cognatio.' In short, whether we consider the form or the signification, we appear to be at liberty to assume the possibility of a dative plural hartiho{s) used in reference to offspring. (10) ATNOACO. This I take to be a contraction of an earlier are-2;o8ft), possibly are-avohoa^ the root being nod or snod, yielding in Welsh the word nod, 'the juice of a vegetable, the sap of a tree,' Irish snodhach, given in O'Reilly's Dictionary probably for a more cor- rect snodhach (with a short o), meaning ' sap or juice '. Modern Welsh has also a plural adnodau signifying ' resources ', suggested possibly by that word. Here we have ad representing the ancient prefix ate which has approximately the same meaning as Latin prefix re. But the meaning I should be inclined to ascribe to the ari;o8co of our inscription would be the adverbial one ' with the shedding of tears, with weeping, tearfully, mournfully '. In trying to get the readings suggested to fit to scale there are two or three things to be noticed. Between ^afMorakos and Avova>- TiKvos there was probably the colon stop: as there may have been also between 2eo-ta and KKafiaKL, whereas it seems to have been absent between BipaKOTcov and Tt . . . . It was probably also absent after Ko/3ptTovAcov, the other word ending with cav. On that sup- position it seems possible to get the required letters in, but even then there will be no room to spare. The first fragment in photographs 12 and 13 requires to be pushed up close to the large compound fragment on its right, so that the split letters, to wit, A in line S and B in line 4, may be closed as far as the damaged edges will help to show how they fitted. The YOYCO piece must not be pushed quite close to the large fragment: room must be left in the first line for : A, and further on for the missing half of the T of AvovcoriKroy, THE CELTIC INSCRIPTIONS OF GAUL 61 the whole of the I and the vertical portion of the K, the rest of which appears on the last or K N C fragment. The pieces should be brought together with due regard to these details in so far, at any rate, as the cementing does not force them out of scale. Then another good photograph should be taken of the whole, and copies of it should be distributed among the Celtic scholars most likely to make use of them. In this way we should soon have the uncertainties of the reading reduced within their narrowest limits. When I visited the temporary museum at Alise I had forgotten to procure permission to scrutinize the inscribed fragments : they were inside a wire en- closure, and I looked at them from the floor. Last April I sought the permission of Dr. Adrien Simon, president of the Semur Society, to examine the inscription more closely, and thanks to him I received all assistance possible, which I take this opportunity of cordially acknowledging. These notes on the inscription may be conveniently summarized here and preceded by the text which they appear to imply, as follows : — CAMOTAAOC : aYOYGJTiKNO C€Cla:KAAMAKt: TAPMA ^ AO) BIPAKOTOJY TICABANNCO KOBPITOYACOYBAPTtB : ATNO That is, freely rendered into English : — Samotalos son of Avvotis (and) Sesia Garma, daughter of Clamacios, tearfully (set up this monument) to their children Biracotus, Tisabannos, (and) Cobritulus. 4. Soon after my visit to Alise in 1909 Commandant Esperandieu and Dr. Epery favoured me with copies of a photograph showing a score of small objects which they had discovered on Mont-Auxois in the course of their diggings there. Among them are two pieces of lead with ^vriting of the graffito kind on them both, in the Greek alphabet : I had a look at them last April. The bigger one measures four centi- metres long by rather more than one centimetre broad and it reads KAPOMAPO. The end of the lead is very jagged, and there is nothing left to show whether or not the name had a final C. Elsewhere it occurs in Latin as CAROMARVS stamped alone on a vessel in the museum at Mannheim, but on others with F {ior fecit) appended, to wit, in the Mainz Museum : for further instances see C /. L., XIII. iii. No. 10010. 461 (p. 174). The name analyses itself 52 PROCEEDINGS OF THE BRITISH ACADEMY into Caro-maro-s, with caro- of the same origin as the Latin cdrus and the second part of Su-caru-s (ibid., 10010. 2408, p. 405), Welsh hy-gar^ 'dear, beloved,' O. Breton -ho-car. The other element is the familiar one of mdro-s, ' great.' Even so it is not quite certain what the compound exactly meant : it may have signified ' a great one who is beloved '. Possibly it meant simply ' greatly beloved \ 5. The other bit of lead is very small, and Dr. Epery suggested the reading DYIfPA, where he would regard the first character as part of an 0. We should accordingly have OYI, which reminds one of the dative termination of certain names of the o declension. I am not quite certain that what figures here as f is not a B imperfect at the bottom. That would recall the Bparovbc formula of the South. Were that reading certain we should be warranted in draAving some important conclusions from it ; but, as it is, we cannot do anything of the kind. Both bits of lead are represented in photographs 10, 11. We may guess that the inscription is part of a humble ex-voto from a temple sacred to the god Moritasgos, *:^ Two Latin inscriptions found by the same gentlemen, for copies of which I am indebted to Dr. Epery, equate the god with Apollo : the first, on a stone carved to represent a person's thigh and knee, runs as follows : see photograph 12, for which I have to thank Commandant Esperandieu : — AVG SAC DEO APOLLINI MORITASGO CATIANVS OXTAI Here the of DEO, of APOLLINI, the final of MORITASGO, and the initial of OXTAI are a modification of omega. See the second Genouilly epitaph, p. 55 below. Whether the same peculiarity is to be traced in the next ex-voto I cannot tell. Dr. Epery reads it as follows : — AVG SAC APOLLINI MORITASGO AVIVS • ALI D/OFANES ER . LIB • P- Perhaps Moritasgos was also named in a more fragmentary inscription, which the same gentleman had brought to light some time previously on the handle of a patera, reading as follows : — SAC DEO APPOL CVS POSVIT • ». » » Plate V THE MONT-AUXOIS lNSCRIPTIOr/S-SAMOTJ\'LOfS © :rAPMA \ c Photo. i)d V/v / r » T/^ V a\V m/ "'^ « Photo. 