Published for the British Academy By Henry Frowde, Oxford University Pressnbsp;Amen Corner, E.C,

155

A. G. van HAMEL

THE CELTIC INSCRIPTIONS OF FRANCE AND ITALY

Last year I devoted the whole of the month of September and a part of October to the examination of the Celtic inscriptionsnbsp;known to exist in France. Last Easter vacation I took the opportunitynbsp;of doing the same with the few which Italy supplies; and on bothnbsp;rambles I had the assistance of Mrs. Rhys. A few of the inscriptionsnbsp;which we ought to have seen in France are not to be found, and somenbsp;there are which we have not yet tried to see. Add to this that whennbsp;the Berlin Corpus Inscriptionum Latinarum has been published fornbsp;the whole of France, the whereabouts of a few more will probably benbsp;made known to the public. I foresee, therefore, that to make thenbsp;list complete we shall have to repeat the pleasure of visiting France;nbsp;but in the meantime I venture to offer the list as it stands to thenbsp;Academy without further delay.

On the other hand, certain inscriptions which have sometimes passed for Celtic are here omitted because I do not think that they are such.nbsp;Two of them belong to Italy: the first is said by Dr. Stokes to be onnbsp;a metal plate found near Verona. I refer to his work on Celticnbsp;Declension, where it forms his No. 4, being read by him as follows :nbsp;Qaninio Qikoremies hisa quasova khik Vcpisones, while Pauli in bisnbsp;Inschriften nordetnislcischen Alphabets, p. 19, gives it, somewhatnbsp;differently, as lt;paniniu4gt;ikuremieshii�4gt;asu'vakhikvipisines. I have notnbsp;seen it, and I am not certain where it is; but so far I have had nonbsp;success in treating it as Celtic. The same may be said of the Estenbsp;inscription, Dr. Stokes�s No. 5, which he reads Tarknovosseno, andnbsp;Pauli, p. 22, as tu ' r � knavas � seno. The piece of pottery bearingnbsp;this is said to be no longer at Catajo where Mommsen found it, andnbsp;I have not succeeded in finding what has become of it. Thus thenbsp;Italian inscriptions which Dr. Stokes made into five are reduced tonbsp;three, and I am not quite sure as to one of those three. Lastly, to comenbsp;back to France, I have been obliged to omit the so-called �Gaulishnbsp;inscription of Poitiers.� It is on a small plate of silver which is now

at the Chateau de St.-Germain-en-Laye and as the readings given are frequently inaccurate, I submit the following as the best I could makenbsp;of it, except the division into words which is mechanically my doingnbsp;rather than the suggestion of the inscriber:�

bis gontavrion analabis bis gontaurio sv ceanalabis bis gontavrios catalagesnbsp;uim canima uim spaternam astanbsp;magi ars sec[? �ei]uta te iustina quemnbsp;peperit sarra.

i. Evrkux. The Museum at Evreux has an inscribed fragment of a table in bronze found in the excavations made at the place knownnbsp;as Vieil �vreux in the neighbourhood: see Stokes�s Celtic Declension,nbsp;No. 21, and the Berlin Corpus Inscriptionum Latinarum, vol. XIII.nbsp;No. 3204. The inscription now consists of portions of seven lines, allnbsp;of which are incomplete at the beginning and several also at the end,nbsp;possibly all. I read the letters remaining, or partially remaining, asnbsp;follows:�

(1)

(2)

(3)

(4)

(5)

(6) 0)

The punctuation here seems to be of two kinds, the one stop being like a long � on a small scale and the other, after RAMEDON andnbsp;FILIA, more like a r or the abbreviation in Latin MSS. for et; fornbsp;I do not suppose it is to be treated here as et, partly because I shouldnbsp;not expect it so early and partly because it seems too insignificant innbsp;point of size to be a part of the ordinary reading; but instances of thisnbsp;kind of stop will be found elsewhere as, for example, in C. 1. L., XH.nbsp;2091, 3693. The first line begins with a portion of a letter whichnbsp;looks like the right-hand side of the lower half of an S ; but so far asnbsp;its form goes it might equally well be the corresponding bit of a B :nbsp;the latter letter is in other respects far less probable than an S as thenbsp;final of the word to which it belonged. The line ends with whatnbsp;appears to have been an I, but as the fracture occurs along thenbsp;perpendicular of the letter a D would fit equally, perhaps P, B, R, ornbsp;E: the appearance of the bronze is not decisive on the point.nbsp;Similarly the second line begins with an imperfect letter, the bronze

having broken off along the perpendicular of the R, so it is hard to say whether RAMEDON is part of a word or the whole. Afternbsp;it comes a wide space in which there is nothing but a small stop r.nbsp;Line 3 consists of what appears to be three whole and separate wordsnbsp;followed by the stop /. Line 4 begins with an imperfect D followednbsp;by 0, but there is nothing to suggest that the whole word was DOnbsp;rather than the end of a longer word. Then comes a space followednbsp;by CARA6IT0NV, the V of which comes so close to the fracture thatnbsp;one cannot say whether the word as we have it is complete or not.nbsp;Line 5 begins with a portion of an N which is followed by a space,nbsp;after which IA come followed by a lesser space. Then we have whatnbsp;I read SELANI followed by SEBODDV with the / stop. From thenbsp;spaces I should infer that the first word ended with N and that thenbsp;next may have been lASELANI. Line 6 begins with a wide spacenbsp;followed by R E MI FILIA with the r stop, Line 7 begins withnbsp;DR VTA Cl SAC I Cl VIS SV. There is nothing to show that thenbsp;entire first name was DRVTA, since a part of the name may havenbsp;been cut off preceding the D. After SV there remains a little of thenbsp;left top corner of another letter which may have been E.

The names in this fragment seem to have been Celtic, and CRISPOS and BOVD .. or BOVI . . have been mentioned in my last paper, Celtaenbsp;and Gain, p. 49 The Celtic portion of the inscription would seemnbsp;to have ended in line 5: the remaining two lines appear to havenbsp;been in Latin. The form SEBOBBV- looks like a lisping of Sebossu-,nbsp;of the same origin as the attested name Sehosus and that of the Alanbsp;Sebosiana or Sebmsiana as in Eq (uites) alae Sebussia{nae) in an inscription found at Lancaster (C. I. L., VII. 287). As we do not knownbsp;what followed SEBOBBV we cannot treat it as a feminine referring tonbsp;Remi Jilia, so it remains perhaps to regard it as the dative masculine ofnbsp;a name Seboddos to be identified with Sebosus supposing that to standnbsp;for Sebossos. REMI would be the genitive singular of the namenbsp;which was so well known in the plural as that of the Belgic people ofnbsp;the Remi. We have in C. I. L., XIII. 3197 deo Gisaco (also foundnbsp;at Vieil �vreux), but the editor thinks that here perhaps Gisac-i wasnbsp;the name of a place, and Holder in his AUceltischer Sprachschatznbsp;treats it in the same way and mentions another Gisacum, called innbsp;French Giso^-la-Coudre in the same department (Eure). But if onenbsp;is to read .. DRVTA GISACI CIVIS SVESSIONIS, as Holder suggests,nbsp;it seems more natural to treat Gi^ac-i as the genitive here of the name

* To the instances of Welsh sp � squ add that of Welsh yspydaden, � hawthorn,� Irish sc�, genitive sciad, both of which Dr. Whitley Stokes derives from a stemnbsp;tkvijdt.

4 PROCEEDINGS OF THE BRITISH ACADEMY

On account of the doubly imperfect state of this inscription the syntax completely eludes me, and also the identity of most of thenbsp;individual � words. A search ought to be made for the rest ofnbsp;the bronze, which must have been a document of unusual pretensionnbsp;for a Celtic inscription.

ii. Alise-Ste.-Reine. The little town of Alise is some three or four kilometres from the station of Les Laumes, about an hour�s railnbsp;before you come to Dijon from Paris. It is situated on a slope ofnbsp;the hill called Mont Auxois, on the plateau of which stands the bronzenbsp;statue erected to the memory of Vercingetorix by Napoleon HI.nbsp;Alise has an interesting museum, among the contents of which isnbsp;a well-known Gaulish inscription which I wanted to examine: seenbsp;Stokes�s Celtic Declension, No. 18, and C. I. L., XIII. 2880; alsonbsp;the Dictionnaire arch�ologique de la Gaitle, where (under � Inscriptionsnbsp;gauloises�) in the plates. No. 7, the stone is described as a � Cartouchenbsp;avec moulures et queues d�aronde trouve sur le plateau d�Alise�;�

(2)

(3)

(4)

(5)

(6)

I EVRV 7 VCVETE 7 SOSKi CELICNON O' ETICnbsp;GOBEDBI 7 DVGI lONTI lOnbsp;'�gt; VCVETIN Vnbsp;IN--- ALISllA

Some of the words are separated by a little triangular mark, and a leaf has been carved in front of ETIC and VCVETIN and afternbsp;A LIS11 A at the end; but it is right to say that according to thenbsp;Corpus this last belongs to the previous line, and follows the pointnbsp;after VCVETIN. The- editor suggests also that there was anothernbsp;leaf in the broken space between IN and ALISI1 A, but I cannot acceptnbsp;either suggestion. The lettering is good, and it has one or two pointsnbsp;deserving of notice; in DANNOTALI the O is bisected by the perpendicular of the T, and the three letters following are of smallernbsp;dimensions and grouped thus, l,, with the two last letters placednbsp;beneath the A. In SOS IN the I consists of a prolongation upwardsnbsp;of the first perpendicular of the N. The 11 have usually been transcribed E, but I should rather represent the words in which theynbsp;occur as dugiipntiip and Alinia�that is, with the vowel i followednbsp;by the cognate semi-vowel or consonant i. This would, in the case ofnbsp;Alisiia, for example, represent a stage of pronunciation correspondingnbsp;to a term petuorua in the series of modulations from petuoria to the

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 5

Welsh pedwyred, pedwreS, or pedwared, �quarta.� So with the ijp (of dugiiontiio) as compared with the yd of Welsh pedwyryd, � quartics ornbsp;quartum,' and with Welsh trydyd, � third,� feminine tryded, which wenbsp;have possibly in a proper name Trittia from Provence (C. I. L., XII.nbsp;316). The other uncertainty attaching to the reading is as to whatnbsp;has dropped out in the breakage between IN and ALISIIa. Thenbsp;editor of the Corpus gives it as his opinion, already mentioned, thatnbsp;it was a leaf, but there is a more natural suggestion to make, namely,nbsp;that IN was not the whole word, but some such longer form as indunbsp;or indo, �in, within�: see Stokes�s Urkeltischer Sprachschatz (p. 31),nbsp;s.v. endo, � in.� As it happens. Holder gives under In no instance,nbsp;except this, which is not such, of a Gaulish preposition in. Thenbsp;Welsh preposition is yn, � in,� and there may have been a Gaulish in;nbsp;but the Welsh word when you add to it becomes ynnof, � in me,�nbsp;ynnot, � in thee,� ynnom, � in us,� amp;c., where the pronunciation hasnbsp;a double n which seems to postulate nd as its antecedent. There isnbsp;another space much smaller than the one last mentioned, but largenbsp;enough to be noticed: it occurs between the I and the rest of thenbsp;word IE V R V in the second line. No explanation of this offers itselfnbsp;except carelessness on the part of the inscriber.

A word now as to the names and the interpretation. Martialis is evidently the Roman name borrowed, and the father�s namenbsp;Dannotalos is well established. It occurs in Etruscan spelling asnbsp;Tanotalos in an inscription from Briona, near Novara, in Northnbsp;Italy: see No. xxxiv below. Holder gives the correspondingnbsp;feminine as Danotala. Compare also Argiotalus, Kaa-a-iraAoy,nbsp;Dubnotalus, and Vepotalos. The element talo-s is supposed to meannbsp;the forehead, but there was also an Evotalis (Irish Eothail), wherenbsp;tal- of a different declension may perhaps be a different vocable.nbsp;In the compound Dannotalos the element danno is to me obscure asnbsp;to meaning and origin in spite of such other compounds as Dannorixnbsp;and Dannomarus, given by Holder. Before leaving this point it isnbsp;to be noticed that the genitive Dannotali means Dannotali Jilius innbsp;Latin. This is one of the regular Gaulish ways of expressing thenbsp;relationship of father and son: another way would have been to havenbsp;called the son Martialis Dannotalicnos, � M. little Dannotal,� as willnbsp;be found done in No. xxxiv. Unfortunately in either case we arenbsp;not given the Gaulish word for son, and there is another formula fornbsp;Gaulish patronymics, but that also eschews the use of a vocablenbsp;for � son � or � daughter.� The next word IE V R V is one of the verynbsp;few Gaulish verbs which have for certain been identified as such, andnbsp;it is treated as pract-ically equivalent to the Latin fecit, � made �;

but in the Corpus, XHI. 1326, the Greek equivalent is given in the imperfect as CTTOCI. Here the Gaulish verb seems to have as itsnbsp;accusative sosin celicnon: that is, a noun celicnon with a demonstrativenbsp;sosin. The former appears to have meant a tower or some suchnbsp;a structure, for the word seems borrowed as kelikn into Gothic,nbsp;where it was used to render the Greek words avdyaiov, nvpyos.nbsp;In Gothic the word was a neuter, and most likely the original innbsp;Gaulish was also neuter. The word is probably connected with thenbsp;Welsh celaf, �I hide or conceal,� and cel, �concealment�; in Mod.nbsp;Irish respectively ceilim, �I conceal,� and ceal, which among othernbsp;things means � a cover,� and from which a possible diminutive wouldnbsp;be ceilm, corresponding exactly to our celicno-n. So this last mightnbsp;be explained in the widest sense as means of hiding or covering onnbsp;a relatively small scale; but the Gothic loan-word indicates that itnbsp;admitted of being narrowed in sense so as to mean a roof or shelter,nbsp;a tower or turret. Perhaps we may call it a cell: at all events thenbsp;Latin cella is usually explained as representing an earlier celula ornbsp;else Celia, from the same root cel as Anglo-Saxon helan, �to cover,�nbsp;German helilen, also Latin c�lare, and the Celtic vocables alreadynbsp;mentioned.

Of the first clause there remains VC VETE, which according to the run of the sentence should be a dative�in fact a dative femininenbsp;like BHAHCAMI, the dative of Belisama�s name in the Vaisonnbsp;inscription. No. vi; and with this declension the accusative Ucuetinnbsp;harmonizes. The analogy of Belesama, dative Belesami, points tonbsp;the fact that the nominative was Ucueta, but another declensionnbsp;seems to be not impossible: Uciieti-s, dative Ucueti, accusativenbsp;Ucueti-n, either masculine or feminine. On the whole I prefer thenbsp;other conjecture, that the nominative was Ucueta, a femininenbsp;corresponding to such masculines as O. Irish cing, �a warrior,�nbsp;genitive cinged, dative cingid, belonging to a declension whichnbsp;Dr. Stokes describes as �weak forms of wlt;-stems��that is to say,nbsp;mostly present participles. The stem in the case of cing, for earlynbsp;cinget-s S we have, for instance, in Cingeto-rm : compare Orgeto-vix,nbsp;and the like, where the syllable et is attached to the verbal stemsnbsp;cing, � to go, march,� and org, � to kill.� Similarly, if we cut offquot;nbsp;the et-e of Ucuete we may expect to have a verbal stem in ucu; andnbsp;assuming such to be the case, there can be little hesitation as tonbsp;what it must be, namely, ud-gu, from which 0. Irish had uccu, uca,nbsp;� a choice or act of choosing.� The prefix ud, od is the equivalent of

* Stokes in his Urkettischer Sprachschatx cites Cinges from Hefner, 280, genitive Cingetos, s. v. Kenget-, p. 77.

the English out, German aus. Compare Irish to-gu, of much the same meaning as uccu; but the stem was in full not gu but giis, thenbsp;s of which is retained before t in Latin gustus, �taste,� a noun ofnbsp;the u declension, represented in the gwst of the Welsh proper namenbsp;Ungwst, Unwst = Irish Oengus, � Angus,� and Gwrgwst = Irishnbsp;Fergus, genitive (in Ogam), Vergoso, for an earlier Verguss�s.nbsp;Accordingly Ucuet-i stands for an earlier Ucuh-et-i, for Ucus-et-i,nbsp;with the sibilant between vowels changed into an h, which eventuallynbsp;ceased to be sounded: compare suiorebe in the Gaulish inscription No. xxxi, from Neris-les-Bains. The derivation of our word isnbsp;phonologically subject to no serious doubt. It is not quite so easynbsp;to fix the meaning of it as name or epithet, which may be eithernbsp;� the choosing one � or � the chosen one,� � the loving � or � the lovednbsp;one,� diligens or d�lecta. On the whole I should be inclined to treatnbsp;it as active rather than passive: compriato, � loved or beloved,� innbsp;the second Rom inscription is not parallel: see Celtae and Galli,nbsp;pp. 41, 42, 46. This was probably not the name, the nomen, of thenbsp;divinity intended, but an epithet�the loving one�understood by allnbsp;the worshippers to whom the cult appealed. Here it may be asked,nbsp;if s became h and then zero in the body of the name Ucueta, whynbsp;it is we have Alisiia with a single s flanked by vowels. The answer isnbsp;that here s represents the sharp sibilant ss, derived probably from cs,nbsp;for which we have the evidence of inscription xxxii, probably anbsp;Celtican one, from the neighbourhood of Bourges. There for � indunbsp;Alisiia � we have � in Alixie.� After a certain period of indecision asnbsp;between s and h, every single s found flanked by vowels in writtennbsp;Gaulish should be read as ss�that is, provided one could rely onnbsp;consistency in the spelling. As one cannot often so rely, one has tonbsp;decide each case according to the etymological evidence : where therenbsp;is no such evidence, judgement has to be suspended.

Thus far the first part of the inscription has been discussed and inferred to mean�Martial son of Dannotalos made this turret fornbsp;Ucueta or �the loving one.� The original covered by this is sonbsp;exactly one half of the whole as to suggest to me that we have herenbsp;to do with two verses of text in metre of some kind. Unfortunatelynbsp;the second half consists mostly of words which are obscure. Dr. Stokesnbsp;translates it � and the work pleased Ucuetis in Alisea,� where etic isnbsp;treated as the conjunction and goiedbi as the verb having the nextnbsp;word as its nominative. There is nothing to say against renderingnbsp;etic by � and,� except that other meanings are possible ; but, taking itnbsp;to have meant �and,� it would seem to contain eti- of the same originnbsp;as the et of eto (earlier etwa = eti-hu-), �yet, again� and �still� as in

PROCEEDINGS OF THE BRITISH ACADEMY

gzoell eto, � still better As to the c of etic one may perhaps equate it with that of the Latin hk, haec, hoc as part of the particle ce innbsp;hicce, ecce which is regarded as of the same origin as Latin cis, � onnbsp;this side.�

GOBEDBI is doubtless the verb of its clause, though it seems highly improbable that it is in the past tense; and instead of sayingnbsp;� and the work pleased U.� I should rather take the words to meannbsp;�and may the work please U.�, that is if dugiiontiio is to be treatednbsp;as the subject, which is doubtful. GOBEDBI is undoubtedly thenbsp;reading on the stone, but I cannot make much out of it. We seem,nbsp;however, to have a choice of easy emendations for DB, such as 6B,nbsp;BB and 66, underlying which should be a lisping of S; fornbsp;I put in BB on the chance of its being sh subjected to a processnbsp;of assimilation. In that case BB may be dismissed as a form of 6B :nbsp;then we have left 6B and 66, and the first part of the verb detachesnbsp;itself as gobes-, which I should treat as goves- �goues- with its semivowel written b as was usual enough in Late Latin. This points tonbsp;a word like Latin gavlsus, the past participle of gaiideo, � I rejoice �;nbsp;but in that case one would perhaps expect a vowel between 6 and B : so,nbsp;rejecting 6B one falls back on 66 in order to treat the whole verb asnbsp;gobeMi=gouessi. The error might be regarded as due to the inscribernbsp;being unused to the letter 6, which is not always easy to distinguishnbsp;from a badly formed B. In the copy given to the inscriber the 66nbsp;might have had the bisecting line carelessly made so short as not tonbsp;have caught his eye in the first consonant at all, and in the secondnbsp;only as a part of a somewhat badly formed B.

The conjecture gouessi recalls the participial feminine gouisa, occurring in the first Rom inscription {Celtae and Galli, p. 37), wherenbsp;should be compared, in other respects, the verbs demti, demtissie;nbsp;also derti, atehotisse, and demtisse in the other Rom inscriptionnbsp;(ibid., p. 41, and, corrected, at p. 97 below).

A better conjecture has the advantage of requiring no emendation ; for Gaulish had no objection to such combinations as db and dg:nbsp;witness Abyevvopiyi and . . ABO . ., in Nos. x and xiv; see also xxiv'¹.nbsp;So it would be simpler to treat the gobed (=goiied) of gobedbi, as thenbsp;equivalent of the gaud of Latin gaudeo and the yqd of the Greeknbsp;yrjdfoo, � I rejoice,� compounded with a form of the Gaulish verb � tonbsp;be.� The compound might be regarded as somewhat parallel to Latinnbsp;forms like ama-bo, ama-bam, mone-bam, rege-bam, and the like. Butnbsp;the absence of an intervening vowel in goued-bi faces us as before:

had such a compound been of old standing, it ought to have become gouepi after the analogy of Ucuete for Ud-guhete. It is probable,nbsp;however, that the shortening of some such a form as gouedo-bi maynbsp;have taken place relatively late, but early enough to be shared in bynbsp;Brythonic. Witness forms like Welsh gwybyd (= gwyd-byd), � donbsp;thou know,� clyhu { = clut-bu-), � audivit,' and the older Brythonicnbsp;hep amgnaiibot, ^ sine mente, without understanding,� in Mod. Welshnbsp;ymnabod with d for the au of amgnaiibot ( = ambi-gnat[o^-buti-)nbsp;because of the stress being at one time probably on the closingnbsp;element: compare pedwyryd from petuonio-, � fourth.�

Next comes DVGlIONTllO in which I cannot see a nominative feminine, or indeed a nominative at all. For had it been neuter, onenbsp;would expect it to have had a final n just as much as the accusativesnbsp;sosin celicnon and Ucuetin. Fui�ther, this retention of final n doesnbsp;not encourage one to assume the discarding of final s in the casenbsp;ofDVGllONTlIO; that is, if one were to treat the latter as representing an earlier nominative dugiipntiios. All that remains for usnbsp;is to regard dugiiontiio as a dative or some other oblique case, to benbsp;construed in an adverbial sense. What that sense should be isnbsp;suggested by the congeners of this word, among which Dr. Stokesnbsp;mentions Greek T(vxagt;, �I prepare, I bring about,� A.-Sax. dugan,nbsp;� to be of value, to be strenuous,� with which go dohtig, Mod. Englishnbsp;doughty, German taugen and tuchtig, also tugend, � virtue,� to whichnbsp;one may add the Lithuanian daug, � much, many.� Thus it wouldnbsp;seem that we might interpret our word as meaning �for good� ornbsp;more exactly � for our good, and to our joy.� The clause wouldnbsp;then run, �And for our good may it rejoice Ucueta in Alesia.�

It has already been suggested that the whole is in metre, and I scan it roughly, thus :�

The metre is accentual hexametre, and the characteristic portion of the lines is the last two feet

I need not discuss it here as it has been treated at great length in The Englyn, which occupies the 18th volume of the Cymmrodor,nbsp;the magazine of the Hon. Society of Cymmrodorion (London, 1905).nbsp;It is needless to say that the discovery, if it should prove such, of this

metre on Gaulish ground is of capital importance, as it substantially establishes the fact that in the majority of its words the Gaulishnbsp;language accented either of the last two syllables as Welsh does tonbsp;this day. To such a rule, however, there must have been manynbsp;exceptions, and the first important kind of exception which hasnbsp;attracted my attention is the case of compounds like Dann�talos:nbsp;for more about them see The Englyn, pp. 6-10.

iii. Dijon. The Museum at Dijon has a patera in bronze, found in the neighbourhood in 1853, and on the handle is the followingnbsp;inscription with the leaf ornament at the end: see Stokes, No. 17;nbsp;Diet. Arch., No. 6 and 6nbsp;nbsp;nbsp;nbsp;; C. I. L., XIII. 5468

We did not send for permission to take the vessel out of the glass case where it is kept, as we could read it perfectly well where it was,nbsp;and see that it has been correctly copied. It may be rendered, � Doiros,nbsp;son of Segomaros, made it for Alisanos,� though it would be somewhat more exact to put it thus: � It is Doiros, son of Segomaros,nbsp;that made it for Alisanos�: at all events it would be so in Welsh,nbsp;� D. fab S. a�i gwnaeth i A.� The same applies to most of the Gaulishnbsp;inscriptions; for the fact of the making is assumed to be evident tonbsp;anybody who sees the vessel: it does not require to be told him, andnbsp;the information given begins with the name of him to whom thenbsp;making is attributed, so Doiros takes the emphatic position in thenbsp;sentence. Had the question been for whom the vessel was made,nbsp;Alisanu would have taken the lead: that is, the sentence would havenbsp;begun with it. At the same time one feels that the English renderingnbsp;with �it is� rather exaggerates the emphasis intended on Doiros.

Doiros is a rare name, while that of Segomaros will be found to come before us again: with the former may possibly be equated annbsp;Irish personal name Doir, Dair, Dair, both nominative and genitive,nbsp;which occur in Irish annals at the beginning of the seventh century:nbsp;see �the Four Masters,� a.d. 619, and the Annals of Ulster, a.d. 623 ;nbsp;compare also 0�Grady�s Silva Gadelka, pp. 56, 57, where we havenbsp;a place called Druim Meic Dair, � Mac D4ir�s Ridge �; compare alsonbsp;0�Donovan�s note on the entry by � the Four Masters,� where he derivesnbsp;Gweedore, a well-kuown place-name in Donegal, from Gaeth-Doir. Itnbsp;is to be noticed that the diphthong appears to have been in Irish oinbsp;or di. Alisanu, the dative of Alisanos, was the name of a god : Holdernbsp;cites the following Cote-d�Or inscription � Deo Alisanu Paullinus pronbsp;Contedoio fil(io) suo v(otum) s(olvit) l(ibens) m(erito).� Hence it

appears that the vessel belonged to a temple of Alisanos, and the maker seems to have manufactured him a thoroughly good article.

iv. Beaune. An inscribed stone found in the eighteenth century near Auxey, in the department of the Cote-d�Or, is now in thenbsp;Museum at the Mairie of the town of Beaune: see Stokes, No. 15;nbsp;Die. Arch., No. 4; C. I. L., XIII. 2638. It reads as follows:�

Here instead of the genitive of the father�s name we have a diminutive formed from it, Oppian-icno-s, that is �little Oppianos,�nbsp;where it is impossible to avoid seeing in kno- the termination whichnbsp;we have already had in the neuter in celicno-n, p. 4 above: therenbsp;it was suggested that icno is reduced in Modern Irish to in. It is, parnbsp;excellence, the diminutive and endearing termination in that language:nbsp;we have it for instance in h�thairm, � a lane or narrow way,� fromnbsp;b�thar, � a road,� and in other nouns like uainm, � a lambkin,� and �inin,nbsp;�a little bird,� from uari, �a lamb,� and can, �a bird.� In Ogam-written Irish we have it as -ign-, for instance in the genitive Cunign-i,nbsp;Welsh Cynin in UB.TL\gynm, �Ecclesia Cunigni.� The derivation ofnbsp;Iccavos and of Oppianicnos is obscure, but they seem to have theirnbsp;congeners in Iccius, Iccio-dnrus, Iccio-magus, and in Oppianos, whichnbsp;occurs as Oppianos in C. 1. L., XII. 1029, 4679, not to mention suchnbsp;related forms as Oppius, Opia, Opiavus, Opiava, as to which seenbsp;Holder. Brigindoni seems to be the dative of Brigindo, the namenbsp;perhaps of a female divinity, to be compared with Brigantia and thenbsp;Irish Brigit. Cantalon I have supposed to be the Gaulish word,nbsp;which in Welsh is cathl, � a song �: see my Celtae and Galli, p. 31. Thenbsp;second a in cantalon is inorganic, like the corresponding vowel in thenbsp;Irish equivalents, c�tol, c�tul, octal, and in Gaulish it does not appearnbsp;to have counted as a syllable. For the inscription is metrical likenbsp;that of Alise-Ste.-Reine, and scans as follows:�

Here Brigindoni departs from the general rule of accenting the penultimate: it is probably the accent of the nominative fixed,nbsp;supposing that was Brigmdd.

V. Autun, An inscription on a stone found in the last century at Autun is preserved in the Lapidary Museum of that city : see Stokes,nbsp;No. 16; Dk. Arch., No. 5; C. I. L., XIII. 2733. The stone is slightly

damaged at the top and the right-hand edge, but the reading is certain, and runs as follows :�

The lettering is said by the editor of the Corpus to be of the beginning of the first century: where O follows C or L it is carvednbsp;small in the bosom of the consonant; the V ending the second line isnbsp;also made small on account of want of room for a bigger letter. Thenbsp;name Licnos is of obscure origin, and is equated by Holder with annbsp;Irish L�n; however that may be, it occurs in other inscriptions foundnbsp;in the Celtic countries of the Continent, as a glance at Holder�snbsp;article on it will show. In the cognomen Contextos the x probablynbsp;means the Greek x or ch, and Stokes refers the second part of thenbsp;word to the same origin as the Latin words tego, tectum, and Irishnbsp;tech, � a house,� with which go the Welsh ty, � a house,� and to, innbsp;English thatch. So he would ascribe to Contextos the meaning ofnbsp;protector. Anualonnacu seems to be the dative of Anualonnacos, thenbsp;meaning of which is to seek. The compound canecosedlon is alsonbsp;obscure, but it has been suggested that it should be rendered � a goldennbsp;chair,� that is, with sedlon compared with Latin sella and Englishnbsp;settle; but I see no proof that caneco meant � gold or golden,� nor cannbsp;one be sure of � chair,� for � saddle,� in Slavonic sedlo, would perhapsnbsp;be equally admissible, but neither sense seems to derive any corroboration from the vocabulary of the Neoceltic languages. On the othernbsp;hand, the compound before us appears to supply us with a Gaulishnbsp;s�dl- closely akin with the Welsh feminine hoedl, � lifetime,� whencenbsp;hjr-hoedled, � shortness of life,� which in Med. Irish is expressed innbsp;the Booh of the Dun Cow (fo. 60��) by gax-s�le Analogy wouldnbsp;suggest for s�le and hoedled a Gaulish s�dliia; but neither does thisnbsp;help one exactly to see how the canecosedlon of the inscription is to benbsp;interpreted. Allowing oneself, however, to be more or less guidednbsp;by the analogy of the transition of meaning from seat, for instance,nbsp;in the sense of stool to a seat in the sense of country seat, wherenbsp;a man spends his life, or by the etymologically suggested relationnbsp;between a settle to sit upon, and that of the length of one�s stay ornbsp;sitting in the world as settled or fixed by destiny, one might venture

* Later MSS. substitute a form garsecle or the like, formed with the help of the Latin saeculum : I have not succeeded in finding the passage in the readingsnbsp;given in Windisch�s Tain.

to think it not improbable that the ancient Gauls may have used sedlon in both the senses indicated. In that case it might perhapsnbsp;be suggested that the key to caneco is supplied by the Irish wordnbsp;cdin, � law, canon, rule,� for an early Celtic nominative cani-x, whencenbsp;possibly an adjective canico- or caneco-. Thus one would be enablednbsp;to interpret the compound word as a �law chair, a judgement seat ornbsp;tribunal.� Since writing the foregoing my attention has been callednbsp;by a passage in M. d�Arbois de Jubainville�s book entitled Lex Druidex,nbsp;p. 5, to two Latin inscriptions at Autun mentioning a Gaulish godnbsp;Anualos or Anuallox. Now our Anualonnac-u seems to be directly ornbsp;indirectly derived from that of the god. This suggests that thenbsp;Anvalonnacos was in some sense or other in the special service of thenbsp;god Anvalos, perhaps his gutuater or � flamen,� and that Licnos hadnbsp;made an official seat for him. While giving this conjecture for whatnbsp;it is worth, I may mention that others are possible, though hardlynbsp;worth mentioning.

