Q.73S 
 
 $B bM 5TS 
 
 THE DATIVE OF AGENCY 
 
 A Chapter of 
 Indo-European Case-syntax 
 
 BY 
 
 ALEXANDER GREEN 
 
 Submitted in Partial Fulfilment of the Requirements 
 
 FOR the Degree of Doctor of Philosophy, in the 
 
 Faculty of Philosophy, Columbia University 
 
 Ketj) gork 
 COLUMBIA UNIVERSITY PRESS 
 
Digitized by the Internet Archive 
 
 in 2007 with funding from 
 
 IVIicrosoft Corporation 
 
 http://www.archive.org/details/dativeofagencychOOgreerich 
 
/ ^ 
 
 COLUMBIA UNIVERSITY GERMANIC STUDIES 
 
 THE DATIVE OF AGENCY 
 
COLUMBIA UNIVERSITY PRESS 
 SALES AGENTS 
 
 NEW YORK: 
 
 LEMCKE & BUECHNER 
 30-32 West 27TH Street 
 
 LONDON : 
 
 HUMPHREY MILFORD 
 
 Amen Corner, E.C. 
 
 TORONTO : 
 
 HUMPHREY MILFORD 
 
 25 Richmond St., W. 
 
THE DATIVE OF AGENCY 
 
 A Chapter of 
 Indo-European Case-syntax 
 
 BY 
 
 ALEXANDER GREEN 
 
 Submitted in Partial Fulfilment of the Requirements 
 
 FOR THE Degree of Doctor of Philosophy, in the 
 
 Faculty of Philosophy, Columbia University 
 
 ::^:»\: 
 
 COLUMBIA UNIVERSITY PRESS 
 1913 
 
y 
 
 XJ.* 
 
 Copyright, 1913, 
 By COLUMBIA UNIVERSITY PRESS. 
 
 Set up and electrotyped. Published December, 19x3. 
 
 •'. ::.«:••; 
 
 Nortoaotr l&ttM 
 
 J. B. Gushing Co. — Berwick & Smith Co. 
 
 Norwood, Mass., U.S.A. 
 
Approved for publication, on behalf oj the Department of 
 Germanic Languages and Literatures of Columbia University. 
 
 Calvin Thomas. 
 New York, October, 1913. 
 
 311456 
 
MATRI AC SORORI 
 SACRUM 
 
FOREWORD 
 
 Of the various functional types of the Dative in the several Indo- 
 European languages no single one, whether original or acquired, 
 has been given so little consideration or has been dismissed with 
 so summary a treatment as that ordinarily designated as the Dative 
 of Agency. Indifferently bandied about, but more commonly as- 
 signed to the original Indo-European Dative, as one of its natural 
 developments, it has in grammatical works of all kinds come to 
 assume the character of a colorless subspecies, to all intents and 
 purposes long since tracked to its lair, duly identified and ticketed, 
 with never a suspicion as to its complete make-up attaching to it. 
 Delbruck, so far as I know, was the first ^ to indicate along the 
 entire line of Indo-European case-studies an ultimate Instrumental 
 origin for Datives of such function ; but in the more recent of his 
 works ^ he has contented himself with declaring the impossibility of 
 a definite division between the two cases, as far as the inherited 
 status of any given language is concerned, and is in fact tacitly in 
 favor of relegating the whole usage to its old place under the Da- 
 tive of Personal Interest. A lately published book giving a passing 
 notice to the subject ^ even seeks to derive from the latter viewpoint 
 additional support for its quite plausible theory of the development 
 of certain Possessive Genitives from the use of Pronominal Da- 
 tives; and tho manifestly a representation of a condition conse- 
 quent upon the one here under discussion, it nevertheless is a 
 significant index of the current conception of the Dative of Agency. 
 
 It has appeared to me that the full value of the construction 
 could more easily be ascertained if it were not subsumed, as 
 merely one of the many constituents, under the general division 
 of the Dative of Personal Interest, nor therefore determined, for 
 
 1 Ablativ localis instrumentalist Berlin, 1867, p. 65 £F., and Synt. Forsch. IV. 
 60, 78. 
 
 2 Cf. Vgl. Synt. I. 300, and Synkretismus, Strassburg, 1907, p. 173. 
 
 * Havers, Untersuchungen zur Kasussyntax der idg. Sprachen, Strassburg, 191 1. 
 
example, from a far-fetched semasiological relation with certain 
 forms expressive of possession, but rather brought in line, as once 
 suggested by Delbriick, with those unmistakable employments of 
 another case which not only harmonize thruout with its specific 
 significative force, but also throw a more or less decisive light on 
 much of its ultimate provenience. It was with this in view that I 
 have undertaken, in this survey of the whole field of contention, to 
 connect, wherever possible, into a vital relationship the auctorial 
 function of the Dative with that of the Instrumental of which it is, 
 in a number of Indo-European languages, the historical heir and 
 representative. It has seemed that if any coalescence or even a 
 contact was to be sought between our construction and any other, 
 aiming at the syntactic location of the former, the analogous use 
 of the old Instrumental was, if not the exclusive, at least the most 
 logical one to resort to. That this idea in itself is not new, I 
 need not repeat ; but it is equally true that it has not yet received 
 that comprehensive comparative investigation in the various avail- 
 able Indo-European languages which its relative importance would 
 warrant.^ 
 
 In attempting to secure for my subject a certain degree of com- 
 pleteness in treatment, I have not shrunk from rehearsing many 
 otherwise well-known and firmly established facts of philology. A 
 reconstructive essay, however favorably circumstanced with respect 
 to illustrative matter, must still labor under a disadvantage in that 
 it lacks the authentic actuality of the more tangible work. In the 
 present case I have in addition been often confronted with that 
 curious situation of comparative paucity of decisive instances which 
 not infrequently robs the argument of its cogency. Many of the 
 changes involved in the discussion must have taken place long 
 prior to any literature we possess, and the remnants of even these 
 earliest linguistic documents are lamentably few. The obvious 
 necessity then of overcoming in the largest possible measure such 
 unavoidable defects is the reason for what may appear to be 
 prolixity. 
 
 A sufiiciency of syntactic evidence, within such limitations, has 
 been thruout my desire for the various periods and texts selected ; 
 at the same time I have purposely refrained from numerical tabu- 
 
 1 Cf. Vgl. Syni. I. §§ 126, 143, and Grundriss"^, §§ 479, 491. Celtic, Arme- 
 nian, and Albanian have been left out of consideration as much for apparent lack 
 of germane material as for insufficient acquaintance with them. 
 
lation as being but thankless drudgery in cases where the state of 
 affairs it might illumine is satisfactorily established. In a large 
 sense these pages do not aim at setting the stamp of finality on 
 the problems under discussion from an entirely new point of view, 
 so much as to indicate clearly where there is ample ground for a 
 departure from the opinions still adhered to; nor has their pur- 
 pose been the chimerical one of solving every riddle at all costs, 
 but rather a definite arrangement, agreeably to the results of 
 present-day scientific research, of whatever in this field at all lent 
 itself to systematic exposition. The number of literary citations 
 has thus been regulated with this particular usefulness in view. 
 
 One word more, in violation of the ancient injunction, " Nee 
 debet prologus enormior esse quam fabula." That thruout the 
 course of this work I have laid many writers under contribution 
 will be fully evidenced by the footnotes. Besides such general 
 acknowledgment, however, I wish to single out for special mention 
 Brugmann and Delbriick as being those from whom I have derived 
 the greatest assistance. I may here and there have taken issue 
 with them on some otherwise hopeless questions, still they have 
 given me not only the much-needed grounding, but, in a larger 
 measure, that inspiration without which these pages could scarcely 
 have been undertaken. My manifold shortcomings, it is needless 
 to add, are the results purely of my comparatively short novitiate. 
 Of those who have lent more immediate aid I am indebted in the 
 first instance to Professor Arthur F. J. Remy for his kindly inter- 
 est in all my work and for the helpful counsel which he has ever 
 been ready to give. Sincere thanks are also due to Professors 
 A. V. Williams Jackson, W. W. Lawrence, and E. H. Sturtevant 
 for generously placing their expert knowledge at my disposal, and 
 to Professor Gonzalez Lodge, besides, for the loan of books other- 
 wise inaccessible. Finally, it affords me pleasure to thank Pro- 
 fessor F. A. Wood of the University of Chicago for encouragement 
 at the inception of this work and Dr. Louis H. Gray for valuable 
 suggestions at its close. 
 
 ALEXANDER GREEN. 
 
 Columbia University, 
 July, 1913. 
 
CONTENTS 
 
 PAGE 
 
 Foreword ix 
 
 CHAPTER 
 
 I. The Concept of Agency and the Passive . . . i 
 
 II. The Force of the Dative and of the Instrumental io 
 
 III. Syncretism of the Dative and of the Instrumental i8 
 
 IV. The Dative of Agency in Latin 32 
 
 V. The Instrumental of Agency in Slavic ... 44 
 
 The Genitive of Agency in Lithuanian ... 44 
 VI. The Dative and Instrumental of Agency in Indo- 
 
 Iranian 51 
 
 VII. The Dative-Instrumental of Agency in Greek . . 63 
 VIII. The Dative-Instrumental of Agency in the Germanic 
 
 Languages 78 
 
 {a) Gothic 81 
 
 Q) Anglo-Saxon 95 
 
 {c) Old Norse 105 
 
 CONCLUSION 114 
 
 BIBLIOGRAPHY 120 
 
THE DATIVE OF AGENCY 
 
 CHAPTER I 
 
 The Concept of Agency and the Passive 
 
 § I. Delbriick's view of the fundamental meaning of the Nomi- 
 native will form the point of departure in these remarks, which 
 must of necessity precede a statement of the case. " In ihn trat 
 urspriinglich, says he,^ jedenfalls der als thatig gedachte, den Tra- 
 ger oder Mittelpunkt der Handlung bildende Substantivbegriff. 
 Erst nachdem sich der passivische Ausdruck entwickelt hatte, 
 konnte der Nom. auch zum leidenden Mittelpunkte der Handlung 
 werden vmd erst auf dieses Stadium passt daher die Erklarung, 
 dass der Nominativ den Gegenstand der Aussage, das grammat- 
 ische Subjekt bezeichnet." A simple declarative such as Fortes 
 fortuna adiuvat presents no difficulties whatever of analysis in lan- 
 guages where, as in the Indo-European family, we find a fully 
 developed, subjectively conceived verb and, in connection with it, 
 a clear concept of the subject and of the object as we understand 
 them to-day. There exists a large number of non-Indo-European 
 tongues which, lacking such a verb, cannot express the real sub- 
 ject in a manner natural to us. In the Malayan languages ^ the 
 verbal expression appears with a possessival agent : ' your search ' 
 stands for ^sought by you."* Others again, like the Thibetan,' 
 equally lacking finite verbs in a subjective sense, express this idea 
 of the bearer of the action by means of a nominal basis for verb 
 and an instrumental for agent : ' I strike you ' is put as ^ your beating 
 thru me.^^ Countless others endeavor to express the relation 
 between the action and the ideal subject by producing forms that 
 are almost pedantically specialized, in that they demonstrate how 
 the bearer of the action, i.e. the agent, is acted upon, instead of 
 
 1 V^l. Synt. I. § 71. ^ Winkler, Zur Sprachgeschichte, p. no. 
 
 8 Ibid. p. 76. * Cf. also G. v. d. Gabelentz, Die Sprachwissenschaft\ p. I02. 
 
grasping the fact so elementary with us, viz. that the agent itself, 
 as the real subject, is to be emphasized as the bearer of the idea 
 designated by the sentence.^ 
 
 § 2. In one sense, however, sufficiently denoted perhaps as ter- 
 minological^ the Indo-European languages are equally deficient in 
 clearness. The Nominative in them does not fully cover the idea 
 of the subject of the action, and Panini's unwillingness to conceive 
 of it as the subject-case, — even tho due to an entirely non-germane 
 reason,^ — is nevertheless interesting in that the Nominative is not 
 the exclusive subject-case. " Der Nominativ — according to Mik- 
 losich ' — bezeichnet dasjenige, was die durch das verbum finitum 
 ausgedriickte thatigkeit vollzieht, undMas man minder genau mit 
 einem der logik entlehnten ausdrucke subject nennt. Man soUte 
 es, da man den ausdruck nicht entbehren kann, grammatisches 
 subject nennen." The Nominative alone, as the form of the agent, 
 is but a potential agent ; the idea of the agent in the real, factual 
 sense, as distinct and separate from a mere formal concept, the 
 logical subject in short, is one whose designation for the science of 
 language is by far the more important one. 
 
 § 3. The grammatical and actual subjects may and often do 
 coincide, but to regard them as equal without further ado is pro- 
 hibited by the circumstance that the subject-nominative of passive 
 constructions is bereft of all connotation of agency.^ In such con- 
 structions the passive subject, the one acted upon, is the grammat- 
 ical, the active the logical. The situation of the latter is as follows : 
 The action of the subject affects some other person or object. This 
 is Active construction. When, however, this person or object is 
 raised to a subject in that we regard the action from the opposite 
 point of view and turn our attention towards the origin instead of the 
 end of the action, we obtain a Passive construction with a logical 
 subject : fortes afortuna adiuvantur} This logical subject need not 
 necessarily be made known, as when we say ' a down-trodden race^^ 
 or locutions like ' dicunt, Xeyovariv, man sagt, on dit ' in various lan- 
 guages ; or, even if known, need not be expressed. Thus, whilst 
 it is safe to assume that in dTi/aa^eTat the logical subject is 6 ari/xa- 
 
 1 Cf. Winkler, ubi supra, p. 137. 2 ygi Synt. I, § 64. 
 
 ^ Vgl. Grammatik d. slav. Spracheriy IV. 344. 
 
 * Cf. Rodenbusch, Bemerkungen zur SatzlehrCy i. Die Agensform als Subjekt, 
 IF, 19, 254 ff. 
 
 ^ Cf. Grimm, Deutsche Gram. IV. 3. 
 
3 
 
 ^wv, the latter may be left out altogether. This omission is par- 
 ticularly common with impersonal expressions.^ On the basis, 
 however, of a synoptic view of all available languages of the globe, 
 H. V. d. Gabelentz ^ declares that the completely developed passive 
 is that which is not only personal but has the nomen agentis as 
 well made clear and definite. 
 
 § 4. We have seen that logically there is no difference between 
 the Active and the Passive ; the contents remain essentially the 
 same, tho the form be changed. Intrinsically, therefore, the Pas- 
 sive is a linguistic luxury, used for the sake of variety and color 
 attained thru the change of subjects,^ and one may well wonder 
 with H. V. d. Gabelentz — Mauthner's characterization of it from 
 the stylist's viewpoint as barbarisch * having its own interest — 
 " dass so viele ganz verschiedene Sprachen sich in der Ausbildung 
 einer solchen Form begegnen, fiir welche eine innere Nothwendig- 
 keit nicht vorliegt."^ It is the general opinion of philologists that 
 the primitive Indo-European language (denoted hereafter as I-E.) 
 had no especial form for the Passive ; at least, with all the means 
 of comparative philology at our command, it is difficult to prove 
 the existence of one.^ That, as far as the passive category went, 
 the medial forms of the verb were employed in such a function, is, 
 however, concluded on the combined testimony of Indo-Iranian, 
 Greek, Germanic and Italo-Celtic. The medium designated that 
 the subject was in some way attracted by the action into a 
 sym-pathy, as Xovw ' I wash,' but Aovofiat * I wash myself or part 
 of my body,' rots ^pasi for instance. All of these languages con- 
 tain I-E. medial terminations in more or less modified forms and 
 variety, Sanskrit, Avestan and Greek being nearest to the original 
 forms, Italo-Celtic perhaps the farthest from them ; the Balto-Slavic 
 
 1 To which cf. Miklosich, Vgl. Gr. IV. 364, and Denkschriften der Kais. Aka- 
 demie der Wissenschaften^ vol. XIV. 
 
 2 Uber das Passivum, p. 540 ff. 
 
 8 De Sacy, Grammaire generate^ Ch. II, 
 
 * Krit. d. Sprache, III. 254. ^ Uber das Passivum, p. 455. 
 
 6 For details of the short account given here of the Passive, cf. the following 
 authorities: Delbriick, Vgl. Synt. II. p. 205 ff., 432 ff.; Thumb, p. 279 ff., 397 ff.; 
 Synt. Forschungen (Delbriick), V. 228 ff.; Reichelt, p. 298; Brugmann, Griech. 
 Gr? p. 458 ff.; Hirt, p. 332 ff.; Synt. Forsch. IV. 67 ff.; Sommer, p. 507 ff., 527 
 ff.; Miklosich, IV. p. 264, 830 ff.; Vondrak, II. p. 183 ff.; Schleicher, p. 99, 
 303 ff.; Kurschat, p. 286, 294 ff.; Grimm, IV. 9 ff.; Streitberg, Urg. Gram. § 212, 
 Got. El.^ p. 137 ff. 
 
cannot be considered at all in this connection, and of the Ger- 
 manic only Gothic has a present-stem medio-passive category. 
 
 § 5. Leaving aside the new formations in Sanskrit and Greek 
 and, of course, all analytic substitutes such as in Germanic or 
 Balto-Slavic,^ the passive verbal forms of the Indo-European lan- 
 guages may be traced back to medial forms. The inference is that, 
 to just such an extent, the passive must have functionally been 
 deduced from the medium. Since, however, we find the perfect 
 also used in a passive sense even in I-E. times, as witness Gk. kc- 
 KkrjTOL ' he is called ' and Sk. juhure ' they are poured,' and very 
 early combined even with the doer of the 'fait accompli ' as Z 56 
 TrcTTon/Tat . . . Trpos Tptuwv, we must grant the perfect also its own 
 share in the development of the passive. What interests us most, 
 however, is the r61e played by verbal adjectives, such as Gk. -ros, 
 
 1 The new Indo-Iranian passive in -ya from intransitives of the to class is fully 
 developed in Sanskrit, but has deteriorated in Avestan. In making use of the ana- 
 lytic expression Avestan requires ah- or bav- with the p.p.p, but Sanskrit regularly 
 omits the copula as tena sarah samdsdditam * by him a pond was reached ' — the 
 highest imaginable development of the passive expression according to Gaedicke, 
 Akkusativ, p. 42. — In Greek the competition of the act. aorist in -drtv with the 
 medial aorist in -utiv (Homeric iKrddrjv beside passival iKTd/injv), resulting in the 
 defeat of the latter, goes back to I-E. beginnings. Specifically Greek, however, 
 is the use of the act. intr. forms in -r)v in passive sense. — Since the Slavic medio- 
 passive has disappeared, with the exception of the p.p.p. and remnants of the 
 pres. p.p. in OBg. and Russian, the OBg. passive is expressed either by means of 
 a newly formed medium reflexivum, nareceth se KKrjdi^aeTai, Mt. 5, 19, or periphras- 
 tically thru the present or preterite pass. part, and dy^i. Periphrastic Lithuanian 
 regularly omits the auxiliary, employing simply the p.p.p., as kardliaus siustas^stni 
 by the king.' — The new personal endings in Latin whose chief characteristic is r 
 might be taken, together with their Celtic parallels, as related to certain active- 
 medial 3rd pi. r-forms in Indo-Iranian, as Sk. babhicv-ur and Av. bdbv-ar', ■< bhu. 
 (Sommer, p. 529; Zimmer in KZ. 30, 224; Pedersen, KZ. 40, 164.) — Outside 
 of the defective medio-passives in Gothic and perhaps an isolated instance in ON. 
 keite, 1st sg. pres. (cf. Sievers, PBB. 6, 561) there are no certain traces in Ger- 
 manic of the I-E. endings. Germanic as a whole has adopted the periphrastic 
 passive with *werpan and *wesan. Entirely new is the ON. reflexive formed, 
 like in Balto-Slavic (cf. Lithuanian vadintis KkKKf\Ta.C) and the Romance lan- 
 guages, of the active verb and the reflexive pronoun, kallask *he is called.' 
 According to the evidence of Greek and Sanskrit a similar formation must have 
 been competing even with the I-E. medium. Cf. Gk. dtroKp-uirTfii iixavrbv beside 
 diroKp^TrTOfiai,, Od. t, i6o; Sk. ned dtmdnam vd prthivim vd hindsdni 'lest I 
 injure either myself or the earth,' Sat. Brdhm, I, 2, 4, 7; cf. also Bopp. Vgl. Gr. 
 p. loii; S.F. IV. 69 and S.F, V. 262. 
 
Sk. -tds^ in this process. Originally indifferent as to diathesis,^ as 
 intransitive pyno^^ srutds but passive ttctttos, paktds, coctus, this per- 
 fect participle was especially well fitted to express passivity because 
 while, as a participle, it denoted a being or object at rest, as a 
 perfect tense it designated the attainment of a state or condition of 
 affairs as well. It is not to be wondered at that we observe it not 
 only forming the analytic passives with auxiliary verbs in Sanskrit, 
 Avestan (cf. Greek AcXv/xeVos w, d-qv etc.), Balto-Slavic, Latin and 
 all of the Germanic dialects, but also, since the earliest periods, 
 appearing with expressions denoting the agency of the action. 
 
 § 6. The force of the pass. pret. participles in -to and -no thus 
 employed is variable. According to their origin they are both 
 adjectives and verbals. It must here be emphasized for the Ger- 
 manic that combined with auxiliaries they form a unit and are not 
 felt exclusively as predicate adjectives, contrary to Grimm, IV. 717, 
 but, by the very nature of the passive formation, are understood 
 together with the verb as a verbal predicate, denoting no more a 
 condition but an action.^ This their verbal nature is borne out by 
 the fact that when they are passives they can, like the finite passive 
 verb, take the same case as the passive verb itself: pitds^ ukids^ 
 opcKTo^ are passives in force, only a construction like pitfbhir dattdh, 
 cf. RV. 10, 107, I, no more exists in Homer.' But to denote pos- 
 sibility the p.p.p. shows its verbal character in that it may take an 
 instrumental dative or a preposition to express the agency, N 323 
 XaA.K(j) r€ prjKT6<s fieyaXoiac re ^apiULhioun ; SO ^ 568, I 526, V 65. So, 
 too, the Sk. verbal adjective in -ya is more verbal than adjectival, 
 for it may be compounded with the same preposition as the verb 
 to which it belongs, and in the passive may be combined w4th 
 the instrumental of agency.^ There is no doubt in my mind that 
 the modern German use of such compound passives as ' ist — 
 worden,''^ with their past participle strongly adjectival, has been 
 largely responsible for the wholesale interpretation in similar vein 
 of the older Germanic participles. A prepositionless nomen agentis 
 of an ambiguous sort expressed with these younger passives would 
 not, of course, admit of any other interpretation but that of a pure 
 dative of personal interest with adjectives. The evidence of the 
 
 1 Cf. also Brugmann, IF^ V. 117 f. 
 
 2 Cf. Kohler, Germania, XI. 287. 
 
 8 Grdr> IV. § 160, p. 484. 4 S.F. V. 396. 
 
 s * Ich bin gebunden worden,' cf. Wilmanns, Deutsche Grammatik, 3I, § 76. 
 
6 
 
 other allied languages shows, however, that for the older stages, at 
 least, of the Germanic dialects such a viewpoint is decidedly one- 
 sided and all but bears the stamp of purposeful isolation. 
 
 § 7. In general, following one of the foremost authorities, a two- 
 fold division may be established as to the syntactical relation of 
 such nomina agentis to the passive verb, according to the mode of 
 thinking of the various languages. They may conceive of the per- 
 son or object either as causing the action or as sources from which 
 the action originates. In the words of H. v. d. Gabelentz,^ " da 
 wo man das Verhaltnis der Richtung, in welcher jede Thatigkeit 
 sich aussert, iiberhaupt nur umgekehrt, also statt des Punktes, wo- 
 ven sie ausgeht, den Punkt, wohin sie geht, zum Subject des Satzes 
 erhoben hat, wird jener — das Nomen agentis — folgerichtig in 
 einem Casus stehn, welcher der Frage woher? entspricht, und durch 
 den Ablativ oder durch eine Proposition ausgedriickt werden, wel- 
 cher die Bedeutung unseres von beiwohnt." Types of the Latin 
 ab^ Romance de^ da etc. are meant, as well as similar usages in 
 non-Indo-European tongues, such as the Hungarian, Tcheremiss, 
 Chinese and various Australasian dialects.^ " Wo aber das Nomen 
 agentis bestimmter als die Handlung veranlassend, als wirkend 
 Oder thatig gedacht wird, da tritt an die Stelle des Ablativs der 
 Instrumentalis oder eine denselben ausdriickende Praposition." 
 These may be taken as the two comprehensive rules of agency to 
 which there are numerous exceptions among the languages of the 
 globe.' 
 
 § 8. It will be noticed that v. d. Gabelentz narrows himself down 
 to but two alternatives. This is just the casus causae et controver- 
 siae. Where are we to place the so-called Dative of Personal Agency 
 which we find represented in so many of the Indo-European lan- 
 guages ? How are examples such as the following to be construed ? * 
 
 Sanskrit: sdkhibhya tdyah^ 01 
 
 prd mt pdntha devayand adrsran^ RV. 7, 76, 2. 
 Avestan: yesnyo vahmyo visppmai aTduhe astvaite, Yf. 5, i. 
 
 yahmai xsnuto bavaiti, yahmai tbisto bavaiti, Yt 10, 
 
 87. 
 
 ^ Uber das Passivum, p. 540. 
 2 Cf. also Hubschmann, Zur Casuslehre, p. 1 24. 
 
 8 For non-I-E. languages I again refer to Winkler, Zur Sprachgeschichte, p. 
 75 flf. et passim. 
 
 * Cf. Vgl. Synt. I. § 143 and Grdr?- II. § 491. & S.F. V. 396. 
 
Latin: meditata mi hi sunt omnia mea incommoda^ Ter. Phorm. 
 248. 
 arcus subspiciunty mortalibus quae perhibetur Iris, Enn. 
 Ann. 409.^ 
 
 but especially 
 
 Greek: roo-avra \i.oi elp-qaOo), Lys. 24, 4, and 
 8afi€v "EKTopi Stft), ^ 103. 
 
 as well as of the Germanic dialects : 
 
 Gothic: ei gaumjaindau mannam^ Mt. VI. 5, ottws av ^avwo-tv 
 
 </«/ saiJvan im^ Mt. VI. i, tt/oos to QtaBr\va.i avroi? 
 Anglo-Saxon : wear'd da him inweardlice gelufod, Aelfric's Horn} 
 
 fromcynne folde weorde'd />tne gefylled^ Caedm. Gen, 1765. 
 Old Norse: Nordimbraland var mesi byggt Nordmgnnum, 
 Fornm. I. Ch. 15. 
 
 vask forimr verum vegin at husi, Ghv. 10*. 
 
 § 9. The general opinion held of these datives — the Germanic 
 varieties have not yet been discussed in this relation — is that they 
 are in and for themselves a mere subtype, " Abart,"' of the dative 
 of personal interest. " Dieser Dativ ist natiirlich ein Dativ der 
 betheiligten Person, wirkt aber als Agens."* Sdkhibhya idyah 
 thus really means * to be praised /7r the friends ' ; that it manifests, 
 besides the force of the dative, a strong auctorial self-assertion — 
 we must render it * by the friends ' — is to be attributed not to the 
 dative but to its connection with the verbal noun which itself con- 
 tains the idea of necessity.^ So Brugmann,^ " Dass die interessierte 
 Person zugleich als Vollzieher der Handlung zu denken sei, ergab 
 nur der Zusammenhang." 
 
 § 10. It is not easy, however, consistently to extend this distinc- 
 tion with reference to all the instances and categories available in 
 the above languages. There are certain well-defined examples 
 which refuse even on the closest scrutiny to reveal a connotation 
 
 1 The references thniout these pages are to R. Y. Tyrrell's edition of Terence, 
 Oxf., 1902, and J. Vahlen's Ennianae poesis reliquiae^ Lipsiae, 1903. 
 * Thorpe's ed. I. 1843: Assumption of St. John^ 1. 16. 
 
 « Landgraf in Wolffiin's ArchiVy 8, 39. * Delbriick, Vgl. Synt. I. § 143. 
 
 6 Ibid. p. 297. 6 Grdr?- II. § 491. 
 
8 
 
 of commodi or incommodi whether inherent in the dative or induced 
 by its connections.^ A Ciceronian dissimillimis bestiolis com- 
 muniter cibus quaeritur, Deor. Nat. II. 48, 128, may indeed admit 
 of but one view, but, to keep to Latin and Cicero for the moment, 
 already an example like Tusc. II. 4, 15 est igitur ad hunc modum 
 sermo ilk nobis institutus demands more than a casual glance for 
 final disposition. Furthermore, how are we to distinguish between 
 an oft-encountered alicui iudicatum est and ad Att. 8, 3, 7, legionem 
 Fans to conscriptam where, in view of the context, one clearly 
 cannot talk of a dative of interest, since Faustus raised the legion 
 not for himself, but for Pompey. — So, too, in Greek, while there 
 may be no difficulty at all about y 138 t<5 8c kc vtKiyo-ai/Tt <^t'A»y 
 K€K\-^(TYf ttKotTis and still less, once its bearings are defined, about 
 Herod. 7, 168 ov <T<f>L TrcpLOTrrirj ia-rl -^ 'EAXas ajroWv/xevr}, can we 
 translate N 168 S6pv jxaKpov, 6 ol KkKrC-q^i XfXtmro with Monro,^ 
 simply as 'which for him was left in the tent,' when the warrior 
 who was the beneficiary of this act was also the very same Meriones 
 who had left the spear behind? And is S 177 ava(T<rovTai h^ifioi 
 avTio, where the * interest '-element is clearly not in the agent, to 
 be put in the same category with y 138 above and its likes ? Meet- 
 ing with such examples in Greek literature, one must pause and 
 inquire if they all should be classed as having an exclusively 
 datival connotation or whether, with due regard to the constituent 
 elements of what we call the Greek ' dative,' we should not be per- 
 mitted rather to fix our attention upon more logical sources, namely, 
 upon the instrumental whose functions in Greek — to postpone the 
 discussion of the Germanic features — have fused into those of the 
 dative. 
 
 §11. Latin and Greek have been selected in the foregoing para- 
 graph, because they typify the whole situation with respect to the 
 Dative of Agency. In the one we find an instance where the func- 
 tions of the Dative of the I-E. mother-tongue seem to have been 
 preserved substantially uncontaminated. In the other, however, 
 
 1 Right here, to avoid a misunderstanding, the remark must be appended that, 
 where such datives fail to disclose the idea of advantage or disadvantage, it by 
 no means follows that they could not have developed their apparently exclusively 
 auctorial connotation from that of pure personal interest probably even before the 
 dialectal separation of the Indo-European languages. This will be made clear in 
 the succeeding chapters. 
 
 2 Grammar of the Homeric Dialect, § 143, 5, 
 
9 
 
 we can point to numerous traces of a usage, namely that of the 
 Instrumental, which is essentially foreign to the Dative as we con- 
 ceive of it at its inceptional stage. The problem that confronts us 
 then is to investigate as far as possible to what extent we may in 
 the leading Indo-European languages resolve into its constituent 
 elements a usage common to two cases that are vitally different 
 from one another but which have, in the course of linguistic devel- 
 opment, met on several common grounds and have in a variable 
 measure merged before historic times. Specifically, to find out in 
 what languages the so-called Dative of Agency may be attributed 
 exclusively to the expansion of the original Dative itself and in what 
 instances it may be traced back to an original Instrumental function. 
 To this end it will be necessary to discuss fully the force and use 
 of both these cases. 
 
CHAPTER II 
 The Force of the Dative and of the Instrumental 
 
 It must be remembered that we are in the dark as to how the 
 eight I-E. form-groups, called cases, came to acquire their termina- 
 tions, hence our comparative ignorance of their basic significations.^ 
 
 The Dative. § 12. The fundamental meaning of the I-E. Dative, 
 as far as we are justified in seeking for one, i.e, abstracting it from 
 the various functional types we know,^ is thus still a matter of con- 
 troversy, and this entirely aside from the general character attrib- 
 uted to it in chronological sequence by Empiricists of the Sanctius 
 school, Categorists of the Hermann type, and Logicians like Michel- 
 sen. The statements of these theorists are to-day merely of historic 
 interest. For our purposes the beginning is made by the Localists. 
 According to Hartung,^ to whom the spatial relations natural to the 
 human mind, viz. those of Wherefrom ? Where ? and Whereto ? stood 
 for the raison d'etre of our oblique cases, the Dative w^as that which 
 assumed the middle category of absolute rest and answered to the 
 question Where? Fourteen years after him Rumpel* denied that 
 the chief cases of the Indo-European languages served local rela- 
 tions and, as against his predecessor's mistaken attempt to transfer 
 the laws of thought into the realm of speech, declared them to be 
 purely grammatical. The Dative thus was designed simply to ex- 
 pand the pure sentence-matter, Satzsubstanz, represented by the 
 union of subject and predicate, whenever it was desired to express 
 the fact that the action denoted by such a ' substance ' was for the 
 benefit or disadvantage of some one. " Dies und nichts anderes 
 
 1 Cf. IF. 17, 40 ff. 
 
 2 « Was man gewohnlich die Grundbedeutung oder den Grundbegriff der Ka- 
 sus nennt, ist der Gebrauchsumfang, den sie in derjenigen Zeit der idg. Urgemein- 
 schaft batten, in die wir durch Vergleichung der verschiedenen idg. Sprachen 
 miteinander zunachst zuriickkommen." Brugmann, Griech. Gr? p. 374. 
 
 3 Ueber die Casus, ihre Bildung und Bedeutung in der griechischen und latei- 
 nischen Sprache, Erlangen, 1831. 
 
 * In his Casuslehre in besonderer Beziehung auf die griechische Sprache, 
 Halle, 1845. 
 
 10 
 
11 
 
 meint man mit dem bekannten Terminus, dass der Dativ die be- 
 theiligte Person bezeichne."^ Rumpel arrived only at a half-truth 
 because he based his investigations on Greek alone, but in the main 
 his findings have not suffered radical revision. 
 
 § 13. It is true that in his treatise De usu dativi in carminibus 
 Rigvedae ^ Delbriick made out a presentable brief for the reconsider- 
 ation of the Dative from the localistic point of view, insisting, how- 
 ever, that the question be altered from Where? to Whereto? and 
 that the Dative be felt as a casus finalis denoting " die korperliche 
 neigung nach etwas hin,"' but in his later works * he turned to the 
 conception advocated by Hiibschmann who, while in favor of the 
 Dative as a final case because of its partly final uses, is nevertheless 
 inclined to see in it a purely grammatical case of the indirect object 
 "welchem die Aussage gilt."^ The pivot of contention this time 
 was the fact that the I-E. dative did not combine with prepositions, 
 something which it would not have failed to do had it originally 
 been a local case. Sanctius already realized this with respect to 
 the Greek and, while Delbriick was the first * to extend this obser- 
 vation to Indo-European, he failed to draw the correct conclusion 
 from it.^ In accordance, however, with Hiibschmann's reasoning, 
 but especially because of Gaedicke,^ he modified his former views 
 even as to the Vedic dative.® As the Dative now stands, viewed 
 not syncretistically in the form of its several constituents, but in its 
 original force, it is the case of interest, overwhelmingly of personal 
 interest.^® 
 
 § 14. There is to-day, at all events, an interesting recrudescence 
 of localism. The arguments of Hirt,^^ supported as they are by the 
 uses of the dative in Slavic,^'^ are the more plausible since dative 
 and locative seem to have originally been identical in form and 
 dififerentiated only in accentual respect, as Sk. dzX.piire, \oc. pitdrt, 
 
 1 Casuslehre in besonderer Beziekung auf die griechische Spracke, p. 171. 
 
 2 Halis, 1867; given in substance in KZ, 18 (1869), ^Ueberden indogermani- 
 scken, speciell den vedischen dativ! 
 
 8 KZ. ubi supra, p. 100. * First in S.F. IV. 53. 
 
 S Zur Casuslehre, p. 137, 213. ^ j^z. 18, 106. 
 
 7 Cf. also Pischel, BB. I. in. 
 
 8 Uber den Akkusativ im Veda, Breslau, 1880, p. 137 ff. 
 ® Altindische Syntax, S. F. V. 140. 
 
 10 Cf. also Winkler, Zur Sprachgeschichte, p. 193 ff.; Uralalt. Volker u. 
 Sprach. ad loc; Zur idg. Syntax, p. lo ff.; Germanische Casussyntax, p. 541. 
 n Handbuch, § 301, etc. 12 Miklosich, Vgl. Gr. IV. 576. 
 
12 
 
 the dative suffix -at seems to be Vollstufe to the local -i, i.e. origi- 
 nally we have a dat. *paterai and a loc. ^paterai} That there is 
 ample ground for a partial disagreement as to the Vedic dative is 
 evidenced by Whitney;^ PischeP interprets it as a terminative 
 case, Speyer* as both a "Z/^" and a "^2^r"-case, Hopkins^ as a 
 case of direction. Solmsen® holds that the I-E. case-form called 
 Dative must originally have had both the ending -ei and that of -di^ 
 consequently it was probably a combination of two once sharply 
 differentiated cases, namely that of interest and also of finality, 
 ibid. p. 167. Brugmann himself^ is not disinclined to a reversion 
 in some details to the older view. For the Germanic cf. Dietrich 
 in HZ. 13, 128, the Latin, Deecke in Progr. Muhlhausen^ 1890, 
 p. 31 f., and more recently Gustafsson, De dativo Latino^ Helsing- 
 fors, 1904, based mostly on verbs of motion. 
 
 § 15. A far graver charge to my mind, as far as fundamentals 
 are concerned, is that the Grundbegriff given by Hiibschmann may 
 be somewhat too abstract for the beginnings of human language.^ 
 Even if we grant that extreme primitiveness as such cannot very 
 well be attributed to the Indo-European linguistic unit immediately 
 before its subdivision, still we feel obliged to second Mourek, in 
 reviewing Delbriick's Synkretismus^ " Solche geistig hoher stehende 
 griinde sind fur die urzeiten undenkbar, da muss man von roheren, 
 ausserlicheren anschauungen ausgehen, und das ist die raumliche, 
 von welcher man viel leichter zu der weiter entwickelten interessen- 
 auffassung gelangt als umgekehrt." So Siitterlin against an anti- 
 localistic view in Wesen der sprachlichen Gebilde, Heidelb. 1902, 
 p. 108. 
 
 § 16. As far as the development of the function of agency is 
 concerned, however, we are obliged to begin with that of interest. 
 Granted, as above, that the latter was not a Grundbegriff but only 
 a secondary and derived stage, it is quite an effort to connect inter- 
 est with a local basis, — the easy interpretation of the dative as such 
 
 1 For a dat. sg. *-oia^ cf. Bartholomae in Sitzungsberichte, 19 10, 5 der Heidel- 
 berger Akademie der Wissenschaften^ phil.-hist. Kl. and a resume of it in IFA. 
 29, 40 ff. 
 
 2 AJP. 13, 285. 3 ubi supra. ^ VuSS. 1896, p. 13. R 
 
 6 TrAPA. 37, 87 ff. and //^ (95. 28, 360. « j^z. vol. 44, p. 161 ff. 
 
 7 Grdr.^ II. p. 474. 
 
 8 Holzweissig, Uber den soc.-instr. Gebrauch des Dativs bei Horner^ Progr. 
 Burg, 1885, p. 7. 
 
 9 Zs.f. d. Alter t. 38 (1908), p. 117. 
 
13 
 
 a local case is made possible only where the idea of finality is 
 already present in the verbal idea, otherwise it appears violently 
 forced, — let alone the explanation of the dativus ethicus and the 
 dativus auctoris from the same source;^ these two types may be 
 said to be the strongest proofs against the dative as a Wohin-c2iS^} 
 But from the dativus commodi vel incommodi^ the dativus auctoris 
 may easily be deduced as but one of the looser relations which the 
 dative case bears to the predicate of the sentence, in that with pas- 
 sive verbals the person interested in the action may be the identical 
 one who also performs the action. The sentence-substance would, 
 indeed, be complete without such an addition, but once it is desired 
 to denote clearly for whom or at whose expense^ advantage or detri- 
 ment something occurs, the dative of agency is the most natural to 
 subjoin. Thus employed it is stronger than the fundamentally 
 allied dativus ethicus (Buttmann) as when Horace, Ep. I. 3, 15, says, 
 Quid mi hi Celsus agit?, stronger even than a pure dative of inter- 
 est. In Cic. in Verr. II. 8, 22 Verres hunc hominem Veneri absol- 
 vit, the action is to all intents and purposes the acquittal of the 
 man ; that this acquittal is on behalf of or in relation to Venus, is 
 without the bounds of immediate consideration. Not so, however, 
 Ter. Phorm. 2, i, 18 meditata mi hi sunt omnia mea incommoda 
 where the dative is part of the warp and woof of the predicated 
 substance and designates as much the person that has caused the 
 action as the one who is a party interested therein. The construc- 
 tion in its purest form is reached in instances where the former 
 concept is the sole and dominating one, as when Sallust writes, 
 Jug. 107, I saepe antea paucis strenuis advorsum multitudinem 
 
 1 Gaedicke, op. cit. p. 134. 
 
 2 Of course, it is not absolutely impossible for the dat. commodi to have devel- 
 oped from a basic local relation, witness the following schemes, (a) *iibi venioy 
 (b) *adiutor tibi venio^ (c) adiutor tibi sum, or (a) *i\d€iv rivi, 'to or for 
 one,' (b) id., *for one,' (c) dyaa-rijpal rivi. 
 
 ^ Rumpel, o/>. cit. p. 286, acutely observed that this term is grammatically un- 
 tenable : all that the dative expresses in this use is that something, be it a person 
 or an object, is interested in the action, participates in it to such and such an 
 extent, whether however to its advantage or disadvantage, is evidenced only by 
 the context, not at all postulated by the grammatical form itself. So Hiibschmann, 
 op. cit. p. 71. The Greek dLvaaTrjval tlvi is * to rise,' but rivl itself does not in- 
 dicate whether this rising is out of regard for some one or, indeed, against him as 
 in anger. Cf. also the Latin Cic. in Verr. II. 8, 22, Verres hunc hominem 
 Veneri absolvit, sibi condemnat, as expounded in Haase, Vorlesungen, II. 147. 
 The term, however, is retained because of its convenience. Vgl. Synt. I. p. 296. 
 