9e A n A Photo. 9 f On Two Pieces of Lead Photo. 10, 11 THE MORITASGOS EX-VOTO Photo, 12 •..• :'••; : V , \&\\t^ VII ON MOULDS IN THE MUSEUM AT jVIOUEINS' (ALLIM)'* • •*''• ^ Photo. 13 -^^IM rx M Photo. 14 . y fell r 0|i4^ .y .^^/'^^ O-U. •^t^J-d^'T- " ' ( '...'S^.X^^ Photo. 15 THE CELTIC INSCRIPTIONS OF GAUL 53 For this see Com. Esperandieu's Fouilles de la Croix Saint-Charles au Mont-Auxois : Premier Rapport (Dijon, Jobard, 1910), p. [277], See also Holder, s. v. Mori-tasgus, where another important inscription is given, the only one then known to mention that god. Moritasgos was evidently the Apollo of ancient Alesia, in the character of the repeller of diseases, as Caesar describes him. The tasgo-s of his name is to be compared with the Tasgia mentioned at p. 27 above ; but what are we to make of mori, which would seem to suggest the sea? It is likely to prove of the same origin with the moV' of Morrigain, which Stokes (Fick, 211) took to have meant ' elf- queen "*, and to have only been distorted by popular etymology into mor-rigain^ ' great queen \ The element mor- equates with mare in English nightmare, French cauchemxir, and German Mahr\ also with O. Bulgarian mora^ 'a witch.' In that case one might guess that Mori-tasgo-s meant a repeller or queller of elves and witches, that is to say, of the evil powers supposed to cause disease, blights, and baleful blasts, and to be routed, according to Irish story, by Lug : see my Notes on the Coligny Calendar, pp. 17, 36. In fact, Moritasgos may only have been a local name of the god Lug in the capacity of healer of sundry diseases of the human body. IV 1, 1. In August, 1909, 1 visited the museum at Bourges, and saw the inscribed stone from Genouilly (Cher), mentioned in my paper on the Celtic Inscriptions of France and Italy, p. 54 : see C /. L„ XIII. 1326. What we have on it consists, strictly speaking, of two inscriptions belonging presumably to persons of one and the same family. The stone is a slab 1"" 60 high by 0°* 52 or more in the widest part. Unfortunately a considerable piece of it is gone from the left top corner, carrying with it the beginning of the two first lines of the first epitaph, which reads as it stands : — OS . VIRIblOS TOC . OYIPIAAIO AN€OYNOC €nO€l Before TOC the Corpus has a short horizontal line, but I failed to notice anything before OS or TOC ; in fact, the left side of the first is gone in the first line ; and the French antiquary, M. de Lau- gardiere, to whom the Corpus refers the reader, seems to have found no more of this line; and both OS and TOC appear to have been 64 PROCEEDINGS OF THE BRITISH ACADEMY followed by a point or stop. A feature of the L in the first line has been overlooked in the Corpus : it has two horizontal bars at the bottom, so that it looks rather like an F upside down, that is to say, it stands for LL, a conclusion favoured by the next line with its A A. On the other hand, one would have expected OYI PI AAI to have had a C at the end : in any case the inscriber had left himself no room for it. Enough of the first two lines is left to dispose me to think that they consisted of the Celtic name of one and the same person, given in Latin letters and in Greek ones, both times in the nominative, and meaning ' So-and-so, son of Virillos \ Then comes, at a larger distance from the second line, the name of the man who erected or prepared the monument, and that not only in the Greek alphabet, but also in the Greek language. The spelling of the verb does not seem so much to argue carelessness in copying cttoUl, as the wish to give the phonetic spelling, cTroet, at a period when ei was already pronounced t : the result was, at any rate, to give us what appears to have been the correct Attic spelling of the word. The name Aveowos enters into the longer name Otuaneuni, genitive of Otuaneunos, which is found on the triumphal arch at Saintes (C /. I/., XIII. 1036). The former probably analyses itself into Av-€ovvo-Sy where av- is presumably the intensive particle : so the compound would mean ' very eovvos \ but what did eovvos mean ? Holder has collected words ending in -uno-, -una, but the u, in stems regularly formed, belongs to the stem, and this is the case here; we may therefore divide the word into Eu-no-s, which occurs as a man's name, Eunus, in an inscription to be seen at Bourges, already mentioned (p. 30 above), and has a derivative Eunius cited by Holder. Thus the stem would be eu, which I regard as standing for an earlier ehu from esu, that is to say from the name of the god Esu-s. In that case An-euno-s would mean ' partaking greatly of the nature of Esus, very like Esus\ Compare for the treatment of vowel-flanked s such parallels as Holder''s Esuggiu-s and Eu>giu-rix, where we have the same element E{s)ugiu- ; also Uciieti-s from Ud-guheti- = Ud-guseti' (p. 31 above, also Celt. Inscriptions, p. 7), and Suiorebe for ^Suihorebe = "^Suisorebe, ' to two sisters ' (ibid., p. 53). The disjointed syntax of this epitaph is emphasized by more space being left between lines 2 and 3 than between 1 and 2 or 3 and 4. In Latin epitaphs, moreover, we are usually informed as to the relationship between the person who has the monument erected and him whose memory it is to commemorate. Here we have no clue, but it is reasonable to suppose that Aneunos was of the same family as Virillos. Plate VIII THE GENOUILLY TOMBSTONES (CHER) Photo. 16 a, 16 /> Pfioto. 