Indo-European Origins of the Celtic Verb I. The Sigmatic Aorist

Citation preview

INDO-EUROPEAN ORIGINS OF

THE

CELTIC

VERB

L The Siematic Aorist

(--) S

.s l-

c s

\) A* s t\ F:\ I

r*{

\

S

*S F.i.

\)

\)

N sr

A

a

O F at1

a

L F.;

S)

ETH !t

^a

ng t> Za NZ trH

H d a a

z p

a Z 3r i l er'l

H

M E2

fr*

#r;fr Fi

5t o Z

lV.oa Cq a -< r-l -t ts\dO

H z\

\J

MN

p Ei a

rl

F

z >z =,. 'is replaced by (as a result, oI' a morphological or morphophonemic transforrnation)' a

.-. zero ending.

Sl.

sz.

THE

POSITION

IN

T}IE

s3. s4. TIID TIIE

YOCALISM SIGMATIC

any stop or fricat'ive

E :

any vowel (ltut, a.g,, €, -

R, :

anY resonant,

to whicli is adcled t[ :

any iil,ryngeal.

the vowel e)

SIGMATIC

ON' THE AOR,IST

I

AOR,IST

LANGUAGES

SICIMATIC AND

s5. TOCHAR,IAN FOR,MATIONS s6. IIITTITE FOR,MATIONS IN THE

SANSCB,IT

CONSTITUTION

TIIE

s8.

IN

AOR,IST

52 61 74

-S.

PR,ECATIYE OT' A

AND

TIIE

90

?AR,ADIGM

.S- ; TIIE ON' A MORPHEME

I 18

MIDDLE

-S-

INDO-DUI],O?EAN

DIFFUSION

T -

ON' THD

INDO-EUR,OPDAN

AND

YOICE

a/.

I follow Kurylowicz, A,popthon'ie,in describing IndoEuropean and Celtic root structures in terms of

REMARKS ON BECONSTR,UCTION HISTOR,ICAI LINGIIISTIC METI{OD

OII,IC+IN AND

97

PART TIIIO : Celtic 59. TEE srrsrnM ox.TrrD eRETERTTErN oLD

107

IR,ISII

S10. sKETcH x,oR, A x'uNcrroNAl OX' THE

Sll. Sl2. St3. Si4.

TrrE TrrE THE TEE

CELTIC

HrsroRy

110 124 145 156

YDR,B

s-suBJUNCTryE BB,rrroNrc suBJUNcrryE ,-pRETERrrr s-PRETERTTE

r74

APPENDIX R,EX'LEXES Sl5. MoRprror.ocrcar, GEAL

IN

Sl6. MoRprroLocrcAr, R,IX'Ll]XES GEAL

IN

CELTIC

INDEX OF WOR,DS

OF

LARYN.

18r

TIIE NOUN

CELTIC:

:

OT' I,AR,YN-

185

TEI] VEBB

193

Prolegomena S 1.

Rnum,rs

oN RECoNsrRUcrroN AND ErsroBrcAr. I,INGIIISTIO

METEOD.

Trn etymological groundwork in the Celtic languages has already been done; the laws of phonological correspondences between Indo-European and the several Celtic languages can be stated. with a high degree of precision, indeed a fair measure of historical certainty. Yet a brief inspection of the relevant portions of the classical works of Thurneysen, Pedersen, or Lewis-Pedersen is sufficient to convince one that the historical morphology of the Celtic languages remains strikingly obscure ; with the exception of the nominal declension, the inflexional categories have never been systematically accounted for in comparative Celtic grammar. Nowhere is this more apparent than in the verb. This formal obscurity, coupled with certain peculiar syntactical traits ofthe Celtic verb, have led certain scholars; such as J. Pokorny, E. Lewy, II. Hartmann, and H. Wagner, to a variety of hypotheses of the presence of sundry ill-defined non-fndo-European substrata in Celtic, particularly on the shadowy elfin-populated Emerald Isle. Without sure knowledge of the presence of such substrate populations, and without any notion of the nature of the languages they might have spoken, such a line of speculation is otiose : it is merely a deplacement of the problem, a substitution of one unknown for another. If, however, by the utilization of more recent techniques of linguistic analysis, we ca,n account for the peculiar development of the Celtic verbal system, as a direct and unmediated successor of the Indo-European verbal system, then the necessity for recourse to such hypothetical substrata simply disappears. It will be the primary objective of this study to show, by the I b

2

PrioLEctoMENA

of the techniques of functional analysis and' utilization internal reconstruction as well as the comparative method, disparate categories how three formally and functionally the s-subjunctive, verb, in the Old Irish and Early Welsh a single unitary from t-preterite, and s-preterite, are derived 'cornerstone' Celtic verbal the Common of formation, a is a direct Celtic in Common system, and a formation whioh Indo' late of aorist sigmatic lineal descendant of the European. Historical and comparative linguistics has had in the past a tendency to concentrate its attention on linguistic o"igittt, and. not on linguistic development. But the Irieh preterite is an excellent example of the principle that the forms of a linguistic system at a given time can frequently be explained historically only by concentrating attention on the process of development, and not simply by framing a hypothesis about the origin alone. Such a h54iothesis, tho Indo-European original, has been tried in the Celtic verb; but the technique of equating a reconstructed IndoEuropean form with an attested Celtic form simply does not work in a large percentage of the cases. Isolated comparisons between two languages are treacherous in historical i;nguistics. Even where they are correct, if the structural contexts of the forms compared are not themselves comparable, we lrnow in essence no more about the history of the languages than before, namely, that they are cognate. The situation is that of a root etymology ; in historical linguistics it is of value only as a starting point. Indo-Europsa,n linguistics in the nineteenth and much the twentieth centuries, while it made fantastic inroads of unknown, did so in essence by continuous refinethe into ment of the hypothetical original, Proto-Indo-European itself. Building inevitably on the more secure elements of the intellectual edifice created try our predecessors' we aro in a position now to study the dynamic aspects of the process of linguistic development, and frame hypotheses about them. By this method we can elucidate aspects of the history of a given language which had previously remained obscure,

RXIMABKS

ON RXICONSTBUCTION

3

when seen only in the light of static comparison. It is to bo expected that such investigations cannot but alter our conception of the original itself. A problem in historical linguistics to'day presents two aspects: on the one hand the establishment, by the techniques of the comparative method, of an arbitrary initial stage of the phenomenon, and on tho other hand a reconstruction of the historical procosses wheroby this original system was more or less eliminated, transformed, or restructured in the form in which it may appear in the different systems of the historical languages. Either of those two aspects of the historical problem may predominate in a given analysis. But in a case such as the problem before us of the Celtic classical verb, where the late Indo-European prototype-tho without established be form of the sigmatic aorist-can it is tho of certainty, much difficulty and with a fair dogree of and reconstruction second aspect, the establishment the through workings and their dynamic historical processes, successive stages of the given language, which assumes primary importance. The problem may be said to be one of linguistic history, far more than of linguistic comparison ; methodologically, tho tools are frequently less those of the comparative method than those of internal reconstruct'ion. 'We are accustomed to consid.er language as a cohesive system, un systime od,tout se ti,ent, functioning in a temporal continuum. Elence historical d.escription and analysis must be founded on the notion of linguistic history as a sequence of successive linguistic systems. Such a view permits us to come closer to an understanding of the totality of the language in its historical development, than does the centering of attention on individual forms alone, in ot'her Formal and words, atomistic etymological speculation. and develop, change, in language categories functional forms which individual the rate than a slower at disappear comprise them at any given moment. For this reason they have been recognized as a surer guide in the reconstruction of linguistic history than the manipulation of isolated Conversely, the inherent plausibility of lexical forms.

PBOLECIOMENA

individual comparisons or reconstructions-Meillet's criterion of an etymology, and still as close as linguistics can usually come to proof or verification-is considerably strengthened if it, can be shown that an individual reconstruction either enters a previously determined formal class, or constitutes with other reconstructions a new and distinct formal class. In this account of the Indo-European origins of certain aspects of the Qeltic verb, we shall consider as the focus of attention the grammatical categories themselves, in both formal and functional terms. The individual forms which comprise them are of secondary concern, except insofar as they fall into distinct formal groups according to their root, structure, with distinct morphological consequences. In practice, the first step in historical analysis is usually the framing of a hypothesis of the origin of the given category: in the particular problem with which this study deals, that hypothesis is the late Indo-Erropean sigmatic aorist. The second step is the verification of the hypothesis. Reconstructing back from the attested Irish forms, can we arrive at our hypothetical original? In the case of the Irish verb, the answer is quite simply no ; and this is the impasse at which traditional Celtic comparative grammar was forced to remain. But reconstructing back to proximate forms is not the only technique of reconstruetion, nor of historical linguistic analysis. In this case, as in many others in historical grammar, the solution lies not in simple lineal reconstruction, but in recovering the workings of the prehistoric morphological transformations which have reshaped the formal and functional alignment of the category in question. Even this aspect, the attontion to d;rnamic historical process, does not exhaust the possible techniques of reconLineal reconstructions of linguistic prototypes struction. tend to run backwards in time ; the symbol < is more Insofar as we regard our asterisked forms common than ). only as symbols of phonological equations among attested froms, it cannot be otherwise. But Indo-Europeanists, like

REMARKS

ON BECONSTRUCTION

phonemicists, may fall either into the 'hocus-pocus ' c&mp, ' or into the God's truth' camp. Let us make the larger assumption, namely, that our asterisked forms try to correspond, on whatever level and however approximately, Then an asterisked with a preexisting linguistic reality. form, or a sequence of asterisked forms, becomes not an isolated symbol, but an entire model of a temporal sequenee of systematic linguistic transformations leading to an attested form. Once this is admitted, then it is apparent that the simple exercise of reconstructing forwards in time may proye a fruitful and productive line of speculatioh. fn reconstruction we must deal both with forms and with functions. To reconstruct forms alone, without attention to their functional position, is first and foremost to create a hopelessly unrealistic linguistic situation. It is dangerous as well; for the functional analysis of two reconstructed forms may reveal them as incompatible in a given reconstrueted linguistic system at a given time. We shall have occasion in this work to examine how the functional position of a particular form in a givon structure can have a profound influence on the subsequent development of that structure, particularly in the case of the third person singular of a finite verbal paradigm ; conversely, there are frequent, cases where the adventitious identifi.cation of a phonological feature of one form with the morphological mark of another class results in a realignment of the function of the first form. An example in Celtic would be the occasional shift of d-subjunctive (of e/o-present indicative) to weak denominative d-present, with loss of the subjunctive function, and significantly the simple disappearance of the original e/o-present and the entire finite paradigm dependent upon it. C. S. Stang has emphasized, in the first chapter of Das slaai,sche wd, balti,sche Yerbum (Oslo, 1942), that in the stufies of each separate language group within fndoEuropean one must seek to ascertain what existing or preexisting functional oppositions may legitimately be set up in that language. His remarks have the great merit of being the first to insist that divergent formal characteristics

?ROI,EGOMENA

imply the previous existence in that language of definite functional oppositions which may have been in the meantime given up. We may illustrate this with a structurally significant feature of Old Irish (vestigially preserved in Early Welsh) : the distinction between absolute and conjunct endings. Their distribution is automatic : when preceded by a preverbal particle, finite verbs have the conjunct endings; when not preceded by such a particle, they have absolute endings. The origin of the forms, seen clearly in the third person, the basic member of the paradigm, is 'hs the Indo-European primary/secondary endings : berid, 'he does not bear' < *bhereti bears' 1 *bher-eti,n[,-beir 'they bear ' *bher-ont'i, n|-berat ' they do not berai,t < bear' < *bher-ont. The consequencesof this situation for historical purposes are manifold.. In the first place, since the conjunct form of the Old Irish present with historically secondary endings conesponds to the Indo-European imperfect, it is not legitimate to attempt to explain the Old Irish imperfect by the Indo-European imperfect, as some have done, e.g., Thurneysen, OIGr.372. As we shall see in more detail in $ 10, the OId Irish imperfect represents the creation of a new opposition of secondary endings in the functional place of the secondary endings in classical Indo-European. Of greater structural significance, howevet, is tho immediate conclusion to be drawn from these present forms : that the Indo-European opposition of primary/ secondary endings was not, utilized as a tense marker in Celtic. Hence when we observe the exact same distribution of absolute/conjunct flexion in the preterite as well as the present and the future, we are not justified in expecting the Celtic imperfect or the Celtic preterite to have ever been charactetized by Indo-European secondary endings alone. The formal mark of the preterite must be sought elsewhere. It is this sort of aprioristic comparative reasoning which succeeded in masking even to as sensitive an observer as Thurneysen the fundamental fact that the desinences of the present indicative, subjunctive, s-preterite, and f-preterite (except in the plural) are identical ; the variation is

REMARK,S ON BECONSTB,UOTION

in the morph preceding the d-esinence. In some cases the morph partially fused with the desinenco, e'g', 3 sg' -s-til-s-t > -ssi/'ss ; in this case there was constit'uted a zero ending, i.e., absolute 'ss-Ti/conjunct 's's'o, with very significant formal repercussions on the rest of the paradigm' But this cannot mask the original identity' The present work has two parts ; t'he first is concerned with th; sigmatic aorist in Indo-European, and the second with the reflexes of this formation in Celtic' Both parts ' take as their formal point of d'eparture the classical' recon' s-aorist of late Indo-European, as traditionally full particular: in Indo-Iranian of structed on the basis -sathematio plus suffix plus root or lengthened grade secondary endings. Part One will be concerned with the origins, in anterior stages of Ind.o-European itself, of the sigmatic aorist in its o classical' form, and the morphological transformations which brought about the creation of this category in ons d.ialect area of Ind,o-European. Both the initial stages of this historical process and the end result, the s-aorist as traditionally known, lie within Indo'European itself, and can be reached' only by the techniques of reconst'ruction, For this reason it is inevitable not by direct attestation' hypothetical and conjectural more the be rvill that Part One in nature, In Part Two the relations are otherwise' The point of d.eparture is the classical sigmatic aorist', the only formal difrerence being the admission of both primary and secondaryendings, as discussed above. But in this case instead of working backward.s in time, to earlier linguistic st'agos, we in time, reconstructing-forwards shall be working-and forms attested palpable the to Celtic Common through dorvn in vestigially (and Irish In Old and Welsh. Irish Early of Early Welsh) we have the previously mentioned three unrelated verbal categories : s-subjunctive, f-preterite, and Our objective here is to demonstrat'e, &s s-preterite. rigorously, exactly, and unequivocally as is possible, rtorilr these three verbal categories d'o, in fact, go back to a sigmatio

PR,OLEGOMENA

aorist formation. fn particular thereby we will be concerned with elucidating the historical linguistic processes whereby the familiar, ' classical' elements of the Indo-European verb 'ivere largely transformed, restructured, and redistributed to form the basic framework of the new, emergent Celtic verbal system. The two appendices treat certain characteristic reflexes of laryngeal in Celtic, which I term morphological, since their historical development cannot be accounted for by lineal phonological reconstruction alone. The first of these is concerned with the development and expansion of certain suffix forms, which are identical in type with Sanskrit set forms, and admit of the same explanation as these. The second appendix deals with the historical analysis of a small group of Old Irish primary o-verbs, which reflect Indo-European athematic laryngeal presents. 'Ihe nature of the particular historical problem, and the solution there presented, may be viewed as a microcosm of the technique and methodology of historical linguistic analysis which I am here advocating. If this solution is accepted, then the efficacy of that technique must, be considered proved.

PAR,T ONE

Indo-European Le terme indo-europ6en ne se rapporte pas d, un moment, mais d, touto une 6poquo enferm6e entre les archaismes linguistiques non-analysables et, les dernidres innovations communes de la p6riode pr6historique. II faut, se contenter d'un classement 6lastiquo en renongantr A d6limiter les cases d'rino manidre trop s6vdre. I(urylorvicz

S 2.

Tnn PosrrroN ox, THE STGMATTc Aonrsr rN TrrE INDo-EuRopEAN LlNGuacns

IN e masterJy article published as early as 1908, A. Meillet demonstrated, critically and exactly, the recent character of the s-aorist, and its very limited role as the mark of a morphological type in fndo-European.l Meillet began by calling al,tention to the purely formal peculiarities which set the s-aorist apart among Indo-European morphs : first, that the suffix shows no vocalic alternations, appearing only as -s- ; second, that it violates the stricture noted by n'. de Saussure,z thal vocalic alternations in inflexion occur only in the predesinential element (including unsuffixed roots), never in the presuffixal element, whereas the Iengthened grade of the active s-aorist (on which see below) occurs 'across ' the morpherne -s- ; and third, that the dialect distribution and extent of the s-aorist varies extraordinarily, and in a fashion totally unrelated to the 1 'Sur I'aoriste sigrnatique', Mdlanges d,eSaussure,8l-106 (paris, f908). One of the rare reviews of this volume referrecl to Meillet,'s article as 'uno dei suoi articoli smaglianti per rigor lucido di qrgomeniazione e felice novitd, d'intuizioie' (L. Vatmalgi, Bollettdno di, Ei.lol,ogi,aClassi,ca, 17.27 l19l0l). 2 Md,moire sur le systd,mepri,mi,tif des ooyelles d,ans les lctngues i,nd,oeuropdennes, 186 (Leipzig, 1879).

l0

rNDo-EURoPEAN

distribution of any other comparable linguistic feature. Meillet then proceeded to the examination of the varied manifestations of the s-aorist in the principal Indo-European dialects, and was able to show that in every one of them, and. the s-forms achieved a level of high productivity extension only well within the period of the individual dialects themselves ; perhaps the most striking case is the near-universal generalization of the s-aorist in Slavic, whereas the intimately related Baltic group shows no trace of it v'hatsoever. In some of these languages the extension of -s- is largely a finished process by the time of our earliest in others, characteristically extended tecords, as in Latin; enough among the oldest attested, such as Greek and Indic, we can observe the gradual extension of -s- as mark of the aorist during the historical period itself. The level of productivity of the s-aorist and its fundamentally intra' dialectal character is most, clearly mirrored in the fact that in Slavic, Greek, and Celtic, in completely independent fashion, it was this morph to which the language had recourse, to form the aorist (or preterite) of derived secondary verbs : OCS -ard,, -6rd', -ind'; Gk' -4ocr,'o)o(1,-euoa, -docL, -Loc-; OIr. (3sg. ahs.) -ai,s, -is ; MW -ris, -e8, -'is, -wys. In these languages, the sigmatic aorist was an expanding, open, ancl productive categorY. 'l'aoriste Meillet's conclusion may be reproduced here : sigmatique est en principe, au point de vue indo-europ6en, une formation secondaire tir6e de thdmes existants dans la langue, et non une formation primaire, imm6diatement rat'tach6e d, Ia racine, et, en second lieu, '.. cet aoriste existait A, peine en indo-europ6en commun et . . . il resulte presqu'entidrement de d6veloppements propres A, chacun des dialectes indo-europ6ens et relativement r6cents.' Though there are minor points to be corrected in this work, Meillet's basic thesis must be considered valid, purely on the basis of the evidence available at the time he wrote ; as we shall see presently, the subsequent coming to light of Hittite and Tocharian has only strengthened his case. But ofreconstructible for an age unready for a'relativistic'view

POSITION

OX' TED A-AORIST

11

phenomena, or for an understanding of the continuity and dynamism of linguistic development, his paper was premature ; little notice was taken of it, aside from mere citation. It is strange to contrast the cogency of Meillet,'s reasoning and the lucidity of his presentation with the following pronouncement, written twenty years later : 'Dieser s-Aorist ist zwei,fellos eine sehr olte Bildung, denn er zeigt die Wirkung des idg. Akzentes, und er muss daher gebildet gewesen sein, ehe diese grosse einschneidende Umwiilzung im ldg. eingetreten ist,. And,erse'itsist er auch jung, da er zu einer jiingeren Schicht von Verben gehctrt.' 3 Meillet's arguments may be supplemented now from many sides, both in the light of new linguistic data, and new or better interpretations of data existing at the time he wrote. He held there were two alternative 6largissements, *-sand *-fs-, both subsequently charged with morphological function.a In the case of -s- he was almost certainly right ; but to his doctrine of the relation of -s- to -is- one can raise serious objections. This -is- is the element identified by Marids in the Armenian aorist morphs -eac < x-is-d,-skeand pple. -eal, < tris-d,-lo-,5and appearing in various forms in the Latin perfectum : Eg-er-am,Eg-er-i,Eg-er-im,Eg-is-sem, Eg-is-se; eg-is-ti,, eg-is-t'i,s,d,g-er-unt.6 To these he added the Indic -i,s- aorist. It has now' been shown by H. Karstien 7 that the -e'made', has nothing of the Armenian forms, e.9., gorcedc 3 H. Hirt,, Ind,ogerma.nische Gru,mmat'ilc 4.245 (Heidelberg, 1928). I{irt's earlier objections to Meillot's article, The italics are mino. Hand,buch d,. gri,ech. Laut u. Formenl,ehrez 556 (Heidelborg, L9l2), cited with approbation by Schwyzer, Gri,echische Gra,mmatilc L.749, are not, supported by any actual contrary argumentation. a Similar statement's may be found elsewhero in Meiliet's work: 214; fntroductions Traitd, de gramma'ire comparde des langues The last, and fullest, oxposit'ion of his class,iquesz 265-6, 349-50. 'Sur le type latin doctrino is the article egr,, egi,stu', BSL 34. L27-30 (1933). 6 Rets. des 4tudes arrn. 10.167-82 (f 930) ; Meillet,, Esq.r,i,sse d,'une grammaire comparde de I'arminien cktssique2 115-6 (Vienna, 1936). 6 Forrns such as 2sg. -stt, fot -si,stt are not ancient, but result from haplology: cf. A. Ernout,, Morphologie h,,istori,quedu latin 2 336 (repr. Paris, 1945). ?'Das slavische Imperfekt und der armenische -ace- .\orist", (Wiesbaden, 1956). Eestschri,JtJiir Mar Vastner,2ll-29

72

INDO-EUR,OPEAN

to do with an *-is-, but comes rather from *-eg(e)- : an explanation much to be preferred on morphological grounds. We may thus discount the Armenian forrns. Since, as will be shown in more detail below, the Indic -ip- aorist likewise has nothing to do with this element *-is-, we are left with the Latin forms alone as evidence for a morph *-'is-. Now the function and distribution of the two morphs -s- and -ds- is in no way parallel. Aoristic -s- functions, in all the languages whore it appears, as a tense or aspect matker, suffixed to a basic stem or root, to which endings are directly added. But -is-, in contradistinction, either appears further characterized by an additional mark (Lat. *-is-a-, *-is-e-, *-is-i-, *-as-se--),suffixed to an already aspectually determined form, as a mark of modal function, or else appears as a desinential component (Lal. -is-ti,, -is-tis, *-is-ont) with partially predictable distribution (before -l-).8 This element -is- is widespread in the Latin perfectum, but in itself it is not a mark of the function of the perfectum ; it is always suffixed to an already characterized base. The mark of the perfectum is in Zg- beside og-, exactly as in d,i,r- (ch,c-s-)beside dic-. In view of these functional divergenciesbetween s-aorists and the reflections of -fs- it is difficult to imagine a direct connection or alternation -s- : -is- in any but the remotest past. The simple fact of the widely differing distribution of the two among the Indo-European dialects is also against it. For rvhile -s- as aorist is found in Indo-Iranian, Greek, Slavic, Latin, Celtic, Illyrian (e.g., Messap. hipad,ese)and Albanian,l0 -fs- may be plausibly identified onlv in Latin 8 As insisted on bv Meillet,. I'or the relation of this phenomenon io the Hittite and Tocharian evidence, see further belo*. e There are seven attested sigmatic aorists in Messapic, though not a,ll are sure. Thev aro conveniently listed in J, Whatmough's Messapic grammar, Pr6,e-Itatic Di,alects i.009 lOambridge, 1933)." 10 I cite Altranian and flyrian separatell', since I think it has not yet been adequately demonstrated that the formor is a lator dialect of the latter, E. Ilamp's remarks, in 'Albanian and llessapic ', Stud,ies presented to Joslnrcr, Whatmough 73-89 ('s Gravenhage, lg57) while interesting, are finally inconclusive.

POSITION

OF TIIE

A.AOII,IST

13

and Armenian, with the latter not showing simple -s_. fn a summary article published only three years before his death, Meillet returned once more to the problem.u lle concluded quite rightly that the element _is- is not a characteristic of the aorist, but stilt maintained the principle of essential identity of -s- and -is- as fndo_Eu"op*o '6largissements '. The fact remains that in the historical languages -s- is an aorist mark whereas -os- is not. On purely formal grounds alone Meillet's analysis is hard to maintain. An alternation -T- - _iT- is completely isolated in fndo-European morphology, and an enlargement with a syllabic center (-f-) runs counter to the very definition of enlargement. Were it a suffix we should expect a full grade alternanl *-yes- beside -os-, as in the structurally parallel comparative *-yeslis- or fhe optative "-Adli- < x-yeHliH-. But we do not have such a form. X.or the time being we can only regard -is- as one of the uncertainties of Indo-European grammar. That there was an Indo_European element -is-, of unclear function and limited extension, is certainly at least possible. But it must be separated from the morph -s-, with which it shares no single formal, functional, or distributional characteristic.lz The morph -is- most emphatically has nothing to do with the Indic -is- aorist. Meillet,s entire justification for taking the -fg- aorist as reflection of his -is_, and not, with the majority of linguists both before and after, as reflectins *-as- (*-Hs-), rested on the interpretation of a very few Gathic aorists showing apparent -ii- : rinaui,id,, baui,{i,, taai,itd,. In Iranian an internal a from laryngeal (shwa) is lost, hence -7.i- if found must go back to *-ii_. But in these Gathic forms, already suspect by their feeble attes_ tation and the graphic ambiguity inherent in this text. ll'1""

Je type latin Egr,Egistr,,, BSL 84.127_80 (fgBB).

.iz One. possible avenue oF comparison for *_is_. though perhaos only tylcological, is the Oseo-Umbrian fulure perfect, *ith i;;li;.i;;;;i;i component e.9., Umbr. 2 sg. benis , u,neris' Z ,E. biiuii -.uenent', ^-us-_, 3^_pI. benur.ent , u?,nerint,. The parallel L;1.*-,is_ont QU a-us-ert (Umbr. _urent, Ose. _ize(n)t) is parbicularlv l-erunt) striking._ In this easo we woulcl have io deal .i-]rfj,'*ilfr'-"-i;;;iii, obscure Italic innovation.

IN!STITUTT IINGVISTISK t j i . , i . i . r, , l i . r . , i ' t , ( } 5 ! , ( }

14

rNDo-EUR,oPEAN

the apparent -d.i- has been shown by Kurylowicz and more recently H. Ilumbach to be merely an orthographic pecuharity'l3 They are to be read, in normalized Avestan, i?nuai,fid. taEln, bdi\td,, and the franian evidence for -i,iaorist simply disappears ; along with it, the only reason for connecting the Sanskrit -rip- aorist with Indo-European *-is-. The absence of an -i,i- aorist in Iranian is in complete accord with the general, if not absolutely regular, Ioss of internal i < *a in that language.la The Sanskrit -a.s- aorist functions exactly like the -s' aorist, as a,n aspect marker, and totally unlike the various functional manifestations of the morph -ris-' Its origin, as has long been recognized,ls lies in the suffixation of -s- to se! roots in full grade form TERH-. The resultanb TERH-s' appears in Sanskrit as Td,Ri-1-,tu e.g', atd,ri,pamt'o tar', IE *terU-. Kurylowicz (loa' ci,t.), subscribing completely to this interpretation, has described in rigorous and elegant fashion the analogical mechanics of the isolation of -'isas a new morpheme, and its grad'ual ext'ension beyond its originally phonologically justified bounds, replacing'8-' BSL 44.6f -(f 948), 13 Kurvlowicz, 'Lo degr6 long en indo-iranion Caf'{tt' See also -hn ni.re"" a.or. rinao|u (or r{nafi{a, as well astaF#i, " ieopnorui" 163-4. Ilumbach. M&nch. Stud,..Spr.,2.7 : 22.14, and most 9.69-71 (1956), profers^to take ni.nauT'#,tas an opt".mid' 2 sg' ""'"uirtty thu form is not' an aorisf in -i'{-' In eithor isrrr*rlcr. "6"e, rn Se" H. M. Hoenigswald in Euitlemce for laryngea'Is 25 (Austin' r960). --iu Han[Ibuch d'es Banslcrit 381 (ITeidelborg' 1.905J ; M. e llnrr-t, L' Il'enou' I95ll; Mavrhofer, Hancl,buch des Palo I.158 (Ileidelberg' de la langue addi'qua 290 (Paris' 1952) ; H' Pedorson, Zas Grii*"1,r6 sis?nati,ques ili aerbe l,;ti,n LB (RDVB Hi'st. -f'1. Med'. 3-5 t192ll). liies h. Bn"to*, The Sanskrit Language 33.8-9 (London' 1955)' prosenls on hand an unclear and complicated amalgam of explanations, ifr"1tfr"" (4{^Se191{ty^ q1' ow1. very^personal by.his largely conditioned -' 191S'22-61' TPS " Schwa " in Sanskrit', a"Jup["d) views on has been neatly scorod -by-P' g*io*'" method of prosentation it i"*u, Lang. 31.428-*O 1tOsS). -EIis apparent basic premise -is th-at in the compound suffix is formed on nominal st'ems iriu -i"-'ao"iit 'liglft'. But this cornparison -;-"-. d,od. he adduces ctroeiqla t rocip' l."cs in cosencv beside the fact that' arociqla is a Bralrnana form ; tho Rie-VeJa his onlv (a\rocd (3 sg' mid), d'rt-trucot, and pple' r'ucand;" only tho Hist'orically, 16 B;t'h quantitios- are attest'ed in Vedic. short vowel'is proper ; tho long vowel is more.recont',^and^analogical See below, Apopkonie 162-3. in origin, a" sito#n by Kur;'lowicz, chapter two.