14 
 
 bene pugnatum, rendered clear by the identical construction in 114, 
 I with ab ducibus nostris^ or Virg. Buc, VI. 72 ara quae maxima 
 semper dicetur nobis, i.e. Xcyctv, or Propertius IV. 14, 41 prata 
 cruentaniur Zetho, about all of which anon. The fact stands clear 
 that the dat. commodi has in it the potentialities of a development 
 into a dat. auctoris. So much so that in all Indo-European lan- 
 guages a dat. of the person, even when it accompanies the nomina- 
 tive of a verbal substantive, is felt as the logical subject of the 
 action : mir ist sorge = ich sorge mich, O. Lat. quid tibi hanc tactio 
 est = quid hanc tangis, and similarly in Greek, Slavic and Sanskrit.^ 
 
 The Instrumental. § 17. The Sanskrit Instrumental — and to 
 the Sanskrit we continue to go back for our syntactical bearings — 
 contains three distinct ideas, (a) prosecution, (b) association, {c) in- 
 strumentality. Schleicher assumed ^ for the original I-E. instru- 
 mental ^two entirely different suffixes, -a and -bhi, cf . Gk. afia, raxa 
 and Homeric -<^t and -<^iv, and accordingly postulated two originally 
 even functionally differentiated cases, one expressing association, 
 the other — tho he is unable to make a sharp line of demarcation — 
 means or instrument.^ This would indicate that originally the 
 instrumental had a form distinct from that of the sociative, called 
 also comitative. Since, however, Delbriick * disclosed the fact that 
 the I-E. comitative functioned also as a prosecutive, practically all 
 the subsequent investigators agree with him in attributing to the 
 so-called Instrumental a primarily sociative force, with the idea of 
 means or instrument proper as much of a logical derivative from 
 this basic concept as the prosecutive for which there has been found 
 or proposed no separate case-form. Those who reject these con- 
 clusions differ not in ascribing another primitive ' Grundbegriff^ to 
 the instrumental but rather in refusing to attach to it any ' Grund- 
 begriff^ whatever. 
 
 § 18. A notable exception is Miklosich, who in a localistic man- 
 ner derives all of the types in question from an original prosecutive,® 
 
 ^ Jolly, Infinitivy p. 265. 2 Compendium^, 577. 
 
 8 This view of Schleicher's, doubted by Delbriick, Vgl. Synt. I. p. 184, note, 
 would give us eight original I-E. cases without counting the vocative, a number 
 which, tho greater than we accept to-day, is in turn less than the number demanded 
 by Miklosich, Vgl. Gr. IV. 449; cf. Hubschmann, op. cit. p. 127, note. 
 
 *^Z/.p.53. 
 
 ^ " diejenigen Theile des Raumes oder der Zeit, fiber welche sich eine Hand- 
 lung ununterbrochen erstreckt," ALL p. 50. 
 
15 
 
 "wie bei den iibrigen casus, so gehe ich auch bei dem inst. von 
 dessen raumlicher function aus."^ As this is merely a theoretical 
 point of difference and in no wise alters the fact that the instru- 
 mental is not original in its function of means, but owes its pro- 
 venience to some other function, we will not here pause to discuss 
 it, but merely point out the comparative difficulty of conceiving the 
 prosecutive as this fountain-head. 
 
 § 19. Panini already^ recognized, at least from the usage of 
 Sanskrit, that the instrumental lends itself to the designation of the 
 personal agent, in so far as it is not already included in the verb, 
 just as well as to that of material instrument.^ If the nominative 
 expressed the sentence-substance as devadattah pacati^ * D. is cook- 
 ing,' the instrumental could put it as devadatfena pacyafe tandulah, 
 'rice is being cooked by D.'* In a word the kartar could be ex- 
 pressed by the same case as that which denoted the karana, the 
 tool or means, as well, datrena lunaii *he cuts with the sickle.'* 
 That these were connected with the sociative rendered by the same 
 case, he indicates in II. 3, 19.® 
 
 § 20. As to the theory of transitions, by extending the notions 
 of concomitancy and accompaniment from spatial and even tem- 
 poral considerations — case-forms must " von jeher der Darstellung 
 ausserer, lokaler, temporaler oder sonstiger sinnlich anschaulicher 
 Beziehungen gedient haben,"^ — to more or less logical categories, 
 we easily reach the inst. of means as that qualificative with which, 
 in the company of which, the action takes place. The distinction 
 is somewhat that between living beings and inanimate objects.® 
 Consider, too, our English usage whereby * with^^ originally merely 
 associative — cf. Whitney's '' W^//>^ "-case — has developed into an 
 instrumental preposition ; similar is the connection between Ger- 
 man ' mW and * mitiels.^ That persons as well as objects, inanimate 
 
 1 Vgl. Gr. IV. 683. 2 Ed. Bothlingk, Leipzig, 1887. » I. 4, 50. 
 
 * In all that follows, the mention of the instrumental in such subjective function 
 is always in connection with the passive construction. Excluded, therefore, are 
 such uses of a subject-instrumental as occur in impersonal expressions in Avestan, 
 Geldner, KZ. 31, 319 ff.; Slavic, Miklosich, Vgl. Gr. IV. 352, 692; and even 
 Old Norse, Pedersen, KZ. 40, 138 ff., against whom cf. Neckel in IF. 21, 182 ff., 
 where, however, the verb is active, as Y. 48, l, yezi addii did drupm v^Tdhaifif 
 podnimdlo Vasilija svjatym diichom, lystr vindinum ofan i holit, very much like 
 the non-Indo-European expressions of agency indicated in Ch. I. § i. 
 
 5 I. 4, 42. ^ Cf. HUbschmann, op. cit. p. 143, note. 
 
 ' Grdr.'^ II. § 451. 8 Giles, Comp. Phil. 269. 
 
16 
 
 or personified, may also be thought of as instruments of action, 
 visvam so agne jayati tvdya dhanam, ' thru thee, Agni ' ; and that 
 this is quite as logically permissible in the passive as in the active 
 construction and thus a sdsydte kavibhih, *he is praised by the 
 singers,' is but a natural step from sdsydte vdcobhih, ' he is praised 
 with, by words;* are so many corollaries to be expected from this 
 circumstance.^ So that the Instrumental of Agency with passives 
 has its origin in the Inst, of Means with actives and, farther back, 
 in the Inst, of Association. 
 
 § 21. The instrumental with the passive — to neglect for the 
 nonce the possibilities of -tos and of the medio-passive — does not, 
 indeed, represent an original use, for the simple reason that the 
 passive expression was not native to the I-E. language, but devel- 
 oped in its very essentials in the various branches into which the 
 * Ursprache ' broke up.^ But even in this separate formation of 
 passives, the thus appended instrumental of agency originally de- 
 noted mere concomitancy.^ Thus when, for instance, the active 
 intransitive verb which, in terms such as * the house burns, ^ does not 
 represent the real causer of the occurrence, begins its course of 
 development towards the passive by attaching to its grammatical 
 subject the real subject of the action, this logical subject is, accord- 
 ing to Indo-European custom, placed in the instrumental. The 
 Greek aorist iSdfirjv signifies merely ' to be tame ' ; U-qXetoyvL Sa/xet?, 
 indeed, 'tamed, killed by P.,' but the latter is an extension of a 
 purely sociative connection, "gestorben unter Mitwirkung des 
 Peliden."* 
 
 § 22. This instrumental of agency based on an older inst. of 
 association and the dative of agency derived from an older dative 
 of interest bear the following relation to each other : In Latin, aside 
 from the prepositional phrases, only the dative is found expressing 
 agency with passive constructions ; in Sanskrit, however, as well as 
 in Avestan, there is both a dative and an instrumental of agency ; 
 in Slavic, — Lithuanian employs the genitive, — on the other hand, 
 the instrumental has been the choice ; lastly, Greek and the Ger- 
 manic dialects, so Gothic, Anglo-Saxon, and Old Norse, evidence a 
 dative which partakes of both a datival and an instrumental char- 
 acter. Save in the instance of these two doubtful branches, we are 
 
 1 Delbriick, S.K V. 135 and VgL Synt. I. § 123. 
 
 2 Cf. Ch. I. § 4, and Vgl. Synt. I. p. 184. 
 
 3 5. F. IV. p. 78. * Delbriick, S.F. IV. ubi supra. 
 
17 
 
 reasonably certain of the interrelations and delimitations of the 
 two means of agency along the whole line of Indo-European lan- 
 guages, and we shall later be in a position to follow up the various 
 steps thru which Agency is reached from its two sources here con- 
 sidered. The amalgamation^ of the dative with the instrumental 
 in these languages next requires our attention. 
 
 1 The tenn is found in Miles, Comp. Syntax ^ p. 31. 
 
CHAPTER III 
 
 Syncretism of the Dative and of the Instrumental 
 
 Syncretism in General.^ § 23. Since Bopp's investigations 
 disclosed that the I-E. parent language must have possessed eight '- 
 well-defined case-forms, viz. nominative, vocative, accusative, geni- 
 tive, ablative, locative, instrumental and dative, with separate case- 
 suffixes for all, save the vocative,^ and the works of succeeding 
 grammarians successfully postulated for each of the derived lan- 
 guages an originally identical number of cases, there have been 
 queries as to why this number was reduced in the various I-E. 
 dialects. Lassen ' first drew the correct inference as regards Latin 
 and Greek, " namlich ausser den sechs Lateinischen einen Instru- 
 mentalis und einen Locativ." Weissenborn's review of Madvig a 
 decade after was a distinct advance on this in that it declared that 
 this locative and instrumental were really expressed in Greek and 
 German by what was called the dative, and in the Latin by the 
 ablative.* Omitting names like Jacobs ^ which spell retrogression, 
 we meet with added confirmation in Pott who was also the first to 
 apply the term 'syncretistic,'® and in Curtius,' who proposed the 
 name '• Mischcasus^ in reference to case-forms and functions like 
 the Latin ablative and the Greek dative. It remained for Delbriick 
 to follow out these indications and, in reliance on Vedic Sanskrit, 
 to set down the norm for the ultimate analysis of the I-E. syncre- 
 
 1 A part of this sketch goes back in substance to Hubschmann, Zur Casus- 
 UhrCy pp. 74-93, and Zieler, Beitrage z. Geschichte d. Lat. Abl. 1892, pp. 6-8. 
 
 2 Cf. Brugmann, Grdr?' II. p. 474 ff. ; for the opposite view that the number of 
 cases was originally less than in Sanskrit, and that the specialization came about 
 after the so-called dialectal scission, see Diintzer, KZ. XVII (1867). 53. Ludwig, 
 Agglutination oder Adaptation^ 1873, also subscribes to this system of accretion. 
 
 3 Rheinisches Museum fur Philologie^ 1834, II. 148. 
 
 * Neue Jahrbiicher f. Phil. u. Pad. 1845, m* 34'- 
 ^ Zs. f. d. Gymnasialwesen^ 1847, f^* '°3* 
 
 * Prdpositionen^ 1859, p. 16. 
 
 ■^ Cf. Verhandlungen der 22 ten Versammlung deut. Philologen zu Meissen^ 
 1863, p. 49. 
 
 18 
 
19 
 
 tistic cases. His Ablativus localis instrumentalh im Altindischen^ 
 Lateinischen, Griechischen und Deutschen^ Berlin, 1867 answers 
 the question as to origins, premising the fact that there had been no 
 corresponding loss of case-functions to parallel that of the case- 
 forms. Modified in some respects, the reply is thus tabulated by 
 Htibschmann, p. 87 (cf. also Vgl Synt. I. p. 189-199): 
 
 I-E. 
 Skt. 
 Lat. 
 Gk. 
 
 Gr. 
 
 Dat. 
 Dat. 
 Dat. 
 
 Loc. Inst. 
 Loc. Inst. 
 
 Abl. 
 Abl. of ^-sterns 
 
 Gen. 
 Gen. 
 Gen. 
 
 Abl. 
 Dat. 
 
 Gen. 
 
 
 Dat. 
 
 
 Gen. 
 
 
 From this it is evident that while in Vedic Sanskrit locative, 
 inst. and abl. are separated, in Latin the ablative is the Mischcasus : 
 (a) gnaivod paired prognaius, (b) iota Asia vagaiur, (c) manu fecity 
 being all represented by it ; but in Greek it is the dative that com- 
 prises the instrumental and locative functions as well, the genitive 
 being the exclusive heir of the ablative^; in Germanic, too, the 
 dative consists of the pure dative, plus locative, plus instrumental, 
 and may in a manner be said to share the lost ablative with the 
 genitive.^ 
 
 § 24. A short history of the etiology of syncretism follows. The 
 first work of importance relative to it is the epoch-making ALL of 
 Delbriick. According to it' the reason for the reduction of case- 
 forms must be assigned to A) the competition of prepositions. Their 
 very being made the existence of case-forms precarious, in that 
 their constant attachment to the latter expressed the desired idea 
 with more exactitude than the bare forms. Attention to these would 
 be in inverse ratio to the gradual importance of such prepositions 
 as the more essential parts of the meaning, interchanges among the 
 deteriorating case-forms would follow and the resulting fluctuation 
 end in linguistic usage — quern penes arbitrium esi ei ius ei norma 
 loquendi — deciding in favor of one or the other of the alternatives. 
 
 ^ Cf. the use of the prepositions d7r6, /card, irapd, irpds, and iird c. gen. to denote 
 the place or person from which something originates. 
 
 2 For the latter statement cf. Synkreiismus, Vorrede. * ALL p. 77- 
 
20 
 
 For the rest, we know that the increasing use of prepositions is in 
 harmony with that tendency towards the analytic stage common to 
 the Indo-European languages. Holzweissig/ following Penka,'' be- 
 littles the destructive work of prepositions and insists that the chief 
 cause of the loss of case-forms was B) the force of sound and accen- 
 tuation laws maiming the terminations and giving rise to their 
 interchange and disappearance. " An erster Stelle hat nicht der 
 fiiessende Charakter der Kasusunterschiede, sondern lautlicher 
 Verfall das Entstehen von Mischkasus verursacht.'" It is needless 
 to point out that his stand is false.* Just as prepositions do not 
 always replace the case-forms they may have helped to displace 
 and, even tho necessary auxiliaries after case-fusions, they do 
 not always deprive the case-form of significance — witness the 
 Greek where it is often the case that determines the sense of the 
 preposition — , and just as the use of prepositions eo ipso has not 
 always sufficed to cause a confusion of cases, cf. the preservation 
 of inst. and loc. in Slavic in the face of concurrent prepositions,^ 
 so syncretism may, indeed, take place as a result of phonetic 
 changes, but it can come about without them and in a manner 
 purely C) syntactic, as in the instance of the Italic inst. sg.® So 
 others, we find, propose as an offset the confusing influence of 
 coincidences in meaning, " auf 6ine Determination sind durch Me- 
 tapher mehrere verwandte Beziehungsausdriicke iibertragen." ^ 
 There may be added to these the working of a principle of D) Im- 
 guistic economy such as Lanman * proposes to account for the confu- 
 sion of the Indie ablative plural with the dative. According to his 
 view the number of occasions requiring the expression of plural 
 ablative relations was so small that ultimately the ablative was de- 
 prived of a separate form and its function was added to that of the 
 nearest infrequent case, the dative. This, while in itself perhaps 
 insufficient to account for the mergence, is still interesting as one 
 of the probable minor causes of syncretism. 
 
 § 25. Our present views on syncretism stand at the junction of 
 B) and C. Delbriick himself has abandoned prepositions as the 
 
 1 Wahrheit und Irrtum der lok. Kasustheorie, pp. 19, 23, 69. 
 
 2 Die Entstehung der synkret. Casus, Wien, 1874. ^ ubi supra, p. 69. 
 
 * Vogrinz, Gedanken z. e. Gesch. des Kasussystems, Prog. Leitmeritz, 1884, p. 17. 
 
 ^ Audouin, De la declinaison, p. 423. 
 
 ^ Zieler, op. cit. p. 56. '' Vogrinz, ubi supra. 
 
 « On Noun-inflection in the Veda, JA OS. X. 583 ff. 
 
21 
 
 sole cause of case-amalgamation ^ and has, five years before Vog- 
 rinz, declared, at least in the instance of the Greek gen.-abl. for an 
 outer and inner motive of syncretism.^ We can therefore speak of 
 morphological and syntactical causes^ i.e. coincidence of form and 
 approximation of meaning.' As to the relative merits of each, it 
 would seem as if fusions due to phonetic alterations of final sounds 
 had not had sufficient influence in the reduction of cases at least in 
 Greek, Latin and Slavic* Again, the syntactic equivalences which 
 have contributed to such case-fusions seem to reach back to pro- 
 ethnic conditions, because their traces may be found in the most 
 conservative I-E. languages. All we know with certainty is that 
 all the languages did not syncretize the same way and, as we are 
 dealing with prehistoric conditions, we must take to heart Delbriick's 
 caution,^ " Man muss sich auch in dieser Beziehung nicht vermessen 
 wollen, das Gras wachsen zu horen."^ 
 
 The Dative and the Instrumental. §26. The process of 
 amalgamation just described, in full force during the two linguistic 
 periods, an older of eight-and-seven and a younger of six-and-five 
 case-forms — i.e. if we are to give credence to ALL pp. i and 75 — 
 has left for the earliest stages of the dialects here considered the 
 following relation between the datival and instrumental functions : 
 the Indo-Iranian languages have both a dative and an instrumental ; 
 similar is the condition of Balto-Slavic ; Latin has a dative, but no 
 instrumental form ; in the Greek and the Germanic dialects the 
 form serving as dative has assumed the instrumental functions as 
 well.^ It would lead us too far to set down — save for the last two 
 branches, where they are of eminent importance — the various mor- 
 phological and syntactical contacts that are still observable even in 
 those languages which have preserved these two cases as separate. 
 A careful survey of them, however, has brought us to the conclu- 
 sion that, syntactically at least, certain approximations reach far 
 beyond the earliest historical beginnings to definite proethnic I-E. 
 conditions. The following formulaic reconstructions exemplified 
 by actual facts will symbolize our meaning : 
 
 1 Cf. Vgl. Synt. I. § 80. 2 s.F. IV. p. 50. 
 
 3 Brugmann, Gr.Gr.^ 375. * Audouin, op. cit. p. 423. 
 
 5 Vgl. Synt. I. p. 199. 
 
 6 Cf. also Grdr.'^ II. p. 479 ff.; Vgl. Synt. I. p. 189 ff.; and Zieler, op. cit. 
 p. 57. 7 Cf. § 23. 
 
22 
 
 2, 
 
 sac ate. 
 
 [8, 2 anyebhih 
 
 RV. I, 164, 19 
 
 yuktah. 
 
 dhura 
 
 (i) 
 
 Indo-European Instrumental and Dative met 
 (i) With verbs and adjectives of association, 
 
 (a) *to uir5 seq^etai — he follows this RF. 
 
 man: Insf. 
 *ekuo(u) uogh5 iuqgeti — he 
 
 hitches two horses to [by means 
 
 of] the wagon : InsL 
 *udn uoinoi meiksketi — he joins, 
 
 pours water to wine : Z>at. 
 smniQS patrV deiu5 — he shows 
 
 similarity with the father, with 
 
 the god : Insf. 
 *sin^os patrai, deiu5i — similar to Aen. I. 589 os umerosque 
 
 the father, to the god : Dat ^^^ '''^'^''' 
 
 (2) With verbs of ruling and commanding, 
 
 *potietai teutai, nrbho^ — he rules ^^^^- Truc.^zx non uinum 
 
 ^ '^ ' • <^uiris^ moderart^sea 
 
 over the city, the men : Dat. 
 
 *potietai teutam, agrois — he com- 
 mands the city, the fields : Inst. 
 
 (3) With verbs of rejoicing et similia, 
 
 *uoino, tueqnobh^f terpetai — he 
 enjoys wine, rejoices in children ; 
 Inst. 
 but *deiuoibho3f, Dat.^ he is favorably 
 disposed towards the gods. 
 
 (4) With verbs of confiding, 
 
 *qretesx bheidhetai — he is confi- 
 dent because of his strength : 
 Inst. 
 
 Ov. M. 13, 866 sic se tibi 
 mis c eat. 
 
 ^F. 6, 48, igsamo devair. 
 
 uiri uino solent. 
 
 RV. 3, 54, 15 indro visvair 
 viryath pdtyamdnah. 
 
 RV. 7, 24, I mamddas ca 
 sotnaih. 
 
 Y. 50, 5 hyat yHimdkdi 
 madrdne vaordzaBd. 
 
 *suneiiai bheidhetai 
 in his son : £>at.^ 
 
 he confides 
 
 Schleicher, 268 nusitiketis, 
 
 devil *auf Gott ver- 
 
 trauen.' 
 Supr. 79, 6 voinii p'hvaj^ 
 
 svojejq silojq 'miles suis 
 
 viribus fidens.' 
 Plaut. Cap. 536 quid rebus 
 
 confidam me is? 
 
 1 X = the final case-formans is uncertain. 
 
 2 It must here be noted that the above formulas do not always aim at the re- 
 production of an actual usage. Thus *potietai nrbhox is not intended to signify- 
 that *potietai does ever take the dative; the phrase is merely a symbol of the fact 
 that there are verbs of ruling and commanding connected with that case-form, cf. 
 Vgl. Synt. I. § 133 and Grdr.^ II. § 488, 5. 
 
23 
 
 Other connections are less certain, in that their ascription to 
 what still must be called the Indo-European language is not war- 
 ranted from sufficiently extensive or conclusive usages. Those 
 enumerated, however, are satisfactory enough to serve as criteria 
 for these languages, Greek and Germanic, where owing to but one 
 given case-form we might otherwise be led astray. 
 
 § 27. Greek. For a morphological account of the fusion be- 
 tween the dative and the instrumental I refer to Audouin, Declinai- 
 soHj p. 234 ff. and Brugmann, Gr. Gr? §§ 260, 434,5, 477. In 
 Greek the instrumental fused with the dative in the singular, the 
 two cases having already been similar in form in the ist and 2d 
 declension. The w in Avkw stands for * oi < *-o -\- at} As to the 
 plural AvAcots, it is an instrumental form — ^-0 + aisy^ois, cf. Sk. 
 vf'kais, Av. vehrkais, Lith. vilkdis with instrumental signification^ — 
 the other datives in o-i, otcrt, etc. are locatives. — The Grundbe- 
 deutung of -</>tv is difficult to determine etymologically. In Sk. we 
 have -bhis inst. pi., -bhyas abl. and dat. pL, -bhydm, inst. abl. dat. 
 dual, -bhyam for dat. sg. and pi. of first and second person pro- 
 nouns. Then Lat. ii-bi^ i-bi etc. must be connected with it. If 
 Sk. bhi-s and Balto-Slavic -mi be equated, then Gk. -<^i(v) is instru- 
 mental in form. Syntactically, it is, besides, loc. abl. gen.,' but also 
 dative, cf. Homeric ws <f>prJTpr] <f>pT^Tpri(f>Lv ap-qyrj, B 363. 
 
 § 28. (i). Verbs of association, like cTrco-^ai, /^lyvwai, t,€vyvvvai, 
 fidx^a-Oai etc. " govern " a dative which may be a representative of 
 an older datival,* nay sometimes even locatival use.^ However the 
 force of the inst. of soc.-com. is still preserved in Homer and after 
 him wherever we observe avv or afm employed with the instrumen- 
 tal, as t 173) ttvrap iyio avv vtjl t ifirj koI i/xot<s kTapotaiv i\6oiv rwv 
 8* dv8pu)v 7r€Lprj(ro/JuiL ; cf. tOO B 534 tireadai (Sanskrit sac) dfJM. TLVL. 
 So verbs compounded with crw could take the dative in both signifi- 
 cations. Xen. J/^w. 4, 3, 12. TO . . . avTovs 17/xtv (rwcpyciv may mean 
 both 'the fact that the gods work with us ' and ' . . . help us.'^ — 
 
 1 The I-E. dat. sg. suffix was -at, if we may judge from the form of the dat. of 
 consonant stems, like Gk. tdixep-ai, Sk. vidmdn-e, Meillet, De quelques innova- 
 tions de la decl. latine. Par. 1906, p. 35, prefers to think it was -ei, because of 
 Oscan Fuutrei, Paterei and Lat. ovi, hominl, patrl etc., but Brugmann, Grdr?- II. 
 §§ 162, 1 63, believes that the Oscan is itself an innovation or possibly a locative form. 
 
 2 Hirt, Handbuch, p. 244. ^ Hirt, ubi sup., § 304; Grdr? § 113, 3. 
 4 S.F. IV. 59. 5 Vgl, Synt. I. p. 247. 
 
 * Audouin, op. cit. p. 240; Grdr.^ II, § 482. 
 
24 
 
 The testimony of Sanskrit is in favor of such a twofold division : * 
 whereas yuj^ yat, krid, yudh, spardh, and has are connected solely 
 wdth the instrumental "^ sac, mi'srd, misray, myaks and miks admit of 
 the dative also. So RV. 4, 48, 10, rduddrena sdkhyd sacchya 'I 
 would join the mild friend,' is the regular construction, i.e. the 
 instrumental.^ But according to Bergaigne,* "L'objet auquel le 
 sujet ou le regime direct du verbe est mel6, uni, est construit, tan- 
 tot a I'instrumental, I. 22, 3 ; 34, 3 etc., tantot k locatif, VI. 29, 2 ; 
 X. 96, 3, ou meme au datif, X. 104, 2, cf. VI. 29, 3." This applies 
 especially to the verbs myaks, miks and misra.^ In the case of 
 marital relations in Greek we surely have to do with the instrumen- 
 tal, cf. RV. 10, 10, 8, anyena mad ahano yahi tuyam ; similarly 
 Latin, Lucr. 4, 1247, admiscetur muliebri semine semen^ So OBg. 
 ozeniti sf, 'to marry [with] some one,' as Mk. 6, 17, jako ozeni 
 s^jejg, on avTrjv iydfua-ev."^ The Slavic also resorts to the inst. with 
 verbs denoting association, ange/omh retith * cum angelo rixantur,' 
 but also the dative, OBg. drugi> drugu retith, 'inter se rixantur.'* 
 These usages may aid us in clearing up the obscurity of the Greek 
 forms. 
 
 § 29. In the instance of the allied concept of similarity, such as 
 coiKtt, 10-0(0, etKo^o) and adjectives like to-os, o/u,otos, etKcXos, draXavTos 
 we can speak of both a real dative in its adverbal and of an inst. 
 in its soc.-com. sense, as the German 'gleich mit etwas.'* The 
 dative is used in Latin with similis, in OBg. with ^'hCbn'b, podo^hniy, 
 on the other hand Sanskrit samd and ^u/ya, which is probably the 
 same as araXavros, govern the instrumental, fena fu/yak, ' similar 
 to him,' Manu 4, 86. Avestan hazaosa and hadam only with in- 
 strumental.^" Cf. also Gothic he nu galeiko pans mans Lk. 7, 31 
 and OHG. iz ist gilih filu thiu, Otfr. 2, 14, 90. 
 
 § 30. (2). As to the verbs of commanding, we have a third com- 
 petitor in the locative which is already seen in the oldest periods 
 whenever it is a question of a ' crowd in which or over which ' 
 one commands." The Homeric dative accompanying verbs like 
 
 1 Cf. Vgl. Synt. I. § no, and Wenzel, Instrumentalis, p. 29. 
 
 2 S.F. V. 131. ^ Wenzel, ubi supra, p. 29. 
 * La Religion Vedique, II. 261, 263 footnote. 
 
 ^ Latin miscere has inst.-abl., Ebrard, de ablativi . . . usu, p. 26. 
 6 ALL p. 55. "^ Miklosich, Fgl. Gr. IV. 723. 
 
 8 Miklosich, op. cit. p. 597. » Cf. Vgl. Synt. I. §§ no and 124. 
 
 10 Grdr?' II. § 485 a. " ALL p. 38. 
 
25 
 
 dvd<r<ro), KpaT€<Oj oLp\<a instead of the genitive, as A 6i, rrao-t ^ler^ 
 dOavdroiartv dvd(T<T€L<s, may be considered a locative.^ For an instru- 
 mental we have in Vedic Sa-nskrit ^dtya/e, RV. 3, 54, 15, indro 
 visvair viryaih pdtyamanah ' Indra who rules over all powers 
 heroic."* The Slavic conceives of the object ruled over as the 
 means whereby the sovereignty is expressed, so OBg. vlas/z Syri- 
 jeja apx^Lv rrjq Svpta?^ but it allows also of the dative as in the 
 German 'jemandem befehlen,' especially in the later sources, 
 cesar\iSivujqstu Dekiju rimhstei vlasti, Supr. 132, 3.^ Latin is also 
 divided : we say imperare, moderare alicui and, in Plautus, even 
 temperare linguae, but potiri aliquo, as the Plautian si ilk hodie ilia 
 sit potitus muliere, originally probably prosecutive as ' to be master 
 over something, iiber etwas hin.' " Ingleichen, says Brugmann for 
 the Greek,^ lasst sich nicht mehr wissen, wie man Toio-t in Homer. 
 Toto-t §€ /xi;^o)v 5px^ ^- ^gl- empfand. Dass auch Ausdriicke wie 
 kv\ Tpweo-o-' dyopcvwv, S 45, cv 8' v/xtv cpeo), I 528, vorkommen, ent- 
 scheidet ebenso wenig fiir den Lok., wie der Umstand, dass dpy^w 
 ' Fuhrer sein ' mit Iv nai nicht vorkommt, fiir den Inst, den Aus- 
 schlag gibt."* 
 
 § 31. (3). Verbs of rejoicing, etc. In RV. kan, mad, bhuj, jtv 
 and sometimes tarp 'govern' an instrumental,' so 5, 3, 10 agnih 
 devdsya sdhasa cakandh, ' rejoicing in his divine power.' The 
 verbal concept 'to be gracious to' is regularly datival; so RV. 8, 
 93, 27 stotrbhya indra mrlaya, ' sei den Sangern hold,'* \/^r^i RV: 
 mrl\ just like Latin favere, graHficari, gratulari, ignoscere, indul- 
 gere, blandiri alicui^ and Avestan mprpzdd- and urvaz-, Y. 50, 5. 
 hyat yusmakai mqOrane vaordzaOa, * since you are well disposed 
 towards your prophet,' ^^ as against avoya ddOrpm dahditi yei^he 
 ddOrahe dditi noit havo urva vdurdza ' wenn nicht seine Seele iiber 
 
 1 S.F. lY. 56. — This syntactic fusion of the Dat. Loc. and Inst, would, of 
 course, be much furthered by formal resemblances, viz. that of the dat. -oi and 
 loc. -oi before consonant, on the one hand, and the inst. pi. *-tt;t5 and loc. -ourt, 
 on the other, into *-wts > ots and -oto-i > our before vowels. To the latter cf. 
 Gr. Gr.^ § 434, 5- 
 
 2 S.F. V. 133 ; Wenzel, op. cit. p. 79. 
 
 3 Miklosich, IV. 700 ; Vondrak, II. 348. 
 
 4 Vondrak, II. 359 ; Miklosich, IV. 584 ; Vgl. Synt. I. § 133. 
 6 Gr. Gr.^ § 462, 2. 
 
 ^ Of interest is Hungarian : birni valamivel * to have power over something.' 
 ■^ Wenzel, op. cit. p. 81 ff. ^ Cf. Grassmann, Wb. p. 1058. 
 
 ® Draeger, Historische Syntax, I. 403. ^^ Reichelt, p. 240. 
 
26 
 
 das Almosengeben froh ist.'^ Possibly the dative is used with the 
 Slavic radovati s^ ' to rejoice ' and cuditi, diviti s^ '■ to wonder,' as 
 ne divite s^ semu fxrj Sav/xd^cTe tovto Jh. 5, 28.^ The choice of Lith- 
 uanian is the causal instrumental, dede pasigerejo tdis vaikdczais 
 *the uncle had his pleasure in the youth.'' Similarly, with the 
 Greek ^ verbs x"-^P^) repnofjiai the instrumental of cause must be the 
 first to be thought of, [Tcpiro/mt ~ Sk. farp'] as in TepireaOaL BCaKOLonv, 
 like Lat. gaudere aliqua re ; but with personal names in the dative 
 we may sometimes see beyond the stage of ' on account or thru 
 whom.' So cf. * 556 x*t/oo>v *AvTtA.dx4>, y 52 dvSpt which might 
 rather be correlated with expressions like iiraLvioi, *to praise,' with 
 the Homeric "EKTopt or, Meisterhans^ 172 A^/ic Iscc, tw Srjixto. I 
 omit the locatival competition as in $ 245 TcrapTro/Mevos rcKeeo-o-tv. 
 
 § 32.(4). Verbs of trusting and confiding. Delbriick^ cannot 
 find a criterion for separating the dat. and the instrumental ; but 
 with regard to instances with -<^i(v) Audouin decides in favor of a 
 causal interpretation * thanks to which, as a result of which one has 
 confidence,'® M 135 xctpco-o-' ttcttoi^otcs rjSe ^i-qt^iv, unless, indeed, a 
 locatival view is preferred, cf. A 303. But with a person the dative 
 is rather to be understood, as * towards whom one has confidence,' 
 w 97 KaarLyvrjTois olaLirep dvijp iriiroiOe ^ in whom a man places trust.' 
 So, indeed, Avestan/m /? V9r9ne ahe daenaya, ' I trust in your law, 
 I profess your law,' Vsp. 53,^ instrumental; but Sk. sradd/ia, Lat. 
 credo, Lith. veryti, OBg. verovati2i\\ have personal datives connected 
 with them.^ ^o RV. 2, 12, 5, srdd asmai dhatta, 'believes him.' 
 
 The same comparative features may be employed to decide some 
 moot questions in the similarly syncretized Germanic dialects. 
 
 § 33. The Germanic Dialects. For a morphological account of 
 the fusion of the dative and instrumental I refer to Streitberg, 
 Urgermanische Grammatik, p. 223 ff . ; Loewe, Germanische Sprach- 
 wissenschaft, p. 75 ff. ; Dieter, Laut- und Formenlehre, p. 534 if. ; 
 Kahle, Zur Entwicklung der consonaniischen Declination in Germa- 
 
 1 GrdrP- II. § 483 c. 
 
 2 Vondrak, II. 362 ; for the inst. with the adj. dovolhffh * contented ' cf. ibid. 
 
 P- 350- 
 
 ^ Schleicher, Lesebuch,^. 126. apud Vgl. Synt. I. § 115. 
 
 4 Gr. Or? § 460 ; Vgl. Synt. I. § 115 versus ALL p. 38. 
 
 5 Vgl. Synt. I. p. 255. 
 
 ^ Cf. Audouin, p. 238, and Walther, de dativi instrumentalis usu LLomerico, 
 p. 49. 
 
 ■^ Hiibschmann, Zur Castislehre^ p. 261. ® Vgl. Synt. I. § 132. 
 
27 
 
 ntschen, Berlin, 1887, ^^^> ^^ course, to Grundriss^ Vol. II, under 
 the various stems. In the Germanic dialects what is functionally- 
 called the '■ dative ' is nothing but an instrumental form in the 
 pluraP and partly locative, partly instrumental in the singular.^ 
 The confusion of ,'the dative and the instrumental can best be 
 studied in the Westgermanic dialects, since Gothic has only the 
 pronominal inst. he and pe^ and Old Norse forms that are syntacti- 
 cally worthless ; as it is, the two case-forms had already been firmly 
 welded together in Pregermanic. 
 
 § 34. In all likelihood it was the instrumental and the locative 
 that first became merged, and later on the dative,' and in this 
 process the -0 stems are given credit for the ultimate cause of the 
 transformation.'* The oldest authentic endings used to denote the 
 Westgermanic instrumental function are -/ or -u. The latter, the 
 choice of OHG. and OS., altho traces of it occur in Ags., too, 
 according to Sievers, Ags. Gramm. § 242, may go back to a form 
 of the I-E. inst. in -0} The former, retained by Ags., really is a 
 Pregermanic -/ < I-E. -«, a locative.® In Ags. therefore we must 
 first conceive of a clash between the forms inst. -u and loc. -/. 
 The victory of the latter in finally assuming both functions is prob- 
 ably due to syntactical reasons, as with expressions denoting trans- 
 portation where either means or place would be admissible, faran 
 skipu or skipi^ an otherwise genuinely I-E. alternative ; perhaps 
 also in temporal relations, dcegu or dcegi. This change would leave 
 an inst. (loc.) -/' pitted against a dative form in -e [old Ags. ce 
 
 1 The oldest *dat.' pi. of the -0 stems is -ms. Iscc. Vatvims, AflimSy HZ. 31, 
 354 ff. The vowel between the two consonants was probably -i. Cf. Streitberg, 
 U.G.f p. 232, 4; therefore the ending is instrumental, Cf. Lith. rahko-misy 
 this form having been adopted instead of the I-E. inst. pi. of -o stems, viz. -dis 
 as IVirois. 
 
 2 Cf. Wood, Ubersichtstabellen zu Lautentsprechungen und zur Kasusbildung 
 des Nomens und Adjektivs im Germanischen, Chicago; Dieter, Laut- und For- 
 menlehre, p. 537, and Brugmann, Grdr? II. p. 280 ff. 
 
 8 Altho cf. Grdr? II. p. 492, " der Inst, war auf dem ganzen germ. Sprach 
 gebiet das zuletzt hinzugekommene Stiick des Mischkasus," based on the circum- 
 stance that we still have remnants of it in Westgermanic. 
 
 * Synkretismus, p. 232 ; Loewe, p. 76. 
 
 5 Streitberg, U.G. p. 228; cf. Lith.^^r^. 
 
 6 Cf. Sievers in PBB. VIII. 324 f., where he proves from the Epinal Glosses 
 of the early 8th century that the -i ending is older than -e ; but Vgl. Synt. I. p. 
 195, footnote, " es sei mir die Frage gestattet, ob in ihm nicht vielleicht die Fort- 
 setzung eines idg. Inst, auf -e anzuerkennen sei." 
 
28 
 
 < Pregerm. -at monophthongized^ < I-E. oi^ as Sk. -5/, Gk. -<S, 
 Lat. -oi, Osk. -ui^ Lith. -uil ; the phonetic transformation -/ > -e 
 finally resulted in an -e case, the common inheritor of the functions 
 of the inst., loc, and dat., in the same way as the *-miz > -m form 
 had functioned since Pregermanic times. ^ — In OHG. and OS. the 
 -u form is the one used without preposition ; the -e form which is 
 found in addition to -u^ and is used with prepositions, is the sur- 
 vival of the locative we have met in Ags.' The ultimate disap- 
 pearance of -u is due probably to prepositional competition weaken- 
 ing the force of the case-forms. 
 
 § 35. What must be noted in this confusion is the circumstance 
 that the process was purely formal : one form served at the same 
 time for several functional types well differentiated in the con- 
 sciousness of the speaker, as Latin dat.-abl. -bus. So that the Ger- 
 man dative of to-day is in function essentially the I-E. dative ; only 
 when it is preceded by a preposition do we deal with other func- 
 tions in it. '* Hatte eine innerliche Absorption des instrumentalen 
 Gebietes durch den Dat. stattgefunden, so wiirde der heutige Dat. 
 auch im instrumentalen Sinne verwendet werden."* 
 
 § 36. The four types of dative-instrumental fusion discussed 
 under Greek, cf. §§ 28-32, might thus be given for the Germanic, 
 making use of the identical criteria as far as applicable : 
 
 (i) Verbs and adjectives of association. The original form of 
 the instrumental is preserved in Gothic, Lk. 7, 31, he nu galeiko 
 pans mans pis kunjis jah he sijaina galeikai tlvl ovv ofioLiiiaro) tovs 
 dv^pwTTOvs T^s yev€a<s ravnys, kol tlvl ciaiv ofxoioi ; which decides at 
 once about Mt. 7, 24, galeiko ina waira frodamma o/xokixtu) avrbv 
 dvSpl <f>povLfJi(o and Lk. 6, 47, hamma galeiks ist tlvl icrrlv o/aoios ; 
 so the instrumental form in OHG. iz ist gilih filu thiu, Otfr. 2,14, 
 90. With verbs of meeting, when it is a question of persons, the 
 dative is likely to be original ; ^ so blandan : ni blandaip izwis 
 horam yJq crvvavaixLywa-Oai tto/ovoi?, perhaps also Ags. mengan, but 
 we clearly have an instrumental in krim and snaw hagle gemen- 
 ged, 'mingled with hail,' Wand. 48, just as gamainjan takes the 
 inst. of the thing, i Cor. 10, 18. niu pai ma^andans hunsla 
 gamatnjandans hunslastada sind Koii/wvot rov OvfULa-TrjpLov elalv. 
 On the other hand gamains must be construed with the real dative, 
 
 1 Cf. Streitberg, ubi supra, p. 228. 2 cf. Synkreiismus, pp. 152, 163, 235. 
 
 ^ MoUer, Ueber den Instrumentalis im Heliandy Danzig, 1874, p. 14. 
 * Synkreiismus^ p. 167. ^ Vgl. Synt. I. § no. 
 
29 
 
 Rom. II, 17, pizai waurtai <rvyKoiv<uvos t^s p^^v^-^ ^^^ verbs rel- 
 ative to social relations, like /tugan, gahorinon and gasibjon, both 
 inst. and dat. are conceivable; cf. for the former Sk. RV. 10, 10, 
 8 and Lat. Lucret. 4, 1247, for the latter Slavic eko ozeni s^ eja on 
 avTTjv iydixrjarcv Mk. 6, 17.^ 
 
 (2) In spite of Kohler' verbs of commanding do not all take 
 datives. Thus, in the instance of Ags./y rice rczdan we deal clearly 
 with an instrumental, a circumstance which, in turn, admits of con- 
 jecture relative to the force of the case-form with ON. raf>a^ as in 
 einn skal raf>a Geirr0f>ar sunr Gotna lande, Grm. 2, or in Alfrmon 
 sigre gllom rapa HHv. 39. In a similar manner Ags. he sceal f>y 
 wonge wealdan, Gu. 674, and OS. so muosta siu mid iro brudigumen 
 bodlu giuualdan, Hel. 509, might be employed in the proper plac- 
 ing of ON. valda veom, Grm. 13, and Got. waldan garda oUoBe- 
 
 0-7rOT€tV.* 
 
 (3) The question as to whether the dat.-inst. with verbs of rejoic- 
 ing is a real dat. or a real inst. is doubtful. Sanskrit and Slavic 
 indications towards the former (cf. §31) are scouted by Delbrtick, 
 who does not believe that they represent older usages.^ Otherwise 
 Erdmann-Mensing, p. 245 ; Winkler, Germ, Cas. p. 4 if. and 30 if. 
 For Gothic gaplaihan^ kukjan and the like cf. Streitberg, Got El, 
 § 248 ; Bernhardt, § 154, and Kohler, Germania^ XI, 270 ff. 
 
 (4) Verbs of confiding. The person or object trusted in or be- 
 lieved is in the dative, ON. ek munda per f>d trua, Hrbl. 96 ; Got. 
 gatrauands ufhauseinai peinai TrcTrotdws ry wraKOTJ a-ov; Otfr. 4, 35, 
 Ikes giloubi thu mir^ altho there seems to be at least one instance 
 of instrumental competition, OS. that erl thuru untreuua odres ni 
 uuili uuordu giiodean, Hel. 1527, unless one adopts Delbruck's 
 translation,^ " dass ein Mann wegen der [allgemein verbreiteten] 
 Treulosigkeit nicht auf das blosse Wort eines anderen hin [diesem] 
 glauben will." The Indo-European similarities are cited under 
 Greek, § 32. 
 