17 THE CELTIC INSCRIPTIONS OF GAUL 56 1, % Some eight inches lower on the stone, but on a higher layer of the surface, we have the other inscription, which reads as follows — the letters are Latin, but the language is Celtic, probably Gaulish : — ELVONTIV lEVRV - ANEVNn OCLICNnv LVGVRI r ANEVNICNn Here are several things to be noticed: for instance, I found what I took to be marks of interpunction : the one occurring after I E V R V is the only one given in the Corpus, but I thought I detected a similar one between the two words forming the third line, and another of a somewhat different shape at the end of that line. All these consist of diminutive triangles ; but there is a point in the Q. at the end of the second line which may also have been meant as a part of the interpunction, though it is more likely to have been merely ornamental. In that capacity the point occurs often enough : see p. 11 above, where C also was found provided with it. In the next place it is to be noticed that the Q. in question is not the ordinary Latin 0, but a form of the Greek omega resembling an on a horizontal line, li, which was doubtless derived from CI by making the straight lines continuous, a very natural simplification which is known to occur often enough in somewhat late Greek documents : so I learn from one of my Oxford colleagues, who is a distinguished Greek scholar. In the case of the omega at the end of OCLICNH, the horizontal line has been produced as a sort of tangent towards the left until it touches the bottom of the N preceding it. The one at the end of the last patronymic is more carelessly formed and looks almost like a A with its two sides consisting of arcs of a circle. The minuscule omega is used four times in the spelling of Celtic names in the Samotalos inscription at Alise, and in one instance it occurs for the long o ending the dative singular of a noun of the o declension, as in the three instances here in question : see p. 49 above. What is remarkable is that the Greek character should have been retained in epitaphs which are otherwise in Latin letters. It seems to prove two things, that the length of the final vowel of the dative was well marked in the Gaulish pronunciation of the time implied, and that the influence of the Greek alphabet among the Celts of Gaul had been by no means of brief duration. The name Elvontiu is obscure, but it is probably a nominative of the n-declension. The patronymic Oclicno-s is derived from Ocli-s, or else Oclo-s, just as Anevnicno-s is seen to be from Anevno-s, but neither Ocli-s nor Och-s is other than obscure. If I could feel sure m PROCEEDINGS OF THE BRITISH ACADEMY that I have read C instead of G I should compare the following inscription : — OGL • AVG • SAC ATEVRITVS SEPLAS • V • S • L • M. M. de Ricci found this on a small bronze pedestal supposed to come from the vicinity of Rheims : see the Revue Celtiqiie, xxx. 268, ^69, and plate. Luguri would seem to be a derivative from the name of the god Lugu-s, Irish Lug, genitive Logo : see my Notes on the Coligny Calendar, pp. 24-30 andi passim. Compare Rivos (an alias for Lugus) and Rivros, the name of the Rivos month, to wit, August, probably for Rivo-ro-s. In any case Luguri must have been the dative of "^Luguri-s of the i declension. This seems to be involved in the Irish tribe-name of Cinel Lugair, to which belonged the head poet or Jlle of Ireland in St. Patrick"'s time, Dubthach maccu Lugir, one of the Saint's most famous converts. The genitive occurs mostly as Lugair, but the Book of Armagh has also Lugir: see Stokes and Strachan's Thesaurus Palaeohibernicus, II, p. 267, where we have suhulcus Lugir, and p. 260, where we have Dubthoch maccu Lugil, with Lugil carelessly written for Lugir, and Dubthoch correctly given as a more ancient form than the more usual Dubthach. Witness the Latin genitive Dobituci and its Ogmic equivalent, Dovatuceas (of the i declension), at Clydai in Pembrokeshire ; in Welsh it is Dyfodwg. To return to Lugair, the editors of the Thesaurus give the nominative as Lugar: the nominative, however, seldom occurs, but Stokes's Gorman, May 11, has Lugair Lobor, 'Lugair the Infirm,' in the nominative, and the Book of Leinster (f. 315*^) has in the nominative injile Lugair, ' the^Z^ Lugair.' Nominative Lugair, with genitive Lugair, argues a stem of the I declension, inflected somewhat like the Irish feminine inis, ' island,' genitive inis. The name occurs also on a piece of black pottery at Saint-Germain, but unfortunately it breaks off short of the case ending, and only reads AoYroYP: see C.LL., XIIL iii. 10017. 77 (p. 489). One word remains to be mentioned, and that is lEVRV, which I equate approximately with the €TTO€l of the Greek portion of the other inscription. The rendering of the whole runs thus : ' Elvontiu made (it) for Aneunos son of Oclos (and) for Luguris son of Aneunos.' M. de Laugardiere's suggestive description of the stone is to be found (accompanied with a drawing) in the Bulletin archeo- logiqu£ du Comite des Travau^ historiques et scientifiques for 1894, pp. 127-37 (plate ix). I had not seen it when I visited Bourges, THE CELTIC INSCRIPTIONS OF GAUL 57 so I wish to indicate some of the diflPerences between his readings and what I have given above from the notes I made at the time. His account of the L in the first line runs thus : ' II est a remarquer que, dans Finscription, la haste de la lettre L parait etre un peu plus haute que le sommet des lettres voisines, ce qui me pent porter a penser que Fintention du graveur etait de lui donner une valeur double,' &c. He seems to have written that from an inexact recollection of the lettering. On the final of the second line he places the horns of a Y, which, according to my reading, would perhaps make OY, and imply a nominative in -ouj. In both lines I thought that I detected a stop. Lastly, he suggests restoring the first name in the first line as [OCL]OS,and the first in the second line as [AT€Z]TOC, which seems to imply that the inscriber cut Z instead of X. I may mention that M. de Laugardiere writes that ' les sont en general sensiblement plus petits que les autres caracteres ', which is probably true as to the actual height of the in these inscriptions, but their smallness here is not so conspicuous as to attract immediate attention, as frequently happens in other instances. Lastly, his paper would suggest, that some mishap had overtaken my note on the distances between the lines of the first of the inscriptions ; for I jotted down that the distance between lines 1 and 2 is only one-half or one-third of the distance between lines 2 and S ; but this is corroborated by M. de Laugardiere's own plate. 