?OSITION

OX' TIIE

S.AOR,IST

15

The more characterized morpheme -rip- will tend to replace the simpler, but iso-functional morpheme -s-, in the terms of his first law of analogy.lT This process of the expansion of -zp- is continued even in Middle Indic : cf. Pd,li and Ardha-Mdgaclhi Prakrit lsg. pucahis(s)adt,'I asked' (Sky. -ipam), Pdli 2sg. gtucchittho (Skt. mid. -ip-!hd,h), 9sg. Ttucch'i,ttha(SkL. mid. -ip-fa), as well as lsg. apucchi,m, 2-3sg. apucchi. Compare also Peli agacchi ' he went ', amafr,fi,issarir' I thought ' (Skt. man-ya-), Ardha-Md,gadhi annesi, 'he knew' (Skt. *ajfrd,yi,gi,t),bhuui, t he was'.18 We have, however, a few very precious indications which show that the formation of the -ip- aorist, i.e., the suffrxation of -s- to root final -i- < -H-, dates from the Indic period. These are the occasional irregular, so-called 'asigmatic' -op-aorists which crop up in Vedic. The lsg. forms akromi,m, aaad,hi,m(both RV), agrabhi,m(T.S.), commonly cited in this connection (e.g., Whitney. Burrow), are indecisive, since they are much more likely to be analogical creations according to such a proportion as dni,cam : dsi,cas tisicot

ddd,m : dkrami,m (replacing dkram,i,pont) ddd,s d,lcrami,s dddt d,lcrami,t. The likelihood of -i,rn being analogical and relatively recent,re alread.y more than plausible by its being confined to later strata of the Vedic language, is heightened by the fact that in Pd,li the regular I sg. of the ' s-aorist ' is -sinx I -si,m, analogical to 2 and 3 sg. -si < -si,t : in other words, a repetition of the same analogical process. r7 'La natnre des procds dits " analogiques "'. Acta Ldng. 5. f5-37 (1945-9) : Esqudsses Li,ngudsti,quas 66-86 (Wroclaw-I{rak6w,

r960).

18 Ma;rrhofer, Handbuch des Pal,o 1 156-9 ; R. Pischel, Gramtnatilt ,ind,o-ari,schen Phi,Iologde u. (Grund,ri,ss d, d,. Prakri,t-Sprachen Altertumskwnd,e 1.8) 359-61 (Strasburg, Ig00). Tho clearost, purely 'Tho Middle fndic descriptir.'e analysis is givon by F. Edgerton, verbal system', Asi,ati,co (X'estschrift Weller), 78-81 (Leipzig, f954). IIe shows that, the repartition is simply 3 sg. -sd (< Skt. -sZr) to stems in long vowol, -d (< Skt. -Zr) to stoms in thomatic vowel -a-. le In fact, the position taken by Meillet, BSL 34,.128, a,nd Renou, Grammaire 29O.

16

rNDo-EUnoPEAN

But we have a few other occasional forms, where no analogistic explanation is possible : compare the hapax 2 pl. aor. inv. aaitdzo 'favor' (RV 7.59.6) beside aaigtdno, from the se! root ao-. Evidence for laryngeal character is furnished by the pple. d,ta- and derivative d,ti,-, both morphologically isolated forms where analogical lengthening is not possible. The most telling, howeyer, is the hapax I pl. atd,rima 'we overcame' (RV 8.13.21) < *terH-me, to the se! root far-, beside the more recent, regular sigmatic form atd,r'ipma(3 exx.) < *terH-s-me. X'or this form, aside from the detail of the desinence and augment, and the analogical Indic Dehnstufe, may be exactly equated with Hitt. torh-wen (tar-hu-en) 'we conquered', 3 sg. pres. tarh-zi, (tar-ah-zi), an athematic root of the shape TERH-. This is the most ancient, state of the present/aorist system : the simple addition of secondary endings to an athematic root, form.21 Historically, it is to a sub-type of root aorists (se! forms TERH-) that the -s- was suffixed, to form the Indic -ip- aorist. This situation is confirmed by the (methodologically ind.ependent testimony important) of a third IndoEuropean language. This is a single extraordinary archaic Celtic form, preserved by virtue of being in isolation as a component of a suppletive pal'adigm : Ofr. do-cer, 3sg. This form has been shown by pret. of d,o-tui,t ofalls'. (OIGr. 437) Thurneysen to be an old root aorist, *lceret (*kerH-t), etymologically and typologically comparable to 'he crushed', aorist Skt. (AV) (a)iari,t of {rnd,ti. Blot -cer is, in fact, a more archaic form than (a)iail't, which is an -as- aorist, i.e., an s-aorist to a set base.z2 Celtic *lcerat < *kerH-t shows an athematic and suffixless (root) aorist ,. Prdrydth" rrrtd, th" f-- i. "@ ". 290. $ 908,Burrow 339,Renou

21 Compare the remarks of George S. Lano, Long. 85.172 (1959). 22 tTnless one wished to take it as an analogical rofashionins of an older athematic and asigmaLic *aiarit; only ihe 3sg. is found. But such a form is without, parallel in Vedic, though it doubtless may *ker-t, -ceq'from To derive Ob. once have existed, a-3sg. with Ti, Wagner, ZCP 28.10-lL (f960), is of course impossible according to the phonology of OId Irish, and in a,ny case ignores the laryngeal (set,) character of the root, xlcerH-.

POSITION

OX' TI{E

S-AOR,IST

L7

formation, and may itself tre directly equated, for the t;rye, with such Hittite preterites 3sg. as tarh-t, #anh-t, iarh-t, walh-t, parh-t, nxarh-t,all laryngeal (set) bases T E nH -. Meillet's basic thesis on the nature of the s-aorist is in no way invalidated by his unacceptable view on the connection with the Sanskrit -dp- aorist; rather it is strengthened. X'or the expansion of the -cp- aorist even down into the Middle Indic period represents part of the same development which Meillet observed so clearly else. where : the great extension of -s- only after the separation of the dialects. That in Sanskrit the form -ig- gradually supplanted the simpler -s- is merely a historical accident, the general spread of se! at the expense of ani! forms; as we have seen above, the basic nature of the suffix in form and distribution remained the same. This view of the late spread of the s-aorist is further strengthened and confirmed by more recent work in comparative linguistics, subsequentto the appearanceof Meillet's study, and not dependent upon it. Thus the peculiar forms making up the Sanskrit seventh (-sa-) aorist, of the type amrkpat I mrj-, have been shown by Kurylowiczzs to be by their origin not thematized s-aorists,2 but rather 'sigmatized' thematic aorists. The relative chronology of the two formations is clear : the thematic aorist is the older, and in this one small formal subgroup (roots in final .{, j, h, and p) the thematic aorist has been partially ousted by the encroachment of the newer and productive-'dominant' in Kurylowicz's sense -morph in -s-. One may point finally to the nature of the characteristic accessory formal mark of the s-aorist : the apparent lengthened grade d, of the root syllable generally assumed in the standard handbooks of Indo-European. It is to the detailed investigation of this phenomenon in the various Indo-European dialects that the second chapter is devoted. 23Inrloirani'crl,Comptesrend'usd'eIaSociitdd',uS"7,""""td,"Liln_,* d,e Wroclau: 3.1-6 (f948). Cf. also Apolthonie 164 n. 21. z+ The difficulty with this superficial explanation was already seon by Wlritney, Sanskrit Grammar $ 919a.

l8

vocAr,tsm on' TEE s-AoRrsr

INDO-DUR,OPEAN

yooAr,rsM ox' TrrE srcMATrc AoRrsr $ 3. Tnn Tnn original study of Meillet on the sigmatic aorist was the first to call attention to the fact that the lengthened grade generally assumed to be characteristic of the active s-aorist violated. one of the strictures noted by X'' de apophony' paradigmatic Saussure for Indo-European alternations According to de Saussure's formulation, vocalic in inflexion occur only in the predesinential element, nover in the presuffixal element. Thus in a paradigm of the structure R + D,t R may exhibit apophony : Skt' as-fd : s-anti,, gtd't : pad'-am. In a paradigm of the structuro .B + S + D, B may exhibit apophony, but not R : Oltl :6cl,-zop-os. But' in the ; s-l'-tis, Gk' 6cl,-zc,rp Lat. s-'i,d-s sigmatic aorist, of the structure -E f S + D, the apophonic variation is found in -R : Skt. aud,lt-p-am: aalc'g-af' This anomaly is unique in Indo'European morphology. If the lengthened grade in the s-aorist is inherited from Indo' European, then it must stand as an aberrant formation' We have seen in the preceding chapter that Meillet's thesis on the lateness of the sigmatic aorist as a category in Indo-European is entirely valid, and' that linguistic studies subsequent to his have universally confirmed Meillet's conclusion. The traditional assumption of Ind"o-European antiquity for the lengthened grade as an accessory formal mark of the sigmatic aorist could itself be used' as an additional indication of the lateness of t'his category in fndo-European. It is abundantly clear from Kurylowicz's Agtophon'i'e that, the lengthened grade is the latest apophonic development in Indo-European, and' as a derivational process in the parent language one of recognizedly limited extent. As I have shown in a review of Kurylowicz's work,z the great florescence of vocalic length as a morphological mark comes about in large measure independently, and from a variety of sources, only in the historical dialect's themselves. It is lsyrnbols: R,: root,r S: 2 Lang. 34.38r-98 (1958).

suffix, D :

dosinonce'

l9

characteristic that this chronologically later formal mark would be associated with a chronologically later formal category such as the s- aorist, just as it is the distinguishing mark of the late and dialectically conditioned. phenomenon of qddhi. In this review it was furthermore questioned whether we were, in fact, justified, with Kurylowicz and others, in assuming that the s-aorist had lengthened grade at all in Indo-European itself; in other words, that the lengthened grade characteristic of Indo-franian, Slavic, and Latin might be a parallel but independent creation in these three dialects alone. Kurylowicz's justification can be summarized. in the commonly given threefold equation Skt. aud,kgam: LaI'. udni,: OCS ods,rl. But the fact remains that both Celtic (in the s-subjunctive and f-preterite) and Greek show no evidence for lengthened grade: eyen accepting the traditional view, it is necessary to assume that the development of a lengthened grade d associated with the s-aorist took place only dialectally within the fndo-European group, and was never universal. Likewise in the context of the traditional view of fndoEuropean lengthened grade in the s-aorist, it should not have escaped the attention of scholars that the three groups which show evidence for lengthened. grade in historical sigmatic forms, fndo-franian, Latin and Slavic, are dialeets which have gone farthest in exploiting vocalic length as a derivational mark elsewhere in their systems, both in noun and verb formation. One need only witness in fndoIranian vlddhi and the secondary paradigmatic apophony d, : a, the numerous long vowel categories of Balto-Slavic nouns and verbs, or the Latin long vowel perfectum forms. Kurylowicz's Apophonie has little to say about lengthened grade in Italic, but there is a great deal in this aspect of Latin to be investigated. f may point out here two tSrpes which seem clearly to reflect a purely Latin lengthened.grade, in both nominal and verbal derivation. The first consists of simple nouns with nominative singular in -ds, and

20

INDO.NUROPEAN

compound nouns with nominative -ds, or the suffix which may be compared with short vowel forms: sed-e8 n1,oL-ea LA,O-e8

ril,yt-ds amb-dg-d,s com-pd,g-Es

ilE-ses,sed,-Are mol-estus labd (cf. ld,bor) runxpo ag6 pctrlgo

con-td,g-Es

tang6, tagat

ad,-d,g-'i,um

a'i6

con-td,g-i'um

tangd

suJ-Jrd,g-i,um

lrang6

The second group comprises the three following verbs, which may in similar fashion be related to short-vowel forms: sop0o

sopor

sd,gi'6

sagdr

mdlior

molestusg

Both of these types, the verbal as well as the nominal, represent derivatives built on an earlier root noun with short vowel ; in the first, group we have extended nominal forms (replacing the earlier root noun, as clearly in sEdd,s), and in the second group, denominative verbs built on the root noun. The root noun serving as base is preserved in a few cases, particularly in composition : d,d-ses,r4m-er, Gk, d-zo{ show the root nouns *sed-, *ag-, *1tag- which aro the base of the long vowel derivatives sdil,-es,amb-dg-d,s, aonx-pd,g-As.One may extend tho reasoning, and postulate

3 Benveniste has likewise suggested orally (Ecole Pratique des 1958) that, Latin uud'o has tho sa,rno Etudes, February, Ilautes verb with to uad'un't; it is a denominative apophonic relation lengthening of the vowel of the nominal base form.

YOCALISM

OX' TEE

S-AORIST

2L

similar root nouns to account for the remaininE forms of the lists above: *l,ab-

-> -) *ag- -) *frog' -> xso?-6 -) *sag-4 -) -) *mol*td,g_4

ld,bEs -td,gEs,-tdg'ium -dgi,um -frd,gi,um s6pcre sagi,re m.onrr, |noLeg.

Ilaving established the existence of this derivational pattern in prehistoric Latin, it becomes possible to account for other Latin nouns of this structure by the assumption of transformed root nouns : cldd,Es I *lclad-, morphological zevo * l c U , - : - c e l l 6< * k e l d , - ; r d d ' - i ' r1 * w r o i l - , grado in a: morphological zeto grade in a : *wfd'-. Compare Welsh 'root' gwrai'd'd,oroots ' < *wraitr-i-, and Got'h. wourts < *wrd'-i,-. Strdgds and nd,b\s probably ad.mit of the samo origin, as does ud,tEs < *wat- (wilh a-vocalism) beside ' ' Skt. api,-aatoti, blows ; inspires', W. ilyweil,ud, speak' (*wet-); but these demand special investigation. The citation of these Latin forms may suffice here to indicate the important and hitherto unrecognized role of vocalic lengthening as a,n accessory derivational mark in Latin. It is clear that we have here to deal with a purely Latin morphophonemic process, and not one inherited from Indo-European. The sigmatic aorist of Common Celtic is reflected in three categories in the Old Irish verb: the s-subjunctive, f-preterite, and s-preterite. These three, with a small number of exceptions which can be accounted. for, are in strict complementary distribution accord-ing to the final element of the root : roots in final -T form an s-subjunctive, roots final vowels, in final -.8 form a f-preterite, and-'roots'in a These root nouns r:nderlio tdgaL*, sagAfr; compare capAt I toot' f,Lou;r &w-caps, *factr,r (fficd,n) : root norn ponti,-fer, and seo also the remarks ur Eaiil,emce Jor La,ryngeals 188-9. 5 Simply oxtended by-es- in sopor.

22

LNDO-EUR,OPEAN

i.e., the secondary, derived verbs, form an s-preterite. fn part II of this work may be found a detailed investigation of these forms, and the demonstration of this thesis. We may assume for our purposes here the existence in Celtic of sigmatic forms of the structure TET-I-, TERT-I-, TER-s-; the roots in final vowel need not concern us, since they represent secondary and not primary derivatives. All of these types in Old Irish show a short, not a long root vowel, where that vowel reflects fndo-Europeart e i * sed,-s-, Old Irish subjunctive stem sess- ; * melg-s-, Old Irish subjunctive stem m,ell- ; * ber-s-, Old Irish f-preterito stem bertroster of forms attested, see the relevant chapters in part II infra. On methodological grounds one must exclude tho types TERT-I-, TER-s-, since it is conceivable that in these forms the vowel could result from the shortening of a tautosyllabic long diphthong : dR > eR. But there is no conceivable process whereby an inherited *sdd-s- would be shortened to *sed,-s-,O. fr. sess-. We must recognize that the original structure is that mirrored in Lble TETroots with e-vocalism : *seil-s-. The root shows e-grade, nbt the lengthened E-grade characteristic of fndo-franian, and occurring in a small number of forms in Slavic and Latin. Compare Kurfowicz, Apophonie 160 n. 18 : 'Or malgr6 la d6cadence de l'apophonie en celtique on s'airergoit que le subjonctif en s ( : aoriste sigmatique A, d6sinences primaires) 6 y pr6sente l'ancien degr6 plein.' This fundamental fact of the existing data of the Celtic verb should be more fully emphasized in Indo-European studies than it generally is : the lengthened, grade found in fndo-European, Slavic, and Latin even in terms of the traditional interpretation must represent a dialectal phenomenon, never fully developed in late fndo-European. o This is inaccurate. Tho s-subjunctive is a sigmatic aorisi with both primary and secondary endings : 3 sg. abs. /ez'ss/conj, -fd < *wedh-s-tif *wedh-s-t. I};Le pattern is idontical with that of the present indicat'ive, savo for the presence of the thematic vowel in the latter : 3 sg. abs. fed,id,lconj. -Jeid, < *used,h-e-ti,l*wedh-e-t, Compare the remarks in t'he Prolegomena, $ I supra.

VOCAI,ISM

OF TIIE

S.AOBIST

23

Now we have in Greek a sigmatic aorist with simply full grade vowel, never lengthened grad"e. It is customary in the handbooks to assume quite simply that the lengthened grade was somehow lost in Greek : $4pical is the initial 'Le grec, sentence of $ 32 of Kurylowicz's Apophoni,e : langue conservatrice sous divers 6gards, a compldtement perdu le degr6 long caract6risant l'aoriste sigmatique non seulement en indo-iranien, mais aussi en latin et en baltoglave.' Compare Schwyzer, Gr. gr. 751, Meillet-Vendryes 206, Chantraine, Morphologie hi,storique d'u grec L94. To account for this peculiar total abolition of such a feature, different explanations have been offered,? principally the supposed analogy of the shortening of the tautosyllabio *e-d\ik-s-r.n would have long diphthongs : a primitive Z6eefa. Kurylowicz rightly saw yielded attested regularly there is simply no reason in this argument the weakness ; *ddi,k-swhich represents a d'ei,k-s-, why a shortening 2 quantity, should then vowel of neutralization phonological *d'Ek-svowel shortening ) d,ek-s' a morphological cause (6ef-, 6/xo1^r.oc). But Kurylowicz himself offers an explanation of the vowel-shortening which is considerably more com' plicated, and hardly more cogent: the loss ofthe lengthened grade is a morphological consequence of the loss of s between vowels. Kurylowicz's reasoning is as follows. The phonetic loss of intervocalic -s- in forms such as *tyeu-o-a ) *Zyeu-ag resulted in a new appreciation of the mark of the aorist : instead of being -o- it became -o- plus the accessory mark -o- 'dans des conditions speciales,i savoir aprds consonne' (*€yeu-a but *36er.rc-o-o)' Under these new conditions tho accessory mark -o- ceasesto condition the root by implying the lengthened grade, since tho appearance of -o- is itself conditioned by the root form (in final consonant,). Hence *laf- (Iex-) is replaced by,\ef- (Hom. )igo4' Subsequently the compound- morpheme -o- plus implication of -o- evinces ? Seo the bibliography in Schwyzer, I. c,, w}:.o himself favourg the analogy- of the subjunctive with short, root vowel. 8 The historically " TeTTd,T-s-. Yet no one appears to have considered the evidence in Latin against such a hypothesis. We have in Latin a number of sigmatic aorists to roots of the structure TeT-, which show clearly a short root vowel, where there is no possibility that the vowel has been shortened: sltero

clep6 gerd (1trem6)

clepsoLo gessi, press|.

Since the present, cedd is ftom *ke-sil-o-,formed on the root *sed,- 'go' ( : d8ds, Russ. rod'it'), the s-aorist with,short vowel cessdprobably represents an analogical innovation' But we can further compare the series of verbs with root enlargement -l-, which universally show a short vowel s-aorist:

fl.ectd nectO pecto plectd

fleni' neni, peui, plert.

We may finally add another verb to this list, where the original e-vowel is obscured by the phonological change of lc*e t'o ao-: cott . coqud This gives us a total of nine verbs in Latin which show evidence for fultr grade -e-, in the sigmatic aorist,, not for lengthened grade -e-. Roots of the structure TeRT-, TeR- (mersi,, contetnltst') have been exclud.ed, since the 10 Thero is no justification for the form clEpi't in Pacuvius Trag. In Cicero' 185, and the standard edition of Ribbeck adopt's cleltsit de Leg. 2.9.22 Yahlen's edition roads cleper'it rapsi'tEte, but Madvig in his 1839 edition of Cicero do X'in. (ad. 5.25.74) demonstrated convincingly that this passage is to bo rea'd clepsi't rapsi'tue, with t'he subjunctive-optativo suffix -si- (cf. Jari't) forrrd in Lir.1y 22.10.5 si, qu,is cl,eps'it. There are of course numerous attestations of tho sigrnatic perfect' clepsi'.

ON' TIIN

,S-AORIST

29

possibility exists that in these a tautosyllabic long diphthong could have been shortened. But in view of the unambiguous cases of aorists TeT-s-, it is far simpler to assume that the sigmatic aorist in Latin inherited. full grade a-vocalism, and that this is preserved as such in the forms: Q,I,-UCqO

iubed

d,0-U181.

O. Lat. iousit

fluo

.fiur|,

and further in the whole series of the type a,nco

41,tI,

d,uco

dl1,1:L

etc.,

where the vocalism is full grade in both present and s-aorist. The evidence of the TeT- roots above points clearly to inherited e-grade, and the simplest account of the TeRTroots points to the same conclusion ; the positing of long diphthongs here is a clear violation of the principle of Occam's razor. The forms in Latin which exhibit a sure lone vowel in the siqmatic aorist are four: regd

reui,

tegd

teri,

crano

trari,

uehd

uexd.

The long vowel is attested by the apex in inscriptions for the first three, and by the testimony of Priscian for the last.11 To these one may add, with Ernout, Morphol. 3I3, 317, intel-legd

i,ntel-ldri, (and other compounds)

tl Cf. X'. Hand,buch Sornmor, Forrnenlehre2-3 556-7 (repr. Ileidelborg, hi,stori,que ilw latin z 315 (Paris, l9tt5).

d,u la,teini,schen Laut----4rnd lgrt8); A. Ernout, Morphologde

30

INDO-EUROPEAN

even though there is no explicit evidence for the quantity. This yields a maximum of five examples.l2 Of these five examples, it is impossible t'hab trd,ri' continues an fndo-European form, by its vocalism in -a-. The root form *tragh- is confined to Italic and Celtic alone. fn Celtic we have only nominal derivatives: OIr. tra'i'g 'foot' q *trag-et-, Gaul. uer-tragos'greyhound' I *trogo-. The long vocalism of the sigmatic aorist assumes only IE e --> E. The lengthened vocalism of trd'ui, is an isolated analogical shift of purely Latin date, just like the hapax perfectum scd,bi,(scd'beraf,Lucilius) to scab6. It can have The same is no independent value for Indo-Europea,n. assumed by Ernout, loc. ai,t., and by Sommer, Hanclbuch 557. We have clear evidence of a verbal stem in original short a root vowel which was not lengthened in the compound (il,)-lici,o -leri-lectus, from *-laci6 *-l'a,ri' *-lactus. The simplex laci,t is attested in Paul. X'est. 116. A *-l'd'ti' *-ld'ctus would of course not have been weakened to -l,eni,-lectus in nonThe form (per)-cussi', even though the initial syllable. phonological treatment is somewhat more opaque, shows clearly a short vowel in the original sigmatic form *-quat-si ( : quati,6; no perfectum is found in the simplex. We are left with r\xi', udn|, td,ai, -ld,ni. Now we can observe in Latin the gradual spread of the s-aorist, and its replacement of various older forms. So we have simplex momord,6pelteroi,Pu?Wd, but in composition praemorsi,, compers-t,compunr|; the reduplicated perfect is replaced by a sigmatic form, since in composition the reduplicating syllable disappears, and the resultant form has no clear morphological mark. But the sigmatic form may also replace an earlier long vowel aorist : we have 12 X'. Solmsen, Btuiltien z.tn latei,nischen Lautgeschickte 3O, 3B 'squat' has (Strassburg, 1894) suggests t}rat conqueri' r'o conqwi'nd'sco a lons rool vowel, and ho is followed by Sornmor, Eand'bwch 554, as Lat. Gra.n't.s 333. Bttt, conquenl' is attested well is Leumann-Ilofmarrr, only once in the texts (Pomponius Atell. l7l), and the peculiar formaiio. and clearly familiar, aLnnost l'ulgar character of the vorb makes date. Since that, it continuos anything of Indo-European it ullikely the. Iong vowol is pure hypothesis, I prefor to discount, the word enorrery.