 § 37. Sufficient evidence, it is hoped, has been submitted to 
 prove an active and comparatively extensive interrelation between 
 the dative and the instrumental. These interrelations, caused 
 partly by early formal coincidences, partly due to semasiological 
 approximations, could easily be discerned because of the similar 
 
 1 Synkretismus, pp. 35, 132. 2 Vondrak, II. 345. 
 
 3 Germania XI, 267 fF. * Cf. Bernhardt, Zs. f. d. Phil. XIII, 15. 
 
 ^ Vgl. Synt. I. § 115. 6 Synkretismus, p. 159. 
 
30 
 
 correspondences of the various allied languages, save in Greek and 
 Germanic where the internal evidence is not sufficient. Testimony 
 of a like kind would also tend to establish a prehistoric connection 
 between the two cases in the (5) domain of Agency, namely, when 
 they are used with the past participle, the early employment of 
 which in a passive sense — as consequent upon the idea of com- 
 pleteness it embodies — has been indicated before. (Cf. §§5 and 
 6.) For, we encounter the following usage : 
 
 Latin: solely dative, si tibi sat acceptumst^ Plant. Most 224. 
 Slavic: instrumental only, nosim-h cetyrh^ni, alpofxevov xnro reaa-a- 
 
 p(i>v, Mk. 2, 3, but 
 Indo-Iranian has both dative and instrumental, as 
 Sanskrit: dative, tho rarely, yds fe drapsdh skanndh, 'welcher 
 Tropfen von dir iibergespritzt ist,' RV, 10, 17, 13, cf. S.F. 
 V.382. 
 iddm ma uditdm krdhi^ ' dass dies von mir gesprochen sei,' 
 
 RV, 10, 151, 2, cf. Havers, p. 10. 
 ya t^ didyud dvasrsta divds pdri, ' welcher Blitz durch dich 
 vom Himmel geschleudert worden ist,' RV. 7, 46, 3, cf. 
 Havers, p. 14. 
 rdtha iva brhati vibhvdn'e krtopastutya cikiiusa sdrasvatt, 
 'erhaben wie ein von einem geschickten Werkmeister ge- 
 machter Wagen, S. ist zu preisen von dem Kundigen,' Grass- 
 mann, Ueb. I. 550, RV. 6, 61, 13.^ 
 instrumental, nrbhih punandh, 'purified by the men,' RV, 
 
 9, 87, I. 
 pitrbhih dattdh, 'given by the fathers,' RV. 10, 107, i. 
 Avestan: dative, anyahmai arsanai varstptn^ 'begotten by 
 another man,' Yt. 17, 58. 
 Cf. also the pregnant datives in 
 aeibyo ratus spnghaitiarmaifis, 'die von ihnen gefallten Rich- 
 terspriiche wird A. verkiinden,' Y. 43, 6, Barth. Wb. 1502. 
 
 1 Even tho Grassmann's textual emendation of dat. vibhvdne to inst. vibhvdnd 
 is unnecessary, there is no compelling reason for Oldenberg's acceptance of Lud- 
 wig's rendering * dazu geschaffen, sich auszubreiten,' Rgveda, textkrit. u. exeg. 
 Noten, p. 406. Since the dative with passive expressions is certainly not unknown, 
 the interpretation of vibhvdne by Bohtlingk-Roth, Sanskrit Worterbuch, VI. 1134, 
 as "</«/. fur instr.^'' is undoubtedly admissible. To me RV. 6, 61, 13 vibhvdne 
 krtb is absolutely similar in force to RV. i, 141, 8 rdtho sikvabhih krto, *a chariot 
 made by artists.' 
 
31 
 
 yaBa hi tat by o dayaha, 'so gut wie durch Dich die Verlei- 
 hung' sc. zugesagt, K 44, 18, Barth. Wb. 744: '• Dat des 
 Urhebers.'' 
 y9 na is to, ' der von uns verehrt,' Wolff, K 56, i.^ 
 instrumental, Y. 43, 10 parstpm zt dwa, ' interrogatum enim a 
 ' te,' Bthl. IVb. 997. 
 
 /iu. k9r?ta asactt, ' den durch Asa wohlgebahnten [Weg],* 
 Y. 34, 13, Barth. Wb. 1536. 
 
 This twofold denotation of the agent with the participle demon- 
 strates the fact that the formulas 
 
 {a) ^deiuo d?tds, ' given by the god,* and 
 
 (b) *tod tot nfai qrtdm esti, * that is done by, for, this man,' 
 
 are historically correct and to just such an extent precursors of the 
 later expressions of Agency with the finite passive verbs. — We are 
 now prepared to enter upon a detailed consideration of the dative- 
 agent and the instrumental-agent in the various Indo-European 
 languages. 
 
 1 For the exchange of gAv. dat. tiS with gAv. ace. nd cf. Reichelt, p. 293. 
 
CHAPTER IV 
 The Dative of Agency in Latin 
 
 § 38. When we first encounter Latin we see it in a stage where 
 many changes had made it markedly different from the related 
 languages. We know very little of this early history, for Latin 
 lacks older works of real national character. The comedians 
 Plautus and Terence and those remnants of Ennian poetry which 
 fortunately have been preserved represent the material that is here 
 adduced for this inceptional stage. The question, too, of foreign, 
 notably Greek, influence looms up in many connections, and so 
 in relation to Latin syntax, especially to the construction here 
 discussed. 
 
 § 39. The opinions of {a) the ancient grammarians may be of in- 
 terest by way of introduction. Priscian ^ volunteers no information 
 as to how the dative became attached to the passive, despite the 
 fact that the question had before him been taken up by Servius. 
 The latter, as is known, attributes the construction to Greek im- 
 portation ; ^'■'■Neque cernitur ulli,* neque ab ullo cernitur; et est 
 Graecum ovhevl opw/xevos, ut scriberis Vario^ id est scribet te 
 Varius," says he in commenting on Aeneid I. 144.^ 
 
 ip) The modern explanation of the dative as commodi vel in- 
 commodi goes back to that period in the history of Latin Grammar 
 when the several uses of the dative in all languages were reduced 
 to an unqualified unit and the ablative was looked upon as an in- 
 terloper.'* So we read back in 1526 — this is the oldest medieval 
 syntax I have at hand — in Aldi Pii Manutii Inst. Gram. Libri 
 Quatuor. Addito infine de octo partium orationis costructione libello 
 Erasmo Roterodamo Autore, p. 10, Ordo I. " Datiuus post uer- 
 
 ^ De Passive, Inst. Lib. xviii. 1 23, vol. III. ed. Keil. 
 2 Hor. lib. I. od. 6. 
 
 8 Comm. in Verg. Aen. rec. Thilo et Hagen, Lipsiae, 1878, vol. I. 
 * Cf. to the latter Priscian, V. 672 ; Pompeii, Comm. artis Donati, Keil, V. 
 181 ; and as late as the i8th century, Reisig, Vorlesungen, III. § 379. 
 
 32 
 
33 
 
 bum. Quoduis uerbum acquisitiue positum exigit datiuum, ut Non 
 omnibus dormio. . . . Mihi istuc nee serifur nee metitur.^^ I have 
 emphasized acquisitiue because of Tillmann's statement^ that it is 
 to Sanctius we must go back for the first conception of the dat. 
 auei. as one of interest. — Voss of Amsterdam ^ is careful to differ- 
 entiate between what he calls true Graecism and none at all, ''Sic 
 nativus sermo est (p. 460) si dicas de Trajano, Est Senatui lau- 
 datus : quia itidem in activo dicas, Senatui laudare. At non 
 aequ^ nativus, est Plinio laudatus ; pro a Plinio : quia in activo ea 
 significatione non dicam, Plinio lau davit : sed, Plinius laudavit. 
 Quare plane distinguendum, inter dativos passivis appositos : quia 
 tunc demum structura est Graecanica, cum in activo pro dativo est 
 nominativus." — The Port Royal grammar' very conveniently 
 places this dative among the exceptions, p. 383. — To Sanctius* 
 '■ deus amatur mihi'' is similar to 'hoc non probatur mihi^' and 
 both have but the same dative as ''da peeunias mihi' " Itaque 
 ''non eernitur ulli' id est, nulli ostendebatur, nulli erat eonspieuus.'" 
 " Mihi tamen hie & ubique dativus aequisitionem significat," which 
 is a viewpoint evidently less sane than that of Voss. 
 
 § 40. Madvig's note to Cicero De Fin, i. 4, 11 ^ forms the basis 
 of all recent discussions, " Exempla bonorum scriptorum prosae 
 orationis aut in participio sunt, ut res ecfecta tamquam externa 
 exstet et ad personam referatur sine commodi sine aliqua eiusmodi 
 ratione \_pertractata mihi sunt^ id est, habeo pertractata, II. de 
 Orat. 146 ; elaboratum mihi est, Diu. in Caec. 40] aut in eiusmodi 
 uerbis, ut non solum ab aliquo, sed etiam aUcui res fieri intellega- 
 tur, ut in quaerendi uerbo." So monographers like Tillmann,* 
 who supersedes Wisseler^ and is in turn supplemented by Schaef- 
 ler,^ fully subscribe to this concept of possession or acquisition of 
 
 1 Act. Erl II. 72. 
 
 2 Gerardi loannis Vossii Aristarchus, ed. secunda, torn. II. 458 f. Amst. 1662. 
 
 * Nouvelle Methode pour apprendre facilement la langue Latine, etc. Dixieme 
 ed. Par. 1709. 
 
 * Franc. Sanctii Brocensis Minerva, with notes of Scioppius — Perizonius, 
 ed. quinta, Amst. 1733, pp. 181 and 396 of vol. 3. 
 
 ^ M. Tulli Ciceronis de Jinibus bon. et mal. Hauniae, 1839, p. 27. 
 
 ® De dativo verbis passivis lingtuie Latinae subiecto qui vacatur Graecus, in 
 Act. Sent. Erl. II. pp. 71-140, 1881. 
 
 ' De dativo cum verbis passivis coniuncto latinis scriptoribus cum graecis corn- 
 muni. Prog. Wesel, 1838. 
 
 ^ Die sog. syntaktischen Gracismen 1884, p. 47 ff. 
 
34 
 
 the type of *«/ mihi aliquid,'' and so do later grammarians, Holtze,^ 
 Draeger,2 Kiihner,' Haase,^ Roby,* and Schmalz.^ 
 
 § 41. A difficulty, however, arises from the circumstance that 
 Madvig's definition of the dative of agency does not apply to all 
 instances. '' Est mihi moriundum^ may indeed mean 'death ex- 
 ists for me,' but certainly not that 'one has to die in or for my 
 interests,^ ^ It covers only cases where the dative is at the same 
 time a dai. commodi vel incommodi? It is already inapplicable to 
 a great number of examples in Livy,* Tacitus ^° and even Cicero." 
 Such a phrase as '■ videnda oratori^ may easily be paraphrased 
 as 'the things that are such for the orator that it is incumbent 
 upon him to examine them ' ; a ' visum est oratori^'' that a certain 
 result exists for the orator from this examination, but no such 
 connotation attaches to the Ciceronian legionem Fausto conscrip- 
 tam^ ad Att. 8. 3, 7, and ibid. 12, i. The question as to whether 
 such uses, substantially equivalent to ab c. abl., are indigenous to 
 the Latin dative or represent foreign and non-datival influence, be- 
 comes thus the critical point in connection with the Latin dative 
 of agency. 
 
 § 42. Narrowed down to fundamentals, the question is as fol- 
 lows: Is the agency idea in the dat. auctoris^ as undoubtedly 
 present in Latin distinct and separate from that of advantage or 
 disadvantage, merely an extension of the latter or the result of 
 Greek influence ? Antiquated views, such as that of Reisig," ac- 
 cording to which the dative-agent is an ablatival construction 
 transferred to the dative shortly after the invention of the sixth 
 case, will not detain us. Tillmann {pp. cit.) and Landgraf ^ insist 
 that the construction was born and bred on native soil, but that 
 its use was extended, under the influence of the similar but better 
 developed Greek form, at the time of the Augustan poets ; whilst 
 
 1 Syntaxis prise, script. Lat. I. 312. 2 jji^t^ Synt. I. § 189. 
 
 8 Ausf. Gramm. II. 239. * Vorlesungen, II. 149. 
 
 ^ Grammar of Lat. Language^ II. 60. ^ Lat. Syntax, p. 246. 
 
 ■^ Cf. Miles, Compar. Synt. of Gk. and Latin, I., Cambridge, 1893, P" 27. 
 
 8 Landgraf s note to Reisig, Vorlesungen, III. p. 628. 
 
 * Kiihnast, Syntaxis Liviand^, p. 139. 
 1° Heraeus ad hist, i, 11 and Nipperdey ad ann. 2, 50. 
 11 Cf. Tillmann's examples, op, cit. pp. 79-84. 
 ^ Vorlesungen, III. § 379. 
 
 13 In notes to Reisig, III. p. 627, and pp. 1-15 of Beitrage zur historischen 
 Syntax der lat. Sprache, Prog. Miinchen, 1899. 
 
35 
 
 Brenous^ stands for Greek influence even before that time. A 
 historical survey will best bring out the facts. 
 
 § 43. There is no reason to suppose with the grammarians "^ that 
 the dative of agency appeared first necessarily with the gerund and 
 gerundive, and only then with the perfect passive participle because 
 of the latter's adjectival force. Such a view might partly be justified 
 by the comparative infrequency of the latter combination in the 
 earliest times, since Ennius has but two examples,' nam quoi quod 
 agat, institutumstj Sc. 236, is diciust ollis popularibus, Ann. 306, 
 and Plautus and Terence* comparatively few, e.g. Phorm. 248, 
 meditata mi hi sunt omnia mea incommoda ; Epid. 467, argenti quin- 
 quaginta mi ilia emptast minis ^ 471, estne empta haec mi? Most, 
 224, si tibi sat acceptumst, etc.,^ but it is of importance to note that 
 these instances appear as early as the gerundival examples, and 
 that the employment of the p.p. p. with the dative is an I-E. feature 
 far antedating even the creation of the specifically Italic gerund 
 and gerundive.^ Of interest here is also one inscriptional example, 
 maiorem obtenui laudem, ut sibei me esse creatum laetentur^ CIL. I. 
 38. That the idea of necessity or completeness is not inherent in 
 the dative ; that thus we can have expressions like * oratio habenda 
 est^ or ^ visum esf without reference to the agent of the action, is 
 but another proof of the ability of the dative to combine with a 
 p.p.p. at any early period. 
 
 § 44. The gerundive and gerund are more frequent with the 
 dat. auctoris, than the past participle, e.g. Ennius, Ann. 45 f., 
 gnata, tibi sunt ante ferendae Aerumnae ; Plautus, Rud. 1298, 
 adeundus mi illic est homo; Bac. 325, nunc tibimet illuc navi capi- 
 undumst iter; Terence, Andr. 166, restat Chremes, Qui mi exoran- 
 dus est; ibid. 152, prope adest, quom alieno more vivendumst mihi. 
 The dative is not surprising with the gerundive, owing to the 
 latter's originally purely adjectival force, since the dative with ad- 
 jectives is one of the commonest phenomena in the Indo-European 
 
 ^ ^tude sur les Hellenismes dans la Syntaxe LaHne^ Par. 1895, PP* '54-i83. 
 
 2 Cf. all those referred to in §§ 40-42. 
 
 3 Cf. Frobenius, Die Syntax des Ennius ^ p. 31. 
 
 * As in the instance of Terence, so in that of Plautus the edition of the 
 Scriptorum Classicorum Bibliotheca Oxoniensis, recognovit W. M. Lindsay, has 
 been used thruout. However, other recensions, especially that of Goetz-Schoell, 
 have also been consulted. 
 
 s Cf. Holtze, Syntaxis, I. 312 ; Lindsay, Synt. of Plautus y Oxf. 1907. 
 
 ^ Cf. §§ 37 and 44, note. 
 
36 
 
 family of languages.* Horton-Smith's plausible theory' of the 
 origin of the gerundive also explains this dative as one of advan- 
 tage, with the idea of * giving ' in the suffix -do^ as " Cic. Off. I. 41, 
 aliorum iudicio permulta nobis et facienda et non facienda et mu- 
 tanda et corrigenda sunt = in the opinion of others there are very 
 many things which give [cause] us the doing, the non-doing, the 
 changing and the correcting, i.e. which should be done by us." 
 As to the passive tinge of the verbal adjective, though it is second- 
 ary no doubt, we can easily think of ''facile dictu * ' easy to say ' 
 developing into 'easy to be said,' the 'the-doing for some one' 
 becoming the ' the-deed-to-be-done for some one.'' " The secret of 
 the passive lies in an indefinite subject: '' vivif is a definite 'he 
 lives,' ' vivitur ' an indefinite ' one lives.' " ' At any rate, the earliest 
 
 ^ For a morphological discussion of the gerundive cf. later, § 86 ; agreement 
 has been reached on the score of its being an Italic formation, preceding in 
 point of time the gerund which is not found in the literary monuments of Umbr.- 
 Samn. [cf. Brugmann, Grdr?- II. § 1103, Anm. 3 and Horton-Smith in AJP. 
 vol. 15, 194; 16, 217; 18, 449 and 19, 413], the opinions of Weisweiler \^Das 
 lat. part. fut. pass. Paderborn, 1890], and Platner \_AJP. vol. 14, 4] based on 
 Terence and Plautus notwithstanding. That the -ndo forms were originally 
 adjectival and that their relation to the verbal categories arose from the context, 
 is the contention of Weisweiler, op. cit. p. 64 ff. On the basis of the latter's 
 work Brugmann, ubi supra, thinks it highly probable, " dass unserm Gerundivum 
 die Bedeutung der zu vollziehenden Thatigkeit von Anfang an zukam." It is 
 more likely, however, that this is a development from its use as a mediopassive 
 present participle instead of an original *-menos, cf. Persson, De origine ac vi 
 primigenia gerundii ac gerundivi Latini, Upsala, 1900. This idea may be seen 
 as early as Perizonius' notes to Sanctius, 15, 8, cf, Landgraf to Reisig, III. 747. 
 As a part. fut. pass, it does not figure till the third century after Christ, tho see 
 Kvicala, Wiener Sttidien (1879), I. 236, to Livy's preface § 6, * ante conditam 
 condendamve urbeni^ whereas to express obligation it appears in Plautus : prob- 
 ably first in negative sentences, non ferendum, what cannot be suffered, then 
 ferenduffty what is, can, must be suffered. (Sommer, Handbuch, p. 650 ff.) 
 Its third use in the sense of ' facile est ' develops only after Cicero. The very 
 earliest meaning it had was probably what the suffix -do means, * doing, causing 
 to be done' or even 'giving,' so that volvendus is 'giving a roll' as in Lucr. II. 
 3, II ; dat motus, i.e. movetur, cf. Horton-Smith, A/F. 15, 213. Bayard, De 
 gerundivi et gerundii vi antiquissima, diss. Par. 1902, sees in the two participial 
 forms of sequor, i.e. secundus and sequendus, respectively, the original active 
 meaning and the more recent passive use of the gerundive. So Roby, Grammar, 
 pp. Ixi-xcvii, conjectures that the passive sense of the gerundive is really due to 
 an attraction and not to any originally passive signification. 
 
 24//'. 15, 213. 
 
 8 E. W. Fay in AJF. 15, 221. 
 
37 
 
 usage of the gerundive with the dative of interest well accords 
 with its passival character. 
 
 § 45. To revert to the p.p.p. whose connection with the dat. 
 auct. is the more important for the latter's development, three 
 points are of importance in relation to it in the pre-classic period : 
 
 {a) The participle is used and felt as an adjective with the 
 dative, and is often replaceable as when in Sil. Ital. V. 334, nulli 
 non saucius hosti is equivalent to sauciatus, just as Pliny's celebres 
 Homero vituliy Hist. Nat 32, 11, 144, is really celebratus. Cf. also 
 adjectives of the -bills type, Haase, Vorlesungen, II. 156 ; Roby, II. 
 § 1 146. When the p.p.p. cannot be reduced to an adjectival 
 force, the only alternative is, as in the periphrastic perfect, that of 
 a fact accomplished, result attained for some one : cognltum mlhl is 
 habeo cognltum in open disguise. These two meanings determine 
 absolutely without exception the early connotation of the participle 
 with the dative. As it is, it occurs rarely. Ennius has but two 
 examples, Instltutus and dlctus^ cf. Ribbeck, Trag? 185, and Cic. 
 Brut. 15, 58, also Frobenius, p. 65 ff. ; Terence only three more, 
 medltatus, Phorm. 248 ; susceptus, Ibid. 967 ; exoptatus, Heaut. 408 ; 
 decretum^ Ibid. 392 ; spectatus, ' tried,' Ad. 893. 
 
 (b) These datives with the participles in early Latin are regularly 
 pronouns of the personal, relative, and demonstrative type. Thus 
 Plautus has 5 personal pronouns and 3 demonstratives, Terence 
 only two personals, but Ennius none. Havers ^ is surely wrong if 
 he means to deny the existence of relative and demonstrative pro- 
 nouns, beside personals, in any but Plant. Merc. 743, nobis coquen- 
 dast, non quo I conductl sumus. No one, indeed, after the acute 
 observation of Seyffert,^ need give a moment's attention to Epld, 
 
 227, 
 
 at trlbutus quom Inperatus est, negant pendl polls ; 
 
 mis, qui bus trlbutus malor pendltur, pendl potest, 
 
 about which both Tillmann, p. 78, and Brenous, p. 166, are at sea,' 
 but surely Ennius, Sc. 236, quol is not a personal pronoun, nor 
 
 1 Untersuchungen zur Kasussyntax der idg. Sprachen, 191 1, p. 1 88. 
 
 2 Cf. Landgraf, Beitrdge, p, 7, footnote. 
 
 3 The real meaning of this passage, which has been the mainstay of those who 
 would see in it an exception to what seems to be a Plautian rule that the dat. 
 com. in auctorial sense appears only with perfect forms, judging by the context is 
 as follows : To the state the taxpayers declare that they are unable to pay the 
 tax; but to those, i.e. the women of whom the two interlocutors are dis- 
 
38 
 
 Ann. 306, ollis popularibus, nor perhaps, Plaut. Aul. prol. 4 and 5, 
 patri avoque. Plautus, Epid, 520, 1 find, as corrected by Pontanus, 
 ^ me itnpune irrisum esse habitum, depeculatum eis^ would be an 
 interesting addition, but the three lines 518-520 are missing from 
 the Ambrosian text and are condemned by Geppert and Goetz. q. v. 
 
 {c) The early possibility of the retention in the passive of a 
 dative which could be attracted to the verb in the active is one 
 more proof that originally this dat. was but one of personal interest. 
 The old formula of mancipation ^ emptus mihi est pretio^^ is among 
 the first we meet in this transformation. So in Plaut. Epid. 154, 
 qui ubi tibi istam emptam esse scibit, and 467, mi hi ilia emptast^ the 
 phrase ^ emitur mihi^ is the passive of '• emo mihi,^ ' mi hi' being 
 identical in both. In Sen. controv. I. 2, 7, * ancillae ex lupanaribus 
 sacerdoti non emuntur' we are not given to understand whether 
 the priestess should or should not herself make the purchase ; all 
 the sentence conveys is that she is not to have and harbor such 
 persons. So the whole host of participles like sumpfus, probatus, 
 comprobatus^ quaesitus, susceptus^ spectatus, visus^ and even con- 
 ductus, for ^mihi servus^ sc. aliquid, conducitur^' Merc. 560, is but 
 the passive of ^mihi servum conduco.'' On Merc. 743 cf. Tillmann, 
 p. 77, who commits the error of classing these datives with the 
 later and real datives of agency. 
 
 § 46. No essential change in the usage of the dative in question 
 appears before Cicero. It is a regular dat. of interest denoting 
 for whom the action must take place, gerundive, or for whom it is a 
 fact accomplished, past participle or its compounds. No other com- 
 bination occurs and the dative is pronominal, overwhelmingly per- 
 sonal. In Cicero, however, the passive verb is found not only in 
 the perfect but also in the present and imperfect ; for this purpose 
 verbs are used other than such as could take a dative in the active ; 
 
 coursing, to whom a far greater toll is paid, they can pay. It is but fair to add 
 that as early as 1845, 1 find, Naudet's translation. III. 400, takes this view, " mais 
 on peut bien leur payer a elles de plus grands tributs " ; so, too, Epid. ed. Gray, 
 Camb. 1893, p. 35, note, " the subject of negant must be the men." ///^V, therefore, 
 is a dat. of ind. object = meretricibus. 
 
 1 Gains, i, 119 ; apud Tillmann, 78: Fragm. iur. Rom. Vatic. 50 in Huschkii 
 Jurispr. Anteiustin. quae super suni'^, p. 692. 
 
 2 Cf. Haase, Vorlesungen, II. 152. 
 
 * As Plaut. Mil. C/. 517 ; " videri, scheinen, ist kein Deponens, sondern ein 
 mediales Passiv, dessen Infin. futuri visum iri heisst, also ist der damit verbundene 
 Casus dem Dativ bei passiven Verbis gleich zu achten," Draeger, I. 405. 
 
the dative is not restricted to personal or other pronouns, but may 
 be a substantive as well. The whole combination is changed, so 
 that many datives can no longer be explained according to Madvig's 
 definition, since while they may be, often unnaturally, fashioned to 
 disclose a subsidiary idea of interest, they denote rather activities 
 that are purely external, in a word, auctorial. Verr. III. 43, 
 Mete II o paternus honos et avitus neglegebatur has not a past verb 
 and ad Att. 8, 12, i, qui ex delectibus conscripti sunt consulibus is 
 not in the interest of the consuls. 
 
 § 47. There are, as we have indicated, § 42, but two ways of 
 explanation. A third one that suggests itself from the contempla- 
 tion of extra-Italic conditions, viz. that the real dat. auctoris in 
 Latin is the inheritor of the twofold I-E. usage typified in Indo- 
 Iranian, with perhaps a latent instrumental force in it cropping to 
 surface under such favorable circumstances as an approximation 
 in meaning with the indigenous dat. commodi, is interesting be- 
 cause of the presence in Plautus of one compound infinitive, and 
 even of one present tense, if with Acidalius and Goeller we may 
 emend Aul. prol. 4 and 5, as '<?/ color patri avoque iam huius '^ 
 and also because of the possible existence ^ of substantives used as 
 dat. auctoris in Ennius : Saturno sancte create, Ann. 627,' and 
 arcus subspiciunt, mortalibus quae perhibentur [Iris], found in 
 Priscian, VI. 259, 5 H. There is, however, no further evidence of 
 the free use of the dative of agency in early Latin nor therefore 
 of any instrumental contamination of it. Besides we must remem- 
 ber, that because a construction occurs in some one language of 
 the Indo-European group, this is no proof that it certainly occurred 
 in another language of the same group.* 
 
 1 Ussing, ed. Havniae, 1878, p. 276, retains * colo'' in accordance with the 
 Codd. Par., remarking " propter dativum, qui sequitur, Acidalius * color ' scriben- 
 dum censebat ; mihi dativus commodi defendi posse videtur." This would 
 equate *r<?/<?'with * incolo,^ and * patri ^ with ^ pro patre^ or "This house for 
 many years I have occupied and guarded for the sire and grandsire of its present 
 owner." The construction ^ possideo et colo domum patri avoque^ however, is 
 scarcely Latin, whilst Uolo'' for * color ^ 'honor, cherish' is perfectly normal as *« 
 guibus diligenier observari videmur et coli, Cic. Mur. 34, 70. So Goeller, ed. 
 1825. 
 
 2 Hoc contra, without much reason, Havers, op. cit. p. 189, footnote 2, as 
 " nicht beweiskraftig." 
 
 2 Furius in IV, "quod genus hoc hominum Saturno sancte create," p. 117 of 
 ed. Vahlen. * Miles, Comp. SyntP- p. 30. 
 
40 
 
 § 48. The numerical differences observable between the lan- 
 guage of archaic and classic Latinity undoubtedly are too great 
 and sudden in proportion to be explained away on the ground of 
 indigenous development. With due regard to the workings of 
 analogy, of concinnity, and of such unavoidable oltto koivov construc- 
 tions as Z>e Senectute^ 11, 2)^^ ^semper in his studiis laboribusque 
 viventi non inielligitur quando obrepat senectus,^ where the dative 
 might equally well go with ^ obrepat ^^ ^ the fact remains that whereas 
 Plautus has,'^ besides six passive perfects and a compound infini- 
 tive, one sole doubtful case of a finite passive form ; the remains of 
 Ennius' works two perfects and no finite form ; Terence the same 
 and then Catullus but one finite form, tereretur 68, 15, on the 
 other hand, Cicero out of 128 instances has not only 11 participles 
 and 63 perfects, but also 8 pluperfects and 2 future perfects, not 
 only 23 compound infinitives, but also 8 simple infinitives and, to 
 cap all, 13 distinct forms of simple finite passives, viz. 9 presents, 
 3 imperfects, and one future. Besides, Plautus deals only in pro- 
 nouns, of which 5 out of 8 are personal ; Terence restricts himself 
 to 2 personal pronouns; Ennius alone, of all writers of antiquity, 
 has one example of a substantive, the ' ollis popularibus ' of Ann. 
 306, and this one is combined with a pronoun.^ Catullus, at a dis- 
 tance from Ennius, out of 9 examples has but 2 nouns, one proper 
 and one common, est enim venuste magna Caecilio incohata mater ^ 
 35, 18, ?indi pluribus ut caeli tereretur ianua divis, 68, 115. Cicero 
 already has 5 proper names and 12 substantives. 
 
 § 49. The disturbing factor appears to have been Greek influ- 
 ence. The historical fact of Graeco- Roman linguistic contact^ 
 would a priori permit the consideration of such an influence. The 
 success of a Livius Andronicus and the Greek medium of the early 
 annalists, to mention but these, must be taken as a token of a pro- 
 Hellenistic attitude even at that period. The very question of the 
 
 1 Haase, Vorlesungen, II. 151. 
 
 2 Cf. the examples of Tillmann, p. 105; Schaefler, p. 48 ; Brenous, p. 154. 
 
 * A circumstance in which Havers, p. 188, sees a transition from the pure 
 pronominal to the nominal dat. of agency. Fliigel in Zs.f. Volkerpsych. XI. 58, 
 indicates the indissoluble unity of the concept of Self and its Name, so that he 
 thinks of the proper name as an intermediate stage in this transition. If, then, 
 Havers is correct, we might ceteris paribus accept Saturno, Ann. 627 as regular 
 for that early period. 
 
 * Cf. F. O. Weise, Charakteristik d. lat. Sprache, 1909, p. 55 ff., and for a gen- 
 eral view Saalfeld's Der Hellenismus in Latium, Wolfenbiittel, 1883. 
 
41 
 
 absolutely pure Latinity of even so-called archaic Latin is not thus 
 exempt from all suspicion. Beginning with the fashion of the 
 poets, Catullus being the first in point of time, to import and imi- 
 tate everything bearing a Greek stamp, from the technic loans of 
 Horace to the recondite Alexandrinism of Propertius, not to speak 
 of the epic and prose writers, the matter of influence upon syntac- 
 tical Latin becomes even more patent and decisive.^ Haase, II. 
 151, is incorrect when he states that " der griechische Gebrauch 
 ist fiir Cicero noch nicht vorhanden," for his contemporary, Caesar, 
 has only two examples of the dative of agency,^ viz. B. C. i, 6, 
 praeterea cognitum compertumque sibi and B. G. 7, 20, victoria quae 
 iam esset sibi at que omnibus Gallis explorata, and those, too, 
 of the ancient variety ; and the other contemporary, Sallust, has no 
 more than one irregular example ,^y//^. 107, i, ^ saepe antea paucis 
 strenuis advorsum multitudinem bene pugnatum ' [rendered highly 
 questionable hy Jug. 114, i, ^per idem tempus advorsum Gallos ab 
 ducibus nostris . . . male pugnatum \^ so much so that I see in 
 * antea ' a corruption of ' ante ' and ' ^ '] , and, lastly, it was Cicero 
 himself, who, besides other examples, wrote ep. Att. 14, 21, 3, ^ sed 
 mihi quidem p€.^LiiiTaLi.^* 
 
 § 50. With the Augustan poets and historians whose style the 
 former admittedly influenced, there is already no limit to personal 
 innovations and the sense of ' mihi c. passivo ' can rarely be differ- 
 entiated from the type * a me,^ the dative appearing very frequently 
 in places where one expects, according to earlier usages, ab c. 
 ablativo. It may be said, in general, that after Tacitus the primi- 
 tive sense, natura, of the dat. auctoris as a dat. commodi is almost 
 
 1 Cf. F. O. Weise, op. cit. p. 191. 
 
 2 Draeger, I. 435, somehow credits him with none. 
 
 3 Haase himself, l.c.^ explains Hist. I. 42, 25 ' quae si vobis pax et concordia 
 intellegentur ' as ' sunt ' or ' videntur.' 
 
 * True, Cicero tells us, Tusc. Disp. i, 8, 15, ** Dicam, si potero, Latine : scis 
 enim me Greece loqui in Latino sermone nonplus solere quam in Graeco Latine^^ 
 but the passage, by its very context, refers solely to simultaneous bilingual prac- 
 tice and gives no warrant for a belief that in his writings Cicero endeavored to 
 rule out Greek constructions. As to the latter, cf. Kertelhein, Ueber Gr'dcismen 
 in Ciceros Reden, Jena, 1894; Lebreton, Etudes sur la langue et la grammaire 
 de Ciceron, Paris, 1901, especially in connection with Tusc. Disp. 2, 7, 19, aspice 
 Philoctetam, cui concedendum est gementi; and Brenous, passim and p. 440, 
 "tandis que Ciceron s'excuse quand il emploie des mots grecs, nous ne voyons 
 pas qu'il s'agisse de meme quand il se sert de constructions hellenisantes," pos- 
 sibly because Greek was to him a second mother-tongue. 
 
42 
 
 entirely lost and that no care is exercised in drawing a dividing 
 line between the two. Beginning with Plautus, thru Vergil, as e.g. 
 Tros Tyriusque mi hi nulla discrimine agetur, Aen. I. 574 and nihil 
 o tibi amice, reliclum, VI. 509, where it is surely not for his benefit ; 
 Propertius, Praia cruentanlur Zelho, 4, 14, 41 ; or Ovid, lactaque 
 nascenti corpus haberet humus, Trist. 4, 3, 46, nobis habitabitur 
 orbis ullimus, ibid, i, i, 127, and the much-cited barbarus hie ego sum 
 quia non inlelligor ulli, ibid. 5, 10, 37, up to Ammianus Marcellinus, 
 there are all together about 1222 instances of both sorts. (Cf. Till- 
 mann's examples.) For the Church Fathers, among whom the chief 
 offender was Cyprian, cf. 'Ronsch., Ilala und Vulgala, 1875, P- 43^- 
 
 § 51. To assert, however, that the construction was altogether 
 due to borrowing from the Greek is equally out of the question. 
 Taken as a whole we cannot say that the dativus auctoris was a 
 construction wholly alien to Latin and one to which the latter took 
 a bold leap. Rather must we say, agreeably to the facts we have 
 detailed above, that it is a construction for which Latin had a 
 latent capacity but to which it crept by slow stages until acceler- 
 ated by a similar Greek model, in the same sense as the Latin 
 * bellum abolendae infamiae ' became the Tacitean ^ proficiscilur cog- 
 noscendae anliquilatis ' not mediately thru a stage of * pugnal abo- 
 lendae infamiae, but under the influence of a Greek dTr^A^e tov 
 yvwvai.^ This is precisely what Brugmann understands by the term 
 Graecism:"^ " Unter Grazismus hat man nicht zu verstehen, dass 
 der lateinischen Sprache etwas ihr von Haus aus vollig Fremdes 
 aufgepfropft wurde, sondern es wurde nur ein seinem Ursprung 
 nach echt einheimischer Anwendungstypus, weil er im Grie- 
 chischen ein von den Romern empfundenes Analogon hatte, nach 
 diesem auslandischen Muster weiter ausgebildet." So Brenous, 
 p. 79, tho somewhat more radically, " Quand nous disons ' Helle- 
 nisme,' nous voulons dire que la construction n'est pas, a la place 
 oil nous la rencontrons, ce que nous attendions, qu'elle en est 
 meme toute diff^rente, et que, ne pouvant pas etre It^gitim^e en 
 latin, elle a sa justification propre dans la construction grecque 
 correspondante dont elle est imit^e, soit directement et sans etre 
 pr^par^e par quelque tour analogue, soit en s'aidant de ce point 
 d'appui." We have ascertained this point d'appui. 
 
 § 52. Other points of contact, besides the fundamental con- 
 nection, were not lacking by any means. A Horatian ' bellaque 
 1 Miles, Comp. Synt. p. 57. 2 /./r y. 100. 
 
I 
 
 I 
 
 43 
 
 matribus detestata^^ Od. I. i, 24, might also be interpreted as 
 ' wars that are an object of detestation for mothers ' and similar 
 expressions of emotion, as amatus, dilectus, spretus, which easily 
 glide into a connection of authorship, can be taken as having con- 
 tributed their quota to the native development of the dat. auctoris. 
 So undoubtedly the datives with intransitive passives, as cadere 
 alicui=.caedi ab aliquo, and iacere ^= prostratum esse, e.g. cui con- 
 sul in armis Crispinus cecidit, Sil. Ital. 17, 305, as a parallel to a 
 tan to cecidisse viro, Ov. Met. 5, 192, both set against TibuU. 1,1, 
 ^^ agna cadet vobis with the clear dative of interest (of. Reisig, 
 III. note 551 a), to which it is not at all fantastic to compare the 
 development, according to Delbriick, S.F. IV. 75, of the Greek 
 dative-instrumentals with aorists in -lyv and -^v. (Cf. Brenous, p. 
 160.) In a word, Latin may be said to have had its own points 
 of view in this respect. Unaided it developed the native dativus 
 commodi to a certain degree from which the construction might 
 naturally have risen to the rank of the dativus auctoris pure and 
 simple. (Even Plautus' Men. 645 '• palla surruptast mihi'' in reply 
 to ' palla mi hi St domo surrupta'' sounds, to my modern ears at 
 least, dangerously near the brink of agency.) It was, however, 
 betimes assisted in this tendency by the entirely homogeneous con- 
 struction of the Greek. In this sense only is the Latin dative- 
 agent with passives a Graecism. 
 
CHAPTER V 
 
 The Instrumental of Agency in Slavic. — The Genitive 
 OF Agency in Lithuanian 
 
 § 53. Just as Latin offers the best illustration of the develop- 
 ment of the dative-concept into that of agency, so Slavic, whereof 
 old Bulgarian, OBg., is selected as the type, does with respect to 
 the Instrumental. It has been indicated before that the Instru- 
 mental of agency is but a development of the Instrumental of 
 means with passive expressions. As such it is directly traceable 
 to the Inst, of association which is the accepted forerunner of the 
 Inst, of means. A series of examples might thus be adduced to 
 represent the hypothetical gradations, but it must always be borne 
 in mind, since we must believe the Passive to have arisen in each 
 language only after the dialectal scission, that in this connection 
 only the idea embodied in the Sk. karana, but not that in the 
 word kartar, can be considered as I-E. with finite verbs. The 
 case is, of course, entirely different, with the past participle, § 37. 
 § 54. Thus we have {a) pure concomitation ^ in Slavic ex- 
 pressed by the instrumental, — altho it is rare by reason of com- 
 petition on the part of the preposition s-b, — and that first of all in 
 military expressions,^ 
 nuzdajemu bease iti voiy Sup. 157, 26, * proficisci cum militibus* 
 
 corresponding to the Lat. abl. of association, as in ' Caesar omnibus 
 copiis Jlerdam proficiscttur,^ B.C. i, 41, 2. 
 
 {b) This concomitation, especially, again, in military expressions, 
 shades off into means or instrument, so cf. 
 
 udariti ratiju na gradh, ' cum exercitu urbem invadere ' 
 and even nearer to the inst. of means in denotation of convey- 
 ances, as' 
 
 / ida Vb pusto mesto korabVem edini, Mk. 6, 32, 'and they de- 
 parted into the desert place by ship, t<5 TrXoto).' 
 
 1 Turns like k-hmotrami svoimi ne sittnilati s^, ' cum matrinis suis non com- 
 misceri,' Cloz. I. loi, are thus omitted. 
 
 2 Vondrak, II. 342 ; Miklosich, IV. 723. » Miklosich, IV. 689. 
 
45 
 
 {c) To the use and extent of the inst. of means there is no limi- 
 tation. ALL p. 58 ff., gives an approximate sketch of it. For the 
 Slavic, Vondrdk, II. 345, offers similar categories. One example, 
 however, must be mentioned here because of its interest in relation 
 to Germanic, cimh odezdem s^ Mt. 6, 31, cf. Gothic he wasjaima, 
 
 (d) It is but natural to expect that since the sociative inst. re- 
 ferred largely to persons, its logical successor, the inst. of means, 
 should likewise refer to persons. In fact, persons are also found 
 employed as means or instrument in a manner equivalent to Latin 
 
 J>er c. ace. : ^ 
 
 tlhkom-h rece, Sup. 44, 12, 'per interpretem dixit,' 
 
 nest-h gospodix nynja glagolal-h Mhnoj'a, Sup. 144, 17, 'nonper me 
 
 nunc locutus est,' 
 de sljejej vesti susedami, * mittebat nuncios ei per vicinas.' 
 