2. In the same paper M. de Laugardiere has described another inscribed stone found at Genouilly : it is of the same material, and he gives the dimensions as 1^ 29 by 0*^ 27 wide at the level of the writing, which consists of only a single word — RVONTV. It is possible that there was a letter or two preceding the R, but I could not find anything there ; and as to the T, that letter is peculiarly formed with the top stroke tilted up behind ; I was not able to make it out to be any other letter. Taking RVONTV as the entire inscription, I would suggest explaining the name as standing for an earlier o stem, Rugonto-, with the soft vowel-flanked g elided just as in the T 1 of the Coligny Calendar, for tigos, ' house, temple."* Com- pare the instances of elision of ^ collected by Holder, I. 1503, 1504, including such as ' deo Mounti ' for Mogonti, ' dis Mountibus ' for Mogontibus, 'deo Mouno"* for * ApoUini Mogouno\ not to mention Latio-ma(g)us, Ri{g)oma{g)us, and (Gregory of Tours') Mantolo- maus, Montalomago, and other instances of the same kind : see also p. 64 below. We have the element rz/g* in a name Caturugi (genitive of a Catu-rugip-s or Catu-rugo-s) in Latin letters on a stone at Merthyr Mynach in Carmarthenshire. We may possibly have the shorter 58 PKOCEEDINGS OF THE BRITISH ACADEMY forms in R^iga and Rugus cited by Holder. The Neo-Celtic words related are Welsh rhu-o, ' to roar, also to bluster,' Irish rucht, ruchd, ' a great shout, clamour,' mcht mikd, ' a soldier's cry,' used of a far- reaching shout for help uttered by Cuchulainn when overpowered by his foes, also of the sound of the sword in use (Windisch's Tain B6 Cualnge, 2960, 5583), Manx Gaelic roogJi, ' to bellow, to roar.' The Welsh rhu-o and kindred forms would seem to indicate that rug- was rvg^ which rather militates against the conjecture that the Welsh word is a loan from Latin rug'tre, ' to roar,' for the u of which compare €pvyriy ' a bellowing ' (Hesychius) and the Homeric kpvyovra, * bellow- ing,' together with other cognates : see Walde. The meaning of Catu-rug-i would seem to have been ' one who is loud in the conflict, a battle roarer,' and that of Ruontos cannot have been very different, Ruonto-s is a species of participial formation to be compared with "^Anaganto-s (in Irish Anagat), Caranto-s, Decantae, and the like (Notes on the Coligny Calendar, pp. 14 n., 60 n.), but it shows the same conjugational vowel o as Nodons, dative Nodont-i, in the name of the god Latinized Nodens, genitive Nodentis (ibid., p. 50, Celtic Folklore, p. 446). Ricontu, with its final w, is the dative, and this brief epitaph means ' To Ruontos, or to the memory of Ruontos '. When I visited Bourges I was met by M. P. Gauchery, engineer and architect from Vierzon (Cher), who kindly vouchsafed me information as to the bibliography of the inscriptions found at Genouilly, and notes of his own as to the locality of the finds. The latter were of such interest that I asked him to be so good as to send them to me in writing at his leisure : he did so in 1909, and I venture to have them printed here : — ' La commune ou ont ete decouvertes ces steles se nomme Genouilly ; elle est a 15 kilometres a I'ouest de Vierzon. Le mouton, motton (petite motte) est une petite eminence a Touest de Genouilly. C'est sur les flancs de cette motte qu'on a decouvert ces steles — au sommet de cette eminence deserte se trouve une sorte de fortin. C'est un ouvrage en terre sans aucune ma9onnerie. II est rectangu- laire et sensiblement deprime a son sommet. Les fosses sont a sec, ils ont 8" 00 de large. La hauteur verticale du parapet au-dessus du fond du fosse atteint a peu pres 4"" 00 ; la depression interieure est de 13 metres carres. Le circuit exte'rieur est de 38"^ sur 32*". Comme il n'y a pas de vestiges de ma^onnerie (quoique la pierre ne manque pas en cet endroit), on peut supposer que le fortin etait en bois comme etaient beaucoup d'ouvrages anciens dans nos contrees, notamment les mottes normandes qui ont e'te reconnues sur les bords du Cher. Mais ici le mouton, qui est une eminence naturelle au-dessus de la plaine, est distant de 7 kilometres au sud de la riviere du Cher. Le sous-sol de cette eminence est constitud par des gres cre'tace's de I'etage cenomanien. C'est une sorte de gres a paves qui se divise en tables d'epaisseurs variables, les steles en proviennent. C'est en voulant utiliser ces pierres pour THE CELTIC INSCRIPTIONS OF GAUL 59 rempierrement des routes que les ouvriers ont cru remarquer des inscriptions sur ces pierres brutes. Au dedans, comme au dehors du fortin, j'ai rencontre' une grande quantite d'eclats de silex et des nucleus de silex d'ou ces e'clats etaient tires. Dans cette contree on ne rencontre pas de silex en place. Ceci fait supposer que les assie'- geants et les assiege's se servaient de pointes de fleches en silex : la presence de nucleus prouve qu'on les fabriquait hativement sur place. J'ai cru reconnaitre aussi, pres de I'endroit ou se trouvaient les steles, des poteries gauloises ; ce sont des poteries grossieres en fragments indeterminables. Aussi je n'affirme rien quant a la provenance.' I owe the photographs 16, 17, to the kindness of M. Gauchery and of M. de Goy, of the Societe des Antiquaires du Centre a Bourges.