VOOALISM

ON' THE ,S-AOR,IST

3t

emd : Emi, but d,6md: d,Em1tsi,,and other forms in composition with preverb in final vowel. The same explanation accounts forintellEr| This form represents simply the suffi.xation of -s- in composition to the historically older perfectum form legi,,with a Latin long-vowel root perfectum. We have, in fact, the older i,ntellEgit still preserved as an archaism in Lucretius 6.17. This sigmatic form spread in Latin in this verb ; beside regular Elegi,we have epigraphic eleni,, and at a later date this -s- spreads also to the simplex : lenerit insc., and cf. Ital. lesse.l3 The form -lEri, thus has no relevance to Indo-European ; it is simply built, on l,d.gi,, with the original point of expansion in verbal composition. The same explanation accounts for rdm, as can be shown from the evidence of the texts themselves. The uncompounded perfectum form rEri, is extremely rare in Latin of the republican period; the infectum regd alono is found in common usage. Plautus, Terence, Ennius, Lucretius, Caesar, and Catullus have only forms of the infectum; in Cicero there are only two occurrences of r4ri,, as against 80 of reg6; the same is even true of Vergil, who has a single instance of rEri, beside some twenty of reg6. This would suggest that regd had only a durative sense in earlier Latin ; where it, was necessary to form a preterite ( : tense of past time) the imperfect regEbamalone was used. The uncompounded perfectum rEri, far from continuing an inherited fndo-European sigmatic aorist thus simply did not exist in early Latin; it is a neologism introduced only at the end of the republican period. In composition, however, this situation did not exist ; on the perfectivizing function of preverbs in Latin compare Meillet-Vendryes, Trai,td 301-3. We expect forms of the perfectum in compounds of reg6, and it is from the compounded forms that the simplex rEri, was subsequently creatod. We have seen that the point of generalization of the sigmatic fown -ld,ui,was in composition, where it ousted the older root form -ldgi,. The same situation recurs in the case 13 See Meillet-Vendryes,

Traiti,

272,

32

VOOAI,ISM

INDO-EUROPEAN

of -r\ri. On the testimony of X'estus, Livius And'ronicusthe perfectum surEgit fot our oldest L[tin author-used This form has been sumari,t (p. 380, 32 ed. Lindsay). curiously neglected or misinterpreted by Latin scholars : Ernout-Meillet, D'i,ct. s.v. regd, call it an analogical creat'ion, as does Sommer, Hand,buch 57L; Walde-Hofmann, Lat' etynt,. Wb. s.v. surg6, simply ignore its exist'ence' Yet it is unthinkable that a perfectum -rdri' of the productive sigmatic type should be replaced by a totally unproductive fype -rAgi,. It is equally unlikely that Livius, who is very careful and indeed archaizing in his Latin, would have used. a neologism. The situation is precisely the reverse ; -rd'ni', by the the non-productive -rEgi, was replaced by -s' t'o the morpherne simple udditiott of the productivo aut'hentic to claim basZ -reg-. The form (su)'rEgi,t has equal form new The antiquity wilh (i,ntet)-ldgi't of Lucretius' -rdgi, rcplacing forms, -rd*i, was generalized. in compounded and fi.nally, only as late as Cicero, was extended to the uncompounded form. We muy demonstrate the spread of 'r6ri' in the historical period itself, as the following tables show' I have listed lhe nu^ber of occurrences of the infectum (stem reg-) and perfectum (stem rdr-) of reg6 and compounds in Plautus and Cicero : a distance in time of roughly 150 years' Plautus :

reg6 comi,gd pergd powig6 surgd regd com'ig6 duiga Eri,gd perg6 porigd surg6

IP to

67 31 1 t4 802 407 L74 tI 25 468 86 822

OF TEE

A.AOB,IST

33

The direction of the trend is too obvious to need comment. We may confidently assume that Latin rE*i, is simply abstracted from the compounded -rEri,, and that -rEri' itself replaced the earher -r4gi. Hence rdui, has no value whatsoever in determining the vocalism of the sigmatic aorist in Indo-European ; it simply repeats the long vowel ot -redn. TIiu" of the five original Latin sigmatic perfects with long root vowel, three have been demonstrated. to be of Latin origin, and a fortiori not inherited. from Indo-Europea,n. Tl, is pri,ma facie urlikely that a supposed morphophonemic process of Indo-European date, the lengthened grade d of the sigmatic aorist, would be continued in Latin by two forms alone, while orrer a dozen forms (v. supra) contradict the rules. The forms tEri and udro are to be otherwise accounted for. The evidence of the texts in early Latin suggests, in fact, that tdri, has the same origin as rEni' : generalization of a sigmatic form from composition, where it had replaced an earlier long vowel perfect in -e-. The perfectum tEri of the uncompounded verb does not occur in the republican period before Lucretius (who has one example) whereas Plautus has four instances of the infectum teg6. In the compounds, Plautus has infectum forms of d,Cteg6,obteg6, and pertegd but only the single perfectum d,EtEri,t(Most. 140, 163). It is only in Caesar and Cicero thab t4rit in the In the simplex and in compounds appears frequently. case of regd Lhe original perfectum form -rdg'it was actually attested in Livius ; we have no attestation of a *(-)teSi,. But the existence of the pairs leg6 : ldgi, regd : (-)r\gi,, makes highly plausible the rhyming pair tegd : *(-)tegi, ; these are the only three verbs in the Latin language of the structvte Ceg-. The previous existence of a long vowel perfect slem *tdg- appears even more likely in view of the fact that the stem tEg- occ;trs also in a nominal derivative ; tEgula ; *tEgi exactly parallels r\gula : -ri,gi,. Compare also d

34

INDO.EUROPAAN

-rAgi,:rEr:legr, as well as s\ilo : sdd,-Es,for the coexistence of long vowel forms in the perfectum and the nominal system. There is one further indication of the parallelism of these three verbs: the long vowel in the fo-participle, which arises from a Latin morphophonemic lengthening of -e- and -a- Lo 'e- and -d" before stem final '. voiced stop f suffix -fo-.15 So we ltave agd d'ctus,Jrango : rego : frd,ctus, cad,6 : cd'sus, and the three rhyming fotms rEctus, Ieg6 : lEctus, tegd : tEctus. It should be noted at the outset, that the analogy of the long vowel participle alone is insufficient to explain the long vowel sigmatic perfect, since both trd,ni, and uenr, have short vowel participles and uZctus (so explicitly Aulus Gellius, trd,ctus (cf. d'Etrectd,re) Noct. Att. 9. 6). Ernout has also suggested (Morphologie 315) that the vowel lengthening in all these s-perfects might have the same phonological (actually morpho' phonemic) basis as in 'ldctus, rActus, tdctus'- stem final voiced stop f morpheme initial -s- being treated like stem final voiced stop f morpheme initial -r-' But this is likewise contradicted by trd'ri, and uEri" since the stem final in ltalic times was neither voiced nor a stop, but a voiceless *trd,n', *u)en' reflect velar fricative lxl . The Italic shems *tragh-, *wegh-, and in such casesthis rule does not operatel 'compare lectum 'bed' < *ler-to-, TE *legh- as in Gk' These long vowel sigmatic perfects t'hus cannot )/xos. 'sigmatizabe of phonetic origin. They must result from tion' of older long vowel root perfects. We may tabulate the parallelism in formations from these three verbs: -+ -rd'ri' -regt' rdctus rEn r\gula reg6 -) -lEri' 'Iegi' lectus lEr leg6 -> *-tegi' -teri'. t\ctus tF.gula teg6 *-ftgi', b:ul the evident' fact The only unattested form is 14 Even if lego arrd lEn were not related in Indo--Europe-an, t'hey treated i"o* the synchronic Latin poinl of view, and '1.6.19 were clearly "o R,omans themselves' : Cf. eicero, de Legibus .""ft t'y ttre "u ,nn, 1i.e,, Ier\ 'illi' Graeco putant nom'ine a suutm cwique tri'buendo eamque eqo nostro a legend'o. annelLatam,, "" tt S"u on"this especiallv M. Niedermann, Phondti'qt'a histori'qwe d'u ao1;y72, 92-96 (Pa,ris; 1945).

VOCAI,ISM

OX' TEE

S-AOB,IST

35

that it fills a hole in the pattern makes its previous existence extromely likely. This view is not new. In Stolz-Schmalz, Lateinische Grommotil*, p. 370 (1890), the same derivation is set forth : 'rdxi,, tEri, < *rEgi, *t4gr,, cf. l6rn < ldgi,.' But the case was not adequately stated for this perfectly correct explanation, and it was subsequently given up in the revision of the Late'inische Grammatilc by LeumannIlofmann. The long vowel root perfectum is a Latin innovation, as has been convincingly shown by Benvenisl,e, Arch. L,ing. 1.16-19 (f949), and Kurylowicz, BPTJ 10.46 (1950). X'or a bibliography of other treatments, see the forthcoming monograph of Warren Cowgill on the Indo-European longvowel preterites. Cowgill's own view, that the long root vowel arises from contamination of the reduplicated perfect with the sigmatic aorist in long vowel, must itself be rejected, since we can show here that the sigmatic aorist did not have a long-vowel in Indo-European. The coincidence of Emi, 1 *Hf-Hrm-r6 and sdd,i' derivational process of lengthening erl- -> Em-, seil- --> sdil-. This pattern was then analogically propagated further in the system : leg- --> IEg-, as assumed by both Benveniste It was simply the incorrect appreciation and Kurylowicz. of the attested form suregit which kep!, reg- -> rd,g- ftom figuring here previously. We ma,y now add the third instance of this analogical morphophonemic process in prehistoric Latin, l,eg- -> *td,g- lo the list, on the clear evidence for parallel patterning of those three verbs, unique in their structure Ceg-. The chronologically later form td.ri, was subsequently formed in compound verbs, by the productive .suffixation of -s- ; the long vowel of teri, simply repeats that of *tdgi,, and can have no Indo-European origin. The fcxtn ftri, was later introduced in the simplex, uncompound verb ; but this is only well within the historical period. There remains udnt,, a cornerstone of the threefold equation 'uero : oadlc7am: a\stc, supposed to establish the 16 Aliernatively, from *e-em,- at a later avoid a laryngeal formulation,

stage, if one wishos to

36

fNDO-AUR,OPNAN

YOCALISM

fhe Indo-European antiquity of lengthened d-grade the The pattern of the attestations sigmatic aorist. texts observable in the case of rEri' and tdri' does not, recur wtlh udui,; in Plautus the perfectum of the uncompounded verb is well represented, in roughly equal proportion to the instances of infectum and perfectum of the compounds ad,ueh6,dueh6, d'Eueh6,'i,nueh6,reueh6. The earliest attestation of the perfectum would appear to be in Naevius (uererant 53 Marmorale). The root does not appear to have been in any sense restricted to an imperfective aspect in We may not suppose the previous existence of a Latin. *-udhi, ) udui, in composition and replacing the simplex. The explanation must lie elsewhere. That, explanation lies in the nature of the root final consonant of the verb stem, which in Italic was the velar fricative /x/. We have seen already that it is universally recognized that the long vowel trdri, is analogical in origin. But trah6 and uehd are the only two verb stems in Latin where the stem final consonant is the Italic voiceless velar fricative /x/ (henceforth vritten h, to avoid confusion with :r : /ks/) : *trah,-, *weh-. The two pattern exactly alike, and we may expect the formation by each of the productive sigmatic aorist ; the sigmatic formation is inherited in *wegh-. X'or the case of *tragh- see below. We would have then the following forms by the regular phononological treatment of Latin :

*trah-

*trah-s-

*trah-to-

->

trah-

*trd,*-

tra,cto-

*weh-

*weh-s- *weh-s-

->

ueh-

*ud,r-

uecto-.

*lak-

*lak-s-

*l'alt-to-

*sltelt- *spek-s- *spek-to-

->

(-li,ci,o -ldxt,

-t

spec-

-lectus)

sltEr- specto-

The participle in -to- is an old formation of Indo-European date ; only its role as a fi.nite passive in the conjugation of the perfectum is an lt'alic innovation. But the sigmatic

OF THE S.AORIST

37

perfect is the new and highly productive formation, par' ticularly in verbal compounds. We may contrast, the roots in final voiced stop with the above type: -> reg*reg-to*rag*reg(')reS- rE'cto' -+ *teg' *teg-to' *ftg*tegt\cto' teg' but in composition, with the new sigmatic perfectum, -) *rEg-s- *reg-to*regrdctorEureg-+ teg*te4'to' *ftg-s*tegtd'ntdcto' two The simplex. the to which was subsequently extended the of alternation morphophonemic roots trah- and.ueh- had a --> -lc's',17 -hperfect' sigmatic in the root final consonant, This agreed with the roots in final voiced stop, which had the change -g- --> -/c-s-, and contrasted with those in final voiceless stop -fr- t -lc-s-,with no morphophonemic change. The result of this pattern was that the vowel length of rEx-, tdr- (from rEg-, *tEg-) was reinterpreted' as an accessory . mark, -g- --> -k-s { Iength, and this was extended to the other case of consonant alternation : -h' -->-k's' * I'en'gth. Ilence

the replacement :r##

,;::: ]

occurred in the new and productive category ofthe sigmatic perfectum ; the older, inherited forms in -to', tractus and uectus, were not affected. The recent character of trd'ni, has been universally assumed by scholars heretofore. It has only been necessary here to show that udri result's from the same innovation in Latin, by an analysis of the analogical process involved' This view is likewise not new. Brugmann in the first edition (fS92) of his Grunilr'i'ss 2.2.1170 gives the equation Skt. oad,kpam: OCS abstc,and states qurte rightly that equation 'zweifelhaft' of Lat,. uEni, is ; but he did not state the reasons for this view, and subsequently, likewise without justification, ga,ve it ap (Grd'r.z 2.3.393). r? The creation of tho sigmatic perfect, antodates, t'ho passa'ge ^of intervocalic -h- to zeto; othdrwise tlere is no way of accounting for the stem final volar beforo -s-, -l-"

38

YOCAI,ISM

INDO-DUROPEAN

We have considered up to this point only the interrral, purely Latin evidence for the lateness of the sigrnatic perfects which exhibit a long root vowel. That evidence points unequivocally to the conclusion that these forms are not directly inherited from Indo-European ; in other words, that there is no evidence in Latin that the sigmatic aorist was characterized by a long root, vowel d. This conclusion may be supported by consideration of the comparative evidence. In the foregoing discussion it was suggested that the stems reg-, leg-, teg- inherited no perfoctum form from Indo-European, and that the long vowel stems r?g-, leg-, xtEg- were first formed analogically in Latin itself. If we look at cognate forms of these three verbs in older IndoEuropean languages, it appears that only the present stem is well attested. In the Rig Veda we have the presents rjyate, rfi,jdte, and the anomalous ,irajyat,i,r8 but no aorist, is attested; the only occurrence of a perfect is the hapax d,nrjdl.t in the Atharva Veda, which is clearly an innovation of Sanskrit by the anomalous reduplication. The only old aorist is Homeric iipe{a; but Greek *tror- | -tran-) which subsequently underwent lengthening to trd,u-, along with *weh-s- (wen-) > u€.r-. 20 This and similar for:ns rec-c,irlo, ret-tuld, ret-tud,i, do not contain a syncopated reduplication syllatrle, but the preverb red- a,s in red,-i,mo, red,-eo. Tl;.e present, is scanled recc'idere llke reccidd (and so written bv editors) ir, ..9.,-Lucretius 1.857, 1063, 5.280, to cito only a, few of numerous instances. See Meillet, Mdlanges Hauet 273-7 (P'aris, f 90g), against, leumann-Ilofmann, Lat. Gr.t g2 and the bibliography given (The latter do not seom to know Meillet's study.) there.

4L

The conclusion remains the samo : LATrN TNHEn,TTED TIIE SIGMATICAORISTX'ORMATION X'ROMLATETTVNO-EUNOPNIU WITII A sHoBT RooT vowEl. Some of the sigmatic porfects of Latin appear with long root rrowel, but these are without exception due to Italic and Latin innovations. The elimination of Latin from the languages showing inherited lengthened grade in the sigmatic aorist is enough to raise some doubts as to the nature of the Slavic evidence, which has been universally regarded as showing an inherited long vowel. Since the end of the nineteonth century, it has been customary to give the threefold equation Skt,. aud,lcpam: OCS abs[t : Lat. ueri,. Brugmann in the first edition of his Grund,ri,ss rejected the introduction of LaLin uEni,; but Hirt in 1900 (Derinilogermanische Ablaut 185) put the three on a par, and in the second edition of the Grunilr'i,ss in 1913, Brugmann had accepted all three (2.3.393). Since that time this threefold equation has been ubiquitously accepted as paradigmatic for the sigmatic aorist with fndoEuropean long vowel. We have seen in the preceding paragraphs that Latin uEnd m:ust be rejected, since it is This leaves only the equation aadkpam: built on *udri. udsd, lo establish an IE *wdgh-s-. This equation itself however cannot stand ; for Old Church Slavonic odszT ftom uezQis a ghost word. No such form exists in Old Church Slavonic, or in any later Church Slavonic texts. The aorist udsd, wherever it occurs in Slavic texts, is always from ',-eilg 'lead' or its compounds, reflecting IE *wedh-.zl A total of seven sigmatic aorists are attested in Old Church Slavonic, which show lengthened grade (the forms 21 Seo llorace G. Lunt, Old, Church Sl,auoni,cGramtnarz 91-2 (f 959) for an exhaustive list, of all the non-productive aorist for:rns fould in Old Church Slavonic. For these and the olher forms found in later manusoripts, see F. Miklosich, Be'itrcige zur oltslouenischen Gram,mat'ik. (1876). Tho later manuscripts simply repeat BbAW Wien 8l.ll0-I13 tho same lexical items from Old Church Slavonic; cf. W. Vondr6k, Verglei,chend,e slacs'ische gram,mat'ilc2 2.125 (1928) on ths s-forms: 'Dieser Aorist kan:r nur im Altkirchonslavischon belest werdon.' Tho only properly Slavic linguistic work to quote a, adsi < aezp is Meillet,-Vaillant, Le slatse cornnlil,mz 240, where the source of 'tho form must be a work such as Brugrrranrr's Grwnd,r'iss.

42

INDO-EUROPEAN

VOCALISM

are given first in bhe present stem, identical in these forms with Lunt's basic stem, then the aorist stem): nesneguEsued,grdsgrebrerrelcti.ntekZarZegbas-. bodAs in the case of Latin, Greek, and Celtic, there is a mote extensive series of forms frorn roots of the structure T'ER-, TERT- suffixed by -s- in the aorist. Slavic would be the only language with a sigmatic aorist where the original putative long vowel in a tautosyllabic diphthong would have left an accentuational reflex. But it is impossible to tell anything from the intonation of the modern South Slavic languages (Serbo-Croation, Bulgarian, and Sorbian alone preserve the Common Slavic aorist), since in theso languages the intonation in all cases simply repeats that, of the infinitive, and has no independent value.2z Aorists of this structure in Old Church Slavonic are: bljud,-

bl,jus-

(*-eu-) ("'en-) ( ,, )

ald,lc-

vldr-

(-el-> -li-).

These five show a simple full grade (-e-) root in both present and aorist, and there are no grounds for assuming anything else. To this group we may add another, where the present stem shows Slavic zero-grade, beside full grade in the aorist. caittitkldnp|ncxn'im-

caxsbi,s(*-e,-)

e8-

22 See C. S, Stang, Slaaoni,c Accentuat'ion f28 (Oslo, 1957).

and with rblLi, < mldzur\zmirsti,r'iIr-

OF

TIIE

,S-AOR,IST

43

erlel, mlbsarbsmrbrstrbrErbu-

(*-el-) (* -er-)

In these forms as well, there can be only hypothetical reasons for postulating lengthened grade in the aorist ; the forms themselves show full grade alone. We have finally a series of sigmatic aorist to roots v"ith Slavic lengthened grade in the present stem as well as tho aorist, which therefore show no alteration: Selc-

8efr-

bdpepagrast-

bspbr- (hist'otically,[. < *od] pasra8-

The problem in these Old Church Slavonic forms is whether the lengthened grade is inherited, or is simply a Slavic innovation. We know that the Balto-Slavic verb shows a, number of grammatical categories with lengthened root vowel, which result from purely Balto-Slavic innovations. One may point especially to the iteratives in -d,ye-, the Baltic preterites in -d-, and the long vowel presents of the type *sEd,-. These have beon discussed in detail in Kurylowicz's Apophonie, and his arguments need not be repeated here. The sigmatic aorist is found only in Slavic, not in Baltic. This fact alone is sufficient to suggest that the sigmatic aorist was not inherited as a productive category from Indo-European through the period of Balto-Slavic community. Its development must, be of purely Slavic date ; the category is Common Slavic, but not common BaltoSlavic. We should expect then that there would be characteristic late features of the morphophonemic alternations observable in the Slavic forms. Such a late feature

44

YOCALISM

INDO.EUR,OPAAN

is, in fact, observable in the o ; a al!,eration of boilp ; aorbasd,. This alternation in the sigmatic aorist coexisting with e : J, is not fndo-European in date: it is only as old as Common Slavic, at the most, Balto-Slavic. For as Kurylowicz has shown at length inhis Apophoni,e (286 ff.), the languages of the Northern group, Germanic and Balto-Slavic, developed an apophonic equivalence e i o : o | 6, where the difference in vowel quantity in o : d was treated as morphophonemically equivalent to the qualitative ablaut e : o. Subsequently, this process of lengthening was extended. to e (yielding Z), and an apophonic situation eo 66 created, in Balto-Slavic times. This lengthening was later extended independently in Baltic and Slavic, with the resultant alterna,tions Slav. e o i il : ealy

Letto-Lith.

e a i u ealu.

This morphophonemic pattern is extremely frequent in the iterative formations in Baltic and Slavic. The independence of these two apophonic systems in Baltic and Slavic may be deduced from the existence in Lithuanian of a number of lesidual forms reflecting a Baltic system eo 6 whiah, after the change o ) ea 6

d (:

Lith. uo)

a, was replaced by a (:Lith.

o).23

It, is this system of alternations which we havo in Slavic in the sigmatic aorist, and nothing inherited from IndoEuropean ; aed,-: ads- : bocl,-: bas- : cult- : cais-. That the latter may contain original full grade *ez is from the 23 I have discussed these Baltic apophonic phenomona in Eaidence Jor Laryngeals 49-50 (Austin, 1960).

ON' THE

,S.AOR,IST

45

descriptive point of view irrelevant; the alternation is Slavic tr : i, like Slavic o : a. fn the case of Lhe verb caitE ca'isti 'bloom', it is in fact extremely doubtful that there ever was an fndo-European aorist *kwe,it-s-, since the verb is a Balto-Slavic innovation, and only feebly represented in Baltic : Latv. lcuit4t 'shine '. The aorist cc'iswas formed only in Common Slavic times, following the Common Slavic apophonic pattern. In Slavic there is a widespread tendency to contrast the present and aorist systems by a formal apophonic mark, as has been noted by Stang, Slau. u. balt. Verb, 9O ff. Such contrast appears notably in forms with an aorist stem in -d,- : pi,39 : p\sa-rd,, pliSq : pilIza-rd. Stang has stated explicitly 'Die fiir das slav. System characteristischen Bildungen sind. die Typen ueilq : ad,sd,,mirq : mrU,rd,,berg : bilrord,. Sie zeichen sich durch das Ablautsunterschied zwischen Praesens und Aorist, aus.' These apophonic contrasts represent living, productive Slavic morphophonemic alternations, and the very fact of their productivity indicates that we have to deal with phenomena only as old as Common Slavic. In no wise do they reflect much older apophonic phenomena of fndo-European date, particular those as feebly attested in the verbal system as lengthened grade. For the aorist the original contrast need only be viewed as full-gradefzero grade. Whore the present stem shows zero grade, the aorist has full grade (mtrp : mrdnd 1 *mer-); where the present has full grade in TeR- roots, the aorist has zero grade plus suffix -d(berg : birard). In the case of the roots of the structure TeT-, no zero or:reduced grade is possible : the 'lengthened grade' fbf + -s- fills the function of the zero grade, and represents a Slavic development. The situation is exactlv parallel to that of the 'lengthened grade' iterative formations in -ayelo- independently developed in Latin and in Balto-Slavic, as discussed by Kurylowicz, Apophonie 301 tr. As in the case of the iterative formation, the contrast zero grade : full grade is sufficient to bring about the creation in Slavic times of a new apophony e : d, in tlne

--

46

INDO.EUR,OPEAN

s-aorist, ; the sigmatic aorist itself was inherited only with full grade vowel : *mer-s- (mrd,rdl : *ued-s-. It was the conirast ptes. mir-e- : aor. *tner-s- which was ultimately responsible for the creation of aid,-s- and the apophonic contrast aed,-e- : *ud,il-s- (adsd), in conformity with the general pattern of morphophonemic relation between present and suffixed aorist in Common Slavic. In the iterative formation in -d,yelo- mentioned in the preceding paragraph it is clear that we have to deal with a process of lengthening of Balto-slavic date : cf. Apophonie 'we have, for 303. n'rom the basic tek-'run example, OCS -th,lcati: Latv. tgkd,t'run, flow '. Now the sigmatic aorist is a purely Slavic formation, not found in Baltic at all. The sum of s-aorists inherited from fndo-European must have been extremely small, and of little'importanee in the Balto-Slavic period. It is only after the separation of Baltic and Slavic that this nucleus of forms achieved productive status in Slavic ; in Baltic they were eliminated entirely. This would suggest that the Balto-Slavic morphonemic alternation e : d, as in the Balto-Slavic iterative, was only in Common Slavic times extended to the new sigmatic aorist ; in other words, that the alternation pres. teltQ : aor. tdrd is posterior to, and probably even modeled on, the alternation pres. telcq : iter. -td,ka'jg. fn Baltic, where the thematic and sigmatic preterites were replaced largely by the preterite in -e-, the latter as well developed a lengthened grade of purely Baltic date: pres. L sg. geriil,, pret. 3 sg. gdre. Indo-Iranian is the one remaining dialect group of the Indo-European languages which shows a characteristic long root vowel for the active sigmatic aorist. The fact that one language group alone in fndo-European shows a given linguistic feature is of course in itself not enough to warrant, considering such a feature an innovation of that group. But if the evidence of all the other relevant languages points unequivocally to the conclusion that such a feature did not, exist in Indo-European, then it becomesvery likely that this feature is, in fact, an innovation in the dialect

VOCAI,ISM

OF

I'IIE

S-AORIST

47

where it appeals. The likelihood of an innovation is even stronger if it can be shown that the feature in question represents an expanding and productive category in the language under study. Such is the situation of the longvowel sigmafic aorist in Indo-Iranian. The distribution of full-grade, lengthened grade, and zero grade in the Indic sigmatic forms in itself indicates that there has been a certain reformation and readjustment of the apophonic forms in the Indo-Iranian and Indic periods. In particular one may point to : (o) the contrast of full grade middle forms to TER- roots (aneplo) and zero grade middle forms to TERT- roobs (arudd,ha),both showing lengthened grade actives (anaipom, arautsam) ; (b) the presence of the lengthened grade in singular, dual, and plural of the active, which contrasts strikingly with the grade (whether athematic present, where the 'strong' lengthened or full) is found only in the singular number. Finally, there are such isolated forms as GAv. q,srd,5ild,nn, mahma'id,i,,RY agasmahi, mastyo, aasdmah'i,JB atas'i, with zero grades Itan. sru-, lna-) a"nd Indic ga-, ma-, 'ud,-, ta,with a 1?, as well as RY amatsuh with irregular full grade.2a Such variation suggests that the observable apophonie situation is not original. One particular class of forms in Sanskrit shows conclusively that the lengthened grade is late: the -zp- aorist. We have ample evidence in the -ip- aorist that only the full (e-) grade is original, e.g., dai€-, d,haanip-, mathig-, Snathig-,Sram'i,.s-, sphari,p-,all from the Rig Veda.25 Beside these one may point to the forms RV agrabhi,t, Icram,igam, auailhdt, asaanit, -LY astari,h, aiari,t, ana,yit, which are all descriptively -dp-aorists. The extension ofd, al,the expenseofa in these forms is a processobservablein the Vedic period ; the mechanism has been described in detail by Kurylowicz, loc. c'i,t. It is evident, t'hen, that the replacement of original basic a-vocalism by the new lengthened grade d-vocalisrn 24 \,Vhitney, $ 887 ; Thumb-Ilauschild, 1.2.305 (Heidolberg, 1959) ; Bartholomae, l.L208. 25 See Kurylowicz, Apophonie 162.

Handbuch d,es Sanskrt'tg (lrund,ri,ss d. iran. Phi,l.

48

INDO.EUROPEAN

is a new and productive morphophonemic process' We the have finally the proof of this in the extension of beside atd'ri'ma, aorist lengthened grade even to the root' thJsigmatic-atd'r'i,9mo. From the hist'orical point of view only The *a-taii-ma is regular for t'his type of root' aorist' morphological secondary, lengthened. oo*"1 is of late, just as is that of all the -'i9- aorists which prienience, iho* t long vowel. The basic type is TaHi'.s'lTaTi'g-' As we have noted above, following Meillet, the apophonic is a clear contradict'ion of situation in Indo-Iranian Saussure's morphological stricture of apophony in suffixed a forms. It is furthermore the unique example of such a such' as verb, contradiction in the Indo-European innovation' Indo-Iranian an words secondary origin, in other In fact, Kurylowicz would" Ui ttte simplest explanation' the solution' indicated already has imhis Agtoythonie(p' 160) t'he, between similarity striking when he points out the of that and' aorist, sigmatic the apophonic situation in the of genesis the of explanation Kurylowicz's rfaini.'u Iengthened grade in the sigmatic aorist and that of vrddhi are-substaniially identical : the contrast' between variable (full or zero grad'e) vocalism in the base form (forme d'e and the,invariant full grade fond,ati'on), te"e ttte present, (forme form fond'd'e),here the sigmatic vocalism of the derived that' the full grade assumes case any in form. Kurylowicz aorist ("f' sigmatic Indo-European is original in the lengthened the of creation the and that c'it'), Agtoythinde,l,oc. g*a" i* later ; in view of the fact that no other language inherited. lengthened grade in the sigmatic aorist' f,t, "r, the it is only necessary to suppose that the generation of in only place took grade lengthened new morphological itself' Indo-European in not and period, the Indo-Iranian The close similarity botween the Indo-Iranian sigmatic as aorist and vlddhi speaks for this as well, since vlddhi a productive morphophonemic process hardly exists outside Indo-Iranian. 16; of the root'nbuns: cf' kd'rs-lk.'f' (/'lihat ' heart 'iiake') "i*rou,ti"--ro-r",,",-tt"-"t51 beside hard-lh'd"

vocAr,rsM oF TEE s-AoRrsr

49

As to the genesis of the lengthened grade proper within the fndo-franian period, it is certainly likely that the point of departure was the contrast between variable base and invariant derivative, as Kurylowicz has suggested. The s&me was assumed in the case of the Slavic forms discussed above. But it is not necessary to demonstrate the existence of a particular phonological channel for the repeated recreation of a given morphophonemic alternation (as here the lengthened grade) in every successiye morphological category where it appears. This seems to me a methodological fault in Kurylowicz's Apophonie, which frequently leads him to overly ingenious explanations of the channel in the case of individual morphological categories. These are unnecessary, and serve only to undermine the credibility of his reconstruction as a whole. The historical problem is not to explain the genesis of the lengthened grade in all the categories in .which it appears, but to isolate the original category in which it appeared, which served as model for all the succeeding instances. Once a morphophonemic alteration is developed ln one part of the linguistic system it can and will spread to other parts of the system, if the functional relations are comparable and the resultant phonological situation is permitted in that system. If fndo-franian developed, or for that matter inherited, a productive derivational pattern between base form in root vowel a, arlr, a,ifi, etc,, and derived form in maximally fifferentiated ('polarized,) d,, d,r, d,'i, etc., in the nominal system ; then that derivational pattern is free to repeat itself in unmediated fashion in the verbal system, where there exists a comparable functional relationship between base (,forme d,efond,at,ion) and derivate (forme fond,de). Ilence in the sigmatic aorist the fndoEuropean contrast, inherited into fndo-Iranian,

I

e, erlr, eifi

rt e , e r , e L

50

INDO-EUROPEAN

was subsequently in Indo-Iranian replaced by the productive morphophonemic pattern of the nominal system:

I

J

a, arlr, aili &,

ar , q,1'

But the fundamental, inherited vocalism of the aorist is full grade. In exactly such fashion as this we haYe assumed that in Slavic the inherited (and non-productive) apophonic relation between basic present stem and derived aorist st'em

i

e (o) erltr i'lE. . .