 {e) Delbriick has intimated for the Sanskrit'^ that it was from 
 this inst. of means with the active that the inst. of agency with the 
 passive had risen. But just as the possibility of a development 
 like ^'- sdmsati vdcdbhth> sasydse vdcobhih > ribhyate vdsis- 
 thaih, i.e. he praises with words > thou art praised with 
 words > he is praised by the Vasisthas " cannot be denied, for 
 the Slavic itself as well as for the other languages, a '■'■per interpre- 
 tem dixit > dictum per interpretem^ i.e. ab interprete " formula would 
 not only be conceivable, but would be directly prior to the San- 
 skrit model because of the finite passive forms in the latter. Such 
 participial forms are not at all uncommon in Slavic* 
 
 ZTb rozdenych-h zenami, Mt. 11, 11, h yevvrjrol^ yvvaiKwv ; 
 iskusa/em'h sotonq/a, Mk. i, 13, ircipa^o/Acvov V7rb rov ^aravS; 
 pravim-h dhvema aggeloma, Sup. 124, 26, 'qui a duobus angelis 
 
 ducitur ' ; 
 ne vidim-h nikymhze, ibid. 159, 28, ' qui a nemine videtur ' ; 
 nosim-h cetyr.hmi, Mk. 2, 3, aLpofievov xnro recra-dpoiv; 
 
 these instrumental of personal agency are doubtless all develop- 
 ments from the instrumental of personified means, like 
 
 trbstb vetrom'h dvizema, Lk. 7, 24, koAxi/aos vtto dv€/xov o-oAcvo/acvos, 
 
 where we have only a personified agent, and Russian pishmo napi- 
 
 1 Miklosich, IV. 693; Vondrak, II. 345. 2 s.F. V. 135 and Vgl Synt. I. 268. 
 
 3 Miklosich, IV. 704; Vgl. Synt. I. § 123. 
 
46 
 
 sano mnojuy 'the letter is written by me,' must be analyzed according 
 to napisano perom-h, 'written by a pen,' and napisal-h perom-h, 'I 
 write with a pen.'* 
 
 {/) Reflexives employed as passives also take the inst. of agency.'* 
 
 cjudith sja vsemi Mojsi^ * Moses is admired of all,' Izv. 615. 
 
 narece s^ preprostyj vhseja braiija^ Sup. 131, 19, ' dictus est simplex 
 
 ab omnibus fratribus.' 
 krhsti s^ tu sastiim-h episkuponvh^ ibid. 146, 10, ' baptizatus ab epis- 
 
 copo qui ibi erat.' 
 
 (^) Undoubtedly some adjectives, especially in -b^ia, etymologi- 
 cally related to the p.p.p. in -no and in all respects their equals, 
 also belong here, e.g. 
 
 imeniti sqste vhsemiza dobra detelh. Sup. 63, 18, * cum celebrarentur 
 
 ab omnibus propter virtutem ' ; 
 mnogymi Ijudhtni cesten-h i slavervh^ Izv. 267, ' qui a multis homini- 
 
 bus colitur et celebratur.'* 
 
 § 55. This is the usage of Slavic. The employment of datives 
 to express the logical subject in impersonal sentences whose verbs 
 are reflexive-passives must not be thought of as an infringement 
 upon the province of the instrumental. OBg. munifh mi s^ Sokcl 
 fxoL; izvoli se mne Iho^k /xot ' mihi visum est,' Lk. i, 3; or Russ. 
 mne dumajetsja instead oija dumaju ' puto ' ; mne chocetsja, * volo ' ; 
 mne sniiosh ' somniabam,' etc. are like Germanic es dUnkt ihm,pdtti 
 ser, where the dative in reality denotes the object towards which 
 the action tends. 
 
 § 56. Passing to Lithuanian we find there an instrumental case, 
 which is even frequently employed, so akimis matyti, ' to see with 
 eyes,' and farther back, in a sociative sense, ve'zimii vaziuti, 'to 
 travel with a wagon.' At the same time a development similar to 
 that in Slavic of the inst. of means into agency cannot be traced, 
 because what few instances occur of the passive defined by a logi- 
 cal subject are expressed by means of the Genitive,^ and in modern 
 Lithuanian even with the preposition nu, ' from.' Thus, 
 
 1 Potebnja, Iz-b zapisolcb po r. gram.^ I. u. II. 467, apud Vondrak, II. 350, 
 
 2 Miklosich, IV. 704 f. 
 
 8 Cf, Miklosich, IV. 704 ; Vondrak, II. 350. 
 
 * Schleicher, p. 273 ; Bezzenberger, Beitrdge zur Gesch. d. lit. Sprache^ Got- 
 tingen, 1877, p. 243. 
 
47 
 
 mdlka ugnes su'edama, * the wood is consumed by the fire ' ; * 
 plaukelei vejo puczamiy ' hair blown by the wind ' ; 
 pastas kardliaus siustas, ' ambassador sent by the king ' ; 
 grdmata man^s raszyia, * letter written by me ' ; 
 avis liuto sudraskyia^ ' the lamb has been rent asunder by the lion ' ; 
 paziur'/, kdd jo arklys suestas vtfko, * he saw that his horse was de- 
 voured by the wolf.' 
 
 § 57. Two explanations are possible for this use of the genitive. 
 A) That the genitive is of ablatival origin. As in Slavic, so in 
 Lithuanian, too, the genitive is the recipient of the functions of 
 I-E. ablative. For the Lithuanian this condition reaches back to 
 the period of Balto-Slavic unity. However, the formal confu- 
 sion of the two cases is I-E. in the sing, of all save the -o stems, 
 where in both Slavic and Lithuanian the ablatival form has been 
 preserved in both functions : Slav, -a < I-E. abl. -dd\ Lith. -o < I-E. 
 abl. -od} The genitive-agent may thus denote an earlier abl. of 
 separation or origin. The fact of its present use exclusively with 
 nu, as jis yr nu kardliaus si^stas^ would point towards the prob- 
 ability of such a provenience; other examples may be found in 
 Kurschat, p. 393, who believes that this use of nu in the sense of 
 ' von ' is a Germanism imported by bilingual Germans. Tugi ne- 
 prissigaudinkite Hiskios^ * do not be deceived by H.,' in Bezzen- 
 berger, p. 243, is a good example in support of the view which 
 regards this construction as ablatival. Brugmann ^ has no positive 
 opinion on this matter. 
 
 § 58. B) On a firmer basis stands the conception of the genitive 
 as one of possession, because it brings the Lithuanian in line with 
 Indo-Iranian, Greek, and Germanic under similar conditions ^ ex- 
 cept that in these languages the force of agency is less pronounced. 
 Thus the formula * deiuosio datos appears in : Sanskrit — in 
 RV. 10, 160, 4 dnuspasto bhavaty eso asya, * conspicitur ille ab eo, 
 eius^ or * ei notus est ille ^ \ pdtyuh krita saft, M. i, no, 11, 'the 
 wife that is bought by the husband,' i.e. 'the purchased wife of the 
 husband.'^ — Avestan, Y. 34, 9, dwahya b^rpxham vlduso^ 'esteemed 
 
 1 The examples are gathered passim from Kurschat, Bezzenberger, Schleicher, 
 and Leskien-Brugman; cf. bibliography. 
 
 2 Cf. Leskien, OBg. Gr. p. 109 ; Vondrak, II. p. 3 ; Grdr.'^ II. § 155. 
 8 In Leskien-Brugman, p. 321, note. 
 
 * For the latter, cf. Vgl. Synt. I. § 170 ; Grdr."^ H. § 513. 
 
 6 Cf. Siecke, de genitivi, p. 28; S.F. V. 153; and Gaedicke, Der Akkusativ, 
 
48 
 
 of him that knows thee,' K ii, "j , pairis.x'^axfym ayaidhahe is not 
 really * enclosed by iron,' but ' the enclosed of iron,' and Vd. 7, 29, 
 aiwiynixta su no is 'the gnawed of dogs.'^ — Greek has a similar 
 adnominal genitive, (T<f>aycU AlyiaOov, Eur. £/. 123, 'slain of A.'; 
 cravSaXiov avTov 7r€cf>op7]fX€vov, Herod, 'worn by him; ' cf. also Stos- 
 80TOS, 'the presented of Zeus.'^ — The same construction is prob- 
 ably seen in Latin ecquod est huius factum aut commissum non 
 dicam audacius,pro Sull. 26, 72, and Terence, legati Romanorum, 
 e IMS dicta, malivoli veteris po'itae male dictis, Andr. prol. 7, 
 where the p.p.p. is really substantival.' — Germanic has few of 
 these genitives ; so cf. Aelf. So. 171, ^,gedo me lufiende &* onfundne 
 pines wisdomes.* The modern German ^ die Gesandten des 
 Konigs {unser Gesandter) ' and '■ wir Geweihten des Schmerzes' 
 have their OHG. prototypes in Otfr. V. 20, 6'j,gi'wihte mines vater^ 
 gisegendte sine, where the verbal nature of the participle is distinctly 
 felt. — Accordingly, the Lithuanian genitive with the p.p.p. may 
 safely be taken as adnominal. That a ' kardliaus siustas ' is actu- 
 ally felt to be ' the king's messenger ' is seen from examples like 
 ^Ba katre mano [instead of ^no manes'] bus supraszyti, fe biis 
 svecziu sule pasoditi,^ ^ ' who will be invited by me.' 
 
 § 59. It is thus seen that the auctorial force of the adnominal 
 genitive with participles is not due to any such force being inher- 
 ent in the genitive itself, but that it is, rather, developed from the 
 context, — a parallel, therefore, to the assumption of a similar tinge 
 of agency by the dat. of interest. A short digression may be in 
 place here concerning the probable interrelations of this dat. aucto- 
 ris and the genitivus possessivus in question. According to Havers' 
 investigations ^ the I-E. pronominal forms * moi, * toi, [* soi^ i.e. 
 Sk. ml, fe, Avestan moi, toi, hoi, Greek jxoi, toi, [ot,] Latin mt, as 
 in mi pater, "^ and * ti, as in O. L. genitives mts, tts, and also OBg. 
 mi, ti, \_d, ] were originally not both dative and genitive forms, as 
 
 p. 42. " Das Pradikat paralysirt die urspriingliche Bedeutung des Genitivs, die 
 Zugehorigkeit desselben zum Nomen, und lasst ihn den Instr., Dativ und Abl. 
 vertreten." 
 
 1 Cf. Reichelt, p. 259; Hiibschmann, p. 270. 2 yg^ Synt. I. § 170. 
 
 8 Brugmann, IF. V. 136. * Wiilfing, Synt. ofy^lfred, II. 22. 
 
 5 Leskien-Brugman, Lit. Volkslieder u. Marchen, p. 275. 
 
 ^ Untersuchungen zur Kasussyntax der idg. Sprachen, Strassburg, 191 1. 
 
 ■^ The mi<,mihi is different; cf. Stolz, Lat. Laut- u. Formenkhre^, 216; 
 GrdrP- II. 406. 
 
49 
 
 stated in Grdr} II. § 312, but simply sympathetic datives^ which 
 in the course of time developed a possessive genitive function. 
 
 § 60. Strangely enough, the dat. auctoris seems to be one of the 
 main sources of this possessive use of the dat, symp. and therefore, 
 indirectly, of the adnominal genitive, and this could have been all 
 the more possible because of the fact that both the dat. symp. and 
 the dat. auctoris originally appear only with personal pronouns. 
 For the Latin dat. auctoris we have already indicated this condition 
 (§ 45 b). Havers strikingly demonstrates it for the dat. symp., 
 Untersuchungen, p. 237 ; so for Sanskrit, ibid. p. 44, Avestan, ibid. 
 p. 60 ; the demonstrative pronoun in Homer must be considered 
 an extension of the usage, ibid. p. 106. The transformation of the 
 dat. auctoris into an adnominal sympathetic dative probably came 
 about thru the adverbal pronominal dative separating itself from 
 the verb, joining the substantive and ipso facto entering upon the 
 road to an ultimate adnominal genitive. Thus : 
 
 Sanskrit, RV. 10, 145, 2, adverbal dat. > symp. dat. in posses- 
 sive sense, sapdtnim m 'e para dhama, ' blow me the neighbor 
 away'; in RV. ibid. 5, ubhe . . . sapdtnim me sahavahai^ 
 *we will both conquer my neighbor.' So 
 RV, 10, 151, 2 priydm bhoj'esu ydjvasv iddm ma udifam krdhi, 
 originally felt as ' spoken by me ' might have become ' this 
 my word,' and 
 RV. I, no, I tatdin ml dpas tad u tayaie is correctly given in 
 S.F, V. 394 as ^ gethan ist mein Werk.'' 
 
 Greek, c 243 ^ows Se ot iji/vro l/oyov, ' quickly was the work done by 
 him > his work done ' ; t 404 TroAvapryros Sc rot iart, 'he is 
 much desired by you ' > " ^r ist dein Heissersehnter.^'' ^ 
 
 Latin, mi hi quidem aetas actast ferme, Plant. Trin. 319, where the 
 dat. originally with the verb could be attached to * aetas '/ 
 uritur cor mi hi, Pers. 800, equals cor meum? Well known 
 are the instances of * alicui in mentem venire ' when really 
 * alicuius ' is meant and felt. 
 
 If then we believe, with Havers, that the dat. auct. was one of 
 the sources of the possessive genitives * moi, * toi, (* soi,) we can 
 
 1 The term appears to be Gildersleeve's ; also called da^. possessivus, Gunther ; 
 dynamicus or energicus ; lastly dat. personae cui studium est, Holtze. 
 
 2 Havers, 74 ; cf. Vogrinz, Gramm. d. homer. Dial., p. 305. 
 * Havers, p. 183. 
 
50 
 
 surely go one step farther and declare that it was this same gen. 
 poss. < dat. auct. which in conjunction with the p.p.p. gave rise 
 to the gen. auctoris sc. possessivus that we have found in Indo- 
 Iranian, Lithuanian, Greek, Latin, and Germanic ; and that just as 
 the dat. auctoris spread from pronouns to substantives, thru the 
 medium of proper names, the genitivus poss.-auctoris must have 
 experienced a similar change and arrived at that stage of develop- 
 ment in which we have found it in Lithuanian and elsewhere. 
 
CHAPTER VI 
 The Dative and Instrumental of Agency in Indo-Iranian 
 
 § 6i. Latin, aside from analytic constructions, expresses the 
 agent with passives solely by means of the dative ; Slavic knows 
 only an instrumental of agency. It is to Sanskrit and Avestan 
 that we must turn to obtain the only view of the simultaneous em- 
 ployment of both cases in the function of agency. That both ex- 
 pressions are independent of each other since the earliest period 
 we have no cause to doubt, and we may take Sanskrit as well as 
 Avestan as preserving that I-E. condition from which both Latin 
 and Slavic have narrowed down to their own individual uses. But 
 that there is observable from the very outset a thorogoing interre- 
 lation between the dat. and inst. in their several functions of 
 agency is equally undeniable. 
 
 § 62. Such coincidences have most likely arisen from purely 
 semasiological connections. Possibly, too, we have the influence 
 of another case, viz. the genitive of agency, to consider as an 
 added impetus in this competition of the other two cases. If, for 
 Sanskrit, we agree with PischeP that the pronominal asnie, gen. 
 [-dat.-loc] plural is sometimes employed as an instrumental, so R V. 
 I, 165, 7 ; 7, 67, 2 and 7 ; 8, 2, 10 ; 8, 82, 6, and with Persson^ that 
 the Avestan gen. sg. ma-na contains the same suffix as inst. tl-na^ 
 then there is that much to be registered for the mediative offices 
 of the genitive.' In addition, according to Wackernagel,* the sg. 
 gen. me^ te are found used as an instrumental in the first Delhi isc. 
 
 1 ZDMG. 35, 174. 2 //r II. 234. 
 
 8 For the gen.-dat. relation cf. § 59 ff . Of Pischel's examples, Delbriick 
 (^S.F. V. 207, 381) is inclined to consider asme \n RV. i, 165, 'j.yujj'ebhir asme 
 as genitive and in the rest as locatival. But surely in R V. 8, 2, 10 ime ia indra 
 somas ttvra asme sutasah, ' these sour Somas are pressed by us,' it is instrumental 
 in sense, because of R V. 3, 47, 3 somam . . . sutdm nah, * the Soma pressed by us,' 
 lit. * our pressed Soma ' ; also 7, 67, 2 dsocy agnih samidhdnb asme, * there shone 
 Agni lit by us,' because with participles in -na the agency is expressed by the 
 instrumental and not also by the genitive (cf. Audouin, Declinaisottt p. 109) as 
 with participles in -ta (cf. § 58). 
 
 * KZ. 24, 599. 
 
 61 
 
52 
 
 of Asoka, Saddavtsati vasa abhisitena me iyam dhammalipi likha- 
 pita} — In Avestan trie and fe when seemingly instrumental may 
 always be interpreted as datives, as J?. 5, 77, yat 1712 avavat 
 daevayasnanqm nijatpm, ' dass von mir so viele D. erschlagen 
 worden sind/ Reichelt, p. 241, comparable to yahmai xsnuto^ Yt, 
 
 10, 87, 'by whom he is satisfied."^ 
 
 Sanskrit. § 63. The dative of agency as such is not mentioned 
 in Panini ; ' evidently he considered it but one of the natural func- 
 tions of the sampradana or dative case, i, 4, 32. The instru- 
 mental denotes the agent, kartr, unless it be already expressed 
 by the verb, and the instrument, karana, as well, 2, 3, 18, deva- 
 dattena krtam, * done by D.,' datrena lunatic ' he cuts with a sickle.' 
 It may denote the agent only with a passive or a causative verb ; 
 with an active verb the agent is contained in the termination of 
 the verb itself; so dlvadattena kriyate, 'it is done by D.,' and 
 pacayaiy odanam devadattena ' he has porridge cooked by D.' 
 Another reference is 2, 3, 71 as to the agent's appearance with the 
 participium necessitatis either in inst. or in gen., so bhavata or 
 bhavatah katah kartavyah 'you must make a mat.' — Examples 
 for the dative usage follow. 
 
 § 64. The dative of agency appears 
 
 {a) With verbal adjectives in -ya^ sometimes called gerundives 
 or partic. necessitatis.* The dative is really one of interest^ 
 but has the force of an agent. 
 
 RV. I, 33, 2 yah stotfbhyo hdvyo dsti yaman^ 'he who is to be 
 
 invoked by the singers at the sacrifice, an object of invocation 
 
 to the s.' ; 
 I, 75, 4 (and I, 189, 7 ; 3, 62, i, etc.) sdkhibhya idyah, 'to be 
 
 honored by the friends ' ; 
 4, 5, 8 pravcLcyam vdcasah kim ml asydj ' what is to be said by me 
 
 of this talk ? ' 
 
 (b) Verbal adjectives in -ayya.^ 
 
 RV, 2, 4, 3 daksayyo yo dasvate ddma a, 'he who is to be satis- 
 fied by the sacrificer in the house ' ; 
 
 1 Corp. Iscc. Indie. I. 106. ^ cf. Hubschmann, p. 223. 
 
 8 Panini's Grammatik, Bohtlingk, Leipzig, 1887, and Liebich, BB. lo, 207 flf.; 
 
 11, 273ff. 
 
 4Cf. Delbriick, KZ. 18, 90; S.F. V. 396; Vgl. Synt. I. § 143; Grdr.^ H- 
 § 491 ; Havers, op. cit. pp. 9, 14, 22. ^ S.F. V. 400 ; KZ. i8, 90. 
 
53 
 
 6, 69, 5 indravctvisnu tat panajayyam vam, 'this is to be praised 
 by you both.' 
 
 {c) Verbal adjectives in -tva} 
 
 RV. 2, 30, 10 vlrya krdhi yani te kdrtvani, 'accomplish the deeds 
 
 which thou hast to accomplish ' ; 
 4, 18, 2 bahuni m'e dkrta kdrtvani, ' much undone is to be done by 
 
 me ' ; could also come under {d), 
 
 {d) With past participles in -nd and -td, 
 
 RV. 10, 17, 13 yds tl drapsdh skanndh, 'which drop is spilled by 
 you.'^ 
 
 For -ta Delbriick cites only RV, i, no, i, but this is really an 
 example of the adnominal genitive, cf. § 60. Additional examples 
 are, however, besides 4, 18, 2 q,v. above, 
 
 RV. 6, 18, 15 dkrtatn ydt tl dsti, 'what you have not yet accom- 
 plished ' ; 
 
 6, 61, 13 vibhvdne krto^ 'von einem geschickten Werkmeister 
 
 gemacht ' ; cf. § 37 note ; 
 
 7, 46, 3 j'i /i didyud dvasrsta divds pdri, ' which thimderbolt was 
 
 hurled by you from the sky ' ; 
 
 8, 77, 9 eta cyautriani t'e krta, ' these gigantic deeds have been per- 
 
 formed by you ' ; 
 10, 151, 2 iddm ma uditam krdhi, 'effect that this be uttered by 
 me.'' 
 
 {e) Delbriick also declares * that he cannot find this dative with 
 a finite passive verb, RV. 10, 65, 4, devah stavante manusaya, 
 ' the gods are praised by men,' being regarded as doubtful both by 
 him and Oldenberg.^ But with Havers, p. 10, one must consider 
 ^F. 7, 76, 2,prd mepdntha devayana adrsran, ' erblickt wurden von 
 mir die gottbetretenen Pfade,' a certain example; similarly RV. 1, 
 175, I, mdtsy dpayi te mddah, 'be merry, thou hast emptied the 
 intoxicating drink.'*® Likewise 8, 51, 9 tubhyam pdviravy ajyate 
 
 1 No instances cited in S.F. V. 2 cf. S.F. V. 382. 
 
 8 Cf. Havers, op. cit. p. 14. * Vgl. Synt. I. 300 and S.F. V. 145. 
 
 5 Rgveda. Textkrit. u. exeget. Noten. Abh. d. kgl. Ges. d. Wiss. zu Gottingen, 
 phil.-hist. a. N.F. XI. 5, p. 406. 
 
 • Cited by Gaedicke, Der Akkusativ^ p. 42. 
 
54 
 
 rajlhy " dir wird beim Paviru Reichthum gefuhrt = du fiihrst dem 
 P. Reichthum zu," Gaedicke, p. 134.^ 
 
 Perhaps 8, 26, id yuvabhyam bhutv asvina may be classified 
 here, cf. Havers, p. 14, because the context does not warrant 
 Grassmann's " Es sei eur eigen, Ritter ihr," I. 439 as the sole 
 possible rendering. 
 
 § 65. In two articles, written in 1906 and 1907 respectively, 
 Professor Hopkins attempts to disprove the present accepted ver- 
 sion of the Vedic dative as originally a purely grammatical case of 
 interest, cf. § 13 ff., and seeks to vindicate for it a local or directive 
 force.^ His contention is of special importance in that one of his 
 arguments relates to the old objection Gaedicke and Hiibsch- 
 mann raised to the dative as local on the ground that, as they 
 pointed out, the dat. sympatheticus^ ethicus etc. and the dai. auctoris 
 could never have developed from such a concept. According to 
 Hopkins, Delbriick admits, S.F. V. 145, that the agent dative is 
 not combined with finite passive verbs. " There remains only the 
 adjectival gerundive, which Delbriick still holds to be construed 
 with an agent dative in its most primitive use. . . . But it will 
 be found that the role of the dative in connection with these and 
 similar adjectives is normally not that of an agent." ^ Moreover 
 the examples cited in S.F. V. 396 and 401 with hdvya, tdya and 
 daksayya are incorrect, because the gerundive should be taken 
 absolutely. Besides, of the gerundives thus given hdvya is found 
 with the inst. of agency 4 times, with the inst. of means 2 times ; 
 it is also found with the genitive of the person, and with the 
 ablative absolute ; 4 times with a dative which, however, depends 
 upon an accompanying verb, and only 3 times as apparent agent, 
 viz. FF. 8, 96, 21 ; 10, 39, 10 ; i, ^^, 2. But of these, sdkhibhyas 
 in the eighth book, he finds, depends on the verb, and the other 
 two instances are respectively in the first and tenth books in hymns 
 belonging to the secondary stage. The question with respect to 
 
 ^ This passage is given by Aufrecht, II. 156, from whom Gaedicke, cf. p. 40, 
 appears to cite his examples, as tirds cid arye rusame pdriravi tubhyet sd ajyate 
 rayih, and in Bloomfield's Concordance, p. 429 and 432 as tiraccid arye rucame 
 partravi tubhyet so ajyate rayih. Paviru, according to Macdonell and Keith's 
 Vedic Index, London, 1912, I. 509, "appears in a hymn of the Rigveda, viii. 51, 
 9, as a Rusama, being a prince or at least a wealthy noble." Cf. Grassmann, 
 Wb. s.v. paviru and Ubers. 502, " Der Reichthum, . . ., der wird heimUch von 
 dir dem treuen Rusama Paviru zugefiihrt." 
 
 * TrAPA. yi, 87 ff. zxi6.JA0S. 28, 360 ff. 8 TrAPA. 37, 109. 
 
55 
 
 idya is similar, the most certain examples being in the first and 
 ninth books. Sakha sdkhibhyas is a stereotyped phrase, ' friend to 
 friend,' and the clauses it is in should not be separated by Delbrtick 
 into sdkhibhya tdyas. 
 
 § 66. Professor Hopkins' conclusion thus is that neither of these 
 two gerundives can stand as an early example of a gerundive with 
 the dative-agent. In fact, the latter arises from a wrong way of 
 looking at the gerundive, and " the agent-dative [a construction not 
 found in Sanskrit, where the gerundive takes either instrumental 
 or genitive] is due partly to native imitation of older forms with- 
 out understanding and partly to modem interpretation of what was 
 not originally conceived of as agent." ^ 
 
 § 67. In reply, it must be noted that Professor Hopkins himself 
 does not deny the existence of some instances at least of the dat.- 
 agent with gerundives, certain to all intents and purposes.^ The 
 only trouble with them is that they suffer in numerical comparison 
 with the absolute uses and the instrumental connections of hdvya 
 and idya. Hopkins himself admits ' that his data do not altogether 
 disprove the interpretation advocated by Delbriick. This fact, 
 combined with his choosing the use of the dat.-commodi-auctoris 
 with the gerundive as a criterion, and reducing such evidence 
 to nil in order to disprove the early existence at all of the dative in 
 auctorial function, detracts much from the force of his argumenta- 
 tion. Surely the employment of the dative with past participles in 
 -td and -nd, to all intents proethnic, cf. § 37, must have antedated 
 the use of the dative with such a specialized form as the gerundive. 
 Moreover, the verbal adjectives in -tva also appear combined with 
 the dative-agent, cf. § 64 c. Lastly, in spite of Delbrtick's admis- 
 sion, there are found, besides those in Gaedicke that seem to be 
 inacceptable, some examples of the dative of agency with finite 
 passive verbs, cf . § 64 e. Strong collateral testimony is offered, 
 furthermore, by the results of Havers' investigations, as given 
 before. For, if the genitive of possession was to have evolved, 
 partly at least, from a personal dative-agent, the former-being found 
 simultaneously with the latter, — the stages of transition being still 
 visible,* — it stands to reason that the latter must have chronologi- 
 cally preceded it.^ So much for the dative of agency. 
 
 1 TrAPA. yj, no. '^JAOS, 28, pp. 371, 372, 373. 
 
 8 TrAPA. -yi^ ^ ^o* * Havers, op. cit. p. 22 et al. 
 
 5 Cf. also Speyer, VuSS. § 46, " Wie im Skt. der Dativ I sein Gebiet mit dem 
 
66 
 
 § 68. The Instrumental of Agency. — The same plan may be 
 followed as in the instance of Slavic in the visualization of the 
 various gradations leading upwards from the sociative function. It 
 must once more be emphasized, however, that while highly probable 
 and certainly acceptable as mirroring actual facts, such a scheme 
 nevertheless is theoretical in character, because when we first en- 
 counter the sociative in its diverse types we find them all existing 
 side by side ; Brugmann then is justified in asserting that " alle 
 acht Gruppen diirften als schon uridg. nebeneinander vorhanden 
 angesehen werden." ^ 
 
 (a) Pure concomitation ; the use of concurrent prepositions like 
 saka, sakam is not the rule until post-Vedic prose, ^ but see Panini 
 2, 3» 19- 
 RV, I, I, 5 devo devebhir a gamat^ 'deus cum dis adveniat.* 
 
 ijf) Equivalent to " with the aid of," 
 
 RV.^, 103, 2 indrena^ or 7, 48, 2 indrena yuja^ 'with Indra 
 as a companion, Indro iuvante.' 
 
 So in mentioning vehicles or other means of communication both 
 
 " with " and " by means of which " are admissible, 
 
 RV. 5, 58, 6 ydt prayasistha pfsatibhir dsvair, vilupavibhir 
 maruto rdthebhih, 'when ye approach, O Maruts, with the 
 piebald steeds, stallions, with the stout-tired chariots ' ; 
 
 a curious transition is offered in 
 
 I, 123, 5 jdylma tdm ddksinaya rdthena^ "in woUen wir besigen 
 mit dem wagen [farend.] "^ 
 
 (c) The wide-spread instrumental of means is found with pas- 
 sive expressions also, (i) of objects, 
 
 Gen. teilen muss, so giebt es auch einzelne Gebietsteile, welche er an diesen con- 
 currenten Casus unwiderbringlich verloren hat. In solchen Fallen hat nur das 
 Vedische den alten Dativ, wie a) zum Ausdruck des Agens bei dem Gerundiv, 
 nur im M. \_Mantra] vgl. AIS [Delb.] 396; schon B. \_Brdhmana] hat hier 
 den Gen., nie den Dativ." 
 
 1 Gr. Gr.^ 407. For the examples following see Wenzel, IJber den Instru- 
 mentalis im Rigveda, p. 21 ff. ; ALL p. 50 ff.; S.F. V. § 83 fF.; Speyer, VuSS. 
 § 32 fF. ; Vgl. Syni. I. § 104 ff. ; Grdr."^ II. § 476 ff. 
 
 2Cf. VuSS. % 32 Aum. 
 
 3 Wenzel, op. cit. p. 63, but otherwise Griffith : " may we subdue him with our 
 car the guerdon." 
 
57 
 
 ^F. I, 35» 4 abhivrtatn kf-sanath^ 'all adorned with pearls '; ^ 
 9, 109, 15 pibanty asya vUve devaso gobhih sntdsya, 'cooked with 
 milk.' 
 
 (2) of things personified, 
 
 I, 164, 14 suryasya cdksu rdjasaity avrtam^ 'the sun's eye thru 
 
 the clouds covered ' ; 
 4, 17, 12 vato nd jutdh standyadbhir abhraih, 'as the wind 
 
 driven on by thundering clouds.' 
 
 (d) Persons are employed as means, 
 
 visvam so agtie jayaH tvdya dhdnam^ ^'he attains thru thee, Agni, all 
 good.' ^ 
 
 (<r) The instrumental of agency may have been produced from 
 the inst. of means with the active solely along the line indicated 
 by Delbriick, " so dass man zuerst gesagt hatte : qdsydte vdcobhih 
 er erfahrt Preis, wird gepriesen mit Worten, dann ^dsydte kavi- 
 bhih durch die Sanger." ^ So Brugmann, "1st das Verbum ein 
 Passivum, so kann im Instrumental auch eine Sache auftreten, die 
 weniger als Mittel denn als die den Vorgang bewirkende und her- 
 vorrufende Potenz vorgestellt ist, wie ' er wird mit Liedern geprie- 
 sen ' = ' Lieder preisen ihn.' . . . Dies fiihrte im Arischen und 
 im Slavischen dazu, auch die handelnde Person beim Passiv in 
 den Instrumental zu setzen."' The probabilities are, however, 
 that, as we have indicated before, § 54 ^, this inst. of agency was 
 first employed with the past participles in -nd and -td. 
 
 RV. I, 163, 2 yamena dattdh, ' given by Yama ' ; 
 9, 87, I nfbhih punandh, ' purified by the men ' ; 
 8, 62, 18 krsndya badhito visa, ' sore pressed by the black horde.* 
 
 The instrumental is found with the gerundive also, where we have 
 already ascertained the coexistent use of the dative and the ad- 
 nominal genitive as well. With regard to the latter, as a rule 
 the inst. is required if the verbal sense prevail, but the gen. if 
 the krfya has the value of an adjective. This is evident from the 
 fact that the subjective genitive is used in similar situations, if the 
 krfya has the value of a substantive (§ 69, VuSS.). For the inst. 
 with idya and Mvya cf. Hopkins, /A OS. 28, 371 ff. 
 
 1 Found in Grdr.^ II. p. 526. 2 sj^', V. 135. » Grdr,^ II. p. 527. 
 
58 
 
 (/) With finite passive verbs, 
 
 RV, I, 92, 7 divdh stave duhita gotamebhih^ 'the daughter of 
 
 D. is praised by the Gotamas ' ; 
 9, 86, 12 sotrbhih puyafe vfsa, ' is purified by the worshipers * ; 
 7, 67, 7 usa ribhyate vdsisthaih, ' is praised by the V.' 
 
 The inst. of means and of agency side by side in 
 
 RV. 9, 87, 7 fydm mrjanti ddsa yosanah sutdm soma fsibhir 
 matibhir dh'ttibhir hitdm, ' accelerated by the priests thru 
 devotion and prayer.' 
 
 § 69. The results of the mutual relations of Dative and Instru- 
 mental are thus as follows. 
 
 With the past participles, both dat. and inst. occur ; 
 With the gerundive in -ya, ditto ; 
 With verbal adjectives in -ayya and -tva, only dat. ; 
 With finite passive verbs, both dat. and instrumental. 
 
 Avestan. § 70. Here the development is similar to that of 
 Sanskrit. The dative, while it is lost in Old Persian, save as to 
 the pron. suff. matyy taiy^ saiy^ and replaced thruout by the genitive 
 as in Prakrit, is a favorite case in Avestan. Following are the 
 facts with respect to the dat. commodi in the function of 
 agency.^ 
 
 The dative of agency appears with {a) the perfect participle in -ta ; 
 cf. § 37- 
 Yt 17, 58 ^5 aom puOrpm baraiii anyahmai arsanai varstpm 
 
 pai&e upabaraiti, ' who bears a child that is begotten by [to] 
 
 another man and exposes it upon the highway ' ; 
 5, 77 yaf me avavat daevayasnanam nijat^m yaSa , . ., 'that so 
 
 many Deva-worshipers have been slain by me ; ' 
 K 43, 10 pprpsaca na ya tot phma parsta, ' et interroga nos quae 
 
 tibi sumus interrogandi,' Barth. Wb. 878. 
 
 An example in Old Persian is Xii, 3, uta tya^maiy kartam, ' and 
 what was done by me,' against a possess, gen. tya^ mana kartafn, 
 ' mein Gethanes,' Bh. iv, 42 f.^ 
 
 iCf. Reichelt, p. 241; Hiibschmann, p. 223; Havers, p. 52; Vgl. Syni.l. 
 § 143; Grdr?H. §491. 
 
 2 Hiibschmann, p. 299; Havers, p. 52 ; cf. also L. H. Gray's note on Tolman, 
 AJP. 30, 1909, p. 457. 
 
59 
 
 {b) Also in periphrastic passives with * bav,^ * to become,' so 
 
 Yt lo, S^j yahmai xsnuto bavaiti, yahmai tbisto bavaiti, *by 
 whom he is satisfied, injured ' ; ^ 
 
 and with finite passives, 
 
 K 44, II azpm tot ais paouruye fravoividey *I was selected by, i.e, 
 for, you for this from the beginning.' * 
 
 {c) With the part, necessitatis, 
 
 V. 6, 32 vaso aiwisx^areda pasubya vtraeibya^ *fur [von] Vieh 
 und Menschen nach Wunsch zu geniessen,' Hubschmann, 
 p. 223; 
 
 K 19, II aetatca rid vaco fravaoce sixsaemca^ hismairimca yaQana 
 kahma icit kdtqm — * and this our word has been spoken to be 
 learned and considered by every one of Uving beings,' Reichelt, 
 P- 334. 
 
 (d) With the verbal in -tua, 
 
 Yt. 13, 153 taca 7)ohu . . . ya frayasaBwa naire asaone, *and 
 the good things that are to be revered by the righteous man,' 
 i.e. ' are for him to be revered.' 
 
 Thus, Avestan likewise offers a parallel to what we have con- 
 ceived to be the development of the dative of interest into one of 
 agency, appearing first with the p.p.p., then, or almost simultane- 
 ously, with other verbals, and finally with the finite verb after the 
 creation of the passive. — A curious point of resemblance is that, 
 whereas we have first met the genitive with the p.p.p. in an * ad- 
 nominal ' relation, § 58 ff., we find in Avestan the later and natural 
 development of an * adverbal ' genitive with the finite passive it- 
 self, so 
 
 Yt. 13, 50 kahe no tSa nqma ay airy at kahe vo urva frayezyat, 
 * by whom will now our name be praised, by which of you 
 our soul revered ? ' * 
 
 1 Reichelt, p. 330 ; Bartholomae, PV6. 929. 
 
 2 Bartholomae, PVd. col. 131 9, and Ar. Forsch. 2, 174. For the form of the 
 verb cf. Reichelt, § 614. 
 
 8 From six^aya^ PFP. of sak^ adj. " was zu lernen ist, discendus," Bthl. Wb. 
 1580. 
 
 * Cf. Reichelt, p. 259, note, and for the form of the verb, p. 299. 
 
60 
 
 § 71. The Instrumental of Agency. — Its development might 
 thus be portrayed : ^ 
 
 (a) Pure concomitation ; in Old Persian the analytic expression 
 with prepositions is the rule. The verb itself may contain the 
 idea, as 
 
 K 32, 2 sarsmno vohu mananha^ * united with V. M. ' ; or other- 
 wise, 
 
 y. 46, 10 tais vt spats cinvato frafra p9r9tum^ 'with all these I 
 will cross the Cinvat Bridge.' 
 
 (J?) Means of locomotion, 
 
 Yt, 10, 67 yd vasa mainyu ham fdsta . . . fravazaife^ ' who rides 
 
 forth with a steed made by the spirit ' ; 
 V. 13, 161 pairitaxto paurvaeibya^ ' running forward on, i>e. with, 
 
 the front paws.' 
 
 {c) Whereas the sociative use, as expressed by the pure instru- 
 mental form, is in decadence, the instrumental of means, soon to 
 pass over into the prepositional construction in Old Persian,^ is 
 preserved in its entirety in Avestan. I mention but one example 
 of this type which is to be found in almost all languages, viz. ' to 
 bind with iron fetters,' from Spiegel, p. 426. 
 
 Vd. 4, 147, ayaidhaeridis fisbis azdibis paiti ava.pasat^ 'with iron 
 fetters should one bind his body.' 
 
 Cf. Earth. Wb. 1029. For rivalry on the part of the genitive of 
 agency, cf. Hiibschmann, 277, as in 
 
 Vd. 7, 74 hak9r9t frasnahaypn maesmana pus hakprpt zpmo uzdd- 
 Oaysn hakdr^t apo frasnahaypn, ' they are to wash it once with 
 the urine [inst] of a cow, once treat it with earth [gen.] and 
 once wash it with water [gen.].' Cf. also Vd. 8, 37. 
 
 (d) Persons used as means, 
 
 K 44, ^ kp ya ma uxsyliii nprpfsaitt, 'who is it thru whom the 
 
 moon waxes and wanes ? ' 
 Y. 44, K^ kp ya usk arpm.pidwa xsapaca, 'thru whom are the dawn, 
 
 the midday and the eventide ? ' So K 41, 4. 
 
 1 Cf. Hiibschmann, p. 254 ; Reichelt, p. 232 ; Spiegel, p. 423. 
 ^ Cf. Hiibschmann, p. 298. 
 
61 
 
 (e) The inst. of means with the past participle is evidently the 
 precursor of the inst. of agency with the past participle as well as 
 with the gerundive in -ya ; so, cf. § 37, 
 
 y^' 5) 93 daxsta daxstavania^ ' mit dem Zeichen gezeichnet,* 
 
 Reich elt ; 
 K 43, 10 parstptn zt Owa, ' interrogatum enim a te,' Bthl. Wb. 997 ; 
 Yt 10, 2t^ frazinte anasita maedanya, * the houses not inhabited 
 
 by posterity'; Hubschmann, p. 261 ; cf. Bthl. Wb. 1706, s.v. 
 
 say. 
 Y. 31, I agusfa vaca spnghamaht aeibyOy "words heard not by 
 
 those wicked men." ^ 
 
 For the gerundive cf. 
 
 Vd. 3, 2^ya karsya karsivata^ 'that is to be plowed by the 
 plowman.* 
 
 (/) With finite passives, 
 
 y. 29, 4 ya zt vavprpzoi daevaisca masyaisca, * which have been 
 
 carried out by Daevas and men ' ; 
 32, 12 yais grphma asdt"^ varata^ ' by whom G. and his followers 
 
 were preferred to Asa' ; 
 43, II hyat xsma ux^ais dldaii^ke^ ' when I was taught the proverbs 
 
 by you.' Cf. also K 43, 6 ; 50, 5 etc. 
 
 If Spiegel's translation of D 3, "Hochbau von Stein, durch einen 
 Clangenossen des Konigs Darius ausgefiihrt " is correct, then we 
 would find a prepositionless inst. of agency even in Old Persian, 
 64 L ardastdna dthangaina Ddrayavahush naqahyd vlthiya^ apud 
 Hubschmann, p. 298, and given in Bthl. Wb. 193 as ardastdna^ 
 aOa^gaina^ ddraya^vahaus viQiyd karta^? 
 
 § 72. In Avestan, too. Dative and Instrumental meet in the 
 function of agency, viz. both are found with the past participles in 
 
 1 Cf. p. 3 of A. V. W. Jackson's A Hymn of Zoroaster ^ Yasna 31, Stuttgart, 
 1888; also p. 17 of the unpublished proofsheets of his Sketch of Avestan Gram- 
 mar, II., in type in 1892, accessible to me thru the kindness of Professor Jackson. 
 For the same view cf. Bthl. Wb. 49, as against Bthl.'s own former opinion in 
 § 55, 2 of Grundriss der Iran. Philologie. 
 
 2 Ablative, Bthl. Wb. 1361. 
 
 * Cf. also Tolman, Ancient Persian Lexicon and Texts, 1909, p. 36, Dar. 
 Pers. c. 
 
62 
 
 'ta, with gerundives in -ya^ and with the finite verb in the passive. 
 As all these uses occur simultaneously from early beginnings, con- 
 taminations of one case by the other, if any have taken place, can- 
 not be traced chronologically. But from what we have seen of the 
 possibility of individual developments on the part of both dative 
 and instrumental, such assumption is neither desirable nor neces- 
 sary. The state of affairs is different, however, in the languages 
 which are discussed in the next chapters. 
 
 i 
 
CHAPTER VII 
 The Dative-Instrumental of Agency in Greek 
 
 § 73. As indicated in § 27, Greek does not possess two separate 
 case-forms for the dative and the instrumental, and the functions of 
 both are discharged by a case-form called dative, which in the 
 singular is partly datival, partly locatival in origin, in the plural 
 partly locatival, iroai, partly instrumental, Tinrots; so that vrjt is 
 loc.-dat.-inst. in function, cf. Sk. navi, nave, ridva; Attic Xvkw is 
 dat. as well as instrumental [as the petrified forms a/ia, TreSa, Trapa] 
 and locatival [as oikoi, domi'\, cf. Sk. vrkaya, vfka, vrke ; lttttol^ 
 stands for inst.-dat.-loc, cf . Sk. vfkats, vfkebhyas, vrkesu ; vava-L for 
 locative (as *A6rjvrj(n), dat. and inst., cf. Sk. nausu, lidubhyds, nau- 
 bhis} Consequently in instances where both a dative and an 
 instrumental function are conceivable, it is extremely difficult to 
 decide between them. This is especially true of the dative of 
 agency, since it might often represent an original instrumental. 
 " In diesem Gebrauche," says Brugmann, " hatten sich Dativ und 
 Instrumentalis beriihrt, und sie sind ofters schwer gegen einander 
 abzugrenzen." ^ 
 
 § 74. As to the opinions of grammarians, I do not find anything 
 pertinent in Apollonius Dyscolus, altho he mentions the fact that 
 TO. y€ fXYjV TrXayMV aTraLTovvra TrdvToyq kol ds TraOrjTtKriv {rrjv iv€pyrjTL- 
 Krjv) SLaOecTLV fJL€Ta<TT'^crtL€Vy crviXTrapaXa/x^avoiiievrj^ yeviKrjs fiera T7J<s vird 
 7rpo6€(T€(o<; . . . KOL avTY) fx\v fJLOvrj ia-TLV Twv TraOrjTLKiov -7 owTa|ts.* 
 Modern writers are, in general, divisible into two camps, the majority 
 being in favor of the interpretation of the Greek dativus auctoris 
 as pure dat. commodi, while the minority give due recognition to 
 the instrumental ingredients of the construction. 
 