Like some of the previous inscriptions this also is in metre, in fact the same metre, for it seems to scan as follows:�

It has sometimes been supposed that the legend is incomplete at the top, that is, that Licnos is the latter part of a longer name; butnbsp;the metre makes this inadmissible, and at the same time it carries thenbsp;accentual hexameter back to the beginning of the first century.

vi. Avignon (1). The Cal vet Museum at Avignon contains among other Gaulish inscriptions one found in 1841 at Vaison in the Department of Vaucluse. It is written in cursive Greek letters: see Stokes,nbsp;No. 6; Diet. Arch., No. 2; and C. I. L., XII. p. 162, where the editornbsp;describes the writing as litterix malis et leviter ineixix: I can only saynbsp;that I should have been glad if the rest of the Celtic inscriptions innbsp;Greek letters had not often been worse. It will be noticed that the sigmanbsp;here, as in most of the other Celtic inscriptions in Greek characters,nbsp;has the form of C, and that the coupling of the A consists not of a singlenbsp;straight line but of two, thus v: the reading is the following:�

ceroMAPoc OYIAAONCOCnbsp;TOOYTIOYCnbsp;NAMAYCATICnbsp;eitUPOYBHAHnbsp;CAMICOCINnbsp;NeMHTON

vowel M as in (icapov, which has come before us hitherto only in its spelling of lEVRV. When the diphthong ou or ow has to benbsp;expressed we have oov as in roovnovs; but at the same time vnbsp;alone, preserving the sound originally intended for it in Greek, isnbsp;used in the diphthong au (pronounced as in German) in Napavo-any.nbsp;The �i of euapov had probably the same sound as that of i innbsp;the Latin spelling ieuru pronounced most likely as a trisyllablenbsp;i-eur-u. Compare Greek ei used even for Latin i in the name EIOY-BIANOS RETOS (C. I. Z,., V. 5408) which Mommsen regarded asnbsp;meaning lovianus Raetus : it comes from the neighbourhood of Comonbsp;in North Italy. As to the history of Greek ei, see Blass Ueher dienbsp;Aussprache des Griechischen (Berlin, 1888), pp. 31, 34, If any distinction was made between e and tj in Celtic names, it must have been onenbsp;of quality or breadth as it was in Greek itself (Blass, ibid., pp. 24-7).nbsp;At all events rj does not indicate a long vowel: it is short in the threenbsp;instances before us, and in two out of the three tj seems to have bornenbsp;the stress accent. Take first BriXria-api, which was the dative ofnbsp;Belisama, the name of a goddess identified with Minerva in an inscription found at St.-Lizier in the Department of Ari�ge and beginningnbsp;with the words Minervae Belisamae sacrum (C. I. L., XIII. 8). Ptolemynbsp;in his Geography, ii. 3. 2, gives the goddess�s name to an estuary innbsp;Britain, which is supposed to have been the Mersey, and the MSS.nbsp;seem to be unanimous in reading BtXiaapa. From the latter Holdernbsp;derives Belismius (in a Caerleon inscription, C. /. L., VH. 97), and fromnbsp;this in its turn the place-names Blismus in the Department of Ni�vre,nbsp;Blesmes in that of Marne, and another in that of Aisne. The interestnbsp;of these forms is that they prove the name of the goddess to havenbsp;been like such masculines as Cintugnatos, Dannotalos, Segomaros, andnbsp;similar quadrisyllables accented on the antepenultimate, Belisama:nbsp;see page 10 above. This is proved by the significant way in whichnbsp;the accented syllable has, in the French place-names, annihilated thenbsp;two syllables which flanked it in Gaulish. The other word spelt withnbsp;7), namely, vepL-qrov, is to be equated with the O. Irish nemed, a glossnbsp;on sacellum: see the Gram. Celtica, pp. 10, 801, and compare Strabo�snbsp;compound Apvviperov or ApwaCptrov, xii. 5 (C 567), and in Ptolemy�snbsp;Geography, ii. 7. 12, AvyovaTovip.(Tov. But this accentuation isnbsp;doubtless Greek rather than Gaulish, which was probably nein�ton.nbsp;But our first name was probably Segomaros with the stress onnbsp;the short o and not on the long a: compare such Greek words asnbsp;amp;v9pC)gt;TTos, which is, I understand, become avdpoiros in Mod. Greek;nbsp;but in Gaulish I should rather expect the change, when it took place,nbsp;to have been towards S�gomaros and even Segmaros.

These words may next be reviewed with regard to their etymology and meaning: in his Urkeltischer Sprachschatz, p. 297, Stokes treatsnbsp;Segomaros as Segomaros with sego, meaning � strength, victory,� of thenbsp;same origin as German sieg, � triumph �; and maros may be taken asnbsp;the ancient form of Welsh mawr, � great,� 0. Irish mar, mor. Thus thenbsp;whole name should mean � him of great strength or power.� Uilloneosnbsp;is supposed to be an adjectival formation from the father�s name,nbsp;which may accordingly have been Uillonos; the derivative should meannbsp;belonging to or related in some way to Villonos, in this instance relatednbsp;to him as his son. The more usual adjectival ending employed innbsp;such cases is that in -io- as in Tarbeisonios (in No. xxvi) and the like;nbsp;for I do not identify -eo-s with the latter, and the question of its originnbsp;is a difficult one. But it seems to claim to be equated with the e ofnbsp;such Latin formations as aureus, � golden,� from aurum, � gold,� charteus,nbsp;�of or pertaining to paper,� from charta, �paper,� terreus, �of earth,nbsp;earthen,� from terra, � earth,� on which see Stolz�s Historische Grammatiknbsp;der lateinischen Sprache (Leipsic, 1894;, p. 473). Stolz \ however, is ofnbsp;opinion that Latin -eo- represents a prehistoric -�io- which lost its i � imnbsp;Uritalischen �; but I should prefer to suppose a still earlier, Italo-Celtic combination -�o-, and to regard it as surviving in Latin andnbsp;Celtic. The difference in the application of the termination oflTersnbsp;no difficulty; from saying, for instance, that a thing is of earth earthynbsp;to the English saying that a man�s child is � a chip of the old block �nbsp;is but a short step. Another instance to compare with OviWoveos isnbsp;Airovjuapfo?, from Litumaros, in No. xvi, and others will be found innbsp;Nos. XX and xxxiv.

With Toovnoay Stokes {Celtic Declension, p. 54) equates Toutiu (better Toutio) of the Briona inscription (No. xxxiv below), butnbsp;the retention here of the final sibilant is rather unexpected, if, asnbsp;he suggests, roocnovs is the same vocable whose Latin genitivenbsp;Toutio\nu^ is restored by Mommsen and Hirschfeld in C. I. L., XIII.nbsp;5278, and whose etymological equivalent is recognized by Dr. Stokesnbsp;in the Gothic thiudans, � king.� With this last, however, the name tonbsp;be equated is rather Toutu or Touto, Latin genitive Toutonis, whichnbsp;occurs in an inscription found at Arles (C. /. L., XII. 852). Thesenbsp;and kindred forms are derived in the last resort from touta, � a peoplenbsp;or tribe,� in Irish tuath, � a tribe,� Welsh tud, � a people�s country,�nbsp;Breton tud, � people,� A.-Sax. theod, � the race, the people,� Gothicnbsp;thiuda. If Dr. Stokes�s suggestion that TOOYTIOYC representsnbsp;toutiuns of the w-declension is to be rejected, one has to regard it as

* For calling my attention to Stolz and to Blass I have to thank my friend Professor Joseph Wright.

16 PROCEEDINGS OF THE BRITISH ACADEMY

a nominative of the M-declension: in either case it was probably pronounced toutuus or toutius, liable to be contracted into toutius.nbsp;Dr. Stokes renders the word by � magistrate �; but, in the absence ofnbsp;sources of information as to the exact meaning of the word, it seemsnbsp;safer to treat it as meaning no more than a citizen or native of Nimes.

Namausatis, like Namamicabo in No. xvii is derived from Namamos, given in Latin more usually the form Nemausus, which the Frenchnbsp;Nimes proves to have been accented N�mausos or N�mosus: possiblynbsp;it reached the Romans through a Celtican channel, while the samenbsp;vocable in its more Gaulish form has to account for the Neu�xrcros ofnbsp;Strabo, iv. 2. 3 (C 191), which was the city afterwards called Augusto-nemetum, at the present day Clermont-Ferrand, in the Puy-de-Dome.nbsp;The Gaulish form also underlies the actual name of Nemours innbsp;Seine-et-Marne One cannot help also seeing that the nemetum ofnbsp;Augustonemetum was probably related to Negoxrtrdy, perhaps evennbsp;synonymous with it, and that phonetically the first a of Namausatisnbsp;and the Namausicabo of No. xvii had taken the place of an earlier enbsp;under the influence of the a of the ensuing syllable: this wouldnbsp;happen all the readier in a syllable, which in an earlier pronunciation of those words, was unaccented. They are to be traced backnbsp;probably to the same origin as Greek re'gos, �a wooded pasture,nbsp;a glade,� Latin uemm; but a more complete parallel�so complete,nbsp;in fact, as to suggest a case of borrowing�offers itself in the oldnbsp;German nimid, � heiliger Waldplatz � (Stokes�s Urk. Spr., p. 192); andnbsp;Holder, s. v. nemeton, cites from the Indiculus Superstitionum et Pagani-arum the heading �De Sacris silvarum quae nimidas [=nemeta]nbsp;vocant�: see Pertz, LL. I. 19, and LL. Cap. I. 223.

As to fiuipov, the CO of the Greek spelling, as contrasted with the eu of ieuru in Latin letters, seems to prove that there were at least twonbsp;pronunciations, but we are not helped by the etymology of the word,nbsp;as its origin is obscure; but it is possibly to be traced to the samenbsp;source as an Irish verb of which we have the Mediaeval forms furas,

� he or she who does, makes, or causes,� nt iurfaithe, � would not be done or wrought �: see Kuno Meyer�s opinion in the Revue Celtique,nbsp;vi. 191, 192, and, as inconsistent with it, Stokes�s in his Celticnbsp;Declension, pp. 62, 63, where he is inclined to refer the Irish formsnbsp;to a compound of the verb orgim, of much the same meaning asnbsp;Latin caedo, �I cut, I kill,� and similar significations. Perhaps in Celticnbsp;one may compare the personal name Andiourus, which Holder wouldnbsp;divide into And-iourus. As to the termination u of the word here

� See M. J. Vendryes�s � Melanges Italo-Celtiques � in tho M�moires de la Soc. de Linguistique de Paris, 1905, pp. 390, 391, and The Englyn, pp. 6, 7.

in question, it is to be observed that the Irish forms argue a weak verb on a level with the Latin amo, � I love,� or canto, � I sing,� makingnbsp;in the third person singular of the preterite amavit, cantavit, whichnbsp;are in Italian amo, cantd. Similarly the u of ieuru may be all thatnbsp;was left of -ouit or -auit in Gaulish; compare logitoe in No. xxxi, andnbsp;camitu, carnitus in Nos. xiv, xxxiv, xxxvi. That this kind of formationnbsp;existed in Celtican we know from voravi and priavi in the Romnbsp;inscriptions: see my last paper, pp. 41, 47, 64. Sosin probablynbsp;meant �this� as in No. ii above, and as it agrees with celicnon andnbsp;nemeton it must be regarded as neuter, and contrasted with thenbsp;Celtican sosio used substantivally as a neuter in the Horn inscriptionsnbsp;no less than half a dozen times.

Lastly, the dactylic rhythm of the latter part of the inscription leads me to suppose that the whole was meant to be verse; but I amnbsp;far from certain that I have hit on the metre intended. On the wholenbsp;it seems to form a hexametre and a half, somewhat as follows:��

2eyo/iia|pos OuiXAojr^oy Tooi;[rfoay Najaanjcrdny ei,\(apov j lt;r�lt;Tiv vt\ii,'^Tov.

It is to be noticed that the short line has the stress accent exactly placed as in Horace�s Insignes aut Th�ssala T�mpe, or Ini�cto ternbsp;pulvere curras in Odes, i. 7 and 28, that is, when read as prose. I shouldnbsp;not venture, however, to suggest that the author of the inscriptionnbsp;knew anything about the Alcmanian couplet, but only that henbsp;imitated the (prose) accentuation of the hexameter, and took alsonbsp;the liberty of appending a part hexameter. But as we are nownbsp;getting southwards, with Massilia not so very far off, a questionnbsp;which I cannot answer presents itself�Was the hexameter, which thenbsp;Gauls set themselves to imitate and to modify in their own 'vay, takennbsp;from Roman poetry or directly from the Greeks

vii. Avignon (2). On the hill overlooking the town of Orgon, in the Bouches-du-Rhone, was discovered in 1866 an inscribed stone,nbsp;which is now in the Calvet Museum at Avignon: see the Revuenbsp;Celtique, vii. p. 450, viii. p. 397�both inaccurate, and C, I, L.,nbsp;XII. p. 820. It reads as follows:�

The lettering, which is not good, has the following points deserving of notice. The Y consists of a perpendicular stroke prolonged belownbsp;the line, and of a short straight line branching from the upper part

of that perpendicular towards the left. The B has its lower part larger than the upper, and the arcs are seldom brought together tonbsp;touch the perpendicular. The P is long, with its semicircle small andnbsp;tending to become a triangle. The second O is smaller than the othernbsp;letters. The sigma (imperfect at the top) and the alpha have thenbsp;same forms as in the Vaison inscription, No. vi; also ov and oov arenbsp;here used as in that one. No intentional ligature occurs, but othernbsp;inscriptions go to prove that what is here clearly an M should havenbsp;been a ligature for NA : that is, it lacks the joining line. Mostnbsp;likely the workman who carved the letters misread the copy givennbsp;him, with the last word correctly spelt KANTEN A, but with the twonbsp;last letters ligatured; unless M be simply a slip, for N.

The name Vebrumaros is remarkable in having the vowel u, not o, at the end of its first element: whether this means anything morenbsp;than an obscure sounding of the vowel, it is difficult to say; butnbsp;compare such datives as Alisanu in iii and Avevvo in xxxP. Thenbsp;meaning of the name is not certain; but the second element isnbsp;probably to be treated as maros and identified with that of Segomarosnbsp;in Nos. iii and vi, and uebru is perhaps to be explained by reference tonbsp;the Welsh word gwefr, � amber.� In that case the compound wouldnbsp;seem to have meant one who was great or distinguished for his amber,nbsp;one who made a display of amber in the adornment of his person.nbsp;Taranou is the dative case of Taranu-s, �a divinity identified withnbsp;thunder�: the Welsh word is still taran, �a thunder.� Thenbsp;declension was probably nominative Taranu-s, genitive Taranous ornbsp;Taranouos, and dative Taranoui, retaining possibly an old Aryannbsp;accentuation Taranouos, Taranoui. It is a shortening of this latternbsp;that we probably have in Taranou, written here TAPANOOY.nbsp;Compare the doubtful case of Tpaa-fXov in No. xiii, and contrastnbsp;Mapeooat in x and Eivovi in xviii: see also xxii.

To come to the other words, dede is probably more or less analogous in formation to the Latin reduplicate verb dedit, �gave.�nbsp;As to cantena, this has been touched upon in my previous paper, p. 34-,nbsp;where the meaning of primitiae, or firstfruits, has been suggestednbsp;in connexion with the entry (on the 13th of August) concernednbsp;with the offering of the harvest to the god Rivos. To the remarks'nbsp;made thereon I would add that while Irish c�t- corresponds tonbsp;a Welsh cant-, the common Celtic combination was probably cnto-,nbsp;whence a strong form canto- and a weak one cinto- (as in Gaulishnbsp;names like Cint�gnatos), in Irish cetu-, ceta-, cita- (Stokes, Urh. Spr.,

' So far as they postulate C . RIV they are to be cancelled, for the right reading is G . RIV : see pp. 86, 91 below.

p. 77), represented in Welsh by cynt, � previously, before.� As a related word may perhaps be mentioned Latin re-cens, re-centis,nbsp;English recent; and the phonetic sequence in Goidelic is illustratednbsp;by the Cantlo-s of the Calendar being represented in O. Irish bynbsp;c�t�l, c�tal, while the Welsh equivalent cathl (for *cantlo-n) �song�nbsp;retains the older vowel throughout.

The word bratude offers some difficulty, for it is not quite clear whether it is made up of bratu-, �a judgement, doom � (Welsh brawd,nbsp;as in Dyd Brawd, � the Day of Judgement,� Irish La Bratha), withnbsp;de, �from,� as a postposition, O. Welsh di, pronounced (as anbsp;preposition) probably di, Med. Welsh y, Irish de, or else thatnbsp;jSparovbf is a derivative from bratu, not involving any postposition,nbsp;but carrying with it the sense of � through or because of a decree,� bynbsp;virtue of its being in an oblique case, say the ablative or instrumental,nbsp;/3parov8-e. Compare in Alixie in xxxii, also the forms Acmodae,nbsp;Bagaudae, bascauda, Cassauda, Sapaudus, collected by Holder undernbsp;-dos, -da, -don, and also such Irish words as crabud, Welsh crefyd,nbsp;� religion,� which Stokes derives (p. 97) from *crab, � piety ornbsp;religion.� In his Celtic Declension, pp. 62-4, he interprets bratu-denbsp;as � by decree,� � by order,� and cites as its Latin equivalentnbsp;ex imperio, from an inscription reading � Matronis Afliabusnbsp;M. Marius Marcellus pro se et suis- ex imperio ipsarum,� fromnbsp;J. De Wal�s De Moedergodinnen (Leyden, 1846), No. cxx (p. 88).nbsp;There it is to be noticed that the originating of the decree is ascribednbsp;to the Mother-goddesses themselves. That is probably the way tonbsp;interpret ^parovbe, but the instance does not settle the question asnbsp;to the composition of the word.

viii. Avignon (3). Among other inscriptions in the Calvet Museum is one from Apt, or more precisely from St.-Saturnin-d�Apt.nbsp;It is an imperfect one on a small altar reading as follows: see Stokes,nbsp;p. 64; C./. L., XII. p. 137;�

The letter here given as the last of the first line may have been either C, that is s, completing a name Valicis, or else O with its rightnbsp;side broken off�the right edge of the stone is all very rough andnbsp;imperfect. Ualicio would probably be a noun of the n-declension;nbsp;there is nothing to suggest ualkios. The next line may have had annbsp;I after the T, hardly any broader letter. An Arles inscription givesnbsp;a potter�s mark (C. I. L,, XII. 5686. 747) as RE ST I O�that is �Rest!

O(fficina),� and according to that one might treat Ualwio Oneresti as meaning Valicio son of Onerestos. Aiouniai is more difficult tonbsp;interpret, except that it is probably a dative of the name of a goddess.nbsp;Her name in the form we have it admits of being explained in thenbsp;following ways: (1) The termination ai is a way of representingnbsp;the e which Stokes sets down as the ending to be expected�hisnbsp;declension, p. 102, gives � nom. r�da, gen. r�d�s, dative r�de (r�di ?),nbsp;accusative r�dirn,' � a chariot �; and the name as a whole may be of thenbsp;same origin, presumably Celtic, as the man�s name Aio ov Aiio ofnbsp;the w-declension, as to which see the inscriptions cited by Holder.nbsp;Aiunia would be a derivative from Aiuno-s (Aiuna, Aiuno-n) and thenbsp;termination -uno-s may be a variant of -ono-s (-ona, -ono-n) whichnbsp;Holder illustrates by means of a long list of names ending with it.nbsp;(2) Treat the terminal ai in the same way as before, but suppose thenbsp;name of the goddess to be the Greek word aliovia borrowed andnbsp;modified in Gaulish into Aiunia. This would require one to supposenbsp;that some of the Greeks with whom the Gauls had come in contactnbsp;had a goddess called the Eternal or the Everlasting One. (3) Treatnbsp;the whole word AlOYNIAI as intended, in spite of the spelling withnbsp;01�, for the Greek dative allt;avCq, and it would naturally follow that wenbsp;should regard the inscription as a whole as Greek, and not as any kindnbsp;of Celtic. Now on comparing other inscriptions, such as the next onenbsp;here, I am inclined to favour the first of these three interpretations,nbsp;but I feel by no means certain on the point.

ix. Avignon (4). In the court of the Calvet Museum is an inscribed block of considerable weight brought there from Gargas in thenbsp;Department of Vaucluse. In the Corpus, XII. p. 137, it is treated asnbsp;being still at Gargas, and I failed to learn when its removal tooknbsp;place. Stokes�s reading, p. 64, is inaccurate, which is owing probablynbsp;to a misprint, as the lettering is good and perfectly plain. It consistsnbsp;of one line close to the top edge of the block, and runs thus:�

eCKerrAIBAANAOOYIKOYNIAI

It probably means � For Escenga daughter of Blandouicunos.� Here we seem to have the same dative ending AI for e, for I see no reasonnbsp;to suppose this inscription to be Greek. Both names seem at any ratenbsp;to be Celtic: the stem esceng stands probably for what might otherwise be expected excing. Compare eCKirrOP�IZ in inscriptionnbsp;XX, and names like Excingomarus, ExcingiUus, -a. But verynbsp;possibly X is here nothing more than another way of writing s or ss;nbsp;that is to say, the prefix ec-s had in pronunciation been reduced to ess,nbsp;at least when it came immediately before another consonant. The

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 21

adjective agreeing with Escenga probably involves the father�s name, which would be Blandouicunos �, and this in its turn looks like anbsp;derivative from Blando-uix, genitive Blando-uicos, to be comparednbsp;with such forms as Ordo-uices, � hammer - fighters,� Brannouices,nbsp;Lemoukes, and others : the ending -uni-ai reminds one of the Aiuniainbsp;of the last inscription. What the blando portion of the compoundnbsp;meant is not evident: if it be of the same origin as Latin ilandus, onenbsp;might perhaps explain the name as meaning a bland fighter, a courteousnbsp;warrior.

Lastly, I wish to mention that the stone bearing the above writing was dressed for another purpose : it seems to have come out of a greatnbsp;building, and it bears on the face of it, what seems to have escaped thenbsp;editor of the Corpus, traces of a long and elaborate inscription whichnbsp;a stonecutter has purposely effaced; but here and there one cannbsp;identify a letter or two. Thus towards the left edge, not far belownbsp;the Greek lettering, I thought I detected Cl I or CV, a little lowernbsp;I OR DO, but the two last letters were doubtful; also a good dealnbsp;lower down, and of a larger size, I detected VC or AVG (with AVnbsp;ligatured) or perhaps N C. I mention these merely in order to callnbsp;further attention to the stone, as somebody with better eyes thannbsp;I have may be able to make out enough of the writing to obtain annbsp;idea to what it related. The placing of the Gaulish inscription sonbsp;close to the upper edge of the block is probably to be explained asnbsp;due to a wish to avoid as much as possible the earlier writing, andnbsp;even the area where it was known to have been.

X. Avignon (5). In the same Museum there is an inscribed column from L�Isle-sur-Sorgue in the Department of Vaucluse. The letteringnbsp;is bad, and the surface of the stone is so irregular, owing to holesnbsp;and scratches, that I have not succeeded in reading it so as to makenbsp;sense of it. In the Corpus, XII. p. 822, it is given asAAPCNNOPir !nbsp;OYeP�T�//MAP�/YI; but I was inclined to think that I detectednbsp;traces of an I after the second f of the first line, also that the letternbsp;following the tall T may be an O. Then comes a gap where therenbsp;should be perhaps two letters. Then I jotted down some strokesnbsp;in which I fancied I found a A and an A, but I concluded thatnbsp;Ilirschfeld�s MA cover the space and the traces of writing morenbsp;satisfactorily. After the second P� I seemed to see a C or thenbsp;beginning of 00, which would fill the gap before Yl, The guessesnbsp;may accordingly be read thus

AAr�NNOPiri

OY�P�TO///MAP�OOYI

The whole looks as if intended to be in the dative case, and if one take Hirschfeld�s OYCP�Te to be the better reading, which it maynbsp;well be, the readiest way to complete the name would be to suppose itnbsp;to have been Oaepereoa (= Uerete~u) dative of Oatpere-oy, � son ofnbsp;OvepeTos': compare OviXkoveoi in No. vi. Lastly, Mapeoovi couldnbsp;only be the dative of a word of the M-declension making in thenbsp;nominative case Mapeovs, which could hardly be anything else thannbsp;the Latin cognomen Marius, borrowed and adapted as a noun of thenbsp;Gaulish M-declension; but why Mapeoas rather than Maptoas is not clear.nbsp;Instances, however, of Greek e for Latin � are by no means wanting:nbsp;witness such ones as KaTrerwAtoa, Xeyewa, and others cited by Blass,nbsp;loc. cit., p. 34, and in Latin inscriptions there is no lack of � for ?, suchnbsp;as fecet for fecit, uteles for utilis, Veatori for Viatori, and many morenbsp;brought together in the Co?jgt;us, vol. XII. pp. 953, 954. The Gaulsnbsp;could doubtless readily pronounce Marius or Marius, but they maynbsp;have had a difficulty in hitting off Marius as a trisyllable, and gotnbsp;used to say Mar��s. Be that as it may, the whole, according tonbsp;the view here advanced, might be translated:�� To Adgennorixnbsp;Marius, son of Veretos.� The Gaulish, it is seen, comes as near asnbsp;it was possible without using a word for son, to the ordinary Latinnbsp;formula in such cases as the following: Aemili Calvini f. Sabiniani,nbsp;� Of Aemilius Sabinianus, son of Calvinus � (C. I. Z,., V. 6527) andnbsp;Devilliae Catulini fit. Titiolae, � Of Devillia Titiols^ daughter ofnbsp;Catulinus � (ib., XH. 2271). If these conjectures should prove well-founded, one might regard this inscription as pointing to the secondnbsp;or first century b.c., when the conqueror of the Cimbri, Teutones, andnbsp;their allies was the greatest name in the Roman world, and especiallynbsp;in the Rhone valley in Southern Gaul.

xi. Avignon (6). A piece of a column from the neighbourhood of Apt bears an inscription in mixed Greek and Latin letters, which arenbsp;now hopelessly illegible in part. Hirschfeld has tried them in thenbsp;Corpus, XH. p. 822, and I agree with him as to the latter part of lines 1nbsp;and 2 ; but he is wrong as to the word forming line 3, which he givesnbsp;as AAE with a suggestion of something to precede the A. The wordnbsp;is no other than the Latin VALE, with V and A ligatured, and thenbsp;L a little damaged. This, together with the portions of the othernbsp;lines fairly legible, will stand thus :�

........NITOYL

.... ARNOC

square, and is decidedly so at the end of the second line. It is the form used throughout the Cassitalos inscription (No. xviii) at Nimes ;nbsp;and it was probably not arrived at by angulating the rounded sigmanbsp;C, but by giving S a straight back : it is here transcribed C. Thenbsp;NO consists of a badly formed N with a little O in the top angle,nbsp;which is made wide like some instances of the Etruscan N in Nos.nbsp;xxxiv, xxxvi below; but the N in the first line is different: onenbsp;might perhaps read lOV rather than NO. The R is very degenerate, consisting of a badly made P with the stem crossed by a straightnbsp;line near the bottom of the bulge at the top: a somewhat betternbsp;instance occurs in the earlier portion of the first line, where I guessnbsp;NI TO VC, to be preceded by N ERA IP. The NE would be a ligature,nbsp;but the N portion is very doubtful: the E R are fairly certain andnbsp;probably end a word. As to A I, this is also Hirschfeld�s reading, butnbsp;it is just possible one should read N. He represents the next letternbsp;as a but I do not think it anything but a badly shaped Latin P,nbsp;and instances of it will be found beginning a potter�s name, Perimos,nbsp;in the Corpus, XIII. 10010. 1525. Before ARNOS Hirschfeld has annbsp;N, but I see there a ligature which might be read either AV or AN,nbsp;and preceding it I thought I detected an M ; but needless to say it isnbsp;very doubtful. I ought to have said that to the right of the fracturenbsp;in the first line Hirschfeld has an italic F, which I suppose should benbsp;the beginning of the line; but I gave it up. My conjectures, broughtnbsp;together with the utmost diffidence, stand as follows :�

In the Corpus this has been classed as a Celtic inscription, but one cannot be certain to what language it belongs, with the exception ofnbsp;the word VALE, with which I leave it.