^ Before proceeding further, I append a few notes and corrections relating to the inscriptions at the Hotel de Cluny in Paris and the Chateau de Saint-Germain. The references are to my previous paper. P. 46, a little of the top corner of the P of PVBLICE is still visible. P. 47, a bit of the final I of VSEILONI can be detected, but the N suggested will not fill the gap : possibly N N is required. P. 49. The scratch before E S V S was possibly a clumsy attempt to convert the latter into lESVS. P. 50. The final S of [C] E R N V N N S is very doubtful : I can find no certain trace of it. As to SMERT[VLL]0[S], I thought I detected a little of the right side of the 0, but if that should prove correct, there would be no room left for a final S. P. 51. The T of FORT[ V N A] is only partly there. The altar face to the left of Fortuna shows Cernunnos with a female figure to his right. On the face to Fortuna's right there is another group of two figures, but in both instances the inscriptions are gone. ^ The photographs have come to hand late and I find that the photographer, anxious about the success of his work, has baffled the three of us. He seems to have traced the letters on the stones with something black before attempt- ing to photograph them. He has for example made €170 € I into CNOei with quite a neat N, which nobody had seen before, and he appears to have mended other letters including the omegas. But if anybody familiar with photo- graphs scrutinizes them, he will detect the top of the Greek TT intact, and he will fix on at least one H, as a friend of mine has done. The photographs show the relative positions of the writing, but it is useless to consult them for the peculiarities of individual letters. 60 PROCEEDINGS OF THE BRITISH ACADEMY P. 52. In the Bratronos inscription the reading RICI is right, the G has its lower part continued some distance horizontally and on that stands an I, the top of which almost touches the top end of the G. The of LEVCVlLo has a tangent ("D) which reaches the tops of the JL. Compare the case of the last 0(=(0)ofOCLICNO with the tangent underneath (p. 55 above). This is not all, for we have here two instances of omega: I find that I noticed them in 1905, but thought that the variation was not deserving of mention. My note in reference to NANTDNICN runs thus — 'The here and in SVIO is circular except at the bottom, which is a straight line.' I have only come across it by accident after studying the Genouilly inscriptions. From this inscription one learns that the oof NANTON- was long, and in the case of SVIQREBE there is no difficulty in proving it long. Witness the Welsh chwior-yd, 'sisters,' Latin soror-es, and Welsh cy-chwiawr^ which Silvan Evans renders ' coequal ; even, like ; participant \ but in the line from the poet L. G. Cothi, which he quotes — * Henri a Siaspar gy chwiawr', it admits of being rendered ' brothers \ For a remarkable parallel in point of meaning compare the German ge-schwister, ' brothers, brother and sister.' P. 56. In the inscription on the Celtic Mercury's shoulders in the court of the Chateau de Saint-Germain I formerly read SO SI in the second line and suggested S S I N . I have looked at it several times since and I thought that I got so far as to trace the N of SOSI N. When I was there last April I had the invaluable assistance of M. Camille Jullian, who thought he could detect the beginning of the N, but that letter, I must confess, is not certain. We came to the conclusion that the third and last line ended with RO, which I had previously thought to be PO. The letters immediately preceding seemed to be MA, though I had some difficulty in tracing them. The three letters preceding the M seemed to be ESO, and preceding these near the edge M. Jullian thought that a depression, which I did not regard as carried through, was the top portion of a C. He thought also that he could detect the return end of that letter behind the foot of the E. On a previous visit I examined these doubtful traces with M. Reinach, who had a good cast made of the inscription, and we failed to read them into a C or G. M. JuUian's reading would make the whole into CESOMARO, better GESOM ARO, for an older spelling of GAISOMARO, nominative Gaiso-maro-s^ ' the man of the great gaesum or spear,' or perhaps ' great in using the spear '. Even now, however, I am not quite certain that it is not the name of Esus which forms what is practically the first element in the compound : the latter in that case might be interpreted as meaning ' great like Esus, or with. THE CELTIC INSCRIPTIONS OF GAUL 61 Esus \ Our reading of the inscription as a whole may be represented thus : — APRONIOS lEVRV . SOSIN (G)ESOMARO It would have to be rendered : ' Apronios made this for (G)esomaros.'' ^^^ Here we may briefly discuss an inscription in Latin containing a name Esumopas into which that of Esus seems really to enter. It is a bronze bust at Saint-Germain, with the inscription : — ESVMOPAS ^ CNVSTICVS V S L M It appears to have been discovered in 1830 in the course of excava- tions made in the forest of Beaumont-le-Roger, near Evreux. The foundations were then discovered of some four dwellings and a small rustic temple to which the bust probably belonged. A short and systematic account of the whole find was published by M. Salomon Reinach in the Reviie Celtique, xv. 413-17, with a plate ; also in his CuUes, Mythes et Religion^ i. 253. The interpretation offers very considerable difficulty. The abbreviation means votum solvit lihens merito ; but who paid his or her vow ? Whose was the name, and whose the bust.^ The obvious conclusion would seem to be that Esumopas was the person who paid his vow. In that case the name of the divinity is not given, but that was sufficiently fixed by the fact that the vow was paid in a certain temple, which was well known to the worshippers as belonging to such and such a divinity. Perhaps the bust was that of the divinity, and perhaps it was meant to be the visible expression of the identity of that divinity. In any case we could hardly expect the bust to have been that of the donor. On the other hand, we should hardly expect the donor to have con- cealed his name. The epithet or surname was doubtless a Celtic adjective cnHstico-s, derived from a stem represented in Irish by cnuas^ ' a collection ; treasure ; recollection, reflection,' and by cnuasaim, ' I collect, gather, glean,' and cnuasaire^ ' a collector, a gatherer ' (Dinneen). The surname may have meant 'fond of gathering, habitually collecting', unless we render it still more simply as treasurer or collector. Assuming then that Esumopas was the donor's name, and