J

e (o) er

was replaced directly by the new and productive apophonic relation between present' stem and derived iterative

t-

Ib

a

As regards the instances of irregular vocalism of the above' the Indo-Iranian sigmatic aorist such as those noted of -zp' a number explanation lies elsewhere. We know that of sigmatization the aoiists owe their origin simply to evident particularly is roots' This original -such root aorists of se! forms as atd'ripma beside atdti'ma; the sequence in *tarip-, whence (a)td'ri,g-(and' (a)tari'is aorist stem *tari- ) most easily by a nonce analogical transformation)' We can masi'ya' agosmahi, alcount for such forms as asrd'Ed'd'm, In every aolists' root of atas'i, as sigmatizations 1)asi,nxa,h'i,, sigmatic the mid'dle, the one of these cases, all forms of athematic form simply repeats the regular zero grade of the the root Compare formed' aorist micldle on which it is a'n1'd'ta' agata' RY aganmahi aorist forms GAv. sraotd' (imv'), atata' In these sg' 3 (non-R'Yl 3 pl. atnata, and the Vedic -s' directly of suffixation the cales the formation represents ga-s', aa't'' tct's" onto the root aorist middle stem : sru'-s''

VOCALISM

OX' THE

S.AOR,IST

5l

The fact that the lengthened grade is found. throughout the activo of the s-aorist (rather than only in the singular, as in the present) has been explained by Kurylowicz (Apophoni,e f 64-5) as imitating the model of the radical and thematic aorists, neither of which show a variable vocalism between singular and dual or middle in the active. This is quite plausible ; and since we know that this fixation of vocalism is an Indic innovation, it follows that the rigid sigmatic aorist vocalism in the active indicative is itself an innovation. There is further the argument from statistics, which can be seen from the comprehensive study of J. Avery, . Verb fnflexion in Sanskrit', JAOS 1,0.Zlg-824(I8S0). It does not seem to have been noticed that in the Rig Veda, out of some 180 occurrences of the s-aorist active, only 19 occurrences of the indicative are found, and these from l0 roots only.z? Of these I0, ayd,samayd,susmust be exclud"ed, since the root vowel is always long, and one, amotsus shows 'irrogular' full grade and no vrddhi. Ifence we are left with only 15 occuruences of 8 roots forming an active sigmatic aorist with vrddhi in the whole of the Rig Veda. ft is probably not a coincidence that all these forms are roots of ihe structure TR(T); the one s-aorist active indicative of the structarc TET does not show the lengthened grade (amatsas), and it is possible that the analogical spread of the vlddhi form had not yet spread fuom TR(T) to TET toots in early Vedic times. The final argument, against the antiquity of the lengthened grade of the sigmatic aorist in Indo-Iranian is the funda, mental fact that this lengthened grade is proper only to the active voice, and not the middle. As will appear from the evidence presented in the following chapters, the original locus of diffusion of the sigmatic formations in fndo_ European is in the middle voice ; the active fortns are secondary. We have already seen indications of this, in 2? Tlae forrn auyk1a,m cited by Avery here is a _so_ aorist ; for the lorm,.see Geldner_ R,ig Ved,a III ad 10.159.5, against Winitrney, Roots, a,nd Grassman, IZb. X'urthor, aDesdn by its endiig cannot be aorist. Ilence f havo reduced his total correspiondingly. ""*igm"tiJ

52

INDO.EUROPEAN

the occasional replacement and renewal of root aorists middle by sigmatic forms. This is particularly apparent in the case of the Gathic forms asr6'5ild'm beside sraotd,. We know that the active form is ancient, by comparison with RV airot; we know as well ihat the zero grade form is historical, by comparison with P.'Y lrud'hi and Gk. rtr0dr' The Gathic form asril'Ldd,m represents an archaic renewal of the aorist middle by a sigmatic form of Indo-Iranian date, which had not yet spread further even in the Rig Veda ; it is only in Br5,hma4a times that the new sigmatic aorist active airaus.r't appears. The basic evidence for the middle voice as the original channel for the penetration of the sigmatic aorist into the system will be considered in the following chaPters. S 4. Trrn srclrltrc

Aonrsr aND Trrn MTDDLEYorcn.

Tnn appearance of a generalized -s- throughout the middle, contrasting with an athematic (or at any rate asigmatic) formation in the active, has some chances of reflecting an since it is mirrored earlier Indo-European distribution independently in several Indo-European dialects. The situation in Greek is at first sight just the reverse, where in the Ilomeric language there exists a series of athematic (from the Greek point of view) aorists rniddle [,,pro,rdAro, &trro, which have sigmatic active forms : d'pp'evos, i).l).,,rcro,A€xro, i)'ercro, tp,,,'xro,xa./rt1xro, Z$0r,ro,Ixtaro, fujpn1v,ty,ttvuro,6lxro, -xrty"evos, tyuro, (oouro. But the universal athematic character of these forms, maintained 3 by Kurylowicz,l Meillet, 2 and Chantraine has not been completely accepted ; for the bibliography one may refer to Schwyzet, Gri,ech'ischeGrammat'ilc I.194 ff., who hirnself favors sigmatic explanation for some of them. In the case of ilpp.evosand f,tpro, t'he existence of an extended present &paptoxat, \pvultt makes a root aorist formation likely' But 1 Eos 32.22Ltr (1929). 2 Meillet,-Vendryes, ()ramrn, Comp. des langues classi,quesz203' 3 Gram,ma'irehorndrique 1.383 (Paris, 1948).

S.AOR,IST AND

THE

MIDDLE

YOIOE

53

the present of rd)ro is a'dltrc,r, a normal -yelo- d,erivative present which regularly in Greek forms an s-aorist in both active (as here zril)e) and middle. The root is in all likelihood se!, *pelH- appearing in Lat. pell6 < *pel-n- (for this treatment of se! nasal infix bases in Latin compare the identical sternd); from the Indo-European point of view an alternation pres. *plH-ye-til aor. mid. *pl.to is surprising. For Greek one would expect rather rdA)\ol*blvlzo from such a type ; cf. pd.A).al€p\ro (A, 675). ft is clear that the aotive s-aorist is chronologically late, a purely Greek development, since it is simply built of the Greek stem zral., not on any Indo-European full grade : rfiAe 1 *r.a)\-o- -ai is attested in I sg. uddd. We know as well that IE -i was combinable with a 3 sg. zero ending, as is proved by the Greek thematic 3 sg. present -eo { -e f c, where -e is simply the thematic vowel. The Ilittite hi-conjugation 3 sg. -i may also contain deictic -o suffixed Lo a zero ending. The deictic -i alone, suffixed to the bare root with zero ending, occurs finally in a very archaic category in Indo-Iranian: the 3 sg. aorist passive. The most archaic form of this class in the Rig Veda is jani 'was boln', which shows the absence of the secondary vlddhi as in jiini,. The augment (ijani,) is likewise secondary, subsequent to and conditioned by the identifi.cation of this form with the aorist system. To my knowledge this explanation of the Sanskrit aorist passive has not been proposed before ; but it will most adequately account for both the formal and the functional situation.T The interpretation of the personal ending as zero, with the suffixation of a deictic particle, explains why the form is only a 3 sg., and why there is no complete paradigm. Tunctionally, the type is only secondarily a passive, and the basic value is that of an intransitive ; cf. Whitney, $ 846, who points out that the form is passive only if the verbal base is transitive. What this implies is that the form is in origin simply the bare root, the neutral verbal notion alone, in the 3 sg., the non-personal form, with zero ending. The -d is only an emphasizing particle which has become obligatory, as in the athematic imperative in -d,hi. The basically neutral, intransitive value of the bare root form \lrith 3 sg. zero ending which has developed to the fndo-Iranian aorist, passive, is thus directly comparable to the basically neutral, intransitive value of the sigmatic aorist in its earliest manifestations, which we have discussed 7 X'or earlier attempts at an explanation see Thumb-Hauschild, The Ho,nd,buch d. Skt. 1.2.298-9, with tho reforencos qiven there, solutions proposed are noither formally nor functidhally satisfactory.

104

rNDo-EURoPEAN

above. The form in -s is likewise in origin a bare root form T.ith 3 sg. zero ending ; the only difference is that the root is more complex, being enlarged by -s. But both types show in origin the same characteristic neutral, intransitive value ; they simply predicate the existence . of a verbal action. The Indo-Iranian aorist passive shows the bare root form with 3 sg. zero ending, suffrxed by the particle -d; the forms of the type d,hd,sshow the bare root (enlarged by -s) with 3 sg. zero ending, but without the particle -d. The parallelism is even closer. X'or the free combinability of -c with zero ending makes an alternation ,E f -s / .E f -s-i predictable for Indo-European. We have seen how a new morphemic cut created a 3 sg. ending *-s, occurring in Hittite, fndo-Iranian, and vestigially in Tocharian (neksa I nek-s I a). In fact, a vestigial 3 sg. *-sd exists dialectally in two Indo-European languages as a 'primary' ending : Tocharian and Germanic. In Tocharian A we may compare 3 sg. Ttiilhti€ 1 *bhl.g-si from piillc- 'light, shine ' : Lat. Julge6, Gk. $AQa. The ending is not found in Kuchean. In Germanic we may compare the Old Norse 3 sg. (and 2 sg.) br{tr 'breaks' < Gmc. *breuti,z, IE thematic -es'i, or Northumbr. f.nd,es (:f,nd,ep).8 This analysis of theso forms as originally 3 sg. eliminates the necessity of assuming a transfer fuom 2 sg. to 3 sg., with Krause, loc. cit., which is contrary to the observable nature of analogical extension in a paradigm. In Old Norse the -r I *-si has spread to the 2 sg. from the 3 sg. and fills both functions, just as Hittite -s or fndo-Iranian -e in the precative ; but it is the 3 sg. function which is basic, as appears from the position of Toch. A -p, or Skb. d,hds. In Hittite the result of the encroachment of -.i on the 2 sg. function was the elimination of an earlier 2 sg. characLeized by a -t-, which is preserved in the 2 sg. mid. pres. and pret. -ta(ti,), the 2 sg. h,i,-ptes. -fi, and in the unique 2 sg. hi,-pret'. paitta 'you gave', which doubtless shows the original form of the ending. On the other hand, .where the -s was not 8 Forms

from

I{rauso-Thomas.

Toch. El,em. 259,

ORIGIN

OX' THE

MORPIIEME

.S.

105

extended from 3 sg. to 2 sg., the original 2 sg. ending in -f- was preserved, as in Tocharian A 2 sg. piillciif, 3 sg. piillctip, and Hitt. 2 sg. pai,tta, 3 sg. pa'i!. The process of croation of a desinence from a root enlargement must havo been widespread in fndo-European. IJ:rb (Idg. Gram. 4.L04) has rightly compared Ved. 3 sg. dhaL dhdti with Lat. sacer-d,6t-,or Gk. 3 sg. tyva : d.yvor-. The process is the same, and structurally d,hd,t-ais identical wilh d,hd,s-a. It is perhaps possible finally that the classical Indo-European 2 sg. ending -s(i) represents the old 3 sg. form, ousted from 3 sg. to 2 sg. by the encroachment of a newer 3 sg. -f(i). Such rearrangements must have taken place : the I sg. -uf -w'i, pamllel to -tl-ti,, -s/-sc which appea,rs in Hitt. I sg. pret. -u-n, Kuch. -w-a, Luv. -zoi, has elsewhere in Indo-European been ousted" from the I sg. by the pressure of the new ending -ml-mi; it survives only in the dual : Skt. -oo, OCS -od. This ending -uf -uti is itself. only a root enlargement ; Ivanov (Tor. Jaz. 17) has rightly compared Kuch. I sg. neltwa with Gk. zercrjs. On the basis of comparisons of this sort, together with an understanding of the functional transformations and transferences which have reshaped these paradigms, we can see back to a far more fluid and flexible state of affairs in the system of the fndo-European verb. The rigid paradigmatic structure for the three persons of the singular, -m(i') -s(i) -f(i), belongs only to the latest period of Common Indo-European, and was completely achieved only after the separation of the dialects. There remains the question of what was, in fact, the original function of an enlarged root form such as *prelcs. To such a question one can offer only more or less unverifi.able hypotheses. The interpretation of such a form as showing a zero ending is sufficient to mark the form thenceforth as verbal, with the consequences which we have described. But when we consider the neutral, intransitive value still present in such forms in the historical period, as well as the formal identity of -s with the mark of the nominative singular, then it is hard to avoid the supposition that the formation was originally nominal.

f06

rNDo-EuRopEAN

We may tentatively suggest that such nominal forms could be used simply predicatively: *Hneks' unnatural death (occurred)' *gtrelcs'asking (occurred) '. The syntactic point of contact between the nominal and verbal expression could have been furnished by the nominal phrase, where no verb at all is present. A good typological parallel is furnished by the history of the Persian preterite, where the finite forms 3 sg. Icard,,etc., are all derived from a nominal form lcrta- 'done ', and the original locus of diffusion was the nominal phrase, as in OP ima tya mand, krtam 'this is what (was) done by me'. Such an originally nominal form thus has been functionally differentiated into a verbal form (Pers. kard,'fecit, ') and a nominal form (Perc. kard,-a ' factus '). By the same token, a,t am earlier stage, we could have had a similar differentiation into verbal forms, *Hnelcs-a, *preks-o (Kuch. Z sg. nelcsa, prelcsa), and nominal forms *Hneks, *preks (Lab. ner, prec-d,s). This view of the functional identity of verbal and nominal expression in such a case as *Hneks, xprelcs, receives some measure of confirmation from Hittite. In the Kikkuli-texte we have as indications of time, such equivalents as mohhan-mo lulckatta 'when it becomes day' (3 sg. verb) andlzckkatta-ma'the (next) morning' (adverb), or mahltan-ma, nelcuzzi 'when it becomes evening', and mahhan-ma nelcuz mehur lci$ar,i 'id.', with 3 sg. nelcuzzi ' becomes evening ,, and 3 sg. k'iiari, 'becomes ' with the nominal nekuz mehur 'evening-time'. Compare also the simple nominal nelcuz rnehur-md, 'in the evening '. The distinction between 3 sg. Iulckatta and adverb lukkatta may be secondary ; and the 3 sg. verb nelcuzz,iI *nsfuID-fzsimply has the deictic particle -i suffixed to the same enlargement -l which furnishes the nominal form *nelcw-t- of l{itt. nelcuz (*nelcu-t-s). Compare also *leuks in both Kuch. 3 sg.lyaulcsa and Lat. nom. sg.ld,r. The difference between these two and Hitt. lukkattais ,"t o"" .f choice of enlargement,, *leulcs : *leukt. "]!1"tgl s A.

r96r).

Iiammenhuber,

,

PAR,T TWO

Celtic La comparaison perrnet, dans la mesuro indiqu6e, do 'rostituer ' la ' Iangue comm.uno ' dont les la,ngues d'ule m6mo lfamille' sont les formes prises au cours du temps. Mais il reste dse rendre compte do ce qui s'est, pass6 ontre 1'6poque do .c6mrnunaui6'ot, la^dato ou les langues commencent d'6tro attest6os en fait, Le problbme est difficilo, souvent insoluble. Meillet.

fl !

S 9 Tun Sysrnu

on rEE Pnnrnnrrn

rN OLD IRrsH

Tnn traditional description of the formation of the preterite in Old Irish, as exemplified by Rudolf Thurneysen's masterly Grommar of Old, Irish (415-37), distinguishes three fundamental categories: the s-preterite, the f-preterite, and the suffixless preterite. Thurneysen further subdivides the suffixless preterite into two groups, a reduplicated and an unreduplicated, both of which share the same endings. The distribution of these three formations of the preterite stem is, roughly speaking, as follows : all weak verbs have the s-preterite ; strong verbs in certain root final resonants or -g have the f-preterite ; and all other strong verbs have the suffixless preterite. Thurneysen's analysis is, in the main, both adequate and efficient. But in his handling of the third group, the unsuffixed preterite, he has occasionally confused historical and descriptive facts in the repartition between reduplicated and unreduplicated classes, leading him to consider as 'forms with peculiar reduplication' certain groups of forms which in fact do not exhibit reduplication at all, but rather a long vowel preterite in -e- (before non-palatalized consonant). On the other hand, his unreduplicated class sets up as parallel formations the preterite in -d,- and the

r08 preterite in -i,-, as well as a very interesting group of anomalous forms, whereas the -i- preterite is by its formation -correctly analyzed by Thurneysen-on exactly the same chronological level as the preterite in -d-, but markedly younger than that in -d,-. Similarly the anomalous group, aside from hesitating about two d-preterite forms, succeeds in masking the Irish evidence for a very significant structural principle of the morphology of the Celtic verb : the regulation of the appearance of the reduplicating syllable. The net result of the procedure is to misrepresent the complexity of the Old Irish preterital formations, to obfuscate certain clear principles of formal organization, and to introduce unnecessary confusions and misapprehensions into the historical problem of the genesis of these formal classes. Like so many other attempts at combining descriptive and historical data and analysis, it succeeds here only in falling between two stools. However, by the technique of proximate and internal reconstruction it is possible to separate clearly these chronologically earlier and later formational types of preterites in Old frish as we have it preserved. This is, in fact, principally indicated already by Thurneysen in his notes on the origin of some of the types, though he does not concern himself with their relative chronology. The later types, such as the -i,- preterite or the -i- preterite, are universally attributable to the immediate or eventual effects of the phenomenon known as lenition. This phonological development has been dated most cogently to the second half of the fifth century A.D.; consequently, the development of these phenomena to the status of morphological marks belongs to the Primitive Irish period, and lies outside the scope of a work dealing with the ori,gi,ns of the Celtio verb. f hope to present at some future dato a detailed analysis of the constitution of the Primitive frish verb. With the isolation of the relatively recent, Primitive Irish layer of preterital categories we are left with a substantial residue of forms, still within Thurneysen's threefold classification. The recent formations are limited in

SYSTDM ON'THE

PRETERITD

109

extent ; the remaining formational types include the great majority of Old Irish primary verbal roots, as well as all the derivative verbs, principally denominati'ie, which constituted the single unsuffixed productivo type in Old. Irish of a- and r,- weak verbs, i.e., with a stem in nonpalatalized or palatalized consonant respectively. Among these remaining forms of the preterite in Old Irish we may distinguish five morphological types. These are, in the terms of traditional frish and Celtic grammar, (f) the s-preterite, (2) the f-preterite, (3) the reduphcated preterite, and (4) the d-preterite, and to which I add, in Indo-European terms, (5) the radical and thematic aorists. In a descriptive analysis of Old lrish, forms from the last category ma,y appear, depending on the individual lexeme, in certain of the first four categories, where the preterital function is preserved, or may also appear as presents. All but the fifth of these are Common Celtic formations, attested both in Goidelic and Brittonic. But the nearly complete abolition of complex primary verb morphology in the Brittonic languages, by the time of our earliest extended records, has reduced tho value of their evidence by and large to only a corroborative one, however precious, Ilence if Brittonic shows no evidence for a t;rpe already represented in the more archaic Goidelic only feebly and that by mere fossilized remnants not structurally organized, this may safely be attributed to chance and the general Brittonic elimination of irregular and isolated archaisms. It is most exped.ient to group these preterital types according to their origin in formal terms. These five Common Celtic preterital categories, with the addition of a sixth morphological type, the s-subjunctive, fall into three distinct, rubrics: (l) sigmatic formations : s-subjunctive, f-preterite, and s-preterite, all historically s-aorists. (2) reduplicated and d-formations : historically fects, with and without reduplication. (3) isolated historical root and thematic aorists.

per-

ll0

cDLTrc

These three types go back to distinctive Indo-European types ; in other words, they are, within the Celtic family of languages as a whole, inherited categories. It is the first of these groups, the sigmatic aorist and the precise details of its history, with which Part Two of tho present study is concerned. The remaining two, as well as the categories discussed above which belong to the Primitivo Irish period, I expect to treat in a subsequent study. S 10.

Srnrcn

roB A x.uNorroNAl rrrsronlr ox, TEE Cnr,rrc vnnn

Irv the Note liminaire to his Altophonie (24 tr.), Le systdme verbal de l'indo-europ6en, Kurfowicz presents a modol designed to illustrato the interrelations of theoretical generic catesories in the verb. ft is this: I (r6f6rence au moment II (r6f6rence ir, un de parler) moment du pass6) l1

Tt lz This is, in essence, a three-dimensional figure : the symbols of each quadrant correspond to distinctions on the a,spectual plane ; I and II correspond to the two terms of the temporal plane. The aspectual states which he sets up are : imperfective (B), perfective (p), 'I'etat r6sultant d'une action accomplie' (7), and 'le pr6sent (temps) g6n6ral' (f). Kurylowicz apparently envisioned this as a purely theoretical model, superimposable on any language,l 1 I havo criticized this proceduro at somo length in a reviow of L'Apophoni,e, Zg. 34.381-98 (1958), to which the reador is reforred. It is also possible to enlargo tho system by a third quadrant referring to the future, aE in Y.V. Ivanov and V.N. Toporov, Sanskrit g8-9 (Moskva, f 960). To this extension the sa,mo objections can be rais€d,

FUNCTIONAL

HISTORY

OX' TEE

VNB,B

tIl

and not as a descriptive model dictated by the exigencies of any given language system. Yet the fact remains that the nature of the aspectual correlations envisaged by Kurlowicz shows that his schema was modeled on the verbal categories of traditional Indo-European as reconstructed from Homerio Greek and Vedic Sanskrit, and what is most significant, not only for the functional correlations of such a languago system, but also for the forms. What we havo in Ifomer and the Vedas leads us to reconstruct for traditional fnd.o_ European precisely such a quadrant as Kurylowicz,s schema I, where each functional structure point has a corresponding formal category. Thus : fl IE (athematic) present, p, fE aorist, 7, IE perfect, Br IE iterative-durative suffi.xed present. X'or this reasorr it is justifiable and indeed fruitful to regard this schema as applicable to fndo-European not merely as a theoretical model but as a descriptive system. A brief inspection of the verbal categories of the earlier Indo-European languages shows that the second parameter of Kurylowicz's schem&, the 'r6f6rence dL un moment du pass6', is only partially complete, in other word.s, that there is not the unitary correspondence of formal oategory to functional structure point which we can observe in schema I, the 'r6f6rence au moment de parler'. The only two points formally expressed. in Greek or Vedic are B, and yr, imperfect and pluperfect, and of these two the pluperfect has long been recognized as an innovation of the individual languages. This leaves for late Indo-European proper only a single formal class (Br) in the whole second quadrant-an unrealistic situation. It is furthermore no coincidence that the imperfect (Br) is the only case in tho system which is formally characterized by the d.esinences alone : the opposition Br : B, is marked only by tho difference of primary and secondary endings, B sg. -til-t. Now on the evidence of such archaic categories as tho Vedic injunctive, which shows only secondary endings, or the Vedic subjunctive, which shows both sets, this distinction did not originally conyey a temporal opposition (present/past) but more likely one of deixis ; the form

rt2 with -d insisted on the hic et nunc, whereas the simple form did not.z A transformation of the optional use of deictic -i rvith the present to the obligatory use of this morph with the functional present is readily understandable, and such a transformation would automatically imply a repartition -(t)i present/-(f) non-present (past). Kurylowicz himself has made what amounts to a tentative suggestion that this distinction of enfings originally reflected the opposition B. : l, (Apophonie 32) ; if this is accepted, then it would still follow as a corollary that this formal expression of the opposition B, : B, must be posterior to that of B, : lr. This suggestion of Kurylowicz's is supported further if we assume that the Vedic injunctive with its secondary endings continues the form proper to the undifferentiated. f1 . As Thurneysen recognized in 1883 (KZ 27.174), tho injunctive represents the oldest layer in the Indo-European verbal system. It does not partake of any temporal oppositions, and the temporal value of present, past or future is dependent upon the context. Looking at the extraordinary functional disequilibrium of the system now usually recognized for Indo-European with the imperfect included, it is difficult to avoid the conclusion that Indo-European proper had originally only a single quadrant of verbal oppositions, and that 'r6f6rence I un moment du Kurylowicz's quadrant, the pass6', is a secondary development. The imperfect, insofar as it is purely a past of the durative-'imperfective' Br, must be a later accretion to the system, as it implies a thoroughgoing functional opposition in a purely temporal correlation, for which there is no other evidence in the system. Originally the temporal past must have been simply a 'variante s6mantique combinatoire' of y1, the perfective 'general form, the Indo-European aorist, and of fr, the present '/injunctive. 'Linguistiquo 2 Cf. also A. Ma,rtinet, structu,rals et gra,mmairo (1956), compar6e', (lraaau,a d,e l"Institut d'a Li'ngwi'sti,qre l.f-2f.

X'UNCTIONAI]

EISTORY

OX' TITE VERB

ll3

Viewed in this light the process of the historical development of the second parameter or quadrant of temporal remoteness-which has become the basic correlation already in the Hittite verbal system, for example-becomes an interesting and significant problem. The directions of development were fundamentally different in different dialect groups ; but it is clear that the creation of the additional parameter, the second quadrant, was not accomplished in a unitary way in the period of common fndo-European. fn some cases it is possible to see how this development has come about. In Greek and Indo-franian the so-called augment tenses of the past, fhe imperfect corresponding to Br, and the aorist corresponding to f, and B, (earlier only to pr) arose out of the use of injunctive ' forms (with secondary ending) of the present and aorist s stems in the function of a narrative tense. This becomes evident when we consider the origin of the augment d-. X'ormally, the augment corresponds exactly to the Luvian sentence connective a- ( < *e) ; its primary and original function is that of sentence connective in consecutive narration. Tunctionally, the augment is equatablo with OIr. no; it is used in combination with secondary endings, OIr. impf. no-bereil : t-6rpe. We know that functionally, Luv. a- is identical wibh H:itt. nu, and that formally, IJ:itt. nw is identical with OIr. no ( < *nu). We may set up a square as follows including the 3 Ivanov and Toporov, Sanslcri,t 115, suggest that tho injunctivo ie tho predecessor of the entire indicative systom, present, imperfect, and aorrst,,a^rrd aorist, They they a,re surely right in maintaining tlie original the orieinal absoncs i,bsoncs tcmporal of oppositione in fhe verb; bu6 the existonco ol the perfoct implios at least ono aspectual correlation in the system, and I liositats to assumo tha,t present and aorist gtem wore originally ono and tho sa,mo. n'or this roason I assumo that a corrolation of perfoctivenoss did orist in at least gomo formg in the system as far back &s wo can reconstruct. The present of aspoctually neutra,lizod injunctivos doos orclude aspoctually not orclude aspoctuallv ma,rked iniunctives. injunctives, Verlic thcre and in Vodic there anc aro approximatoly as many injunctives from tho present stem as from tho aoriet stem. Cf. L. Renou, 'Les fonnes dites d'injonctif dans ls Rgvoda', Etrennes Benaeni,ste 65 (Faris, lg28),

7t4 augment, idontities:

CELTI(

illustrating TE C

tl

OIr. no

both

the

TUNOTIONAI,

formal Lt: IIilt.

and

functional

.a nua.

