L, IC^DI
Determinative prefix of
personality.
** Hittite "(?).
Determinative prefix of
individuals.
**Toestablish", ''make".
Determinative affix of
parentage or race.
''This'Y^o^.^;.
Determinative affix of
plurality (?).
iuwe, *' prince ".
Determinative of supre-
^^ Ir
macy.
khaiiu, khat.
Determinative of action
or power. Signifies
*'great".
Determinative of image (kue).
Determinative of speaking (e-we).
i?a (?),** and".
li II «\ //
we, ve.
Ideograph of *' king " (tarkus).
eSj nominative suflBx.
Ideograph of *' country ".
er.
me.
'' High-priest ".
tu.
''Priest".
ku.
tf~^
"Heaven", ''above".
kus.
'' Earth ", " below ".
kus.
" The sun ".
•ji
kuice, kue.
*'Sear\'' inscription".
sur.
^'Tocommand".
**Great".
sar.
tar.
VU
Sanda, *' the god Sandan ".
bp
ère, er.
icr
Tarku, ** king ".
<sd
ewe (?).
er{^).
tt d
Dimej '* shrine ".
the determinative, précèdes the first proper name. I would now call the character the determinative of personality.
The pronunciation of the ideograph of '' king " was tarkus, and it interchanges with the crook which I hâve doubtfully read scu^me. Both thèse facts resuit from a study of the inscription of Mar'ash. If we dénote the fîrst ideograph (tlie royal tiara) by X^ and the second (the crook) by Y, we read as follows : line 1, tcu^kus (the goat's head, the value of which is known from the bilingual boss) X; line 2, tar + tarkus^ Y, iuterchanging with tarkus in line 1; line 3, tar-kus Y; line 4, X kue Y. Comp. H. iv. 1, tar-kue taj'ku(s),
The determinative of supremacy is written over the determinative of divinity at Gurûn. At Ivriz, as well as in J. i. 2, 5, it is attached to a human head w^hich is crowned with a tiara. In J. i. the head is preceded by the ideograph of " king"; at Ivrlz (ii. 2.) the ideograph of '' king " does not appear. In place of it we hâve a character formed of a horizontal line below which is the lituus carried by the priest at Boghaz Keui. I conclude, therefore, that the crowned human head when preceded by the determinative of supremacy signifies " high-priest ".
The determinative of supremacy must also ideographically dénote * ' heaven ". In the fragment published in plate XI of Wright's Hittites we find (fig. 4, line 4) what must mean " earth and heaven ", or better " what is below and what is above ". The ** earth " is represented by the boot (kue); that this signifies " below ", ** ground", and the like, is shown by the complicated symbol at Boghaz Keui in which the priest who supports the winged solar disk on his head stands upon a boot. The ideographs of " earth " and *' heaven ", '* below " and '' above ", hâve attached to them the suffix kus with a staff underneath; this indicates, perhaps, the passive participle. The copulati\'e conjunction is expressed by the head of an ass.
The character which I hâve read se should rather be me. At ail events in J. i. 3, it is the phonetic complément of a character which is probably to be identified with the dimeoi the boss of Tarkondêmos. If the identification is right, we must read Dime-meku-er by the side of Dime-ku-er in the same liçe.
The altar which 1 hâve supposed to hâve the value of si is rather to be read kus. It interôhanges with kus (e. g. J. ii. 8 compared with M. front, last line), and frequently is added to kus (e. g. J. ii. 7). A cumulation of characters, I may observe, is one of tlie characteristics of the Hittite mode of writing. The same syllable is often represented by a determinative or ideograph as well as by a phonetic character, and two or even more phonetic characters may be used simultaneously to express it. Thus the word tarkus may be written tarkus -f kus 4- kus -f ku -k- es -¥ ideograph.
The character which I hâve given in Wright's Hittites p. 177 as having the value of sar occurs on the British Muséum seal 5, but is misprinted in Wright's plate (XIII). It follows, on the seal, the ideograph of the god Sandan, and for reasons which I hâve given elsewhere forms part, I believe, of the name of the Kilikian king Sanda-sarve whose daughter was married to Assur-bani-pal. The name on the seal will read Sanda-sar-vees. The character is identical with that which follows the name of Sandan in T. 3, B. 1, 2,
where the proper name seems to be Sanda-sar-e'me-kuisyes , Sanda-sar-me-ku.