that the first part of it was the name of the god Esus, we have mopas left to be considered. In the first place, the first vowel may be o or a. Witness the Latin genitive Agedomapatis (Acedoma- patls) cited by M. Reinach, also Agedomopatis (compare Agedillus, 62 PROCEEDINGS OF THE BRITISH ACADEMY Agedovinus) and others given by Holder. The form with a, mapas, for mapat-s, is doubtless of the same origin as Gaulish mapo-s, ' son ' (in Maponos), Welsh map, mab, ' boy, youth, son,' and mopas for mopat-s cannot be severed from it. Now if we substitute the consonants of Goidelic or a language resembling Goidelic, we have a stem maquat-^ which according to the phonology of Goidelic must eventually reduce qu to c. This latter, subject to lenition, would become ch ; and subjected to further lenition it would result in gh, written g. Similarly the t would become successively th and dh, written d. Now this, with the vowel o preferred, would give us exactly the stem and declension of the Medieval Irish word mog, genitive mogad, dative mogaid, which is explained as meaning a servant or a slave : see Stokes's Three Middle-Irish Homilies, p. 70, Windisch's Irische Texte, p. 694. But there was an Irish word liable to be confounded with it, and that was mug, genitive mogo, moga, 'a slave, a servant': see Stokes's Oengus, pp. 4, 5, 347, 348, where we have De mog, corresponding to the Welsh meu-dwy, ' a hermit or Culdee,' literally servus Dei : see p. 3 above. Here we have the etymological equivalent of the Gothic magus^ 'a boy, a child, a slave.' Whatever the exact signification of mug, genitive mx)go, may have been originally, that of mog, genitive mogad, was probably not slave or servant, but lad or young man ; for words starting with this sort of meaning frequently acquire that of service. Take, for instance, such a word as Welsh gwas, 'a young man,' which is now mostly used in the sense of 'a servant man'; not to mention the English and French page derived from the Greek naiUov, ' a boy, a slave boy.' Keeping in view the idea of age as originally underlying mogad, and giving the preference this time to the vowel a, we can account for Irish maccdacht in the term ingeii maccdacht, 'a grown-up girl.' The latter word was borrowed into the Brythonic dialects when the Irish pronunciation was machathecht or machadecht, and it makes in Medieval Welsh machteith, ' a damsel, a young woman.' Allowing here for the influence of the word mace, ' boy, son,' and for the preference for a, we see that maccd- is a short equivalent for mogad- = mochath- from macat- for maquat-. This last with o in its turn making moqiuit-, would be the early equivalent of the Gaulish mopat- of the name ' M. d'Arbois de Jubainville in his Premiers Habitants de V Europe, II. 344, suggests that this was one of the Celtic words borrowed early by the Teutonic nations ; another according to him (p. 343) was the adjective free for early frijo-s from a Celtic priio-s, which appears in Welsh quite regularly as rhyd, ' free.' Hence it would follow that Celtic at the time of the Teutonic borrowing had not done away with Aryan p, and that the borrowers had not done provect- ing p into/. ». • e • a * :v^::?^;' ^ Plate IX :/; C'^A FC^TSHERD FROM LEZOUX (PUY-DE-DOME) Photo. 18 a A POTSHERD FROM ANNECY (AIN) Photo. 18 6 THE CELTIC INSCRIPTIONS OF GAUL 63 Mopates, applied by a Nervian citizen to his own Mother Goddesses : see Holder. From this use of such a name one may infer that they were regarded as being in the bloom of life. VI 1. In the thirteenth volume of the Berlin Corpus, part iii, p. 459 (No. 10012. 19), the editor, Dr. Hirschfeld, has brought together under the heading Vasa Gallica Oriiata two inscribed fragments of two earthenware vessels. One of them is in the Plicque Collection in the French National Museum in the Chateau de Saint-Germain, and comes from Lezoux in Auvergne. The other is in the museum at Annecy (Ain). Thanks to the courtesy of M. Salomon Reinach, I have before me a cast of each in plaster of Paris. The fragments do not belong to the same vessel, but they give portions of the same writing. That writing was made with a stilum on a mould, and then stamped on the vessels while they were still soft. So the inscription on the vessels stands in relief and reads from right to left. The Lezoux fragment has its surface well preserved for the most part, while the Annecy one had its lettering a good deal levelled and flattened by a clumsy touch of the potter"*s hand. In one direction the fragments overlap, so that it is possible to combine them into one, giving nearly the whole of the inscription. The writing is of the graffito kind and runs thus, as will be seen in photographs 18 a and 18 6, which represent the casts and should be scrutinized at every point : — (1) CALIA . VII . .. (2) BIVSA^NITI (3H OBIIRTII . M (4) OVNO » (5) CALIINI (6) OFICINA Line 1. After the V II near the end one detects the bottom of the perpendicular of an fJ and there is room for a vowel or I . Line 2. The only thing to remark is that Dr. Hirschfeld and M. Plicque read the fifth character as a conjoint AN, which is probably right; it would be possible perhaps to read Tl instead of AN. Line 3. M. Plicque read the fii-st character as X, and Dr. Hirschfeld as I or X (queried). To my thinking X is out of the question, and the letter meant is either I or a form of €, to wit,-(^: I give the latter 64 PROCEEDINGS OF THE BRITISH ACADEMY the preference, though the vowel I would do, as will be shown later. In support of ^ in the same volume, one may turn to p. 479 (No. 5) and p. 480 (No. 8), in each of which several instructive instances of this kind of € will be found to occur. Line 4. At the end of this line in the Annecy fragment is some- thing that was evidently meant to mark the end of the first part of the inscription. Dr. Hirschfeld suggests an ivy leaf. Line 5 appears to begin with a C differing at the top from the C in lines 1 and 6. Only the top of it appears in the Lezoux fragment, while two bits of it may be detected in the Annecy