 § 75. Kiihner* says " Da der Dativ uberhaupt den person lichen 
 Gegenstand bezeichnet, der bei einer Handlung betheiligt ist, und 
 
 1 Brugmann, Krz. Vgl. Gr. II. § 531. 2 Cr. Gr."^ 209. 
 
 2 P. 404, 2^ Grammatici Graeci, Ap. Dyscoli Ile/oi SuvTa|ews Bt/SXta T€<r<rapa, 
 cd. Schneider et Uhlig, Leipzig, 1910. 
 
 * Ausfiihrliche Grammatik d. griech. Sprache^^ 1870, II. 368. 
 
64 
 
 zwar so, dass er bei'der auf denselben gerichteten Handlung des 
 Subjekts zugleich selbst als thatig und personlich entgegenwirkend 
 erscheint ; so driickt die Sprache die thatige Person beim Passive 
 haufig und bei den Verbaladjektiven auf tos und tcos welche passive 
 Bedeutung haben, regelmassig durch den Dativ aus." Kiihner- 
 Gerth ^ admits " ebenso beriihrt sich beim Passiv der Dativ der 
 thatigen, beteiligten Person mit dem Instrumentalis des Mittels." 
 Like Riemann-Goelzer,^ to whom the dative-agent is "la personne 
 pour laquelle telle ou telle action est, sera, a ^t^ un fait accompli," 
 so K. W. Kriiger ^ sharply differentiates between the instrumental or 
 * dynamic ' dative * and the dat. of persons with passives which he 
 considers a dat. possessivus. So, too, Hirt, without reference to 
 an instrumental component of the dative.^ For Brugmann, how- 
 ever, cf. § 73, and also Gr.Gr? p. 400 ff. 
 
 § 76. As against the foregoing, we find, as early as 1829, 
 Bernhardy speaking of the instrumental constituents of the Greek 
 dative and describing the personal datives with passives, like 
 &TroK€KpLTcu fioL, as instrumental.^ So, too, W. E. Jelf,^ " the logical 
 subject of the impersonal verbal adjectives, the agent or person by 
 whom the verbal operation is to be performed, stands in the instru- 
 mental dative, as in the passive voice." More recently, R. Wagner* 
 under the ' Dativ als Instrumentalis,^ " Auch der gewOhnlich als 
 Dat. des Interesses erklarte Dat. das tatigen Individuums beim 
 Passivum wird, wie der des Arischen und Slavischen, vielmehr hier 
 einzuordnen sein : kyjoi TroXefxrjTiov iartv heisst ' mit meinen Mitteln 
 an Kraft ist zu kampfen,' iravS* ruxlv TreTrotrjTai, nicht ' alles ist fiir 
 uns = von uns,' sondern ' mit unsern Mitteln an Kraft geschehen.' " 
 This would agree with Delbruck's indication ' of § 434 in Curtius' 
 Sc/iulgrammatik*, — which is *' die tatige person beim passiv," as 
 belonging to the functions of the instrumental dative. 
 
 § 77. Each side has its own justification. Working a priori 
 from the undoubted existence of an I-E. instrumental ; from the 
 existence, also, of one of its functional types denoting agency with 
 passives ; and, then, from the subsequent historical fusion in Greek 
 
 1 Gramm? II. i. (1898) § 422, 4. 
 
 2 Gram. comp. du grec et du latin, 1897, P- 248. 
 
 8 Attische Syntax^, 1862, § 48, 15. * To which cf. Quintilian, Inst. I. 2, 46. 
 
 ^ Handbuch, p. 227. ^ Wissenschaftliche Grammatik, p. 103. 
 
 ' A Grammar of the Greek Language^, Oxf. 185 1, ed.* 1866, § 613, 5. 
 8 In Freund's Triennium Philologicum 2\ 1908, p. 147. * ALL p, 73. 
 
65 
 
 of that I-E. instrumental with the dative, one would be inclined to 
 reduce all Greek datives of agency to an original instrumental. 
 Such a stand, however, would be as false as that view which, satis- 
 fied with the mere disguise of instrumental functions under a 
 * datival ' garb, would without further ado class such instrumental 
 datives under genuine datival usages. This is one of the situations 
 where a well-meant ' lo non posso ' is preferable to a presumptuous 
 dictum. Follows a tentative reconstruction of a few probable facts 
 in this vein. 
 
 § 78. This so-called dative of agency may be used in three 
 different ways, viz. : 
 
 {a) with verbal adjectives in tos, 
 
 Tots ovK i^LTov iaTLv, * quibus non licet exire,' Hes. Theog. 732 ; 
 {b) with verbal adjectives in -t€os, 
 
 ov o- ^ t TrepLOTrrirj iarl 17 'EAAas OLTroWvfieirrjf * they must not permit 
 Hellas to perish,' Herodot. 7, 168 ; 
 
 (c) with finite passive verbs, mostly in the perfect tenses, 
 
 &v TovTo viKiofxev, irdvO* rifxiv TreTroLrjTai, 'everything has been accom- 
 plished by us,' Xen. An. 1,8, 12. 
 
 These three ways will now be discussed in detail. 
 
 The Verbal in -t6s. § 79. The Greek suffix -to? is the same 
 as the participial suffixes Latin -fus, Sk. -fas, Avestan -fa-, Slavic -A?-, 
 Lith. -fas, Germanic -da ; so that, for instance, 
 
 Gk. kXvtos '^ Sk. srutds '^ Lat. in-clutus 
 
 fiaro^ gatds ventus < *gUt'^tds ; 
 
 in Greek, however, -r6% did not come to take the place of the past 
 participle in just such a degree as in the other languages and re- 
 mained a verbal, developing very early a nuance of possibility.^ 
 It does not seem to have possessed a distinction of voice at the 
 beginning of the specifically Greek period, cf. Plato TrXavTyros errans, 
 altho its employment as a p.p.p., like Sk. uktds in both passive 
 and perfect sense, must have been quite prevalent, because of the 
 extension to the domain of derivative verbs, as wviyros purchased, 
 
 1 dXvTos, indissoliibilis > Xvt6s solutus and dissolubilis, cf. Gr. Gr? 200 ; 
 Grdr? II. p. 207. 
 
do-Ki^Tos manufactured, dyaTn/ros beloved, etc. How closely allied 
 these two ideas were is seen in the readiness with which the verbal 
 lent itself to being made a p.p.p. in Latin. 
 
 § 80. Delbriick^ states that in combination with the dative- 
 agent the verbal -ros to denote agency is found first in Hesiod. 
 However, cf. li 620 TroAvSaKpvros Se rot lorTai, * much lamented by 
 you,' or T 404 TToXvapT/Tos Se rot caTt, ' much desired by you.' Be- 
 ginning with Hesiod, however,^ it is a regular combination. So, 
 
 S. Ph. 33 (TTetTTTi; yc <^vAAds, ws cvavXt^ovrt tw, 'crushed as if by 
 
 some one resting on them ' ; 
 Xen. An. 3, 4, 29 ^viirCterov rjv rot? TroXc/xtois ; 
 Xen. Q'/'. 3, 2, 25 TOts dAAois ravr* cvktci etr]. ] 
 
 § 81. What this dative represents cannot be decided directly by 
 the testimony of the related languages. Latin dative balances 
 against Slavic instrumental [with p.p.p. in -wb], and Indo-Iranian 
 is non-committal because of the presence of both. I venture to 
 think of the dative as the original case in this connection, with the 
 verbal denoting that the deed accomplished was at the disposal of 
 some one. Such a dative of personal interest with a verbal denot- 
 ing completion may easily be conceived as one of the most primi- 
 tive locutions, certainly antedating all idea of passivity, and more 
 primitive at all events than the comparatively complex and sophis- 
 ticated notion of an instrumental of personal agency, as distinct 
 from and more advanced than the equally primitive instrumental 
 of material means. That the latter was bound in time to develop 
 into personal agency, we have noted before. That it developed 
 so after the use of the dative with the p.p. was established, we 
 cannot of course demonstrate chronologically, since at the earliest 
 moment we find in Indo-Iranian both expressions of agency exist- 
 ing side by side. Possibly a careful analysis of the respective 
 number of instances in relation to the comparative antiquity of the 
 documents, on the basis of such indications as Ch. III. and X. of 
 Arnold's Vedic Metre in its Historical Development, Cambridge, 1905, 
 and in Avestan by the testimony of the Ga^as (cf . Reichelt, p. 9), 
 might throw a faint light on the primitive connections between the 
 two cases. Here, however, we must content ourselves with re- 
 marking that a simple statement of individual personal interest, 
 
 1 Vgl. Synt. II. 484. 2 Cf. Theog. 732, above. 
 
67 
 
 one nearest to the primitive man's mental make-up, in a deed ac- 
 complished by him, probably antedates an idea at which he in all 
 probability could arrive only after a relatively complicated process 
 involving a transition from material to personal considerations, 
 from an active to a passive mode of thinking. '■ Pitrbhir dattdh^^ 
 to my mind, represents a considerable and significant advance 
 over ^ml krtam.'' It must be added at once that, should this 
 reasoning be applied not to the earlier but to the later stages of 
 I-E. which, far from being primitive, presupposes an already pro- 
 longed development of the instrumental, an admixture of the force 
 of the latter can easily be imagined. 
 
 § 82. For Greek itself such a blending of the two case-forces 
 must actually be postulated, if for no other reason, then surely 
 because of the historical fact of a dative-instrumental amalgama- 
 tion. Significant in this connection is the use of instrumental 
 datives of the type described by Brugmann as " die den Vorgang 
 bewirkende und hervorrufende /<?/<?«s " (Grdr.^ II. p. 527) with this 
 same verbal in expressions of possibility, as 
 
 I 526 SioprjTOL T lirikovTO TrapdpprjTot re €7r€€0"(riv, * how he by gifts 
 was softened and soothed by accents due,' and 
 
 N 323 ^a\K(S T€ p7jkt6<s fJ.eyd\oL(TL t€ ;j(6p/M.a8toto-t, 'who may 
 by brass be wounded or huge stones flung afar ' ; so <I> 568.^ 
 
 The Verbal in -tcos. § 83. Ascoli's view^ that the form origi- 
 nates in *-TLo [^ereo — Sk. saf/a'], so that Hesiod's <^aT€tos < *<f>aT€- 
 fLo^, held even by Brugmann in Grdr} II. 1421, is now no longer 
 accepted. G. Meyer ' would derive it from -ros ; Hirt ^ believes 
 that both Sk. tavya and -reos are based on /z^-stems. Brugmann* 
 has -T€os < *-Tc/ro? < infinitive form *-T€/:ai, Vedic -tave and -tavai^ 
 just like participial -/acvos < infin. -jutcmt. — The employment itself 
 of the verbal, denoting necessity, is a post-Homeric usage appear- 
 ing particularly in Attic. There is no trace of it in Homer;® in 
 the few instances in which its form occurs at all it seems to be 
 exchangeable with -ros, as vrTyarcos, B 43, H 185 : ^veiyyaros ; Pporir), 
 T 545 \Ppor6<i\ ireov, t 176. Hesiod's <^aTc[i]os, Theog, 310, 'Ao-tt. 
 144, 161, is the first example of it as a part, necess.^ Meyer is 
 
 1 Cf. Vgl. Synt. II. 486. 2 Sprackwiss. Briefe, p. 69 ff. 
 
 8 Gramm? p. 671. * Handbuch, p. 438. ^ Gr. Gr? p. 525. 
 
 * Vogrinz, Grammatiky p. 141 ; G. Meyer, p. 671. " Schulze, Qaest. ep. 433. 
 
68 
 
 probably incorrect in discarding the Hesiodian instance and assign- 
 ing the first use to Herodotus. With the later writers the employ- 
 ment of -reos is quite frequent, the agent upon whom the obligation 
 rests to perform a given action being denoted by a dative, and 
 never, as in the case of Lat. gerundive and ab c, abl., by vno c, 
 gen, Thus,^ 
 
 Xen. Oec. 7? 35 ols av IvSov ipyaa-riov, rovriav <roL lirurraTrfiov'^ 
 Dem. 6, 28 Trcpt tcov v jxIv TrpaKTeoiv varrepov /^ovXevaecrOe ; 
 Thuc. 8, 65 fxeOeKriov tw irpayfrnrmv ttXcloo-lv. 
 
 § 84. The morphological proportion suggested by Brugmann, 
 *-T€po<i [> -T€os] : ^-rcpai : : -/Ltevos : -/xevat gives a good starting point 
 for the investigation of the origin of the dative. For if the formula 
 holds good, and it seems to leave nothing to be desired, the case 
 used to designate the agent may be brought in line with the wide 
 appearance in the Indo-European group of languages of a dative of 
 agent with a predicate infinitive. — What the Greek verbal denotes, 
 the Sanskrit renders by a verbal adjective in -tavya. This, how- 
 ever, is only post-Vedic^ and may be resolved into -tav-{e) and 
 -ya, cf. also stuseyya-s < stuse^ ' to be praised < to praise," which 
 would amount to the adjectivization of the infinitive stem or even 
 form by means of a suffix. Now, in the Vedas not only the gerun- 
 dive with this suffix -ya is employed,^ in J^ V. to the number of 40 
 [Whitney], but the bare infinitive is also found used as a part. nee. 
 with the termination -e, -tavai, and -tave.^ Like the Greek verbal 
 in -Tos denoting possibility and the Latin gerundive denoting neces- 
 sity, this predicate infinitive first occurs in a negative sense, as R V. 
 I, 54, I nahi fe dntah sdvasah parindse^ 'the end of thy might is 
 unreachable.' Whereas post-Vedic -tavya takes only an instru- 
 mental agent, M.S. 1,5,7, agnihotrina nasiiavydm, ' an A. should 
 not eat,' in the Vedas -tave requires that case only when the mean- 
 ing is strongly passival, RV, 6, 56, i nd t^ena deva adise, 'the god 
 is not to be scorned by this one,' cf. too 2, 16, 3, so that, discount- 
 ing the everywhere competing genitive, solely the dative-agent is 
 in evidence, RV. 10, 125, 6 brahmadvise idrave hdntava u, 'so 
 
 1 Cf. Moiszisstzig, Quaestiones de adiectivis graecis quae verbalia dicuntur^ 
 4 parts, Prog. Konitz, 1844-68 ; Kuhner-Gerth, p. 447 ; Gr. Gr.^ p. 524 ; 
 Krviger, Synt. p. 229. 
 
 2 S.F. V. 398. 8 Krz. Vgl. Gr. II. 605. 
 
 * S.F. V. 396. 6 Vgl. Synt. II. 460 ; S.F. V. 410 flf. 
 
that the arrow may strike the hater of prayer,' or, without any 
 final tinge whatever, but purely predicatively, RV. 2, 11, i sjama 
 te ddvdfie vdsunam, 'who must be well received by you'; so 2, 11, 
 12. This fact seems to be a significant indication of the original 
 datival character of the agent with such infinitives.^ — For a similar 
 infinitive in Avestan cf. K 29, 3 nditvtduye ; K 45, 4 tidit diwzaid- 
 yai; possibly K 45, 5 is an example of the dative-agent,^ tho I am 
 inclined to place it elsewhere, cf. § no. 
 
 § 85. Similar combinations of the dative with the predicate in- 
 finitive are found in Slavic, the dative denoting the person on 
 whom the obligation rests,^ nesth nam-h ubiti, ' occidere non debe- 
 mus,' Sup. 325, I, and in Russian, a jemu sderzati carstvo moskov- 
 skoje, '■ and he is to rule over the Empire of Moscow.' In Lithuanian 
 the agent with a similar infinitive may be left out altogether, as 
 af eit ar ne, ' shall I go or not ?,' but when it is expressed it is in 
 the dative, ^ de waiku aniemus mirti, sine liberis moriendum illis 
 est.'* 
 
 § ^6. In Latin this infinitive construction has, it is well known, 
 been taken over by the gerundive. It is interesting that the sema- 
 siological process is similar to that observed in the transference 
 of the idea of necessity from the Vedic infinitive to that post-Vedic 
 gerundive which was formed from it by means of a sufiix. One is 
 tempted to see a perfect analogy between the two languages, in 
 that there is high probability of the Latin gerundive form itself 
 originating from an infinitive. Brugmann^ — to mention but the 
 latest attempts at explanation, tho cf. also Dohring, Die Etymologie 
 der sogen. Gerundivformen, 1888, and Havet in Mem. de la. Soc. de 
 Ling. VI. 6 — at first ^ identified the suffix -ndo with Balto-Slavic 
 -tno [Lith. siik-tina-s ' torquendus^^ to infin. suk-ti\ then, corrected 
 by Conway,' he propounded a new theory,^ according a Proto-Italic 
 character to the formation of the gerundive and referring it to the 
 infin. ace. in -w, like Umbrian fdsiu\_m\ Lat. fero\fn\ somewhat 
 on the analogy of NHG. ' der zu lobende, ein zu lobender ' beside 
 ^ zu loben,^ or Sk. sravay-iyas beside ^sravai. L. Horton-Smith * 
 
 1 For examples, etc., cf. S.F. V. 399, 422, and KZ. 18, 93. 
 
 2 Cf. Reichelt, pp. 337, 340 ; Bartholomae, KZ. 28, 26. 
 
 8 Miklosich, IV. 859. 4 Schleicher, p. 312 ; Vgl. Synt, II. 461. 
 
 fi Add to this sketch § 44 above. « Grdr.^ II. § 69, p. 152. 
 
 ' Class. Rev. V. 296. 8 Grdr.^ II. § 1103. 
 ^ Cf. references in § 44, footnotes. 
 
70 
 
 expresses doubts about Brugmann's conjecture as to the possibility 
 of -ndo arising from a postpositive relation of a suffix *dd/e, mean- 
 ing ^ to, zu,^ ':> -md-> -nd-, like Umbrian /d!«^ and Lat. quam, and 
 substitutes for it a suffix -do^ derivable from either ^/dd or ^ dKe^ 
 or both, attached in Primitive Italic to the Prim. Ital. infinitive, 
 a substantive in -m, and governing the infin. as its object ; so edum 
 •^do> edundo, + nom. masc. suffix s > edundus, and edendus on 
 the analogy of pres. part, edent-. Fay ^ refers Lat. fer-en-d-ae to 
 an I-E. bherndhdi, Sk. bhdr-a-dh-yai, cf. Gk. <f>€pe(a)$-m, with the 
 form in -en being an accusative to a root-noun, in effect an infini- 
 tive combination with the root ^d/ie, so that bhdradhyai originally 
 meant ' for putting into bearing,' a possible case of survival of -ndae 
 in infin. function being Epid. 74 Puppis pereunda est probe, if the 
 iQ2i6mg pereundae is taken, ' is for perishing.' 
 
 § 87. It will be noticed that these leading explanations all meet 
 on the common ground of referring the gerundive of the Latin to 
 an original infinitive. For the regulation of this infin .-gerundive 
 with a dative agent, cf. § 44 ff . The conclusions from the fore- 
 going remarks establish, it is hoped, the analogous stand of the 
 dative with the Greek part. nee. within reasonable certainty, as 
 being virtually equivalent to that in X 76 laaojxivoun irvOiaOai where 
 the dative appears with an actual predicate infinitive. Nor is 
 there any reason to think of an instrumental contamination, be- 
 cause even in Sanskrit a material havyair agnir manusa trayddhyai, 
 * is to be quickened thru the sacrifice of men,' RV. 4, 2, i, develops 
 into the personal agent but sparely, and mostly in post-Vedic litera- 
 ture.' It is significant, besides, that when the verbal -reos first 
 appears with this dative, — it is in Herodotus, — the latter is solely 
 in the form of a pronoun, as I. 191, 3 to iroL-qriov ot rjv,^ To which 
 
 cf. §§ 45 ^. 60. 
 
 The Finite Verb in Passive. § 88. In contradistinction to the 
 verbal adjectives we must postulate for the dative-agent with finite 
 verbs a partial instrumental force of more than conjectural stand- 
 ing, for the ease with which the impersonal, i.e. material, use of the 
 
 1 " The Oscan-Umbrian forms bear upon the much-disputed question of the 
 origin of the Gerundive to this extent, that they are unfavorable to any theory 
 which assumes that the original form contained ndh." — Buck, Grammar of 
 Oscan and Umbrian, Boston, 1904, p. 181. 
 
 2 AJP. XV. 217. 3 s.F. V. 399, 422. 
 * Cf. Helbing, E. u. S. Dat. p. 39, note 2. 
 
71 
 
 instrumental can be extended to the personal use, which we have 
 noted in the related languages, must apply" equally to the Greek 
 passives. The entire series of presumably I-E. transitional pro- 
 cesses is faithfully reproduced in Greek, naturally in the guise of the 
 so-called dative, both in the earliest poet, Homer, and the earliest 
 prose writer, Herodotus.^ Thus we find, cf. §§ 54, 68, 71, 
 
 (a) pure concomitation, naturally with verbs like ipi^oi, hroimi and 
 o/xiAe'o) ; ^ fidxofmi alone takes in Homer 80 bare dat.-instrumentals, 
 Walther, 19. Examples of a bare instrumental of association, as 
 apart from verbs, are rare, cf. 
 
 S506 TOLO-Lv CTretT* Tjiacrov SC. crKrjrirpa €;)(ovt€s, 
 
 A 163 ^ vvv 8r] Tpoirjdev aXwixevos ivdaff tKaveis vrjCre Kal erdpoLO-L 
 iroXvv xpovov, 
 
 where vrjC re koL irdpoLa-L belongs to <iA.<i>/x€vos in spite of a 182 
 (cf. S.!*". IV. 58); accordingly erdpoLai is sociative. That it has 
 no (Tvv is undoubtedly due to the influence of vrji. Cf. also 
 
 P 460 tTTTTOis dto-orwv and 8 8 LTnroia-L Kal apfxa(riv ttc/x-ttc vcec^at, 
 
 where the idea is, however, non-personal.' — The dative combined 
 with the attributive pronoun avros, in Homer only non-personal, 
 ^8dAA' avTOts LTTTroLa-i Kal app^acrtv aairov lovres . . .j but post- 
 Homerically also in turns like avrots rots avSpao-t, cf. Thuc. 4, 14, is 
 also considered sociative."* Avros originally belonged to the main 
 idea and by some inexplicable syntactic metathesis became at- 
 tracted and attached to the inst., so that avrot iTrTroca-L lovres > avrots 
 LTnroLo-L lovres. — In military expressions the troops, ships, etc., are 
 so many companions of the leader, hence the sociatives a-rparw, 
 vav(TLV, iTTTrevaL, etc. For HerodotUS cf. 3, 54, l o-ToA<p fieydX<o ws 
 
 OTTLKOVTO, but 
 
 (^) already in 8, 17, 7 €<rrpartvero avhpdcn re SirjKocrLOLorL kol oIkyjIy) 
 vrjL we can talk no more of a sociative sense. Cf. too. Soph. At. 
 
 1 Cf. "Walther, de dativi instrumentalis usu Homerico^ Breslau Diss. 1874; 
 and Helbing, tjber den Gebrauch des echten u. sociativen Dativs bei Herodoty 
 Diss. Freiburg, 1898, and Der Inst, bei Herodot, Program Karlsruhe, 19CX). 
 
 2 Add the following authorities : Kiihner-Gerth, Ausf. Gr.^ 430 ff . ; Kriiger, 
 Syntax^ p. 75 ff.; Vogrinz, Gram. d. homer Dial.^ p. 312. ff. 
 
 8 Cf. Holzweissig, Uber den soc.-inst. Gebrauch d. griech. Dativs ^ 1885, p. ii ff. 
 '^ S.F. IV. 58; Vgl. Synt, I. 236 ff.; Walther, p. 16; Gr. Gr? 408, and 
 Hentze, Philologus 61, 71. 
 
72 
 
 767 ^eots Kparos KaraKTrja-atTo, * by the aid of the gods.' So, nearer 
 to the inst. of means are 
 
 K TOLS LTnrOKTLV €Tro^<T€rai 
 
 P 448 apfiaa-LV 8at8aA.€otortv iTro^T^aeTOL, 
 
 (c) For instances of the instrumental of means cf. Walther, 25 ff., 
 and Helbing, Der Inst, bet Herodoi ; also Gr. Gr? 410, Kiihner- 
 Gerth^, 435. They are innumerable. The expression a-fjitXTf 
 •jrtTroL-qfxevov, ' made with the graving tool,' shows the construction 
 with the perfect participle. 
 
 (d) Persons and personified objects are often employed as 
 means or instrument, 
 
 S. Ant. 164 vfxas 8' €y(b 7rofJLTroL<rLv t(Trei)C iKeaOai, per nuntios ; 
 
 Eur. Heracl. 392 ovk dyycXotai ror? cvavrtovs 6/aav ; 
 
 with passives Th. 4, 39x0,5 Sc oAAas [ly/tAcpas] rots co-ttAcovo-i \ddpa 
 Si€Tpic}>ovTo, ' durch die heimlich Einfahrenden,' Kiihner, Gram} 
 378: "vTTo Ttov kd'wX'e.ovTiiiv personlich aufgefasst." 
 
 So t 445 Xa;)(vw OTTCtvo/xevos Kat e/AOt', TrvKtva <^pov€OVTt, 'thru me'; 
 also a 280, H 475, and H 546. This concept is as near to that of 
 agency as it is possible to approach it without crossing the divid- 
 ing line. It is, for instance, difficult to decide in examples like 
 the following which type we have at hand, 
 
 Xen. An. 6, 4, 27 fjivXaTTOfievoi tKavois <^vXa^t; 
 S. At. 539 Trpoo-TToAots <^vXao-o-€Tai.^ 
 
 Certainly they are not far removed from such examples of agency 
 3iS V 28 Xaola-L TCTt/xeVos, or the one cited by Meisterhans from an 
 isc. (Gr.2 156, 172). 
 
 ttTTo Tov w<f>Xr]ixivov 5<i)7r(iAi8i apyvpiov, 'of the money owed by 
 [not due to] SopoUs.' 
 
 (<?) That with finite verbs there should be no doubt at all as to 
 an instrumental possibility, is further evidenced by personified 
 instruments found with passives, 
 
 Homer, M 207 Trerero ttvoly}^ dve/xoLo; 
 
 <l> 524 (f>€p6iJLrjV oXoots aviixoia-iv\ 
 i 299 [253] Boperf aV€fx<o. 
 
 1 Cf. Fairbanks, The Dative Case in Sophocles, p. 117, vol. 17, TrAPA. 
 
73 
 
 (/) Still, when we come to the question of passive verbs with 
 personal agents, we are confronted with an almost absolute uncer- 
 tainty. We possess in Greek a parallel series of gradations, lead- 
 ing up to a legitimate instrumental of agency, of stronger cogency 
 than even in Sanskrit or Slavic, but unfortunately the topmost 
 point itself in the development is obscured by two circumstances, 
 at least. The first one is that the very juxtaposition of a ' dative ' 
 form to any verb is sufficient to render its datival force eo ipso 
 paramount and even exclusive. This statement is not intended to 
 deny the potentialities, inherent in the dative itself, of an eventual 
 transition from the function of interest to that of agency ; at the 
 same time it is manifest that the remains of an originally I-E. in- 
 strumental force are obscured by that very process of case-substi- 
 tution which has brought the dative to the fore. In a word, the 
 matter is one of popular linguistic feeling: had the reverse oc- 
 curred, had the instrumental remained a living force and the dati- 
 val function merged into it, we would to-day undoubtedly claim an 
 instrumental connotation in all instances under consideration with 
 as much insistence as some grammarians at present champion the 
 cause of the dative. — Secondly, the fact that this dative of inter- 
 est, standing apparently for viro c. gen. appears in the overwhelm- 
 ing majority of instances with the perfect and the pluperfect and 
 seldom with other senses (so cf. Kiihner-Gerth^, 422), is calculated 
 to prejudice one against instrumental interpretations. If we recall 
 that the Greek perfect, as in ri tti'Kpa.KTai rots aA.A.ois, Dem. 2, 27, is 
 a present perfect which denotes equally the circumstance that an 
 act has been done by some one and that the act serves some one's 
 interests, we can see a striking similarity to the dative with the 
 Latin periphrastic perfect of the type of *mihi facta sunt,' a use 
 which we found to be I-E. Superficially, then, every Greek verb 
 in a perfect tense may be said to have a dative of agency with it 
 which is really a dativus commodi. 
 
 § 89. But to exaggerate such indications and to insist on the 
 exclusive interpretation of all datives of this kind as datives of 
 interest, is manifestly erroneous.^ Surely, each verb must be studied 
 separately, both as to its meaning and as to its form, then brought 
 
 1 As when Monro, § 143, 5, renders Homeric 56/)u jjtxiKpbv, 8 ol K\i<Tlrj<pi \4- 
 "KenrTo, N 168, by "which for him was left in the tent." According to this 
 reasoning, what one forgets to take along with him is left somewhere /or him. 
 Utinam semper! 
 
74 
 
 in line with linguistic similarities, defined as far as possible by its 
 Indo-European bearings, and only then can a decision be made as 
 to the character of the dative with it. This decision itself will be 
 conditioned by the degree of verisimilitude in the conjecture. 
 The court of last instance, in this matter at least, must not be the 
 Greek Sprachgefuhl at its acme of development, but all those often 
 intangible tokens which will enable the grammarian to trace that 
 Sprachgefuhl to its fountain-head. 
 
 § 90. Thus, for instance, if the verb could command a dative 
 in the active, the possibilities are strong that the dative-agent with 
 that verb in the passive is a retained dative, on the model of Latin 
 ^ emo mihi servum — >- emptus mihi est servus.'' Three categories 
 especially lend themselves naturally to this interpretation, where- 
 fore they are selected to illustrate the case of the dative. 
 
 {a) Verbs of finding, evptcrKO), — t 42 1 ct rtv' kraipoia-iv Oavdrov 
 \v(nv yS* ifiol avTio evpoi/xrjv and <f> 304 KaKov 61 avT(o are media, but 
 in Hdt. we find the compound dvevpLo-KCLv used passively; thus 2, 
 82, 5 ripaTOi re TrXeo) (r<f>L dvevprjTca rj tolo-l oAAown aTracrt dvdp^iroKTL ; 
 of. I, 8, 16 dvOp(i>7roL(n with iievptarKeLv. So Men. Sent. 51 raXTjdh 
 dv9p(x>'7roi(TLV ovx evpLcrK€Tai, etc. 
 
 (d) Verbs of saying, calling, like elprja-OaL, ScSrjXuxrOaL, XeX^xOai, etc. 
 Herodotus uses in a formulaic manner ws dprfaC /xot irporepov, as i , 
 130, 15.^ Similarly i, 18, 9 ws kcu irportpov p.01 ScSTjAwrat, 2, 155, 
 8 o)S Kol Trporepov uivopjaa-rai fxoi, 5, 62, 3 dirriyqTa.L jjlol. Homer, ^ 
 795, ov fiev TOi fifXeo^ ciprjo-erai atvos, V 138, viK>;cravTt <^tA.ry KeKX-qcrri 
 ttKOtrts, to which cf. Soph. O.T. 1359 Pporois iKXrjdrjv, O.T. 8 Trao-t 
 KoXov/xevos. 
 
 {c) Verbs of doing, iroUa), tct^o), etc. Z 56 ^ a-ol dpiara TTCTTOoyTat, 
 i; 342 ireTTOvrjTai 8e tol cvviy ; Herodot. 2. no, 7 ov ot TreTroirja-OaL tpya 
 olaTrep Seo-wo-rpt. — E 446 oOl oi vt/os y€ T€Tvkto, A 671 a)5 ottot 'HAet- 
 otcrt KOI -^fuv vctKos iTvxOr], etc. — irpdrruv is frequent in Attic, cf. 
 Xen. An. 7, 6, 32 kiriirpaKTO vfuv. 
 
 § 91. Besides the fact that with these verbal concepts there are 
 but very few exceptions from the above-mentioned rule concerning 
 the use of the dative of agent with the perfect, pluperfect, and 
 aorist passive, the pronominal nature, in the great majority, of the 
 datives to be found accompanying them, is another proof of the 
 real datival character of these cases. Helbing notes 109 instances 
 
 1 Cf. Helbing, EuSD. p. 35 ff. 
 
75 
 
 of pronouns out of a total of 141 such datives. On a similar con- 
 dition in Homer Havers bases his arguments for the evolution of 
 the sympathetic dative in Greek into a possessive genitive.^ If we 
 add that with the three categories vtto c. gen. never seems to ap- 
 pear instead of a dative of the agent, the intensely personal nature 
 of the connection is reasonably established.^ 
 
 § 92. So much for the dative specifically, besides the remarks 
 in § ^Z f. On the other hand, an instrumental provenience may 
 safely be vindicated in at least one instance, viz. with the aorist 
 Saixrjvai. This verb is used absolutely, as a 237, but more fre- 
 quently with the agent, personal or material, in the dative. Curi- 
 ously, also, it appears with its agent in vtto c. dativo, as N 668 vtto 
 Tp(oe(T(TL SafiYJvai and others,* even w e/Aot as in <^ 213, E 646 ; in 
 Hesiod, too, vtto Kevravpownv Sa/ActTy. The fact that KT€Lv<t} in pas- 
 sive has a similar regimen, as 11 490 vtto HarpoKXta Kretvoyotcvos, 
 makes the following remarks general for all verbs of subjection. 
 At any rate, La Roche, p. 356, is justified in rejecting 11 434 and 
 II 452 VTTO IlaTpoKXoLo SafirjvoL* because " bei Safma) findet sich sonst 
 nur VTTO mit dem Dativ oder der blosse Dativ " and not vtto c. gen, 
 'ESdixrjv, cf. § 5 note, originally must have meant */ became tame^^ 
 which already in Homer's time was equivalent to */ was con- 
 quered.'' The relation of such really intransitive verbs to the 
 agent of their action is mutual ; for not until their agent is clearly 
 expressed, do they appear as definite passives : ' the house burns ' 
 is intransitive until it turns to ' the house is burnt up by some one.'' 
 Frequently, therefore, the force of the agent makes the verb pas- 
 sive. For Latin parallels, as agna cadet vobis, cf. § 52. 
 
 § 93. Delbriick^ thus pictures the case of iSdfjirjv, From * to be 
 tame ' it could be applied to one fallen in battle, as iSaixrj, ' he suc- 
 cumbed.'' The conqueror was conceived of as an orig. instrumental 
 of the sociative type, as U-qXetiovL Sa/^cts, ' gestorben unter Mitwir- 
 kung des Peliden,' in the same sociative-locative sense as 11 543, 
 
 1 Helbing, loc. cit. p. 37 ; Havers, p. 73. 
 
 2 For Herodotus cf. Helbing, p. 37; for Homer, La Roche, Uber den Ge- 
 brauch von virb bei Horner^ in Zs. f. d. oesterr. Gymnasien, 1861, p. 348, and 
 Crusius, Worterbuch, s.v. 
 
 8 Cf. La Roche, ubi supra, p. 349. 
 
 * These are two similar expressions within a score of lines of each other and 
 somewhat doubtful, cf. Ludwigh's annot. II. II. 184, 185 sub. lin, 
 6 S.F. IV. 79; cf. also Kuhner-Gerth8, 123. 
 
76 
 
 Tov 8' vvb HarpoKXw Sdfma ty^d xoXkco? "Aprjs, ' durch P. liess er ihn 
 seinen Tod finden ' ; or A 243 cvt' av ttoXXoI v<f>* "EKropos avhpo<f>6voio 
 OvrjcTKovTcs TrtTTTaxTt, ' wann unter dem menschenvertilgenden Hektor 
 todt hinsturzen die Schaaren.'^ This would be one explanation 
 completely shutting out the datives. That XliyXcWt and the per- 
 sonal datives in similar examples are not mere datives of personal 
 interest, but have primarily a locatival connotation, is evident from 
 the fact that often these datives are strengthened by vtto. Thus, 
 compare the instances cited in § 92, N 668, etc., with 11 854 x^P^^ 
 8a/x€VT beside the frequent virb x^P^' SafxrjjMevaL * unter den Han den 
 und durch dieselben,' as B 860, 847 iSafir] inrb x^P^-' ttoSwkcos AiaKt- 
 Sao; as well as X 246 Sovpt besides E 653, A 444, 748 e/Aw vtto Bovpl 
 
 8ap.€L<s. 
 
 § 94. A development of the inst. of means, however, is also con- 
 ceivable. So (T 54 Lva TrXrjyrja-L Sa/ACt'cu, or X 271 tyxei ifx<S 8a/xaa 
 or n 854 xep<TL, X 246 SovpL, A 99 j8cA.€i, would lead up to per- 
 sonal concepts like Y 294 HrjXctoiVL and P 2 T/owco-o-i, or substan- 
 tives like 11326 Sototcrt KaaiyvrJTOia-L, F 301 aWotcrL and T417 
 deS Tc KOL dvepif thru such natural gradations as M 37 Aios fia- 
 (TTLyt a personified S 352 vttvo) KaX (ftiXor-qTi or, of a river-god, 
 ^ 291 TTorafiS. Other examples are 244, X 55, S 432, X 176, 
 S 103, o- 57. A nice example from the N.T. is James 3, 7 Sa/xa- 
 ^erat kol SeSa/xaoTai rrj <f>v(T€L ry av9p(07rivrj, * can be tamed by the 
 human genius.' "Es musste also der Instrumentalis der Personen- 
 namen hertibergenommen worden sein von Sachnamen," Vogrinz, 
 Grammatik d. horn. Dial. p. 307. 
 
 § 95. Of other aorists in -r[v we have l'K\'r)yy]v with a participle 
 always passival, and so with the inst. of means /x 416 TrXy^ydaa kc- 
 pavv^ ; so ervTrryv, A 206 hovpX tvttcis rj y8A>;/xevos iw. I find none with 
 the personal agent. Like Sa/x^mt are to be judged the passive of 
 dvao-o-o), 'to be ruled over,' as 8 177 dvao-o-ovrat 8' tfxoX avTw, where 
 the ' interest '-element is clearly not in the agent, and KTetveaOaC 
 TivL, * to be killed by some one,' as E 465 i<s rt In KTeivea-daL ida-ert 
 Xabv *Axaiots. Kuhner-Gerth's opinion ^ that all the above are pure 
 datives with passives, in which Monro, § 143, 5, concurs, are to be 
 rejected. (Cf. to this point Brugmann, Gr. Gr? § 471 Anm. ; 
 § 456, 2 Anm. ; § 523, p. 464.) 
 
 1 Ehrenthal's translation ; Voss is more advanced, " wenn sie in Schaaren, 
 vom mannermordenden Hektor niedergestiirzt, hinsterben." 
 
 2 Gramm? § 423. 
 
77 
 
 § g6. The results arrived at for Greek are thus, by the very- 
 nature of things, of a doubtful character. In the instance of the 
 verbal adjectives in -ros and -reos the dative was given preference 
 over the instrumental, in that of finite verbs the impossibility of 
 localizing and even locating the instrumental force was emphasized, 
 and a method was indicated for separation wherever feasible. The 
 Sanskrit could not be drawn upon for aid. After all, it is not the 
 amalgamation of a Sanskrit dative and a Sanskrit instrumental 
 that we have in Greek, but in the last analysis the fusion of an I-E. 
 dative and an I-E. instrumental without the pecuHar features of 
 Indo-Iranian additions and extensions. And this is just the reason 
 why we are forced to bow to a semblance of things : the dative is 
 in charge of the function of agency thruout; the interpretation, 
 therefore, of the dative-agent as an original dative is the easiest 
 procedure. Even Safirjvcu tlvl might be suspected of analogy with 
 VTreUeLv tlvl, as in /x i i 7 Ocolctlv viret^eat aOavdroLcn ; while the ques- 
 tion of advisability as to such a reasoning is doubtful, such reason- 
 ing is only natural. For, had not the dative itself been suited to 
 express agency, it would neither have crossed over into the domain 
 of the instrumental nor monopolized this special function of agency. 
 Language — a living growth, not a system with iron-clad regula- 
 tions — abhors violence. At the same time we must not forget 
 the original I-E. development of the instrumental of means and its 
 compelling conclusions relative to the personal agent with passives ; 
 and whereas one must not and cannot go so far as to assert with 
 Fairbanks ^ that agency is a new and restricted idea with respect 
 to the dative, one is constrained to agree with him that it would 
 be in accord with the facts to extend, wherever possible, the instru- 
 mental of means in certain connections, so as to apply to persons 
 as well as to things, and not endeavor to vindicate everything in 
 sight for the dative alone. " The simplicity of the extension of the 
 impersonal use of the instrumental to the personal is greatly in 
 favor of that procedure, and it has what support is to be gained 
 from the Sanskrit parallel." ^ 
 
 1 The Dative Case in Sophocles^ vol. 17, p. 117 of TrAPA, 
 
 2 Fairbanks, ubi supra. 
 
CHAPTER VIII 
 
 The Dative-Instrumental of Agency in the Germanic 
 Languages 
 
 § 97. The Germanic languages offer almost an exact parallel to 
 Greek, not only in the details of the phenomena of syncretism, but 
 in the extreme difficulty of attaining an)rthing like definitiveness in 
 the fixation of individual usages affected by such phenomena. 
 Mourek, in a review of Delbriick's Synkretismus^ gives an excellent 
 characterization of the protean nature of the material, " wo man 
 es anfasst, entwindet es sich dem griff und verwandelt im handum- 
 drehen seine schon sicher geglaubte gestalt. man muss sich daher 
 iiberall nur mit einer der geahnten wahrheit moglichst nahe kom- 
 menden wahrscheinlichkeit begniigen." In the particular subject 
 under discussion the situation is further complicated by a well- 
 developed tendency towards analytic substitutes for the synthetic 
 expressions predominant in other branches of the Indo-European 
 family ; especially is this the case with Old High German and Old 
 Saxon, which employ exclusively prepositional phrases for the ex- 
 pression of agency with passive verbs : fona gode chisalbot, Is. 3, 2 ; 
 // uuas bi thesun liudiun giduan^ Hel. 5026, Ik geuuisso gesazt bin 
 kuning van himo, Ps. 2, 6.^ Since a detailed consideration of such 
 uses is outside the limits of our investigation, this chapter will con- 
 fine itself to Gothic, Anglo-Saxon, and Old Norse. 
 