The crook, to which I had assigned doubtfully the value ofsarme, is, as we hâve seen, an ideograph of '' king", and consequently was pronounced tarkus. But it also appears to hâve acquired the Assyrian value of sar from the Assyrian sarra "king". At ail events four of the British Muséum seals (2, 6, 7, 8) which were found in the archive-chamber of Assur-bani-pal along with the seal of '* Sanda-sarves " contain an inscription which I should now read id. Sanda-sar-u ** the seal of Sandan-sarves ", the syllable sa/' being denoted by the crook.
The boot reads kue or kuwe only, never mesi as 1 had supposed. Instances of its interchange with Àra-e, ku-we, and ku-u^e are fréquent.
We hâve already seen that the numéral 4, whih corresponds to the Assyrian ]^ in the bilingual inscription, niust be read oe or œe, not me. Hence its interchangeability with u-e and even u.
The ideograph of a pôle the upper part of which has been carved into the form of a human face does not signify '' worshipping ", as I imagined. The corrected copies of the inscriptions of Ivrîz hâve revealed its true meaning and at the same time hâve thrown a flood of light on the interprétation of several of the texts. The ideograph appears as the determinative of the word ku-u-e and ku'Va{^) in H. v. 3; cfer. also H. IV. 1, 4. kU'We-ku, Now ku-e is the first word of the two inscriptions (1 and 2) attached to the figures of the god and his worshipper at Ivrîz. In the case of the figure of the god it is foUowed immediately by the determinative of divinity and the name of the god Sandan ; in the case of the figure of the worshipper by what we shall see is a word signifying " prince " and a proper name which occurs in T. i, A. 3 and B. 2. It must ar.cordingly mean " image *' or '* likeness ", a meaning which fuUy explains the form of the ideograph which sometimes accompanies it.
The discovery of the signification of the word kue (which reminds us of the Vannic gis " an image") clears upagood many passages in the inscriptions. In I. i. 1. it is associated with the column followed by the suffix of the nominative (es). Similarly in J. m. 4, we hâve ku-kue-u-e followed by the name of a god and the column with the suffix es. Now the column must convey the idea of stability, and since the wings of the solar disk are supported by two columns in the complex emblem found at Boghaz Keui, we may conclude that it signifies *' to establish ", '* build "or *' make ". The first words of the Ivrlz inscription will accordingly mean *' the image of Sandan making " or ^' having made ", the last word being a participle in the nominative, agreeing with the proper name which foUows.
In H. V. 2 kue is coupled with a word which is represented by two altars, below which is written the character kus and a character which varies in the différent places where the word occurs. We hâve seen above that the altar had the value of kus ; I believe therefore, that the stem of the word in question must be read kus, the two altars being idéographie. The word must signify either *' altar " or *' shrine " ; perhaps the fact that two altars are always depicted, and not one only, indicates that the meaning is " shrine ". Moreover in J. m. 2, 4, kus or gus '' a shrine " is associated with an ideograph which
1 have identifiée! with the dime of the Boss of Tarkondêmos, and which seems to represent a house \ The translation of the two passages wouldbe : (1) " altars I have erected and (ué) a temple [dimé) [I have built] ", (2) '* an image of the god X making and a temple... making ". The word I have rendered " I have erected " is expressed by the picture of a mason's angle and the determinative of active opération (a hand holding a dagger).
We can now proceed to decipher a little more of the Ivrlz texts. In I. u. 1, between kue and the name of the prince there intervenes a word which is represented by the phonetic character tu and an ideograph which seems to dénote a fringe. In B. 3 the same two characters recur, but above them are the determinatives of supremacy and personality. This enables us to identify the group of characters with the second word in H. I, n. and in., only hère ihe place of the ideograph is takeri by the phonetic characters we-es, The whole word consequently reads in the nominative tu-we-es; it foUows the word " Says ", is separated from the proper name which succeeds by the oblique line which dénotes the end of a paragraph, and is preceded by the determinatives of supremacy and personality. It must, therefore, signify ** prince". The.form of the word reminds us of Tou, the king of Hamath in the âge of David. The ideograph must have been pronounced tuwe or iuwes since it is written without the phonetic complément in B. 2.