one, namely, the tip of the top of it just over a letter A and the lower tip joining the lower end of the right limb of that A. That vowel has a trace of the underneath stroke shown in the final A of Calia in the first line. Then follows an I of the type k and after that a = II, that is E. In all these the strokes look very clumsy and broad, the ridges of all having been accidentally pressed flat in the making. The line is completed by the addition of Nl, but these two letters, unlike the others, are scratched in the surface of the vessel to complete the lettering wiped off, and the two previous ones have their outlines traced in the same way. I ought to say that this line of letters as given in the Corpus differs considerably from what I seem to read on the cast of the Annecy fragment. Line 6. Here the paste which stood out forming the first I of oficina is gone, but the area covered by it is distinctly traceable. The N is like that of N I T I . The A is crowded into a corner, and only two out of the three strokes of an A are present : they have to be read conjointly with the N preceding. So far these notes have mainly had to do with the writing and individual letters. The next thing is to complete the separation of the letters into words, which I do as follows : Calia • Veni\bitis Anniti | e oberte * M ouno c^ \ Caleni \ Oficina !. The first word to notice is Calia, which may be masculine of the a declension, or else entitled to a final s. The latter would ety- mologically represent Caliat-s and make as its genitive Caliat-os, nom. plural Caliat-es (like Mopates, which I have just mentioned), meaning the individuals of a Calian gens, clan or family, or the inhabitants of a place known by some such name as Calion. In either case we may accordingly render Calia{s) as ' Calian ', and treat Venobius as his name. Other names like Calia, treated as Calian, occur in the same volume of the Corpus, witness the follow- ing: REXTVGENOS SVLLIAS AVVOT ^ Rechtugen the SuUian, Proprietor', C.I.L. XIIL iii. p. 474 (No. 85); AONNIAC, p. 489 THE CELTIC INSCRIPTIONS OF GAUL 65 (No. 76); MATI€PIAC, p. 489 (No. 78), which reminds me of an Ogam inscription in the neighbourhood of Lismore, in Co. Waterford. The probable reading is Ercagni Maqi Miteres or Mitereas, where the last word is presumably a genitive of a name Miteri-s. Whatever the reading of the last letters may prove to be, the name seems an early form of J/if/ir (usually indeclinable). The bearer of the latter used to be treated as one of the chiefs of the Fairies of Ireland. Possibly it was originally an ancient divinity of the race represented by them in so far as they may be regarded as historical, but compare Mitis (p. 9 above). Another name of this class is Nettas, which will be mentioned presently. For more instances see Holder, s.v. -ati-s. The next word to demand attention is Venobius^ which coupled with qficina would seem at first sight to prove the inscription to be in Latin ; but these cannot be compared for weight with the Celtic tense form oherte. The presence of qficina can only show that the Latin word in the form qficina was borrowed early by the Celts of Gaul : in French it was reduced to uisine, now usine, ' a workshop.' It might be conceded that what follows the nondescript stop, namely, Caleni Oficina, was meant as Latin, but I see no evidence that the potter was conscious of it. Even had he been so, it does not involve our having to regard the previous portion of the inscription as Latin : we should merely have to regard the whole as bilingual. Venobitts we may state with confidence to have been a Celtic name which took that form as a contraction of Veno-bivos, which would be in Irish Fian-beo, Fian-bheo, meaning the 'quick or alert warrior '. The first element occurs in other names, such as possibly the following cited by Holder, Veni-carus, Veni-mancs, and Veni-touta : compare the Lepontian feminine Venia Metelicna discussed by Danielsson, loc. cit., p. 18. The other element is in Irish beo, ' quick, alive,' Welsh h/w of the same meaning. It occurs in such Irish names as Beo-aed^ Buad-beo (early genitive Bodi-bev-e, in Ogam), Chth-beo, Find-beo, Beoc, Beoan, in Old Welsh Biuan, Biuuan, Biuguan, and contracted Bican, as an ordinary adjective bnun, ' quick, swift.' It is due to their translating Celtic names that the Dessi had such Roman ones as Vitalis, VitalianuSy and Vitalinm : see the Cymmrodor, XXI. 48-50. But to return to the potsherds, it is merely an accident that we have here a name Venobius which looks like Latin. It does not prove that the inscription is Latin; and it is probably to be explained as a contraction of Venobiuos into Venobius of the u declension. Anniti seems to be the genitive of Annitios or Annitos, and this can be referred to a group containing such instances as Annios, Anniccus^ Annicco (fem.) cited by Holder. Our genitive is followed by e (or i) V g5 66 PROCEEDINGS OF THE BRITISH ACADEMY and the sequence is exactly parallel to Latin Marcif., ' son of Marcus.' The potter has been careless about the punctuation— he has only two points in the whole inscription plus the leaf or nondescript stop: to be consistent he ought to have inserted either more or none at all. I venture to treat the present case as if it had been Anniti f., and to translate it ' son of Annitios \ The inscriber might have left the genitive standing alone or used some such patronymic as we have had in ClXovkvos: he was tempted by the Latin formula to suggest one of the vernacular words for ' son ', indicated here by the abbreviation e (or i), which would have been y* or y*. if he had been writing Latin. The question now arises, what that word was. The answer is given by a bilingual inscription at Eglwys Cymun in Car- marthenshire : see the Archoeohgia Cambrensis for 1893, p. 285 (also for 1889, p. 6). The epitaph is probably of the fifth or sixth century, and there the Goidelic equivalent for Latin ^Zm is given in Ogam as inigena, the spelling in old manuscript Irish was ingen, now inghen, * daughter, girl.' The