Old Irish no reflecLs both the primary funtion of sentence connective and the secondary function of accessory mark of the narrative tense. The first is continued in the use of no es an empty preverb to infix pronouns ; originally the sentenee began wilh nu, followed by enclitic object, and tho verb came at the end of the clause. This structure is still faitMully mirored in such Old Irish sentences w"ith otmesis' ausno-nl, choimmd,'i,u-cotma'the Lord cherishes me', which can be directly compared with a Hittite sentence such as nnt-nlu d-L{tar GASAN- ru kane*Ean harta ' my lady Ishtar But the function of sentence connective cherished me'. was lost with the fusion of verb and preverb in Celtic, and nn was simply subordinated to the other preverbs ail, com, etc. In the secondary function, OIr. no appears with the a These are not, the only gentence connectives of Indo-European, rror tho only evidenco fot IE *e, *nu. The latter is preserved in one form in Latin : nud,i,us tertius 'day beforo yesterday'. This adverb is actually a nominal sentenco finu dius tert'iusS 'and it is the third day', as astutely recognizod by J. B. IIofmann, IE 42.77 (1924), It is directly comparable to a llittite nominal sentence suc}n a,s finw 'and it is the tenth ITU l0 KAJ\[# month'. Othor sontenco connectives in fndo-Europoan a,ro *lo, *so, which .a,ppear in Archaic JJ.ittite ta,.irz. Both recur in Old Irish : lo (functionally in complementa,ry distribution wil}: no) ; se in se-ch a's 'and it is' : no-ah i,s'id'. Both lo and sa function cornrnonly as sentonce .corrnectives in Vedic, particularly in the Brahmanas. The form *so is preserved in Myconean Greek as a connectivo, in the form ho; it is the hitherto obscure o (yb) prefixed to verbs and othor words, at aentence initial position, e.g., ho-agrise ' and he took ' (Ventris-Chadwick 176). The conlectivo value for ho is proved by tho negation of the vorb in a parallel passage in tho sa.me tablet, : ow-lctne agr-ese ' ar;Ld he did not take ', when -lcwa clearly has tho samo content, as ho. Aside from Mycenean, fto appears as a component of tho domonstrativo oizos (Boeot. oijra, o9rov) I #ho-u-to-s. Originally only tho -ro wes the pronomi-nal stem ; ft,o is a corrnective, occupying initial position, and ea is an enclitic particlo regularly in second position, as with Ved. w, Golh. u. These conloctives a,nd rolated syntactic problems in fndoEuropean will be taken up in detail in a papor to appear in Celti,aa, vol 6 (1962).

EISTOBY

O.[' TED

YXIRB

It6

Old lrish seconda,ry tenses : imperfect, subj. II, fut. II. fn this secondary function, OIt. no is identical with the augment @occurring in Greek, Armenian, and Indo-Iranian, where it appears with the tenses with seconda,ry endings : imperfect, aorist, pluperfect. In the latter three languages, tho augment is always directly prefixed to the verbal form ; it cannot be in tmesis. But the situation in Luvian, whero @is separated from the verb, shows that this is a secondary development, and the same can doubtless bo inferred from the unaugmented forms in llomer and the Ved.a, where it is likely that the separated ' augment' C, originally a sentonce connective in initial position in the sentence, has been replaced by some other connective or copulative (e.g., Gk.3t, ze). It is the'univerbation' of 6 and injunctivo forms of the present stem (i.e., with 'seconda,ry' endings) which is responsible for the creation of the formal' imperfect , in Greek, Indo-Iranian, and Armenian: 1Sepe: d,bharat : laor.) eber. But it is only in Greek and Indo-Iranian (and perhaps independently in each) that the formal opposition of such forms built on the present stem with similar forms built on an aorist stem was completely carried out, with the development of a thoroughgoing functional opposition of imperfect and aorist of the formal type ZLe,,zreli),rne, drecatldr,icat The form eber in Armenian was at the outset simply a preterite, opposed to the present berd < *bhereti. It is only with the creation of ber\r (1 *bheretor?6) lhat Armenian developed a functional contrast imperfect/aorist (bererleber), collectively opposed to tho present, just as in Greek and fndo-frania,n. The imperfect of Greek and Indo-Iranian, as in €$epe,dbharaf, is an innovation from the Indo-European point of view, just as much a,s Armenian berEr, OCS nesd-ard,, or Lat. legd-bam. The imperfect *6-bher-et is not an Indo-European form. In Indo-Europea,n there were in most cases differences of formal expression for each of these functional structure 5 Meillet,

Esquisaez

I27.

116

cnr,rrc

points, particularly of the general present and the aorist. We would expect the same in any natural language. But the functional position, the relation to other elements of the system, remained the same. Celtic, like all other IndoEuropean languages, sho$'s evidence for this formal diversity, particularly in the formation of the present ; but by and large, as is to be expected in a group of languages attested. only to any large extent in the Middle Ages, such variants were largely eliminated ; the tendency was to create a uniform derivational type. In the strong (non-derivative) verb, the only case whore we may properly speak of inherited forms, the present still exhibits numerous divergent formations (suffixes -yo/d-, -na-, -nu,-, nasal infix), but the basic type is the single thematio formation -elo-, e.g., beri,il, < *bhereti,. The Celtic forms filling the other three fndo-European functional structuro points are essentially unvarying. The iterative, relatively rare, is the familiar type *l,oulc-eye-,with root vocalism o and Celtic suffix -yol|-.6 The perfect continues the Indo-European perfect, though with different distribution of the reduplicating syllablo, and a later and independent development of a long-vowel perfect fdT-. The reduplicated perfect and the d-perfect were originally in complementary distribution according to the structure of root, as is still essentially the case in Old frish: roots with structure TeT- (i.e., with root, vowel e) formed. the d-petfect,, and possibly roots with structure TeR- as well (subsequently eliminated in favour of tho sigmatic aorist ) f-preterite). The d-perfect from the beginning lacked the reduplicating syllable; in its distribution the situation strongly recalls that of Germanic. The origin of the -d_ is to be sought in a new lengthened grade (a + d) of the rnorphological zero grade in a (TaT) of the perfect, after the abolition of paradigmatic ablaut. Typologically it o tr'or tho gum. n . (Paris,ropr. 1950).

X'UNCTIONAL

EISTOR,Y

OF TEE

VER,B

II7

corresponds thus to the preterite plural of the Germanic fifth class, e.g., Goth. gebum, which also shows a morphological lengthened grade, though the detail is obscure.T The aorist, in conformity with the trend. of development in numerous fndo-European languages analysed by Meillet and diseussed in Part One, is almost universally sigmatic. There are only a handful of Celtic forms derived from IndoEuropean aorists which do not continue the sigmatic type ; such are Oh. d,o-cer'he fell' < *kerH-d (athematic se! aorist) ; Luid,'he went' q *luil,het,perhaps buig'he broke' 8 (thematic aorists) ; ro-kt, suppletive perfective present to -cuirethmr 'puts', the original root aorist of the structure ReH- to the nasal infi.x present *!-na- (sbructure R-n-eH-) remade in ail-ella 'visits'.s But these verbs are residual , fossil forms, and have no place in the living formal organiza. tion of the verbal system. From the fndo-European situation as described, then, Common Celtic inherited an identical functional alignment of categories ; formally the configuration was as follows, for the basic type:

(thematic) present \. ,/ ,/\ ,/\ matic raorist o-grade iterative terative sigmatic \ /\ t\

,/'

reduplicated or long-vowel perfect. ? For a similar caso of morphological longthoned grado in -zi-, cf. the Latin typos stig-i,o, com-tdg-ium,, cld.d,-?s, disclotsed in $ 3 above. 8 Tho preterite -bui'g may also be an s-aorist to a,n itorative'breaks ' causative *buigid, besida bongi'd' ; cf. to-luigi' beside fo-l,oing 'supports', da,ns los langues and M.-L. Sjoostodt, Les it6ratifs-causatifs Ety'rnology in celtiques, MCl,a'nges Vend'rges 323-40 (Paris, lS25). fact supports this analysis ; bongi'il is cognato with Arm. belcanem,, *bheg-, to which an itorativo would bo in Celtio Sl -E-, -6- (some forms analogical): tf,agu 'I go' teg-scon-rig 'binds' rEg-s. nigi,d, 'washes' nAg-ssnigid, 'drips' sneg-sro-f,tir 'knows ' wEd,-sad,-fdt-'relates' wdd,-silo-i,n-fdt' inspires o swEzil-s'fights' wd.lc-sfi,i,chid, 'oppresses' dd,g-sfor-d,ing ci,ngiil 'goes ' lc4g-sIi,ngid, 'leaps ' 6 Idg-soid,id, 'lends ' oil-sbongi,il 'breaks' bog-ttongid, 'swears' tog-s'supports' log-sfo-loi,ng as-boi,nil'refuses ' bdd,-sroind,id, 'reddens' r6d,-s(ibid 'drinks') ldg-s-

6 It is structurally not, consistent verbs in -nn, aosrn OIGr. 389. It

kerd,-smelg-s-

to

class this

form

with

the

130

:

,S.SIIBJIINCTIVS

O,DLTIC

Roots in -rcz: selsssEssgl4ssgressbrEsstess-

'pursues' ' d,o-gexnn o sennid, plays' 7 'learns' fo-gleim d,o-greinn'pursues'

d,o-bruinn' offervesces' tenn'id,'iuts' 8 To these we may add one root in -s: g6sstlo-goa 'chooses'

Senn-sSwenn-sglenn-sgrenn-sbrenn-stenn-sgdss-s-.

In the Brittonic languages we have only three roots which preserve the s-subjunctive : MW d,uch'diicat' < xileulc-s-; 7 Thurneysen (OIGr, 353) is surely correct, in taking these two distinct, against Pedersen, VKG 2.625, Tho form as etymologically is from IE *sutenH-: Lat,. senn'iil ' plays a musical instnrment' sondre. RY tisoamit, The association with musical instrumonts is in Latin as old a,s Ennius, aes sop'it (Scaen. l8I Vahlon); in India it is already in the Rig Yeda z god,ful,pdri, sami,gaa,ryat'the gut-string sounds' (8,69.9). Compare also f,ilEs ' string, stringed instrument' with sondre 'play' in Ofr. sennid 'was !. Propertius 4.7.62. The transitivo use doveloped in Itialian suomdre; compare the mwical independently 'pursuos ', wit'h simFlex senn'id,, hae terrm sonata^ Tho verb d,o-se'inn 'plays ', though it has fallen semantically nothing to do with sennid, together with it in form, by adopting from the latter its lenited initial danhzi,, form /-. This verb is to bo compared with IE +senH- irt }Iitt. 'seeks; plans; demands'Yod. sanoti' wins, gets; gfants',Gk. &lia '. 'accomplish The apparent variation ; get (something, somewhero) in moaning among all those forms is actually gra,rnmatical in nature, tho sarne action can be viewod oither as rafher than lexical; 'Beok, pursue ', or a€ completed : 'got, accomplish'. uncompleted, 'mado for'lro-si,aaht The situation is commonplaco : cf. OIr.. siacht 'reached', or Russ. redal; zad,aCu'work on a problem'lre{i,b' zadalu 'strikes' 'golve a problom'. given in t}re RIA Contribb, The moaning condifionod s.v. 2 se'inm,id,represonts sirnply a spocialized application, 'so that he struck his hair from by the context ': co seba'ind' a Jolt d'e him' (LU 5678) and sirnilar examples, all from the T6,in. This paralwa,ys with preposition de or ass, is paralleled ticular application, 'seek' also has the spocialized meaning whero Sanhin Hittite, ', ' cleanse, wipe particularly with tho prevoib arha ' off i. Such aro corr)mon with verbs of this somantic specializations diverqent sort fthe cognate &"vii rn colloquial Attic Greek (Aristophanos passdm) '. The basic meaning in Old Irish of the simplex mea,ns I hurry up 'obscure ' in the senn'iil, appeats in tho sories.of examples labelled 'got (something, somowhero) ' as ia RIA Contribb., where a value 'ho (}k. d.v{u is indicated : roilom-s'ibaea sech riaga will get, me pasf RC 20.414 (Amra Choluim Chillef ; senna'it rodt ruadtiessa torments' 'they go a road of fiorco battles'. Tho latter example, with "road' object, is exactly paralleled by uses of f,he cognate &rr1o as 'in;er' 'way, road ' in Groek with xlleudos ; zrol)thv xll 'plays'). *stodss- (senn'id, This resulted in some more instances of an apophonic contrast present o/subjunctive d, some (the roots in -nn) of present e/subjunctive e-, and 'comes' one of present c/subjunctive i, : i,ss- to -'icc < *irs- 1 *ink-s- (cf. ri,chtu I *rirt- *r-ink-t-, OIGr. 128). The chronologically later development of Common Celtic eu > ou > o-, completed well after the separation of Goidelic and Brittonic, probably after the beginning of the Christian Tho era,, created a parallel apophonic situation -u-l-o-. effect of this was to create a,n analogical prossure towards long vowel s-formations ; as a result, th.e TET- formations s Lewis-Pedersen 287. gutntl }nas beon dissimilated 'lot us make'.

The initial of gtunech (l sg, pres. ind, from gwr-, still attested in OW gwrdgurl

r32

,S-SUBJIINCTIYE

CEI,TIC

with root vowel -a-, originally outside the apophonic situation even in those fndo-European dialects with long vowel sigmatic aorists, lengthened that -a- in the sigmatic form : sog-s --> sdg-s-, to conform to the apparent structure -splus accessory vowel lengthening, thus filling in the pattern of sigmatic forms : TEss-, Tdss-, Td,ss-. This, in turn, opened the possibility u --> d, (ld,ss- : lo'i,mm' draught ' with original short zr), an occasional formation. But this development was never extended to the group of roots in e-vowel and final stop, TeT-, which preserved their characteristic fndo-European shape. The sigmatic aorist was inherited from Indo-European with a short vowel : the development of all the other long vowels appearing in the Old Irish s-subjunctive is later, and probably completed only quite late in Proto-Goidelic. As regards the distribution of the suffix -s-, its appearance with roots in final dental or velar is cloar. The roots in -nn constitute only an apparent exception to this, X'or in these cases -nn- results from an assimilation of an earlier -ndn-. Typical for the formation is -glerut-: OCS gl$,ajq gleila,ti, 'look'. The original form of the lrish verb is *glenil-nelo-, with suffixation of a present, forming -nelo-. Historically this is merely a repetition of the -n- of *glend-, originally a nasal infix present. The type is that common in Greek 'I leave') and Armenian (e.g., (e.g., )ry,zrd.v -ps- -> -ks- [-os-] must have produced a thoroughly anomalous situation: aorist, garspresent gab'i,rs-. ibIn view of its typological rarity this was inherently unstable, In the case of and liable to analogical transformation. take place, and did not aorist these two verbs the shift from -dretained in was (to tools!) TETthe old optative in orrsted of being instead function, the subjunctive modal -eba.75 -gaba. inherited These are from the system : OIr. 15 More commonly

the subjunctive

of i,bi'd' is the suppletive

I'zZss'.

I36

CDLTIC

.forms, on the parallel evidence of Osc kahad,, Lat. bi,bat. The sigmatic aorist forms remained. in the preterital function, krut in renewed form ; *gars-,*,irs- wete replaced by *gab-oss_, !ib-ess-, with the ney, productive preterite morph whose origin we shall discuss in the chapter devoted. to the s-preterite. This essentially nonce substitution of -ass_/_eos. for -s-, with the selection of the two doubtless conditioned by vowel harmony, had the advantage of preserving the identity of the root form, and. thus achieving a linguistically stable situation: present gab-goli-, ib-elo- preterite gab_ass-, 'ib-ess-. . The examination of these few roots in final labial, .together with the roots in original final -s such as *geus-, *zues-,shows that tho process s-aorist -> s-subjunctive was originally proper to essentially all roots of the structure fET-, TRET-, or TERT-. Originally all of these roots formed an d,-optative (later d-subjunctive) ; where the original e-aorist assumed the subjunctive function, the .form in -d- simply disappeared from the system. The only case of hesitation was those roots-limited. in number-ending in -.s, where both the -s-subjunctive and the -d-subjunctive are found. The tendency to use of the -d- subjunctive in these cases is probably not very old in frish, and in all tikelihood conditioned by the loss of intervocalic -s-. X'or this resulted in the creation in the present of a hiatus root, Thurneysen's class A fII: cf. *wos-eti ) disyllabic fdaid,. Now this class of hiatus verbs, sufficiently unified in Old frish to warrant their being treated as a single group, is of diverse origin. The hiatus in most cases results from the phonological loss of a consonant in intervocalic position before the verbal desinences (which all begin with a vowel). This lost consonant may have been either the final consonant ofthe root, or the initial consonant of a present forming suffix. The consonants subject to loss are fE -?-, -s-, -U-, -w-. The loss of -p- is a much earlier development, Common Celtic in date, and there are actually no hiatus verbs attested in Old frish which appear to go back to an Indo-European root, in final -1r : ind.irect evidence

S.SUBJUNCTIVE

137

for their original existence is ihe adjective tee td, ' hob', n. pl. disyllabic tei,t, from a participle *telglent- : Skt. tapont-'hot.', to the verbal root, of Skt. tapati, Lat. tepe\. The great majority of hiatus verbs in Old Irish result from the loss of intervocalic -s-, -w-, and -y-. Of these three the loss of gr is found both in intervocalic and in absolute initial position, and its loss is probably much earlier than that of the other two, as Thurneysen has suggested (OIGr. l2l-2); ihe situation is quite analogous to the treatment of *rn and *y in Greek. fntervocalic -s- and -w- were lost at a much later date, and were preserved in initial position. Now the number of hiatus verbs from roots in original -s is small compared to the number with original -u or -y : lhe predominance is of the order of 4 or 5 to l. The roots in *s are of the structure IET-, hence we expect the formation of an s-subjunctive TET-s- ; but roots in -w or -y are of the structure TER-, and for these the s-subjunctive is not expected, but rather the conservation of the d-subjunctive fER-d-, as we have seen above. Before the loss of -s- and -w-, the situation was stable:

present TUs-elopresent TEw-elo-

subjunctive TEs-ssubjunctive TEw-d,-.

The subjunctive was structurally determined by the present, in Kurylowicz's sense. But after the loss of intervocalic -8-, -w-, this relation became a source of ambiguity, since the presents TUs-elo- and TUw-elo- had coalesced into the hiatus type TU-elo-. Phonological loss of -,r.u-yielded. in the subjunctive f E-d-, whereas in TEs-s- the geminate I was simplified and preserved as a new phoneme /s/. Ilence the new situation appeared in structural terms as follows :

present r*-eto{ xirxnzTr$? The type TE-d,- was statistically prevalent in the hiatus verbs," as well as structurally more in accord with the hiatus type ; the d-subjunctive was furthermore the productive subjunctive formation in Old frish, that employed with the weak a- and f-verbs. The forms TZ-s- in hiatus verbs,

138

CNLTIC

on the other hand, were rare and residual, for in the normal situation of Old frish a subjunctive 7.&-s- presupposed a present fXlT-, not TE-. Hence it is predictable that the ?.&'-s- subjunctive of hiatus verbs would tend to be forced out of existence, and replaced by the productive d-subjunctive. We have seen that gdss-to -goo ' chooses' is the only such for.m continuing in Old Irish. In general the s-subjunctive simply gradually disappeared, and the verb formed a prod.uctive d-subjunctive. 'sees ' (active), with deponent a-subjunctive Compare ail-ci I sg. ad,aeor, 3 sg. -acwil,ar. In the case of this verb we have two indications of the previous presence of an The first is the subjunctive passive form s-subjunctive. -acca,star,where the occasional indicative use in the older language (twice in Wiirzburg, cf. OIGr.386) would indicate that it was already on the way toward being ousted from the subjunctive system. The second and. conclusive indication is the preservation of the hapax active 2 sg. subjunctive form for-a'i'c'i,s (from for-ad'-cl,) in the very archaic law text il,in Techtugud,,AL 4.I8, 21. This form is a venorable archaism both because of the conservation of the s-subjunctive form, and because of the proservation of active inflexion ; in itself it constitutes additional .evidence for the great antiquity of this legal text which its heptasyllabic verse would suggest. We have examined the history of the Old Irish s-subjunctive as a category ; its origin in the Common Celtio s-aorist, and its subsequent ousting from the aspectotemporal plane toward a modal acceptation, in the case of roots of tho structure TET-. There remains to be considered the question of the constitution of the paradigm itself, the formation of the desinences. The Indo-European sigmatic aorist was athematic ; the desinences were added directly to the morpheme -s-. The original paradigm was in late Indo-European, with secondary endings alone : sg. -s-fn, -s-s, -s-f, pl. -s-me, -s-te, -s-qt't (-s-elont). The exact form of each ending is not, important ; only the fact that they were athematic.

S.SUBJUNCTTVE

139

Two forms speak for the fact that Celtic inherited an athematic paradigm from Indo-European. The first is the 3 sg. of the s-subjunctive, which in Old frish must be reconstructed to a proximate athematic form : 3 sg. abs. tdis < *(s)tei,gh-s-ti,,3 sg. cjt,. -td, I *(s)tei,gh-s-f,or 3 sg. abs. re'i,s < *ret-s-t'i, 3 sg. cjt. -rd < *ret-s-t. There is an additional testimony of the athematic The forms have character of the s-aorist/subjunctive. independent value, being in the Old Irish period divorced from the paradigm of the s-subjunctive. They are the six originally 2 sg. s-subjunctive forms which function as 2 'arise ! ' sg. present imperatives : (OIGr. 375) at-r6, < 'in'spare ' *reg-s-s; no-m-ain me ! < *aneg-s-s; a'icc v o k e ! ' < * a , i l , - g e d ' - s -t so;g ' c h o o s e l ' I * t o - g d s - s - s t; a i r 'hulp ! ' (MS -f6ir) ' come ! ' a < *to-are-inlc-s-s; *foi'r *wo-ret-s-s. These forms can only be explained as containing athematic 2 sg. -s. And since they function as imperatives we may assume that they were taken over after the transition from aorist to subjunctive, in other words, 'jussives'. that they are by origin The original athematic paradigm of the s-aorist was preserved only in the 3 sg.; elsewhere in the paradigm the athematic endings were replaced by the corresponding thematic endings. We have seen in chapters $ I and $ I0 that the Indo-European opposition of primary/secondary endings was utilized in Celtic solely on the basis of a mechanical repartition between non-compounded (absolute) and compounded (conjunct) forms of the same verb. The desinences themselves had no function as temporal markers, and it was only late in Celtic prehistory, after the separation of Brittonic and Goidelic, that a set of'secondary' endings with a temporal ('remotive') function was developed. The 3 sg. s-aorist has absolute/coniunct -s-til-s-f just as the 3 sg. present has themat'ic -etil-et. The sole difference is that the present, with its thematic vowel, has innovated Where the aorist was vis-i-vis the athematic s-aorist. not formed with -s-, as in the forn ro-la discussed in the preceding section, the endings of present and aorist were

140

,s-suBJuNoTrvE

identical, *ld-til-ld,-t, which doubtless facfitated the ahilrzation of this form, etymologically an aorist, as a perfective present : n6-ro-lui,m, OIGr, 470, and cf, also the presents eon-|,a, ,i,n-la. It is then hardly surprising that we should find the same identity in the. other persons of the subjunctive paradigm, from an s-aorist, with the other persons of the present indicative. We may compare the paradigms of the eonjunct of the present of. berid, and the subjunctive of guid,id,: (OIGr. 360, 390) -gess (-lius with z-colour marked) sg. I -b,i,ur -ge,iss 2 -bir -gd 3 -beir -gessa,nx pl. | -beram -gessid, 2 -berid, -gessat. 3 -berat The paradigms are identical, save for the athematic/thematic difference in the 3 sg. We may reconstruct them in Common Celtic as follows: ber-d, ber-i,L6 ber-et ber-omo ber-ete ber-ont

ged,-s-d, ged,-s-i, ged,-s-t geil,-s-omo ged,-s-ete geil-s-ont,

The mark of the aorist (and later subjunctive) is the -s- ; the desinences themsolves in Common Celtic did not contribute to this distinction. Ilence given the situation of identity of -til-t in the 3 sg. of both tenses (present and aorist) there was no functional reason to stand in the wa5l of the replacement of the athematic endings of the s-aorist outside the 3 sg. by the thematic endings of the present. n'or the reason why this did not take place in the 3 sg. of the aorist, see further below. The parallelism is even closer when we consider that in the present itself, there existed a number of athematic back to *-e'i' and'is cosnatowith Gk' -er(s)' Lttff ;T:?"2u."s;"::.sou"

l4l

formations inherited from fndo-European, which were renewed by the sa,me process of thematization within the Common Celtic period. The thematization of the s-aorist outside the 3 sg. was in all likelihood contemporaneous with, and simply one aspect of the general process of thematization in the verbal system as a whole, the replacement and renewal of older athematic forms by newer thematic forms. In a linguistic system where the tenses and aspects were marked formally either by root apophony or suffixation or both, it is hardly surprising that the renewed form of the desinences would be the same in the several categories. We may examine the evidence for these formerly existing Common Celtic athematic presents. Old frish, like many other Indo-European languages, preserves athematic flexion only in the verb 'to be ' in historical times. But there ie good evidence that athematic presents did exist in Celtic in earlier times-which in general we should expect. Eric P. Hamp, is his study of ' Ofr. gaib-, Welsh gafaet, caffael, cael, cahel' (ZCP 24.229-33 [f954]), has made it very likely that the proliferation of forms in Welsh results from a crossing of three separate athematically conjugated roots. in Common Celtic, *gab-. (OIt. gai,bid,, gab-yoli-), *kalt(Lat,. capio), and *lcag- (Osc. kahad,). We have also a more interesting series, showing a certain paradigmatic unity of shared morphological features, which warrant their being treated as a group, and justify the hypothesis of unitary typological origin. There are six of these verbs : fi,chid, ' fights' , ni,giil ' washes ', l,igid,' licks ,, s n i g i d , ' d r i p s ' , m l i , g i d , ' m i l k s ' , a n d d , l i g i d , 'i s e n t i t l e d t o , . They share the interesting formal characteristic of being presents formed on the zero grade of the Indo-European root, with. thematic conjugation, as follows (respectively) : *wih-elo- (*weik-), *nig-elo- (*nei,g-), *lig-elo- (*teigh-), * snig-e o- (* sneigwh-), * mlg-e o- (*melg-), *d,!g-e o- (*d,etgh-). f I I Mqillet, with his characteristic aclrtity, was the first to note that lhe zero grade vocalism of some of these forms is likely to reflect an original athematic present. He treated the Celtic forms mlig,id, and ni,gid,in two articles of the same

t42

S-SUBJUNCTIYN

year : 'Les formes verbales de l'indo-europ6en *nnel,g(MSL 17.60-4 [fgfl]), and 'De quelques for"traire"' mations de pr6sents en indo-europ6en' (MSL 17.193-T [I9fl]), where he demonstrates that both continue IndoEuropean athematic presents. His suggestion is strengthened by the rule we can formulate about the treatment in Celtic of paradigmatic apophony, of which the athematic present is in fndo-European the example par excellence : that Celtic universally generalized the zero-grade. There are no exceptions to this anywhere in the Celtic verbal system ; the apparent reflexion of the Indo-European o-grade (proper only to the singular !) commonly assumed in the reduplicated perfect of the type ci,ngid,: cechairryis, as I shall demonstrate X'or two other verbs in in a subsequent study, illusory. our list original athematic conjugation in the present is 'licks ' beside historically attested : cf. Vedic reQhi Ofr. 'it is snowing' beside OIr. sni,giil li,giil, and Ol,ith. sniegti The comparative evidence, in conjunction with the principle of the Celtic generalization of the zero grade in paradigmatic apophony, entitles us to assume that in an original athematic paradigm as follows: weilc-m'i' weilc-si wei,lc-ti

wik-mo wilc-te wilc-onti,

the zero grade *w'ik- w&s generalized throughout the paradigm. Subsequently the paradigm was renewed by the qeplacement of athematic endings by thematic endings, yielding thus the immediate prototype of OIr. fi,chi,il, ( < *wi,h-eti nppc- prra 'e.rnlcunf r4p,m -Icl+?cl4 ] -pn7Ecac'rIO ueo.'K+oq+s€rluoc Eurlec or1+uT quoredde sT slq+ i suolleurro; qcns ur e1qe11-{s -gdnpo.r oqg Eurpunoclurocet 5o por.rod e q8noeqq lergred Jo r'Io.{to^ Fulpecerd orl+ Jo 1ue.u dlguepr^o qsrrl o^rlprlrd