The proper name which foUows tuwe in I. ii. is spread over two Unes. It appears in tlie nominative in T. 1. where the inscription begins with the words : D. P. E'U-{ri)e-es D. P. X-e-me-Kus-es ID. ID. Que-es " Eu(r)es the son of ...emes the Quan king ". The value (ri) which I have assigned to the third character in the name is purely conjectural : its sole support is the fact that the king of the Que in the time of Tiglathpileser III. was called Urikki, which may be the Assyrian équivalent of Eure-kus " the son of Eures ". At Andaval (A. 3) we find E'U'{n-)e, as also in I. m. and B. 2. In 1. 1. 2. we have E-U'{ri'')e-tar-es ; the same form occurs in B. 1. Hère the words which immediately précède throw light on its signification : '' The speaker is Sanda-sarwe-kus this man of the king, the son of Er..., the son of Sanda-sarwes, the... of the king, who belongs to Eu(r)es {Eure-tar-es) the Quan ". Consequently the suffix toi' dénotes ''belonging to ", or perhaps more specially " ministerof ". Since tcu^kus "king" terminâtes with the patronymic sulïix kus it is probable that the root tar signifies " to govern ". In Vannic tar is '' strong ".
I have rendered the '' twist" which is found above the human face after the name of Sanda-sarwe-kus by the pronoun '' this ". That such must be its meaning is évident from the inscription on the bowl from Bagdad (Wright pi. XXV.) which commences with this character followed by the picture of a bowl. So, too, in J. 1. 1. which begins with the words : '' Says the prince ([tu-]u-es) X-we-tarkus the king of the Hittite (?) country, the son of the god : this image... " Cp. also H. v. 3. where after the name of
(1) It will be seen that I have thas been led by a différent line of reasoning to substantially the same conclusion as that at which M. Menant bas so acutely arrived in bis ** Études hétôennes " (Recueil, XIII, 3, 4, p. 141).
ihe god Uva (?) -tarkus {lar-kue-tarkus-kus) we read " this image " or " thèse images ''.
It is now possible to translate the greater part of the first two inscriptions of Ivriz. The first runs : ** Imaginera dei Sandan faciens rex fecit minister Eu(r)is E... tu... es ". The last five characters of the text I must leave untranslated. The second inscription is "' Imaginera principis Eu(r)i8 pontificis (feci) ".
The sensé indicates that the last word is the first person singular of a verb which must therefore terminate in -ku, The Haraath texts confirm this conclusion. In H. i. 3. and II. 3. the suffix -ku is attached to the colunm, which, as we hâve seen, signifies " to make ", and the word/' altar " or ** shrine " foUows immediately. As both inscriptions commence with the words : '' Says the prince ", it is obvious that we must translate in the third Une " 1 hâve made the shrine ". Ku, accordingly, will be the suffix of the first person singular, and consequently ku-DET.'We-ku in H. iv. 1. will mean " I hâve made the iraage ".
From H. i. 3., ii. 3. we can also infer what was the suffix of the third person singular. After ku8 (or gus) *' shrine " coraes tu; then an ideograph representing a hand grasping a staff with the phonetic coraplement va (?) -e, and then the narae of a king who is not raentioned elsewhere in the inscription. In i. 3. the royal narae is replaced by a patronymic. The hand with the staff can hardly signify any thing else than "command, order". The translation of the passage must therefore be : '' I hâve made the shrine which X (or the son of Y) the king ordered ".
It would appear that the prince who records his narae on the stones of Haraath acknowledged obédience to a higher lord. Now the father of this higher lord, Sanda- *ues has the same narae as the king to whora belongs the inscription on the lion of Mar'ash, tho' it is true that an additional eleraent seeras to enter into the narae of the latter. But the narae would seera to be Koraagenian. Are the Haraath texts, then, records of a Komagenian conques t of Syria ? It is even possible to read the narae of the father of the Hamath king as Kustaspis, a name borne by two Koraagenian sovereigns according to the Assyrian raonuraents. At ail events the name begins with Kus and ends with e, two characters alone intervening. If this is the right reading of the narae, the deraonstrative " this " will hâve the value of tas.