corresponding word for ' son, boy ', would doubtless at one time have been inigenos. Both probably meant a child that was in-born, that is to say, born in the house. Compare the Latin ingenuics, ingenua, 'a free-born man or woman.' Both of these occur also as cognomens in Roman nomenclature, and we can match them in the case of ini-geno-s, but in the slightly different spell- ing eni-genO'S, as in an inscription from the neighbourhood of Grasse (Alpes-Maritimes), which reads SECVNduS ENIGENI F(ilius) : see Holder, s.v. Eni-g{e)nus. His instances show that the second e was liable to be dropped, leaving the name shortened into Enignos, geni- tive Dnigni, With regard to the other e we may mention that the Celts of Gaul seem to have treated the prefix not as ini but eni^ which recalls the Greek hi, ' in.' Among other instances cited by Holder are the following : Eni-boudiiis, Eni-cenius, Eni-cus, Efii-ca, fundus Eni- anus. This suggests a personal name Enianos which is in Welsh Emion and appears in the Marches in the English garb of Onions, while Ab Ein(i)on yields Beinon or Beynon. One is tempted to ask, whence came John Bunyan's surname ? The next word to require notice is OB II RT II, which is the verb of the sentence. It doubtless meant ' gave ' or ' offered ', and stood for od'ber-te, from ber, ' to bear or carry, ferre,'' with the preposition ud, ' out.' But the analogy of Ucuetis, from Ud-gtts-etis, would lead one to expect not oberte but operte, which was possibly one of the ways of spelling the word. The reason for hesitation may have been that the consonant resulting from the assimilation of db was not exactly either p or b but an explodent of intermediate force between those THE CELTIC INSCRIPTIONS OF GAUL 67 two labials. At any rate one finds the same uncertainty in the ortho- graphy of the kindred words in Irish. I select some instances from the Grammatica Celtica}^^ where we have two groups involving od-ber. (a) One of them prefixes the preposition ad (ed, id, and iodh), 'to,' as follows : ad-opuir, ' ofFert ' (885*), ad-opart, ' tradidit ' (455*), ad- ohartat, ' obtulerant ' (885^), ad-oparar and ad-obarar, * offertur ' (471^, 885^), ad-oparthe, ' ofFerebatur ' (480^). Some of the instances have the particle ro inserted, such as ad-r' -opart, ' obtulif (885*), atropert = at-ro-qpert, ' obtulit ei ' (885*). To the foregoing may be added the forms of the verbal noun ed-part (shortened from ad-opart), as follows : nominative singular feminine adbart, edpart, idpart, ' oblatio ' (5*, 59*), genitive idbairte, *oblationis' (242^), dative edpairt (5*, 59*), accusative adbirt, audbirt, idbirt, 'oblationem' (7^), genitive plural idbart, ' oblationum "* (245^), dative edpartaib, ' oblationibus ' (5*, 59*), accusative idbarta, ' dona ' (246*). The modern spelling is iodhbairt, which is treated as the nominative, though etymologically it represents the accusative ; that is, the accusative is used as both accusative and nominative, which happens often to feminine nouns in Modern Irish. (b) The other group has the preposition di (do), ' from,' prefixed to od-ber, as in the following instances, likewise from the Grammatica Celtica: do-opir, 'privat' (430*, 885^), di-oiprid, 'fraudatis' (885^ 994*), na ti-ubrad, 'ne privet ' (365^, 885^), di-uparthe, 'privatam' (885^), di-upart, ' decrease ' (885^), oc di-upirt or oc di-nheirt (484^, 885^). Now such a form as the Irish absolute preterite birt, * tulit,' is sup- posed to be derived from ber-ti (Thurneysen, p. 392), and on the other hand the bert of such a compound as du-bert, ' dedit,' and the od-ber-t of at-ro-opert harmonize exactly with our ober-te with -te rather than -ti. Compare also Welsh cym£rth (from com-ber-t), ' took,' differth, 'defended' (also cymyrth and diffyrth)-, and gwan-t, 'wounded,' can-t, 'sang,' which seem to be derived from uan-te and can-te. So we may perhaps regard -te as one of the original termina- tions of the forms in question in Irish and Welsh, just as it was in our Continental oberte. Mouno is the dative of Mounos, the name of a god, given also in the dative, as DEO MOV NO in an inscription from Risingham in North- umberland (C /. L., VII. 997). Practically the same name occurs in the longer form Mogovnos, as in an ex-voto found at Horburg in Alsace, reading Apollini Granno Mogovno (C. /. L., XIII. 5315) : other spellings are Magounus, Magonus : see Holder, s. v. A related form Mogon-s yields the dative Mog{on)ti in an inscription from Netherby in Cumberland (C /. L., VII. 320). From Mogont- comes 68 PROCEEDINGS OF THE BRITISH ACADEMY a shortened form Mount- as in DIS MOVNTIBVS in an inscription from High Rochester, now in the museum at Newcastle (C /. L., VII. 1036, cf. 321). In this connexion may be mentioned the statement that one of St. Patrick's names was Magoniits, that is, Magonius^ which by dropping the vowel-flanked g became in Welsh Maun: see Mommsen's 'Historia Brittonum cum Additamentis Nennii' in the Chronica Minora, III. 195 (§ 51). An ancient instance of g being made a spirant and reaching the vanishing point occurs in the Coligny Calendar in the word for house or temple, namely, tio for tigoSj whence the woman's name Tioccia Peregrima (C. /. //., XII. 3897), which may be compared with the Latin Domestica and Domesticv^. For other instances see Holder, I. 1503. There remains to be mentioned the genitive Caleni, of which I have but little to say, except that the reading is not very certain, but Caleni, if approximately correct, may be regarded as belonging to the same family as Calia. In any case Calenos, latinized Calenus, is a name known in the ceramics of the Continental Celts : witness the Vascula Gallica in the thirteenth volume of the Corpus, part iii, p. 167 (No. 10010. 404), where we have as one of the *tituli signa- culo impressi ' the following : C A LE N VS from Saignes (Cantal), in the museum at Limoges ; CALENVS^ from Trion, in the Guimet Museum in Paris, a stamp which I have seen ; CALENVSor F(ecit)from Tongres, in the Brussels Museum; also one said to read GALENVS