Before quitting the Calvet Museum I may mention that a cast was shown me of an inscription supposed by some to be Celtic: I firstnbsp;heard of it from M. Maruejol, Conseiller General at Nimes, whose letternbsp;describes it as an � inscription rupestre, gravee sur un rocher qui hordenbsp;la Durance a Cavaillon (Vaucluse),� and he gave the reading

On looking at the cast I felt inclined to read A instead of A in the first word; but I missed visiting the original, which, though onlynbsp;about twenty minutes� walk from the railway station at Cavaillon, is

very difficult to find. This I learnt from a letter from another Nimes archaeologist, Dr. Michel Jouve, Conseiller at the Cour d�Appel;nbsp;who was spending his vacation at Cavaillon, and would have kindlynbsp;guided me to where the inscription is covered by the mud of thenbsp;Durance. Since he rediscovered it some years ago he has acted asnbsp;guide thither, among others, to M. Labaude, head of the Museenbsp;Calvet at Avignon, who has, I am told, contributed a note on it tonbsp;the M�moires de F Acad�mie de Vaucluse, 1903, p. 164. Dr. Jouve�snbsp;letter reached me too late for me to avail myself of his kind offer. Innbsp;any case, I do not suppose the inscription to be Celtic.

xii. Malauoene. An inscribed stone found at Beaumont, near Vaison, was taken to his house by an antiquary named M. denbsp;St.-Bonnet, and finally removed to his home at Malaucene, where itnbsp;and other antiquities collected by M. de St.-Bonnet are preserved bynbsp;his courteous and hospitable heir, M. Chastel. This inscription is allnbsp;in Roman letters, and as a whole it is in the Latin language: seenbsp;Stokes�s No. 24, where the initial letter has been read I instead ofnbsp;S, and C. I. L., XII. 1351, where the reading is more correctly given,nbsp;as follows:�

The right-hand top corner is broken, but after N there is still to be seen the foot of some such a letter as I, possibly E. The whole of thenbsp;right-hand edge of the stone is rough and broken, but I fanciednbsp;I detected a stop at the end of the second and third lines: in any case,nbsp;I do not suppose much is gone. As to the lettering there is nothingnbsp;much to be said, except that the S is long and sprawling; the tail ofnbsp;the last one passes below the line almost beneath the V preceding.nbsp;The F has its horizontal lines very near one another, and of aboutnbsp;equal length. This is one of the inscriptions which have been paintednbsp;in red, and as usual incorrectly painted, whence it is perhaps that thenbsp;long initials was misread I.

The interpretation offers considerable difficulties as it admits of being construed in several ways: (I) F probably stands for the Latinnbsp;vioxA fecit, and the Latin proper name Uirius may be taken as thenbsp;nominative. (2) Qoreto-uirius would make a Gaulish patronymic,nbsp;meaning, according to the analogy of other instances, �son of Voreto-viros �: the latter seems to be exactly represented in Welsh bynbsp;gwared-wr, � rescuer or deliverer,� from uoreto-, Welsh gwared, � deliver-

ance,� and uiros, Welsh (g)�wr, ^vir.' That being the meaning of uoreto, it does not appear how to construe it except as part of a compound ^oreto-uirms rather than attach it to Sumeli. (3) This latternbsp;word may be taken to stand for a nominative Sumelis, and one wouldnbsp;then construe Sumelis Uoretouirius as �Sumelis, son of Voretoviros�nbsp;and treat it as the subject of the verb fecit. But (4) it seemsnbsp;improbable that Subroni and Sumeli are to be severed, and if they arenbsp;construed together we have the following three possibilities to takenbsp;into account; (a) We construe Subroni(s) Sumeli(s) Voretovirius asnbsp;the designation of a single man : that would give the inscription onlynbsp;a minimum of sense. (5) Say Subroni and Sumeli go together as thenbsp;names of two members of the family of Voretoviros, and you may construe thus: � Sumeli(s) son of Voretdviros made this for Subro�, butnbsp;that sounds forced as you ought to have a word for brother, sister,nbsp;mother, or the like, (c) Take the alliterating names Subroni andnbsp;Sumeli to have belonged to a single person, and the whole may benbsp;rendered thus: Voretovirius made (this) for Subro Sumelis.

Of the possible interpretations�I am not sure that I have exhausted them�I give the preference decidedly to the last mentioned. It is doubtful, however, what it was that Voretovirius made; butnbsp;it was probably some kind of a building in the wall of which thenbsp;inscribed stone had been duly placed. So the building may have beennbsp;either a temple or a tomb: at any rate Hirschfeld thinks the inscriptionnbsp;admits of being regarded as a sepulchral one. Unfortunately thenbsp;names do not help one to decide whether the bearer of the two firstnbsp;was divine or human. Sumelis, however, seems to point to a femalenbsp;rather than a male. The prefix su or so, Irish so, Welsh liy, is largelynbsp;used in the Celtic languages to make adjectives implying the qualitiesnbsp;or characteristics of what is suggested in the ensuing portion of thenbsp;�'M-word. Take such examples as the following:�Gaulish su-carusnbsp;{C. I. L., XIII. 10010, 2408), O. Breton hocar (in Eu-hocar), Welshnbsp;hy-gar, � amiable, apt to be loving or friendly,� from car-, � to love �;nbsp;Irish so-chruth, � good as to shape, fair of form �; Welsh hy-hrydnbsp;(unattested), the contrary of Irish do-chruth, Welsh dy-hryd, � deformed,nbsp;hideous, ugly,� from cruth SLndp')yd, � form, shape �; Welsh Hy-wel, Ho-wel, �conspicuous, easy to see,� from gwel-ed, � to see�; Welsh hy-law,nbsp;� distinguished for his hand and its cunning, dexterous, handy, eiixd-Pinbsp;from Haw, � hand �; hy-barch, � venerable,� from parch, �respect �; and liy-fford, � having knowledge of the way,� whence hyfordi, � to put one onnbsp;the way, to direct and instruct.' Having regard to the majoritynbsp;of this category of words one might reduce them into two groups:nbsp;(a) those in which the prefix has the force of � good or desirable,� and

26 PROCEEDINGS OF THE BRITISH ACADEMY

(6) those in which the idea of � good or desirable � does not appear, but only the neutral one of � capable of being, doing, or suffering innbsp;a certain way.� Some, however, of the instances are hard to classify ;nbsp;take for example hydraul, which Davies renders �tritu facilis, con-sumptu facilis, �wptTrros �; that fits well enough into the second groupnbsp;until you come to the passage he had in view when he added thenbsp;Greek word. It occurs in the Mabimgion (Oxford edition, p. 55) andnbsp;reads � a hen ditlat hydreul tla�t ymdana�,� meaning � clad in oldnbsp;clothes, threadbare and poor.� This signification of hy-draul fromnbsp;traul, �wear, waste,� will not fit the first group though it may benbsp;Englished as �well worn,� and not into the second except by forcenbsp;of throwing back the capacity for being worn into the past. Now asnbsp;su- is supposed to have originally meant � good or well,� an inferencenbsp;drawn from the fact that su- and hu- occur in the same sense innbsp;Sanskrit and Zend, we have to refer Sumeli(s) probably to the firstnbsp;group of words with su and guess its meaning as well as we can.nbsp;What words there were in the Celtic languages with the combinationnbsp;mel one cannot tell, but one such for certain was the word for honey,nbsp;which in Welsh is still mel. If one fix on that I should gather thatnbsp;Sumelis meant one who had the leading attribute of honey, namely,nbsp;sweetness, let us say in his or her speech and disposition; that is,nbsp;Sumelis would have meant honey-like or sweet-spoken; the derivativenbsp;somilse meant in O. Irish � sweetness, dulcedo?

So much as to the meaning of the name; the examination of the inflection places one on a surer footing. In early Brythonic andnbsp;Gaulish the declension was probably not very different from whatnbsp;it was in early Goidelic, and the latter can be constructed as farnbsp;as one wants from Mediaeval Irish which was nominative mil, genitivenbsp;mela, dative mil. The Latin was mel, genitive mellis, and the Greeknbsp;pAi, genitive pi^Xtroy, both neuter, while Welsh mel is masculinenbsp;and the Mod. Irish feminine; but supposing the Irish and Welshnbsp;were also originally neuter, the Irish ferms were, for early Goidelic,nbsp;nom. meli, gen. mely-as, dat. meli. If the word was not neuternbsp;then the nominative would be 7neli-s, and whether mel was neuter ornbsp;not the personal name would be Sumelis with a dative Sumeli, thenbsp;form which we have here standing in the Latin inscription as thenbsp;epithet or surname, as I take it, of Subroni.

Analysing this latter name into Su-broni in the dative case, the question is what we are to make of bron-i. One mechanically thinksnbsp;of the Greek ac�cfipaiv, neuter cr�(f)por, genitive (rdeppovos, dative ardtlipovi,nbsp;� of sound mind, discreet, prudent.� However, I am not rashly goingnbsp;to identify the Greek o-co- in this word with our Celtic su-, as I am

aware that there are difficulties in the way; but the second element remains attested by such words also as ev(jgt;p(iov, evtppov, �cheerful,

making glad, well-minded, kindly,� -npoipputv, -ov, gen. 'Trp�(j)povos, � well-wishing, gracious, zealous,� aifipoov, -ov, genitive a(fgt;povos, � senseless, frantic,� (ftpovricris, �purpose, high character, good sense, practicalnbsp;wisdom,� all connected with 4gt;priv, gen. ^pev�s, �the midriff, the heart,nbsp;praecordia, the breast, the seat, as it was supposed, of the mentalnbsp;faculties.� So one may regard Subroni as of the consonantal declension�nom. Subro for an earlier Subron-s, and dative Subron-i whichnbsp;appears to be the reading in this inscription. But the cognatesnbsp;are not confined to Greek : we have them also in the Celtic languages,nbsp;and among them may be mentioned the Welsh bryd, � mind, thought,nbsp;purpose,� Irish breath, �judgement, verdict�; and Stokes (s.v. bera, bra)nbsp;interprets the Gaulish vergo-bretos as �iudicium exsequens'�, fromnbsp;Welsh hyd is derived ded-fryd, � a verdict,� and hy-fryd, �to one�s mindnbsp;or liking, pleasant, agreeable,� but explained by Davies as �hilaris,nbsp;amoenus, (v(j)pagt;v.' Another derivative from the same root is to benbsp;found in the jBparovb^ already noticed in connexion with No. vii, p. 17nbsp;above, where the jSparov portion of the word has been referred to thenbsp;same origin as Welsh brawd, �judgement or doom,� Irish brath, thenbsp;equivalent of which Stokes finds in such Gaulish names as Bratu-spantium, Mandubratius, and Cassibratius. In connexion with thisnbsp;suggestion his editor Bezzenberger suggests the possibility of referringnbsp;to the same origin the Greek word ^prjr and 0. Norse grunr, � ahnung,�nbsp;and gruna, � beargw�hnen ��this, should it prove sound, would go tonbsp;show that our Celtic words could not have anything to do with thenbsp;root from which Latin fero, Greek (jgt;�pagt;, � I bear,� and their congenersnbsp;come, inasmuch as it suggests as stems rather ghtiron, ghuren. However that may be, the Neoceltic languages have also cognates in whichnbsp;the nasal appears, namely Welsh barn (fern.), �judgement,� and Irishnbsp;bam (masc.), � a judge.� But we have no stem exactly to match thenbsp;bron of Subroni as the Greek forms do. The meaning of the namenbsp;would be �good at thinking, possessed of sound judgement, prudentnbsp;and wise.�

xiii. Notre-Dame du Geosel, formerly Grasellus in Latin, near Malaucene. A little beyond M. Chastel�s house one comes to thenbsp;church of Notre-Dame du Grosel or Groseau, situated in one of thenbsp;most picturesque nooks I have ever seen; and there in front of thenbsp;entrance is a mutilated inscription on a stone which forms part of anbsp;structure to hold a cross. According to one of the authorities quotednbsp;in the Corpus, XII. p. 824, the stone served some time previouslynbsp;as the support of the Roman altar (Tautel romain) in the little chapel

of St. John the Baptist, said to be the oldest chapel of the Groseair. Further, a document quoted by Holder, under Grasellus, appears tonbsp;carry that form of the name back to the beginning of the seventhnbsp;century, and speaks of a � monasterium in loco nuncupato Grasello �:nbsp;it seems to be the same place, though Holder does not refer to thenbsp;inscription there. The spring of the Groseau is a little further thannbsp;Notre-Dame, and for lack of time I had regretfully to leave itnbsp;unvisited. The reading given in the Corpus is as follows:�

I ought, however, to say that for typographical reasons this ascribes a little too much to the Corpus, which gives only the first limb of thenbsp;final A of KANTENA, and only the second limb of the first A in thenbsp;second line; but even that was rather more than I can be quitenbsp;sure of. With regard to the original, I found the bottom of the Y innbsp;the fourth line damaged so that the letter now looks more like a V.nbsp;I could not be sure of the B of jSparovbe, though it must have beennbsp;there formerly; and the same remark applies to the T at thenbsp;beginning of the third line, but I thought I could trace the Pnbsp;following it, also the inner outline of an I at the edge where thenbsp;reading quoted by Stokes, No. 12, gives a B, and makes the wholenbsp;line into MACEAOYB. The reason for that reading is not evident,nbsp;and Hirschfeld is probably right in making it into FpacrfAov, whichnbsp;one may possibly complete into rpaaeKovi.

The C given in the Corpus at the end of the second line is scarcely to be traced now, and the letter preceding it is given as O. At firstnbsp;I took it to be E, but on examination it appeared to be an O, thenbsp;circle of which has been squared : in fact most of this inscription ha snbsp;been tampered with and scratched in order, I suppose, to renovate it.nbsp;I jotted down the whole of the line as yielding traces makingnbsp;I AAIAKOC, but that can hardly be the correct reading: it has toonbsp;many lambdas. Possibly the A A should be regarded as representingnbsp;a M : we should then have -ipikKos, but I cannot guess what thenbsp;whole word would be. The same difficulty would meet us if wenbsp;treated IA as traces of f A and the third A as A, for that wouldnbsp;yield -yaXtaKos: what might be the whole word.'� A simpler conjecture would be to read -lAAIAKOC, which might be completednbsp;perhaps as B/gt;iXXta/cos; Holder gives Brilliacus as the name of more

than one place in Gaul, but none near Malaucene. Better than all these guessings perhaps would it be to regard the beginning of thenbsp;second line as occupied by the end of the name beginning in the firstnbsp;line : this would allow us to read the second vocable as simply lAXia/co?nbsp;related to such names as Illius, fem. Illia, for which see Holder�snbsp;Altcelt. Sprachschatz, as also for instances of Illio-marus and Illio-marius: see likewise IXXarortaKos in No. xvii. What remains for certainnbsp;of the first line is AOYC, but immediately preceding the lambda therenbsp;are traces of a letter which may be TT, but I feel far from certain.nbsp;Now if XotJ� ended the name in the nominative, which ought here tonbsp;be the grammatical case, it would be a noun of the w-declension ; butnbsp;the chances are rather against a proper name of that declension�stillnbsp;more against two such in one brief inscription�and in favour ofnbsp;regarding the name as ending in the next line.

The treatment here of ov calls for a remark in passing. Before a consonant, as in Xovs and ^parovbf, it had probably the usual soundnbsp;of u, while in Tpaae\ovL or Tpadekov it must have been the diphthongnbsp;OM or ow, unless we are to suppose that it had been reduced in thenbsp;pronunciation to �. The whole may be represented as follows ��

And interpreting the last two words as was suggested in the case of No. vii one may render the inscription thus�^////// lusos Illiacosnbsp;(gave) firstfruits to Graselus by his decree.� The verb dede is herenbsp;left out, but the construction of the sentence is perfectly clear, thanksnbsp;to the case endings.

The third line giving the name of the recipient of the cantena naturally attracts attention in spite of its fragmentary state; fornbsp;apart from the question as to the ending, the dative is here thenbsp;centre of interest, because it seems to identify the name of thenbsp;divinity with that of the spring, whence that of the �locus nun-cupatus Grasellus,� was derived^. The nominative would accord-

* This relates to Aredius, bishop of Vaison in the seventh century, and the context will he found given in Pertz�s � Monumenta Germaniae Historica,nbsp;Diplomata, i. 6� (= p. 67). In later times the place became a favourite resortnbsp;of the Popes of Avignon ; but now hardly a solitary cyclist finds his way there.nbsp;Doubtless this will not always be so, as an enterprising society of Frenchmennbsp;has lately been organized to acquaint their countrymen with the beauties ofnbsp;French scenery.

30 PROCEEDINGS OF THE BRITISH ACADEMY

ingly have been Tpaa-fXovs, Graselus, of the M-declension, and as the spring is now le Grosel or le Groseau the divinity of oldnbsp;was probably a god rather than a goddess. With regard to thenbsp;change of vowel from Graselus to Grosel it is relevant to mentionnbsp;that I ascertained the fact that the local pronunciation still makesnbsp;the vowel in question more a than o: it is decidedly nasal, for thenbsp;word is sounded Gra^zeu. Possibly the ancient form of the namenbsp;would have been more correctly written Granselus in spite of thenbsp;Grasellus of a later document: compare cesor, Masuetm, and mesisnbsp;for censor, Mansuetus, and mensis in Roman inscriptions frequently.nbsp;Lastly, when one bears in mind the connexion of the stone withnbsp;the Roman altar in the ancient chapel of St. John the Baptist, it isnbsp;natural to infer that the early missionaries adroitly converted thenbsp;rustic water-god Graselus into the Baptist of their own faith, thoughnbsp;the former may have been a distant relative of Apollo Granus.

xiv. Saignon. Some four kilometres from the town of Apt, in the Department of Vaucluse, nestles the village of Saignon under thenbsp;threatening crest of a once fortified cliff, and within the church is anbsp;hopelessly mutilated inscription. It is in the wall near the door asnbsp;you enter, and I found it above the level of my eyes. It appears tonbsp;have been plastered over for a time, which explains why the editor ofnbsp;the Corpus treats it as lost; in fact it is not very long since it wasnbsp;rediscovered by M. Ginestou, the hospitable and learned cure ofnbsp;Saignon. The editor states (C. I. L., XH. p. 822) that the stone wasnbsp;originally found in the gardens of the presbytery about the yearnbsp;1867, and the reading he gives shows much the same letters andnbsp;portions of letters that I thought I saw. The following are mynbsp;guesses :�

The last letter visible in the first line may be regarded as C or an imperfect 0: the Corpus has the latter, and indicates that the linenbsp;did not end with it. The I in this line may have been a T with thenbsp;top imperfect, but I copied it as I. The B seemed imperfect also atnbsp;the top, but it has the shape characteristic of that letter in ournbsp;inscriptions, especially those at Nimes. The Corpus marks the A asnbsp;imperfect, which did not attract my attention : there is no doubt,nbsp;I think, as to its identity. The coming together of AB is remarkable,nbsp;and recalls such Gaulish names as Adbogius, Adbucillus, andnbsp;Adbucietus. Now, AABOTIO- will not fit the lacuna in thenbsp;middle: probably AABOKETO- or AABOKIETO- would do better;

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 31 but we should want it longer at the end, some such a form asnbsp;AdjSoKeroovi^, or else AS^oxteroy, followed by the beginning ofnbsp;a patronymic genitive ending with the OYEI, let us say, of thenbsp;second line; this last line ends with a ligature which has to benbsp;resolved into AN. The Corpus makes the last limb of the ligaturenbsp;too nearly perpendicular, as if it were A with an I accidentallynbsp;attached to it. This is possibly the case with the beginningnbsp;of the third line : it is rather imperfect, but it looks almost morenbsp;like AI than AN, the whole word being as I read it AlOTEI ornbsp;ANOTEI. It is right, however, to say that the Corpus reading isnbsp;AlOYEI, where we differ as to T or Y, as to A or A. I was notnbsp;aware of the reading with Y, and I copied the letter as a good T, sonbsp;far as I can gather from my notes.

To these details of the reading must be added some attempts to interpret the whole. The first word to claim one�s attention is thenbsp;verb KapviTov or carnitu, which occurs also in Nos. xxxiv and xxxvi,nbsp;where Dr. Stokes translates it by � congessit � and � heaped together.�nbsp;The tense in u is the same as that of ieuru, which has been noticed atnbsp;pp. 5, 16, and the syllable it may be compared with that of Latinnbsp;habifo, vocito, as compared with habeo, � I have or hold,� and voco,nbsp;�I call.� We have earn- left us, which is doubtless of the samenbsp;origin as Welsh earn, � a heap or cairn.� Thus it would seem at firstnbsp;sight as if we might render the inscription in some such a way asnbsp;� Adbogio-uix has here buried Vimatica�; but that will not do, sincenbsp;in the two instances where the accusative is expressed, namely twicenbsp;in No. xxxvi, it is not the accusative of the person but of the thing.nbsp;We .must accordingly find an accusative of the latter kind in thenbsp;second line. This leads to two interpretations, in which the digraphnbsp;e I has probably to be treated both times as pronounced i, as in thenbsp;case of �i�gt;pov, at p. I�:�(a) Without a genitive, one would have tonbsp;treat IJimatiea as a word meaning a grave, or perhaps a pyre, andnbsp;consider the legend to have run somewhat thus ; � Adbocietouix pilednbsp;up a uimatiea for Annotis.� (h) With a genitive, such as Mogouinbsp;(nom. Mogovios) or Anovi (compare Holder�s nom. Annous,nbsp;Annovos); and a dative Anoti (nom. Anotis : compare Holder�snbsp;A nnotius), the rendering would have to run on this wise : � Adbocietosnbsp;son of Anovos, piled up a matica for Annotis.� The proper names arenbsp;inserted simply to help to indicate the syntax of the sentence : I cannotnbsp;make a more definite suggestion, because I do not know whethernbsp;uimatican or matkan is to be regarded as the accusative, or what either,nbsp;in case of a decision being made between them, would mean. Onenbsp;might, however, guess that the idea of a funeral pile, if expressed by

a collective feminine matica (better in that case mattica), would harmonize with the Mod. Irish word maide, meaning �wood, a stick,nbsp;a staff.� It is supposed to come from an O. Irish maite, for an earlynbsp;combination mazdio-s, of the same origin as English mast and itsnbsp;congeners : see Stokes�s Urk. Spr., p. 203.

I may mention that the cure called my attention to a remarkable benitier in the church: it is of white marble, and its shape suggestsnbsp;to him that it was originally a druidic altar. One of its marginsnbsp;bears an inscription so far effaced that I could make nothing of it;nbsp;but I was so tired and so afraid of missing my train for Avignonnbsp;that I hardly gave the lettering a fair trial; so my failure is no proofnbsp;that it cannot be read.

XV. St.-Remy-de-Peovence (1). The little town of St.-Remy (Bouches-du-Rhone) is associated with the ancient Glanum, and hasnbsp;in its immediate neighbourhood some of the most interesting ofnbsp;Roman remains ; but the inscriptions which I went to see are in thenbsp;Museum at the Mairie. One of them is on a stele reading asnbsp;follows: see Stokes, No. 11 ; C. I. L., XH. p. 127 : �

That is Urittacos Elusconios, � Vrittacos, son of Eluse� or Elusconos.� Urittacos is a name of the same origin as Ate-uritus, Ate-urita, alsonbsp;At-urita, together with related forms duly recorded in the Corp^is,nbsp;XHI. 10010. 2096, 2907 ; and from Ireland may be added the genitivenbsp;Ape-vritti: compare also xxxiii�� below. Of the patronymic Elusconios I have nothing to say.

xvi. St.-Remy (2). Another stele at St.-Remy has its top broken off, but the inscribed portion of the stone is intact, and thenbsp;lettering, though ruder than the previous one, is clear enough. This,nbsp;however, does not spare one considerable trouble with the very firstnbsp;name, for it presents a ligature which has not been satisfactorilynbsp;resolved. As far as concerns the strokes involved, they would benbsp;covered by supposing the ligature to have meant NN, but this isnbsp;unlikely, as the next letter is M, and the whole name would be in thatnbsp;case BINNMOC. The reading adopted by Stokes, No. 10, is BIMMOC,nbsp;and other authorities have treated it in the same way; but by sonbsp;doing they omit one limb of the combination. Two other readingsnbsp;are mentioned in the Corpus, XII. p. 127, as BIM V MOC and B1N V MOC,

CELTIC INSCRIPTIONS OF FRANCE AND ITALY S3

both of which introduce a Latin V into a legend which is otherwise all in Greek letters. A less violent conjecture would be to take thenbsp;M which is there to have had a line joining its two first limbs so as tonbsp;make it into a ligature for AM, and to assume that this line wasnbsp;overlooked by the inscriber who had not carefully scanned the textnbsp;given him to carve. The inscription would then stand thus:�

BINNAMOC

AITOYM

Apeoc

That is, �Binnamos son of Litumaros,� Litumareos being a patronymic derived from Litumaros: compare OvAAoreos in No. vi, pp. 13, 15.nbsp;Both Litumanis and Litiimara are cited by Holder. Binnamos onenbsp;would possibly have to derive from the same source as Irish hinn ornbsp;hind, � sweet of voice, melodious �; and as regards the formation of thenbsp;w'ord compare such Gaulish personal names as Bladamus, Cavvama,nbsp;Clutamus, Uxama, as to which see C. I, L., XIII. 1316, and Holder,nbsp;s.v. -amo-, -ama.

xvi�. The Corpus mentions another Celtic inscription which should be at St.-Remy, but the Maire, who is familiar with the antiquities innbsp;the local Museum, knows nothing about it: he was very ready to assistnbsp;in the search, but it was all in vain. In the Corpus, XII. p. 127, it isnbsp;given from a manuscript (Romyeu, f. 95^) as follows�ON �OYOHOnbsp;AlOYI � BRATOY. Here the O possibly means an O, for thatnbsp;vowel is now and then ornamented with a point in the centre. Thenbsp;gap before it may be due to careless copying or else ON is the end ofnbsp;some longer word, the rest of which was illegible or broken off. Thenbsp;seeond and third Y have been copied as a V standing on a shortnbsp;horizontal line, but the peculiarity is probably due to the copyist.nbsp;The inscription is imperfect not only at the beginning but also at thenbsp;end, which was probably fipaTovbe Kavrtva, as in the case of the Groselnbsp;inscription, which with its FpacrfXoui /3parovde ravreva helps one tonbsp;construe the present one. In fact one perceives at once that 8 tout isnbsp;in the exact position for a dative, and this proves to be so; for wenbsp;have only to suppose the not uncommon substitution of i for � and wenbsp;have d�7ti, the dative of the word d�ui-s or dvui, � goddess.� In thenbsp;first Rom inscription I have found the vocative asnbsp;nbsp;nbsp;nbsp;= deutii,

which has been read deei. In Irish we have the word in a genitive dca in Fir Dea, �the Men of the Goddess,� meaning the Tuatha D�nbsp;Danann, and dea Dcchtiri, �the goddess Dechtire�s� (Bk. of Leinster,nbsp;f. 123��); for deui would yield in early Irish the genitive deuy-as,nbsp;which according to the prevailing rule in that language had tonbsp;become d�a in the course of time.

M 3

Further search should be made for this most interesting inscription, beginning with the examination of the Romyeu manuscript, to whichnbsp;the Corpus refers.

The Corpus, XH. p. 127, places among the Celtic inscriptions one found �prope Glanum in cippo quadrato iuxta villam quae diciturnbsp;le Mas de Durand^ which I did not visit; but it adds that the language may be Greek, and the reading given is ///YP/AKA/HOC/Y� A,nbsp;which suggests the name of Heracles, but nothing Celtic as far asnbsp;I can see.

xvii. Nimes (1). One of the most remarkable inscriptions in the Museum at Nimes is one that was formerly to be seen in the templenbsp;of Diana : it is stated to have been originally found in 1742 near thenbsp;great spring which forms one of the most remarkable features of thatnbsp;ancient city: see C. I. L., XH. p. 383, Diet. Arch. No. 1, and Stokes,nbsp;No. 7. The letters are here and there imperfect, and in two or threenbsp;places wholly gone; the Corpus produces it accurately so far asnbsp;it goes, but the Dictkmnaire arcMologique de la Gaide, No. 1, hasnbsp;tried to complete it, and has not done it satisfactorily. To thenbsp;best of my belief the original was as follows ;�

The work of restoring the inscription is rendered comparatively easy by the MS. notes of Dardalhion, dating about 1745. M. Maruejolnbsp;kindly enabled me to find them in the public library. The secondnbsp;line is all legible, and the chief lacuna in the first line nearly coversnbsp;the letters POSI which Dardalhion�s reading supplies. But even thenbsp;perpendicular of the P can be traced in its proper place, and the lowernbsp;end of the I is also there: the P has wrongly been guessed to be a B.nbsp;The top of several of the letters that come later in this line has beennbsp;damaged, among others that of the Y which Dardalhion has accordingly copied as an I ; but as a matter of fact one can still detectnbsp;the beginning of the fork of the Y on the stone. The K of KOSnbsp;is also imperfect, but of special interest is the fact that the lowernbsp;slanting arm of that letter does not reach down to the ground-levelnbsp;of the lettering as a whole: this makes it possible to suppose that thenbsp;inscription began with a K and not with a gamma as has been usuallynbsp;thought. In any case, whatever it was, it had become very uncertainnbsp;before Dardalhion saw it and copied it as a doubtful I. This last,nbsp;however, could not fill the space, and the more fitting character seemsnbsp;to be one of the following, r, H, K, ff, P, Y, Without any. reason

of special force I have selected K, but I should be glad to see a better case made for any one of the other letters. This was the firstnbsp;inscription I found with the sigma written S, and I may mention thatnbsp;the E is here formed with its perpendicular protruding in bothnbsp;directions beyond the horizontal bars. Lastly, the sides of the trianglenbsp;of the delta are produced upwards so as to cross one another and formnbsp;a forking at the top; the base of the A is produced also at both endsnbsp;and twisted a little upwards. Altogether the lettering is morenbsp;pretentious than in any of the previous instances. I may addnbsp;that Dardalhion gives me the impression of being an accuratenbsp;man, and as an instance I would mention his copy of the deltanbsp;of BPATOYAE with a short tag which hangs, as it were, from thenbsp;middle of the base of that letter; I cannot suppose it, thoughnbsp;well defined, to have had any meaning, but anyhow there it is,nbsp;cai�efully reproduced by Dominus Dardalhion.