6tT

fiariloNnfs-os

cr.\toJJrufl

SurueqlEuel d.rolesueduroc q?rar lsol sr -l- oroJoq -rl- I€pour pue '-'lI- se s.reoddu -ls-x Itsrpour qsrrl ur ecuts '-pn7c1at1* Jo -pnqxs* flJ +cogor d1qco.r1p+ouu€c ', sEIp , pxpxblc rJror; -pnlEcoc '1o.rd .ro ( so>lrr+s p?p?n1slruorJ svyas- '+nJ s€ r{cns , 'ern?nJ pu€ o?rra+ord pelecqdnper oq+ u1 o1qe11.{s rrrroJ y Furlecqdnper er{1 ra}Jts rlslrl pIO ur +uetu?€or+ uor+[uel relnEo:r.rroq+ .{q po,r.o.rdsr 'o1ep o1e1d1e,rr1r1oJts1B 'qro.,l oq1 ur ,{.rrapunoqeruoqd.rour +€ €uoruoueqd e.rnlcunf lucr8oloqd.rour pedole,rop .{lluepuodepur se8en8uel cI+IoC eq+ +€q;. 'o,rn1cun[ +noq+Ind-6os-* se po+cu+suocer oq u€c purs '4ueuosuoc Eurpece:rd€ Jo ( uorlco,lo.rd , go oct?J+ou 'uosrred sd.oqs (ttpo)bca- xlg.ns e^r?€uruouop qslrl pIO eq1 'o.rn1cun[ '-6nsJ--* -ruoc ,{g Eurpoce.rd qry,* xsns eq? sr '.onco luonEulal '+eJgO ', Jo ruJoJ cruo++rJfl orl[ +u€no1, , gufiaEuruncs 119 : -6ns-* cruol+rrg orl+ uT srncor luoru+Bor+ lecrFolouoqd otut?soqJ 'uoI+ISu€r+u1 ue1s.{s € qcns uor+cogoreq} eq +sntu cruo++rrg Jo ur soxgrns elrlepedns pue elrlenbo oq+ eroJoq euoruoueqd 'QtZ 't7rc) oenlcunl 3:o eouu"reedde eq; qslrl pIO ur uor+coBur Jo ocuosq€ oq+ puB uor+cnr+suoc o^r+Burruou d.role8qqo aqq ''3'e 'uosr;redruocJo surroJ orl+ Jo serlr.rerlncod Ir?uor+cogur put cr+c€+u.{s eq1 dq pep.roge sr uorsnlcuoc our€s oq+ 2 n- 't1g oq pesoddo sE ,qcn- xsns o^r?€redmoc r{slo^,1 elgqcnpo.rd oq+ dq osIB u,r4.oqs sr se 'suor1€aouur quepuedopur Jo roqunu € uro$ s+lnsor se8ranEuul crq1e3 I€ro^os oq+ q errrlcofpe or{+ Jo uosr.reduroc or{+ Jo tuolsr{s 'uor+€aouur cruo++rrfl e d1.ree1csI 'u1Fpo orr+ue eqJ +nq ur orncsqo sr ruroJ srqf 'taq- L\IN: elrqenbe oq+ sr puocos 'o1 .reprurs o.re oq; 'ed,{1 cr11e3oq+ 'q+p!t +ou +nq l€cr?uopr (snuetsst - ) -o'tu.b s-sx- * pue (-owns-o"t6t d * > snutot n 6od) -oxLLn s- * ur?€T : roplo +ou sr +r ruroJ slq+ q 1nq'-ownsrIna9 prrc Ma'JIO 'cI?IoC '-orunsx.* ul Eur.rrncel uoruuroC sr xsns sftIf oq+ q +s.rgrerIJ > {o,t- I6\IN ',run7- ,1,1O : o-a.r1e1.redns 'ruro; xeldruoc Jo xsns € ol€q o/. osorl? Jo II€ uI 'punodrnoc € Jo roqtuow puocos oq+ oron 11q8noql s€ po+€eJ+ 'quetuqeell cnuouoqdoqdrour sr xsns l€uor+B,t.rrop€ orer{.4d. ewr?s oq+ Eur1e.r1sn1psos€c roq+o cruo+?rrg uI o^€rl e11\ CIITIC

87I

Jo onl€^ eq+ Bu$rnss€ +ceJ"redrorlrtsa er{+ Jo +uourqc€orcuo 'I€+uep ro rBIo^ orl+ Jo ernsserd l€uor+cunJ eqJ' I€us uI s+oor : e^rlcunfqns-s qsnl plo eq+ uI +r3rl] q+ud' -{Ioslcerd ur po+ceger q+rnoJ eq+ q+I/d leqlo8o1 'qcaum0 seer8€ 'ot1capfi,.t' ' sotoot6' tlcnp' qslo r(1.re g ur sorrpcunfqns-s Surrrl,rrns eerql 17y aq+ Jo +u€uosuoc I€uu loor Jo sruro+ m uol+nqrr+slp oq.l -ssa! 'tqns '.r19 i -s-qal* ', soog , oqcapfil' ,'pL p!1pa! o+ 'uoll€aouul cluo++Irg ts s€ 'Es g 'tqns 'J.red Tygq eq+ sI srqJ 'uorleuilo3: roJ s+I louu€qc or{+Jo uor+€clpur r€elc ouo qslo^\ 'c11eplo9 ur punoJ sr olqereduoc dpeg u1 o^€q e,$. +ng Surqlou ! c111e3uowruoC +ou sr -9s-x uT elrlcuntqns y '/doloq possncslp aq ilYrt qc1rlla''uorqeut,ro3:sue.r1 lecr8oleue req+rnJ t euo8;Iepun seq 1I 'o.,lr1cun[qns qqer11 oq+ sollrepun ruroJ erues or{J "oq1nq '+org 'oC ,tpco.np plop( ppon qclll.,la '(e.r5u1',r '1o,ttorre .z(q pepeoerd) -gs-x sI uErs o,trlcuntqns eq1 roJ uol+cnrlsuocor '-8- uro4 -?- q+pe '-oq- sI snol^qo oq+ 'ord.?esoql Eumlquoc cruo++rrg ur eallcunlqns oq+ Jo >Ir€rrr eq+ +€q+ uoos oslr!'o>IrI 'uo1e.rg u1 -o- earS puts qsruroc e^€q o/v\ '0- t{sl+I,rg erl '-9' ur ellrlcunfqns € ,(1.re1n3o.rplno/a qclq^\ Jo cruo++Irg -ogord ur ocue+slxe eq1 elord oaoq€ pessncslp srtrroJ ot{J '881 '.tgIO '1c 'amoq* ) '2,'v ,moq* tLE SEHT 'nq 'qetd 'Es € &\trtt oq+ q+I-4&pelenbe eq +couor q+og 'oq' '1tc 'Es g eqt u€c pue 'eql.reqerdplo oq+ sluoserder nq'qcee s€orerl r 'qcoyredrnl plo oq+ sluoserder Dq'sqa'Es 6 oq;, oq .redo.rd sruroJ eq+ Jo uor+nclrr+slpor € ur po+Insor t{clr{,vr 'qslrl uI elndoc orl+ Jo egrelerd pue lcesredurl Jo uol+€Euoc I€uor+cunJ e pol.rod ro+€I +€qnd.eruos€ 9e dllueprlle ssat e.roq1i esuo+ +coJrodrul ue o? peruroJsutsr+se,^l.flluenbesqns '*e'I qclq/A 'euo 1repoureq+ s€td. uor+cunJ cls€q or{J, "pq'qcogledu4 (-pnqq* flI ruo+ q+oq) qclqrd eq+ u1 s.rtodde oII€+I -DQa oq u€c 11 oculs '.rep1o peqenbo qqr,rn uotuwoC ct1v,71 'GAl 'tOtO) uev(eu.rnq;, q1r,tr uelo sr -pe* c\l;rrC uomwoC '(6I'/.7 'punore pu€ oq+ teqlo .deal +ou 7gg) eqque,rueg 'lgrsair > poo qsled[ e{II .{q pelerlsuouop ueoq s€q sB 'qce;rodmr o+ turoJ I€pour urorJ s€.4rqsul q reJsu€r+I€uol?cunJ 'e+€p q cI+IoC uouruoC sI +I (oa.4e1do ) ) ealpunt oq; -qns € Jo uorlcunJ or{+ uI 'i.oloq roq+rnJ eos 'surroJ r{slo^\ I9I

fiArfct{Ilffios

'o72o- sqr,esetd. eq? s€

s€ ., solrlp

po6n o7

olqa- errr+nI fie.tr , eq?sausrurnr ",rr*"1AfJ" qsllrplo "t l'".:,'ffilIut -,tl&r{fl".18'1"t"fl##.$i"",

., ueeqcnj eq1 q8nogr 'Surcurauocun .{11euroy ere .167611 :A fZ.l SCf -pua qsl/+lrB ur qceg.redu_rregJ.. .asnts.ry .76;o slueunB.re bq; 6 '(196I) .,uegdo.rne-opul ue OZ-II.l,t ?SvB. Jl+r+do ?o ?Irg?grd, 8

eql rof, 'Dg '?q: elndoc orl+ Jo eg"n1"r*1*g"eau1 pu€ oar+cunfqns qsr.r1 pIO oq+ "q"p"q* q+I.4,LItscquepl q +I .kAZ d.rnluec q+xrs eq+;o EuruurEoqeql .dq -0q .+pg SUHT)'c'v Suraf -gQ*ol >1cr:qseo8 pue ,-oq'oC uI sreedde ruroJ srq+ Jo 'o^oq€ uelr8 ueeq seq ur3rpeeucl snolcnu crseq oqJ or{;, ', eq. , loq go elrlcunfqns oq+ sr cruo++rrfl ur oarlcunfqns-4r oq+ Jo uorl€^Josuoc orl+ solord qcrq,ra.urJoJ I€uS oq; 'se8enEuul cruo+?Trg eorq+ oq? q saurp I€cuolsrq ur pedoleaep .{puepuedepur ere,&r.qcrq,r 'o- ur stu.rog:.Es g quece.reroru oq+ roJ slunocc€ '-oqow€s oqJ uI odr(1 elrqcunlqns olrlcnpo.rd oq1 3o ,.E.e .'Es g ocuengur ro+€I € lueserdo.r,u,o(t1)ya.1d lecr8opue 1 eq+ opTslno ur8lpe.red or{1 Jo +ser oq? uT -?- q+pt surroJ eqJ 'ftgg so1qe11.{sIBug Jo ssol oq+ 3o EuruurEoq eqX Sgnf) so+tspuos>Ic€f eroq/$. ,.1uororn r€J sr pue ,(699 SgHI) uos>Icef :(q pleqcln q ,(0€7.6 ,EL.T OXA) uasroped dq posodo.rd 'pao to1 uorleueldxe Ellq.I I'Io"t?roJ e+sruoauog

cn{of,Ilus

or,f,ltuo

09I

'urrJd oql oJoJoct s?r ur po+Insor oal3rl +Iqrts7rrJ omE- xoueill![ orn?cunf € Jo ocuosoJdeq1 ecurs '!-psn-* sonur+uoc,{11ce"rrp +I +r?q+olq€Arocuoc sI +l +nq I Eurpue-o.rozE Jo uor+r?orcoql .{ldurrs Surquese.rde.r'ur8r.ro ur lecrSoleue eq deru lLnE- wtro1 'Es g eqg, '?t-gsn-n ) pnnE- : earlcunlqns eq1 lcuntuoc 'Es '(616 'rg g olnlosq€ ?cogorII€ surroJ eseql Jo sluroJ dA ', a4vutt ueqce"rlg) (uorqco-,r.o"rd q+I,K) , Ie^€r+ 1ya. , pmntdfr,mgg ', ''E'e 'pnnq- '3s g ur III^,, pnnEm.t{nb >I€erq llrna, pnnE%olq sruroJ ern?nJ a\I[ eq+: e.r11cunfqns€Jo uor?cunJorll uI +orr gnq 'qs1er11o1ppll,1trur srncco xgns srq? Jo xogor crleuoqd '11o.4A. ,re1n8e.ry s€ qsle4,1 olr+rrurrd ur ruroJ slq+ q +sffo plp -Qq(p)-x xgns oq+ +Bqg.uorlecrpur snolceed euo el€q onA+nq : (-aot1--) -oE- e,rrlcun[qns qEor!\ oq? roJ +unocc€ '-oq- w uo+orfl put .{1qco.rrp?ouu€c ruroJ xgns € qcns qsproC ur pe.reeddepuv '-Qt1(n)-o.l.u8qsl+Irfl uI qcrqnl,'-pso-* odeqs eq1 Jo xsns earlcunlqns ts pe+cnrlsuocor o^€rl od\ '+qrnos oq o+ erB xsns srq+ punorl3 euourouoqd ornlcunl .rrarlncodoqg go ur8r.roeq+ g.€q?'eEenEuel oq+ q qrol ,t.reae z(lerluesse o+ -psp-* yo pee.rds oqel elrnb ssol+qnop pue snolrnblqn oq+ q sI +I '-?s-* +u€Ir€a xoldruoc ssel '.rep1o eq1 peculde.r flelelduioc +sorule pue 'spunocl 'd8o1eue ,ltBI +srg s.zcu.o1.{.rny 1eurEl"ros1r puo"{aq peo.rds Jo q?L&r pJocce ur ute8e '-psn-x xsns nd.ou srqJ 'D- Iarrgt ur sloor o\ -gs-* Jo uol+€xgns eq+ ruo4; -4so-* emeqd"rour ea,rlcunfqns re^tou € Jo uor+€erc orl+ sI uolg€ruJoJsutsJ+ puocos oq;. '-?2s-*xgins elrlcunfqns a,ou € Jo uor?€eJc oq+ sr uor+ -€ruroJsu€r+ eq+ e.rrlcunfqns oruo++rrfl-o+ora oq? uI lsrg 'sruroJ irue 3:o esuedxe eq+ +B srn.ro3lies ;o prae.rds oq+ pr€^!o+ .{cuepuoq ?rDIsu"eSle.reueE eq+ Jo enr+ sr our€s 'sqoor ({es) 'q{S oq; 1eo8u.d.re1oE€ sr -3Jo ocrnos oq? i q.r"rqsuegur +srro€ -dp-otl+ .{q qsr.roe-s- oq+Jo queurecelder oqg 'cr?Ioc uI Jo +€q? q+r,rd.I'ecr+uoprsI crllac w uor+€n+rsoqJ lue8urt.rel Jo soxegor 1ec18o1oqd,roruuo socrpuedde oq+ se 'eqr.reletd-s 'uor+€urJoJsu€r+ eq+ uo .reldeqc oq+ ees IIo,{,!.s€ 'spunoq leclFoloqd.rour sII+ Jo slr€+ep or{+ rof, lecr8oloru.{1o 'leur8uo s1r puo.(eq pozqe.reue8 pue '-so- 1 'kZt 'c1pq : errqcunlqns I aI !-ao cwrqcrc) {-n* Atp8s lueso.rd pu€ edr+l3crpurlueserd ueeid,loq ursrqdtourosr cr+Bru -3rpe.red euo sun urolled er{;, 'uoos ea€q eat s€ +u€q.InsoJ Jo '1gnq se.r ruElpe.rud oq+ 'fr,mq lueure+Ie rog'€I s+r Jo +sor eq+ qtq 'goq ru.ro; 'Es srqq urord '?bqx 1 ? * Qa* | QZE dI) I 'Es lo-ggq MO ur uoos 6 e,rrlec1pq oq+ Jo l€ug eql ro^o ue{€+ stsq 0gr. luurEr.ro oq+ +€q+ q ?oq ,!\O ro3: uorgeueldxo f1e>1q qsour eq1 uoruldo ,{ur u1 '6lV'2, OXA 'Zg, 'tOIO ';c : seEen8u€I cr+IoC oq+ uT ruroJ uI rer{go qc€o poouongur elrr.q -naqq* pue -sa* sqoo;r ueedornfl-opuJ o^r+ oq; 'srrl+ Jo +uoru -doleaop clleuoqd .ru1nBe.rI EI oq pu€ '!oq MO sl rrrroJ IAW 'e.Lr1cunfqns € s€ posn '(poq 'Es g crs€q oqJ L\IN) ur.ro; earluredurl oqq .{ldrurs pooqqo>lrl 11€q sr r- q wroJ eqq 'orlr.q osel{+JO 't!,oq rc ?oq MO qou ,(lureg.roc pue '(nnq* lcedxe p1no,.lreuo) og A\.IA[ple!( 'rea.eaa.oq'1ou p1noa. q1 'oq 'oC Io 99I

sl\IloNn

fsns

.s-61* mDE-^rntoJ xgns gounfuoo_orl] roJ ?unooc€ plnoi& our€s eqJ 'ed{1_ sq+ solq€[dsouour ur snelcnu ) , Bq4 , ?Itt IA, u1 se Jo 1ea.b,r. rror+€ rase.rd eq1 dols or1? oq+ ssol €q? ournss€ 1nq Jo Itsug Jo I.*r'gfl ', qro11 qslo1y\ oqq uo suor?€^resqo .nBI+loN ,ropo.LeilunC i t6g'3 , 't[ dq po+oolloc oouoprlo roqpnJ €rl+ dq pelroddns er€ suorsnlcuoo '6-LrlV 'qrun7 'qcrv ', 11 d8o1o1rqd crarud3 q solpnls , sr slg

pnnq- rtlse,^,lqcunfqns qsl-o,y! e5,l ol po?Blor{m or€ qcnB prn'?,ff}1; ", .sp.roal squese,rd-r Jo surJoJ 'sqe '3s g dlduris e.re srrrJoJ uo?eJg eq1 "reqqd -r7- eq1 pre8er o1 peldruel sr ur ! ro{r€rr sn}€rq € se .dldrurs pE?.t? Jo sru.rog euo 'e.trlaunfqns oq,] eq1 ur ,(1uo €sr{Jogto -?- sr!4.oqs Jo 1'r:.rn1d 's?rnqnJ dllecgrceds_ lorr €r€ '(Am uo?org eculs qr;reneqnqrq .18 al) 'qe.rqF '1! goquc '+olopar '13 gotaag[''8'e 'pnbqvr sorn?nJ qsle,1l lo??!.dsoc r ? l€ql o?oN ar "fL

,l-pq* Jolsecu€ o+€rperrrrrrreq+ q ruroJ slq;. ,'lilQq* "1t"gt uaeq o^€q +snur cluo+?Irfl uI ruroJ +soplo oqJ .I.taq,ooq p1g 'og 11tII : poq Io e,rrlcunfqns oqt roplsuoc o1 .t.resseceu sr lI qslo6t.{1.reg u1 ao(r7)-Eurpue earlcunlqns oqq ureldxe o;, '(qsle76 'cg.o '{o- .8s .Es g uropotr\J ur pozrl€roue8) 1 Ic€q oF espro.e>11 4snur I€u$-ure+sc[€co^ uT s+oor II€ fq peur.ro; sr qclrl^r 'eqr.re1e.rd-s erll l€q? ruururerE oarleredruoc c$IeC Jo 3u1uu1Eeqeq+ uro4 pezru8ocer uoeq s€q 'r€IoA pu€ +I puY l€?uop uI s+oor : sorur+IBclro+srq uT qslrl pIO q e.,lrlcuntqns-s orl+ Jo uor+nqr4slp oq+ q po?cogor IF+s 'eallcuntqns-s uol+€n+rs oq+ ocuosso uI sr srqJ I€clro+sTtl oq+ o+q fqqepour o1 lcods"e uro.r; roJsu€.4 leuorlcunS: e .{q pedolel'op qcFl/d 'cI+IoC uourruoC ur uor+€urroJ crleruEls e peiloqs ,L- IBluAur s+oor l€qro^ .(reurpd cI+IoC11€,tlequesse 'se11rpq1ssod ou er€ eroq;, roqlo 1€q+ uoos ,tpee.r1eo^€q odl 'fl- 'u- 'itr- q pue pFoc cr+Ioc uoululoc Y ?oor l€clro^ 'pou$n€xo oq ia.ou .,teur ecueprao srq+ Jo orn+€u or[[ 'qslrl pIO ur {pepcrped 'soEenEuel cI+IoC oq+ q elr.reqe.rd-1 oqq Jo uognqlr+ -srp rBrlncod oq+ q se11slseql0drtq sItI+ roJ ocuopr^o crsBq oq+ +€q+ po+ou osrid.o>lqe^€rl ed\ '+srroe cgqeurFp oq+ uI elr.relerd-4 oq+ Jo ur8po eqq : ororl po+uoseedeq oq srseqlod.{q oq+ po+B+$",{peo.r1eo^€rl odrpoEesr,rue oq u€c uol+€rrlroJ sIrI+ u13go eqq go ruelqord oq? o+ uol?nlos roq+oue ?o21 Jo '+cerrocul peroprsuoc ocl eroJe.roq+ +snur slseqloddq l€q1 pue '.ro; pequnocc€ oq ?ouu€c d1du4s +I +slroe cr?€uroq+€ goo.r ueodo.rnfl-opul uB Jo srseqlo&(q oq+ {fl ol;[aso

Ei[IEfifgtrat-,z

69f

'u-r€lrocrm

68

qlnmI.un Jo ,tEolourf+o

pu€ uorqtsrurod

6

uI fp€e{€ lq,l orcpq pezrl€rqneu selc /4/ pan l8l uoone+oq ocloa Jo uorqrsoddo oq? 1€q? i&ou{ oie 'ece1d 1s.rg oq+ uI '.{repuoces pu€ o+€I f1ea.11e1ersr fur s+oor .fq oql.reqe.rd-toq? Jo uol+€ruroJ oqq +€q+ suorl€crpuI snororunu or€ ererl? pu€ 'pequ1os1 f,11e1o1,tlectuouoqd '(ouo1e -oqdroru pue ,(lecrleuoqd e.re f- ug sloo.r eq;, 6-) ,n- pue (u- +ou +nq u- I- r-) A'- : sl€ug loor Jo soss€Ic oll1 dn +os utsc e^ uor+€Inuro3: s,uesferunq;. tuoJd isruro+ 'ueeru orn?cnr?s +oor Jo smle+ ur elr.reqord-4 lecr8olouoqd u1 soop l€q.&t eq+ Jo uor+nclrr+srp cr+slro+c€rBq" 'Jc ".{+ '8np?.1 '+E : 'oyba,t,* > : suolsnJuoc ecnpo.rd tqcatI plnoo oarss€d eqreelo.rdoq+ q epe.rE 1ry oq+ Jo ocuerreodde oq+ +Bq+ 'relncrlred u1 6- u1 s+oor o+ 'elrroqo.rd-1 orI+ Jo suol+€ruroJ.{repuoces pus re?€I eqq u1 z(1uoq +I '-ofiry* gI > -ol!II* ) qFTc-'.116 : eper8 oroz eq+ uo peuroJ ser$.oarssed eq+ s€oreqrr (-s-pq* ''E'e) loo.r ope.rE 1q: oq+ o9 e11.reqe.rc[ pexsns se,ra" qd;roru ?srJo€eql s+oor qcns ur Jof, '?- pu€ l- es€c .due u1 'lueuosor IBug q s+oor Jo dnorE ltur8rro eq+ eprs -qno eculd uo>I€+oa€q fpo ppoc eqr.relerd-1oq+ pu€ -ot- rr\ erlssed e11lo1e.rdaq+ Jo uol?€cgpuopl oq+ +BtI? pocgou eq ppoqs 11queurdoloaep slq+ Jo sseuo+€Ioq+ o+ ocuoroJer uI 'porred I€cIro?sI{ orl+ Jo o^o oq+ uo +sotul€ poruroJ 'cpleC r€lnsur Jo soqcu€rq oia.+ oq+ uI suor+€aouu11o11e.red z(ldrurs ore .r(eqg.'se1qe11,tsI€uS Jo ssol oq+ o1 rogolsod ! opo11r,rg pu€ crloprop ltsclroqslq ur olrreqo,rd', eq+ Jo sruJoJ +uocer +sow oq? luosercler. -o!-* u1 elrssed eg.rolo.rd oq+ o+ lecl8oleue srrrroJ I€roaos or{J, 't?c- 'IIIN to '1uomb 'luryJ sp z(lgcexe -ol-qnp* 'suor1eru.ro3: a'-oy-clnxsdeqred pue 'Wfr(npfrp sruroJ otIJ lerdrcrgeed ea.rssed ol {c€q oE qgfr,nrun 'pe+so?+€sI aclnop* frT > cnp eyrelerd oq1 ,tpo (ufr'np'fr'p ur pepunoduroc) , psol ufi,np xeldmrs orl? JoJ pluv 'cnraa 'cnpfi'p 'Es g selpeqe.rd .re1n8e.req? ureq+ eprsoq e^€q surroJ r{+og 't1gfr,muta qgfr,npfr,proJ po+€crpm os!.r&o{Tlsr urEuo .{repuoceg ', Euns stsra,, !?c- oa,lssed eg.role.rd oq+ Jo s1se1sod.{q u lqtsrjrloppU\I ,{1.reese ur8Fo ,(ldrurs : (e8gl gg) Euts t?c, , 'se1qe11ds I€ug Jo Itsol+uepl Jo orts qunn6 pun rupc lqlIa/ ssol oq+ .rege (o1dlc4red -o1-*) eanssed eqr.relord pue 'Es g

'(996I)

0S-662't

ttc!,?1a1t'r(qor.rg 'y

'C eas 1rrourg€org oq? rod I

e1r.rele"rd-1 dlrquepr orl+ oq+ .{ldrurs 11pq uo ;o lopow Jo '1ecr3o1eueputs o+€I 010.{LsruroJ qsloA\ erl+ +€q+ pezruEoce.r -{pn1s luur8rro srq uI uosdou"rnqg 'uon6a6'ucnrlcac: qsrrl pIO 'qln(nlan ur o1r.re1e.rdpelecqdnpe.r € oatsr{ o/d"++srg eq; 'Ttfi'mpfr'p 'lunntb 'gunc 'uorqe1se11e ;o uorldecxe eq+ qlpr pu€ uor+nqr4sTp uI qslrl pIO qy,r.r. ee.r8c surroJ oseqJ, '(,tto1eq ', eos o,r.r.+ qcrqn uo) , pe,1ce1o.rd , t11fi,mtan +qEno.rq, q1fi'npfl'p : (-6at,' r1g) -fre.r tp pourn+er, (soEcau,ufl,p ) "rom6. lpa 'Otnrutot-op 'p1,Q,owt1g) -6nu, > ,U!?nlu, lyyg

, 8 u € s , > l u o c .+r^€urcs€J, '?ergo : (pt"taq 'qg) -raq qool , t11tau.tfi,c +tn-onn-rD ) , : (pe4.rern se ldecxe 'tlsle1{. e1pplni) e1r.re1e.rd-1 eq+ Jo surroJ 3uralo11o3orl+ o^Br{ e,ro.se8enEu€I cmo++rrg oq+ urord 'oqrrelerd-1 eq+ urroJ I- pua l- ur sqro^ Suo.rqs11€+ErIl peurnsse f1.4e5 eq feur +I 'roqrunu w II€rus or€ ses€cqcns ?ng '(OfS gl 'spulq '?oII '€16 Jc puc zgqfl) Uas-o.tro'(OZOSnt) sn77a6o"t '!' 'p : orncsqo dldrnls or€ tp6-pur^r'-pp +nq solpeqe.rd'1 e{H {ool qclqa 's+xo+ oq+ ur punoJ sruroJ I€uors€cco ore oroq+ esr,{.o{rT 'g"tn-* ,lcedxo ppoqs ond qclq/l roJ o, sq8nold p?qn sts qcns 'pe+so+?€sr , 11€ ?€ elrreqerd ou qcnl.,!droJ eruos or€ oror{+?- pu€ l- uI sqro^ Euo.rls eq1 Euomy 'qsIII pIO rno$ uor+€?ueuncop ro[nJ o/d. p€q 6suroJ ^l.eJ pepue+xo oq pFoc ?s{ sH+ +€q+ elqraqo,rd sr 11 e fq

.1^OIS, UO-

, qqEnos, !Ecn?.s. poqclor+s , tqca,r( pas€orcur , lEcD1.il. pelcelo.rd , lqcouo.paL tqco-

-6.to -6ns '6at

-frmn -6aun -bp : f- ur sqoor

, po.rnod, lps.4oo+, I?'