 § 98. The character, moreover, of the Germanic dative has 
 given cause to some difference of opinion. And this entirely aside 
 from the now exploded theories of Diintzer' and Ludwig^ accord- 
 ing to which the Indo-European languages have not systematically 
 syncretized their cases so that few appear complete, but that the 
 number of case-forms was at first actually less in number and in 
 
 1 Zs.f. deut. Altertum, 38 [1908], p. 118. 
 
 2 Heyne, Kleiner e altniederdeutsche Denkmaler^ Paderborn, 1877; Graff, Die 
 ahd. Prdpositionen, Konigsberg, 1824, p. 236 ff; Behaghel, Syntax der Heliand, 
 
 1897. P- 131 ff. 
 
 ' KZ. 17 [1867]. * Agglutination oder Adaptation, 1873. 
 
 78 
 
79 
 
 distinction than in Sanskrit, and the specialization came about 
 only after the dialectal scission. To-day, of course, we no longer 
 hold that in the various aspects of the Indo-European case-system 
 we have to deal with instances of accretion. Rather do we stand, 
 cf. § 23 if., for the fusions of original case-functions and explain 
 the present position of Indo-Iranian with 8 case-forms, Balto-Slavic 
 with 7 [no abl.], Latin with 6 [no inst. and loc], Greek, Germanic, 
 and Old Irish with 5 [in general no inst. loc. and abl.], purely on 
 such a demonstrable basis of case-reductions. But recently a new 
 champion of the discredited " Ausdehnungstheorie^'' has ap- 
 peared. Winkler's Germanische Casussyntax, which deals specifi- 
 cally with the Germanic dative and instrumental, is directly opposed 
 to the so-called " Vers chmelzungs the orie,^^ as best advocated 
 by Delbriick in ALI.^ Vgl. Synt. I, and Synkretismus. This theory 
 of course explains the Germanic dative to be of syncretistic origin, 
 a Misch-kasus, arisen thru the assumption of the specific functions 
 of other lost case-forms ; a case, therefore, in which the force of the 
 I.-E. dative has coalesced, fused, merged with that respectively, of 
 the I-E. ablative, locative, and instrumental. As Delbrtick puts it 
 Vgl. Synt. 1. § 234, " Den Trager all dieser Bedeutungen nennen 
 wir Dativ." More specifically, in the plural the Germanic has lost 
 the dative and the locative, and it is an instrumental form, called 
 'dative,' which functions for all three ; whereas in the singular the 
 dative is most often expressed thru a locative, sometimes thru an 
 instrumental form, in only a few instances perhaps thru a dative 
 also. Cf. § 2>Z ff- 
 
 § 99. Far from accepting the syncretistic explanation, Winkler 
 considers the Germanic dative to be the direct and uncontaminated 
 descendant of the pure I-E. dative, essentially a case of personal 
 interest, '' casus der beteiligung ohne ortliche nebenvorstellung," 
 save that in the course of time it had, by reason of its basic signifi- 
 cance, extended the boundaries of its original sphere and had come 
 to acquire the functions already specialized and expressed by other 
 cases in the parent language. In this condition it has come down 
 to the Germanic, where we meet it in all its various applications. 
 It has, then, besides its original force, locatival, instrumental, and 
 ablatival signification, but these, save the last, are merely seemingly 
 such, inasmuch as their origin can easily be connected with the 
 real character of the dative case, interest. The pure dative in 
 Gothic, for instance, never becomes either locative or ablative, 
 
80 
 
 never performs any ablatival, p. 76 f., or locatival functions, p. 68 f. 
 and p. 142, never even approximately represents a real locative or 
 a real ablative ; wherever instances occur when the latter seems to 
 be the case, they will, on closer inspection, turn out, in the last 
 analysis, to be the real dative of interest in a special sense. Like- 
 wise it only appears to perform instrumental functions because, 
 owing to its original force, and the flexibility of its use, it is suitable 
 for such purposes. " Selbst hier ist es moglich, die vermittelung 
 mit dem reinen casus der beteiligung zu finden, wobei von wichtig- 
 keit ist, dass er nie comitativen sinn hat, sondern direct vom 
 casus des interesses zum casus der hinsicht und des mittels 
 geworden ist," p. 2, 90 if. and elsewhere. 
 
 § 100. Diagrammatically expressed, the difference between the 
 two views would be somewhat as follows ; 
 
 Delbruck Winkler 
 
 I-E. functions of Dat. Loc. Abl. Inst. I-E. functions of Dat. Loc. Abl. Inst. 
 
 ^^ 
 
 Germanic Dative Germanic 
 
 Dative 
 
 Winkler's whole trend of thought may be summarized under 
 the catchphrase of "Germanic emotionalism." According to him 
 the Germani must be thought of as having a peculiar predilec- 
 tion (besondere Vorliebe) for the case of personal interest, the 
 case of reflection, and are psychologically inclined to conceive of 
 lifeless objects as living beings interested in the moment and dura- 
 tion of an action whereby they are affected. To the lively Greeks 
 who grasped the purely actual fact, almost any verb can connect 
 itself with the accusative ; but to the inwardly directed conscious- 
 ness of the- Germani the reason for, rather than the action itself, 
 is of importance, and thus there results a tendency to represent, 
 even in expressions of vigorous action, the object acted upon as it 
 is affected, p. 26. With what consequence ? That the conception 
 of personal interest dominates the Germanic mind throughout, per- 
 sonifies the most inanimate things, renders them living partici- 
 pants in the action, enables them to feel, makes them suffer or 
 
81 
 
 react, contend or submit. The dative thus becomes a casus uni- 
 versalis in Gothic, for, besides the purely subjective relation, it 
 energetically invades the spheres of other cases and takes posses- 
 sion, wholly or partly, of the functional domains of all the other 
 oblique cases and even of prepositional relations of whatever 
 nature, — and everywhere it remains the same case of interest. 
 Indeed, here and there Winkler concedes some instrumental sense 
 to the dative and is not averse to lending an ear to the claims of 
 syncretism, but even then he conceives of a natural connection be- 
 tween the two. In the words of a somewhat sarcastic reviewer ^ 
 " wodurch man jemanden totet soil auch als das angesehen wer- 
 den, wofiir man die handlung ausfiihrtll Also man erschlagt 
 einen im interesse des beiles." 
 
 § loi. We are not inclined to concede such sweeping conquest 
 to the Germanic dative, however personal. This acquisition of 
 functions was not so much aggressive and expansive as receptive 
 and combinative, or else the morphological and syntactical data 
 of Syncretism with its evidences of passive, accidental and logical 
 absorption of one function by another case-form go for naught.'* 
 But how does all this affect the discussion at hand ? In this way : 
 If Winkler's theory be tenable, the function designated as that 
 of agency can instantly be resolved, in conformity with the postu- 
 lates of that theory, into a relation merely of personal interest. 
 For, if it be true that " was von jemand gethan wird, wird in irgend 
 einem sifine fur ihn gethan,''^ p. 80, the whole question of the I-E. 
 origin of the construction discussed would be side-tracked at once. 
 It is proposed to demonstrate in the following how far such a 
 theory is in error with respect to the so-called Dative of Agency. 
 
 Gothic. § 102. The treatment of any syntactic problem in 
 Gothic necessarily involves at least two important considerations, 
 namely 
 
 {a) That the chief, and almost exclusive, source of our knowl- 
 edge is not an original work, but only a translation into the West- 
 Gothic language of the Greek texts of the Gospels and of the 
 Pauline Epistles, as well as fragmentary renderings of portions of 
 the Old Testament, generally attributed to the West-Gothic bishop 
 Wulfila, 311-383 A.D. ;^ 
 
 1 O. Mensing, p. 552, vol. 30 of Zacher*s Zs. 
 
 2 Cf. here, in addition, O. Mensing's review of SynkretismuSy I FA. vol. 22, p. 47, 
 8 For syntactic purposes the Skeireins and the documents of Arezzo and 
 
82 
 
 {b) That, as a result, for Gothic we do not, as in the case of 
 other Germanic dialects, possess the materials for tracing the his- 
 torical development of the language, a knowledge of which alone 
 makes possible conclusive decisions concerning syntactic difficul- 
 ties. What we do have of the Gothic is — aside from the so- 
 called Crimean Gothic of the i6th century — a completely isolated 
 literary monument, a mere snapshot, as it were, taken of the lan- 
 guage in the middle of the 4th century a.d. ; besides this, only a 
 few insignificant inscriptions remain of the language of the once 
 powerful Visi- and Ostrogoths. 
 
 § 103. This lack of material is all the more to be deplored in 
 view of the importance of determining Wulfila's attitude towards 
 his sources. Whether he slavishly imitated the style and construc- 
 tions of the Greek original^ or preserved that measure of inde- 
 pendence indispensable for an idiomatic rendering is a disputed 
 question.^ A careful consideration of all the evidence pro and con 
 
 Naples, etc., are not generally recognized as authoritative, unless corroborated by 
 other evidence; cf. Winkler, Germ. Cas. p. 136, and the introductory remarks to 
 Streitberg's Got. El.^ ; but see also Lenk, Die Syntax der Skeireins, PBB. 36, 
 237 ff. 
 
 1 Streitberg, Got. EL^ p. 30, indicates the strong probability also of the influ- 
 ence of the Latin Itala upon the translation, " am schwachsten ist sie in Mt. , 
 starker in den ubrigen Evangelien, am weitesten aber in den paulinischen 
 Briefen." So Bernhardt, Vulfila, Einleitung, xxxviii, and Bangert, Einjluss der 
 lat. Quellen auf die got. Bibeliibersetzung, Prog. Rudolstadt, 1880. 
 
 2 In a general way, the striking similarity between original and translation 
 cannot be denied. Wulfila evidently was not inclined to depart radically from 
 the Greek text ; his close adherence to it at times, resulting even in such ab- 
 surdities as imitations of anacolutha, certainly minimizes the number of cases of 
 difference, and would tend to reduce their significance when compared with the 
 multitude of instances in which his version agrees absolutely with the original. 
 On the other hand, whole series of grammatical differences and syntactic devia- 
 tions have been adduced to prove Wulfila's comparative independer.ee, not only 
 in these respects, but also in his peculiarly Germanic interpretation of Bibhcal 
 passages. Thus he is unquestionably unrestricted in the application of his verbal 
 forms and has a tendency to a peculiar use of modes and of the dual to attain a 
 linguistic finesse not possessed by the Greek text. His ready inventiveness is, 
 furthermore, remarkably well illustrated by the fact that he often turns the very 
 forms and idioms which he sometimes studiously avoids and, again, uses almost 
 indifferently, against the usage of the original text itself. In a word, it is a case, 
 modified of course, of * Quandoque bonus dormitat Homerus* Cf. Gaebeler, Die 
 griech. bestandteile d, got. Bibel, Zs. f. d. Phil. 43, i ff. ; Stolzenburg, Zur Uber- 
 setzungstechnik des Wulfila^ Diss. Halle, 1905 ; G-L's Prolegomena, etc. 
 
justifies the conclusion that the translator, anxious to remain faith- 
 ful to his source, either because of his reverence for it, or for rea- 
 sons of literary expediency,^ undoubtedly went to the farthest limit 
 of what was admissible in his own language ; but that, if he was 
 not to defeat the purpose of his work, even if he did not contem- 
 plate a popularization of the Bible, as Friedrichs claims,^ he had 
 to avoid doing violence to the genius of the Gothic tongue. Simi- 
 larly Bernhardt,' " Der gotische gebrauch der casus erweist sich 
 durch die vergleichung des Ahd. Altn. Ags. Alts, im ganzen als 
 echt germanisch." Therefore the occurrence in Wulfila's Bible of 
 constructions that might be suspected of being Grecisms, aside from 
 probable and rare instances of unconscious imitation, can be ex- 
 plained on no other assumption save that, in such cases, the 
 Gothic could follow the Greek without at all becoming unidiomatic. 
 This was quite possible if, as we have reason to believe, the two 
 related Indo-European languages, at that period, stood both on 
 nearly the same syntactic niveau, having in common the loss of 
 the ablative, locative and the instrumental.'* 
 
 § 104. It is precisely that intimate interplay caused by these 
 conditions, the almost systematic identification of Gothic with 
 Greek, that renders it difficult to determine with absolute exacti- 
 tude the character of certain syntactic constructions found in Wul- 
 fila, especially if the limited literature at our disposal offer but 
 few examples of their actual application. Such is the case with 
 the so-called Dative of Agency without preposition. As a rule, agency 
 combined with a passive verb in Gothic is expressed by the prepo- 
 sition * fram ' ^ connected with the dative of the person, e.g. Mt. 6, 
 2 ei hauhjaindau fram mannam, ottods ho^aaOoicnv virh twv dv 
 ^pwTTwv, Ags. />£et hi sin gearwurff ode fram mannum^ or Lk. 5, 
 15 garunnun hausjan jah leikinon fram imma, kol BepatreveuOcu 
 There are found, however, well-defined instances where, 
 
 ^ Cf. Kauffmann, Zs. f. d. Phil. 32, 316, and Dietrich, Die bruchstucke der 
 Skeireins, 2 v. Strassburg, 1903, p. Ixxvi. 
 
 2 Die Stellung des Pron. personate im Got. Jena, 1891, p. 3 fF. 
 
 8 Zs.f. d. Phil. 13, I. 
 
 * Cf. also Curme, Is the Gothic Bible Gothic? JEGP. 10, 156. 
 
 ^ Grimm's * ideal preposition ' with passives, Grammatik^ IV. 947. 
 
 6 The Belles-Lettres series : The Gospel of St. Matthew, J. W. Bright, 1904. 
 
 ' Other examples in Gering, Zs.f. d. Phil. 5, 411, and Winkler, Germ, Cos, 
 p. 157- 
 
84 
 
 instead of this prepositional construction, a pure dative of agency 
 seems to be used. 
 
 § 105. Before we undertake a systematic examination of them, 
 the opinions of grammarians will be of interest. Of those who to 
 a greater or lesser degree have touched upon the subject, Gabe- 
 lentz u. Loebe ^ attempt a fair solution of the problem, but are led 
 astray partly by a hypothetical intransitiveness with which such 
 datives are connected, partly by an inclination to treat the con- 
 struction as a mere imitation of the Greek dative used with passives 
 for VTTO c. gen. At any rate they offer no discussion of examples. 
 For the former of these views they seem to have followed in the 
 footsteps of Jac. Grimm, who in his few words on gaumjan ^ sug- 
 gests ' appareant, videantur hominibus ' as a meaning for Mt. 6, 5 
 gaumjaindau mannam. Kohler,^ deploring the lack of material 
 that might furnish evidence, reluctantly assigns the datives with 
 the passives of gaumjan and saihan to the category of the farther 
 object. Erdmann-Mensing do not mention the subject at all ; 
 nor does Streitberg who must be understood as taking the construc- 
 tion for a Greek imitation, since his syntax is based only upon the 
 differences between Gothic and Greek.^ Bernhardt^ and Wil- 
 manns® content themselves with adopting Winkler's view. Lastly, 
 van der Meer^ thus sums up the situation, " Ofschoon hier 00k in 
 de meeste gevallen het Grieksch een datief heeft, laat zich in ver- 
 band met de bovengenoemde gevallen deze constructie ok van 
 Gotisch standpunt verklaren, a Is we den datief opvatten als een 
 dativus commodi et incommodi." Winkler's view, as might be 
 expected, is characteristic,* " Noch einen fall nahen zusammen- 
 hanges bietet der ausdruck des agens beim passiv, wo wir gewohnt 
 sind das ausgehen der handlung vom agens betont zu sehen, gr. 
 VTTO, lat. ab, got. f ram : vielfach begegnet uns auch der reine instru- 
 mental, daneben der dativ sporadisch in den verschiedensten 
 sprachen, so ziemlich oft in Avesta, hier und da im gr. und latein. 
 und desgleichen im germanischen. JDer dativ ist hier Uberall rein- 
 ster ausdruck der beteiligung.''^ No writer, so far as I am aware, 
 has made a systematic effort to connect this Gothic dative, from 
 
 1 G-L. p. 226, § 231, 2. 2 j)gj^i^ Grammatiky, IV. 699. 
 
 ^ Germania, xi. 237. * Cf. Got. El.^ Vorwort. 
 
 5 Krz. Got. Gr. p. 84. « Dg^t. Grammatik, III. 2, § 289. 
 
 ■^ Got. Casussyntaxisy I. Leiden, 1901, § 38. 
 
 8 Germ. Cas. p. 80. 
 
85 
 
 the syncretistic development of this case, with the instrumental of 
 agency known to other languages. 
 
 § 1 06. As in the instance of Greek, so here the very survival 
 and supremacy of the so-called ' dative ' contributes much to give 
 everything it represents a strong datival coloring. In the discus- 
 sion following, therefore, the cooperation of the dative will be taken 
 for granted, with the important proviso, however, that modern Ger- 
 man Sprachgefuhl must not be too heavily drawn upon for the 
 determination of such an old stage as Gothic represents. In any 
 case of doubt we must hold as authoritative rather those dialects 
 which have the most important bearing upon Gothic syntax, viz. 
 Anglo-Saxon and Old Norse,^ as well as the usage of the older 
 periods of other cognate languages. 
 
 § 107. ei gaumjaindau man nam, Mt. 6, 5, oirin^ av <jxLvS)(nv rots 
 avOptoTTOLs.^ Two considerations present themselves here. The verb 
 is by many regarded as used intransitively, tho of passive form, 
 and this seems to be borne out by the Greek </>av(u<nv. Now, 
 altho the results of the action are the same whether the verb be 
 regarded as intransitive or as passive, the direction of the action in 
 the one case is essentially different from that in the other : in the 
 one a merely subjective status is mirrored, in the other an objective 
 action towards the subject is emphasized. The intransitive usage, 
 then, cannot denote that shade of meaning which is inherent in the 
 passive verb ; and given, in any language, the possibility of both 
 variants, there is no reason why the one expression should be 
 taken as the exact and necessary equivalent of the other. The 
 possibility of such a conception cannot be denied, but its necessity 
 has yet to be shown. The evidence supposed to be furnished by 
 the Greek text is of no weight whatever in the light of the fact that, 
 to be consistent in their interpretations of the dative, commentators 
 consider obviously passive verbs in the Greek text itself as being 
 used intransitively. As to the dative mannam, observe that gaum- 
 jan is one of the verbs in Gothic which, aside from instrumental 
 
 1 Bernhardt, Beitrage z. d. Phil., Halle, 1880, p. 73. 
 
 2 These citations are from Streitberg's Die Got. Bibel ; the Gothic text, there- 
 fore, is substantially that of the Codex Argenteus of the 5/6th century, whereas 
 the accompanying Greek is that of the recension current at Constantinople in the 
 4th century, as reconstructed by scholars like Lagarde, Kauffman and others. 
 Pp. 23-33 of Goi' ^^-^ and, more extensively, S. xxxi-xlvi of Got. Bibel offer 
 detailed information on this question. 
 
connections, as Jh. 12, 40, ei ni gaumidedeina augam, tva firj tBuxri 
 Tots 6<f>daX/x6L^, take the dative-object instead of the accusative, e.g, 
 Jh. 9, I gaumida mann blindamma, Ihtv avOpwTrov. Inasmuch 
 as the dative appearing with it bears the same relation to the verb as 
 the accusative-object to tfs verb, with the passivization of the verb 
 the dative-object necessarily becomes the subject of it^ — and this, 
 indeed, is the usage of the Gothic, the formation of personal 
 passives by verbs governing a dative-object.^ Thus the construc- 
 tion ei gaumjaindau mannam presupposes an active ei mans gaum- 
 jaina im. There can be no talk, then, of a dative-object retained, 
 as in Jh. 17, 15 or Mt 9, 17 ; a retained dat. of indirect object, as 
 l>atei giban ist mis, tyjv Sodcta-dv /aoi, Col. i, 25, is impossible; a 
 pure dative of interest is excluded by our view of the passives. 
 As a practical proof of the soundness of this reasoning let me offer 
 Mk. 16, II jak gasaiJvans warp f ram izai, koI idedOrj vir avr^s,' 
 I^a/a * et visus esset ab ea,' which shows, despite Winkler, p. 35 note 
 to gaumjan, that the passive of a verb of seeing must no more be 
 intransitive in Gothic than in Greek ; that the dative connected 
 with it must not be a dat. of interest; that, finally, the idea of 
 agency, as such, may and can be expressed with it. 
 
 § 108. Anglo-Saxon pcet men hig geseon offers no aid here.'* 
 Piper ^ assigns mannam to his table of dat.-instrumentals, Erdmann® 
 strongly dissenting. Kohler ^ remarks that if gaumjan is taken in 
 its real passive meaning, 'beobachtet, bemerkt werden,' then this 
 is a case of the dat. with passives '■'■ wie wir im Lateinischen gar 
 nicht selten beim Passivum . . . finden." Mt. 6, 16 and 18, ei 
 gasailuaindau mannam fastandans, ottws (^avwo-iv rot? dvdpwTroi^ 
 vrja-Tevovre^ and ei ni gasailuaizau mannam fastands, ottws /w,^ <^avijs 
 
 1 For similar use in Gk. cf. Kriiger, ^/A Syn^. p. 137. 
 
 2 Cf. Go^, EIJ^ §§ 241, 255, 286 ; van der Meer, p. 3; esp. Kohler, Germania, 
 xi, 285, 287. 
 
 3 For Gk. cf. also Xen. Cyr. 3, 3, 31 and Oec. 8, 11. 
 
 * In accordance with the opinions of Bernhardt, Beitrage z. d. Phil. Halle, 
 1880, p. 73, the Anglo-Saxon version of the Bible is constantly drawn upon for 
 collateral testimony. Whilst the Gothic and Anglo-Saxon translations were not 
 made from the same text, the prototype of the latter being the Vulgate, the evi- 
 dence of the Anglo-Saxon is nevertheless important, as an index of Germanic 
 " Gesinnungsweise" in clearing up the Germanic character of ambiguous con- 
 structions in Gothic. 
 
 6 ijber den Gebrauch des Dativs, p. 29. « Zs. f. d. Phil. 6, 1 21. 
 
 ^ Germania^ xi, 287. 
 
87 
 
 Tots dv9p<jiTrois vrj(TT€v(Dv, may after the preceding lines be passed by, 
 save that Ags. lends added weight to their interpretation in the 
 above light: whereas in \(} pcBt hig cefeowun mannum fcestende is 
 a real dative of interest because of its verb, * to appear,' iB>,AcBt/>u ne 
 sy gesewen fram mannum fcestende has the idea of agency in its 
 most distinct form of expression with a verb of active seeing. 
 
 § 109. Du sailuan im, Mt. 6, i, Trpos to OeaOrjvai avrots. Bopp, in 
 an interesting passage ^ intimates that t'm is not an ordinary dative. 
 " Ausserdem kann oft im Gothischen nur aus dem Zusammenhang 
 und durch den dabeistehenden Dativ [allein oder mit /ram, von], 
 welcher im Gothischen hdufig die Stelle des skr. Instrumentalis vertritt 
 — italics mine — erkannt werden, das der Infinitiv nicht die 
 gewohnliche active, sondern passive Bedeutung hat. So erhellt 
 Matth. 6, I, aus dem Dativ im von ihnen, dass der vorangehende 
 Infinitiv passive Bedeutung hat, und du saihan im, welches wir, um 
 die Construction nachzuahmen, durch * zum Sehen von ihnen ' 
 ubersetzen miissten, ubertragt das Griechische Trpos to 6ea6rjvai 
 avTois." Grimm (IV. 104 f.) evidently under the spell of NHG. — 
 Tatian in his paraphrastic thaz ir gisehan sit avoids the dative alto- 
 gether — expresses his astonishment that the Gothic infinitive 
 should not only be used in a passive sense, but even connected with 
 an oblique case thru which that passive is defined. 
 
 § no. In spite of Bopp, I am inclined to vindicate for im an 
 original datival use, and that because the form with which it is 
 found seems to bring it in line with a general Indo-European phe- 
 nomenon, the use of the infinitive in a final sense. Not, indeed, 
 according to Jolly ,^ who, citing Benf ey's rule ' as to the attraction of 
 the dative-infinitive, would see a parallel between vrtraya hantave, 
 * Vritrae occidendo, ad Vritram occidendam,' or drse vi'svaya, ' omni 
 videndo, ad omne videndum,' and du saiJvan im} There would, 
 indeed, be a similarity if du saihan im meant * ad eos videndos * 
 or, in Benf ey's scheme of attraction, '■ eis videndis.' — The final 
 sense of infinitives, originally case-forms of verbal abstracts, 
 nomina actionis,^ undoubtedly based on an I-E. usage, is common 
 to a number of languages. So Sanskrit, with a meaning of pas- 
 sivity similar to the Gothic, RV. 7, 31, i a vo vahistho vahatu 
 
 1 Vgl. Gramm. III. § 872, p. 305. 
 
 2 Gesch. des InfinitivSy p. 163, tho cf. also p. 265. 
 
 8 Krz. Sk. Gr. p. 237. * Cf. also KZ. 18, 104. 
 
 ^ Cf. Bopp's Konjugationssysteniy 1816, p. 71. 
 
stavddhyai rdthah, * in order to be praised, for praise ' ; RV. i, 123, 
 II avis tanvdm krnufe drse kdm, 'so that it may be seen,' is the 
 exact counterpart of du saiJvan. These examples are taken from 
 Vgl. Synt II. 464 and S.F. V. 410 iff. For the A vest an, cf. Rei- 
 chelt, § 702 tat nidi vicidydi vaoca viduye mpncd daidyai. Similar 
 infinitives of purpose are found in Homer, 7176,11373; passively 
 5 507 Kfxro 8* ap iv fiea-aoiaL 8voi ;(pvoroto ToAavra, tw Sofiev, * in order 
 to be given to him,' os etc.* For Bal to-Slavic, cf. Miller, Ueder 
 den letto-slavischen Infinitiv, Kuhn &* Schleicher 8, 165. The 
 agent with these infinitives is in the dative; it is not 
 so much an agent as a dat. commodi. RV. 9, 4, \ puni- 
 tdna somam in dray a patave, 'for Indra to drink,' 'to be drunk 
 by Indra,' as we would say to-day; 10, 14, 12 drsdye suryaya, 
 *for a view to the sun, to be seen by the sun.' So i, 113, 5. — 
 Avestan, '' uksne krathwe, zur Vermehrung fiir den Verstand," 
 Wilhelm, Infin. p. 96. Cf. above moi, also Y. 45. 5 sruidyai ma- 
 rptaeibyo, 'for men to hear.' — Greek, x 47^ /xiySea t i^epvaav, 
 Kvcrlv w/xa BdcracrOaL, A 76 Koi k<T(Toix.€voi(n TrvdiarOai. Slavic, sila 
 christosova jest-h videti, ' in order to be seen.' ^ Lithuanian: ^J2 
 girdejau, kdd til tun dtikteri smakui praryt, ' a daughter to be swal- 
 lowed by a dragon, zum Verschlingen fiir einen Drachen." 
 
 § III. The Germanic infinitive in such instances, consequently, 
 may be taken as one of these final infinitives. Since the preposi- 
 tional infinitive bears the same relation to the simple infinitive — 
 itself but a nomen absolutum — which the prepositional cases bear 
 to simple cases, the preposition du serves only as a preservative 
 of the original, worn, and tarnished basic signification.^ In other 
 words, im parallels the datives in such noun-connections expressive 
 of purpose as Mt. ^, ^ du weitwodipai im, eis frnprvpiov avrols, ' for 
 
 1 Cf Gr. GrJ^ § 457. 
 
 2 Miklosich, IV. 849 ; Vondrak, II. 366. 
 8 Cf. also Schleicher, p. 311. 
 
 * The W. Germanic has an inflected infinitive [gerundium], an expansion of 
 the orig. suffix with -w-^ preceded by a preposition, Ags. to^ OS. to, te, OHG. zi 
 followed by a dative, to dlosannea, to bindanne, zi bintanne (Dieter, Altgenn. 
 Dial. p. 379; Wilmanns, 3, i, p. 127). It is doubtful whether a dative form 
 like this existed in Gothic, because Lk. 14, 31 du viganna is a conjectural form, 
 cf. Streitberg, Got. Bib. II. 174. On this see Winkler, Germ. Gas. p. 289, and 
 Jolly, p. 176, who hold that the Germanic infinitive was originally a dative, tho 
 Streitberg, U.G. § 198: I-E. -{p)no-m > Qxm. {a)n, as I-E. *bheronom, Sk. 
 bhdranam. Gt. bairan. therefore a neuter nominative- accusative. 
 
a testimony to them'; R. lo, 4. du garaihtein a II aim paint ga~ 
 laubj andain^ eis hiKaxo(Jvvt\v iravrX tw TrttrTevovrt. On the Other 
 hand, compare 2 Cor. i, 20, where agency is denoted, pure and 
 simple, gupa du wulpau pairh uns, t<5 ^ew Trpos ho^av hC ^/mwv. 
 So, ^« saijvan im really means '/^r « seeing, a view to them.'* 
 The passive meaning attached to it to-day is artificial and based 
 on analogy with the modern German and perhaps with Latin, 
 a procedure warned against in principle as early as Grimm, 
 IV. 66. 
 
 § 1 12. Anglo-Saxon pcet ge sin geherede fram him does not corre- 
 spond at all to the Greek, — it means ' laudare, celebrare,' as in 
 Ps. Th. 43, 10 — and for Mt. 23, 15 tt/jos to OeaOrjvaL TOts dvOpfOTroLs, 
 which is not in our Gothic text, we have in Ags. pcet men hi geseon. 
 The Gothic i Tim. 4, 3, fianzei gup gaskop du andniman mip awili- 
 udam galaubjandam jah ufkunnandam sunja^ a 6 Oeos Ikti- 
 <T€v eU ixcToX-qij/Lv . . . Tois TTLo-TOLs kt\., Teprcscnts to my mind the 
 same construction as du saihan im, 
 
 § 113. I Cor. 15, 5 ataugids ist Kefin jah afar J?ata ]'aim ain- 
 libim, ui^Or\ Kr)(f>a etra rots StoScKa ; and I Tim. 3, 16, cf. also Mk. 
 9, 4, ataugids warp paint aggilum, w^O-q dyyeAots — admit of an 
 intransitive interpretation in the first instance, and Mk. 9, 4 of the 
 Ags. pa cetywde him He lias mid Moyse substantially covers jah 
 ataugips warp im Helias mip Mose, to parallel io^O-q avrot?. At 
 the same time some considerations are at hand towards a contrary 
 explanation. The Gothic presents the curious complication of em- 
 ploying both the passive of ataugjan — 'be shown, appear, be 
 seen ' — and the reflexive ataugjan sik to render the Greek opdofmL. 
 Thus, Mk. 9, 4 ataugips warp im H. w<f>97j avrot?, [also i Tim. 
 3, 16 and I Cor. 15, 5, as against i Cor. 15, 7 ataugida sik 
 Jakobau, w<l>9r] 'laKw/Sw; so i Cor. 15, 8. To the latter must be 
 added Mk. 27, 53 ataugidedun sik managaim, If^avLaOyja-av iroX- 
 \o2s, and, because of its meaning, Mk. 16, 9, ataugida Marjin, 
 icfjavrj. It is to be noted that the expression w<f>6ri with the dative 
 recurs four times in close succession, but that the Gothic does not 
 render it alike in all instances ; in the first two Wulfila translates 
 by the periphrastic passive, in the last two, however, by a reflexive 
 construction ; probably this variation is simply due to a desire to 
 vary the style. Kohler's statement, " Der got. Ubersetzer scheint 
 gefiihlt zu haben, dass er seiner sprache etwas ihr eigentlich Frem- 
 des zumuthet, wenn er den Dativ beim Passiv setzt, und das Re- 
 
90 
 
 flexivum vorgezogen,"^ is gratuitous in view of the Indo-European 
 bearings of this Gothic dative with passives. Ataugjan sik is too 
 palpably and correctly used in its literal meaning ' to show one's 
 self ' either to necessitate an intransitive ' to appear ' or to be 
 equated with a passive ' to be seen.' This much cannot be said of 
 I Cor. 15, 5 or I Tim. 3, 16. It might be that Wulfila did try to 
 avoid a similar Greek construction, but the inference need not be 
 that a suitable substitute is an alternate of exact correspondence 
 as well. That desire for variety played a large part in Wulfila 's 
 choice of words is evident from the same sentence-unit where an- 
 other ^Bk] in the series is rendered gasaihans ist managizam^ 
 i-rraLVio, 1 Cor. 15, 6. Furthermore, the impression gathered from 
 I Tim. 3, 16, where we have another series of Greek passives, is 
 that, alive to the finer requirements of stylistic technic, Wulfila 
 varied his verbs accordingly, but that, despite Kohler's view above, 
 with the passive of ataugjan he really wished and could render the 
 Greek passive with a dative-instrumental of agency. In mere in- 
 transitive turns, he resorts indifferently to sik ataugjan or in siunai 
 wairpan to render ffxxvrjvai or kavTov Sei^at. 
 
 § 114. Mt. 5, 21 and 33 )7atei qij7an ist Jjaim airizaim, on 
 ippWrj ro2<s dpxaLOL<i. Our modern English version, ' ye have heard it 
 was said by them of old time,' encourages the view of agency, such 
 as appears in Lk. 2, 21 -with /ram. Neither the Greek original 
 nor the Anglo-Saxon version based on the Latin Vulgate can help 
 us out. So the Anglo-Saxon has ^ Ge gehyrdon pcBt gecweden wees 
 on ealdum tidum, cwiden, but the Northumbrian glosses the Latin 
 * dictum est antiquis ' with acueden is to deem haaldum, whereas 
 the Mercian reads, /^cette cwceden wees p(2m iu-monnum. The in- 
 tention of the Church Fathers was to take it 'to the ancients,' 
 Chrysostomos, Horn. xvi. ; accordingly Meyer^, Kommentar, I. i, 
 1858, p. 144 is correct in paralleling ippiOrj to2s dpx^'tbis with 
 Xeyo) 8c v/juv in verse 22. So, too, Tholuck, loannis Calvini in 
 N.T. Comm} 1838, I. 147. Luther, " dass zu den Alten ge- 
 sagt ist." 
 
 § 115. In Rom. 7, 10 bigitana war]? mis anabusns, /cat evpidr} fxoi 
 17 ivToXr}, and 2 Cor. 12, 20, jak ik bigitaidau izwis swaleiks swe ni 
 wileip mik, <{>o/3ovfuii /ai^ttws . . . xdyw evpWo) v/juv olov ov Oikere, cf. 
 
 ^ Germania, XII. 64. 
 
 2 From The Holy Gospels in Anglo-Saxon, Northumbrian, and Old Mercian 
 Versions, synoptically arranged . . . iy W.W. Skeat, Camb. 1871-87. 
 
91 
 
 also Rom. lo, 20, we have to deal with two instances of the dative- 
 agent developed from an original dat. commodi. Both Gk. evpCa-Ko- 
 fjMi and Germanic *bigeian'^ admit of a dative of interest in the 
 active, hence \uoi and v)u,tv, mis and izwis might well be retained 
 datives in a developed sense. To my mind G-L/s translation ' et 
 inventum fuit mihi praeceptum ' savors too much of an adjectival 
 sense of the participle. That in Gothic the participia praeteriti 
 must be considered as adjectives, especially when connected with 
 the substantive verb,^ I do not beUeve, because all past forms of 
 the Greek passive, aside 'from their rendition by means of Gothic 
 actives, must be expressed by periphrasis with wisan and wairpan? 
 Consequently these forms are finite passive verbs first of all.^ 
 
 § 116. Mk. II, 17 J7ata razn mein razn bido haitada a Halm 
 Jjiudom, on o oikos /xov oikos Trpoa-evxrj^ KXyjOi^creTaL TraaLv rot? eOveaLv, 
 is one evidence of the errors practised in the interpretation of 
 transitive verbs. Even if one allows that active transitive verbs of 
 seeing in passive form must necessarily give the intransitive sense 
 'to appear,' a verb of naming and calling can remain nothing else 
 but naming and calling even when passivized. Nothwithstanding, 
 G-L. suggest "es sollte a//en Voelkern als ein Bethaus gelten." 
 Ags. fecet mm hus fram eallum P'eodum bid genemned gebedhus^ 
 rendering the Vulgate ' vocabitur omnibus gentibus,' clearly shows 
 instrumental agency. Again, it may be that Gothic, as apart from 
 the evidence of Ags., merely followed the prototype, in which case, 
 of course, we have to do with a dative-agent as in § 115. For both 
 cf. § 90 a, b} Piper, op. cit. 29, classes piudom among the instru- 
 mental datives. The OHG. ther uuas uns io giheizan, as often in 
 Otfrid, means, of course, ' promised to us.' 
 
 § 117. Rom. 14, 18 walla galeikaij? guda jah gakusans ist 
 man nam, evapeo-ro? tw ^ecu Kttt SoKt/xos Tots dvOp(i)7roL<s. The Greek 
 seems to have influenced the Gothic. But, then, guda is governed 
 by ga/eikan, as ' placet domino ' ; gakiusan, on the other hand, is a 
 transitive verb, SoKt/Act^ctv, hence mannam cannot be accounted for 
 except on the ground of being a dative of agency, ' probatus est 
 hominibus.' That it is not a dat.-inst. of agent, in spite of rather 
 late Kgs. gecoren wcerefram gode, Blickl. Hom. 187, 20 — Caedmon 
 
 1 Cf. Synkretismus, p. 38. 2 Grimm, IV. 717. 
 
 8 Cf. Got. El.^ § 285, 3. * Cf. Kohler, Germania, xi. 287. 
 
 6 The Hebrew text of Is. 56, 7, n^m:fh'^ K-il^": n^BITn^a "'rT'3 is identical 
 in sentiment with the Greek ; cf. also Tob. 3, 11 and i Maccab. 7, 37. 
 
has Dan. 92 metode gecorene — is shown by the consistent Gothic 
 rendering of dSoKt/Ao? by ungakusans. 
 
 § 118. Mk. 10, 12 jahliugadaanparamma, koL yafx-qOfj a\\(om2iy, 
 according to §§ 28 and 36, (a), he taken as a real dat.-instrumental 
 of agency. So, also, Piper, op. cit. 29 ; even Winkler, p. 84, " konnte 
 auch bei den passiwerben mit dem dativ der handelnden person 
 genannt werden." 
 
 § 119. Jh. 6, 13 Jjatei aflifnoda J^aimmatjandam, 5 cireptWcvorav 
 Tots pe^poyKoa-Lv and Lk. 9, 17 /^/(?/ aflifnoda im gabruko. The 
 intent of the Greek passages is given by Alford, The Greek Test. I., 
 ad loc, as that they collected the fragments for their own use, each 
 in his Ko<^tvos.^ So Meyer*, Kommentar, Johannes, 1862, p. 217, 
 " und jeder Apostel fiillt mit dem Uberfluss seinen Reisekorb." At 
 the same time Wulfila may have had a different interpretation of the 
 matter and, if not for the ambiguous character of the intransitive- 
 inchoative verbs,^ van der Meer's "wat overgelaaten werd door 
 hen " could be justified by an example to be found in Skeir. 7,27 
 which is another translation of Jh. 6, 13, but reads />atei aflifnoda 
 at paim ; matfandam is lost in the fragment. That this at is the 
 preposition found with verbs of taking, receiving, and finding in 
 contradistinction to the Greek which in such cases emphasizes the 
 * direction from ' by dTro, irapd, etc., is supported by i Thess. 4, 9, 
 unte silbans jus at guda uslaisidai sijufoj avrol yap v/Acts deoBCBaKTOL 
 ia-re, * von Gott belehrt ' ; i Tim. 6, 5 at f^aimei gatarnip ist sunfa^ 
 Kttt airea-TeprjiJLiviov r^s aX.r]6eui<Sf ' by whom truth is perverted ' ; Lk. 
 10, 7 driggkandans f>o at im, sc. gibanona, TrtVovres to. irap avrdv. 
 Consequently the possibility of Jh. 6, 13 and Lk. 9, 17 being each 
 a dative-instrumental of agency is by no means excluded. 
 
 § 120. In contradiction to van der Meer, p. 87, in Mk. 2, 26 
 J^anzei ni skuld ist matjan niba ainaimgudjam, and in Lk. 6, 4, 
 the dative should be taken with skuld ist, a dat. commodi, as in 
 Mk. 6, 18 ni skuld ist fius haban? The Ags. pe him ne alyfede 
 n^ron to etanne, buton sacerdon anum supports this view. There- 
 fore ' quos non licet manducare sacerdotibus.' For Lk. 9, 14 cf. 
 Kohler, Germania, XII. 450 f. This exhausts the fist of the Gothic 
 examples available for discussion. 
 
 1 Cf. Juv. Sat. 3, 14 quorum cophinus foenumque supellex. 
 
 2 Cf. G-L. p. 138 and Streitberg, Got. El.^ p. 188. 
 
 ^ Cf. Winkler, Germ. Gas. p. 24; Gering, Zs.f. d. Phil. 5, 420 f. 
 
93 
 
 §121. To sum up : All together there are not many instances in 
 Gothic of passive constructions with the subject designated ; those 
 with a prepositionless dative are very few. Of the verbs neverthe- 
 less found thus connected the majority are those of seeing, or find- 
 ing, or calling, concepts, in a word, which could suggest an a priori 
 explanation of these datives as datives of personal interest.^ In 
 the instance of the verbs of finding such an interpretation was 
 indeed given on general I-E. grounds as well ; in that of a verb of 
 calling the intent of the Church Fathers was invoked to confirm a 
 similar conclusion. In the discussion of du saiJvan im the testi- 
 mony of the Indo-Iranian, Balto-Slavic, and Greek traced the dative 
 back to a common I-E. basis. Still, in the course of this investiga- 
 tion, each example being considered by itself, enough has been 
 shown to prove that Gothic possesses both datives of agency that 
 go back to original datives and such as present a variety distinct 
 and separate from the usual run of dativi commodi and which, in 
 fact, are traceable to an I-E. instrumental. 
 