Thus far of the lettering: the inscription means � Cartaros II-lanuiacos gave (this) to the Nemausian Mother-goddesses by their decree.� What Cartaros did give was probably the gift of a buildingnbsp;in whose wall the inscribed stone was inserted or else some gift whichnbsp;was not intended to be separated from the stone. Cartaros I shouldnbsp;refer to the same origin as the plural carti in the first Romnbsp;inscription, where I have conjectured that it means strong ornbsp;powerful {Celtae, p. 38). lUanuiacos has been treated by Dr. Stokesnbsp;as the genitive of Illanoviax, but I am not sure of the existencenbsp;of such genitives in Celtic, and it seems preferable to regardnbsp;WkavomaKos as an adjective agreeing with Kaprapos, and so withnbsp;lAAtaKos in No. xiii. What may have been the precise force ofnbsp;formations in -ci/coy when used in this way I am unable to say. Holdernbsp;supplies related names in Illanuissa and Illanuo, gen. (Latin) Illa-nuonis, which he dates at Cologne in the first half of the fimtnbsp;century of our era. Dede we have had in No. vii, and as to matrebonbsp;that corresponds to Latin matribus, while the feminine adjectivenbsp;derived from NamaiLSOS has its dative plural answering to such Latinnbsp;forms as dominabus, Jiliabus, and the like. Bratude we have hadnbsp;before, and both times with the dative coming immediately beforenbsp;it; see also xxii.

xviii. Nimes (2). A stele found in the city in 1876 has the writing on two contiguous faces of the stone, and so placed that each linenbsp;is partly on one face and partly continued across the edge on thenbsp;other face, an arrangement which the Corpus seems to suggestnbsp;in the case of a Latin inscription at Nimes, namely XII. 3656:nbsp;I do not recollect having noticed either this or No. 3964. The

reading of the present inscription is as follows: see Stokes, No. 8, and C. I. L., XH. p. 383

The lettering reminds one of that of the one from Apt: see No. xi above. It has the square sigma derived from Z as is indicated bynbsp;the protrusion of the two horizontal lines a little behind the perpendicular, which is especially visible as a kind of heel at the bottom.nbsp;The koppa form of the P in that inscription has here become a simplenbsp;loop on a stem. On the other hand, one is reminded of the Saint-Remynbsp;inscription from Romyeu�s copy by the fact that here the upsilon tendsnbsp;to take the form of Latin V, from which it is, however, distinguishednbsp;by the horizontal finish of the T being retained. Dr. Stokes endsnbsp;his reading with the A A of the fourth line, and adds that the restnbsp;is � almost certainly wrong.� It is repeated, however, in the Corpus,nbsp;and is right except that it ends with T11 the last of which is notnbsp;a letter but an accidental scratch. At any rate that is what it seemednbsp;to me to be, and I thought I found a point before A A and beforenbsp;EINO. It looks as if the author of the inscription treated thenbsp;familiar sequence dede bratude cantena as requiring no punctuation,nbsp;and as if he reserved it for the part which was special to this case,nbsp;Aagt Eivovi, Lastly, I ought perhaps to mention that in the Corpusnbsp;the O of OYEPCI and of ATOY is provided with a little tag stickingnbsp;out of the highest part of the circle of that letter. This shouldnbsp;be the apex: it escaped me, and I do not understand what it cannbsp;have meant, at all events in the former instance.

There is room here for the same doubt as to the pronunciation of the two syllables ovi as in No. xiii (p. ^9), that is, whether they werenbsp;sounded ui or oui. But in either case the word would be an instancenbsp;of a dative corresponding to which the nominative must have beennbsp;Etrovs or Einus of the M-declension, which draws no distinction betweennbsp;masculine and feminine. It is of obscure origin and meaning, but wenbsp;have possibly a kindred form in /no-reixs, cited by Holder, andnbsp;especially in Sp(urius) Inus, and in ... INOVCI . A(eSe) which henbsp;cites from the same part of France. Ewow should be an adjectivenbsp;qualifying Aagi, unless they are the names of two different persons,nbsp;between which we should naturally insert a conjunction. So we come

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 37 to Aajut, and that is somewhat more promising. The nominativenbsp;should have been either Lama or Lami-s : the latter would be epicenenbsp;while the other would probably be exclusively feminine. In eithernbsp;case one is reminded of the Latin and Greek Lamia �a witch,nbsp;a bugbear, a blood-sucking monster.� From Benwell near New-castle-on-Tyne we have an ancient inscription of a delightfullynbsp;brief character LAMIIS lt;r TRIBVS (C. I. L., VII. 507) which one isnbsp;tempted to render � To the Witches Three,� such witches as the weirdnbsp;sisters whom Shakespeare pictures meeting Macbeth. However, thenbsp;word Lamia is less simple than Lama or Lamis, for either of which itnbsp;would fit as a derivative. If we are to look for an Aryan etymologynbsp;for these names, one would be inclined to compare the Goidelicnbsp;deponent verb lam\�-r, � I dare,� rn-laimuv, � avdeo^ ni con-Zai/nemmar,nbsp;�now audemus^ and the Welsh ZZo/asu, �to dare�; see Stokes�s Urh.nbsp;Sprachschatz, p. 240, and Zeuss�s Gram. Celtica, pp. 7% 438*.nbsp;Cassi-talos is analogous to Danno-talos and other compounds withnbsp;talos, supposed to mean brow or forehead. Dr. Stokes has treatednbsp;it as meaning � fair-brow,� as suggested by M. d�Arbois de Jubainville;nbsp;but the meaning of cassi- is not at all certain. Uersicnos is a patronymicnbsp;signifying � son of Versos,� but the etymology of the latter is obscure :nbsp;Dr. Stokes compares Sanskrit varsh�yas, �upper,� and other wordsnbsp;supposed to be cognate with it. Without going into these detailsnbsp;the inscription may be rendered � Cassitalos son of Versos gave first-fruits to Lamis Finns (or Lamis and Finns) by her (or their) decree.�nbsp;Taking them to be two, I should not suppose them to be of the classnbsp;of Mother-goddesses, for in the latter case they should be three rathernbsp;than two. So one would have to treat them either as a god andnbsp;his paredra or, better, as a goddess and her son. Lastly thenbsp;Irish man�s name Mug-Lama, � Servus Lamiae^ decides, by meansnbsp;of its genitive Lama (= Early Lamjjas), for Lamis as againstnbsp;Lama.

xix. Nimes (3). On a small piece of brick-like substance in the Nimes Museum is the following fragment: see Stokes, p. 64; C. I. L.,nbsp;XII. p. 833

The M is mostly gone, so is the T in the second line, and somewhat less so in the last line, which ends with N, and not M as it has sometimes been represented. The second line suggests TOOYTIOYC, sonbsp;the inscription was probably Celtic.

XX. Nimes (4). Another Nimes inscription, now in the Museum, reads as follows: see Stokes, No. 9, and C. I. L., XH. p. 383 :�

ccKirro

peizKo

NAIAAe

That is, � Escingorix son of Condillos �: it was probably his tombstone. The stone is a narrow one about a yard long with the top rounded. The lettering is clear and well cut. Note that here et wasnbsp;probably pronounced f, and compare No. vi (p. 14): Celtic isnbsp;probably to be everywhere pronounced rix, genitive rlgos.

The man�s name should mean king or leader of escingi, and the latter word, according to M. d�Arbois de Jubainville, should mean thenbsp;warriors who sally forth to attack the enemy. Escingos was a fairlynbsp;common name and it stands for Ex-cingos: see p. 20. Condillosnbsp;should be a derivative from some such a name as Condos of whichnbsp;Holder gives instances.

xxi. Nimes (5). A fragment of a vase found at Nimes is now in the Museum, and shows a nearly illegible inscription which may possiblynbsp;be Celtic: this is my copy of it:�

MATIACo ....

KONNoYBP....

But I must explain that of the first letter there is hardly any more left than would cover a lambda. Then the TI are somewhat doubtful,nbsp;and may perhaps be IT ; the A following is also doubtful, and the Cnbsp;is very open, having in its bosom what may be a small o thoughnbsp;it looks more of the shape of a D. So the whole line may have beennbsp;AATIACO or AAITACO, but I prefer the former guess. As to thenbsp;second line, what I have transcribed NN has this appearance, C'C', andnbsp;I know not what to make of the character except a sort of exaggeratednbsp;Etruscan ['I : possibly it may be a gamma. The letter following thenbsp;B has the appearance of the little triangle forming sometimes the topnbsp;of a P, but I could not trace the stem below and detect therenbsp;another /3parov8e. After KONNO or KOffO, I thought I sawnbsp;a small point, but it was too low in the line to have, I think, beennbsp;intended. I have not succeeded in finding this fragment in thenbsp;Corpus; but it somewhat reminded me at first of No. 5885 Ad. onnbsp;a stone found near Collorgues (Gard). The editor gives it as

A A TT A 0 and suggests that it is a factitious production intended KOAAOTPr, for the glory of Collorgues.

xxii. Nimes (6). At St.-Come or Cosme, near Nimes, was found, in 1886, a fragmentary inscription, which is now in the Museum there :nbsp;see C. I. L., XII. p. 833. It reads as follows with the first portionnbsp;of both lines gone, perhaps one third of the whole;�

......AAPESSIKNOS

......YIBPATOYAEKA

Whether KANT ENA was written in full one cannot say, but the formula coincides partly with that of xvii (Nimes 1), and more closelynbsp;with that of Le Grosel, No. xiii; for it has Ka{vTeva) and may havenbsp;had 8e8e also. The Yl is doubtless the end of a dative in oai or oovinbsp;of the name of the divinity to whom the gift was made: what thatnbsp;name was one knows of no means of discovering. The lettering is verynbsp;like that of the former inscription: it has not only the same B butnbsp;also the same K and A ; the curl at the ends of the base of the A isnbsp;here to be detected also in the A.

Adressicnos means the �son of Adressos,� a name which seems to claim kinship with such forms as Reso, Ressius, and Ressi-maros, whichnbsp;appears to be the same name that is found written also Redso-maros:nbsp;see Holder�s instances.

xxiii. Nimes (7). A stone similar in shape to that of Escingorix, No. XX, except that the top is hollowed, is to be seen in the Museum,nbsp;and reads simply KPEITE in clear letters. Holder refers it tonbsp;Redessan in the Department of Gard, and adds that it dates notnbsp;before the second half of the second century of our era. It recallsnbsp;the Irish woman�s name Cr�d, from which another female name, Cr�ide,nbsp;seems to be a derivative : see 0�Grady�s Silva Gadelica, i. Ill; ii. 498.

xxiv. NImes (8). An inscribed stone was discovered more than twenty years ago in the wall of the hermitage ofnbsp;Notre-Dame-de-Laval, near Collias (Gard), and is now (1)nbsp;in the Nimes Museum. See Stokes, No. 13, and- C. I. L., iC OM Onbsp;XH. 5887, where two readings are given, the better CPIOYnbsp;of which runs as given here in the margin, with the MANnbsp;initial character represented as a ligature consisting //ANnbsp;of a reversed E and a K : the reading is M. Rochetin�s. /// O ///

How he and M. Germer, the discoverer of the stone, N Kj/ 0 A satisfied themselves that they found OAl in the �A6 BPATOnbsp;first line I am not quite able to understand, unless YAE KANnbsp;they were in some way influenced by the name of TEN//nbsp;Collias. This version has been improved by M. Maruejol,nbsp;who has coloured the lettering on a cast which is placed near thenbsp;original in the Museum; it runs as in (2), except that the character be-

(2)

(1) 3(//AI0

(3)

(4)

(5)

(6)

(7)

(8)

(9)

CPIOY

MANe

OCAN

AOOYN

NABOA

GAeBPATO

YAGKAN

TGN

tween the initial ligature and AI is rather a nondescript filling the lacuna which I leave there. I examined the inscription with great care, andnbsp;found that I was almost in agreement with him, excepting as to thenbsp;first line which I will leave alone for a moment. I am not convincednbsp;that the G at the end of line 3 is there, and in anynbsp;case you have to look for it over the edge, while atnbsp;the beginning of the next line I detected an I. Herenbsp;the inscribe!� has given us the neuter singular Kavreigt;,nbsp;not the plural Kavreva: compare No. xxiv** andnbsp;such forms as Latin nomen, plural nomina. Thusnbsp;far our difference attaches to the patronymic,nbsp;which I regard as being PIOYMAN|IOC, while M.nbsp;Maruejol reads PIOYMAN|G|OS. I prefer PIOY-MANIOC, and render it �son of PlOYMANOS,�nbsp;that is Riumanos. To go back now to the firstnbsp;line, I cannot improve on the suggestion that itnbsp;begins with a ligature of 3K : it might possibly be ZK for ZK, butnbsp;that is not probable. Next comes a combination which baffled bothnbsp;M. Maruejol and me, but since then I have come across it in thenbsp;thirteenth volume of the Corpus, namely, in No. 5465, in the Dijonnbsp;Museum. The editor suggests the values N, Nl, or IXI, but, it is sonbsp;situated, that it seems there to mean NI. The name in the inscriptionnbsp;seems accordingly to read DASILLINI, but the whole has unfortunately not been interpreted. The combination consists of an Nnbsp;with a long diagonal, and with an I bisecting that diagonal at rightnbsp;angles, with the result that the whole looks rather like IXI, thoughnbsp;it really means |X) = NI or IN. In the Dijon instance, the valuenbsp;required appears to be N I, while in ours that of IN seems to fit better.nbsp;The letter which follows is so faint that I have not made it out withnbsp;any certainty: it may be another N tagged on, or merely an I. Sonbsp;the name would be EKINNOC or EKINIOC; but if M. Maruejolnbsp;should prove to be right in reading A I, the spelling would have to benbsp;regarded rather as EKNIAIOC. From the ingenuity spent on thenbsp;carving of the name, it is highly probable that the inscriptionnbsp;was cut by the bearer of that name with his own hand. Thenbsp;name, if we have it approximately correct, suggests kinship withnbsp;that of the ancient Eceni of East Anglia, called Iceni by Tacitus.nbsp;The inscription will now stand as in the margin on p. 41. Thatnbsp;is 'EiKivvoi Pioviiavios Avhoovvva^o �e�e ^parovbe Kavrev which means�nbsp;� Ecinnos son of Riumanos gave firstfruits to the Andounnas by theirnbsp;decree.� Enough has already been said of the uncertainty of thenbsp;first name, but the patronymic has as its first element riu, standing

perhaps for riuo, to be identified with the name of the god Riuos in the Coligny Calendar; the etymology is obscure, but seenbsp;Celtae, p. 34. Andounndbo is a dative plural feminine (3)nbsp;like Naniausicabo, and is probably the name of a group 3lt; txl/////0nbsp;of Mother-goddesses. Holder, under Riumanos, gives CPIOYnbsp;a reading differing from the others�EkAio� Ptor/[;iaigt;[o]j MANnbsp;Arroei'aKo[s] 6�8[�] ^parovhf /lt;arrer[a]�which, besides IOC ANnbsp;other slips, fails to name anybody as recipient of the AOOYNnbsp;offering; but under Andounnacos, which is more correct N ABO Anbsp;than his later ArrowaKo[s], he explains the former as CACBPATOnbsp;meaning, � aus Andaon, j. Ville-neuve-l�s-Avignon.� YACKANnbsp;This Ville-neuve is the once flourishing town which, T�Nnbsp;from the cliff of Avignon or the Rocher des Doms, as it isnbsp;called, you see over against you on the other side of the Rhone. It isnbsp;hard to avoid the inference that it derived its older name of Andaonnbsp;from the Andounnas, or else that the goddesses derived theirs fromnbsp;Andaon.

xxiv�. Before leaving Nimes I may mention one or two inscriptions which I did not succeed in discovering. One of them is represented innbsp;the Corptis, XII. p. 383, as being � in vinea Guirandi notarii,� and as

if we suppose it to stand for Y one would at first sight perhaps expect OY, making the whole name into KATO-OYAAOC, but verynbsp;possibly we have to pronounce the name as Catudlos, with the accentnbsp;moved on to harmonize with the Gaulish general rule of accenting thenbsp;penultimate. That would explain the shortening of Catu-ualos intonbsp;Cat-uados, as it does in the case of the Catvallauna cited by Holdernbsp;from South Shields, s.v. Catiivellauni. Compare such forms as Atpilosnbsp;from At�pilos, Adhogios from Atebogios, Adgennorix from At�gennos,nbsp;and similar instances which were doubtless comparatively late, asnbsp;otherwise the assimilation of tp, th, tg would have been pushed a stepnbsp;further. Contrast Uciiete (p. 6 above), with its c for etymological dg.nbsp;The whole list of Gaulish names requires to be carefully examinednbsp;from the point of view of their accentuation: an excellent beginningnbsp;was made in 1901 by Meyer-Liibke in the transactions of the Viennanbsp;Academy; see The Enghjn, p. 6. In Celtican the tendency wasnbsp;probably in the contrary direction, resulting in Catualos, and the like.nbsp;The name of which this inscription consists is in Mod. Welsh Cadwal,nbsp;and in Irish Cathal.

xxiv*�. Lastly, the Corpus, XH. p. 383, has a facsimile of a copy published of an inscription in the Histoire de 1'Acad�mie Royale des

PROCEEDINGS OF THE BRITISH ACADEMY

Inscriptions (Paris, 1743), vol. xiv, p. 106, plate I. The inscription was found ad fontem at Nimes, and it is now unknown; but even whennbsp;found it seems to have been broken off at both ends, for the copynbsp;reads as follows:�

Evidently the copyist was not much used to Greek letters, so he dropped off into the Latin equivalents every now and then, as in thenbsp;D and the L, whereas the lambda after D E A E probably stands fornbsp;an A. The first line is rather hopeless; what his inverted C meant isnbsp;hard to say, or his t, unless it was an O with the apex. One wouldnbsp;like to have known whether his OYAB was not followed by a smallnbsp;o ending a dative plural feminine like Andounnabo or Namausicabo.nbsp;The point in the third line cannot have been a part of the originalnbsp;inscription, and the F represented preceding it must have been K withnbsp;the two short arms characteristic of that letter in some of thesenbsp;inscriptions. This was by no means an unnatural error for him tonbsp;make, that is, to suppose it an F, though the original seems to havenbsp;been KANTEN as in No. xxiv. From this we know where we are,nbsp;and how his OYAE has to be corrected into OYAE as the latter partnbsp;of BPATOYAE. If the manuscript facsimile of the inscription stillnbsp;exists, it should be closely scrutinized in case it contains somethingnbsp;which the printed version fails to suggest.

What can be made out here seems to supply us with another variant of the formula Avith the Avords dede, hratude, and cantena; fornbsp;here we appear to have the recipients� name placed just before dede,nbsp;while the nominative to that verb seems immediately to follow givingnbsp;the donor�s name beginning Avith al. We have no means of completingnbsp;the name; Ave are not much better off in the case of the recipients�nbsp;name, but if one may venture to regard their name as ending in thenbsp;dative plural feminine, they may be supposed to have been Mother-goddesses. In any case the length of the description of them Avouldnbsp;offer no serious difiSculty, as they may haA'e had more than one epithetnbsp;applied to them, not to mention the possibility of their being associatednbsp;with a god Avhose name stood as the first Avord of the dedication.

XXV. Gu�ret. Gu�ret is the chief toAvn of the Department of La Creuse, and in the Museum there, or rather outside it in a placenbsp;used by workmen, we found after a long search an inscribed stonenbsp;said to have been discovered in 1864 at Sazeirat, not far from Marsacnbsp;in the same Department: see the Corpus, XHI. 1452; Stokes, No. 19,

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 43

and the Bulletin �pigraphique, 1881, p. 38. The inscription reads as follows;�

The first part is Gaulish while the latter is Latin, which, written in full, makes Votum solvit luberts merito, and suggests that we have herenbsp;to do with a dedication to a god. The Gaulish portion has beennbsp;interpreted in more than one way:�(1) Pictet and Stokes havenbsp;rendered it �Sacer Peroco made (these) porticoes,� which makes annbsp;accusative of DVORICO, but if an accusative plural it should havenbsp;been DVORICA, and if singular it should have been DV OR ICON.nbsp;There is also another kind of objection to this rendering, namely, thatnbsp;it fails to name the divinity to whom the dedicator paid his vow.nbsp;Here Sacer, probably borrowed from Latin, is taken with Peroco asnbsp;describing one and the same man, which implies, and, as I think,nbsp;rightly, that Peroco, of obscure origin, is a nominative of thenbsp;71 declension which would have been Perocon-os in the genitive.nbsp;(2) By taking Peroco, however, to be the dative of a Peroco-s, andnbsp;treating it as the god�s name, one might translate it � Sacer made fornbsp;Perocos this porticum.' Plere there is the same grammatical objectionnbsp;to DVORICO as an accusative, not to mention that the severing ofnbsp;Sacer and Peroco seems somewhat forced. On the other hand it hasnbsp;the merit of not leaving the god without a name. (3) But it occurrednbsp;to the French epigraphist, M. F. Vallentin, that DVORICO is thenbsp;god�s name in the dative case. Assuming that to be right, the sentencenbsp;construes without a hitch, � Sacer Peroco made (this) for Duoricos.�nbsp;It seems impossible, therefore, to accept either of the other translations.nbsp;The inscribed stone was probably inserted in the wall of the buildingnbsp;made for the god, and that building was possibly no other thannbsp;a portico: witness C. I. L., XIII. 2872, where we have �Deo Moritasgonbsp;porticum poni iussit.� Only in the inscription before us there is nonbsp;word for porticum : Dmricos was the name of the god himself.

Duorico comes from the same origin as Breton do7-, Welsh d�r, a feminine meaning a door, that is the means of closing and openingnbsp;a doorway, as the English word itself, and its congeners German thiirnbsp;and tlior, Latin foris, � out of doors,� Greek Gvpa, � a door.� Morenbsp;interesting still as retaining the v are such forms as Old Bulgariannbsp;dvir/, � a door,� and Sanskrit dvara, dvar, dur, of the same signification.nbsp;From dtior we have the Irish dortis, Welsh drws, which mostly meansnbsp;the opening which is shut by means of a door, sometimes the doornbsp;itself. These seem to postulate a dtiorosto-n, not dvorestu, which will

not explain the Welsh form. The latter part of diJiprosto-n would seem to consist of a vocable of the same origin as Latin ostium, and tonbsp;have been accented in early Brythonic duorosto-n or duorosto-n; fornbsp;dii6rosto-n could hardly yield drws in Welsh.

Within an enclosure inside the Museum, and so fenced that it could not be opened, we saw a Latin inscription which seemed to read

enough to be sure whether one should read FILI with a small I in the bosom of the L or FIL only. The letter between E and BROT alsonbsp;eluded my attempts to fix it. I mention this stone as I have notnbsp;stumbled across it in the Corpus. One may add that this littlenbsp;Museum requires to be reorganized: it would be easy to make thenbsp;inscriptions more accessible and far more safe.

xxvi. Vieux Poitiees. Leaving Poitiers by a train going to Tours, we got out at a station called Les Barres, and crossed the Clain. Thennbsp;we walked some two miles, or less, down its right bank until we camenbsp;in sight of Vieux Poitiers, and found the stone we wanted standing innbsp;the middle of a field to our right, and about a quarter of a milenbsp;from the river, which was to our left. A short distance further thenbsp;Clain empties itself, we were told, into the Vienne. The stone isnbsp;Stokes�s No. 14, and in the Corpus, vol. XHI, it is No. 1171 : itnbsp;reads as follows :�

RATN BRIVATIOM

FROITV TARBBISONiS IBVRV

Among the peculiarities of the lettering may be mentioned that the I of R AT IN consists of a prolongation upwards of the first perpendicular ofnbsp;the N, thus N ; the VA of Brivatiom form a ligature; the NT consistnbsp;of N with its second perpendicular provided with the top stroke of T ;nbsp;the E in both instances is peculiar, being B; and the IO consist ofnbsp;a little 0, with a little I standing on the top of it. Let me add thatnbsp;certain of the letters are damaged: thus there is a hollow extendingnbsp;irregularly from the middle of the first T to the N following, and thisnbsp;has been construed into a sort of horizontal I by the readers who havenbsp;missed the real I as part of the ligature for IN. After RAT INnbsp;comes another horizontal hollow, where there may have been a marknbsp;of punctuation, but I could not detect one, and I imagine the hallownbsp;was there before the writing and that it was the excuse for a longernbsp;space than usual between ratin and the next word. The top of thenbsp;second I of Brivatiom is slightly damaged. At the right-hand top ofnbsp;the V of Frontu, there is a hollow which can hardly be regarded as

a mark of punctuation, though Hirschfeld gives us one thereabouts, but he appears to have only studied a cast of the stone, and the resultnbsp;is not satisfactory. Where I thought I found ISON, he prefersnbsp;L with a little S in the bosom of the L and followed by a littlenbsp;O, that is L80. The lettering is damaged here, but I thought the SOnbsp;of the usual size. The previous 1 is unusually close to the precedingnbsp;E. What follows Tarbeiso looked, at first sight, a ligatured A andnbsp;V, but on examining it I thought the connecting groove too low tonbsp;make an A, and it slants in the wrong direction. I came to thenbsp;conclusion that the letter is only a damaged N.

It means, � The rdth for the bridge people Fronto son of Tarbeiso made.� To take the words in their order ratin is the accusative ofnbsp;a word rati-s or rati of the masculine or neuter gender, which is provednbsp;by the adjective Briuatiom, which is in concord with it. The wordnbsp;means a fortifieation of some kind, as in Irish the word raith or rathnbsp;meant mostly a place surrounded by an earthen rampart. The Welshnbsp;is the rhawd in bedd-razod, bedd-rod, � a tomb rath or sepulchre,� andnbsp;gaeaf-rawd, which seems to have meant the place where things werenbsp;stored for use in winter. Briuatiom, has an unexpected final m for thenbsp;usual n: the modification has been explained (see Holder, s. v.) asnbsp;due to the following �, though the inscriber has not been consistent in having carved ratin, and not ratirn, before Briuatiom.nbsp;However, the nasal has not the same sound before b and �: in thenbsp;former it is m with the lips closed, while in the latter it is neithernbsp;m nor n but a dentilabial, formed by bringing the upper teeth innbsp;contact with the lower lip. It is possible, however, that � was pronounced as a bilabial in Gaulish, so that the nasal before it wouldnbsp;also be a bilabial, that is an m as in this case. The adjectivenbsp;Briuat-io-m is derived from Briu-at-es, �people who have to do with^nbsp;a bridge or bridges or live near them, men who have to guard them.� vnbsp;The loeality on the peninsula, between the Vienne and the Clain, isnbsp;suggestive of bridges: we found to our inconvenience that the lacknbsp;of one at a suitable point forced us to go back the way we had come.nbsp;There is nothing to suggest that the fortification was immediatelynbsp;connected with any one bridge: it was rather, I imagine, to be ofnbsp;service to those who lived in a locality which depended much on itsnbsp;bridges, probably several bridges on the two rivers. The longer wordnbsp;is derived from Gaulish briua, of the same origin and meaning as thenbsp;English � bridge �: compare Old Bulgarian br�v�, � a brow, a bridge.�

Frontu is the Latin name Fronto, borrowed : it is not certain either that � was a sound which occurred in native Gaulish words : in thenbsp;few Welsh words of native origin f (now written jf) seems derivednbsp;from sp or perhaps rather sp'h-. compare the Welsh feminine ffer,

� ankle,� Irish seir, acc. dual di pherkl, of the same origin probably as Greek acpvp�v of the same meaning, and see Stokes, Urk. Spr., p. 301.nbsp;Tarbeisonios would seem to be derived from Tarbeiso or Tarbeisonos:nbsp;compare the place-name Tarvisium, Tarvisus, cited by Holder as beingnbsp;now Treviso, in Venetia. In any case the ei of Tarbeisonios wasnbsp;probably pronounced i, the way for the digraph having been preparednbsp;by such spellings as that of eicopov, and Drutei and Druti side bynbsp;side in Roman letters: see Nos. vi and xxxvi. Lastly, the value ofnbsp;the b in Tarbeisonius is not certain. It may have had either thenbsp;ordinary sound of b, in which case one could not compare the namenbsp;with any derivatives of the Gaulish taruos, � a bull �; but B may havenbsp;been introduced here for V, as frequently done in late Latin : comparenbsp;gobedbi, p. 8 above, and Dibona, p. 95 below.

xxvii. Paeis. In 1710 there were found beneath the quire of the Cathedral Church of Notre Dame four altars, which are now in thenbsp;Museum of the Hotel de Cluny: see Stokes, No. 26, and C. /. L.,nbsp;XHI. 3026. Following the order in the Corpus, the inscriptions arenbsp;as follows, beginning with Altar 1 :�

Front.

Back.

Right side of Jove.

TIB-CAESARE�

AVG� lOVIOPTVM// MAKSYMO'S-fnbsp;NAVTAE- PARISIAC///nbsp;//VBLICE� POSIER//nbsp;�N//

EVRISES

SENANI VSEILO///

This may be read with the abbreviations expanded : � Tiberio Caesare Augusto lovi optumo maxsumo summo nautae Parisiaci publicenbsp;posierunt �; that is. When Tiberius Caesar was Augustus thenbsp;mariners of Paris for Jupiter the best, greatest, and highest, set (thisnbsp;altar) up at the public expense. The ends of the lines are imperfect,nbsp;for instance, the 0 ending optumo is gone. The next line seems to endnbsp;with SV, with a little 0 on the second horn of the V. This is alsonbsp;the reading in the Corpus, where, on the other hand, no indication ofnbsp;the S is given: the letters SVO, written as above, stand probably asnbsp;an abbreviation of SVMMO�nobody suggests suo as � their own.� Thenbsp;final I of Parisiaci is gone, and the initial P of publice in the next

line, which ends with a V, of which only the upper ends remain now visible. The N T followed below, but the T is gone, the stone havingnbsp;been broken off close to the N : the latter is preceded by somenbsp;hollows, one of which looks like the punctuation mark r, but inverted.nbsp;The other imperfections of the lettering on this face need not benbsp;dwelt upon. Next as to the opposite face, the tops of all the lettersnbsp;there are gone, but the reading EVRISES is hardly to be doubted.nbsp;The side to the right of Jupiter, that is, the side on the reader�snbsp;left hand, has senani, followed by another vocable which is partlynbsp;illegible, the letters S E and 0 being almost gone, but the lower partnbsp;of the E is still visible: the other two letters would fit perfectly, andnbsp;that such was the reading rests on evidence, which is referred to in thenbsp;Co7'ptis. The reading of the last letter as M requires to establish itnbsp;more expert evidence than we have, and I am disposed to think thatnbsp;it must have been either Nl or intended to be. The whole wouldnbsp;then be Senani Useiloni, whatever that should prove to mean.