The prince who caused the inscriptions of Haraath to be engraved, and whose narae ends with . . . er(?)-me-e5, gives himself a title, which occurs frequently in the inscriptions (e. g. M. 2, 3, J. H. 7, J. VI. 2, T. 2), in positions where it must signify '' great " or " powerful ". It is represented by an out-stretched arm, with the deterrainative of personality below it, and usually (tho' not always) alsoabove it. At Haraath, the gentille deterrainative affix (kus) is attached to the ideograph. Consequently the first Une of H. ii will read : E-voe d. p. d. p. tu-toe-es : X-Y-er(2yme-es id. d. a. ku-es Kus-ta^{iypi(^)' Says the prince ...er(f)mes thegreat the son of
e^kus ID. ID. Kustaspes (?) the king of the country.
The narae of the country over which he ruled cornes in the second line, and seems to read Er-^. Now the buU's head had the value of er. This is clear frora a comparison
RBCUBIL, XIV. 7
of J. I. 2 and 3 where in the royal name we hâve e-BULL's HEAD-er in the one instance and the buirs head alone in the other \ The king at Carchemish calls himself " the king of the country of the bulFs head " (J. n. 1) where we must accordingly read " the country of Er " or ''Ere ". In B. 2 mention is made of '' the man of the country of Er-e ", and since this is conjoined with '* the man of III. ", while at Carchemish '' the country of the bull's head'* is similarly conjoined with *' the country of III. ", we may conclude that Er or Ere was the name of a district which adjoined Carchemish, or within which Carchemish was situated. According to the Assyrian monuments Ara was the name of a district of Hamath, and in the North Syrian List of Thothmes III. the country of Ar is mentioned (no. 134). Is it possible that in J. ni. 3 we should read : D. P. E-re - covNTRY - kus carchemish ' -e - tos va (?) lïl -kus,
** the man of the country of Ere, of Carchemish and of the Hittites " ? In this case we should hâve determined the value of a character (as re) which seems identical with that which I hâve above conjecturally read pi in the supposed name of Kustaspes.
The Word ewe **he says", at the beginning of the inscription, has already been suflBciently discussed by me in Dr. Wright's bock. I hâve there, however, read it eme instead of ewe, Since -wê (also written -ue) is a suffix, the root is e which corresponds with the Vannic a '' to say ", ''speak". -É't/je is either expressed by the ideograph of " speaking " folio wed by we, as at Hamath, or by the ideograph foUoweid by the fully written word ewe, as in J. i. 1, or by ewe without the ideograph as in M. 1. At Mar'ash and Jerablûs the suffix -kus is added to ewe, the signification being, I believe, "the speaker (is) ".
In J. II. 1, Mr. Rylands has long ago recognised that the place of the ideograph is taken by the hieratic outline of the human face. Between the latter and the word e-wekus is found in B. 1. the head of a sheep which would consequently appear to hâve the phonetic value ewe. It is not certain, however, since the human face is shown by a number of passages to hâve signified '' hero ", *' great man ", or something similar, and when it signifies *' to speak " it ought properly to be represented with the word issuing from the mouth, as in J. m. 2, 5. At the end of the inscription of Mar'ash, for instance, the hieratic outline of the face, followed by the suffixes we-kus and preceded by the determinative of personality is conjoined with the word . . .kue-kus, which is shown by the ideograph comprised in it (a hand grasping a dagger) to signify ** powerful ". In J. ii. 8 — where the inscription ends with the same two titles as that of Mar'ash — the ideograph of the face is replaced by the phonetic characters e — sheep's head, the determinative affix of active opération (the hand holding a dagger) being attached to it. The affix would indicate that the word means ''hero" rather than "speaker", although the phonetic reading would appear to be e-ewe. In the preceding Une of the inscription the word
(1) On a bronze bowl, found at Toprak Kaleh near Lake Van, and now in the Muséum at Berlin, are four Hittite characters, the first of which is a bull's head followed by an obliterated symbol. Bronze shields, bearing the name of the Vannic king Rusas the son of Erimenas, bave been discovered along with the bowl, and it therefore appears probable that in the head of the buU we must recognise the first élément in the name of Eri-menas.
(2) As discovered by M. Menant.
e-ewe(l) intervenes between . . .kue-kus '* powerful " and the epithet which bas been shown to signify " great ". Cf. also A. 1.