 § 122. That this is possible has been demonstrated, it is hoped, 
 beyond a shadow of doubt in connection with the similar develop- 
 ments of other Indo-European languages and the establishment of 
 the facts of Syncretism. I have, in accordance with this Indo- 
 European development, constructed a similar scheme, deriving 
 those of the Gothic dative-agents that may not be classified with 
 the idea of personal interest, from an original sociative instrumental, 
 thru the media of material means and personified agents.^ 
 
 {a) There is not a single example in Gothic of what we have 
 called the free comitative, since, altho in Pre-Germanic it surely 
 must have been in use with verbs of motion like ^ijjdm^ it had 
 gradually come to be displaced by the competing preposition mif>. 
 Then, as van der Meer, p. loi, aptly remarks, " dergelijke geval- 
 len komen ook niet voor in den Griekschen grondtekst." The 
 Gothic thus cannot duplicate Ags. folce gestepte, ' he proceeded with 
 an army.' This is Winkler's reason, the lack of comitatives and of 
 distinct instrumental forms in Gothic as against Ags., for deriving 
 the Got. instrumental function from that all-embracing case, the 
 
 1 Winkler, op. cit. p. 8i. 
 
 2 Delbriick in Vgl. Synt. I. § 123 totally disregards the possibility of such a 
 gradation in the Germanic field; in Synkretismus^ § 23, however, he seems to 
 favor it. 
 
 8 Cf. Synkretismus, § 19. 
 
94 
 
 dative, whereas in the instance of Ags. he allows for it a comitative 
 descent ; Germ. Cas. p. 82, 90, etc. 
 
 (b) Numerous examples are found of the material instrumental/ 
 e.g. 
 
 Jh. 19, 2 wasijai paurpurodai gawasidedun ina, attested by the 
 pronominal form in Mt. 6, 25 he wasjai/f, tl iv8va-qa-9e. With a 
 passive verb, Lk. 8, 28 bundans was eisarnabandjom, iSea-fxelTo d\v- 
 a-ea-iv ; cf. § 7 1^. The Document of Naples contains ufmelida handau 
 meinai four times in a formulistic way. Gaumjan appears but once 
 with the instrumental, Jh. 12, 40 augam; the assumption of a similar 
 inst. of means with saihan is justified by ON. ef ek hann sjonom 
 ofsek, Hgv. 151, and Ags. eagum segun, Cri. 536. 
 
 (/) The means employed, however, could represent not only a 
 material object but also a spiritual something, an event, a circum- 
 stance, an abstraction. Thus, the common haitan namin, ovofmn, 
 as Ags. sume worde hef, OHG. thaz hiazi er to then worton. So 
 daupau afdaupjaidau, Mk, 7, 10 ^avaro) TeXevraro) ; cf. also Lk. 7, 
 29 and Mk. 10, 38. 
 
 (d) In Mk. 10, 38, daupeinai f>haiei ik daupjada, the dat.-inst. 
 is seen with a passive verb, and, in fact, denotes both the means 
 and the agent, Brugmann's Potenz {Grdr? II. 527) of the action. 
 The next and penultimate stage in this evolution of means to 
 agency would be instances like 2 Tim. 3, 6 f>oei tiuhanda lustum, 
 dyofxeva iTriOvixuiL^ TTotKt'Axtts ; Gal. 5, 18 jabai ahmin tiuhanda^ ci 
 irvev/xart ayeo-^e ; Eph. 4, 14 uswagidai jah usflaugidai winda Jvam- 
 meh, ttovtI d/x-c/Ao); Eph. i, 13 gasiglidai waurfiu/^ ahmin gahatfis, 
 Tw TTveofwTi T^s cTTayyeAuxs ; Lk. 4, 38 was anahabaida brinnon miki- 
 lai, rjv (Tvvexpixevrj irvpcTw ixeydX(o, etc. etc., where Gothic and Greek 
 alike leave no doubt as to the agent of the action, which is no 
 more tangible than abstract human agency, but is as much felt in 
 its results as any ordinary instrument or tool of action. In fact, 
 to have an instrumental of agency in Gothic — as apart from the 
 dative-agent < dat. commodi — would be no more inconsistent with 
 the genius of the language than to have a dative of instrument, the 
 former by its very nature being but the latter in a transferred, 
 quasi-demateriaHzed, and then personalized sense, to denote the 
 subject-in-action as against the subject-in-inaction. 
 
 1 Cf. Vg/. Synt. I. § 107 ff.; Synkret. § 22 ff.; Winkler, Germ. Cas. p. 100 flf.; 
 van der Meer, § 81 ff.; Piper, 26 ff. 
 
95 
 
 § 1 23. In the light of the foregoing — designed partly to counter- 
 act the tendency of ascribing all of Wulfila's agreement with his 
 prototype to mere imitation — the second difficulty, reference to 
 which has been made in the introductory remarks of this section, 
 may be easily removed. For, by that same argumentum ex silentio 
 which, with perfect justification, has heretofore been applied to 
 quantitatively unsatisfactory Gothic, the lack of sufficient instances 
 of a dative-instrumental of agency ought not to preclude the possi- 
 bility of its existence in that language. Delbrtick, to cite but one 
 example of such reasoning {Synkretismus, p. 240), acutely observes 
 that, judging by the testimony of kindred Germanic dialects, the 
 comitative must undoubtedly have been in use in Pre-Germanic. 
 That, however, it is absent from Gothic should not be used as an 
 argument against its ever having existed in Gothic. As it happens, 
 Wulfila's Greek text is equally devoid of this construction. This 
 lack, then, is a mere accident, due to the limited amount and the 
 peculiar quality of the literature at our command. "Hatten wir 
 ein Epos, so wurde darin auch wohl der Furst * mit grossem Ge- 
 folge ' auftreten und dieses letztere im Komitativ stehen." Accord- 
 ingly, we cannot persuade ourselves, in possession if not of conclu- 
 sive direct, yet, at least, of highly significant collateral, testimony on 
 the part of related languages, either to deny the existence of a dat- 
 inst. of agency in Gothic, or to rest content with a verdict of igno- 
 ramus concerning it. Scarcity does not mean absence. In our 
 case we simply must deplore the lack of further indisputable 
 evidence. 
 
 Anglo-Saxon. § 124. The trouble Delbriick complains of {Syn- 
 kretismus, § 73) in connection with the Gothic dative-agent, in that 
 it always corresponds to a Greek dative and thus renders it difficult 
 to grasp both its Germanic character and its original force, should 
 partly be removed by the fact that a similar construction, even tho 
 not abundantly exemplified because of analytic tendencies, occurs 
 in early Anglo-Saxon also. The present investigation has restricted 
 itself of necessity to the poetical works in this language, " was — 
 to say with Synkretismus, Einl. p. i — wegen ihrer altertiimlicheren 
 Ausdrucksweise gerechtfertigt war." 
 
 § 125. Little attention has been given to this phase of Ags. 
 syntax. What we have consists practically of short notices in syo.- 
 tactical compendia to Readers, like Baskervill and Harrison, 1898, 
 p. 51, or Sweet, 6th ed. 1888, p. 87. Kress, Ueber den Gebrauch des 
 
96 
 
 Inst, in der ags. Poesie, 1864, neglects to mention it, as does also 
 Nader in his Dativ und Instrumental im Beowulf} For Caedmon, 
 Hofer accords it a place, but not satisfactorily.^ 
 
 § 126. Beowulf. — This earliest epic in any Germanic tongue 
 is remarkable in that it contains, besides prepositional turns, a very 
 rich abundance of such prepositionless datives as go back to 
 former locatives, ablatives, and instrumentals more often absolutely 
 reducible than not. A special form for the inst. exists only in the 
 sg. masc. and neut. of certain adjectives and pronouns as well as 
 of the present participles, but its functional fusion with the dative 
 is evident also from instances where the latter, tho different in form 
 from the inst., is employed to express the function of means. This 
 dat.-inst. is found in all functions germane to our discussion.' Ex- 
 amples are not at all infrequent of pure concomitation with verbs 
 of motion, like cuman^faran, ge-l^dan etc., as 923 tryddode getrume 
 micle ; so cordrum micluniy cordre, heapum, herge, folce etc. Tran- 
 sitional types are present, as 2936 bescet sinherge sweorda lafe^ *■ sur- 
 rounded with an army those that had escaped the swords.' The 
 inst. of means is again unlimited in application — persons are also 
 used as means, 1018 Heorot innan wees freondum dfylkd — and 
 is found likewise with passive expressions, 406, 2087 slowed smuSes 
 orf>ancum ; 775 iren-bendum besmi'dod. More personal instances 
 are 217 winde gefysed ; 21 12 eldo gebunden ; 3073 wommum gew'it- 
 nad etc. Very near the actual agent is such a ^ potential example 
 as 991-2 f>a wees handum — foltnum gefrcetwod, or 31 17 panne str^la 
 storm strengum geb^ded^ ' von den sehnen geschnellt.' 
 
 § 127. The idea of agency itself, as relating to persons, is rare 
 in Beowulf, /furh is not yet found in this sense, fram has only a 
 locatival signification, as 1. no. Passive expressions are frequent 
 enough, but the terseness of the style seems easily able to dispense 
 with such defining elements as agency offers. The following ex- 
 amples have been selected for discussion.^ 
 
 § 128. 1. 12 J7^m eafera waes aefter cenned geong in geardum — 
 Heyne-Schiicking (HS.) : ' dem wurde ein Sohn geboren.' A simi- 
 
 1 Reviewed in Eng. Stud. VII. 368. 
 
 2 Der synt. Gebrauch des Dativs und Instrumentals" in den Ccedmon beigeleg- 
 ten Dichtungen, Diss. Leipzig, 1884, and also in Anglia, Vlt. 355. 
 
 3 The tenth Heyne-Socin-Schiicking edition oiBeowulf, Paderborn, 1913, is used. 
 * For others of the above types I must refer to Winkler, Germ. Cas. p. 448 ff.; 
 
 Nader, op. cit.; Synkret. § 19 ff. 
 
97 
 
 lar instance is 1. 1357 hwcederhim ^nigwces ^r acenned dyrnra gasta^ 
 for which cf. Winkler, p. 379. These lines bring up the question 
 as to whether verbs of procreating, erzeugen, must be taken intran- 
 sitively when in the passive. That there cannot be laid down an 
 absolute rule governing the thought of all Indo-European languages 
 need not be emphasized. However, consider Sanskrit which per- 
 mits two points of view, ablatival, RV. i, 123, 9 sukra krsnad 
 ajanista, ' out of the darkness born,' and the other instrumental, of 
 authorship, sdhasa yo mathito jayate nfbhih^ ' begotten by men ' ; so 
 RF. 2, 25, I. Avestanhas ablative, cf. V. 2,41 (Reichelt, § 475 ; 
 Hiibschmann, p. 234); so the Latin, Telamone creatus^ Ov. Met. 
 13, 22 ; similarly Greek by means of an ablatival genitive, o Atos 
 iieyivovTo, E 637. However, Slavic, again, uses the instrumental 
 for the agent, in contradistinction to material origin which is in 
 abl.-gen. (cf. for latter, Vondrdk, II. 334), so Mt. 11, 11 vb rozde- 
 nychi> zenami, iv y^w-qroh ywatKw (Miklosich, IV. 704). For the 
 Germanic, where datives represent both ablatives and instrumen- 
 tals, either of the last two cases is permissible,^ the former being safest 
 in expressions of ongin, as OHG. sie sin Alexanders siahiu. Gothic 
 has no examples at all.'^ The Old Norse correspondences will be 
 enumerated in the next division, cf. § 146 ; it is premised, however, 
 that Grp. 3, 3 borinn Sigmundi, is not to be rendered simply by ' born 
 to S. ' because of Hdl. 25,3 bornirfrd Jgrmunreki. That in the Ags. 
 we have to do with agency and not interest, is evidenced by the 
 later more precise parallels, El. 775 sl-'de on rode wees ond f>urh 
 Marian in middangeard acenned weard ; Blick. Horn. ^1,24. se ilea 
 sunu wcBs Er eallum tldum acenned f ram God Feeder, se yElmihtiga 
 frompon ^Imihtigan; so 59, 25 ; 93, 28 ; 167, 20. For ^Ifric, who 
 uses ^instead oifram^ ci. I. 10, 2, M ure heelend crist acenned wees 
 of peem halgan meedene marian ; so 14, 75 etc. For the Bible Jh. 3, 6, 
 peet pe of gaste is acenned. Of course, in Beowulf we cannot as yet 
 expect such prepositions, but it is safe to conceive of an unaltered 
 and uniform mode of thought thruout. As to 1. 1357, Trautmann's 
 emendation ^ of it as hweeper hie anig hafap ^r acenned, ' ob sie, die 
 ungeheuer, irgend einer friiher erzeugt hat der finstern geister,' is 
 
 1 ALL p. 13 ; Synkretismusy § 23. 
 
 2 Cf. Grimm, IV. 714 to Mt. i, 16 and Jh. i, 13. 
 
 * Fischer, Der synt. Gebrauch der Partikeln of und from in ^elfric's 
 Heiligenleben etc., 1908. 
 
 * Bonner Beitr'dge zur Anglistiky XVI. 76. 
 
unnecessary. Ten Brink's interpretation^ is more satisfactory, 
 " den hatten die Erdbewohner Grendel gennant ; man wisse nicht, 
 ob sein Vater vor ihm schon irgend einen andern im Dunkel hau- 
 senden Geist gezeugt habe," but Schiicking's translation in ed. lo, 
 s.v. a-cennan, is different. For the construction cf. 1. 1180. 
 
 § 129. I am inclined to render 1. 646, wiste J7am ahlaecan 
 hilde gej^inged, because of the context, as ' he knew that a fight was 
 contemplated by the monster.' Hrot5gar left the hall not because 
 
 * er wusste dem scheusal im hochsale kampf bestimmt ' by his own 
 followers, but because he knew that Grendel calculated to appear 
 sid^an hie sunnan leoht geseon ne-meahton. On the other hand, a 
 translation like ' he knew that a fight was appointed unto the mon- 
 ster,' by Fate, that is, is fully in accord with the Anglo-Saxon con- 
 ception of such occurrences. 
 
 § 130. 1. 1068 Finnes eaferan haelea Healf-Dena feallan scolde 
 — Zupitza's Autotype, p. 50, shows eaferum, but eaferan is ac- 
 cepted by Klaeber'^ following Trautmann.^ If, with Wiilker, 
 Wyatt, Garnett, Hall, Tinker and Child, we do not follow Klaeber- 
 Trautmann in making 1. 1068 the end of the sentence, but con- 
 sider it the beginning of the Finn-song, we can retain the MS. 
 reading: 'thru [the instrumentality of] Finn's sons was Hnaef 
 destined to fall.' This would obviate the change of healgamen^ 
 to heal-guma, ' a hall-man,' and the sense of the passage ends suit- 
 ably with mcsnan, of which the subject is kea/gamen, meaning 
 
 * when HrotSgar's bard was to proclaim joy in the hall along the 
 mead-bench,' Clark Hall. This would also eliminate Thorpe's 
 de and Socin^s /ram, — the latter I cannot accept because in no 
 other place in Beowulf would /ram be found thus used, — which 
 disturb the meter, and would still allow ea/erum to be taken in the 
 same sense, as a dative-instrumental of agency. The resulting begin- 
 ning with 1. 1068 of the song, so taken also by Holthausen and Sedge- 
 field, in no wise alters the rationale of the otherwise obscure episode. 
 
 § 131. 1. 1151 ]7a waes heal hroden feonda feorum. In its un- 
 changed form, cf. Zupitza, it has but one meaning, ' then was the 
 hall strewn with the foemen's corpses,' i.e. /eorum is a dat.-inst. of 
 means. But this is not in harmony with the context. Bugge's 
 correction roden^ ' reddened,' recommended also by Sievers,® Rie- 
 
 1 Beowulf, 1888, p. 95. 2 Anglia, 28, 443. 
 
 * Bonn. Beitr. II. 183. * Made by Trautmann, Bonn. Beitr. II. 183. 
 
 6 Tidskr.for Phil, og Paed. VIII. 64, 295. « PBB. IX. 139. 
 
9d 
 
 ger,* and Gering,^ is altogether correct despite its non-acceptance 
 by Wiilker, Socin and Wyatt. Schucking's tenth edition adopts 
 roden and translates the passage, " da war die Halle rot von er- 
 schlagenen Feinden " and " von der Feinde Leichen gerotet," pp. 
 189 and 257. The rendering of either ' the hall was adorned with 
 corpses ' or ' the hall was covered with corpses ' is inadmissible in 
 the place. Trautmann ' effects even a further change into D5 wees 
 heal stro den feonda foicum,* and Barnow^ to an impossible fia wees 
 heakroda, feond afeorred. Holthausen's suggestion® that since 
 feorum cannot here denote ' corpses,' it is miswritten for dreore^ as 
 1. 447 d. fdhne, is really beside the mark. Add also Holtzmann, 
 Germania, 8, 492. Whether we adopt Bugge's view and translate 
 with Klaeber^ 'the hall was made red by living beings,' as in 
 Exod. 384, or Trautmann's emendation into ' darauf ward die 
 halle gepliindert von den haufen der feinde,' we have a dative-agent 
 in both ^feorum ' and '■fokum,^ with more instrumental force in the 
 MS. form. 
 
 § 132. Trautmann's conception of 1. 11 03 as ]7a him swa ge- 
 peahtod [not HS. ge]7earfod] waes would give the translation ' then 
 it was decided by them ' but this is undoubtedly too bold. — Gu. 
 1274 husle gereorded^ ' eucharistiae recreatus ' ; And. 385 pa he 
 gereorded wees, 'cum cibo refectus erat' (Bouterwek, 115) and the 
 meaning given gereordian in ^Ifr. Gram. 26 ' prandeo,' * vescor,' 
 ill agree with the HS. rendering of 1. 1787, J^a waes eft swa »r 
 ellen-rofum flet-sittendum faegere gereorded, as 'da ward wie 
 vorher den Saalsitzenden das Mai gereicht.' The meaning is 
 rather, 'there was suitable feasting by the strength-renowned 
 heroes, the hall-company ' — 1. 1696 hwam J^aet sweord geworht . . 
 arest waere — cannot, on the one hand, be taken as an instance 
 of agency, even tho the maker's name does appear as on the 
 oft-quoted horn of Gallehus ^ ek hlewa^yastiR. holtinaR. horna ta- 
 wi'do, because of the presence in the sentence of arest, if, that 
 is, the signification ' at first ' is to be attached to this word. If, 
 on the other hand, ^rest is to mean ' in olden times ' as in 
 1. 5 of Caedmon's Hymnus ^^ he mrest sceop eordan dearnum^^^ 
 
 1 lacker's Zs. 3, 404. 2 /^^-^ ^2, 124. 
 
 ' Bonn. Beitr. 16, 65. * In Bonn. Beitr. 2, 190, he suggests /^rww. 
 
 6 Textkritische Untersuchungen^ 1902, p. 232. 
 
 6 Zacher's Zs. 37, Il6. ''' An^lia, 28, 445. 
 
 8 Cf. Noreen, Altisl. u. altnor. Gram. 1892, p. 257 ff. 
 
100 
 
 and elsewhere, then hwam may be accepted as a case in 
 point. 
 
 § 133. 1. 2035 ]7onne he mid faemnan on flett g^3, dryht-bearn 
 Dena duguSa biweneSe — HS. would make of duguda an ace. pi. 
 " wahrend ein edler Spross der Danen die Ritter bewirtete." For 
 other interpretations cf. Kluge FBB. 9, 190 if. ; Rieger in Zacher^s 
 Zs. 3, 404 makes it an instrumental genitive, 'beneficiis adsuefac- 
 tus'; Bugge, FBB. 12, 98 accepts Grein's duguda bi werede^ 'in 
 der Hoflinge Schaar ' ; Holthausen, in Zacher's Zs. 37, 119, is for 
 dugude bi werede^ dugude being a genitive dependent on werede = 
 werode. Thorpe's dugude bipenede, 'by the noble served,' is near- 
 est of all to the sense of the passage ; but, perhaps, Trautmann's 
 rendering of the obscure word in the MS. as beweotede is a better 
 transliteration,^ and the change of the agent to dugudum is also 
 more desirable, hence 
 
 " wenn er mit der frau in die halle geht, mit der 
 edelmaid^ der Danen, der von ihren degen bedienten."' 
 
 Finally, if we accept Sedgefield's text, even Trautmann's bold 
 correction may be avoided, " From what follows it is evident that 
 the young Danish bride is accompanied to her new home by a 
 band of her father's thanes, dugude biwenede,''^ p. 173, and the fol- 
 lower referred to wears the captured Heathobard sword. Klae- 
 ber's equation of he with dryhtbearn is unconvincing, both because 
 it leaves Dena unattached and because it would make a Danish 
 warrior of Ingeld. 
 
 § 134. 1. 2435 "^^s ^^^^ yldestan mieges dadum moriSorbed 
 stred, shows the passive verb with both a dat. incommodi and a 
 dat.-inst. of the potential agent: m^ges dcedum is tantamount to 
 m^ge. Rieger's * change of stred to styred on stylistic grounds is 
 unwarranted. — A similar clear case of a dat-inst. of agency is 1. 
 2842, Biowulfe wears dryht-maSma dsel deaSe forgolden, 'by B 
 his share of lordly treasures was paid for with death.' So Traut- 
 mann's 'Von B. wurden bezahlt.' 
 
 § 135. 1. 2957 }?a waes aeht boden Sweona leodum, segn Higelace. 
 Trautmann changes ceht to cefst, Klaeber J/Z^. 20, 85 disagreeing, 
 to mean ' then was disaster bidden to the folk of Swedes, good for- 
 
 1 Bonn. Bntr. 16, 114. ^ Noble descendant. 
 
 8 Cf. besides Wyatt, adloc. p. 86; Eng. Stud. 39, 464 f. and Klaeber in Mod. 
 Phil. 3, 255. * Zacher's Zs. 3, 409. 
 
101 
 
 tune to Hygelac' Sievers' substitution, PBB. 9, 143, of sacc Hy- 
 gelaces for segn \_Hygelace\ proposed by Bugge, Tidskr. 8, 61, is 
 not accepted by the latter in PBB. 12, 108, tho it would harmonize 
 better with cRht, cf. Klaeber, Mod, Phil. 3, 240. Schroer, Anglia, 
 13, 346 ff., points out that ^^/ occurs nowhere else in the sense of 
 ' Verfolgung ' given to it in this passage by Sievers and insists on 
 its real meaning ' possession, treasure ' = ' hord ' two lines before. 
 The meaning of the whole passage would suffer if this were accepted. 
 Bugge's opinion, PBB. 12, 18, that segn Hygeldces is in the same 
 construction as ^ht^ therefore * then chase was offered to the Swed- 
 ish folk, the banner of H. raised,' — " Das erhobene Banner ist das 
 Merkmal der Verfolgung " as in Hdrbar^sljod 40 ek vark i hemum, 
 er hingat g0rdisk gncefa gunnfana geir at rjo'da — which would leave 
 as the subject of ofereodon the Swedes, understood — so, too, Sie- 
 vers, PBB, 9, 143 — has been neglected as to the second part by 
 Holthausen, who makes segn the subject of that verb, * the banners 
 of H. overran the fastness.' I propose a modification of Schroer's 
 view of a tentative compromise before the battle, '"wces ceht boden^^ 
 and translate not as he ' then was the treasure of the Swedish folk, 
 their banner offered to H., as a ransom,' but 'then was treasure 
 offered by the Swedish folk, their banner to Hygelac.'* This ren- 
 dition has the following advantages : 
 
 (a) It obviates Sievers' change of MS. Hygeldce to Hygeldces; 
 
 (J?) It obviates Schroer's change of MS. leodu{m) to leoda; 
 
 (c) It retains the original and only meaning of ^ht^ i.e. < dgan, 
 to possess ; 
 
 (^) It makes unnecessary Trautmann's cefst for MS. aht. The 
 subject of ofereodun would be as in Sievers-Bugge above; leodum 
 an instance of the dat.-inst. of agency. 
 
 § 136. In conclusion I would Hke to draw attention to 1. 2983 
 J?a him gerymed wear3 and 1. 3088 J^a me gerymed waes, as being, 
 at least in the former, amenable to an interpretation involving the 
 original signification oi geryman^ as in 1086 pcet Kie him oder fiet eal 
 getymdon, ' to clear, open.' So^very often : ic him Itfes weg gerymde, 
 El. 1249 ; se weg bif> us gerymed^ An, 1582, etc. If now 1. 2983 is 
 considered in this light, a translation ' when the power over the 
 battle-field had been allowed them ' is wholly unnecessary, because 
 
 1 For the banner offered to H., tho in a different sense, cf. Cosijn, Aanteek- 
 ningen op den B.y 1891, ad loc. 
 
102 
 
 of the alternative of a more literal ' raised him quickly up when the 
 place was cleared by them so that they could control the battle-field.* 
 This brings to a close the list of examples in Beowulf. 
 
 § 137. With the remark that most, if not all, of the emendations 
 offered or adopted above are absolutely necessary, in view of the 
 deplorably inaccurate condition of the MS. — " B. may, I believe, 
 be conscientiously pronounced the worst," Thorpe — and of its 
 often one-sided interpretation, we pass on to the real and alleged 
 Caedmon whose Hymn is placed by Trautmann chronologically in 
 line after Beowulf, at c. 600-700.^ The question of authorship, of 
 course, in a purely linguistic study of given texts, is of no impor- 
 tance whatsoever; reference may be made, however, to Korting, 
 Grdr. d. Geschichte d. engl. Literatur, 1905, p. 42 if., and p. 357 ff. 
 of Bethge's Ergebnisse und Fortschritte d. germanist Wissenschaft 
 im lehten Vierteljh., Leipzig, 1902. 
 
 § 138. As against Cynewulf, the next author to be considered, 
 C^DMON ^ still possesses a prepositionless dat.-inst. to express con- 
 comitation, as Gen. 1798 drihiweras dugudum geforan ; Gen. 2454 
 hie behcefdon herges mcegne Loth, etc., tho that same prepositional 
 competition seen in Beow, 11 28 mid Finne, so 1. 242-3, is making 
 a greater headway here, e.g. Ex. 501 Faraon mid his f oleum; so 
 Dan. 67, Sat. 203, Jud. 170. The prepositionless dat.-inst. of 
 means is also present. Gen. 2550, and persons are likewise used as 
 means, Gen. 95, 1655. Impersonal agents with passives are also 
 frequent, as Gen. 1293 synnum gehladene ; Dan. 295 lige belegde. 
 Entirely in the sense of Brugmann's Potenz is Dan. 277 deaw-dnas 
 weorded winde geondsdwen^ or Dan. 406 f>u gebletsad eart Hdlgum 
 mihtum. 
 
 § 139. The following examples have been found for the dat.-inst. 
 of agency : 
 
 1 Cf. Trautmann, Bonn. Beitr. I. 121. Caedmon, it is true, died in the last 
 quarter of the 7th century, but then the redaction of Beowulf \% much later than 
 the poem itself, altho L. Morsbach " Z«r Datierung des Beowulfepos^^ {Nachr. 
 d. Gottinger Ges. d. Wiss.j pkil.-histor. Klasse, 1906, p. 251 ff.) sets the date of 
 Beowulf zs late as 700-730, therefore about three hundred years before the MS. 
 Cf. also A. Brandl, Geschichte der ae. Literatur, Pauls Grdr? II. 991. 
 
 2 No distinction as to genuine and spurious works is attempted here, entirely 
 aside, of course, from the question as to Caedmon's real authorship of any of the 
 poems. For dissertations, cf. those of Hofer, Oldenburg, Kempf, Dethloff, and 
 Meyer on various syntactical points. All available editions have been drawn 
 upon. 
 
103 
 
 Gen. 1 765 fro mcynn efolde weorded ptne gefylled ; 
 
 Gen. 1967 pa wees gudhergum be Jordane wera Welland geondsen- 
 
 ded, folde feondum; 
 Gen. 1999 gewitonfeorh heora . . . fieame nergan, secgum ofslegene; 
 Sat. ^^^ pa wees on eordan Ice drihtenfolgadf oleum ; 
 Dan. 92 metode geeorene^ 150 se wees drihtne gecoren ; 
 
 so 736 ; Jul. 604 etc. ; to be so taken because of the intent evi- 
 denced by the later use of ''fram ' [as Blickl. Horn. 187, 20 ''gecoren 
 weerefram gode^ etc.] which here is used only locally, cf. Kempf, 
 p. 37 ; Dethloff, p. 57 ; Meyer, p. 12. Other examples are not as 
 clearly definable. Hofer {op. cit. p. 40 f .) approximates what seems 
 to be the truth when, in speaking of the dat.-inst. of means he re- 
 marks, " wird . . . die aktive konstruktion in die passive verwan- 
 delt, so hat man zwei falle von einander zu scheiden : Entweder 
 das handelnde subjekt des aktiven satzes wird auch im passiven 
 erwahnt, oder es kommt im wegf all. Im ersten falle bezeichnet auch 
 bei passivem verb der vorhandene dat.-instr. dasjenige mittel, durch 
 welches vom logischen, handelnden subjekte, — italics mine, — als dem 
 ausgangspunkte, die tatigkeit des verbums auf das leidende subjekt, 
 als den zielpunkt, tibertragen imd ausgefiihrt wird. 1st dagegen 
 die letztere bedingung erfiillt, so erscheint das friihere mittel im dat.- 
 instr. jetzt als veranlassende ursache." If we add that besides and 
 beyond this ' Potenz ' that same acting subject itself may, and in 
 Beowulf and Ceedmon exclusively does, occur in the dat.-inst., his 
 statement is more correct and complete. 
 
 § 140. After Caedmon the tendency in poetry is entirely toward 
 the analytical. While, for instance in the Riddles, which Sievers ^ 
 assigns to the first half of the eighth century ,2 occurs but one doubtful 
 example of the dat.-inst. of agency, ed. Grein xxi. 20 ne weorpe'd sio 
 mEgburg gemicledu eaforan minum, that is, if the sense 'be mag- 
 nified by my posterity ' can stand, in the contemporary Cynewulf^ 
 this is unknown. So in Elene, a typical poem, all personal author- 
 
 1 Anglia, 13, I ff. 
 
 2 So Tupper, The Riddles of the Exeter Book, Albion Series, pp. Iviii and Ixxix. 
 
 3 Both his signed ^oems, Juliana, Christ, Elene, The Fates of the Apostles, and 
 those attributed to him, Andreas, Guthlac, Phoenix and Dream of the Rood, have 
 been read according to various editions. For a discussion of authorship, date etc. 
 cf. A. Brandl in Pauls GrdrJ^ II. 1040 ff.; C. W. Kennedy, The Poems of 
 Cynewulf, Lond. 19 10; Korting, Grdr.^ p. 49 fF.; Bethge's Ergebnisse, p. 364 ff.; 
 Bonn. Beitrdge, I. and XXIV. for bibliography up to 1908. 
 
104 
 
 ship is expressed by the preposition purh}- which in Caedmon was 
 not found in such use.^ So 
 
 EL TT^ se-de on rode wees ond f>urh Marian in tniddangeard 
 
 acenned weard in cildes had; 
 840 pa wees hige onhyrded purh p(2t haiigg treo ; 
 1058 purh gdstes gife gecorenne, should throw a light on Dan. 92, 
 
 150. 736: 
 1 1 53 wees se witedom purhfymwitan sungen etc. in spite of turns 
 
 like 
 1243 ic wees weorcum fdh^ synnum as^led, sorgum geweeledy bitre 
 
 gebunden, bisgum beprungen, cf. also 1264. 
 
 § 141. *■ Fram ' is not yet used for denoting the personal agent, 
 despite Kent's glossary, ed. 1899 of Elene : 1. 701 is simple means ; 
 1. 1 141 is purely locative and 1. 190 swa fram Siluestre larde 
 w^ron^ while undoubtedly auctorial in intent, permits the locative 
 force to be still distinctly felt, cf. German ' von seiten.^ Analytic 
 uses, however, as such are quite common ; mid is employed with 
 both datival and instrumental forms to express not only comitation 
 but pure means. Andreas, a work much influenced by Beowulf, 
 was examined for statistics of the notation of means, — in Beowulf 
 regularly prepositionless. The result: 166 instances of pure da- 
 tival or instrumental form, in the broadest sense, including manifold 
 repetitions of the same words, such as mihtum, crceftum, meegene^ 
 which are really poetic stock-words, and also modals like hludan 
 stefne^ heapum etc. ; purh^ 42 times ; mid^ 10 times ; of^ 3 times. 
 And so on in succeeding works. In the Byrhtno'd of later date, c. 
 991, the pure instrumental is almost entirely given up, to be circum- 
 scribed by mid c. prep. This, 2ig2.\nstjudilh, for instance, where the 
 old usage is still predominant in a 2 : i ratio, is significant of the 
 progress of analyticism. So, in turn, the relation oi Judith to Exo- 
 dus is 4 : I, and of the latter to Beowulf, as above. In OE. Orrmu- 
 lum, c. 1200, the instrumental usage, with the exception of petrified 
 locutions, has entirely disappeared.^ Naturally personal agency 
 
 1 Cf. Simons, Cynewulfs Wortsckatz, Bonn. Beitr. III. p. 144, for examples. 
 
 2 The dissertations of Taubert, Schiirmann, Conradi, Reussner and Rose are 
 of interest in this connection. 
 
 * Cf. Funke, Kasus-syntax bei Orrm und La'j^amon, Diss.TMunchen, 1907, 
 Einl. IV. and p. 55 fF.; Swane, Studien zur Casus-syniax des Friihmitteleng- 
 lischen, Diss. Kiel, 1901, p. 61. 
 
105 
 
 with passives is expressed by means of prepositions, as Orrm, 12846 
 patt he mass pewwtedd unnderr ce. purrh Issracele peode, and we meet 
 with no more instances of the dat.-inst. of agency.^ 
 
 Old Norse. § 142. The investigation of ON. was thought to be 
 important both ' an und fiir sich ' and in confirmation of results 
 reached in the discussion of the Gothic. " Ich zweifle nicht, says 
 
 1 The various stages of auctorial expressions with passives in English might 
 be given as, (a) prepositionless dat.-inst.; (Ji) purh with the accusative; (c) 
 /ram with the dative ; (^) ofwithjhe dative^ and lastly (<r) be or by^ as to-day. 
 For {a) cf. Beowulf; for (3), Cynewulf. As to fram, from, there are all to- 
 gether about seven examples of it in Ags. poetry, to wit Cyn. El. 190 fram Sil' 
 uestre ; Crist 161 7 scyle from his Scyppende dscyred weortSan ; Soul and Body 46 
 ic Wees gsest on pe from gode sended, really locatival ; Metra 20, 245 swd him 
 lyfed wees from pam celmihtigan ; Psalms 67, 23 from pdm pine gangas wseron 
 gesewene, 113, 23 wesdS ge fram gode gebletsade ; Pater Noster 3, 87 pe fram wife 
 and fram were wurdon dcenned. (Cf. pp. 2 and 21 of Wullen, Der synt. Ge- 
 brauch der Prop, fram, under in d. ags. Poesie, Kiel Diss. 1908.) In Alfred 
 fram is the rule, of the exception. (Cf. Harstrick, Untersuchungen uber die 
 Prdp. bei Alfred. Diss. Kiel, 1890, p. 13.) Interesting is ^. 3, 14 he wees fram 
 eallum monnum lufad. (For the Blickl. Hom. as 187, 20 pu (Er gecoren wcere 
 fram gode, cf. p. 86 ff. of Fischer, Der Synt. Gebrauch der Partikeln of und 
 fram in ^Ifric's Heiligenleben und in den Blickling-Homilien. Leipzig Diss. 
 1908.) In Chaucer and onward from is only locatival, as at the very beginning, 
 from Pluto sent, at requeste of Saturne, II. cf. Einenkel, Streifzuge durch die me. 
 Syntax, Mvinster, 1887, pp. 44. — 6>/fluctuates for a while with /row, so iElf. O. 
 154, 28 seo burg wees getimbred of Lcecedemonium, but O. 164, 10 sio wees getimbred 
 from Elisan peem wifmen. (Cf. Hardstrick, op. cit. p. 5 and Jacobsen, Der 
 synt. Gebrauch der Prdp. for, geond, of und ymb in d. ae. Poesie, Kiel Diss. 
 1908.) "With acennan of is regular in jElfred. Certainly less frequently used 
 in OE. than fr a m, of begins to gain ground on the latter preposition in Sax. 
 Chron. R, F and the interpolations of A. (Cf. Bedtker, Critical Contributions 
 to Eng. Syntax, Christiania, 1908, p. 6.) The extensive ME. use of of as in 
 Chaucer 6, 309 thi wille fulfillid be of thi sone, is attribute4 by Einenkel, op. cit. 
 
 162, to the influence of OF. de, as Joinv. 232 il estoient si pressei des Turs que 
 etc. — Be, modern by, is the regular preposition of agency in MnE., rarely of, as 
 against ME. of, rarely by. Despite Einenkel, p. 131, be is found denoting the 
 personal means before Chaucer, as ^Ifr. 2, 170, 14 sum eawfeest man sende twe- 
 gen butrucas mid wine be anum cnapan. (Cf. Gottweiss, Die Synt. der Prdp. 
 eet, be, ymb in den ^Ifric-Homilien, Anglia, 28, 353.) Nay, even in the sense 
 of the personal agent, Beda (>yj, 3 ^a tSe be him geddne weeron ; so Blickl. Hom. 
 
 163, 27. Farther back, in Csedmon, Gen. ^Oi% pat wurde pegn swd monig for- 
 leedd be pam lygenum. (Cf. also Dusenschon, Die Prdp. eefter, eet und be in 
 der ae. Poesie, Diss. Kiel, 1907.) Its modern use, in spite of its infrequency in 
 Ags. and OE., is attributed by Einenkel, p. 132, to the influence of French par, 
 as Chev. Lyon 5127 que ia par toi nHert reconte.') 
 
106 
 
 Bernhardt,^ dass diese beiden zweige am fruhsten dem gemeinsa- 
 men stamme der germanischen ursprache entwuchsen. Nicht 
 minder zeigt sich in der syntax zwischen beiden iibereinstim- 
 mung, und aus der reicher fliessenden quelle des altnordischen ist 
 fiir das gotische oft verstandnis zu gewinnen." To establish such 
 a similarity, the entire poetic Edda^ has been covered as more 
 important for case-syntax than the prose-literature ; ' quite a num- 
 ber of examples, however, have been adduced from the later prose- 
 literature, a comprehensive syntax of which is now being prepared 
 by Streitberg.^ 
 
 § 143. In the instance of the Edda we have not to deal, as in 
 Gothic, with a text that is open to the charge of vitiation by depen- 
 dence upon a foreign original, but in return we are confronted 
 with the difficulty inherent in late recensions. How far the oldest 
 MSS. we possess may be taken as an exemplar of the language of 
 the time of composition and how far we may speak of alterations 
 in the texts made by some officious meddler to bring the old lines 
 more in accord with the linguistic status or metrical technique of 
 his time, will always remain one of the moot questions of Eddie 
 research. Certain it is that the text transmitted to us shows con- 
 siderable corruption : " gegeniiber dem torichten gerede, dass der 
 Eddatext, wie die handschriften ihn bieten, ' von gebildeten Islan- 
 dern und Norwegern im 13. und 14. jahrhundert verstanden und 
 gewiirdigt worden sei ' — muss dies einmal mit aller entschieden- 
 heit betont werden."^ Hence the absolute necessity of textual 
 revision, a fact which in following paragraphs must always be kept 
 in mind. 
 
 § 144. Morphologically Old Norse has only such formations as 
 are syntactically worthless.® The language of the Eddie poems, 
 as such,^ is essentially the same as that of the older Skaldic poetry. 
 
 1 Beitr. z. d. Phil. Halle, 1880, p. 73. 
 
 2 Ed. Hildebrand-Gering3, 191 2. 8 cf. Synkretismus, Einl. I. 
 
 * Cf. IFA. for 1906, announcements. ^ Gering, Introd. to Hildebrand^. 
 
 6 Noreen, Gesch. d. nord. Sprache^ 609 ; Synkretismus, p. 152 ff. 
 
 ■^ The consensus of scholars in general is that they did not arise all at the 
 same time, but that centuries must have elapsed between the oldest, such as 
 VglundarkviiSa, Hdvamdl, and the youngest poems, none, however, originating 
 before the first half of the ninth century, all being composed in the Viking period, 
 i.e. c. 800-1050. Cf. Mogk, Gesch. d. Norwegisch-Isl'dndischen Lit?- 1904, 
 Pauls Grdr? II. 567 ff.; Sijmons, Die Lieder der Edda, Einleitung, § 28; 
 Bugge, The Home of the Eddie Poems, 5, 2 ; Finnur Jonsson, Litt.-Hist. I. 47. 
 
107 
 
 To reduce the oldest of the former to Proto-Norse forms would dis- 
 turb their metric arrangement.^ In fact, such a thorogoing lin- 
 guistic unanimity exists between all of them that it seems as if no 
 decisive change at all had taken place in the language during 
 the Viking period. As against this, witness the manifest advance 
 in case-syntax, especially with regard to prepositional usages, 
 from the Edda to the oldest, tho nevertheless younger, prose- 
 works.^ 
 
 § 145. For the genealogy of the dative-instrumental of personal 
 agency, as established in the foregoing divisions and chapters, the 
 following examples may be offered.' 
 
 (a) Pure concomitation is extremely common in Icelandic. For 
 the Edda, Vsp. 36, 2 g fellr austan sgxutn ok sverpum ; 51, 4 Peim 
 es bropir Byleists ifgr; BHv. prose 12, Helgi ok Atli Idgu skipum i 
 Hatafirf>i ; Akv. 17, 2 sent hjglmum aringreypum at sea heim Atla. 
 But already here mef> competes. Vsp, 51,3 fara fiflmeger mep 
 freka aller ; so Hdl. 5,2. 
 
 {p) Nearer the instrumental is the sociative found so strange by 
 Winkler, p. 470, " rida mar innum melgreipa Myrkvid okunnan 
 (Akv. 3, 2, ripa mar enum melgreypa Myrkvip okunnan)^ auf dem 
 gebissknirschenden rosse durch den unbekannten schwarzwald ; " 
 so Hrbl. 53 ro pu hingat bdtinum. Non-Eddie, rem gllum skipu- 
 num, Hkr. 225, 16 ; ri'dr Nott peim hesti, Sn, E. 7 (cf. Lund, p. 86). 
 Because of phrases like reid a vargi, HHv. 35, i, a locatival con- 
 ception is by no means impossible here.^ 
 
 ic) There is nothing that cannot be expressed as an instrument, 
 " als werkzeug des schlages konnen auch personen dargestellt 
 sein." (Dietrich in HZ. 8, 62.) 
 
 (rt?) Passive expressions with the common inst. of pure means 
 are common, sleginn sessmeWum, Akv. 14, 3 ; Atli slegenn rog-por- 
 num, 31, 2 ; Guprun hlapen haismenjum, Am. 43, 4; so Hkr. 346, 
 16 hann var gyrdr sverdi ; 2, 168 var hgndum tekinny etc., etc. 
 