The inscription on the front of the altar occupied no fewer than six lines, while the others consisted of one line each, but the spacenbsp;was filled by a number of figures illustrating the legend, as one maynbsp;suppose, in each of the three instances. Dr. Hirschfeld describesnbsp;those under the heading of Eu7'ises in the following terms :�� Hominesnbsp;tres barbati pileati cum peltis et lanceis ; unus (ad dextram) praetereanbsp;manu dextra circulum gerit.� This cmulum is described to me bynbsp;M. S. Reinach as a hoop representing possibly an offering innbsp;process of being presented to the divinity�s temple, and the conjecture is borne out by a photograph with which M. Reinach has kindlynbsp;favoured me. The figures under Senani Useilo7ii Hirschfeld speaks ofnbsp;as � Homines tres mutilati, medius comutus videtur.� And thosenbsp;under the lost heading as � Homines tres imberbes pileati cum scutisnbsp;et lanceis; unus (ad sinistram) paene deletus.� Take the wordnbsp;Eurises first, which seems to imply a nominative E7iris or Eiwisi-s;nbsp;but the medial 5 here may stand for an earlier ss from x = cs, as itnbsp;almost certainly does in Useilo7ii: compare also Alisiia in No. ii,nbsp;and Esanekoti for Exandecotti in No. xxxiv. We should have innbsp;that case to operate with eu7ix, genitive e7irixos, nom. pluralnbsp;eu7-ixes ; compare the plural in -ixes of the name of certain goddessesnbsp;cited by Holder from the neighbourhood of Como, and perhapsnbsp;such forms as Durotix, Calitix, and the like, which he gives under -ix.nbsp;Now eurix, eurixos would correspond exactly to the Welsh e7L7'ych,nbsp;� a worker in gold, a goldsmith, a worker in any metal, a tinker.�nbsp;The word is so come down in respectability that it is oftenest heardnbsp;now in the colloquial saying fel dau eurach, � like two tinkers,� which

is said of two persons quarrelling with great wealth of abusive oratory. For eurych sounds in Welsh like a plural, so a singularnbsp;eurach has been made for it, from which in its turn is formed thenbsp;contemptuous plural eurachod. More to the point here, however, isnbsp;the fact that the first syllable of eurych inevitably recalled the Mediaevalnbsp;Welsh eur, � gold,� borrowed from Latin, as is also Irish 6r, � gold,�nbsp;and that the similarity, though due perhaps to accident, suffices tonbsp;account for the meaning of goldsmith coming sooner or later to benbsp;associated with eurych. Discounting the gold accordingly, we getnbsp;left as the original meaning that of worker in metal. The direction,nbsp;also, in which to look for the etymology of eurych and euris(s)es willnbsp;be found indicated by the verb i-eur-u, n-cop-ov, meaning � �woi'ei,nbsp;fecit, made,� already noticed more than once; see pp. 6, 14, 16 above.nbsp;Here the t-, ei- has long since been marked off by Stokes as anbsp;prefix or preposition. See his Celtic Declension, p. 61, also thenbsp;Comptes rendues de VAcad�mie des Inscriptions, D�cembre, 1880,nbsp;where M. Mowat shrewdly cites the Latin wna, and comparesnbsp;fictilia, � pottery,� from fmgo, � I form, fashion, make.�

If we have sailors and artificers mentioned and figured on this altar, the probability is that the other faces of it also represented somenbsp;leading groups of the citizens of Paris in the time of Tiberius. Butnbsp;what is one to make of Senani Useiloni, supposing that to be the bestnbsp;reading ? Dr. Stokes would connect the second of these words asnbsp;useilom with Gaulish uocellos, � high,� in Welsh uchel, and in Irishnbsp;uasal, � high-born or noble.� There is another possibility, and it isnbsp;that the Parish had borrowed the Latin word vexillum, which undernbsp;the Gaulish accentuation they may have shortened from uexilh-n intonbsp;uxeilo-n, and made to serve as the basis of a derivative, uxeiUno-s,nbsp;plural uxeilamp;n-i, with approximately the same meaning as the Latinnbsp;term vexillarii, and having its ei pronounced i as in Tarheisonios andnbsp;fiiopov, pp. 14, 38, 46. In that case possibly senani, derived fromnbsp;seno-s, � old,� may be treated as meaning veterans, and the whole, innbsp;a quasi military signification, as the veterans who were under thenbsp;vexillum, or flag. In any case one should notice the absence ofnbsp;any trace of horses. If one, however, connect useiloni rather withnbsp;uxellos (better uxelos), �high,� the interpretation would, perhaps, benbsp;� aged men or veterans of high birth.� It is needless to say that, asnbsp;to these Paris groups generally, what has been here suggested isnbsp;mere conjecture.

It seems, at all events, beyond doubt that Euris{s)es is no kind of a verb, and that the three words here in question were not intendednbsp;to form any kind of sentence, though the contrary has sometimes been

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 49

supposed. Taking the three more or less legible faces of the altar into account together, one notices that the first with the longestnbsp;legend is in Latin, while the two others are Gaulish: in that sensenbsp;the altar is bilingual, and the author of the inscriptions must havenbsp;been conscious of the fact.

xxviii. Paris, Hotel de Cluny, Altar 2:�

Left side

There is nothing much to say about the lettering, except that it has been tampered with by some modern idler, who has made lOVIS intonbsp;LOVIS, that is, I suppose, Louis, and also added an oblique linenbsp;to the back of the E of Esus, which makes it look somewhat likenbsp;Vesus: the object is not apparent.

It is not evident whether the author of these headings regarded lOVIS as Latin or Gaulish; for the Gauls may have inferred fromnbsp;the oblique cases (genitive lovis, dative lovi, accusative lovevi)nbsp;a nominative lovis for use in Gaulish. In fact a nominative Jovisnbsp;was not unknown in Latin itself; and if we treat Jovis here as meantnbsp;to be Latin rather than Gaulish, we have a sort of parallelism withnbsp;the previous altar, where the face assigned to Jupiter is inscribednbsp;in Latin and two others in Gaulish.

Taruos was the Gaulish for bull, a noun of the o-declension, and trigaranus appears to be a compound adjective meaning � withnbsp;or having three herons� in reference to the three birds standingnbsp;on the bull. Garanus is to be equated doubtless with the Welshnbsp;garan, � a heron or crane,� and we learn from the ending in us thatnbsp;it was a noun of the �-declension as was also the name of the nativenbsp;god Esus. Add to this that Volcanus had probably been takennbsp;over into Gaulish, and treated simply as another noun of the Gaulishnbsp;�-declension ; had the dative occurred, that form would most likely benbsp;Uolcanoui, with which may be compared Mapeoot^t, dative of Mapeovynbsp;for Latin Marius: see p. above.

This altar, like the previous one, has figures beneath the names, under the first of them a half-nude Jupiter holding a sceptre innbsp;his left hand, while an eagle is to be seen to his right near his feet.nbsp;Under the second, one finds a bull adorned with a � dorsuale � andnbsp;having three cranes standing on him among the leaves of a willownbsp;tree, one on his head, one on his flank, and another near the rootnbsp;of the tail; the first two look forwards and the last one backwards.nbsp;Volcanus is represented standing helmeted, with a hammer in his right

50 PROCEEDINGS OF THE BRITISH ACADEMY hand, and a forceps in his left; and Esus is a woodcutter graspingnbsp;with his right an axe, with which he is hewing away at a tree of thenbsp;same kind as the one extending over the bull and the cranes. In factnbsp;it appears to have been meant as a portion of the same tree, which isnbsp;rendered probable by an altar discovered in 1895 on the left banknbsp;of the Moselle, now in the Museum at Treves. This was firstnbsp;pointed out by M. S. Reinach in a brilliant article on �Tarvosnbsp;Trigaranus � in the Revue Celtique, xviii. 257, plates. There he describesnbsp;one side of the Treves altar as showing a willow among whosenbsp;branches figure a bull�s head and three birds with long beaks, whilenbsp;below appears the woodcutter hewing at the trunk of the tree. Hisnbsp;name is not there given, and it may not have been Esus, butnbsp;M. Reinach suggests that the Parisians of the time of Tiberiusnbsp;identified their Esus with the hero of a lost story once widely knownnbsp;about a cosmic tree whose foliage threatened to deprive the worldnbsp;of the light of the sun. The bull, the cranes, and the willow suggestnbsp;a river divinity; but these also may, without losing their localnbsp;importance on the banks of the Seine, have been fitted into thenbsp;wider story familiar to the Parisians who set up the altar.

In the ensuing Revue Celtique, xix. 245-50, M. d�Arbois de Jubainville advanced some parallels to prove that the story innbsp;question is substantially no other than the Irish epic tale ofnbsp;the Tain B6 Cuailnge, which may now be studied at length innbsp;Prof. Windisch�s elaborate edition.

What remains of the lettering presents nothing which requires explanation, but some of the letters are now illegible, and the topsnbsp;of many of the others are gone. To begin with CERNVNNOS,nbsp;the C is now all gone; and the bottom of the E is all I could tracenbsp;of that letter. The left limb of the V is also hard to trace, andnbsp;there is very little of the final S still visible. The tops of the STnbsp;are imperfect, and where Pollux�s name should come there is nownbsp;nothing legible. The tops of all the letters of the remaining facenbsp;are damaged. The SM are imperfect, and as to VLL I could makenbsp;those letters fit exactly, but I could not say that I saw them, and thenbsp;same may be said of the final S. In spite of the bad conditionnbsp;of these inscriptions, the reading of three of the names rests onnbsp;evidence which to all intents and purposes makes them certain.

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 51

As in the former instance, it is doubtful whether the inscriber was conscious of writing anything but Gaulish, for Castor hadnbsp;probably been borrowed from the Romans, while Cernunnos andnbsp;Smertullos were in any case non-Latin and retained the Celticnbsp;ending in os. It would have been interesting to know what formnbsp;the name of Pollux assumed in Celtic. Underneath CASTOR isnbsp;the figure of a helmeted youth wearing a lorica and carrying a lancenbsp;in his left hand; to his right is his horse, whose bridle he holdsnbsp;with his right. The other figure is just the same, so there can benbsp;no doubt as to its being intended for Castor�s twin brother Pollux.nbsp;The figure underneath CERNVNNOS is bearded, and provided withnbsp;stags� ears and stags� horns, from the latter of which rings arenbsp;suspended. The figure suggests that the m-n of this name is tonbsp;be interpreted by means of the Welsh word corn, �a horn,� andnbsp;the Galatian Kapvov for � trumpet,� literally � a horn.� The relationnbsp;between the vowels in these words is hard to explain: in fact Welshnbsp;has, besides corn, � a horn,� cam, � a hoof,� also a horny substance,nbsp;and cem, � the back part and outline of the cheek,� the front and fleshynbsp;part being called boch from the Latin biicca, � mouth,� On the wholenbsp;Cernunnos is probably to be interpreted as the Horned One: asnbsp;to the horned god of the Celts, see my Celtic Heathendom, p. 78nbsp;and passim, also my Celtic Folklore, pp. 552, 553. Smertullos is hardernbsp;to explain : in point of form it looks like the short and fond formnbsp;of some such a compound as Smerto-rix, Smerto-maros, or Smerto-litanos. Smerto- seems to be derived from smer- of the same originnbsp;as Mod. Irish smior, genitive smeara, defined by Dinneen as � marrow,nbsp;pith; strength, pluck ; the best part of anything �; in Welsh the word,nbsp;having lost the s, is m�r, of much the same meaning as in Irish:nbsp;so smerto may have meant � possessed of marrow, pith, and strength.�nbsp;Thus it would seem that Smertorlx may be interpreted as �strongnbsp;king, or king of the strong,� Smerto-mara as �greatly strong, ornbsp;strong and great,� Smerto-litanos as �strong and broad, or strongnbsp;and exercising power far and wide,� Similarly the name of thenbsp;goddess Ro-smerta, the paredra of a Celtic Mercury, may havenbsp;meant �her of pre-eminent power.�

This was an altar of the same description, having on each of its faces an inscription over the figure of the divinity intended;nbsp;but the traces of the lettering are very precarious. One has beennbsp;read FORT, that is probably Fortuna, standing above two goddesses ;nbsp;so there were possibly two names. The back face has a name endingnbsp;in V S standing over the figures of Mars and a female divinity, so here

52 PROCEEDINGS OF THE BRITISH ACADEMY

also there may have been two names. Judging from the other three altars, some of the names on this also were probably Gaulish.

xxxi. Pauis (5). In an obscure place in the Museum of the Hotel de Cluny is another Celtic inscription found in 1836 near Neris-les-Bains (Allier): see Stokes, No. 23, and C. I. L., XHI. 1388. I examinednbsp;it several times, and the last time with a candle in my hand: mynbsp;reading, which differs from both Dr. Stokes�s and the one in thenbsp;Corpus, is to the following effect:�

BRATRONOS NANTONICNnbsp;EPADATEXTonbsp;RIGI � LEVCVLfonbsp;SVIOREBE � LOCInbsp;TOB-

The NT in line 2 form the usual ligature: the A0 in line 3 are not very clear; the G in line 4 shows a trace of the straight linesnbsp;characteristic of a square C : the perpendiculars of the LL stand onnbsp;one continuous base, and the 0 is a smallish one in the bosom of thenbsp;second L. We now come to the last word, which presents severalnbsp;difficulties: Dr. Stokes, on the basis of M. Mowat�s description, readsnbsp;it LOCITOK ; but, setting out from the same, I was rash enough tonbsp;suggest LOCITOV {Celtae and Galli, p. 38). On seeing the stonenbsp;itself I had no hesitation whatever that the last letter is an E.nbsp;That is, I read E instead of the I of M. Mowat�s reading citednbsp;by Dr. Stokes in the Revue Celtique, v. 119, 120. This E hasnbsp;the abnormal feature that its middle bar is prolonged undulynbsp;and made altogether more conspicuous than the two other bars ofnbsp;that letter. This is illustrated by the reading in the Corpus, which isnbsp;LOG I TO I h, a puzzle not to be wholly disposed of, except on thenbsp;supposition that the last I had been intended in the editor�s notes tonbsp;be deleted. My difficulties are not there but earlier; they begin withnbsp;the GI of RIG I, for those look as if they made either a G without I, ornbsp;a C with a short I; but G1 must, I suppose, be the lettering intended,nbsp;unless the dative rigi was sometimes shortened in pronunciation tonbsp;rig, just as we have Tapavoov for Tapavoovi in No. vii. In LOG I, thenbsp;G has a horizontal tag joining it to the following I, and the readingnbsp;Gl is pretty certain. But as to the next letter, the first of the TOE,nbsp;I have to take it a good deal on trust, for I am not quite convincednbsp;there is more of its top left than would legitimately go to completenbsp;an I. However, LOG I TOE is far more probable than LOGIIOE.

meaning � son of Nantonos or Nanto,� and taking the reading to which the preference has been given, the whole would run thus: � Bratronosnbsp;Nantonicnos Epadatextorigi Leucullo suiorebe logitoe�; that is to say,nbsp;�Bratronos, son of Nantonos, made this grave (or lying-place) fornbsp;Epadatextorix and Leucullos, and for his (or their) two sisters.� A wordnbsp;now as to the individual words: Bratronos seems derived from brater,nbsp;which must have been the Celtic word for � brother.� Nantonicnosnbsp;comes in the last resort from the personal name Nantos, which occursnbsp;as Nantus in C. I. L., XIII. 805; we have also Nantiorix, 5485,nbsp;and other related forms. The long name seems to analyse itself intonbsp;Epad- for Epasso-, meaning, possibly, a horseman, from epo-s, �anbsp;horse,� and Atechto-rix, which seems to mean a king of Atechti. Thenbsp;latter in the singular, Atechtos, is probably to be analysed intonbsp;ad-techto-s, like the Con-texto-s of the Autun inscription in No. v;nbsp;both names probably mean � protector.� In that case our Epadatextorix would mean �him who is captain of protecting horsemen.�nbsp;M. Jullian suggests interpreting it as �a knight of the Ala Atechto-rigiana,' as to which see Holder, s. v. Atectorix. Leucullo is thenbsp;dative of Leuculh-s, either cognate with or borrowed from thenbsp;Latin Lvcullus. In the passage to which I have already referred,nbsp;I ventured to interpret suiorebe as a dual standing for an earliernbsp;suihorebe = suisorebe, meaning � to or for two sisters.� Whose sistersnbsp;they were, the inscription does not make clear: they may have beennbsp;the sisters of the two men with their names in the dative, or ofnbsp;Leucullos alone, or else of Bratronos, in which case the two men maynbsp;have been their husbands, and brothers-in-law to Bratronos. Logitoenbsp;I should take to be a variant of the logitu suggested by the analogynbsp;of ieuru and carnitu, with oe representing -atiit or -oiiit as in Latinnbsp;amavit, as does also probably the u of ieuru and carnitu, as suggestednbsp;in the notes on Nos. ii and xxxiv. The syllable it in carnitu has beennbsp;touched upon at p. 31, and accordingly the form logitoe analyses itselfnbsp;into log-it-oe. Further, one may say that, just as carnitu � cam-it-unbsp;derives from a nominal base cama, � a heap,� so logitoe comes from loga,nbsp;�a grave, a burial or lying-place,� the accusative of which, loga-n,nbsp;occurs in No. xxxvi, which see. It is a peculiarity of this nominalnbsp;verb that it governs the dative case, wherefore I have ventured thenbsp;translation, �made a grave for.� The element it in these verbs hasnbsp;already been compared with the same in Latin verbs like habito, vocito,nbsp;and its force in Gaulish may not have been frequentative so much asnbsp;durative or progressive, referring to an activity which occupied somenbsp;time. The question, however, suggests itself, What was there in thesenbsp;instances to call for verbs with that connotation.? and it may be

answered in part by asking another: Why do Greek inscriptions not infrequently have, not inoCrja-fv, but the imperfect �woiei correspondingnbsp;to the fecit of Roman ones ? I am assured it is the case butnbsp;there may be a closer connexion between the Gaulish use of verbsnbsp;with it and the Greek use of the imperfect: the latter may havenbsp;suggested the former, that is, the Gaulish -it- may have been intendednbsp;as a sort of equivalent for the Greek imperfect. At all events, thisnbsp;will serve as an excuse for my giving here a Bourges inscription, whichnbsp;I have not yet seen, but the Corpus, XIH. 1326, gives the readingnbsp;on the stele in three parts, as below: see also Rev. Celtique, xv. 237.

in Latin and in Greek letters, that is, Oxtos son of Virilos: the firstnbsp;part ofthe first name is gone. Thennbsp;come two lines in Greek, whichnbsp;mean that Anevnos made it. Thennbsp;lower down comes a continuationnbsp;of four lines, in Gaulish thisnbsp;time, and in Latin letters, conveying the following sense��Eluontiunbsp;made this for Aneunos son of Oclosnbsp;and for Luguris son of Aneunos.�

This trilingual inscription shows that the son of Virilos had a stone set up to his memory by Aneunos, or very possibly Aneunosnbsp;procured the plot of burial-ground for the son of Virilos, and had hisnbsp;name placed on the stone to show his right to it. At a later date,nbsp;however, a friend or relative of Aneunos, named Elvontiu or Eluontio,nbsp;made the stone commemorative also of Aneunos himself, and of a sonnbsp;of Aneunos, named Luguris. Most likely all the men named belongednbsp;to one and the same family, and represented two or three generations,nbsp;perhaps four. The names are all obscure, but the patronymic yiriliosnbsp;or OvipiWio�why not OvipiWios ?�may point to the father�s namenbsp;not as Uirilos or OvipiWos, but as the Latin cognomen Virillio borrowed.nbsp;Among the points to be noticed in this remarkable inscription is thenbsp;fact that the name of the man to whom the stone was originally putnbsp;up is given twice, in Roman letters and in Greek, which is singularnbsp;among our Celtic inscriptions. Another thing also to notice is that

* I have before me some statistics kindly given me by Mr. Tod of Oriel College, Oxford, together with a reference to an interesting passage in point innbsp;M. S. Beinach�s Epigraphie Grecque, p. 436,

this name is put in the nominative case standing in no immediate syntactical relation with what was added to it and cut at the samenbsp;time with it. This is by no means peculiar, and we have importantnbsp;parallels in Nos. xxxiv and xxxv, both found in North Italy. Lastly,nbsp;the second piece of the inscription ends with the Greek imperfectnbsp;CTTO�l, for the more usual spelling kitoUi, �was making.� It is notnbsp;certain whether any special significance attaches to the fact that thenbsp;later portion of the inscription is entirely in Gaulish.

xxxii. Chateau de Saint-Germain (1). A vase of Gallo-Roman ware, found at Serancourt near Bourges in 1849, is now in the Nationalnbsp;Museum at the Chateau, and reads as follows round the neck of thenbsp;vessel: see Stokes, No. 25:�

That is to say, � Buscilla placed this in Alisia for Magalos.� The verb legasit is an aorist from the root leg of the same origin asnbsp;English lie and lay; here it has the causative meaning of � laid � ornbsp;�caused to lie.� It has the vowel e in the stem, whereas lagan,nbsp;� a lying-place or grave,� in the Gaulish inscription No. xxxvi, has thenbsp;vowel o like the kindred Greek word Xo'xoy, �a bed,� as contrastednbsp;with Ae'xerat (explained as Koi/iarai). With regard to the place-name,nbsp;the X stands either for the sound of cs or ss, and as to Magalu, thatnbsp;was probably a word of only two syllables, Maglu, the dative ofnbsp;Magios, which in Welsh became Mod, and in Irish (as a commonnbsp;noun) mal, � a prince.� The word sosio would seem to be the neuternbsp;demonstrative for �this,� and as it occurs also in the Celticannbsp;inscriptions of Rom, it suggests, as pointed out in Celtae and Galli,nbsp;p. 48, that the language here is not Gaulish but Celtican. Thenbsp;Gaulish demonstrative which we have had was sosin, which was alsonbsp;neuter: it is possible that Gaulish had likewise a neuter sosio, but itnbsp;is more probable that it was the Celtican form alone, and that thisnbsp;inscription is in Celtican. Such a view is corroborated by the factnbsp;that we have had the same place-name in the ablative case in No. iinbsp;(pp. 4-7), and it was ALISIIa, that is Alistia, whereas it is herenbsp;Alixie with a different termination, as to which compare Celtae, p. 48.nbsp;There is also the difference between the prepositions, the reading innbsp;the former inscription being probably � indu Alisiia,� not � in Alisiia.�nbsp;The accentuation would also be different, the stress being placednbsp;probably on an earlier syllable in Celtican, that is, either Alixie ornbsp;else Alixie. If as seems probable it was intended to be in metre,nbsp;it was scanned somewhat as follows :�

xxxiii. Chateau de Saint-Geemain (2). Some years ago a stone statue of Mercury was discovered at Lezoux in the Puy-de-Dome, andnbsp;was acquired in 1901 for the National Museum; it now stands innbsp;the court at the Chateau. It seems to be No. 1514 in the thirteenthnbsp;volume of the Corpus, where it is wrongly described as statua ahenea,nbsp;and the only legend there given is the one on thenbsp;M E R C V RI o god�s chest as in the margin. These letters are closelynbsp;ETAVGVSTO packed together within a moulding with queuesnbsp;S A C R V M d'aronde. So the two final 0�s are small, and thenbsp;first one has the I standing on it and not by its sidenbsp;as represented in the Corpus. This is, however, not all, for the godnbsp;has on his back and shonlders a Gaulish inscription which the discoverer, M. Plicque, attempted to read, but without great success: seenbsp;Dechelette�s � Vases ceramiques ornes de la Gaule romaine,� pp. 144-6,nbsp;where M. Plicque�s reading is given�I have not seen his own work onnbsp;the god Eug. Having examined the shoulders with a candle verynbsp;carefully, I am able to improve a little on his reading, as I findnbsp;beyond doubt that the second word is teuru-, but my reading alsonbsp;is incomplete: the whole should be scrutinized again, and an excellentnbsp;cast which M. Reinach has had prepared will prove of great help.nbsp;The following is what I made of it:�

The S at the end of the first line is very faint and so, even more so, is the I at the end of the second line. That letter was probablynbsp;followed by another letter which I have failed to trace at all, thoughnbsp;one naturally thinks of sosin or some demonstrative approximatingnbsp;that form. The next line ends seemingly with V (possibly with O), butnbsp;I thought I detected traces of an N or M after the V. Lastly, thenbsp;statue seems to have been standing erect when this inscription was cut,nbsp;for the first line slopes downwards as if the workman had a difficultynbsp;in reaching the last letters of the name APRONIOS, With greatnbsp;diffidence I guess the original to have read Apronios ieuru sosin Esun%nbsp;that is, � Apronios made this Esus.' In that case this monumentnbsp;identifies Esus, not with Mars, bnt with Mercury, of which evidencenbsp;is also supplied by one of the ancient comments on Lucan, i, 444-6,nbsp;cited by Holder (s.v. Esus) to the following effect: � Hesum Mer-curium credunt, siquidem a mercatoribus colitur,� amp;c. There is nonbsp;reason to suppose Apronius a name of Celtic origin, but it is here sonbsp;far naturalized in the Gaulish language that it assumes the form

Apronios at a time when it would probably be more commonly Apronius in Latin itself. The evidence that the god was meant tonbsp;be a Mercury does not wholly depend on the Latin inscription: hisnbsp;whole get-up favours this view, though most of his accessories have beennbsp;damaged. The god holds in his right the inevitable money-bag, butnbsp;the other hand is broken off, now partially restored. Lastly M. Plicquenbsp;detected on one of the folds of the god�s dress traces of a third inscription, which he read APRO TASGI. ... My attention was directednbsp;to this too late to do it justice.

Altogether this is one of the most grotesque and clumsy gods I have ever seen; and a workman who observed me looking at the statue asnbsp;restored volunteered the remark, � They are all like that �: of coursenbsp;he meant the Auvergnats.

xxxiii�. For the sake of comparison it is necessary to mention certain inscriptions which I have not been able to see. One of thesenbsp;is the Nevers inscription which cannot be found. The readingnbsp;according to Stokes, No. 20, and the Corpus, XIII, 2821, was asnbsp;follows in the margin:�

ANDE

CAMV

LOSTOVTI

SSICNOS

lEVRV

That is to say Andecamulos Toutissknos ieuru, which means � Andecamulos son of Toutissos made (it).�

in the Corpus, XIIL 10010. 2097, the second line is read CIUCOS, which, however, must have meant Cl NCOS. A more serious questionnbsp;here arises as to the division of the words: Dr. Stokes treats thenbsp;whole as Uritu or Vritu Escingos = � Excingos made (this),� Hirsch-feld on the other hand compares potters� names Vrittius and Vritves,nbsp;which he regards as suggesting a compound. So on the wholenbsp;a Gaulish preterite uritu cannot be regarded as established and readynbsp;to be placed by the side of ieuru, camitu, and logitoe. As to thenbsp;whereabouts of this vessel the editor of the Corpus states that it wasnbsp;at Bavai, and adds : � Ibi fuit apud de Fourmestraulx, iam in castronbsp;Gussignies apud de Moras �; but I have not succeeded in elicitingnbsp;from the Chateau any reply to my letters of inquiry.

xxxiii�. Dr. Stokes�s No. 27 is a gold ring, outwardly octagonal, said to have been found in the neighbourhood of Thiaucourt (Meurthe-et-Moselle), and to be �in the collection of the Acad�mie des Inscriptions

et Belles-Lettres.� But I have not been able to trace it, and it appears from the Comptes Rendus of the Acad�mie that it was presented to that body on behalf of the late M. L. Maxe-Werly;nbsp;A DIA but this, as has been pointed out to me by more than onenbsp;NTVN member of the Acad�mie, only meant presented in the sensenbsp;N E NI probably of being submitted to that body, which was donenbsp;E X V E by M. P. Charles Robert, who began his account of thenbsp;RTIN ring with the words �Je pr�sente a 1�Acad�mie, de la partnbsp;IN AP de M. L. Maxe-Werly, une bague d�or,� amp;c.: see the Comptesnbsp;PI SET Rendus for 1885, p, 33. At all events it never belonged to

V the collection of the Acad�mie: in fact I am assured that the Acad�mie has never possessed a collection. It has beennbsp;suggested to me that at M. Maxe-Werly�s death his collection wentnbsp;to the Museum at Bar-le-Duc, that the ring is probably included, andnbsp;that it will be found when the things come some day to be unpacked.