To retum to the three sbort inscriptions of Hamatli. The oblique stroke which marks the end of a paragraph shows that the first part of the text ends in the second line with a Word which, I believe, dénotes " kings", the sensé being '' the king of kings". The plural is represented by two semi-circles placed back to back and the phonetic complément va (?). We then corne to a phrase which is written differently in the three inscriptions, with the exception of the first word in it, kus{'î)-ku, This last is the first person singular of a verb, and may signify " I am ". Next follows in H. i. 1. sar-u-tu-e-Kus-es '' the son of the sarutu *\ It is difficult not to see in this the borrowed Assyrian sarutu (sarrutu) " sovereignty ", more especially as in H. ni. 2. we find in place of it the ideograph of a throne foUowed by the phonetic complément es. In H. ii. 2, its place is taken by a curions ideograph. In H. m. 2. the word is foUowed by u-e-kus which I should explain as signifying " being" and connect with ue "and". In H. n. 2. itis foUowed by Arw5(?)-i?a(?)-e, which ought to be the third person of a verb ('* [whom] the king of the Hittites bas appointed ") ; in H. i. 2. by no corresponding word at ail. Finally there cornes "the king of the Hittites ", as I hâve suggested in Dr. Wright's book.
The whole inscription, therefore, will be : '' Says the prince, . . . er(?)-mes, the great, the son of Kustaspes (t), the king of the country of Er(?)e . . . , the . . . king of kings. I am the royal heir, being king of the Hittites (or, [whom] the king of the Hittites bas appointed); I bave made the shrine which . . .s (or, the son of Sanda-*-ues) the king commanded".
I bave only to add that the suffix me seens to dénote the accusative case (see J. III. 2). We hâve already seen, from the inscriptions of Ivriz, that the genitive terminated in a vowel, as in Vannic. A case which I cannot détermine was expressed by the suffix er.
The numerals up to " nine " were expressed by perpendicular lines. At Gurûn we find '* nine " so written; cp. also J. n. 2, 4, 5. An inscription, acopy of which was kindly sent to me by M. Menant shows that the ** tens " were denoted by a répétition of the character X, and the '* hundreds " by a répétition of the character ^ .
So far as I bave succeeded in determining the forms of Hittite grammar they seem to agrée in gênerai character with those of the Vannic inscriptions. The nominative of the noun is represented by the suffix -s, the genitive (and perhaps the dative) by the vowel of the stem. The accusative, when immediately preceding a case governed by it, (as in kue tuve '* the image of the prince '*), is without a suffix; at other times it is probable that it is characterised by the suffix me. The adjective agrées with the substantive in its terminations, and follows the substantive with which it agrées. The plural appears to be marked by the suffix which I read va, A common adjectival suffix is -ku8 (Vannic -kas), which expresses origin and relationship, Khattu-kus^ for instance, being ** Hittite", Sanda-kus " the son of Sanda".
The last example shows that the suffix also dénotes a patronymiCj and since in this case it is represented by kh in the Assyrian inscriptions, as in Khattu-khi for Khattu-kus, Tar-khu for Tar-kus, we may connect it with the Vannic patronymic suffix ^khi. As
in Vannic, moreover, the suffix îs connected with the root by the suffix -ue- or -w?e-.
The latter suffix forms the third person singular of the verb, as in e-we " he says ", and in the verb which in J. n. 1. is denoted by the ideograph of two persons joining hands foUowed by the suffix -we, There is nothing corresponding to this in Vannic. In the case of certain verbal stems the third person is represented by the suffix -va(1)e. By the attachement of the suffix -kus the verbal form may be changed into an adjective, e-we-kus " the speaker " or rather " speaking " being substituted for ewe.
The first person singular of the verb is characterised by the suffix -ku.
The copulative conjunction is expressed, sonietimes by the word uej sometimes by the character which I read va or wa. It corresponds to the Vannic ui '* along with ", u "and".
The sounds of e and i do not seem to hâve been very clearly distinguished. The cuneiform texts give us Er-wê by the side of Sapalul-wi and Scuida-sar-wi or Sandasar-we. I believe, further more, that mediae and tenues were confused together as in the language of Mitanni and the Kypriote syllabary. The fact that the guttural of the suffix -kus is represented by kh in the Assyrian transcripts of proper names shows that tenuis was confounded with the aspirated guttural as in the Kypriote syllabary. This increases the probability that the origin of the Kypriote syllabary is to be sought in the Hittite hieroglyphics. It also explains the comparatively small number of phonetic characters in the Hittite inscriptions and the fréquent occurence of certain of them.