-ura8 -ula : ae- ur s+ooJ

g,I,luflIfiEar-,2

99r

v9'r

.(616I 'sT.rua)

sanbun\ sop rnotVuVfi so'L??cor'n)'+otIeutr

691 ,sanbcuou,r'.t'ab'u'tfi'1V an0uD\nI ap ?t?A +eilo'\l-l&"::f";is_ffrr"_ :gubc ''t'e 1ouz7n7 '(.

-prrt as* lnc , -pozrs)

: ere

'elqelclpe.rd

+nq

'pelse+le

+oN tt

'cI€+I ur pul? cr+Iec uI olq€Il.,(s Eurleclldnper oI{+ Jo scnuouoqdoqd.roru eq1 3o ,fpn1s poF€+op € ororl/rd.oslo +uosord II€qs I i {pnls +uosord eq+ q o1 'ssocord qcns rorllou"e dpo popnfl"e eq .{po u€c orn+€oJsrql '-ur7oc '-?o{ c-?o't' s€r& o+rro?ord eq1 ur -t- Jo uol+€xgns eq; '-{qpElqure s+u€urell€ qro^ord cnuouoqdoqd.row eq1 s€ qcns sql pIoA€ o+ socr^op l€crl€Iuru€JF snolr€a o? osrnocoJ p€q qsrrl pIO '(ern1q: pue oar?cunlqns uee,la'qeq-{pepuns) elrrelerd pue lueserd uoa/d+oq suosred ur€lroc u1 .{lrquep1 ur po+Insor qc-rqa 'spunoduroc l€qro^ oluololord ur e1qeffs 'y1co1,s-(ot') Bullecrldnper oq+ Jo uolsse.rddns oq+ q sIqI eprsoq ypn-(oa'op)'E'g pue ltpnut'-('to{-op)Jo q+oq ro+c€r€qc qcrqr[ +uown8ee .req1.rn;€ sr oreql r(repuocos eq+ sr'ld'oqs 'pozrro+c€reqc z(1purlsrp +ou snq+ pu€ +u€r.roqe -{1eur.ro; s€a qclq/a eqrrole.rd € o+ {r€ur cguouoqdoqd.rour re,reelc 'lqcno.s 1 bops* € o^!E o? 'wroJ eq+ ezlrlqBls o+ ropro u1 o+ pepp€ .t1du4s s€,{d''?- qd,rou oq+ ?€q+ +uopr,re q +I '!' '-bo1,s* 'uroqled l€qro^ cqloc eq+ q po+€Iosl ,f1e1o1s€^T'qcrq ruroJ sn+€n[ +u€r.Ieq€ oq+ q pe+Inser 's- cq€co^Io+uT Jo ssol .re1n8e.req;, 'tEcplls elrrele"rd qslrl pIO eq+ sorlropun qcrq/& rrrroJ or{+ ?c€J uI q sIqJ "frns.ts*l-botas* eql.relerd 's{oos pt'6ms) '6os € ruroJ plnoqs (, ro5 so4etu pe1uc11dnpe,r , +Bql uoq+ +codxo pFoqs ei11 zr'cru€urroC pue ur1e1 t(q oslr3 por€qs sI +T : seEtsn8u€I ueedorng-opul uro+so,w oq+ Eurs, Jo orn?€oJ lecrEoloqd.row cl+srre+c€r€r{c€ sr srqJ ,r'(. -unc) -umpac '(,4eolq , -PnM) -pvut'aul'(o8Ip , 'pn1c) -pn7t1cm 'q,m1 -6o"tq) 'bn.tqaqorts qslrl plo q seldurexe eq;, '(. ?eolq , 'e1r.re1erdpelecqdnper € uroJ oq polcodxe eq fiew -un6 ro -;Ln[t orn+cnr?s oq+ Jo ruslltscoa-1,t{?Ii ' +ool cl+Iec Y '-frnw,aw polecqdnper lcedxo p1no.4deuo 'pe1so11e 'eEe1s,t1"reeue qe ptbtQtn o,art+v xeldrnp eq+Jo elrro+o.rde orely1 -urruouep oq? q+InapesnJuoc oru€coq srq+ oculs '(-.tlofr,-6nw*) olq 'uorqrsoduroc xoldurrs eq+ roJ pelse++€s1eg.reqe.rcf p,1,6tota 'e,rrsssd o? po+cl4sot sr 1tnuttof-op Io lEcvru- osr,ld'o{rT elrlelerd oq+ o+ lecgEoleue oq .{em qro^ srq? 3o e11.re1e.rd-1

r.9I

'-buo.tp '-6ut2'-Bu.oq '-6uol '-62f '-bt2ta '-6uo! '-6uolp '-6uop '-6.0u'-61,2'-61,7s '-60..t,o-6a.t'-Botq : er€ oseql or

oq+ ?€q+ polso8Fns s€q Jlosurrq ues.{eu"rnq;, 'snlc?.t, '+ET 'olrssed elrrole"rd : -o1-6at* rrrroJ l€ufiuou oq+ ruo4 .{1"re1n3o.r s€ .{po soaJos lqcal prrv'6onta"r o1r.re1e.rdpegecrldnpe.r ', ', s s€q pc6o,r,xoldr4s orIJ |€ode.r , -6a,t-,t?n-El!,?2osr.r€, -6a"t-ssa : uorlrsodruoc ur .{1uo srncco !Ecai,- rrrrol 'Es g eqg, 'ocuopr.to lecr8olopqd l€uro+m r(1e.rndruorJ olEI oq o+ u/d.oqsocl rue"c6- ur s+oor dc1 elr.ra1e.rd-,or{? Jo uor+€rrrroJ orl+ ereq,il. ses€c IBnprIIpuI Jo reqrunu € oJomroq+rnJ oJ€ eroqJ 'sruroJ +uBiloq€ xrs oq+ roJ +unocc€ o? +uolu -doleaop -{.repuocos oruos roJ Euplool ur pagrlsnf or€ o,td.udq+ 'eq1.re1e.rd-4r ro polecqdnpe.r peq.oedxooql urroJ Jar{+Br +nq or'qou op f- u1 sq.rel Euorqs qsl"r1pIO uoo?Jgroq+o eq? olrqd. 'egl.reqerd-1€ ruroJ f- ur sqre,r qslrl pIO xrs +€q+ elresqo ord. uorlr ecuell 'e1r.reg.e"rd pelecgdnpe.r € wroJ '?r- : U- uoqntr ',tr'g - g' uaqid. A'- m pu€ Ie.rU ur sloor roq?o 11€ltu't o1r.req.e.rd-1 € ruroJ E+oor -[Eif, 1e f ery.re1e.rd.-p ue turoJ s+oor 'suorldecxe -Jl,tr or€r rl+I,rd. : pepuedxo eq {eur +sTIsIqI lla 'e-,rrlcunfqns-? u€ urJoJ s+oor -ggf,t 'e,rr1cunfqns-su€ rrrJoJ 11e s+oor -fty[t 11e '-ro3 po]unocce oq u€c qcrqa,r.-suorldecxe 'qu€uosuoc Jo reqrunu fl€rus fte,r B rl+I11l leug-+oor eq+ fq l(pepcrlred '1oo.r oqq Jo orn?cnr+s oq? t(q polep8er f1qcr.4s sI s+utsrtrroJJo uor+nqrrlslp oq+ 'qrel c111epoq+ Jo ser.ro8olec lecrEoloqd.roru roq+o ur oeceld puoces oq+ uI '+srro€ crt€tuoq+€ u€edorng -opul u€ Jo +uopuocsop l€ouq € sts pe,rd.erasr froEeleo orl+ JT 'cI- Iatuqgcr+IoC uournoC u1 s+oor ruo,{I se1rrelord,3 ou oa€r{ pFoqs ond. ?€q+ z(1e41pn .{1euro.r1xesr +r uoql o!-q- sB.{11cexe po?€or+ sr 1-f- crqlep uoururoC JI .uog€n+Is crmouoqdoqdrour lercgr?re u€ slueso.rde.rr(en ouros, q +nq 'lecrEolouoqd 'snorcldsns 1ou sr uorlnqrrlsrp oq+ +Eq+s^4,oqs+I ,(UBtq q '/x/ .qc- lralo^. ssolocroa ur ro^ou pue lfl f- te1oa. pacrol e ur Eurpue s+oor uo ,tpo peuroSi 'l+x-llEc- ur Eurpuo seqr.role;rdJo sorros € qslrl pIO q ea€q ond.+€q+ ?c€J or{I 'pr,6uco 'u'a, 'tIO '-olf6auo* :uaorl su,plu,.olqauv Jo Incnuo 'lql p qu€rr€a Nd rls{ntsC eq+ er€druoC len+xo?uoc e '[x] s€ peeeedde q+oq ! porred cIlIoC uoruruoC eql OI.ITfiO

fiJrufl,f,filril-,Z

99I

oroqA +€q? nq uorsnlcuoc roq+o ou id"€Jp utc aL\ '-81upue c-s?- '-s.l- ueo,,u.1oqwsqoll€rud Euro8-q8noroq+ oq+ Jo id.orl uI '(t-OS't OXA oas srar{?o ro1) gsmp OHO : -o1ill,* > -ols"tol* ) , 1sr1qg, lrnl 'rIO -{Eo1om,(1eolq€Irtsss€un oq+ dq pe,ro"rd se '-gr,geqeldurocorl+ sI cr+IoCt4 -tst[I- ecuonbes erl] Jo +Inser or{J :Eulpuocr+€ruoq+€q+I^\'(?)r-s-g[Ltr sE fluurEr.ro.reedde plno,{r qcrq,r 'sqoor qcns Jo +BrJoBcrleurErs 'Es g oqg s1 lurod 'o+ue+o.rd-1eq13o uor+nqrrlsrp orl+Jo urolqo.rd eq1 I€rcnrc oq; o+ uorgnlos eqg seppord qcrq,r.r,doqs eroJeq uorlrsod ur cr?Io3 q -sg- sre+snlc er{+ Jo soxogor crlouoqd eq+ q +I '(na:nra) -o-d>t '(ndnaa) -o-d>n '(nyraras) -o-y)Lo (ndogf>) -o-dt7f s€ qons s+srJotscrleruBls >IooJCuoruruog ereduroc .{em euo {gecrEolodr(9, '-s- ,(q poxsns -U[Iif, orn+cnr+s oq? Jo s+oor eurnss€ .r(eru e,r. ?sIJo€ cl+€tuErs crq1e3 uoruuroC or{+ Jod 'cI+IoC uouuroC uI uor+€rrrroJ-, or{+ Jo snolcnu leul8rro eq1 'elr.regerd-1 'l- ';e luil.l.o1 wpu€ lueserde.r ,{eq; leug ur s+oor '.rog polunocce I€qro^ freu4.rd 11€+€q+ pu€+s u€c o1nr oq+ suolqdecxe ,t.reqrlos o,!e+ osoq+ q+pe ! uorltru.rog: +srroe c11eru81se se peureldxe oq u€c rr,Dw?I- pue 'ureo; Eururr(qr e fluo sr ?srg oq+ 'tcntany- puv rxnw.pp- sruroJ oa+ oq+ JO 'I€cl+uopl ere s8urpuo eq1 'pw6- ppfr- p?6- 'uxnqcac-unEcac-%nqcac-oar+c€ 1e.rn1doq+ uI re+Jarxvurpl-. I "rnutpl- 6 rou&pf 1 'Es gueuodep : ell"reqord-q pu€ pe+€crldnpe.r e,rr1c€ er{+ Jo suosrod eerq+ ?srg oq+ uI f911enb +ueuosuoc Jo uor+nqrr+slp eq+ o+ o+€p ro+€I qcnur B Jo uor+€Furrsse lecgEoleue ue slueso.rde.r 'qcunfuoc plr€ o+nlosqe uooao.+oquor?cur+srp oq+ Jo ssol eq+ pue 'e1r.ro1e.rd guouodep eq+ urElpered oqJ 'I€cr?uopr or€ +srro€ cI+IoC Jo IInJ cr?IoC oql oq+ pu€ sEurpuo roq?o eq+ 1e se lsnf luoserd 3o 'oro:4 ) r?p- olnlosqe o,rrssed lueserd qsl{ pIO oq+ Jo q€qx rl+IntrI€OI+uepI sI ?lo-* Eurpuo s1q;, 'o.t'p-'Es g e-l.rssedorperu o+I++IH oq+ qqud. pelenbe flqceerp oq z{eur o;o-* Eurpuo oq;, '-oL.ilp-1 -osutp- sr 't1g +uow?€or+lecr8olouoqd eqqt| (p1,6ws '-ssgrs: -6ns) ottXcunlqns-s qsr,r1pIO oq? ur ueos.s,r3'-s- eroJoq 't,to-s-tag7* -??Io,ro^ +oor pouoq+Euo1.{geclEoloqd.rour ryg,u. queuodep +srroe cr+€{u8rs e se peugeldxe oq o+ sT ruroJ sH;. 'rxnu,t?I-uoq+ sr roJ pa+unocc€ oq o+ ur+o3;dpo eq;

69I

flJrufl,tfua-.,

'!?-* @+!rc+ol.du qoedxe ppoqs 6tVL 'po+rmocsrp eq ,(eur ?I ecuoq 'qsTrI ur e,reqa.{ue polso+1B 6rts sur.rog qoegred eo eq.rrele.rd ou +ng '(OSOf) Oa-Sf .gg.6un'J eos qcq.4a uo '. 'rI?Jg € uorlrppB ur sr oJogJ +eeJ Jo spuBr{ seqsts.dr, ut,tD-u'?, er

'uor+€ruroJor lecrEoleue o1 pefqns snq+ pu€ u.re11ed oql eprslno 'noonrloga {oor0 pul? ll?ruJou '++IH uI se '(t)ysntan?a rerfr€o uB ?z-Fotunp Jo +uornocelde.r B oq -{eru (luepocolue sgr .ro) quul,ppIIe/d. lecrEolouoqd '. uloq svu, .r,toug0- Jo+J€ ,{,ou eq; , , lqEnoql , "tr,Du?,til'Jc : qro^ qslrJ eq+ ur luenbel;ur +ou er€ s.ned eru,{q.r qcng ,ktV'.tOtO) slso8Ens uos.{eurnq;, poopw ee ,tzau.og- uttog Eurur.{q.r eqq o+ lecr8oleue sI rxuurpp- qrql {p>111dleruerlxe s1 1r 'o1uo1e.rd eq1 ur r(po queuodep s\ p?Mlop s€orer{^\ rurErpered eq1 lnoq8norqq queuodep sr roElaul?4?l-oouls '+uoprco€ l€crro?srq € oq +snlu qslrl plo uI ssBlc sH+ uI ecueserd s?I pu€ 'qco1s leur8rro eq+ go q.red oq +ouu€c -lLart etn+cnr+s eq+ Jo s+oor rerl+o oq+ 11v Euouoe -gng ern+cnr+s or{+ Jo +oor € I leufg,ro oq +ouu€c ecuo8.re.,lrpqcns 'e1r.re1e.rd-g oq+ uqll4yqueuodep ere ,toqq (g) ! dop e qou 'lueuoso.r € ur spuo +oor oq+ (Z) : -a- qou (-tunp ,-utn1) -n- sr Ioid.o^ wo+s crs€q oq+ (1) : suosee.roorq+ roJ sselc elr.reqord-gr .Jxou,pT ,.m,ou"r,ppur snol€ruou€ or€ swroJ o.ad.+esorlJ : elreoqerd-grlueuodep B turoJ asorl+Jo r11og .(-tun1) , se.rep, &nEpurxnfor pu€ (-tayp) , s?rrupe , p?ur?Dp *., : oqrre4o.rd € rrrroJ r{crrltd. qsl{ pIO uI au- ur Bqroa Euorls .reqqo onq {po or€ oroq+ (-was'-uta) eqr.releed-4 B oa€rl - uI sqro^ orrd.+o1q/d. ro; 'Eurpeolsrru sr ese.rqd eq6 .eqrrelerd-4 eq1 rrrcl ar- ur s+oor . ouros , ?€q+ q +uowo+€?ss,uos..{eurnq;, 'LuI€US ur sgoor fq peuroJ seqr.reqe.rd-3 er{l uoq? roprsuoc sn +oT '14-. to '1- t- q +u€uosor 'queuosel +Bqg erorl/d. IBUU ur s?oor otl+ dq pelueso.rde.r sr uor+BrrrroJ-?cIlIoC uornmoC orl+ Jo 4coqs leurEr.ro og+ ?€q? s1 queogruErs sr +€q.dtqnq I queurdolo^op .{.rupuooes sFI+ .roJ iunocce o1 elqlssod sI +T+€q+ /toleq oos 11tsqsod1 .t(.roEeqec I€qro^ cr?IoC uotutuoC slq+ o+ uor+orcc€ o?€I pue drepuoces e squese;rdo*6- u1 sloor ,tq oqrrele.rd-1 eq+ Jo uo.r+€wroJ og+ +€q+ uorssncsrp Euro8orog oq+ ruorl lueaedde sr 11 . OIfTfiO

89I

'(lSOf)

S-ffS'gg

'6un,J eeg s1

'y,aq- 'Bs 'N?q- 'Bs 'pnoq- 'Es 1 'srnro; 6 t lcunfuoc r{slrl pIO oq+ q qcns s€ dpcexe ponulluoc q +I '(orez : 'Es g) uoNuNrssIolur o+ l(ressocouunq +I ?€q+pue '-o1o-*'-ssg-*ur ed.{1 oloqm erl+ JoJ Iopow s€ poaJos osoq+ sts qcns srrrroJ lteJ B ?tsq+ elqrssod ,-?Ep-panlx sr ?r pu€ '-?Is* wI > !,11?o,t-sp flI > rpeAorloq, xlalc- ete seldruraxg '-g- ueedo.rnfl-opul o+ >IctsqseoF qr r"tt.t1ota +€q+ r€elc sr +r sos€c ulr?+recq +nq i g1.t1oru SCO m se u'sq,rer ( cr+€rueq+-Jleq,cr.l.tsIserl+ Jo -g- o.qqqqrar pegenbo .serrypqrssod oq ppoc Ar eceld ?srg orl+ uJ l€ro^os Jo .Io.4d.oa -?- orl; s1rrupt -ol0- '-ssxcr?€ruoq+ueodotng Jo ,-ssa-* -opuJ erl+ quese.rdor -a- eq+ +€r[+ r€olc +snlr;u-op-* Jo sr qt 'd.reurr.rd sr -g- leqg I ,r,r.ar.ts.loilrom qsure8e sre ?!?-x 6/.I

I 34afi,a-*'Fsg crq1e3€ Jo ,{rora srq qdocc€ o.&\+ou ro rorl+eqlA 's-r.rssedoq? q -or- pu€ +srro€ oq+ ur (s)s- eroJecl sr€edd€ qcrq.{,\,-?- s1t{+ q +I +€i{+ pun '-ofr,-a- squese.rde.r-ofia- xrg'ne (e.,rrge.re1r-) o^r+tssn€coq+ +€q? sepnlcuoc oH '([S€6I] EgZ'09 ZX) , runlr+€sn€X ueqcsru€urro8opul urng, pu€ ([616I] ( €^r+€sney uorlcsl4 rop Eunppquurelg EB-?;67 ZII) rnz , : urolqo.rd slq+ o+ sorpn+s o,1a+po+o^op peq ,(u.ro4o4 'sor?Incgrp snorrosou lueso.rd-g- pua -a- slo^rol wo+s oq; 'ples€ ror{?rtsJpu€ cr}Ioc unl+r/d. q+oq '!+xeu oq+ o+ e8en8uel ouo ruorJ dlrs.relrp Jo +unotu€ ur€+roc e qcodxo 's?colerp ?snw od. ocual{ l€npi^rpur oq+ Jo uoueuroueqd €

8I

SruOlS

crI€COA

,{repuOoeS

JOrl+O

pUB

OAI1€UITUOUOp

o? s+srroe 3o lueurdolo^op oq+ i ueedo.rng-opul-olord ur poqsrldruocc"e+ou s€.{rsqra^ poarrep yo tu8rpered aqolduroce yo uolln+r+suocoq+ +€r{l ,raou{ o11yss oq+ eroJoq -?- '-!- '-a- '-p' slo,rtroaruo+s eq+ dpo .rogpolunocce ocl o+ ur€ruor oJot[J 'uouourouoqd crlqn8uq ''3s g eq1 ur Eurpuo eraz E oq+ lureue8 € Jlos+r Jo uor+n+r+suoc go ecuenbesuoc r? sr uJn+ ur uor+€urroJsue.rl lecrEoluut3 slrl; '.r\1?Ir? s€ srq+ e1els .{ew on I uorqeruroJsu€J+ lecr8oleue ue u1 ur8rpe.red oq+ Jo +sor orl+ oururre+ep UA rIcH-&\'Es g oq1 3o ruroJ or{+ s1 q1 'spfi,- '8s g .re13eun-spfi,- lmnun*'3s 'urou Ear1u)urtnul 'uuu?uxn e1uuEoc qsIJI €q;, 'rg1n '{oua 'o3 rn d1lee1c +sour ueos '-uo1'u-Huag* EI > .'pl , ffoua ,,Ir\os "-HuTfr os€q -uour-Dun* 81 U.LroJarIJ '-uour-* xgJns oqq. q\r , ur ol€q e11l_ lee8u-{ae1 ruo4 urroJ qons ouo cr+IeC ur pe,rrese.rd lct3J 'crllec 8€ xeldraroo ', ('repoddns) slql Jo +uetu+€er? crleuoqd ourus oq? 8mn'oqs >IeorC o\IIeC de;r1s, zorln1rr'>I0 Jo +Br{1{l+o€x€ sr eddl ol1.tr t'-uoru-nvq1 ernlcnr?s er{} Jo ?oor gI l'HgEJ ur {1.ruprEe.r epler{, -uout"71gg1 -uout'n'* pe8ud,rel ts o+ pexggns au.l*\ -uou,t'- €Joqa sas€o osoq? uJ sorl '+ooJ erl? 'epe.r8 r4 -rr- oql go m8r"ro eq;, 1n; m d11eur;rou sr qcrq$' oq ,t11oe.rrppepp€ 'xggns .{reurtad e EI -uolu- : -d.orl Jo qulod u"eedo'rng -opql eq? uro.r; leuorldeoxe eq plno.d. esec due q r{cTq'!a 'I€ug ure?s

'##;; r€ururoupto u€ B€porneldxe€q +o..u€creaeidorl

j]_-_"i{*"n*

-ueurFes crureuoqdoqd,roru rel€I oq? lnq '-uo11t'13-ru.ro;: elduns eq1 t431'ro qte't vrc1s-olofr' ur .t1aee1osr xgns erII '-uout'n-?'in* : , sq8nold , p'1."r'.on dpce.rrp pe^Irep 8! qclt{^\ '. ueuq8nold , liloJto sr {rrJoJ eq? ruo{ '-uourco- f,'q -uou'r,n'esllrq oq; 3o lueureoelde'r l€qrolr -{1ernd e uto:g eql roJ Iouu€qo eq? se?€rlsn11l qclr{ia eldurexe r€elc ouo e^€q €lV\ 'elqulorperd et xgns earqxpord oVI sts -uoutnx-* fq -uour'n-* Jo +uoru -ece1de.r oq? pu€ 'eloq€ pessncsrp ?tsq? q?r.6. I€crquopr sr ssaco.rd eq;, 'o+e'-uourD?-Izxrr* rrorsrarp eureqd.rour luenbesqns E plua o-uourn-L?D1o1* '-uoutn-oynyt* sed"{qoqo.rd qlraa' ', drqsp.rol , q??n{ } , p;io1 'rc1nt, u'at4.on{ ', ern+€orc s€ , 2??p > . roq€oro , Iilo'tkp : ss€Ic srqq roJ sunou estsq pe^ro6 qcrqlr sure+s-? aq? uro4 pe?c€r'1xo se peureldxe eq o1 .d1"ree1cs1 -uurtol,-* Jo ? I€I?1u.reqJ s', sa8pnf , r?Ur?p?ut| -uoutD-fvaw* > . '+slnt[,

88I

.\IIION

+"::'E;i:,,fll'::"W,I1Xf":,:, Xllt F ?i:::tr::" s"r+"p ",r) "1,?,:

('rol?€ur srql ur r€elc +oy sI tg .r01O) .b? crqe11.{ssrppo?ouep .{11em8r}o +I +€q? neJ,r eq1^Surldeoce pue ..sor;roqeql ,'qsii1 '-ro_yo or?sr;req.o€.fl?qo e se pee.rds sq.r Smle.rlsuou;;op ,tunopto! ur'se' orrres ortr1 q!.rld, arurv oql p mtl1y@v) Fmpua .ssed'Bs g eq1 passn,isip ;?!Ilq.u4nloJ .rrorq.tsrrrroJeu€r? .{lluenbasqns uesdeulnq;, gq_ZSI]_ 9-n'tI dCZ) IBDrtoIarrB ol olqsrl snrll i< ogpuoJ eurroJ. eql -uoutD_ plrts ,, uorq-epuoy €p ourJoJ, Oq+ eurooeq Ew.7-uounx- 1tsrl1 8I _uoutDl,_N _uo,u.&D.Ic;orx?.rl -uarogrp l€urroJ or{.}Jo +lnsor oqJ .pelcadxe eq ol sJ srrq7, .uourotq)r*q* ruo.r; peledocuz(.s ,{1.rep8ar s uDzuaUt!.rq .Es- .ued iri1in,tsr.r1 ,eijqil '-uoutDa-.re8uno.,( eq? Jo ocuengm 1ecl?oleug eq+ o+ onp sr _uou,tD_luarrts,r sexegar r{srrJ pIO aqf q edocu,ts Jo eouesqe er{l +€q? Burar,oqs I?q1o ?r11Jo r€Incr?r€cl uT p-TI€ -ourn?- llJ:oJJ -uo.1,u,L- Smqsrn8mgsyp csu6rqerr,iasqb eseq+ papu6.?xe (lStOtl 9-Zte'98 5rg) .repue"rqsielT i(96-9 nT : t668il zSl'02 ^Cpr) 9I $ elllqO unnloJ €rury creqc.re-aqf-"m'. ieaunsiJir'. .Bs ,ue8 ure8o eq1 o1 utD?p!,oJ'ds 'ruou oq1 01 pu€ -.(ure1s-o) rwvrrvrn uorqueqt€ poil€o or{^4, ,(qSO 'a ,6t_ .g16I 1!\VdqS) 6S $ AI .mg 'ra(oln dq pel.roddns se,r,r pue-.i19.i-fi>frf.l "pun1Wor11 s,.re"iap"j .ZLI .,|DIO o63 sr -uo|,u,nI,-* rroT?onJ?suocoJ eqJ z Itf:-":Lt:! II€mErJo

urnpaur sr ?sor€elc eq? qorq.l^, Jo suuoJ rIsIrI pIO .&!.eJ€ u] pea.rese.rd p1s '-uou.tD-* Br cr?IeC m xgns eql Jo rrfioJ eargonpo.rdrm pu€ ,r6plo oq; u', Smsuedsrp , wp > , ;reaeaqdnc , Mlt?pp.. q.ueure8pnf , qlatq ).egpn!, uaqf.lq : surols Itsmurou o1 dgedicur..rd peppe xgJ:ns rmou lueEe ue se (ute- '8s 'urou .t1g\ -uotuno-* ur"ro.I oq+ ur (cqeprop-oqoa4 eqrl .ro) qsuT oar?runr4 m dqnrlcnpord ;:o por.rad o oaeq {11ue.redd, p1p +l 'qerJl pIO oreqcr€ ur rrolo ?nq e,rrlcnpo;td .re8uol ou Br xgJns srr.I; 'znrlo>Lh '{C : tadtkh uI s€ -uoru-re gI.Io xegor eql . rep€el .I€murou .xgms : sr uor?€uiloJ eq; lou eq? olq I€qJoa pele.rod;rocm pur pezqereue8 ueoq s€q (n se) aq+ pu€ ,-Gigg,f epe.r8 1ee8u.{.r€I 1ny ur e.rsedde ?oor eq? oror{.la'crlToC ur -4,ro8elec qcns ouo ?s€al }ts e^€q otv! 'euo1e .{.roeq1 uo pes"Bq ,-HVEft sts flo.la. sB ,-Vq,tr .g.utsuosarts +ou Er ur sqroa orl? 11€ o? or?IoC ur pezqe.reue8 uoaq el€q ppoaa, pe8u.,(re1 I€ug pue 1oo.r epe"rE 1ng o1 :redo,rd xogor eq? 1eq1 uorqdumsse eq;, .?rr€Aolal +orr sr Brrll qnq .uorlelueur8es JerI+JnJ lrmpts .lnc ,{eru ;1os1r ?oor orIJ luonT?suoc e?€rpeunur ?s4} oq1 Jo surro+ ur dldrms elq€ugep ocuor{ erts n1lSns p:ue ?oot i osuos ueedo"rng_opu1 er{? TrI loor € €q peou eq+ rtselc ?ou e{Bru ( osIB ?oor, +€q1 seldurexe ?r d+uourolo eBerIJ 'snxrnlfi ed.t'1 eq1 Jo -?- .DI1uS ed.{q eq1 Jo -2IBug-toor e 3o uorqe;rodaocul .{q orm+ qc€o .-ou!)!.- oq urnl uI -out- prte ._oltoo1 pepuradxe sr -ozr- xgns aql oroqa .ryp?gs!,lq7---!ul?!tnry__11r?,1n5 '?€T dq pa?tsr?Bn[r ecuenbes oqg ur lueurdole^op l€or?uop1 eq1 3o {po >lurti? peau eug 'd;roeql lee8udrel q?ytr op o1 Burqlou s€q +r : ern?cnrg.s leorSoloqd.rour go ed,{1 srql go sa8en8u€I ur soxgtns Jo uor+€uuoJsutsrl oq+ lnoq€ .r.ei le.reue8 € sr srq+ 1eq1 pezrseqdrue eq pFoqs ?I .xBns -?-?re,lrqcnpoJd eq? s€ eceldea : serldde al?I IIT,|& +srg s,zc1a,o1,,t.rny plrts '-I- puts -lD- sqdrour xgrns Sqleduroc ol€q eu, luiod sFIl gV TYEONAUYT