The inscription reads continuously and Stokes translates it: � Nappi-setu (gave this) to Adiantunnena (daughter) of Exvertinios.� Thus he treats Adiantunneni as the dative of a woman�s name Adiantunnena.nbsp;There are two ways of explaining the patronymic: it may simply benbsp;the genitive of the father�s name, after the analogy of Doiros Segomarinbsp;in iii; his name in that case would be Exuertinos or Exuertinios withnbsp;Exuertini standing as the genitive of either form of the name. Ornbsp;else one may treat it as an adjective in concord with Adiantunneninbsp;and standing for a dative feminine Exuertin{^i, nominative Exuer-tinia: this would imply that the father�s name was Exuertinos;nbsp;compare Tarbeison-ios, xxvi, and ^iril-ios, xxxi�. Related to thenbsp;woman�s name may be mentioned as cited by Holder, Adianto, dativenbsp;Adiantoni, from Bale in Switzerland, and also a number of namesnbsp;without the first nasal, especially Adiatunnus, which is given bynbsp;Caesar, iii. 22, as the name of a chief of the Sontiates or Sotiates,nbsp;a people of Aquitania. The Adiantunn- of the present name seemsnbsp;to equate with the Adianton- of the name from Bale and to derivenbsp;from a stem ad-ianto-, which is represented in Welsh by ad-iant,nbsp;�a longing, a wish or desire,� just as the ad-iat- of Adiatunnus, isnbsp;probably represented in Welsh by ad-iad, of much the same meaningnbsp;as adiant; in fact adiant and adiad ^ are probably derived from the

* Why these words have not become eidiant and eidiad, I do not quite see, especially as we have Welsh aid, � zeal, fervency, enthusiasm,� whence eidig,nbsp;jealous, a jealous person,� from ad-ies-, involving the same root ies as the Greeknbsp;ft'o), � I boil,�nbsp;nbsp;nbsp;nbsp;� ardour,� feoror, � boiled, sodden,� Eng. yeast, and Welsh

ids (fem. = ifsta), � a thrill, whether hot or cold �; but the association with boiling is not forgotten in the language. Witness such words as rhoi ias o ferw

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 59 strong and weak cases of one and the same stem represented bynbsp;the Gaulish ad-iant- and ad-iat-. Without the prefix we should havenbsp;iant- and iat-, and the former occurs in the Gaulish lantu-mdros, withnbsp;which Dr. Stokes has equated Irish �tmar, � zelotypus �: see his Urk.nbsp;Spr., p. 222, where he refers the Celtic forms to the same originnbsp;as Sanskrit yatna, � effort,� and Greek C^re'co, � I seek.� The derivativenbsp;syllable m, in Adiantunn-en-a, better perhaps Adiantunn-eni-a,nbsp;reminds one very much of Goidelic proper names like Adamnan�snbsp;jEmm-e and the Ogmic genitive Doman-en-i. Compare also Dairene,nbsp;the name which �the Four Masters� (a.d. 619) give to Dair�s offspring:nbsp;see p. 10 above. The in of Exuertmi is common in proper names, bothnbsp;Welsh and Irish, and the first part of Exuert-in-i seems to warrantnbsp;its being referred to the same origin as Welsh eh-orth, �orth,nbsp;�assiduous, strenuous, energetic.� It is remarkable that both Adian-tunnen-i and Exuertin-i are not of the mere compound kind whichnbsp;Gaulish inscriptions usually illustrate. The next name, Nappisetu,nbsp;baffles me, but it is perhaps a nominative of the n-declension, derivednbsp;in part from what appears as SETV on a silver coin of the Volcaenbsp;Tectosagi; compare Setonim and Setu-bogios, also the place-namenbsp;Setunia, Setuna now Stonne in the Department of Ardennes. Thenbsp;other element occurs in the woman�s name Nape, cited also by Holder,nbsp;from an inscription found at Tiermes in Soria, Spain; C. I, L., II.nbsp;5795. The symmetry of the legend on the ring would seem to suggestnbsp;that Nappisetu has a p too many. Lastly, Nappisetu may benbsp;a neuter of the w-declension, and not a proper name at all, but a wordnbsp;of some such meaning as that of a gift or present. In that case onenbsp;would have to construe thus: � The gift of Exuertinos to Adian-tunnenia.� In any case it looks more Celtican than Gaulish.

xxxiv. Novara, North Italy. In a cloistered court of the Cathedral of Novara is a Celtic inscription found in the neighbourhood; itnbsp;is placed in the wall, and surmounted by a label inscribed � Brionaenbsp;in territorio vici S. Bernardini dum vetusta ibi silva excidebaturnbsp;a. 1859.� It is Stokes�s No. 2, and No. 10 in the Diet, arch�o-logique de la Gaule, where a photograph of it is given : see also C. I. L.,nbsp;V. p. 719, and Pauli�s Inschriften iwrdetrushischen Alphabets, pp. 12,78.nbsp;The letters are Etruscan, with K, T, P for both those letters andnbsp;for G, D, B; so the transliterator has to distinguish between them,nbsp;also to insert nasal consonants when coming before those other

i�r llefrith, � to give the sweet milk a thrill of boiling,� that is to say, to bring it just to the boiling-point and then stop ; the derivative adjective iesin, fromnbsp;meaning � productive of thrills � of delight, has been weakened into � delightful,nbsp;beautiful, fair, nice.�

He interprets it as follows: Kvi{n)tes asoioi ken Dannotaliknoi, Kvi{n)tos Legatos, Andoko{m)hogios, Setubogios, Esandekotti, Andare-visseos, Dannotalos kamitus. Tekos toutki. He has added the followingnbsp;translation:�� (This sepulchre) the grandsons (?) of Quinta, to wit thenbsp;Sons of Dannotalos, (namely) Quintos the legate, Andocombogios,nbsp;Setubogios, (and the sons) of Exandecottios, (namely) Andarevisseos,nbsp;Dannotalos, heaped together. Tecos the magistrate (lies here).�nbsp;I need not mention that I have nothing to say by way of criticizingnbsp;the Celtic forms of the names suggested by Dr. Stokes: they seemnbsp;to be well established. But the reading especially of the firstnbsp;horizontal line and that of the cross line to the left offer difficulties,nbsp;which inevitably make the interpretation a matter of considerablenbsp;uncertainty, A more correct notion of the whole monument maynbsp;be got by representing it as standing with the cross line as thenbsp;head line, and the other ten lines as reading downwards in thenbsp;direction of the length of the stone: that must have been its originalnbsp;position and not lying down as in the wall at Novara.

The letters do not exhaust the points of this monument, for in front of the ten vertical lines, and between them and the top line, therenbsp;is a row of four closely packed circles with eight radii or spokes to each.nbsp;Are we to regard them as representing chariot wheels or even entirenbsp;chariots ? I cannot answer, but they remind me of the earlier stagenbsp;when the owners of war chariots were interred in them as in thenbsp;well-known instances found in the neighbourhood of Market Weighton,nbsp;Driffield, and other places in Yorkshire. If that is so, it is butnbsp;natural to regard the four circles or wheels as representing twonbsp;chariots, and two warriors as the number of men buried. In thatnbsp;case the wheels might be regarded as an instance of early heraldry.

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 61

Pauli has tried to begin earlier and has fared worse. Dr. Stokes gives K(vi)tesasoioiken, with vi in brackets as supplied by him; but asnbsp;a matter of fact the V is still just traceable, and the same wouldnbsp;doubtless have been the case with the I but for a slight breakagenbsp;which has made that letter impossible to trace. I agree furthernbsp;with him in reading Kvitesasoioik, for though the last k is damagednbsp;it is not doubtful. Dr. Stokes has read the next letter as E, butnbsp;it may be A, though I am on the whole inclined to E. We arenbsp;agreed also as to the next letter which is N ; but it is, I think,nbsp;followed by an I ending the line. However this is not yet the wholenbsp;of the line, for there are traces of writing before Kvites, Thenbsp;K of this name stands opposite the first A ofTANOTALIKNOI, andnbsp;before it I seemed to trace the equivalents of IN A, but I wouldnbsp;not be sure of them. What is certain, I think, is the presencenbsp;there of traces of writing. The whole line is near the right edgenbsp;and the lettering gets worse towards the top, that is treating the stonenbsp;as standing upright.

Without going into the question of the origin and descent of the Etruscan alphabet, it will suffice so far as regards this inscription tonbsp;treat the letters as if they were merely clumsy forms of the Latinnbsp;ones, with the exception of two or three, such as X which stands fornbsp;T (with the top as it were fallen half-way down), as D which standsnbsp;for R (derived probably from P with the Greek value), and as M whichnbsp;seems to be a sort of double S, and to represent probably the sharpnbsp;sound of ss; otherwise there is here no doubling of consonants. Thenbsp;other S in the inscription varies considerably: it is like Latin Snbsp;in lines 4, 5, and 7. In most of the other instances it is more open,nbsp;except that at the beginning of 6 it is a sort of a wriggle resemblingnbsp;a corkscrew, and that at the end of 3 it is reversed. It is reversednbsp;also at the end of 10, where it is rather imperfect and faint.

The difficulties offered by the top line are greater as both of the corners are gone, and especially that opposite the reader�s left hand.nbsp;The reading given by Dr. Stokes is TEKOS TOVTIV. But the firstnbsp;letter X, that is T, stands close to the broken edge, so that one cannotnbsp;say whether it was not preceded by one or more characters. The nextnbsp;difficulty is the identity of the next letter : it looks like our F upsidenbsp;down or our E without the top line, but that line was never there, asnbsp;no damage can be traced there, and the top end of the perpendicularnbsp;is complete and of the proper depth. On the whole I am morenbsp;inclined to treat it as an Etruscan A upside down. The latter I oughtnbsp;to have said looks like an F with its arms drooping a little: it fulfilsnbsp;the conditions better than any other character I can think of.

According to this guess the first word reads XFKOS, which seems to be followed by a shallow stop . , after which come the letters XOVXI.nbsp;The next letter seems to me to be an 0 rather than a V : then comenbsp;some cracks covering probably another stop, and rendering the nextnbsp;letter partly imperfect, as to which, however, there is no doubt thatnbsp;it is S. It is followed by a V, after which one detects the two leftnbsp;ends of X. To me the rest of the line is illegible, though Mommsennbsp;has suggested several letters more, and a lacuna which he could notnbsp;fill; for in his transcript he represents this part of the line as � ositnbsp;. . where the letters, except perhaps the t, are at variance with thenbsp;drawing accompanying them. So he fails to help us, except in testifyingnbsp;to the presence of more writing than is suggested by Stokes or Pauli.nbsp;My own guesses would stand thus: XFKOS � XOVXIO- SVX . . . .,nbsp;which may be transliterated as follows:�JA^OS � JOVJIO � SVJ ....nbsp;In selecting the alternative equivalents one cannot be wrong in treating the middle word as a nominative Toutio, corresponding to whichnbsp;the Gaulish genitive would be Toutionos, which, as already mentionednbsp;(p. 15 above), Mommsen restored as (Latin) Toutionn, equated sincenbsp;by Stokes with the Gothic thiudans, � king.� The first word is morenbsp;difficult to fix: at any rate two possible treatments of it are possible,nbsp;(a) Either take it to be tagos, which recalls the latter part of thenbsp;name of the Ecenian king Prasu-tagus, alongside of which Holdernbsp;places a genitive Ito-tagi (C. I. L., IV. 2451), which may, however, benbsp;jfor Ito-tagi-i: compare Tagim, fern. Tagia, assumed by Holder onnbsp;the strength of C. I. L., XIH. 3456.nbsp;nbsp;nbsp;nbsp;(^�) Another valuation of the

Etruscan characters is quite possible, yielding Dagos, as in Gaulish names like Dago-rix, � good or brave king,� Dago-uassus, � good ornbsp;brave youth,� and others, including Bitu-daga, cited by Holder. Innbsp;Welsh the word has been reduced to da, �good,� Med. Irish dag-, asnbsp;in dag-dtime, �bonus homo,� dag-per, �bonus vir�=Welsh dewr,nbsp;� brave man,� and adjectivally � brave.� If I am right in supposing thenbsp;last word to begin with SVX, it would probably mean some name^nbsp;Su-t... or Su-d . . . , beginning with the prefix su, with which wenbsp;have already had to do ; see p. 25 above.

I do not believe a photograph would be of much use, and I have not yet succeeded in getting a squeeze of this remarkable monument;nbsp;but Signor Tarelo, the most learned archaeologist connected with thenbsp;museums of Novara, has kindly promised to do his best, in addition,nbsp;that is to say, to the valuable help which he most readily gave me,nbsp;both when I was there and before I arrived. The foregoing guessesnbsp;of mine will be found put together in the arrangement on the nextnbsp;page.

The heading or the important portion of the monument seems to have been the separate line over the four wheels. The names ofnbsp;which it consisted are in the nominative, as was the case withnbsp;No. xxxi% where they stand conspicuous in no syntactical relationnbsp;to the rest of the inscription, I should take the words to mean:�

latter was perhaps somebody of lesser importance, but seemingly the personages were two and no more, which agrees with the conjecturenbsp;that the wheels represent the two war-chariots of the deceased.

�n

o

In the first of the vertical lines K VIX ^ S seems to be the Gaulish genitive of the borrowed name KVIXF, that is Quinta, just asnbsp;K VIX 0 S stands for Quintas in the third line. Dr. Stokes has treatednbsp;asoioi as the noun on which the genitive depends, and suggests as thenbsp;translation �grandsons of Quinta.� The singular should be asoios,nbsp;which is probably to be analysed into a-so-w-s, where the root wouldnbsp;be so or s�, which has already been noticed in connexion withnbsp;atehotisse, with hot for sot of the same origin as Irish suth, �birth,nbsp;offspring,� Celtae, p. 43 : the English word son and its congeners arenbsp;of the same origin, and the soios portion of our word recalls abovenbsp;all the Greek vlcf?, � son.� The prefix in asoios may have been ad, ornbsp;else the a which we have in such Welsh words as a-dysg, � instruction,�nbsp;and a-def, �to confess�: so the whole word may well have beennbsp;expressive of relationship, and may have specially meant a grandson;nbsp;if not that, at any rate a descendant. Dr. Stokes was inclined tonbsp;treat asoios as standing for an earlier asovios, which, should it benbsp;found phonologically preferable, would fit this interpretation just asnbsp;well or better. The termination oi of the plural is matched by that

of XFNOXFLIKNOI, that is Dannotalicnoi, a patronymic meaning �the sons of Dannotalos' The intervening word I am inclinednbsp;to read KENI: it admits of being transliterated either ceni ov gem,nbsp;and it has been taken to mean �to wit, namely, even.� Thenbsp;fourth line gives us another borrowed Latin word to place by the sidenbsp;of Quintos and Quinta, as it is legatm borrowed: it is not certainnbsp;whether it is used in its Latin sense or simply employed as a personalnbsp;name. Apparently the descendants of Quinta are here divided intonbsp;two groups, the sons of Dannotalos and the sons of Exandecottios.nbsp;Dannotalicnoi as a patronymic meaning the sons of Dannotalos offersnbsp;no difficulty, but it seems somewhat harsh to suppose that alongsidenbsp;of it we have in ^SFN^KOXI, that is, Exandecotti, simply thenbsp;genitive of Exandecottios (if not that of the simpler form Exande-cottos, like the Cottos from which it is derived) without any noun onnbsp;which that genitive might be said to depend: I should prefer to treatnbsp;it like Dannotalicnoi, as standing for a nominative plural Exandecottii,nbsp;resembling such patronymics in -ips as Tiavpavios and Tarbeisonios.nbsp;The objection to this has doubtless been that the other two pluralsnbsp;end in -oi not in -i; but leaving that for the present, let us proceednbsp;to the verb which is KFDNIXVS, that is, carnitus, the plural of thenbsp;form KAPNITOY of the Saignon inscription, as to which see p. 31nbsp;above and No. xxxvi below, also Celtae, p. 47. If the final u ofnbsp;ieuru, �fecit, knotei,'' represents what was in Latin -auit in forms likenbsp;amavit, then the -us of carnitus should correspond to the -auis-underlying -au�re, -avxrunt in the forms amavere, amaverunt of thenbsp;plural in the same verb and the like.' The plural nominative to thenbsp;verb carnitus consists of the nouns in the vertical lines. It is morenbsp;difficult to find an accusative to represent the object of the verb;nbsp;on the whole I am inclined to think that there is no accusativenbsp;expressed in the sentenee. At any rate the doubtful beginning ofnbsp;the first line, where I have guessed INF, is more likely to be an adverbnbsp;than the object of the verb, that is a word meaning � here, below,nbsp;hard by,� or the like. At the end the uncertain element is ceni or geni,nbsp;which, whatever it exactly meant, is not very much like an accusativenbsp;of any kind. Accordingly my attempt to translate the whole willnbsp;stand thus, and with it should be compared No. xiv, p. 30 above, andnbsp;No. xxxvi below :�

� It is right to say that Brugmann expresses himself as not quite certain as to -erunt = isont{i) : see his Grundriss, II. �� 841, 1023,1079.

Quintus the Legate, Andocombogios, Setubogios, (and) the Sons of Exandecottos, (namely) Andarevisseos, Dannotalos, piled up a cairnnbsp;for them.�

Lastly, a word as to Exandecotti as a plural: there is a prima facie objection to this, arising out of the fact that we have by its sidenbsp;two plurals in -oi of the same declension asoioi and Dannotalicnoi.nbsp;Of the two endings oi and � of the nominative plural, the latter is thenbsp;one that won the day in Latin and Celtic, while in Greek oi held itsnbsp;ground as in aheKcpoC, xp�voL; and just as in Latin one finds citednbsp;only pilumnoe poploe (for the usual populi), so in Celtic these twonbsp;instances asoioi and Dannotalicnoi seem to stand alone; no othernbsp;certain example appears to be on record. There must, however, havenbsp;been a period of transition when both -oi and -i were in use side bynbsp;side, and to that period the Briona inscription would seem to belong.nbsp;I cannot help adding that this pair of instances of the plural in -oinbsp;marks this inscription as an early one: it is possibly the earliestnbsp;Celtic on record,

XXXV. Beescia. There is here a bilingual stone of possible interest to Celtists: it was found built into the wall of the belfry of one ofnbsp;the small churches in the neighbourhood of Limone near Lake Garda,nbsp;and is now inserted into an inside wall of the Brescia Civic Museumnbsp;of objects of the Roman period. It is conveniently placed fornbsp;inspection, but the letters have been painted dark red, and here andnbsp;there mispainted as usually happens in such cases. This forms a greatnbsp;difficulty when one wishes to make use of photographs. The inscriptionnbsp;is Dr. Stokes�s No. 3: see also C. I. L., V. No. 41883, and Pauli,nbsp;loc. cit., p. 15. The reading is as follows :�

Dr. Stokes has rendered it continuously as one sentence ; � Tetumus (filius) Sexti, Curator Sassarensis, me addixit Obuldino Tino.�nbsp;Besides other differences between his interpretation and mine, I treatnbsp;the first four lines as Latin in spite of the character for ss, and asnbsp;giving the names of the owners of the ground or the tomb. Thosenbsp;names are put in the nominative case as in the Briona inscription, andnbsp;we may treat them as probably those of husband and wife, Tetumusnbsp;son of Sextus, and Dugiava daughter of Sassadis. Of these names

Dugiava is undoubtedly Celtic, and on looking up the word in Holder�s Altcelt. Sprachschatz, it will be found that most of the instancesnbsp;and kindred names come from the same district and from Piedmont.nbsp;Tetumus will come under notice later, and Sassadis has a number ofnbsp;seemingly related forms cited by Holder, such as Sassus, Sassa, andnbsp;Sassius, Sassia ^; but Holder�s own account of the name is that innbsp;SAM AD IS the D has to be treated as Etruscan R so as to readnbsp;Sassaris, but in spite of the occurrence of Saserus and Sasirus, thisnbsp;does not appear obligatory or very easy to accept, seeing that thenbsp;previous line has in Dugiava a D which has practically to be givennbsp;its ordinary value in the Latin alphabet. At the same time therenbsp;is no denying that the inscriber was very mixed in his alphabets,nbsp;perhaps even more so than appears at first sight; but more of thisnbsp;presently.

I come now to what I regard as possibly Celtic, in which some of the symbols require special notice: among other things the f whichnbsp;should have been X is indicated by five points. In the first instancenbsp;it serves to mark where the non-Latin portion begins, and in thenbsp;second one it occurs at the division between two words; but, in thenbsp;latter position, it is right to say that the five points are placed nearernbsp;to the preceding letter than to the one next following, so that evennbsp;there we are not obliged to treat them as a stop. The spacing helpsnbsp;to mark off the letters ina as making perhaps a separate word, withnbsp;which the first of the vertical lines on the Briona stone, p. 64, seemsnbsp;to begin. It is to be noticed that if the five points formed a merenbsp;punctuation mark (as in C. I. L., XH. 1416), they should have beennbsp;placed immediately after Sassadis and not at the beginning of thenbsp;next line. Next must be mentioned a sort of an arrow-head whichnbsp;appears in both lines, for it is the same symbol in both, though at firstnbsp;sight there seems to be considerable difference between them. Thatnbsp;difference, it should be pointed out, is due to the fact that the first ^nbsp;is damaged, and then misrepresented by him who put on the paint.nbsp;I may add that the damage reaches upwards to the D above, and that

* With the forms with ss Pliny�s supposed sasia, � rye,� with vowel-flanked s can hardly have anything to do. Holder makes it Ligurian, and the Welsh for barleynbsp;is haid for an earlier heid, which suggests a Gaulish sasUo-n : the Welsh would henbsp;successively sasHo-n, hehtdo-n, heid, heid, haid, Breton heix, hei, all masculinenbsp;now. The MSS'. of Pliny�s Nat. Historia, xviii. 141 read, however, not sasianbsp;but asia after an s {sub Alpibus asiani), and this latter or rather astio-nnbsp;would also fit the Celtic words : compare Welsh haeam, �iron,� from eisamo-n,nbsp;eihamo-n, heiam, haeam. In favour, however, of the emendation of Pliny�snbsp;word into sasia, one could not help pressing the Sanskrit sasyd, � feldfrucht,�nbsp;Zend hahya, 'getreide.'

there our painter has given that letter the look of a very modern D. Dr. Stokes reads the arrow symbol in the lines of Etruscan letters asnbsp;meaning d, and Pauli makes it into a z. In my opinion it is notnbsp;a letter but a stop, and whether it should be called an artow-head ornbsp;an inverted twig I can hardly say; but for comparison I need onlynbsp;refer to some of the varieties of instances given in De Rossi�s firstnbsp;volume, such as Nos. 317, 339, and 661, also 352, 360, 395, 477, 494,nbsp;542, 585, 586, 588, 612; compare 689 and especially 722, where thenbsp;twig has no less than four pairs of little branches; the number ofnbsp;them is just double that in the present case, but the shape and directionnbsp;are the same. This does not sum up the difficulties of these two lines,nbsp;for the first of them has good Latin CA followed by the Etruscannbsp;form of the letter A. Lastly, we have probably to suppose the Wnbsp;to mean an M upside down; the N is inverted in both the instancesnbsp;into H, which may be said to mean also that it is more Roman thannbsp;Etruscan. The L has the form of the Greek A, which according tonbsp;Pauli is its form also in the Este alphabet of Etruscan.

TOME � EC/1AI

OBAL � ANAT INA

On the hypothesis that this is Celtic�and it is only a hypothesis �I string together alternative conjectures, showing how one mightnbsp;essay the interpretation, (i) In the first place let us assume thatnbsp;OBAL, which, by the way, might perhaps be transliterated oval ornbsp;omhal, meant � and, also, likewise.� The whole might then be renderednbsp;thus: � Tome (daughter) of Ecaaios also waits here.� Tome mightnbsp;be regarded as based on the name Te-tumus and as borne by a membernbsp;of the family of Tetumus. Tome�s name is followed by her patronymic, in which one seems to detect a form of Eccaios, which Holdernbsp;cites from various Celtic coins, including among them some whichnbsp;are ascribed to the Transpadan Roii. But the two a�s offer a difficulty :nbsp;What is one to make of them ? Various conjectures occur to me:�nbsp;1. Take the two a�s to mean a: to say the least of it, that was hardlynbsp;to be expected. 2. Suppose that the inscriber made the mistake ofnbsp;cutting an A instead of a Greek A, then we should have to correctnbsp;his spelling into Eclai, a name which would derive little confirmationnbsp;from Holder. Eccai would have been more to the point if thenbsp;inscriber had not been averse to doubling consonants. 3. One mightnbsp;assume that the two a�s were not intended to be there, that thenbsp;inscriber, hesitating between the forms of Latin A and Etruscannbsp;A, inadvertently cut both on the stone. He had just cut a Latin

C where a K was to be expected, and he went on cutting a Latin A ; then he discovered his mistake and proceeded to make it worse bynbsp;placing an Etruscan A by its side. At any rate, if one of the two isnbsp;to be cancelled, it is doubtless the first, as his vowel is otherwise thenbsp;Etruscan one, which occurs four times in the next line. 4. Lastly,nbsp;suppose he cut not an A but a A, and on discovering his mistake drewnbsp;a line through the middle of his A�a short line, as he did not wishnbsp;to disfigure his work�the result would look an A, while in reality henbsp;regarded it as a deleted A. Whether this is what Pauli meant bynbsp;copying it as an italic I with a point underneath, I, I do not know.nbsp;These two last conjectures come practically to the same thing, namely,nbsp;that the reading intended was Ecai, the genitive of Ecaios, or as thenbsp;coins give it Eccaios. For the present I pass by the word O B A L, innbsp;order to mention that anat would make a good Celtic verb of thenbsp;same conjugation and position in the paradigm as Latin amat.nbsp;Anaim, �I remain, I wait,� is one of the most common verbs in Irish,nbsp;and anat would here have to be taken as proof of the inscription being probably Christian, expressing the idea of waiting fornbsp;the resurrection or the coming of Christ: compare De Rossi, 1.nbsp;No. 317, �expectatque Deum superas quo surgat ad auras�; Le Blant,nbsp;Inscriptions chr�tiennes de la Gaule, No. 478 �diem futuri iudiciinbsp;. . . letus spectit�; and Le Blant�s Nouveau Recueil, No. 17,nbsp;� expectantque diem nunc Domini properam.� Lastly, the adverbnbsp;ina has its in- probably represented by the yn of Welsh yn-a,nbsp;�there (near you), then (of time),� and yn-o, �there, then.� Thesenbsp;words are pronounced y-na and y-no, which separates them from thenbsp;preposition yn, � in,� as in yn-nof, � in me,� yn-nom, � in us,� yn-nocJi,nbsp;�in you,� in which the y is blocked by the consonant and not leftnbsp;open : see p. 5 above.

(ii) As one might search the Celtic languages, probably in vain, for a word like OBAL with such a meaning as that of �also,� let us try itnbsp;with that of offspring, say �son,� �daughter,� or �grandchild�: thenbsp;interpretation would then at first sight be �Tome, daughter ofnbsp;Eccaios, waits here.� We are, however, not bound to treat Tome asnbsp;nominative, and if we try the genitive we have: � The (grave) ofnbsp;Toma: (she the) daughter of Eccaios waits here.� This improves thenbsp;syntax from the Celtic point of view, but it leads to anothernbsp;conjecture which claims a mention, namely, that Toma, genitivenbsp;Tome, is not so likely to be a feminine as a Celtic way of treatingnbsp;Thomas, genitive (Latin) Thomae or Thome. In fact, putting thisnbsp;and Tetumus together, one may expect the key to the whole puzzlenbsp;to prove to have been � Thomas who is called Didymxis,' in the New

CELTIC INSCRIPTIONS OP FRANCE AND ITALY 69

Testament. That is, Tetumus was Didymus, or AtSujaoj, borrowed with the d changed into t in the pronunciation, unless, indeed, onenbsp;should treat the t as an Etruscan X to be given here the value of d.nbsp;Of course, in the latter case, it ought to have been written with Latinnbsp;d; but it has been seen already how little one can calculate on thenbsp;inscriber�s sense of consistency. According to this guess the interpretation might be: � (The grave or urn) of Thomas : (he the) offspringnbsp;of Eccaios waits here.� The possibility of admitting the idea ofnbsp;an urn is warranted by the next inscription to be mentioned.

(iii) One would probably have to regard O B A L as a neuter borrowed from another language, but to borrow a word for son or daughter, ornbsp;even grandchild, would seem less likely than for a particular kind ofnbsp;tomb or urn. The interpretation would be much the same as before,nbsp;except that here the word for tomb or urn is to be treated as given,nbsp;namely, as ohal. In other terms, though the ground or the tombnbsp;belonged to Tetumus and Dugiava, the first person actually buriednbsp;there was Thomas, son of Eccaios, the deceased being a membernbsp;of their family, possibly a grandson. The fact of the relationshipnbsp;may be regarded as sufficiently indicated by the association of thenbsp;names Thomas and Didymus, that is, supposing they went togethernbsp;in the Christian family concerned.

I have sufficiently indicated how I should treat this inscription if I felt sure that the latter portion is Celtic. Having misgivings onnbsp;that point, and thinking it might possibly be Etruscan, I wrote tonbsp;Professor Danielsson, of Upsala, the well-known Etrusean scholar, andnbsp;I asked him to tell me if he thought the two last lines could be claimednbsp;as Etruscan. He kindly replied at once, expressing his view that itnbsp;does not seem to him to be Etruscan. So far as this goes one isnbsp;encouraged to think the words in question may be treated as Celtic.nbsp;But Celtic and Etruscan do not exhaust the list of languages spokennbsp;formerly in North Italy.

xxxvi. Rome. In the Gregorian Museum of Etruscan antiquities in the Vatican is to be seen a bilingual inscription, brought thithernbsp;from Todi, in Umbria. The stone, with nearly the same doublenbsp;inscription on its two sides, forms No. 1 in Dr. Stokes�s Celticnbsp;Declension; C. I. L., I. p. 262, No. 1408, and Pauli�s No. 26,nbsp;pp. 23, 84. A good deal of the top of the stone is gone, with thenbsp;whole of what probably constituted the first line on the side whichnbsp;I call A, and the whole of the first two lines on side B: there arenbsp;other lacunae, but those to which I have alluded are both in thenbsp;Latin. The whole may be given provisionally thus :�

The first question which all this suggests is, why one should have practically the same thing four times over on the same stone; butnbsp;that may be left to be answered, so far perhaps as it can be answered,nbsp;by the details on which one must now enter. Version A of the Latinnbsp;begins with what seems to be the lower half of El ending DRVTEI,nbsp;which occurs in full in line 4 of version B: the point after D R VT]E Inbsp;in version A is a guess of mine. Following this, in line 2, we have thenbsp;lower portions of letters which have been read VRDVM, and guessednbsp;to have signified sepulcrum, if indeed that very word was not the onenbsp;intended. But it needs very little attention to see that the D isnbsp;impossible, and my first guess was that the letter was R, helping tonbsp;make another unknown vocable V R R V M ; but on scanning closely annbsp;excellent squeeze made for me by Dr. Nogara, the head of thenbsp;Museum, who assisted me in every way, I see clearly that the letternbsp;w'as f'l, with its first limb longer than the second, as regularly occursnbsp;in the Etruscan lettering. This being so, it becomes clear that thenbsp;correct reading is urnum, a faulty rendering of the Latin accusativenbsp;urnam, governed by the verbs in lines 5 and 6: on the other sidenbsp;they are �locavit et statuit.� Ategnato Drutei, if that is to be thenbsp;restored commencement of the inscription, stands, as we learn fromnbsp;the Celtic version, for Ategnato Drutei JUio, and it may probably benbsp;regarded as an imitation of Gaulish, as in Doiros Segomari�� D. (son)nbsp;of Segomaros � on the Dijon saucepan : see No. iii above, and others.nbsp;The edge where the first letter of Coisis or Goisis should stand is gone,nbsp;but not so in the case of the F of FRATER : very close to the firstnbsp;R of RATER there have apparently been attempts to scratch an F ;nbsp;but I do not feel at all sure that it was there originally, the F at thenbsp;end of line 3 being intended as the beginning of FRATER, leavingnbsp;Jilius unrepresented even by F, just as with Jilio after Drutei innbsp;line 2. However, worse was to come ; for, when the inscribe!� reached

if .-KI

'r i j f'*', -

gt;-�'.-..kNi^'� Aquot; �* 1.;. nbsp;nbsp;nbsp;quot; y 'nbsp;nbsp;nbsp;nbsp;� i,_nbsp;nbsp;nbsp;nbsp;' %' � �' #'� '�'

V'4�.''�'�i' nbsp;nbsp;nbsp;�.nbsp;nbsp;nbsp;nbsp;* �'�f�f i�

'kK 4'lt;c ''��'i h\,' ��' �' ,� � *.|i

' nbsp;nbsp;nbsp;'gt;' '.gt;'v' .V �w-'-- v.-�i*'.-

. nbsp;nbsp;nbsp;W.r',A� � �nbsp;nbsp;nbsp;nbsp;-nbsp;nbsp;nbsp;nbsp;'nbsp;nbsp;nbsp;nbsp;� . - ' .

PHOTOGRAPH OF DR. NOGARA�S SQUEEZE OF THE TODI BILINGUAL IN THE GREGORIAN MUSEUM OF ETRUSCAN ANTIQUITIES IN THE VATICAN.

Side A.

Beiiveenpp. yo^ yi.