In spite of our inability to read the proper names which occur in the texts, it is possible to draw some historical conclusions from the monuments we possess. First of ail, the name of the Jerablûs king, * - ♦ - we-tar-kus, is found at Gurûn (G. i. 2.) That the same king is meant i's made clear not only by the fact that the Gurûn prince is s^id to be king of the same country as the Jerablûs prince (G. i. 3. compared with J. i. 1), but there is the same référence both at Gurûn (G. i. 5.) and at Jerablûs (J. ii. 1.) to " the suprême nine ". The king therefore to whom the extant monuments of Jerablûs are due made his way as far north as Gurûn, and there engraved inscriptions on the rock to three of the gods he worshipped. The fact has an importance in its bearing on the date of the Jerablûs inscriptions.
I hâve already drawn attention to the fact that the royal names found at Hamath seem to be Komagenian. This excludes the supposition that the Hittite inscriptions of Hamath can be subséquent to the conquest of the country by Tiglath-pileser m.
I hâve further pointed out that Eu(r)es who appears as " prince " and " highpriest *' at Ivriz is king of the Que in the inscription of Tyana and probably also in that of Andaval. His name also occurs at Bulgar Maden, just as the name of his subordinate, Sanda-sar-me-kus, occurs at Tyana. We must infer, therefore, that the monuments of Ivrlz, Tyana and Bulgar Maden were ail executed by the same king and about the same time.
Now there is an cTctraordinary similarity between the inscriptions of Jerablûs, Bulgar Maden, Tyana, Ivrlz, Mar'ash, and Hamath. The forms of the characters are practically identical, the same groups of words are found in similar positions, the same
suffixes are used and the same titles are employed. At Bulgar Maden and Jerablûs we find mention of the same deity whose name is expressed by two ideographs, the second of which is a bird. As the title of tarkus " kîng " is attached to the name of this deity in J. III. 5. I am inclined to identify him with the god " U-va tar-kus ", " Uva the king ", found at Hamath (H. iv. 1, v. 2, 3.) With Uva or Uwa must be compared Uas in the name of the Tibarenian king Uas-surwi (? " Uas is king ") in the time of Tiglathpileser m. The name of Sanda, again, is found at Hamath (H. v. 3.) as well as in Asia Miner.
It is therefore évident that not only were the same system of writing and the same language used throughout the tract of country which extended from Gurûn and Mar'ash in the north to Hamath in the south and frrm Jerablûs in the east to Bulgar Maden and Ivrlz in the west, but also that the inscriptions we possess in this writing and language belong to the same era. The only différences I can find between the inscriptions is that in those of Hamath there seems to be a more fréquent employment of phonetic characters than elsewhere and that the forms of some of the characters in the inscription of Mar'ash are slightly peculiar.
The only period when such monuments could hâve been executed is one in which the Hittite race and language dominated from theTaurus mountains to Hamath. During the Assyrian epoch, so far as we know, Hamath was under Semitic rule, and consequently, though certain reasons still make me reluctant to abandon my old view of the comparatively late date of the existing Hittite monuments, I do not see how we can avoid referring them to an âge when as yet the Phœnician alphabet was not used in Syria. Thîs takes us back at least as far as the âge of David et Solomon.
I hâve said nothing about the inscription of Ilgùn, because both in style and language it appears to dififer from the Hittite inscriptions further east. Some of the characters which occur in it are not found elsewhere, and I fail to discover the suffixes which are characteristic of the other Hittite texts. On the other hand the ideograph which I hâve conjectured to signify " Hittite " occurs in it (Une 3). It may be that in the inscription of Ilgùn we hâve a différent language from that of the other monumental memorials of the Hittite race.
A.-H. Sayce.
P. S. — A comparison of the three last characters of line 4 in J. ii. with the three last of line 2 makes it évident that in line 4 we must read " the 9 great gods ". Consequently we must assign the signification of " god " to the character which twice occurs in H. IV. 2. in combination with kuskus " temple " and kue " image ". It will further be noticed that the word found at the end of H. i. 2, (consisting of tar with two boots, back to back) is also found in J. ii. 4.
TEXTES RECUEILUS DANS QUELQUES COLLECTIONS PARTICUUÈRES
PAU
Georges Legrain