[H'I

.f,O SEXfiT,f,EE

681

OJnlonrls I'oltlrropt Jo -!,r,D?*,qro^-?, dr€u,,rA e sesoddn;T:i WY;t > moto, 'Jo^o^roq !los,+r unou l€qro^ erlJ .ogz.I 0)I/. .ues.repe4 dq possnoslp 68 'mmo,Lol* 'namo,rD!* 'ozy'uo?* > (uorSourqelq) s9ma4 .qetd 'umnl1 'jIrr?s, mwnt ( llr , prn€, n1our I .JIo ! srrrtsur€J , pxDnou, 'tlo r . soqsTrrolq, p?DuD* 's.+Eor.,pxouo 'JIo ( 8m€rrroJ .JIO , sele,redes , p?o,Luce :8uua.o11ogerl+ er€ sqrol 'IZ-gIg '. aqJ uosqocfry umilolf .ro4 sqrel-r, i(.rsun,rd qslrl pIO eq+ 66lor+rtseqq ur o8u s.ree,t euros pe+Bco^pts ;o ,{.ro1s1qd;eururrle,ld V , I qcrqa 'suuoJ eur€s oq+ Jo ouros 3o uorleurldxe Jerlrtso ue ecolde"r o1 popuo?rrt sr ?f 'sqre^-r, d;reurrrd cT?IoCJo dnorSe;o srs.{1euu l€cuo?srq orll Jo urelqo"rd eq1 o1 uor+nlos id.ou € ?u€serd 1ua, xrpuedde sru;, SEEA : OI.ITIIC

NI

TY$ONAUYT

gET

.f,O sEXfiTJIfirI

TYCIOOTOE'I1IOtr\I

'9I

$

'(a.aqqon{) -uo,un)-L?nlm* rr xSnB elll€ururouep'f.repuooes ofreurr.rd urorJ ot, (tanpaut) -uouto-poul* rrr xgns olrleqJolop +JTqs eq1 ecuenbesuoo -{.resseoeu € s€ pop€?ue qcrqal '. (tU.on{ 'qO) dgqsp.rol t eltu* , -g?r)fn* sts surJoJ lerrrurou ,(1e"rnd qcns dq uorqcuny ltsu.rurou vI -pa'tu* B€ rlcns unou +ooJ crl€ureq+B oql Jo +u€urecelde.r 1errr.ro; go sseco:rd aqq sl +I '9oor aru€B er{? uo ?Inq srrnou pe8.re1ue puu pepueqxe dleur.rog ?nq 'I€cr?uopr ,,{lleuorqoung Jo +ueuftlceoroue eq1 ,{q peltsulu4lo .{panper8 ere.^d sunou ?oor aq? 'se8en8uel ueedo.rng-opul ror{lo 11€ ur 'ss?p-patil u€cso ur s.rredde lqcryzt -paur* rrnou sts cr?IeC 14 lng +oor cr?rlureq?€ oq? q?r-daI€crluepr Sureq -paurr( +oor I€qJeA eq+ .e,r.oqe . e8pnf , -uoutn-f)out.* sr eldurexe poo8 y 'rrrJoJ erues er{l ptsq unou looJ eq+ ooqs

98r

flUflA

IIIIJ

'pn?s 'rr {3 'aeu4: .','*o"nyu*fi;i"r:";;''{{"ffi';u;": "' '(SZO1 .[eneS]-(srer €IIBI1) gZ qs?,tl pIO to ia9]nar In?nnls?H s

jo' eq+q?p,ruor?oeuuoo -ur,e-rsz rruoqaoav\f;tiH3f /#"Tffi

?rr{sues srql yo lueurdole,rap xelduroc oql Jo sre.{1eue pellelep V s '?uol€l\rqur€ ouft|ceq e1qe11,{s .loo.r -;[E;L ts o? pex$ns I€orptsr eql uoq+ ,{1ee"rgpue ps+ctsJlxe Eela. -uo'tuo- xgins oql oouo ?nq ! epe.rE o.rez erreq r(lleur.rou plnoja Bestsc rlcns rn rmou +oor cr+€rueqqe €q+ ecqs .+oor .-uoul,U7[ snon8rqurerm u€ uo +Fnq oR -'troem)Ugil sedf,1 eqg, 'solr?€ rrop .{1ernd .reql.rng ,tq rmou +oor eq+ go lueurecelde.r I€rrturou luenbesqns oq+ puts ounou +oor cllaureql€ eql qllrd +oor I€qJeA eq? Jo l111uap1 l€urroJ Jo Bestsour eoual€,rrqurts eq1 .{q pameldxe eq o? Br ouo e^rl€rnurouep uferpuooes a o? x$ns 1ecrpeJ.{.ruun.rd e rro.rg -uou,tU- or+Ioc Jo lJp[B eqq i qoor Fqro^ eq? uo +1rnq pue .z(.reurr,rd sejl xgns eq+ ueqit eunl e 1e eould ue{ts1 o^Bq ganut (-uow- Brncqde;r) -uqlltllo uorlezrlereue8 eqg, .pe?se+?€ qsrrl +seplo 6111.uJ .uuoJ e,rrlonpo.rdrm u€ 4gns eq+ puts .peqelduoo f,pee.r1c Br Burels-? r;uolE -uolu,D!,- Io uorlezllu.reue8 eq1 go sseoo.rd puoces oq! roJ .orllaC u.r f,I,rep ,t1e,l.r1"a1e.r eoeld ue>1e1el€q lsnru -uown- Jo uorlezqe.reue8 eq1 go sseco.rd eq;, .glg uosqoqpf uDlu,ou ,toI 'Jo .8uo.rq.s u(11eq81.ro se.aa.qroa : se r?Bluep ltsmurou s11 .{q u.lcoqs se '14 .{q ua.oqs se ., pe,(o.r1sep eq; sT .B+Ieru, plofloL .tIO ,-6aI* a'ee,tetrq ) . {lou , utoba1 il +oot. -JgJ oI -uouro- 14 uoJ+€rrrroJ prlql v '16 IIT ilI 'se{q1s 'pe) gZ uu€unrv rTgC (.E.o .ern+Bre}rl r{BrrJ elpprtr{trq wS.ssod S pogse+?B er urownbag o?lv I r{sr.rl r(1e.rnd su pep.re8e.r eq oqe ,(ew 1eq1;de eqa 'ryutobag* sea. .{grn-,r.rp eql Jo rrlroJ qslptsp eq1 lsoE8ns plnoj&'o3 proJral€ll ur .,tuo.roq our€B oq? ruorJ or€ suorldr.rosrn qloq---+cerroc Jr qclqa ', slue.r8runm r{sTln€p Jo +ueruol1+6s o?tsI ts Jo .{uourrpel , Fmar8 'ur81.ro qsl1ntsp m s€ svNoxrrross rrr.rogureEo eqq.spae8e.r n,(u.ro o4 '3 plog rei.olg , sn6ota-o7p7g opsoq u.mu,@?p?g-ourtlu ,suo1lr,-obas prmoduroo eq+ uxorJ r[.roJ I€coI oq+ u] uees ed.t1 eq1 Jo clqsrrocod..(q r sr peqe.rd.relm eq oslts .{uu .req1e16rll +nq ! suorldr.rcsu-r u1l€T e^g q pelso+1ts sr€J\l Jo leqqrde ue ,owo6eg eur€u ernlJp u-r?€T -o11€p eq? ur osla sdeq;red puts (009 ,€gZ .sou .0 .I .I .p ..re1s5ecetrq) strNorrrrous ru.rog .reln8urs e.trlrue8 eq1 ur suorqducsm ure8o ortd.+ur errreu eur^rp Jo leuos.red B€ pa+so?1€ ,so6ts 'qqoyl -o6as .1nep .apqns .+qS '-q6as'-uou,n6os * qI uro4 * er ed.f;,4 7, g a eq1 yo elduruxe req?orry .(-umtuxrDqp) -uow1- ur.ro; 9.1 3uo1 lecrSoleue rq+IM 1es eql .1o pee"rds oq? fq porncsqo dle8rel ueeq Br'q, (gq;tr -HUETI '-,tory* - -gua0*) -uDuu,pelp- -uDut'!.up! uorl€ure+Ia lucrfiolour,{1e pge leur8r"ro eq1 ororll( -unur- xgtrs eru€s oq+ qllid. ll;rrlsu€S m gueurdoleaep r€Funs € o^€q or!\ .-uou)D-pout* > lu,Dpaw.{O rr.r sa .-JEfr +oorB o,+ .{lenque,re pue '-ZI Etr 1oor,iruu ue o? pepp€ dlee;r; eq w@Wplr:roo -uolnn.-gy[t xgrns pe?n?r+suoo .{1,rl,eu eq6 qoor ts o] poxgins -uourr,D-WW TYfiONIIVT,{O

SfrXUTJIfirI

78I

'ereq 8n uJocuoc qou peeu seouoursep eq? rrrroJ esmoJd eql Jo 'uorlcnJ+suoceJ 'pesoddnserd +€q+ lqnop o+ uos€er orr el€r{ ell pue od-{q ueedo.rng-opul eq? e{11 Srnqlou q sFI'} +ng '!1ry01* ol€q

e.& qcrqr

u17Utur* ?t?'Inur* ??PuD* 9l?u0 * uDrpqs* 'Es g eqtunxo.rd e sr oB uec ea. qsoqlr€J oq? '6urroJ qsrrl eql uro4 {c€q Smlcnrgsuooell 'Bprocor ltscrrolsF{ 6ql rrI Joedd€ feqq qolqlrr ur r[roJ oqg e.rrnbce o+ €ur€c BqJe^ crlloC ue^eB Jo xls eser{+ 'por.rod oTlleC eql mq?rl. '/.oq r4eldxe o? sI sn 8u11uo.r;uoo urelqo.rd eq;,

b wugl?-Hu'tr -71-HULL -?1.u-HUiL

?c-Hggfi !s'HUg,tr ?ur'Hgq'tr

: se sur8rpe.red qons eurnsse .(eur er11 'p.rn1d eq1 ur sErnpue eqq uo pu€ '.re1n8ms eq1 ur e1qt1,{s +oor eq+ uo ll€J lueoc€ er{J '(epplu, eqq lnoq8no.rqq puts I€np eTI+ ur s€ lleja se) 1crn1d oql ur epe.r8 oraz puts 'reln8rns oql rn qooJ ep€r8 IInJ q+ud. 'quose"rd eq1 ur ur8rpe.red oruoqdode uts urroJ ueedo,rng-opul eql roJ '?ueuosuoc :oq1o ,tue se ,(ldurrs po?€oJ+ lcedxe ppoqs eit r{ons BV uorgrsod srql rn qea8udr€I oq? 1-,IgT,tr o.re.a,r. ernlcnrls goor eq9 Jo slueso.rd ?oor cr?tstuoql€ Jo sstsIc .ro3.re1 eqq .;o ed,{1qns e .{lereur e.re.aa. eq+ urord sluese.rd 1es orquureqle ^.e: Jo lurod ueedo.rng-opul 'eu4? u.r EpIBA-roJ squesord ies olleureqle Surq.analsuooe.r3o enbruqcel eq1 .(q 'cr'11e3urr{?rdrr. 3o luotudoleaep lecrrolsrqerd eqq Furrnurexo ,(q pequ.rlsrrourep eq .{eru srseqlod,(q srq1;o.{1iprrc,t oqg al.orl o6s II€qs e,u,f sluese.rd 1es crleureqqe ueodo.rng-opul enur?Troc Bqro^ cI+IeC eq+ +€q+ : srseqlod.{q .rno 'padoloaep eroi. r{slet6 .{1.rtg pue qsr;1 € €ur ll€rls eia 'uoq+ 'srr{J pIO m pelse?t€ uloq? o^€r{ ea s€ sqre^ oserl? Jo suIJoJ eql l€r{? 'cl.}leC uounuoC o?ur po+rrerlur 'sluesord cr?€ureq?€ qcns uro.r;: si6.d. +r +Brlt roqpnJ pue 'ueedo.rng-opul 14 sluase,rd cl?€ruer{?€ poulroJ sqre^ osor{+ '-guoEH*'-Huozg* ,t1e,tr1oedse.r 11€ ?€q? d1111q1qs9doq+ s?srxa eter45 slooJ rrro4 oln^a 'uro6 puts 'llxup '1:ig : se8en8uel I€orJo+sEI oq? ur poqBag?tssq;ro,r eleuSoc cr?€tuoq+€ ol€q oa sqre^ eseq+ Jo o,o{} rod 'eseq+ 'adlq xgur Jo drr€ roJ ocu€prle I€B€u eq+ €pnlcxe op deqq ?nq o,rr8 lq8rs lsJg ?E lou op sollesuror{g surroJ cr?Ie3 er{J 'suor+€oJc Jol€I pue drepuoces -{1e.reue8 ore esoq? qnq'cg-afr,-pgS, E€ qcns 'se8en8uel '(?t-H[I-u-gg,) xgrl l€stsu ro lecrro?srq eq1 ur ;reedde deur sed.{1 rcq+O fluflA

A8I

y'?rt1'wfr,ge'fc?e '+el,,otrX-norrrg .{1e1cedseees ,{lurey .Iq? ,rgJ': "

*

crl€rueq+€ reqqre fgensn e.rera. deql i squesead Jo spr4{ &l-t1gnt, 'ueedo.rng-opu1 u1 Jo requmu pe+$u11 e ,t1uo peur.ro; sloor 1es 'dno.rE ueedo.rnf-opul eq+ ulq?1.{r or?IeC Jo uorgrsod or1srnErnl eq+ Jo Erelqord eloqal eql o+ur lqSlsur areu sa,^"r8puB .or+IeC uouruoC Jo eJnl€u lern8r.ro eq+ ?noq€ uol+€urroJu .releer8 ,(11e1qucqs 'alqrssod eJoJo?aroq s€irt u€ql z(.rolsrq -qns Bn eezrrE ouenl uI 'EII{J osuorq.BuuoJsu€J+ 4eq+ Jo .aa.er,r.egrce.rd eJorrr J€J € l€ ezr.rJJ€ u€c eia leorSolouoqd pue lecrEoloqdrou qloq Jo surr€1 ur surroJ osotl+ Jo g.u6ur -do1e,rep eq1 Squolsraue .(q 1ng 'do1s o1 luo?uoc uooq el€q serpn1s ItscrJo+srq snor.re.rd eJoq^r Br ser3o1our.,{1e }ooJ Jo ?uourqsTlq€?so srql -ueedo.rng-opul q?I,ld Jou orlloC qq.ua roq?r€u Smp.rocce uor+nlos B pequese.rd qorqal-,tpnqs rorlrtso Tr.&r.o.,(ru go uolldecxo oq+ q?rtvl , qrlr, U s . r r o < l l . > d > t ' 1 . DI r r y p f f i ! ' ( t L D , m ? . 1

,purrB, ga|yn 'qteg '.\g\ou 'asrt[ 'nouc '^+BrI 'o?,lpu 'qq.1 luarueer8e '( sqru .tl4s I npu.t, Itsuor+tsnq.ueoc€ €Ioq& oq? prrB ) Llourut M 'tndnno'>I-0'oJD/o or+€ureqgts.rrropt pxbun*.r1o . plocs, ?Drloro sor/ran .4p ,, soq?tsorq , gc.r,up$rcser.d in* '+r{S i pQ,DuD.lIO uor?tsuoqm e+nc€ qqr.4. , elu.redes , ?Ultclsp3,,txtls'LlIr.It poDrrtcs'rIO .1oo.r 1eeEu.4.relroJ €ouoprle tlqlqxe qcrq..rt seleuSoc 4oq? puts su.rroJ cr?IoC eq? Srnsodelxnl ,(q '.{So1our,(gegoor € r{sqq€lso deur or11 .-HtIg,L looa (1es) 1ee8u,t.ru1uro.r3: erts sqr€a oseq+ Jo e^g lstsol ?a ?€q? ocueprao eldure slsrxe e"req; 'orn?cn.rge ?oor r€Incr?rtsd srql luese.rd qolq^\ crqloC ,(pee ur sq.re,r olr+€uurouep-uou .{1uo oql ere eseq? ecuecyru8rs +Br1l Jo ssel,}qnop sr '-g-Un(il) o;rn+cm?s .{.relrun .rreqq .(q ss€Ic 1I ltsruroJ n Be pe?runlep pu€ pougop eq deur .{eqq ! sq.re,t-n er{l Jo +sor erlt uro.r; .{1ele.redes pe?€€rl Smeq .rreqq sogTlsnl .{.retrn.rd er€ Bqrol esoq+ +€ql +cBJ oqJ '( Bpurrs p?|aw 'rlo q9ye nlDu.tJ , M prrts 'n,tau.t, '+B.I rl+ra\ poolour lstsJ?rroc oJourJoq+rnJ {eur ea, oI}IeC eprs?no

>

.sq8rs, nuso-uoc';c

',,ra8ne ,IW,LDI , III (.?neu) .qsJrrrelq, ?/o (Dun-sso-uoc Eurueatu eq1 .ro1) .rnaatq, Ipul) ('cseru) .3u14odun, tucs ,rIO

: r{?Ild .{Ie^Ilcodser sqJol €Berlq e.reduro3 'uor,+€ro+I€ crureuoqdoqd,roqr qsrJl 3ur,rq due dq peureldxe eq 1'ouu€c rlcrqa, 'urJoJ ?uoBJa^Jp € .ll.or{s ,?srxo .{eq1 su .re3:osrn 'crqlap .os€q urq,trru surroy eleuSoo eq? .reqq€u l€ufiuou drre uro.ry pe^rrep era deqj '(ry 'rpIO) Es.BII-D e,rrlcnpo.rd ., >1eell , eq+ uI lorr 'selr+Buturouop s€ pe+cagur q8noq+lv .dreurr.rd or€ Bqrel eseq? IIV .

uHf

TYIIf)I{-IUYT

.f,O SUXfiTf,flU

98r

'polcoJJarm s€-1. uSrp€rBd eq+ Jo ornlcnr+s crs€q 'crlloc uouruoc m .{Iree o}mb uooq e^€r{ .{€ur uorln+r?sqns or{+ lng ]vq+ | x(D)rccs 'rIO ecueq.&\ 'pelnlT+sqns lurod euros ?€ s€.!{ -r- od€qB elrieqojd eq+ Jo-314pue rie '?s- '3s U cl+eurar{?e leu-r8T.roeq+ roJ q elerpeunm

otll

'r{slrl plo wr p'Lntncs 3;o luese.rd eq1 3o ed'{1o1o'rrl urSlpe.lud oqq qll,la Itscrluepr sr ?I s€ elqr?cnrlsuocer '?|uo-rnqB* ??-?'tDtlg*

?u.t-pDqstF

-aur-DJnqg*

s Q)'?'rozls*

-ar?rDcls*

snql

q

ur8rperod

'uor!€1rrmss€ crureqd,rour € Jo rr€q? uolln+Jlsqns 4eeds ueae sdeqrod deru e.aa.??lJrDcIso+ ?\runJls eureqdrour eaT+€mruouop a,rrlcnpo.rd {1eure.rlxe -r- qd.rour lrssoJ 'e^rlcnpo.rdun 'peqelosl aq,} Jo 'lueurecelde.r er[? Br surroJ osaq? Jo 1ueurdolo^ep '!'tuo'JDtlS*

lu€llnsor

eqJ

orureqdrour € Jo ssol r[.or| Surssed u1 "?' pu€ uounuoc eql .{q 'esue1 luase.rd eq1 rn eqq 3o e8els I€ug orIJ

Ll-u,Dclg*

-at- DJDtIs*

xs-Dl,nqs*

-aur-D,tDrIg*

ola-n,rDtlg*

'(e^oq€) : cr+tsuro?n€ oq IIta ur8rpu.red 3rn-aa'o11o3er{+ dq +r 'ueq? 'luaureculde.r eq;, ueos o^€q e,u. su u€rpered cr11e3-o1o.r4 eq+ Jo 'epe.r8-orez 'ue,rr8 se 1r e:1e1 -{eur e/a: srql oq suorldeoxe ou oJ€ eroqJ '?nalqv crleru8lpe.red 'sp:ro,t eq+ pozrl€reuo8 dles.ro,urm stsq cIlIeC roq?o ul 'epe.r3 o.rez pue epe.rE IInJ uaeal€q uor+€rrrollB lBuollounJ B urorJ peqrrequr ur.Srpu.rud-{.re-te rn 1uq1 po+Iqlqxe qcrq.4l u€adorng-opql 'urelsds crls1n8tn1 crlloc-o+orc{ erl? Jo Ia^eI ](ue 1u elqerls 1c€J oq? sI +I -uoruep dppeo.r prrB 'ouol€ or+IoC o+ Jadord uor?€unoJsrr€r+ crurouoqd 'uor+€nlrs -oqd.rorn a BI sTqJ 'seuorlro+ur ro+ctsJ reqlouo eroq lng cporldode Jo lxaluoo oq? qqll4 , olqts+s € s1 srql 'su€rp€r€d '?tuo'@cts* 01-DJ,aqs* -o1-uDcls*

?8-DJOXIS*

-aut-o,rpcls* xL&-o.toclB* 'IBrnId puoces nr8lpe.red eq1 Srn,terqc'e snq+ pua lsrg eql ur r:r,Lp epar? otez 1ec13o1oqd.rour eq+ .dq p.t' eper8 ol,ez crleuoqd oq? Jo gueurocelder eq1 o1 Sinpeel 1sd1e1uc B€ +ots -{Fr€ssecou 'oldur€xo pTnoh\ ?truo-rncls* 1u.rn1d prlql 6q1 surroJ oorun eseq+ ulq}Il!! ',ld.oT^ uroJJ Jo ?ulod s.zcl^\olfrnx I€opI uB -tlJeeu -{.rel oq pForl . 11 '+uaurecalder poeprl leclSoioqd.rour € qcns o1 ealcnpuoc +sour pur3 elq€rno^€J ?solu +tsq? ,tleqce.rd sr ereq sruaoJ 1e.rn1d eqt trt uol+tsn+rs 8u1urng, puocos Jno o1 qe8e eqq +etl+ eos u€o en 'ur8lpt.red _ 'peII€,,\.oJo ?, 'ep€r8 'lsrxo xeBor oql 'BurroJ pe+ts^llour-uou uI ?ou plp sJq? eroqlvl oJez eq? pue ep€r8 IInJ et{? uoealoq uo11€Ier lecrSoloqd.rour 3uyt11 € pe+slxs ereq+ Jo^ereqia '-o|-yt roplo eq1 pecelde'r -o?-D'tD!-4aeu eqJ

68r

fiEflA

-o?-D,rD|. -O-,tOt

-OI-OJAIa -o-to|

uorl.rodold aq? Jo q.uourqs{qts+se eql o? pel oo'rql lsrg eqg ur se8ueqo lecrSolouoqd .re1nEe.r€ql 'cr?IeC Surpnlcm 'so8en8uel 3o dno.€ u.reqqnos eq? q .,ra.oN.I€r?rur I€?u€uosuoc qlpl eureqdroar : -o!- 'Iar+rrn oqtscol q1r-aa.eureqd;rour : -o- 6re{n& -o1-Er1

-o'H'q

(epe,r8 orez) -opul

uI

'IBrlIu-r

-o?-Hro? -o-H"[41 (epe;8 1ng) : ser.res Euraa.olloJ eql p€q eat ueedo.rng qlr_d\ seuror{dJoru xgns puB .I€r?rm

la+u€uoBuoc

oq€oo^ q+r.d, seureqd.rour xgns Jo uorlsoddo eq+ u.r seq loor eql Jo urJoJ or{+ o1 -fe1 eq1 'elqrssod se ,{1drms s€ +nd I uor?€r?suouep oq? Jo II€+op esloerd oq+ olur oB ol oreq ..t"resseoeu?ou sl ?I .se8en8uel €seq? uJ b Io.d&o^ or{+ Jo uorsue?xe eq+ qlgn dn punoq fleqerullur .sqoo.r sr epe.r8 oaez leorEoloqd.rour srq; leo8uf"rel go epe;rE o.roz l€clSolor{droru € pue oi}ouoqd e uoell.leq qsln8m+srp +sruu euo-ollloC pu"a'r4?€a'{eerp'u€iuourry-se8en8uul u€edorng-opul rrroq?nos eql ul +€q1 po+tsrlsuourep ,{lqueqp.rq seq zoraa.ol-4lnx .sostsq lee8ud"rel go epe.r8 orez eq? Jo urroJ eqg uo (.g 61 $ aouotldody) zcra,o1.{.rn14go {ro.6a +rrooer eqq dq uaroqs ueeq s€q ,{qlnognp srrll Jo ?no dea. eq;, 'urErpered i&o{B B splerf qcrqll Uuo-r,Dqs*

'Id g

dq pegeldruoc eq +snur u€rpe.red eq+ eoueq : ro oIlIeC r4 q le^r.o^ eroJeq lT,, Jo xogor erl+ +ng .-a?-p.r,cls*.ld Z -eur-pr,qs* 'Id I .?", sr cr?IoC ur goedxe qq8rur e.4. ecuolf +uruosuoc aloJoq H& Jo xeger p€+Blosr I€rqFIro oq;, ,dlin8rrpue ue .dpreqe el€r{ ot& 1e.rn1deql 14 1ng .u-Dnaqs* xs-n.rcqB* xur-D.Iaqg* .suorl€urroJsuts.r1 : elduns sr .reln8urs eq; pcrSolouoqd ,t1e.rnd pue .re1n8e.rfq 'ur8rpe.red srrlq cr+IeC-o?ord er{? sI de+s puooes eq; +InBor Jo 'o1u9Hrcle * ?1'H.t'?cls* ?1-H.t'qs* ?8-H.t'?qe* -?11rH.ttls* ?ur-H,t?cIB* : ru8rpelrd 3ura,o11o.;: eq+ o^tsrl o11\ '. Bel€J€des , pr,b,mcs .qro^ qsul IBn+oE rre oq Surpuodse.r.roc uorlcn,rlsuoceJ a osn o/, +€q? dluo e,res .peqsTlq€?se dlsnoiae.rd su 'sedd1o1o.rd ueedoang-opul oq+ q?IA ur8eq feur e1y1 'eldrurs .{1e-lrqe1e.rq srrrroJ osor{? urelqo.rd oql o+ uorqnlos eq+..suol+€ur Jo -roJsu€r? orureuoqdoqd.rour cTro?sqerd go s3m4.roa" eqq. 8m,re.toce.r dq p.te 'eqn+ ur BpJ€MJoJ Surqcnrqsuooe.r 3o enbpqoel eql ,tq 1e;1 TYfiON.T.EYT

lf,g.I

frO SflXtrT.iTflT

88I

XgONI

'uor+tsurJoJsu€J? erues eq1 Jo eldrrr€xo reqtou€ ,(ldrurs sr -ptncF*