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 71

LOCAV, he cut after the V a letter which he seems to have erased, thereby producing a hollow where it is impossible to read anything.nbsp;Then follows an E, the bottom of which is partly gone owing to thenbsp;edge having been damaged. Whether that E was followed by a T tonbsp;make ET one cannot tell, or whether, in case the T was there, the in-scriber regarded it with the E as making the conjunction it is impossiblenbsp;to say, for his nearest approach to locavit may have been locavet. Innbsp;any case he did not leave room for the IT required to completenbsp;LOCAV into LOCAVIT before he cut the E, which is now the lastnbsp;letter to be read in the line as it stands. The next line begins withnbsp;faint ST, and the whole of it reads ST ATVITQ_yi, with a decided Inbsp;at the end, where one ought to have had E as part of -gue, �and.�nbsp;After these departures from ordinary Latin one is not surprised tonbsp;find that the author of the A version gives us in Latin the twonbsp;spellings of the genitive Drutei and Druti: the variation is, however,nbsp;of importance as suggesting that Drutei was probably an archaism;nbsp;it will therefore not materially help to prove that the inscriptionnbsp;was a very early one. One may now enumerate the eccentricities ofnbsp;the Latin in version A as follows :�1. Drutei for Drutei f.nbsp;2. Urnum for urnam. 3. The inconsistency of writing Druti fornbsp;what had been written Drutei in the previous line. 4. A blundernbsp;either in the spelling of locavit or in the use of et when a -que was tonbsp;follow. 5. The spelling of the latter vocable as -qui instead of -que.

These peculiarities of version A seem to supply a reason why it was thought necessary to have the same legend put into more correctnbsp;Latin. As a matter of fact no solecisms of the kind noticed are to benbsp;found in what is left of version B, which may be assumed provisionallynbsp;to have read when it was perfect ATEGNATO ] DRVTEl � F. VRNAM |nbsp;COISISI DRVTEl -F. FRATER | EIVS | MINIMVS � LOCAVlIT � ET �nbsp;ST AT V IT. The letters intact begin with the SIS of Coisis. Thenbsp;top corner of the D of the next line is gone, and similarly a portionnbsp;of the top of the R at the end of that line is damaged. But as to thenbsp;lettering generally it is very different from that of version A wherenbsp;the letters are of the ordinary shallow kind. Here they are cutnbsp;comparatively deep into the stone, and the sides of the cuttingnbsp;are perpendicular, so that in the squeeze the paper becomes morenbsp;or less detached at the edges of such letters, and the letters themselves are incompletely jointed. Thus a V is represented by andnbsp;even O tends to be O. Moreover the K in lines 8 and 9 is notnbsp;quite such but 1C, consisting as it were of I and C placed verynbsp;closely together, but not actually joined. But here, as in the othernbsp;version, some of the T�s tend to be very like T, and the whole

appearance of the letters is more thick and stumpy than in version A. Everything goes to show that the inscriptions on the two sides werenbsp;carved by different hands.

Before proceeding further it is convenient to discuss very briefly the proper names found on this stone: first comes Ategnatos, of whichnbsp;Holder gives other instances, together with the feminine Ategnata.nbsp;The name begins with the prefix ate, followed by gnatos, whichnbsp;Dr. Stokes equates with the latter part of the Greek Kaaiyi/rjros,nbsp;AioyvrjTos, and the like compounds, and with the gndtus of the Latinnbsp;agnatus and cognatus. Druticnos practically means � son of Drutos,�nbsp;literally, � little Drutos,� and the latter, if it was pronounced Drutos,nbsp;as is probable, equates with the Welsh word drud, in Med. Welshnbsp;dnd, �a hardy man, a hero.� Stokes and Holder cite also thenbsp;feminine Druta, as to which see p. 2 above, and as to the formativenbsp;-icno- compare pp. 6, 11, 61 above. The name of the youngestnbsp;Druticnos only occurs complete in the Etruscan letters as K OISIS,nbsp;which Dr. Stokes has transcribed Coisis: Holder corroborates himnbsp;by citing a single doubtful genitive Coesi from the Berlin Corpus, V.nbsp;No. 5621; but I am inclined to transcribe our instance as Goisis, andnbsp;to equate goisi with the goi of G�idel, Gaedel, Gaoidheal, � a Gael,�nbsp;Welsh Gwydel, � an Irishman.� But the meaning of goisi- is obscure,nbsp;unless we may assume that we have here to do with a form from thenbsp;same origin as Gaulish gaiso-n, gaiso-s, �Vergil�s gaesum, a spear,�nbsp;Irish gae, � spear,� gdide, � armed with the spear, pilatus^ a regularlynbsp;reduced form of Gaisid-io-s or Gdisid-ip-s: as regards the diphthongsnbsp;ai and �i compare the Irish equivalents of Doiros, p. 10 above.nbsp;With a different affix we should have Gdisid-elo-s or Goisid-eh-s:nbsp;compare the names which Holder cites under -elo-, feminine -ela,nbsp;such as Antelus, Bittelus, Cemenelum and others. He remarks thatnbsp;M. d�Arbois de Jubainville considers the termination to have beennbsp;also Ligurian. The meaning of G�idel or Gdidel would probably benbsp;parallel to that of gaide��one who is armed with the spear, anbsp;spearman, a gaesatus\

Let us now take the Celtic portions of the inscription in detail, beginning with version A, which, unlike the Latin above it, does notnbsp;appear to contain blunders: the author of the inscription knewnbsp;Celtic, but he was shaky in his Latin. In the first place let menbsp;remark that the first two or three letters of the three lines are gone atnbsp;the left edge, also the ISIS of Koisis of line 8; while the second K Nnbsp;of line 7 are barely legible. There is, however, no doubt that thenbsp;original reading w^as the one which has been suggested : this is established by the other Celtic version in which every letter is certain.

though the bottom of all the letters of line 11 at the end is gone owing to breakage. The two inscriptions placed side by side innbsp;Roman letters stand provisionally as follows:�

The two versions differ only in the accusative, one having logan, a feminine singular, and the other artuass, which probably is anbsp;feminine plural derived from an earlier artuans: I take logan andnbsp;artuass to be alternatives, neither of them being necessarily a blunder.nbsp;As to the former see Dr. Stokes�s comparisons in his Celtic Declension,nbsp;p. 53, where he cites the lo occurring in two Ogam inscriptions innbsp;South Wales. We have lo, also for log-, in the Med. Welsh golo,nbsp;�bury, burial, interment,� for an early uo-log-, and gwely, �a bed�nbsp;(for early iw-log-ion), plural gwldu (for gvlag-eu�m-lag-) withnbsp;change of vowel as in troed, �foot,� plural traed, while Irish hasnbsp;laige, � concubitus,� also luiglii, pointing to the same double stemnbsp;logio, lagio. The other accusative artuass, meaning perhaps stonenbsp;chambers, seems related to the Irish word art, � a stone, a gravestone �:nbsp;see Stokes, loc. cit., and tiolder, s.v. artos. The two versions havenbsp;been rendered by Dr. Stokes respectively as follows :�

� Ategnati Druti filii tumulum congessit Coisis Druti filius �: �Ategnati Druti filii lapides sepulchrales congessit Coisis Drutinbsp;filius.�

Here it will be seen that Dr. Stokes has not ventured to follow the original order which would have given us � Ategnati Druti filiinbsp;congessit tumulum Coisis Druti filius,� and, for my own part, my habitnbsp;of Celtic syntax makes my difficulty as to the sequence of the wmrdsnbsp;in the original very considerable. In fact I cannot readily believenbsp;that the accusative logan was meant to be separated from the genitivesnbsp;dependent on it by the verb carnitu, and I am forced to think thatnbsp;a construction which would require one to take that view cannot benbsp;the right one. The two last inscriptions, both found in Italy, seemnbsp;to point the way to construe this: they begin with nominativesnbsp;standing alone without verbs. Here also a noun comes at the head andnbsp;stands alone, but in the geniti^^e case, which makes this instance morenbsp;closely articulate than the others: it reminds one of the almostnbsp;exclusive use of the genitive in the Ogam inscriptions of these Islands.nbsp;In other words I would construe thus :�-

The other version with artuass would, of course, run parallel, and whatever may be said of the Latin, �Congessit tumulum Goisisnbsp;Druti filius,� the sequence of the words in �Carnitu logan Goisisnbsp;Druticnos� is idiomatically Celtic, with the verb at the head of itsnbsp;sentence, at the same time that it evidences a preference for lessnbsp;complicate syntax than Latin, as might naturally be expected in thenbsp;case of a people like those of Gaul, who were much less given tonbsp;writing than the Romans were. The two versions are possiblynbsp;in metre, and the division which has just been suggested derivesnbsp;confirmation from the probable division of the legend as follows,nbsp;with the same word ending both parts: take version A : �

Finally there remains one or two comparisons to make between the Celtic and Latin versions : among other things the fact that thenbsp;Celtic commences with the genitive Ategnati, not with the dativenbsp;Ategnato hitherto accepted for the Latin, renders it probable thatnbsp;the Latin began so likewise, and that version B ran thus:�� Ategnati Drutei f. urnam Goisis Drutei f. frater eius minimus locavit etnbsp;statuit.� Next, there is the question, what the words �locavit etnbsp;statuit � precisely meant: did locavit refer to providing the locus, thenbsp;plot of burial ground, or else to the loculus, the special compartmentnbsp;or shelf in the tomb, which was to receive the urn ? I am inclined tonbsp;the former view, and it is for that reason I have proposed thenbsp;equivalent of the Latin locus as the noun on which the genitivenbsp;Ategnati Druticni is to be understood to depend at the beginning ofnbsp;the Celtic versions. In other terms, the Latin gives one to understand that Goisis both purchased or otherwise provided the plot ofnbsp;ground and set up his brother�s urn in the tomb erected there. Thenbsp;Celtic, on the other hand, seems merely to say that the ground wasnbsp;appropriated for Ategnatos, and that Goisis built up some kind ofnbsp;an erection there. The two statements are consistent, but the Latinnbsp;seems to make no direct reference to the erection made there, andnbsp;the Celtic no reference to the urn mentioned in the Latin,

To return to the question put at the outset, it is hardly necessary now to suggest that the spectator did not read the same thing four times:nbsp;doubtless the trial side A was wholly concealed by the stone beingnbsp;inserted in a wall. This carries with it the probability, that thenbsp;verb carn-it- meant not the mere heaping together of stones or timbernbsp;but orderly work, the construction of a regular building.

76 PROCEEDINGS OF THE BRITISH ACADEMY

Nom. EsuSj xxviii. hipviTovs, xi.nbsp;TOOVTlOVSf vi.

trigaranus, xxviii. Uolcanus, xxviii.nbsp;Acc. Esun, xxxiii.

Singular :�

No7n. Elvontiu, xxxi�.

Frontu (Latin), xxvi. Nappisetu, xxxiii�.nbsp;Peroco, XXV.nbsp;OvaXiKto, viii.nbsp;toutio, xxxiv.

Eo'Kt'y'yopei^, XX.

Castor (Latin), xxix. Uritues, xxxiii'gt;.

Brigindoni, iv.

Subroni, xii.

Abyevvopiyi, X. Epaffatextorigi, xxxi.nbsp;KavT�p, vii (?), xxiv, xxiv�.nbsp;obal, xxxv.

Dat.

Acc. Neut.

Dual ;�

Dat. suiorebe, xxxi. Plural :�

Nom.

Dat.

Acc. Neut.

' It should probably be sorted into two declensions at least, but I do not know how : more data are wanted. The spread of the case vowels e and t tooknbsp;place from the genitive and dative, and in some nouns it reached the nominative and accusative hut not in all; see The Englyn, p. 13.

Out of the number of the inscriptions which have here been discussed a group of eight is suspected by the learned Celtist M. d�Arboisnbsp;de Jubainville of not being Celtic at all, but of belonging rathernbsp;to some dialect of Italy ; they are all written in Greek letters, andnbsp;most of them are in the museums of Avignon and Nimes. Butnbsp;as a preliminary to discussing this question it will be convenientnbsp;to have them and the other inscriptions grouped in two lists fornbsp;comparison. Let us begin with those which are not contested innbsp;the same way :�

Etic gobedbi dugiiontiio Ucuetin in(du) Alisiia (Alise-Ste.-Reine, Cote-d�Or).

in.

iv.

vi.

viii.

ix.

xi.

xii.

xiv.

XV.

xvi.

xix.

XX.

xxi.

xxiii,

xxiv�.

XXV.

Licnos Contextos ieuru Anualonnacu canecosedlon (Autun). Seyogapos OvtWoveos toovtlovs NagaucraTi� �icopou BjjA.r)crapunbsp;aoaiv v(\xt]Tov (Vaison, Vaucluse).

xxvii. (1) Tib. Caesare Aug. lovi optum[o] maxsumo su(mm)o Nautae Parisiaci publice posierunt. (2) Eurises. (3)nbsp;Senani Useilo[ni]. (4)........(Notre Dame, Paris).

xxxi. Bratronos Nantonicn(os) Epadatextorigi Leucullo suiorebe logitoe (Neris-les-Bains, Allier).

xxxi�. (1) Oxtjos Uirilios. Ojxros OvtpiXXio. (2) Avfovvos etrod. (3) Eluontiu ieuru Aneuno Oclicno Luguri Aneunicno.

xxxiii�. Adiantunneni Exuertini Nappisetu (Neighbourhood of Thiaucourt, Meurthe-et Moselle).

xxxvi. (Ategnati Drutei f. urnam Goi)sis Drutei f. frater eius minimus locavit et statuit.

Ategnati Druticni carnitu artuass (or logan) Goisis Druticnos (Todi, in Umbria).

These thirty-five inscriptions make up the longer list, concerning which there has been no serious controversy as to their celticity;nbsp;there are a few of them, however, which are partly in Latin, namely.nbsp;Nos. xi, xii, XXV, xxvii-xxx, and xxxi� which is also partly Greek.nbsp;The contested inscriptions are the following:�

To these should perhaps be added a fragment of an inscription found at Substantion near Montpellier and partly restored by Holder

Nimes, St. Cosme or Come, Collias not far from the Pont du Gard, Malauc�ne on the left side of the Rhone, St.-Remy, and Orgonnbsp;near the Durance ; the area implied as belonging to the tribe or tribesnbsp;that set up the inscriptions of this group does not appear to havenbsp;been a very large one.

An article by M. d�Arbois de Jubainville in the Revue Celtique, xviii, 318-24, may be taken as embodying his reasons for thinkingnbsp;that this group of inscriptions is not Celtic. He enters first intonbsp;questions of chronology and arrives at the conclusion that the Gaulishnbsp;occupation of the district in question may be compared in length withnbsp;that of the French domination in Alsace, and adds the followingnbsp;words: � Deux si�cles ne suffisent pas pour imposer definitivement dansnbsp;un pays I�emploi exclusif de la langue du peuple conqu�rant.� Innbsp;answer to this we have, however, to say that one is not clear as to thenbsp;date of the inscriptions in question, and that no chronological argument can be of much avail here until that date is more narrowlynbsp;defined than has as yet been done.

M. d�Arbois�s next argument is intended to prove that the forms of the individual words in these inscriptions fit into the pronunciation ofnbsp;spoken Latin from Ennius to Cicero or later. This he does withnbsp;comparative ease, but when he tries to go further and show that theynbsp;fit better into Latin than into Gaulish, he is less successful; fornbsp;besides a number of minor points on which he is perhaps open tonbsp;criticism, his argument is inconclusive because nobody knows enoughnbsp;about early Celtic to be able to say what forms were inadmissible.

80 PROCEEDINGS OF THE BRITISH ACADEMY

The data for one side of the comparison are too slender: in other words no safe comparison at all is possible as to the details.

The same remark applies to his discussion of /Sparov-Se, where he treats de as a postposition as in the Latin combination voltis-cum andnbsp;quo-ad, to which he adds from Cicero � Quibus de scriptum est,� andnbsp;from Horace � puellis de popularibus,� which, by the way, I cannotnbsp;admit as a parallel to ^parov-bf; and from Lucretius � tempore denbsp;mortis,� together with others of the same kind in Latin. He citesnbsp;authority for the frequent occurrence also of postpositions in Oscannbsp;and Umbrian; but who is to say that they were not as frequent innbsp;Celtic or more so ? He settles this with the rash negative: II n'ynbsp;en a pas d^exemple celtique. According to some scholars the Celticnbsp;language most exactly in point would be Welsh and Breton, and herenbsp;are a couple of Welsh instances at once, pa-h-am (for pa-am), andnbsp;py-rag or pa-rag, as in paham y deuthoch ? � why (literally � what for �)nbsp;have you come ? � and Own paham y deuthoch � I know why you havenbsp;come.� This paham is one of the commonest combinations in thenbsp;language, but the other py- or pa-rag is now obsolete in Welsh,nbsp;whereas it is in common use in Breton as p�-rah � pourquoi.� Fornbsp;similar locutions in Modern Irish see Dinneen�s Dictionary, s.v. cdnbsp;and as, � out of, from.� Comparing roughly the Romance languagesnbsp;with Latin from which they derive, one finds that they make theirnbsp;prepositions into postpositions less often than Latin did ; so perhapsnbsp;one would not be wrong in guessing that such cases were more usualnbsp;in early Celtic than in the Celtic languages of the present day, namelynbsp;Welsh or Breton, and Irish Gaelic. Lastly, the possibility is not tonbsp;be wholly forgotten, that ^parovbt is not to be explained at all withnbsp;the help of de � from �: see No. vii, p. 19 above.

The learned Celtist sums up his case in the following terms : � Ma conclusion est que les inscriptions pr�cit�es nous mettent en presencenbsp;d�un dialecte italique, usite dans la Narbonnaise sous la dominationnbsp;romaine, concurremment avec le latin et avec le gaulois, sans parler dunbsp;grec chez les Marseillais.� But what a peculiar people they were, who,nbsp;though speaking an Italian dialect like Latin, preferred to have itnbsp;written in Greek characters. It is stranger still that they chose allnbsp;to be known, not by Italic or Greek names, but by Celtic ones. Fornbsp;M. d�Arbois admits this when he says; � Les inscriptions pr�cit�esnbsp;paraissent �tre des d�dicaces. Tons les noms des personnages quinbsp;font ces d�dicaces semblent gaulois ; quant au reste du texte de cesnbsp;d�dicaces il appartient, suivant moi, a une langue italique.� Thenbsp;celticity of the nomenclature is a fact which is, it seems to me, wellnbsp;nigh impossible to get over; but the significance of it is not fully

expressed in M. d�Arbois�s admission. For the names not only seem Celtic, but the patronymics also are Celtic in their formulae, whichnbsp;is still more convincing. Take for instance, Kaa-o-irakos Ovepa-iKvos,nbsp;� C. son of Uersos,� and ASpeo-truci'oj, � son of ASpea-o-oy,� in xviii andnbsp;xxii, and compare the case of Iccavos Oppianicnos, �I. son ofnbsp;Oppianos,� in iv, found in the Cote-d�Or, or Andecamulos Toutissicnos,nbsp;� A. son of Toutissos,� in xxxiii�, belonging to Nevers: both havenbsp;the verb ieuru and are supposed to be Gaulish. Next may be mentioned the Collias inscription. No. xxiv, beginning with Exirros Pior-pavLos, � E. son of Riumanos,� where the formula is the same as thatnbsp;of Frontu Tarbeisonios, � F. son of Tarbeiso,� in xxvi. One mightnbsp;here also take into account the forms in -ukos, such as Kaprapos lAAa-vovMKOs and IWiaKos in xvii and xiii respectively, which claimnbsp;comparison with such a name as Anualonnacos in No. v, which is annbsp;ieuru inscription : see p. 12 above. The points of identity to whichnbsp;I refer mean vastly more for the view here advanced than the merenbsp;use of Gaulish names. To them must be added the weight ofnbsp;evidence supplied by the occurrence of the peculiarly Celtic word biovi,nbsp;that is diui = d�ui, � to the goddess �: see the remarks made at p. 33nbsp;on inscription xvi% lost at St.-Remy.

M. d�Arbois de Jubainville has done a service to Celtic epigraphy in challenging the celticity of the group of inscriptions in question :nbsp;to me at any rate they now appear more certainly Celtic than theynbsp;did before his verdict led me to examine them more closely. Instead,nbsp;however, of making them into an Italic group, as M. d�Arbois denbsp;Jubainville is desirous of doing, I am greatly inclined to regard themnbsp;as Celtican. They unfortunately supply us with no obvious testnbsp;words, but that leaves it possible for us to regard them as being innbsp;the language of the Coligny Calendar and of the Rom Defixiones.nbsp;The geographical area, be it noticed, which the challenged inscriptionsnbsp;cover may be said to take in the neighbourhood of Apt, where wenbsp;have, at Saignon, an inscription with carnitu. It is but naturalnbsp;accordingly to suppose that verb to belong to the same language ;nbsp;but that identical form has been found in the Todi inscription, whilenbsp;its plural occurs on the Briona stone now at Novara. This wouldnbsp;mean that Celtican once extended across the Alps far down into Northnbsp;Italy. In another direction we have the fragment at Evreux and thenbsp;Buscilla legend on a vessel dedicated to a divinity at Alesia: that theynbsp;are both in Celtican was suggested in my previous paper. Theirnbsp;interest for the moment, however, is eclipsed by the fact that thenbsp;language which I have been obliged to call Celtican seems to havenbsp;covered the area which, par excellence, belonged to the ancientnbsp;Ligurians.

Besides the foregoing inscriptions, the Coligny Calendar in the Lyons Museum claimed a large share of my attention. Soon after itnbsp;was discovered, I had a passing look at it without being much thenbsp;wiser, and most of my paper read to the Academy last year wasnbsp;devoted to it. So last September I was determined to examine itnbsp;from beginning to end. With the kind permission of M. Dissard, thenbsp;learned head of the Museum, I spent a week collating the fragments,nbsp;with the � Reconstitution � of them into months by M. le Commandantnbsp;�mile Esperandieu, and with the coloured plate or chart of thenbsp;whole published in 1898 as a supplement to the Revue �pigraphique,nbsp;No. 90. It may here be mentioned that another edition of thatnbsp;chart was issued in the Revue Celtique for 1900; but it lacks finalnbsp;revision by M. Esperandieu.

I may begin my corrections by mentioning the fact that in my former paper I forgot to say that the statue of the god, whom wenbsp;may now call Rivos, had figured in more than one publication: fornbsp;instance, in M. Salomon Reinach�s R�pertoire de la Statuaire grecquenbsp;et romaine, vol. iii (1904), p. 234, where he has placed it amongnbsp;the Apollos. It appears also in the Piot Monuments et M�moires ofnbsp;the Acad�mie des Inscriptions, vol. x. pp. 61-90, where it has anbsp;plate (No. ix) devoted to it, and an elaborate article, written bynbsp;M. Joseph Buche, to prove the god a Mars. I may add that I havenbsp;asked the opinion of several of my colleagues, and they also arenbsp;inclined to call him a Mars ; but M. Reinach sticks to Apollo, andnbsp;suggests to me a luminous parallel between Augustus (in the role ofnbsp;Apollo) giving his name to the month of August, and Rivos (the godnbsp;of the Coligny Calendar) giving his to Rivros, approximately the samenbsp;month. For references to Augustus as Apollo, see Mr. A. B. Cook�snbsp;�European Sky-God� in Folk-Lore for 1905, p. 310.

Colunm 1, with an Intercalary Month beginning with the second line (Esperandieu�s � Reconstitution,� p. 3^).

The big letters MID are followed by a bit of the top of the next letter, which cannot, I think, as formerly suggested to me, be an A :nbsp;it looks rather as if it had been X.

To discriminate, if possible, between G and C was one of my chief objects in collating the Calendar.

Day ii. The second line has nothing in sight after SONNA, and I conclude it was treated as a complete word.

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 83

Day ix. The letter following E D V TI seems to have been C, C, or O, possibly the upper portion of an S.

The letter beginning the word following INNIS cannot, I think, have been R, P, or D, but rathernbsp;I, V, M, or N.

In fact the TO of At. viii seem so close to the TIT of the previous line that there appears to be nonbsp;room for a line between. That is, Messrs. Dissardnbsp;and Esperandieu�s arrangement proves correct as farnbsp;as I could judge; for some difficulty was occasionednbsp;by the right-hand fragment with the ends of thesenbsp;entries, having fallen about two lines out of its place,nbsp;as the result, I should suppose, of shifting the glassnbsp;case some time or other since the placing of thenbsp;pieces by M. Dissard. My guessings in my Celtae,nbsp;p. 19, have to be corrected accordingly. The notenbsp;ending the intercalary month has POG, with a distinctnbsp;G. The C of COB is not decisive in its form. Thenbsp;X of OXT is imperfect, and the T is gone ; imperfectnbsp;also is the first limb of the first A of ANT I A.

V. There runs a fracture right through the supposed X of RIXRI, and it is continued through the second I of INIS in vii;nbsp;the shadow cast prevents one from seeing clearly eithernbsp;letter through the glass, but I have no doubt about the Inbsp;and not much about the V, instead of what appears throughnbsp;the glass as X. I should have mentioned that the line of

vii. the fracture is not given correctly in the Chart. I have a note that the whole entry * for vii is N DVMANN InIS-R.

* M. Dissard was kind enough to promise, that, in case of my notes proving incomplete, I might write to him to be reassured on various minor points whichnbsp;might appear doubtful. Such have an asterisk in these remarks. Here my querynbsp;is whether the entry begins with N or with N D as in the ' Reconstitution.�

viii. There is something which looks not very unlike an angular S just before MO ; but it may be no part of the writing.

N.B. The detached bit provisionally placed near the bottom of Dumannios in the first edition of the Chart has been sincenbsp;removed to the Atenoux. of Dumannios in Col. 11 of the 1900 Chartnbsp;and in the � Reconstitution,� which would be here indicated asnbsp;� Esp., p. 5^,� or fourth year on the page representing the monthnbsp;Dumannios : it contains the word RIV RI three times.

Col. 2, beginning with Rivros^ (Esp,, p. 6'). xiii. This numeral is not there, and the reason for its absencenbsp;was not lack of space: was it objection to the numbernbsp;13? The whole line is DEVO RIVO RIVRO(S): partsnbsp;of the RO are still visible, though the S is gone: therenbsp;was room for it. The first arm of the lower V ofnbsp;DEVO is doubled, the outer line being thinner thannbsp;the other : the engraver seems to have hesitated and tonbsp;have thought of a way of doubling the V, thus V, which,nbsp;however, was not what he finally adopted.

At. ii. Of this entry there remains a D ; but it may have belonged to the previous day, for the numerals are gone.

At. xii. Where there should have been a D there is a patch of verdigris which makes it impossible to trace the letter.

N.B. Near the top of this column is now placed a piece which in both Charts is to be found in Col. 14 (Anagantio.s): we shall returnnbsp;to it when the latter is reached.

ix. Here the B of AMB is actually situated beyond the peg-hole. Both go to prove that the peg-holes were made after the lettering, contrary to what I rashly suggestednbsp;in my former paper, p. 18.

ii. The whole entry is gone except the final S of IV 0 S, which stands beyond the peg-hole; that is in Dumannios,nbsp;day iii, in the column to the right, as shown in the 1895nbsp;Chart: it is accidentally omitted in the later Chart.

At. ix. This seems to have begun with it I and ended with R, which is to be found beyond the peg-hole in thenbsp;contiguous month of Anagantios. What can the wholenbsp;entry have been? Hardly itl N INNIS R, since it Inbsp;and the two other arrangements probably refer to certainnbsp;hours of the day as indicated by the sundial, and donbsp;not occur before N or N S, which seem to stand for a wordnbsp;for night. It is more likely to have been like the entrynbsp;in Simivisonnios^, which is also At. ix, namely, it I Dnbsp;A M B R ; but as we want a longer entry, it had possiblynbsp;a month�s name inserted somewhat thus, itl D EDRINInbsp;AMB R : compare Simivis.�, At. iii, with D EQ_VI AMB.

ii. I thought I could identify the MD of this entry, though they have been omitted in the Chart and almost whollynbsp;in the � Reconstitution.�

86 PROCEEDINGS OF THE BRITISH ACADEMY Dumannios^ (Esp., p. 5^).

i. The entry seems perfect and complete, SAMON PRIOVDIX 1 VOS, without anything to suggest a division or abbreviationnbsp;in PRIOVDIX.

V. The entry is . . RINN LAG IT with a G, but query*RINNI. viii. The I of SAM ON I is, I think, there.

i. , , NAG A NT with a very decided G. iv. ...G RIVROS, with the G of BRIG ; but see cols. Sand 11.nbsp;V. There is. . . NIS R there: the latter has been accidentallynbsp;omitted in the Charts and the � Reconstitution.�nbsp;viii. I have a note querying* the L of LOVD in favour of I.nbsp;xiii. What is left begins with the latter part of some suchnbsp;a letter as M, and after a space comes I VG RIV : thenbsp;G seems here certain. Unfortunately the correspondingnbsp;entry in Col. 11 is not in a state to help us on thisnbsp;point. This entry looks as if originally DM I VG Rl V.nbsp;At. viii. The entry is . . . IVX A NAG with traces of the bottomnbsp;of some more letters of P E TIV X : compare P E TIV Xnbsp;two lines lower, and PETI RIVRI ANAG in Col. 8.

Anagantios* (Esp., p. 7*^),

The headline has a very decided G in AN AG AN like that of OGRON at the head of the ensuing month. The name of the former isnbsp;doubtless to be everywhere corrected into Anagantios, and thenbsp;etymology suggested in my Celtae, p. 36, cancelled.

At. iii begins with til, and the usual sequence suggests in At. iv the combination itl, but the engraver blundered into lit,nbsp;and then he made a lower horizontal line through thenbsp;second and third perpendiculars: At. v is normal, I it.

N.B. In the 1898 Chart both Ogronios and Cutios were here left empty, but in the 1900 Chart the portion of Cutios placed in Col. 12nbsp;in the former Chart (and in the Reconst.) is found transferred herenbsp;in the latter Chart, but it has been shifted again, namely, to Elena-bivios in Col. 13 where I found it.

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 87

N.B. The fragment with what remains of days xiii-At. iii has been sliifted by M. Dissard to the corresponding place in Equos in Col. 16.

It is difficult to make out what letter ends the name of the month in the head line: I do not think it can be S. It is more like a tallnbsp;O ; so one should suppose the original letters to have made E D RINIO.

xiii. nbsp;nbsp;nbsp;The reading seems to be it I, inexact for itl, and then, I think,

xiv. nbsp;nbsp;nbsp;Between this entry and the next there is a blank space which

would suffice for about two lines of writing: it is the measure of the extent to which the engraver had got out ofnbsp;his reckoning. The ATENOVX following proves to be onnbsp;a level with that heading in other months.

At the end is DIVORTOMV, but the final V is difficult to trace on account of the verdigris.

iv. The entry is MD BRIG RIV, with room for one or two more letters, but there is verdigris where the othernbsp;letters of RIVROS should come. The last letter ofnbsp;BRIG looks somewhat dubious, G or C.nbsp;viii. I can only read PR IN I LO. The LO are at a distancenbsp;from PR IN I, and close to the edge of the column ; theynbsp;are by no means certain. As to the wide space intervening compare PRINNI LAG in Anagantios, a littlenbsp;lower in the same column.

At. viii. The entry is D PETI RIVRI A NAG, but the I of PETI is not quite certain, as it is in a break badly jointed.

88 PROCEEDINGS OF THE BRITISH ACADEMY

At. X. This reads as follows: N RIVRI D RIVRI lit M. The top of the N is there; the D has a dab of verdigris,nbsp;which makes it look at first sight like a C reversed. Anbsp;part of the M is gone with the punching of a peg-hole.

Anagantios� (Esp., p. 7�).

The name of the month as abbreviated in the head line was AN ACTIO; the G is certain, and so may the AN be said to be,nbsp;though they are no longer intact. A N ACT 10 stood for Anagantio-s :nbsp;had it been a complete word (Celtae, p. 5) it ought to have beennbsp;written AN AXT10, and not AN ACT 10.

i. What is left of this entry is M [D] RIVRI EXOIVO. There is a narrow piece lost between M and RIVRI, where there wasnbsp;probably a D. A peg-hole of the second intercalary monthnbsp;encroaches on the O of IV 0, but it only covers a part of it.

vi. PR INN I LAG has the wide space to which attention has already been directed: it would seem to indicate that thenbsp;things represented by PR INN I and LAG respectively werenbsp;not essentially connected with one another.

Ogronios^ (Esp., p. 8**).

Col. 9, beginning with a four-line introduction to the Second Intercalary Month (Esp., p. 3^).

Line 1. As far as I could judge through the glass, this line would be rendered complete by inserting a V, so as to make itnbsp;CIALLOS BVIS.

Line 3. The corner of a letter at the beginning of this seems to be the top of an A : so the reading would benbsp;AMMAN � M � M XIII. I have no note of my findingnbsp;a point between the last M and the numeral.

between LAT and the numeral, but the presence of the point is somewhat doubtful. There is room fornbsp;some equivalent of�id est� before LAT.

Line 5. What is left of the line is ANTARAN-M, but close before it one observes the top corner of a preceding letter,nbsp;which I have failed to identify ; so I guess the writingnbsp;to have consisted of an abbreviation of the name ofnbsp;the month, followed by Cantaran-M., for canta-rcmnin

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 89 matus, signifying that it was lucky as to its first part.nbsp;The compound would make in O. Irish c�trann, laternbsp;c�adrann, � first part,� and here it seems applied to thenbsp;first fifteen days of the month as distinguished fromnbsp;the ate-nouxtion, � the subsequent series of nights � innbsp;the month. Let us call the month Ingendios, and thenbsp;syntax of the line will be this : Ingendios canta-ranninnbsp;matus, � I. (as to) its first part lucky,� which impliesnbsp;that the other part was not always lucky: possiblynbsp;this means that the month varied in length, contrarynbsp;to my suggestion {Celtae, p. 7), from one lustrum tonbsp;another. The Calendar now proceeds at once to the daysnbsp;in their order ; but here occurs a most serious lacuna,nbsp;which I suggest filling in outline somewhat as follows:�

At. ii. Read tTT MD Q_VTI IN OCRO. This last 0 is not quite certain, but I thought I could trace it. Whethernbsp;the initial symbol was meant to differ from 111nbsp;I cannot tell.

Here one finds considerable space between Giamoni and Elembi and the next Giamoni, the engraver having to makenbsp;this intercalary month fill the room ofnbsp;two ordinary months as in the case of thenbsp;other intercalation.

90 PROCEEDINGS OF THE BRITISH ACADEMY

This name, abbreviated GIAMON, begins with a decided G, and CIA in the next line is to be corrected into GIA ; so also in othernbsp;instances of the name.

viii. EQ_yi PRINNI LAG; the C is there, but has been encroachednbsp;upon by a peg-hole.

These are all on a twisted piece of the bronze, and for that reason I could not seenbsp;the foot of the second L through the glass :nbsp;I have little doubt that it was not I�thenbsp;inclination was wrong for T or I.

The N for M in A N B is there, and is to be put down as a slip on the part of thenbsp;original engraver.

vii. SAMON PRINI LOVD. The last might possibly be read I OVD, but the base of the first letter looks large for I, andnbsp;falls too far below the horizontal direction.

N.B. In the Atenoiix. of this month M. Dissard suggests that there should be placed three detached fragments, the last of which has annbsp;entry beginning with D Tl.

say for certain whether we have a G or a C here; but compare Col. 5 and note the abbreviation stops here. I propose to read in full iiio-gotuatro, and to translate � To the zuos-priest the crops.� This agreesnbsp;absolutely with the first-year entry, except that it is there morenbsp;piously put with the god Rivos as the dii�ect recipient. The ivos-priest was presumably the one who had to do with the many feastsnbsp;or functions held probably in the god Rivos�s honour, and marked innbsp;the calendar I VOS, I VO, IV. As to gotuatros, �priest, a divinity�snbsp;mouthpiece,� see Holder, s. v. gutuatros, and M. d�Arbois de Jubain-ville, Les Druides (Paris, 1906), pp. 1-7. The IVG of Col. 5 (p. 86)nbsp;suggests a shortening of the compound into iu-gotuatro to be comparednbsp;with Riumanios from Riiwmanios in Inscription xxiv.

v. nbsp;nbsp;nbsp;N INIS; I could not trace the R to complete the formula,nbsp;vi, vii, viii, ix. D . . . . IN NI ; The original in vi may perhaps have

been OGRONI as in the three next days, the first two of which are still legible, while only NI of the third is so.

At. iv. Here A M B, which M. Esperandieu rightly found inadmissible, turns out to have been cancelled by means of a horizontal line across each of the three letters: I am notnbsp;sure that the B had ever been completed. The engravernbsp;made another slip below, namely, at the end, where henbsp;placed an X in front of the w ord DIVIR T O M V.

The name of the month in the heading is mostly gone, and what is left is puzzling, being ROM, which would seem to mean that thenbsp;engraver had made Nl into M.

N.B. As arranged when I saw it in September the Calendar has nothing left in this month except a fragment which, in the 1900 Chartnbsp;and in the � Reconstitution,� will be found placed in Dumannios* innbsp;the Atenoux. from iii to x. in Col. 11 ; it is also the detached piecenbsp;near the bottom of Dumannios in Col, 1 in the 1898 Chart. As itnbsp;has three consecutive days with the word RIVRI, it fits into one ofnbsp;the Dumannios columns in point of season, and not at all into Cutios.nbsp;But I understood M. Dissard to say that he had been trying to adjustnbsp;the fragments with due regard to the thickness of the bronze whichnbsp;varies considerably. Furthermore the contents of Cutios^, as givennbsp;in the 1898 Chart and in the � Reconstitution,� appear in the 1900nbsp;Chart in Cutios in Col. 5; but unless I have blundered in my notesnbsp;they have been since moved to make up Elembivios in Col. 13. Itnbsp;is needless to say that this latter move also would not suit thenbsp;almanac view of the case.

SiMIVISONNIOS* (Esp., p. 12 *). i. GIAMO PRINI LAC. The last G is encroached uponnbsp;by a peg-hole, which has caused a slight curling ofnbsp;the bronze where that letter ends, and makes it hardnbsp;to distinguish G from C.

vii(.?viii). M D TIOCOBREXTIO. The T is made in both instances taller than the other letters; the C isnbsp;certain, the two ends being given the same form,nbsp;whereas in a G the bottom differs from the top.

Col. 13, beginning with Eavos* (Esp., p. 12*). ii. PRINI LAC, but the last letter being imperfect one has nonbsp;ocular proof that it is G.

N.B. This month is empty in M. Esperandieu�s � Reconstitution,� but in the Charts it has a piece with incomplete entries covering thenbsp;days from vi to xii. When I saw the Calendar this had been pushednbsp;down the column to the corresponding place in the next month,nbsp;Edrinios, and there it has prefixed to it a fragment with numerals fromnbsp;i to V ; but the compound fragment contains no month name: I havenbsp;identified neither piece in M. Esperandieu�s � Reconstitution.�

At. xiv. M D. This is doubtful, and may have been NSDS. Rivros� (Esp., p. 6�).

the bottoms of CO or GO. Then we seem to have the bottoms of the letters RIVRI without much doubt. Sonbsp;I would read GO � RIVRI, perhaps OC GO (? OCO) � RIVRI,nbsp;meaning that the gotuatros or priest has his crops safelynbsp;at home by him on that day, that is, two days earlier thannbsp;his fellow tribesmen would have theirs.

(v) nbsp;nbsp;nbsp;Nnbsp;nbsp;nbsp;nbsp;INI R to the top of Col. 3, that is, into

impossible it would seem to fit exactly into Anagantios in Col. 5, that is, into that month in the second year. The remaining piecenbsp;given in the Chart to Anagantios At. iii-xiv. has now been shiftednbsp;to the right to the earlier half of the contiguous month of Simivi-sonnios, but as it contains no month name the difficulty is not of thenbsp;same acute nature.

Of the month�s name in the head line only MOM is left, seemingly supplying another instance of M for NI.

N.B. Here comes the bit removed from Equos in Col. 6, and extending to At. iii; see p. 87.

abbreviations to SIM and SI of the month name Simivi-sonnios occur, it seems, nowhere else in the Calendar.

In the course of the same ramble I made an excursion from Poitiers to Rom, in the Department of Deux-S�vres, in order to examinenbsp;the two inscriptions on the lead published in the Revue Celtique,nbsp;xix. 168-176, by M. Camille Jullian : in my paper they come at p. 37nbsp;and occupy the subsequent ten pages. I put these references forwardnbsp;because I wish to confine my remarks as nearly as possible to thenbsp;corrections which I have to make. First of all let me say hownbsp;to find the locality. You take the train for Angoul�me and Bordeauxnbsp;and descend at a station called Couhe Verac some twenty miles fromnbsp;Poitiers. The village of Rom is thirteen kilometres from the station,nbsp;and the little town of Couhe itself is six kilometres, and runsnbsp;an omnibus to meet the trains. At Couhe I found one of thenbsp;sons of M. Blumerau awaiting me with a carriage; M. Blumeraunbsp;is a public notary living at Rom, and he owns the land onnbsp;which he has found the inscriptions and numerous other ancientnbsp;remains, which are to be seen at his house. M. Blumerau is greatlynbsp;interested in them, and he and his family entertained me mostnbsp;hospitably during the day and a half which I devoted to the studynbsp;of the inscriptions. The country around is flat or undulating,nbsp;but I did not see it to advantage as the weather was bad. I wasnbsp;especially disappointed to find the bed of the Dive green and nearlynbsp;dry; but M. Blumerau assured me that the width and depth of thenbsp;channel of the Dive prove that it must once have been a much morenbsp;considerable stream than it is now�it is a stream still in winter. Henbsp;thinks its reduction has been caused by the clearing of the countrynbsp;around of its woods. To this should perhaps be added that itsnbsp;course is in a limestone district, which may leak, and does so herenbsp;and there. Our Dibona is called la Dive du Midi, as there is anothernbsp;Dive somewhere between Tours and Nantes.

On carefully collating Inscription A with M. Jullian�s reading, I found exceedingly little to call in question; my own reading, whichnbsp;was only rendered possible by his, is as follows :�

9. nbsp;nbsp;nbsp;rorio POUR A......

11. nbsp;nbsp;nbsp;A.UIT NAUOUreiA

The A in these inscriptions never has the cross bar; the E in this one is always rounded, and the R is left open, being altogether ofnbsp;a decidedly cursive form. The T when not in combination withnbsp;another letter tends to have its stem twisted forward at the bottom.nbsp;The V varies from that form to a fully developed U, and affordsnbsp;ample room for the marking off of a small V inside it thus, �,nbsp;so as to represent VV, uu or w, as in deuui, for that and notnbsp;deei is the reading in line 6. The D in line 3 approaches thatnbsp;shape, but the others come nearer our d: they come still nearernbsp;to Le Slant�s instances, dated 568 (Narbonne) and 582 (Truilhas).nbsp;The B is a peculiar character, forming a modification of that formnbsp;reversed and resembling Le Blant�s second of the two first instancesnbsp;(from Rome), under the heading � Inscriptions non dat�es,' p. 12, onlynbsp;that ours is better formed. Line 2 begins with a somewhat brokennbsp;�, and it contains two instances of a ligature which is the samenbsp;for NT and ANT, an ambiguity readily understood when it is remembered that the A has here no cross bar. In etiheiont it is nt, but innbsp;Caticanto we require a vowel before the nt: the character is N, withnbsp;the top stroke of a T on the upper end of the last limb of the M.nbsp;Unfortunately it could also be treated as an A with a T joined to it,nbsp;but that seems to be ruled out by the fact that in the first part ofnbsp;the name the A and the T are written separately : it remains that onenbsp;should regard the word as Caticanto rather than Caticato. In linenbsp;3 the Ul are somewhat imperfect, and the L is disfigured by accidentalnbsp;scratches. In line 4 it is hard to say whether LILA or LILLA is thenbsp;right reading: here we have the ligature for N T, as also in line 5,nbsp;where the first B and the C show some accidental scratchings: the

96 PROCEEDINGS OF THE BRITISH ACADEMY

final 0 is imperfect. Line 6 begins with fl for N, and the P of PI A has the lower end of its stem twisted forwards like that of the X '� Lenbsp;Blant gives no less than four undated Instances of this form of P. Innbsp;line 7 0 P are damaged, and the second SI are partly gone. Line 8 isnbsp;rather worse: both l^s are gone except the characteristic top twig:nbsp;the I before the first ^ is also very slight. Line 9: the P resemblesnbsp;the one already mentioned: the A of POUR A is imperfect, and I havenbsp;failed to read what follows it. Line 10 is also bad, and I couldnbsp;scarcely say that I could trace the MTI. I could read nothing fornbsp;certain after the A, but I could see nothing inconsistent here withnbsp;M. Jullian�s reading. Line 11: here the second U is rather indistinct,nbsp;and the last three letters are very much so. Another great difficultynbsp;attaches here to the third character between the U and the N, for itnbsp;is not an N as read by M. Jullian : it is more than N for it ends withnbsp;the top stroke of a T. Are we to read dunt or duant ? Tire formernbsp;might be part of a longer verb ending in -unt = ont, while the latternbsp;might be the whole of a verb�in either case we have a plural; butnbsp;there is a third possibility, namely, that we are to read A and T joinednbsp;together, which would give us a singular form dual or -duat. Againstnbsp;this must be placed the fact that we have AT written separately innbsp;line 2, as already pointed out. I am not prepared with an interpretation, but I am inclined to prefer duant, and in any case tonbsp;suppose that in what follows we have a dependent negative clause.

N.B. The extreme corner of the lead with the greater part of the initial d of this line is ready to drop off: there is no writing on thenbsp;other side of it, so it is possible to mend the lead on that side without any damage being done to these invaluable documents.

Inscription B on the other face of the piece of lead is much more irregular in its lettering than that already described. As a rule thenbsp;letters of side B slope more and show more minuscule forms; but thenbsp;stem of the T is mostly straight. One of the most troublesomenbsp;letters here is G : one of its cursive forms is which we had innbsp;govisa in Inscription A, and the other is which tends in some ofnbsp;the instances to become sprawling and untidy. It occurs here in thenbsp;case of the spirant which I represented by a Greek f in the formernbsp;paper, p. 4)1; but its being so used is perhaps only an accident, fornbsp;we have ZJ also in H�27^0, and alone in POZ^CA, My reading ofnbsp;the inscription is the following:�

3. nbsp;nbsp;nbsp;t,0 ATA/^A I� COM

In line 1 the ligature for AU is badly joined, but its last limb bulges slightly outwards and is altogether unlike the tall sloping I ofnbsp;Imo. In line 2 M. Jullian gives a ligature which he reads as ANT ;nbsp;but I found that the T is all there, so the ligature is confined to thenbsp;AN preceding it. In fact there is in this inscription no instancenbsp;which one is obliged to read as ANT, and we have trouble enoughnbsp;without it. For we have H to be read AN : the only time an n occursnbsp;here out of combination with another letter it is found written n,nbsp;namely, in line 5; and we have an N with the top stroke of a T,nbsp;a ligature to be read sometimes as NT and sometimes as A and Tnbsp;merely joined together. We have the joined AT in the first syllablenbsp;of Atanto and the ligature for AN in the second. The last threenbsp;letters of this line are very puzzling ; the middle one seems to be annbsp;� preceded by what seems to be the first half of an H, but I cannotnbsp;trace the horizontal line produced quite to join the �, though thatnbsp;seems to have been intended. The last letter seems to be a G of thenbsp;kind already mentioned as and I think we have a somewhat simplernbsp;instance of it in line 6, where M. Jullian has taken it to be eithernbsp;a T or a G : I regard it as more probably the latter, with the horizontalnbsp;line not joining the top of the stem but intersecting it�a form of 2^.nbsp;Line 6 ends with an uncertain 0, which M. Jullian thinks possibly annbsp;�. In line 8 the letter ^ is a very straggling specimen in the firstnbsp;instance, while as the last letter but one it is very much simpler: I agreenbsp;with M. Jullian in thinking that the same letter was meant. Innbsp;line 9 the ligature for AU comes somewhat nearer AV,but the secondnbsp;joint is damaged so that it is not easy to judge of the exact shape,nbsp;except that it is still different from N. In line 11 I cannot read NAnbsp;because the first of those two characters, the N, finishes with the topnbsp;stroke of a T ; and it seems to me now that this is also the probablenbsp;reading of M. Jullian�s copy, though he has preferred NA. Atnbsp;the end of the same line I looked in vain for traces of I to make SS�nbsp;into SSI�, but I found none, and the verb seems to be demtisse,nbsp;on a level here with atehotisse. Of line 12 I could make nothing but

V Pe, followed by traces of which I could only read an A at a distance ; but M. Jullian suggests, subiect to a triple query, the followingnbsp;reading: UZIETIA 0 ... PA ... A,

So far of the reading of Inscription B: other questions, however, present themselves, and among the first that of the symbols for thenbsp;spirant sound of gh, which we have here represented by in CH;^A,nbsp;by innbsp;nbsp;nbsp;nbsp;and by a sprawling variant in ONT�^ATIM

and �2^01^1 A : the reading of P02^�A in line 6 is too uncertain and the origin of the word too obscure to be of any help. One seemsnbsp;safe in drawing the inference that the author of this inscription feltnbsp;it to be desirable to distinguish the soft spirant sound of gh from thenbsp;stopped sound of g-, but he hesitated at first between andnbsp;and at last made up his mind for the modified form of G which is herenbsp;represented by ij, and to use that alone without the aid of H or 7^.nbsp;If we may treat this as the case, we see at once the extent of the errornbsp;which I ascribed to him in my previous paper. There was no pause,nbsp;or hardly any, between TG andnbsp;nbsp;nbsp;nbsp;so h was admitted in the

hiatus, making the words in pronunciation into TC H�ZJi^O, whereas I MO was not so closely taken with GH^A, and therefore there was nonbsp;occasion for a hiatus aspirate. He only made one slip�he wrotenbsp;�Hz;A ATANTO for what should have been �H;^0 ATANTO. Innbsp;the next place demtisse, while parallel to atehotisse, differs from thenbsp;devitissie of Inscription A. It was a difference of spelling whichnbsp;had perhaps to do with a sound like that of English sh�possiblynbsp;there were two pronunciations, one with ss and one with sh. Thisnbsp;completes the certainty that the same man did not write the twonbsp;inscriptions.

It is not improbable that ata demtisse is to be treated as a subordinate clause in the affirmative and corresponding to na demtissie in the negative. In that case ata may have had the force of Latin at,nbsp;and be in some way related to the ate of ate-hotisse. At all eventsnbsp;we should have to distinguish both ate and ata from the prefix eti innbsp;eti-heiont in Inscription A.

Both in the Calendar and in the Defixiones I have kept as closely to the text as it was found convenient. A revision of thenbsp;conjectures in my last paper concerning these documents would takenbsp;up too much of the space at my disposal, not to mention that it wouldnbsp;most likely prove premature; for it is only now that those conjectures are beginning to be discussed. When Celtic scholars havenbsp;given their opinions, I expect to find some positions to defend, andnbsp;some, doubtless, to relinquish.

�Rcste a savoir si nous dirons cfxtp: ou i-iguee: c�est I�eternelle question.� So ends one of M. Camille Jullian�s �Notes gallo-romaines � in the July number of the Revue des �tudes Anciennes,nbsp;and the interest which he feels in that question makes him return tonbsp;it every now and then. That is not all, for since the foregoing papernbsp;was written and presented to the Academy, I have read his contribution to the volume of M�langes recently dedicated to M. d�Arboisnbsp;de Jubainville on his seventy-eighth birthday. It is entitled � Lesnbsp;Salyens Celto-ligures,� and it goes, I think, some way to solve thenbsp;Celto-ligurian question. He lays it down that in the centurynbsp;following the foundation of Marseilles the native peoples of Provencenbsp;were Ligurians, and that the Ligurian tribe of the Salyes ornbsp;Saluvii possessed the town of Arles, and had the command ofnbsp;both banks of the Rhone. Now the territory held by the Salyes onnbsp;both sides of the Rhone, together with that of kindred Liguriannbsp;tribes up to the latitude of Orange, let us say, would include easilynbsp;the localities already mentioned as signalized by the PpaTOvbe-Kavrevanbsp;group of inscriptions (pp. 77-81). The conclusion is, therefore,nbsp;hard to avoid that it was the Ligurians of a later age, but still pagan,nbsp;who set up those inscriptions. Add to this that the carnitu inscriptions, which by their provenance seem to be inseparably connected,nbsp;indicate, as has been suggested at p. 81 above, that the same languagenbsp;extended across the Alps far into North Italy. In other words thisnbsp;means that what I have called Celtican was practically one and thenbsp;same language as that which M. d�Arbois de Jubainville callsnbsp;Ligurian. In fact, I may say that ever since that distinguishednbsp;scholar wrote to show that Ligurian must have been an Aryannbsp;tongue, I have had the idea present to my mind that this was tlienbsp;Continental idiom akin to Goidelic, as Gaulish was to Brythonic. Sonbsp;to me it becomes more and more a question of names, whether it isnbsp;to be called Celtic or Ligurian. I received a month or two agonbsp;a letter from one of the most brilliant of living Frenchmen suggestingnbsp;that the proper name for the Celtic family is Ligurian; and he basednbsp;his opinion on a passage of Lucan�s, which he quoted. That is, however, not the usual attitude of those who are attached to the termnbsp;Ligurian : they seem inclined to treat it as a terminus or as a symbolnbsp;standing for an unknown quantity, but most of them are little concerned to try to work out the equation which should give usnbsp;approximately the philological value of their X. There is, however,

one other great exception, and that is, again, Camille Jullian, who writes to me as follows; � Que d�ailleurs cette langue ligure ne futnbsp;point trop diff�rente de celle des Celtes, je le crois,� and then henbsp;proceeds to indicate his reasons for that conclusion by referring tonbsp;Varro, Strabo, and Livy. In other words, his careful reading ofnbsp;ancient authorities lead him to a conclusion practically identicalnbsp;with that which I draw from studying the Coligny Calendar and thenbsp;Rom Defixiones. It is this; whatever you call the language of thosenbsp;documents, the key to it has proved to be Goidelic. Some of mynbsp;critics would say not Goidelic but Brythonic: even so the keynbsp;remains Celtic. The same conclusion follows from M. Jullian�snbsp;examination of such a name as that of the Ligurian tribe of thenbsp;Segobrigii. He is probably right in treating it as Ligurian, but nonbsp;glottologist whose attention had not been drawn to Ligurian wouldnbsp;have dreamt of its being anything but Celtic. It is a compound thatnbsp;reminds one of the personal name Netta-Segamonos, which occurs nonbsp;less than three times in the Ogam inscriptions of the Goidelic tribenbsp;of the D�ssi of Co. Waterford. It seems to have meant thenbsp;�Champion of Segamo,� that is of the Mars Segomo, whose cult,nbsp;as shown by Holder, extended from Lyons to the Cote-d�Or and fromnbsp;Nice to the Jura.

P. T. To the kindness of Com. Esperandieu, who is conducting the excavations proceeding at Alise, I owe what seems another instancenbsp;of etk. The reading of the first letter is not quite so certain in thisnbsp;instance as that the word in it means � and.� It comes between twonbsp;datives closing a recently discovered inscription there, which endsnbsp;thus:�

P. 12. Instead of treating Jioedl as related to -s�dhn, I should probably have equated them ; for the former was at one time hoedl,nbsp;as stated by Dr. Davies, s. v., and as proved by such an inscriptionnbsp;as that of Gwnnws ending with � Hiroidil filius Carotinn,� H�bner,nbsp;122, and by such names in �The Book of Llan Dav� as Guorhoidil,nbsp;p. 189, and Rihedl, Rihoiihil, or Rioidyl, pp. 149, 152, 155, 156.nbsp;Add to this that Casnodyn, a poet who lived early in the fourteenthnbsp;century, makes hoedyl alliterate with hedwch, that is hedwch: seenbsp;J. Morris Jones�s Welsh Grammar, � 189, vi. He suggests thatnbsp;the fact of hoedl being feminine is owing mainly to the influence ofnbsp;its feminine synonym oes, � a lifetime, or generation.�

P. 41. 1 forgot to point out that the Collias inscription (No. xxiv) appears to be in metre: compare the one at Alise, p. 9 above.nbsp;The former seems to scan as follows:�

The same treatment may have applied to Nos. xvi�, xxii, and xxiv��, but they are too imperfect to be pronounced upon. With No. vii itnbsp;is different, for there one would only have to correct K A N T�M intonbsp;KANT�N, rather than KANTeNA, and treat TAPANOOY asnbsp;TApavoov. In any case the remark on p. 18 as to the accentuationnbsp;of that vocable is to be cancelled as probably inapplicable to Celticannbsp;forms.

P. 69. Since the conjectures on pp. 67-9 were printed, it has occurred to me that ohal equates with the Irish uhhall, �an apple,�nbsp;in medieval spelling uiall or uhull, formerly neuter; but this yieldsnbsp;no satisfactory meaning unless we assume, that, besides the sensenbsp;of �apple,� the word had that of �offspring or child, Kapirbi rrjsnbsp;6(r4gt;vos (Acts ii. 30).� That we may do so becomes a certainty whennbsp;the fact is recalled, that POMMIO, a word of the same origin,nbsp;doubtless, as the Latin pomum, � apple,� occurs in the sense of � offspring or son� in one of the Rom Defixiones; see Celtae, pp. 41,nbsp;42. In the light of this comparison obal may be treated as anbsp;neuter singular: compare navrfv, plural navreva (p. 40 above), and asnbsp;to Irish ll, the names Conall, in Welsh Cyrvwal, from CVNOVALOS,nbsp;and Domnall, in Welsh Dyfnwal. The O. Welsh for �apple� wasnbsp;abal, now afal, while aball, later afall, was � an apple tree,� whencenbsp;aballenn, now afallen, the actual word for that tree. The O. Irishnbsp;for the same appears to have also been written aball, '�fiaec malus��nbsp;(Gram. Celtica, p. 769�), which survives in abhall-ghort, �an orchard,�nbsp;Sc. Gaelic abhallghart; but abhall- seems to be ousted by ubhall-,nbsp;�apple,� except perhaps in Munster where the word for orchard isnbsp;pronounced abhl�rd: see Dinneen�s Irish Dictionary. The Irishnbsp;aball looks like a loan from Welsh, where aball admits of an easynbsp;explanation as a feminine formation derived from abal-ia. Needlessnbsp;to say, this leaves unreduced the difference between Welsh abal andnbsp;obal or uball, � apple.�

The Celtic Inscriptions of France and Italy

Professor John Rhys

London

155

A. G. van HAMEL

THE CELTIC INSCRIPTIONS OF FRANCE AND ITALY

4 PROCEEDINGS OF THE BRITISH ACADEMY

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 5

PROCEEDINGS OF THE BRITISH ACADEMY

16 PROCEEDINGS OF THE BRITISH ACADEMY

eCKerrAIBAANAOOYIKOYNIAI

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 21

AAr�NNOPiri

OY�P�TO///MAP�OOYI

........NITOYL

.... ARNOC

26 PROCEEDINGS OF THE BRITISH ACADEMY

30 PROCEEDINGS OF THE BRITISH ACADEMY

CELTIC INSCRIPTIONS OF FRANCE AND ITALY S3

BINNAMOC

AITOYM

Apeoc

M 3

ccKirro

peizKo

NAIAAe

oc

MATIACo ....

KONNoYBP....

......AAPESSIKNOS

......YIBPATOYAEKA

PROCEEDINGS OF THE BRITISH ACADEMY

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 43

RATN BRIVATIOM

FROITV TARBBISONiS IBVRV

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 51

52 PROCEEDINGS OF THE BRITISH ACADEMY

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 61

o

o

o

o g ^

o o

o

TOME � EC/1AI

OBAL � ANAT INA

CELTIC INSCRIPTIONS OP FRANCE AND ITALY 69

S'

'r i j f'*', -

' nbsp;nbsp;nbsp;'gt;' '.gt;'v' .V �w-'-- v.-�i*'.-

. nbsp;nbsp;nbsp;W.r',A� � �nbsp;nbsp;nbsp;nbsp;-nbsp;nbsp;nbsp;nbsp;'nbsp;nbsp;nbsp;nbsp;� . - ' .

Side A.

Beiiveenpp. yo^ yi.

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 71

76 PROCEEDINGS OF THE BRITISH ACADEMY

80 PROCEEDINGS OF THE BRITISH ACADEMY

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 83

86 PROCEEDINGS OF THE BRITISH ACADEMY Dumannios^ (Esp., p. 5^).

Anagantios* (Esp., p. 7*^),

CELTIC INSCRIPTIONS OF FRANCE AND ITALY 87

88 PROCEEDINGS OF THE BRITISH ACADEMY

Anagantios� (Esp., p. 7�).

Ogronios^ (Esp., p. 8**).

N CIAMONI ELEMBI

N CIAMONI AEDRINI

90 PROCEEDINGS OF THE BRITISH ACADEMY

9. nbsp;nbsp;nbsp;rorio POUR A......

11. nbsp;nbsp;nbsp;A.UIT NAUOUreiA

96 PROCEEDINGS OF THE BRITISH ACADEMY

3. nbsp;nbsp;nbsp;t,0 ATA/^A I� COM