 1 HofFory, Eddastudien, p. 36 fF. 
 
 2 Cf. Gebhardt, Beitr. z. Bedeuiungshhre der awn. Prdpositionen, Leipzig 
 Diss. 1896. 
 
 8 For further information, cf. Detter-Heinzel, II.; Dietrich in ^aw/^'j Zj. VIII. 
 23 ff.; Lund's Ordfajningslare ; Nygaard's Eddasprogets Syntax and Norran 
 Syntax, finally Winkler, Germ. Cas. p. 454 ff. The various abbreviations are 
 those used in Hildebrand-Gering^. 
 
 * Cf. ALL p. 58. 
 
108 
 
 (<r) Passive expressions with instruments that are nearer to the 
 notion of agency than that of means, Gf>r, III. ^,20/ borin verk- 
 jum; ^ HH. II. 37, 3 dyrkalfr dgggo slungenn ; Vgl. 4, pa vas grund 
 groin gr0num lauki ; so Vegt. 5, var ek snivin snjovi ok sleginn regni 
 ok drifin dgggu ; the concept of agency appears very clearly in Akv, 
 34, 2 / gar^d />anns skripinn vas innan ormum, * perreptatum angui- 
 bus ' ; Fornm. Sud. 70, 26 gll hgllin mun vera skipud hrcedilegum 
 ormum, to which cf . Sn. E. I. 496 frd IdH Finnum skripnu ; real 
 auctorial in Nj. 153, 52 hann var vd vinum ho f Jinn, so in Drop- 
 laugarsona Saga, 34 ; Egilss. ed. F. J6nsson, 1886, VIII. 1. 20, 
 hann vir'diz par vel huerjum manni ; similarly in Hitd, ed. Fridrik- 
 son, 1847, 4 virdist konunginum hann a/b ragd s ma'd r vihosQ mean- 
 ing is attested by Stjorn. 458 hann var virdr minnzt of peim. Nor- 
 "dimbraland var mest byggt Nordmgnnum, Fornm. I. 23, to which 
 xf. Olafssaga in Flat. I. 16, ^, er kallatt af Nordmonnum sidan 
 bygdu pat? Follows the discussion of examples from the Edda. 
 
 § 146. Vm, 25, 2 en N^tt vas Ngrvi borin is explained from the 
 connection with § 128 under Ags. acennan. Already Grimm, IV. 
 714 recognized the instrumental nature of the datives with verbs 
 denoting " das erzeugt und geboren werden, wobei freilich in uns- 
 rer sprache fast nur das part, prat., weil die passivflexion aufhort, 
 in betracht kommt." So Delbriick in ALL p. 13, and Synkretis- 
 mus, p. 173. Not so, however, others who class whatever examples 
 of this kind are known to them under the functions of the real da- 
 tive, as Dietrich in HZ. 8, 53 and Nygaard, Edd. Synt. I. 17 
 " medens man i Udtryk som nott var N'orvi borin har egentlig Hen- 
 synsbetegnelse." Right here may be enumerated all similar exam- 
 ples from the Edda, besides Vm. 25, 2. 
 
 Vm. 38, 5 ok vas at hann ^sum alinn 
 
 Alv. 29, 3 hve su n^tt heitir, en Ngrvi kenda 
 
 Rp. 43, I Upp oxu par Jarli bornir 
 
 HdL 12 pu est, OttarrI borinn Innsteini etc. 
 
 Hdl. 25,2 allir bornir J g rmunrekki 
 
 1 Cf. Dietrich, HZ. 8, 53. 
 
 2 The dative in combination with vera and the present participle to denote 
 necessity must naturally be taken as a real dative of interest (cf. Lund, p. 119 ff.; 
 Winkler, p. 463; Nor. Synt. p. 99; HZ. 8, 52). at ySr se pat vel geranda, 
 Hkr. 357, 26 ; er per pess ecki biQjanda, * derom bor du ikke bede,' * hoc tibi 
 petendum non est,' Egilss. 60, 22 ; er ongum dugandi manni er geranda, * hoc 
 nuUi homini faciendum est,' Alex. 47. 
 
 I 
 
109 
 
 Hdl, 29, I borinn Hr0rekki 
 
 Fj. 6, I hverjum estu, sveinnf of borinn ; cf. Fm, i, i 
 Vkv. 2, I Hlapgu/^r ok Hervgr Hlopve bornar ; So Hildebrand- 
 
 Regius: borin var Hlopve, and J6nsson^ : vas Hlopve borin 
 Grp. 3, 3 Sigurpr heitik, borinn Sigmundi 
 Gpr. I. 24, 3 of borinn Bupla^ brofiirminn; cf. Sg. 55, 5, and 
 
 Sg 15, 2. ^ ^ 
 Od. 10, 4 j^w vit br0prum tveim bornar vdrim 
 Hm. 2, 5 ^J hvaiti Gu/>run^ GJuka borin 
 Grt 9, 4 br0/>r bergrisa, />eim erum bornar. — Both ala and bera 
 
 are extensively used in the active in the sense of 'beget, bring 
 
 forth ' ; so 
 Hdl. 42, I olulf Loke vi/> Angrbofio 
 Hdl. 37, 3 nio bgro pann jgtna meyjar^ etc. 
 
 And yet the nature of our language is such that we are unable to 
 render them correctly not only in the passive but even in the ac- 
 tive; Cpb. is constrained to translate HH.\. i, 3 /ds hafpi Helga 
 Borghildr borit^ * H. was born of B.' — Further notes follow. 
 
 § 147. Vm. 25, 2. For Grimm cf. § 146. That, at any rate, ON. 
 linguistic consciousness did not conceive of such examples as pure 
 datives of interest, not to speak of dative-objects,^ is evidenced 
 from instances like the Flateyjarbok version of Hdl. 25, 2 allir 
 bornirfrd Jgrmunrekki. This example would indicate a later clear- 
 ing up of the I-E. confusion of the instrumental with the ablative * 
 — since we have no frd in the Edda in this sense — in the same 
 way 2&fon has become predominant in OHG. with verbs of descent.' 
 But since af is the regular Norse preposition in this use, as in hon 
 var af Most kynjud ok fcedd^ Fornm. 10, 384 ; cBttadr af Arabia^ 
 Alex. 39, Detter-Heinzel's comment upon the passage in question, 
 II. 627, is not at all improbable. " Die Construction mit '/ri * 
 statt des nackten Dativs deutet vielleicht an, dass sie nicht Sohne, 
 sondern Enkel, Urenkel Jbrmunreks waren. S. oben 8 koma frd^ 
 unten 25 uera frd. Aber koma frd wird unten 38 von directer 
 Descendenz gebraucht." 
 
 § 148. Vm. 38, 5 ^suin alinn, in Regius and Arnam. — Hilde- 
 brand's and Sijmons' insertion of mep before gsum is gratuitous. 
 
 1 Gislason, Eft Skrifter, II. 23. 
 
 2 Cf. ALT. p. 12. ff. ; and § 128 of this dissertation. 
 
 * Cf. Erdmann, Untersuchungen uber die Syntax der Sprachc Otfrids, Halle, 
 1874-6, II. 245 ff.; Graff, Die ahd. Pr'dp. 236. 
 
no 
 
 Delbriick, ALL 13 and Synkret. 173, translates the line, "er wurde 
 nicht von den Asen erzeugt " ; so Vm. 25^ " die Nacht war von Ngr 
 gezeugt." Detter-Heinzel, II. 165 " Der nackte Dativ kann im 
 Passivconstructionen statt des mit ^verbundenen stehen." Lund, 
 120, classes this example among the uses of the dat. of interest. 
 So Dietrich, LIZ. 8, 53 " weil hier zugleich ein besitz des geborenen 
 vorhanden ist " ; for Winkler, cf. Germ, Cas,^\o. Wis^n, Om ord- 
 fogningen^ 40, holds an ablatival view, somewhat like Erdmann's 
 conception^ of Ot. i, 5, 23 thu scalt beran einan . . . fatere gibo- 
 ranan ebanewigan as " von, oder besser aus dem Vater geboren als 
 ein gleichewiger." This would also be supported by Fm. 3, 2 ^ 
 hvety'u vast undri alinn. However, if the material in § 128 has 
 any weight, then, besides the ablative ^ we must surely allow an 
 equally strong instrumental possibility — a datival sense we must 
 grant a priori because of the morphological aspect of the case — ; 
 and, for the OHG., Grimm's suggestion, "schoner ware /^/^n^ " 
 for Otfrid (cf. Grimm, IV. 714) is not now quite as decisively to be 
 rejected as Erdmann would have us believe. LLdl. 19, 4 presents 
 still another verb of begetting, enn LLildi vas Llglfr of getinn, the 
 reading being that of Bugge in Ark. i, 249 ff. The active is repre- 
 sented in 42, 2 en Sleipni gat\Loki'\ vi/^ Svapilf^ra. 
 
 § 149. LLLiv. 21, 3 ef mir's alhugat. Cf. Glums. c. 4. mer mun 
 mest um hugat. For personal datives with impersonal verbs 
 Dietrich, LLZ. 8, 51 gives a rule, "Dies ist besonders der fall bei 
 gesinnungen und seelenzustanden die wir nicht machen, die iiber 
 uns, an uns kommen, so wie bei ahnlichen leiblichen zustanden, 
 dann auch namentlich bei sonstigen absichtslos und zufallig vorge- 
 nommenen handlungen, und bei den naturvorgangen in denen eine 
 uns fremde macht, ein es da \sic\ wir nicht begreifen, uns niitzlich 
 oder schadlich wird." Clearly there is a danger of subsuming too 
 much under this category. Much will depend on the nature of the 
 verb itself ; /<a!««J/ mgnnum mikit um hana, LLkr. 27, 7 may be 
 a propos, but a similar impersonal, dvaldist hdnum i peiri/erd, Flat. 
 I. 42, might be taken as a dat.-inst. of agency. So, while mental 
 states like Pykkir mer (cf. Gothic hja izwis pugkeipy Mt. 26, 66), 
 mer skilz, mer minniz and even honum skytzt, * he loses his way ' etc. 
 are surely datives of interest, they cannot be classed with instances 
 like konunginum vard litit, Fa. 3, 456 ; var'd henni opt gengid, Od. 
 
 ^ Untersuchungen, II. 245. 
 
 2 Misteli, in Zs.f. Volkerpsych. u. Sprachwiss. x. 160, scouts this alternative. 
 
Ill 
 
 3, 49 ; \>rdndi ok \>orgrimi vard mart ialat^ Flat. I. 556, 5 ; vard 
 f>eifn ok mart taiat, Laxd. 248, 24 ; honum mundi helzt misgert i 
 vera at far a at mer, \_Nj. 124, 114 ^ af ham vilde der isaer vaere 
 handlet ilde ved at,' Norr. Synt. p. 99] ; potti hdnum ser f>d skjotara 
 fyrnast lifldt Droplaugar [^ snarere glemmes af ham^'' Lund, p. 1 1 9]. 
 Dropl. 9. Of course HHv. 21, 3 is a dative of interest, but 31,3 
 hvVs per, stillir! st0kt or landi W\\\ have to be explained differently. 
 Cf. Detter-Heinzel, II. 359 ff. 
 
 § 150. BB. II. 8, 6 }7vi vas a l0gi [m^r] litt steikt etit. — Gering 
 omits the mer of Regius ; Bugge and Heinzel retain it, so Munch 
 Grundtvig and J6nsson in their several editions : mer is the logical 
 subject. But in 8, 5 the Cod. Reg. er sagt m^ cannot be taken as 
 mer, because that is not the required sense of the passage, despite 
 the Copenhagen edition and that of Hagen. Hildebrand's m&r is 
 much better. 
 
 § 151. BB. II. 18, I esat }?er at gUu, alvitr! gefit. Gering's 
 rendering of this line, Wb. s.v., " dir ist nicht in jeder beziehung 
 gliick beschieden gewesen " is correct only in case the above read- 
 ing is accepted, and even then uncommon for the Edda. But the 
 get of the Codex Regius is the usual contraction not for gefip as the 
 MS. would imply, but for getit} If this reading is allowed to stand, 
 then per must be construed as a dative-agent. The sense of the 
 passage in this form is entirely satisfactory : The situation is that 
 of the valkyrja Sigriin who is alvitr and aids Helgi in his battles, 
 coming to meet him after the carnage in which he could not help slay- 
 ing his antagonists, Signin's father and brother. He receives her 
 with such words, "All turned out well, hut you did not get everything 
 your way, all-wise tho you are, I have killed your father and brother ; 
 but then you could not prevent it, — some of this was Fate's doing." 
 This interpretation of the lines is borne out by the sense of 20, 3 
 vanntat vigi, vas per vg skgpup, for which cf. Detter-Heinzel, II. 
 374, and has the advantage of a correct reading without the neces- 
 sity of an unusual translation. 
 
 § 152. Fm. 21, I raj>'s J»6r ra]7it. So Regius; Grimm's (ed. 
 18 1 5) and Ettmiiller's reading of mer for per"^ which Cpb. follows 
 
 1 Cf. Hildebrand-Gering^, p. 263 fn. and Cpb. notes ad loc. Wimmer og 
 Jonsson's note to the MS., on p. 147 of their phototype edition of the Edda: 
 " 1. 25, gepip'] rettet fra /; stregen over p synes at vise, at der, saledes som ogsa 
 Bugge formoder, s. 196, forst har vseret skre-vet gel y igeHi." 
 
 2 In Germania, 17, 12. 
 
112 
 
 is unwarranted and unsatisfactory : it is SigurJ)r that speaks these 
 words to Fdfnir, who in verse 20 has been telling him raj>k per nu, 
 Sigvgrpr, en pu rap nemir. Hence Nygaard ^ is entirely correct in 
 interpreting per as ' of dig. ' 
 
 § 153. The following two examples are of interest because they 
 might give the impression of containing analytic substitutes for the 
 expression of agency in the Edda. Gpr. II. 4, 4 gll vgru sgpuldyr 
 sveita stokkin ok of vanip vast und veggndum. Cod. Reg. has 
 of instead of und^ adopted by Bugge, Grundtvig etc. as against 
 Rask, Copenhagen ed., Munch and Ettmuller's af ; cf. Detter- 
 Heinzel, II. 493. But af to denote agency, tho common later,^ as 
 Flat. I, 69, 14 pa var Gudefridus drepinn af sinum monnum, is un- 
 known to the Eddie poems. The only other example to be con- 
 sidered is Gpr. II. 34, 2 pann munk kjosa af konungum ok po af 
 nipjum naupug hafa, but since naupug is an adjective, the relation 
 is local, * tho coerced on the part of my relatives,' or, if differently 
 interpreted, causal, ' him will I then choose among kings and have, 
 tho constrained, simply because of my relatives ; ' for the latter 
 cf. also Ifdi. 43, 3 varpr Loptr kvipogr af kono Hire, where we 
 have an adjective with af. 
 
 § 154. Ghv. 10, 2 vask Jjrimr verum vegin at husi, 'I was 
 led home by three men,' may well be another instance of a dat.- 
 inst. of agency. The Gothic gas andjan sik, 'to be led,' takes fram 
 in this use, as in 2 Cor. i, 16. Even Winkler says, p. 476, '' eigen- 
 tumlich ist der instrumental des mittels, falls nicht der dativ der 
 beteiligung vorliegt, mit dem ausdruck des personlichen agens z. 
 b. in der folgenden stelle ; var ek primr verum vegin at husi = von 
 drei mdnnern heimgefuhrt.^^ 
 
 § 155. Bm. 7, 2 b0kr V0ru inar J?enar blahvitu valundum 
 rol^nar, flutu i vers dreyra. Such is Hildebrand-Gering's reading, 
 against Cod. Reg. ofnar vglundum, apparently because of the sup- 
 port of Ghv. 4, 3 d0kr vgru pinar enar blahvitu ropnar i vers 
 dreyra^ folgnari valblopi, the sense being 'thy blue-white bed- 
 covers were reddened by the deadly wound, they swam in the 
 blood of thy husband.' Cpb. even substitutes folgnar for flutu, 
 ' bathed in his blood.' Because of the fact that ' *valund, f. todes- 
 wunde ' ^ is a aTra^ Acyo/xcvov, it were best to return to the reading 
 
 1 Norran Syntax, p. 99. 
 
 2 Probably under Latin influence, cf. Falk og Thorp, p. 170. 
 
 3 Cf. Gering, Worterbuch, s.v. 
 
113 
 
 of the Codex, accepted by Hagen, Rask, Copenhagen ed., Munch, 
 and Detter-Heinzel. So Lund, p. 119 ' vaevede af Kunstnere,' fol- 
 lowed by NygB.rd, ££^dasj>r. Synt. p. 17; ' ab artificibus contexta ' 
 in Copenhagen ed. Cf. Detter-Heinzel, II. 576 " Deine von 
 Kunsilerinnen gewebten Betttiicher wurden von dem Blute deines 
 Mannes besprenget." For vglundum cf. Fritzner, Ordbok, and 
 Sturl. I. 278, ])tdr. 69, 82. 
 
 § 156. Such are the examples from the Eddie poems. In the 
 light of the testimony of Gothic and Anglo-Saxon as to the exist- 
 ence of a dative-instrumental of personal agency, they cannot be 
 mere accidents. "Anomalien im satzbau — says Dietrich in HZ. 
 13, 124 in another connection — konnen, so lange sie nur verein- 
 zelt in einem alten texte vorliegen, desto mehr zum zweifel an der 
 iiberlieferung oder ihrer auffassung, beziehungsweise bei einem 
 schriftsteller zum corrigieren veranlassen, je mehr sie gegen die 
 gewohnliche logik moderner oder alten sprachen verstossen. So- 
 bald eine solche erscheinung indes durch noch mehrmaliges vor- 
 kommen in demselben dialecte oder durch auftreten auch in einem 
 verschiedenen dialecte stiitzen empfangt, muss sich der zweifel in 
 forschen nach dem grund der abweichung verwandeln." This 
 might be taken as a precise statement of the case with respect to 
 the Germanic dialects. The dative and the instrumental, in their 
 respective functions, were found to be ill-defined and confused. 
 This was especially true of them in the domain of agency. The 
 datives that denoted the personal agent with passive verbs could 
 not all be reduced to the basic concept of a dativus commodi ; in 
 the case of a considerable number sufficient reason was produced 
 for a dissent from their older interpretation and in favor of their 
 classification as original instrumentals. The collateral testimony 
 of other Indo-European languages lent added weight to these find- 
 ings in confirming the development of the Germanic sociative-comi- 
 tative into the dative-instrumental of agency. 
 
CONCLUSION 
 
 § 157. The deductions gained from the preceding chapters may 
 be summarized as follows : 
 
 {a) The Indo-European languages express personal agency, i.e. 
 the logical subject, with passive verbs either by means of an un- 
 aided case-form or thru a prepositional phrase. As a rule the 
 prepositional locution is the direct product of later analytic ten- 
 dencies, whereas the prepositionless devices are resorted to in the 
 older or synthetic stages of a language. The Indo-European lan- 
 guages may be said to employ four case-forms to denote the agent 
 of a passive action: the instrumental (Slavic, § 53), the dative 
 (Latin, § 38), the genitive (Lithuanian, § 56) and the ablative 
 (Armenian).^ Of these the instrumental-agent is really a personal- 
 ized instrumental of means ; the dative-agent in reality designates 
 personal interest ; the ablative emphasizes the idea of separation ; 
 finally, the genitive of agency is but an adnominal genitive of 
 possession. 
 
 (b) Judging by the combined testimony of Latin, Slavic, Sanskrit 
 and Avestan, § 37, the instrumental and the dative case-forms met 
 in the function of agency already in the Indo-European Ursprache : 
 the perfect (passive) participle, namely, could be connected with 
 both an instrumental and a dative of agency. Whilst, however, we 
 may speak of a competition in that domain between the two cases, 
 this rivalry is not to be thought of as one of complete synonymity. 
 Both could represent the logical subject but, at the same time, each 
 stood for a different type of this subject. The instrumental ex- 
 pressed the agent as an outer, moving, and directive force, with 
 whose cooperation the action took place. The instrumental-agent, 
 in the last analysis, was a comitative agent, §§ 20, 21. The 
 dative, on the other hand, developed the function of agency from 
 its basic signification of personal interest; it is not merely the 
 author, Urheber, of the action, but a partaker that is vitally in- 
 terested in the outcome as one that may be of benefit or disadvan- 
 
 1 Cf. Lauer's Grammatik d. classiscken Armenischen Sprache, Wien, 1869, 
 p. 86, and K. H. Gulian's Elementary Modern Armenian Grammar^ 1902, p. 72. 
 
 114 
 
115 
 
 tage to him. Consequently, the dativus auctoris is essentially a 
 dativus commodi vel incommodi, § i6. The statement of H. C. v. d. 
 Gabelentz^ that in some of the non-Indo-European languages as 
 well, notably in Manchu, Mongolian and Japanese, the relation of 
 the nomen agentis with the passive can be symbolized by means of a 
 dative form, throws an interesting light upon the similar evolution 
 of the Indo-European dative. 
 
 (c) Indo-Iranian is the sole linguistic group that has preserved 
 this simultaneous employment of the two cases. Both Sanskrit and 
 Avestan attest an instrumental of agency not merely with a perfect 
 passive participle, but with verbal adjectives and finite passive 
 verbs as well, §§ 68, 71. Its combination with the last category 
 cannot be declared * Indo-European, because the formation of 
 passive verbs belongs to the period consequent upon the " dialectal 
 scission," §§ 4, 5. The Indo-Iranian usage, however, demonstrates 
 that the instrumental of the personal agent is the direct heir of an 
 original sociative-comitative force. The various stages of its descent 
 are clearly visible, since thru the comitant denoting military asso- 
 ciations, conveyances etc. one can logically arrive at the instrumen- 
 tal of means, pure and simple. To the apphcation of the latter 
 there is no limitation : persons also can fimction as the means or 
 instrument of the action. The instrumental-agent with finite pas- 
 sive verbs may thus have two immediate predecessors, §§ 64*?, 
 68 e. (a) the use of the instrumental of agency with the past parti- 
 ciple, developed from the instrumental of personal means, as if 
 * per interpretem (inst.) dixit ' > ' dictum per interpretem ' > * dic- 
 tum ab interprete ' (inst.) > * dicitur ab interprete ' (inst.) ; (/8) the 
 instrumental of means with finite passive verbs, as * he praises 
 with words ' > * thou art praised with words ' > ' thou art praised by 
 your friends.' Because of the late creation of passives, the process 
 described under (a) must be held chronologically prior to the other. 
 
 {d) The dative-agent of Indo-Iranian derives its auctorial force 
 purely thru its context ; its grammatical form postulates only the 
 concept of benefit or detriment to some participant in the action. 
 This overwhelmingly personal nature of the dative-agents in the 
 earlier stages of Sanskrit and Avestan is shown by the fact that 
 these agents are predominatingly of the pronominal type ; ^ the evi- 
 dence, however, of both the Rigveda and the GaOas is at hand to 
 
 ^ ijber das Passivum, p. 541. 
 
 2 Cf. Havers, Untersuchungen^ pp. 10, 14, 44, 60. 
 
116 
 
 prove that this personal agent was capable of attaching itself not 
 only to the *I-E. past participle, but in turn to the verbals and 
 finite passives as well — an evolution of the same sort as that which 
 took place in Latin. 
 
 (e) Whilst Slavic offers the example of a language that has only 
 an instrumental of agency and can demonstrate all the stages of its 
 probable provenience, § 54, Latin restricts itself to the dative as the 
 prepositionless logical subject. In the beginnings of the language 
 this Latin dative, of a strongly pronominal character,^ is found 
 combined only with the past participle or its compounds and the 
 verbals, denoting /^r whom the action is an accomplished fact, or 
 in whose interests it must take place, § 46. Its combination with 
 the participle must, of course, have preceded its use with the 
 gerundive, because the latter is a specifically Italic formation. 
 The agency idea, synonymous with that denoted by ab with the 
 ablative, which develops after Cicero, and especially in the usage 
 of the Augustan writers and attaches itself not only to pronouns 
 and verbals, but to substantives and finite passive forms, is a 
 Graecism in that the native tendency of the dativus commodi to 
 rise to the rank of a pure dativus auctoris, devoid of all personal 
 interest, was assisted by the homogeneous construction of the 
 Greek, § 51. 
 
 (/) Greek and Germanic, the two syncretistic languages par 
 excellence, present a complication in connection with what may be 
 called the dative-instrumental of agency, i.e. a case-form which 
 combines the two separate ideas represented by the Sanskrit instru- 
 mental and dative of agency. The existence of such a composite 
 force, as distinct from the view formerly held of these ' datives ' 
 with verbals and passive verbs, must be postulated because of the 
 fact that both the Greek and Germanic dative represent an agglom- 
 eration of syntactic uses, among them the function of the instru- 
 mental as well as that of the dative, § 73. 
 
 {g) The interpretation of all the Greek dative-agents as original 
 datives in force cannot, in the last analysis, be called false, both 
 because it harmonizes with the Grundbegriff of the dative and be- 
 cause the history of the dativus auctoris in Latin offers a tangible 
 proof of a similar possibility in the other languages. In fact, the 
 personal datives with the verbals -rds and -rcos lend themselves 
 
 * Cf. Havers, op. cit. p. 237, 188. 
 
117 
 
 quite well to such an explanation, especially with -rcos, because 
 the morphologically demonstrable similarity of that suffix with other 
 verbal adjectives tends to bring its dative in line with the dative- 
 agent of predicate infinitives in other Indo-European languages as 
 well, §§ 84-87. But in the instance of finite passive verbs, at least, 
 the partially instrumental force of the dative-agent must be insisted 
 upon because of the ease with which the entire series of the transi- 
 tional processes leading away from the dative-instrumental of asso- 
 ciation can be reproduced in earliest Greek, § 89. The argument 
 in favor of a new interpretation of these datives is one of analogy 
 and evolution. The instrumental of association is the fons et origo 
 of all the functional types of the instrumental ; the instrumental of 
 subjective or personal means is one of its descendants in the same 
 logical relation as is that of the objective or material means ; in 
 Sanskrit, Avestan and Slavic, being in a position to demonstrate 
 the whole genealogical relation, we are permitted to acclaim the 
 inst. of personal agency as the last and highest typed expression 
 of the inst. of concomitation ; there is no valid reason, therefore, 
 why, in full possession of similar facts in Greek, from the inst. of 
 association onward, not only to that of material means, but also to 
 that highly personified stage which Brugmann calls an active 
 potency {Grdr? § 479, p. 527), we should upon the presentation 
 of examples of personal agency, the logical fruition of all preced- 
 ing developments, stamp their dative-agents as datives of personal 
 interest, when it is well known that the so-called ' dative ' in Greek 
 represents an instrumental force in an equal measure to its own. 
 Nor should it detract from the logical cogency of the processes in 
 question that satisfactory explanation of these agents can be fur- 
 nished upon a purely datival basis because, on the one hand, the 
 linguistic consciousness of Old Greek could hardly have failed care- 
 fully to distinguish between the instrumental and the datival con- 
 stituents of the various functions of those case-forms which after 
 the syncretism of the two, § 26 if., came to be called * dative ' ; nor, 
 on the other hand, could the dative of agency have so thoroly 
 assumed the auctorial functions of the instrumental if, during the 
 period prior to their amalgamation, it did not widen its basic signi- 
 fication from within on the analogy of the competing instrumental 
 of agency, as in the instance of Latin dat. auctoris^ § 51- 
 
 (Ji) In the Germanic dialects the general situation with respect 
 to the dative-instrumental of agency is exactly the same. To our 
 
118 
 
 modern Sprachgefithl every originally instrumental form is over- 
 shadowed by a datival connotation ; it cannot be doubted, however, 
 that tho one form served at the same time for several functional 
 types, these types were well differentiated in the consciousness of 
 the speaker.^ And since Gothic, Anglo-Saxon and Old Norse both 
 offer dative-agents that may not be classed under the traditional 
 idea of personal interest and admit of the derivation of personal 
 agents, thru the media of material means and personified agents, 
 from an original sociative instrumental, §§ 122, 126, 145, it would 
 seem but logical and perfectly in accord with analogous develop- 
 ments in related languages for the Germanic also to extend the 
 instrumental of means to apply to persons as well. In the field of 
 reconstructive philology this strong evidence of the possibility of 
 such a construction must be regarded as tantamount to a proof of 
 its actual existence. Moreover, if the argument — frequently under- 
 lying the interpretation of these ' datives ' in decisive instances — 
 that the Germanic past participle is in most cases an adjective (cf. 
 § 115 for Gothic) be tenable, then we could not speak of a Ger- 
 manic passive at all and the similar periphrasis of Avestan and 
 Slavic, too, would mean the complete inability of these languages 
 to express passive verbal relations. And yet we have in related 
 tongues evidence of the bare participle, without the copula so essen- 
 tial in Germanic, combined with instrumental forms in instrumental 
 function to denote the logical subject. So in Sanskrit yamena dat- 
 tdh, kvestdin frazinte anasita maeOanyd; but especially in Slavic 
 with the bulk of its passives expressed solely by the participle (cf. 
 Miklosich, IV. 704 ; Vondrdk, II. 349) we have not only locutions 
 like trhst\> vetrom'h dvizema, vwo dvifxov, Lk. 7, 24, but also pravim-b 
 dhvema aggeloma, * qui a duobus angelis ducitur,' Sup. 124, 26, so 
 reminiscent of ON. vask f>rimr verum vegin, Ghv. 10, 2 ; ne vidim-h 
 nikymhze, ' qui a nemine videtur ' ; mnogymi ljud\>mi cesten-h i slaven-h, 
 * qui a multis hominibus colitur et celebratur,' — expressions which, 
 under the present system of interpretation would, if paralleled in 
 the Germanic by a dative-inst. of agency, be rendered as ' zwei En- 
 geln gefuhrt,'' ^ niemandem sichtbar,^ and ^ vielen geehrt und gefeiert.'' 
 While the Gesinnungsweise of one language cannot simply be 
 saddled upon another, and each must be conceived of as capable 
 of its own uses and interpretations, attention must be drawn to this 
 
 1 Cf. Synkretismusy p. 167. 
 
119 
 
 fundamental similarity between the Slavic and Germanic passive 
 expressions and to the danger of the too subjective application of 
 that theory which would intensify the Germanic dative, regardless 
 of its syncretistic bearings, by too much '• InnerlichkeW (cf. also 
 SynkretismuSy p. 237 f.) and would isolate it from the testimony of 
 the related languages. Despite the negative or, at least, non-com- 
 mittal character of the criticisms thus far made in that regard, I 
 w^ould still fain hark back to Delbriick's opinion expressed away 
 back in 1867 {ALL p. 73), " Um die deutsche casuslehre durch- 
 sichtig zu machen, bedarf es vor allem einer vergleichung mit der 
 litauischen und slavischen syntax." 
 
BIBLIOGRAPHY 
 
 Only the principal works used in the preparation of this dissertation are listed here. 
 Periodicals are omitted ; for the abbreviations of their titles cf. Jahresbericht iiber die 
 Erschcinungen aufdem Gebiete der germanischen Philologie, and similar publications. 
 
 GENERAL THEORY 
 
 Gabelentz, G. v. d. : Die Sprachwissenschaft. 2d ed. Leipzig, 1901. 
 Gabelentz, H. L. v. d. : Uber das Passivum. Vol. 3, 449 of Abhand. d. 
 
 phil.-hist. CI. d. Kgl. sachs. Ges. d. Wiss. Leipzig, 1861. 
 Holzweissig, Fr. : Wahrheit und Irrthum der localistischen Casustheorie. 
 
 Leipzig, 1877. 
 HUbschmann, H. : Zur Casuslehre. Erster Theil. MUnchen, 1875. 
 Rumpel, Th. : Die Kasuslehre. Halle, 1845. 
 Winkler, H. : Zur Sprachgeschichte. Berlin, 1887. 
 
 INDO-EUROPEAN 
 
 Audouin, E. : De la ddclinaison dans les langues indo-europdennes. Paris, 
 
 1898. 
 Bopp, F. : Vergleichende Grammatik. {Vgl. Gr.) 3d ed. 3 V. Berlin, 
 
 1868-71. 
 Brugmann, K. : Kurze vergleichende Grammatik der indogermanischen 
 
 Sprachen. (^Krz, Vgl. Gr.) 3 v. Strassburg, 1902-04. 
 
 Grundriss der vergleichenden Grammatik der indogermanischen 
 
 Sprachen. Zweite Bearbeitung. {Grdr.^ or Grundriss^.') 2 v. 
 Strassburg, 1897-1911. Sometimes Grdr^ is also cited. 
 
 DelbrUck, B. : Ablativ Localis Instrumentalis im Altindischen Lateini- 
 schen Griechischen und Deutschen. (^ALI.) Berlin, 1867. 
 
 Vergleichende Syntax der indogermanischen Sprachen. {Vgl. Synt.^ 
 
 3 V. Strassburg, 1893- 1900. 
 
 Giles, P. : A short manual of comparative philology. London, 1901. 
 Havers, Wm. : Untersuchungen zur Kasussyntax der indogermanischen 
 
 Sprachen. Strassburg, 191 1. 
 Jolly, J. : Geschichte des Infinitivs im Indogermanischen. (Jnfinitiv.) 
 
 MUnchen, 1873. 
 Schleicher, A. : Compendium der vgl. Grammatik der indogermanischen 
 
 Sprachen. {Compendiufn^.^ 2d ed. 1866 cited. 
 
 120 
 
121 
 
 SANSKRIT 
 
 Aufrecht, Th. : Die Hymnen des Rigveda. 2d ed. 2 v. Bonn, 1897. 
 Delbriick, B. : Altindische Syntax. Halle, 1888. Vol. 5 of Syntaktische 
 
 Forschungen. (S./*". or SynL Forsch. V.) 
 Gaedicke, C. : Der Akkusativ im Veda. Breslau, 1880. 
 Grassmann, H. : Worterbuch zum Rig-Veda. Leipzig, 1873. 
 Oldenberg, H. : Rgveda. Textkritische und exegetische Noten. I. -VI. 
 
 Abh. d. Kgl. Ges. d. Wiss. Gottingen, phil.-hist. CI. N. F. XI. 5. 
 Speyer, J. S. : Vedische und Sanskrit Syntax. (VuSS.) Strassburg, 
 
 1896. 
 Thumb, Handbuch des Sanskrit, I. : Grammatik. Heidelberg, 1905. 
 Wenzel, H. : Uber den Instrumentalis im Rigveda. Tubingen, 1879. 
 
 AVESTAN 
 
 Bartholomae, Chr. : Altiranisches Worterbuch. (IVd.) Strassburg, 1904. 
 HUbschmann, H. : Zur Casuslehre. Zweiter Theil. Mlinchen, 1875. 
 Reichelt, H. : Awestisches Elementarbuch. Heidelberg, 1909. 
 Spiegel, F. : Vergleichende Grammatik der altdr^nischen Sprachen. Leip- 
 zig, 1882. 
 
 BALTO-SLAVIC 
 
 Kurschat, F. : Grammatik der littauischen Sprache. Halle, 1883. 
 Leskien, A. : Handbuch der altbulgarischen Sprache. 4 Aufl. Weimar, 
 
 1905. 
 Leskien-Brugman : Litauische VolksliederundMarchen. Strassburg, 1882. 
 Miklosich, F. : Vergleichende Grammatik der slavischen Sprachen. (^/. 
 
 Gr.) IV. Syntax. 2d ed. Wien, 1883. 
 Schleicher, A. : Litauische Grammatik, und Lesebuch. Prag, 1856. 
 Vondrdk, W. : Vergleichende slavische Grammatik. II. Gottingen, 1908. 
 
 GREEK 
 
 Bnigmann, K. : Griechische Grammatik. {Griech. Gr^) 3d ed. MUn- 
 
 chen, 1900. 
 Curtius, G. : Griechische Schulgrammatik. 5th ed. Prag, 1862. 
 
 Erlauterungen zu meiner gr. Sch. 2d ed. Prag, 1870. 
 
 Delbriick, B. : Die Grundlagen der griechischen Syntax. Halle, 1879. 
 
 Vol. IV. of Syntaktische Forschungen. (S.F. or Synt. Forsch. IV.) 
 Green, S. G. : Handbook to the grammar of the Greek Testament. New 
 
 York, 1904. 
 Helbing, R. : Uber den Gebrauch des echten und sociativen Dativs bei 
 
 Herodot. {EuSDat.) Freiburg, 1898. 
 
122 
 
 Hirt, H. : Handbuch der griechischen Laut- und Formenlehre. Heidel- 
 berg, 1902. 
 
 Kuhner-Gerth : Ausfiihrliche Grammatik der griechischen Sprache. 3d ed. 
 2d part. Leipzig, 1898-1904. 
 
 Monro, D. B. : A grammar of the Homeric dialect. 2d ed. Oxford, 1 891. 
 
 Vogrinz, G. : Grammatik des homerischen Dialectes. Paderborn, 1889. 
 
 Walther, E. : De dativi instrumentalis usu Homerico. Breslau, 1874. 
 
 LATIN 
 
 Brenous, J. : Etude sur les Helldnismes dans la syntaxe latine. Paris, 1895. 
 
 Draeger, A. : Historische Syntax der lat. Sprache. 2d ed. 2 v. Leipzig, 
 1878-81. 
 
 Frobenius, R. : Die Syntax des Ennius. Nordlingen, 1910. 
 
 Haase, F. : Vorlesungen uber lat. Sprachwissenschaft. Vol. 2. Leipzig, 
 1880. 
 
 Holtze, F. W. : Syntaxis priscorum scriptorum Latinorum usque ad 
 Terentium. 2 v. Lipsiae, 1861-62. 
 
 Landgraf, G. : Beitrage zur hist. Syntax der lat. Sprache. Munchen, 1899. 
 
 Reisig, C. K. : Vorlesungen liber lat. Sprachwissenschaft. 3d vol. : Syn- 
 tax. Beriin, 1888. 
 
 Schmalz, J. H. : Lateinische Syntax und Stilistik. 3d ed. Munchen, 
 1900. 
 
 Soramer, J. F. : Handbuch der lateinischen Laut- und Formenlehre. Hei- 
 delberg, 1902. 
 
 Tillmann, H. : De dativo verbis passivis linguae Latinae subiecto, qui 
 vocatur Graecus. Act. Sem. Erl. IL (1881) 71. 
 
 GERMANIC 
 
 Delbrlick, B. : Synkretismus. Ein Beitrag zur germanischen Kasuslehre. 
 
 {Synkretismus .) Strassburg, 1907. 
 Dieter, F. : Laut- und Formenlehre der altgermanischen Dialekte. Leipzig, 
 
 1900. 
 Erdmann-Mensing : Grundzuge der deutschen Syntax. Stuttgart, 1886-98. 
 Grimm, J. : Deutsche Grammatik. ist ed. IV. 1837. 
 Streitberg, W. : Urgermanische Grammatik. (C/. G.) Heidelberg, 1896. 
 Wilmanns, W. : Deutsche Grammatik. Strassburg, 1893-96. 
 Winkler, H.: Germanische Kasussyntax. {Germ. Cas.) I. Der Dativ. 
 
 Berlin, 1896. 
 
 GOTHIC 
 
 Bernhardt, E. : Kurzgefasste gotische Grammatik. Halle, 1885. 
 Gabelentz, H. C, und Lobe, J. : Ulfilas. 2 v. Leipzig, 1843-46. II. 
 Pars posterior grammaticam linguae Gothicae continens. (G-L.) 
 
123 
 
 van der Meer, M. J. : Gotische Casussyntaxis. I. Leiden, 1901. 
 Streitberg, W. : Die gotische Bibel. 2 parts. Heidelberg, 1908-10. 
 Gotisches Elementarbuch. 4th ed. {Got. El^), Heidelberg, 1910. 
 
 ANGLO-SAXON 
 
 DethlofF, R. : Die Darstellung der Syntax im ags. Gedicht " Daniel." 
 
 Rostock, 1907. 
 Heyne-Socin-Schiicking {HS.^ : Beowulf. loth ed. Paderborn, 1913. 
 Hofer, O. : Der synt. Gebrauch des Dativs und Instrumentalis in den 
 
 Caedmon beigelegten Dichtungen. Halle, 1884. Also Anglia 7, 355. 
 Kress, J. : Uber den Gebrauch des Inst, in der ags. Poesie. Marburg, 
 
 1864. 
 Nader, E. : Dativ und Instrumental im Beowulf. Wien, 1883. 
 Wulcker-Grein-Assmann : Bibliothek der ags. Poesie. 3 V. 1883-98. 
 Zupitza: Beowulf. Autotypes of the Unique Cotton MS. Vitellius A. 
 
 XV. in the British Museum. EETS. London, 1882. 
 
 OLD NORSE 
 
 Detter, F., und Heinzel, R. : Saemundar Edda. 2 v. Leipzig, 1903. 
 Fritzner, J. : Ordbog over det gamle norske sprog. 2. udg. 3 v. Kris- 
 
 tiania, 1883-96. 
 Gering, H. : Vollstandiges Worterbuch zu den Liedem der Edda. Halle, 
 
 1903. 
 Gislason, K. : Efterladte skrifter. 2 v. Kebenhavn, 1895-97. 
 Hildebrand-Gering. Die Lieder der alteren Edda. 3d ed. Paderborn, 
 
 1912. 
 Lund, G. F. V. : Oldnordisk Ordfajningslaere. Kebenhavn, 1862. 
 Nygaard, M. : Eddasprogets Syntax. 2 pts. Bergen, 1865-67. 
 
 Norr0n Syntax. Kristiania, 1905. 
 
 Sijmons, B. : Die Lieder der Edda. Halle, 1906. 
 
 Vigfusson, G., and Powell, F. Y. : Corpus poeticum boreale. 2 v. (Cpb.) 
 
 Oxford, 1883. 
 Wimmer og J6nsson : Codex Regius af den aeldre Edda, i fototypisk og 
 
 diplomatisk gengivelse. Kabenhavn, 1891. 
 Wisdn, Th. : Om ordfogningen i den aldre Eddan. Lund, 1865. 
 
VITA 
 
 Cassoviae natus sum, Superioris in oppido Hungariae, die un- 
 decimo mensis Februarii anno MDCCCLXXXVIII, patre Armi- 
 nio Gruenn, matre loanna e gente Englaender. Primis litterarum 
 dementis in ludo imbutus, Praemonstratensium Canonicormn gym- 
 nasium aliaque per annos sex frequentavi. Deinde, maturitatis 
 testimonio nondum instructus, anno MCMIII in Americam septen- 
 trionalem emigravi receptusque a Collegio Urbis Novi Eboraci 
 septem post annos Baccalaurei Artibus in liberalibus gradum 
 cum honore sum assecutus. Socius ibidem ad linguam Latinam 
 docendam factus, in Universitate quoque Columbiae in philologiae 
 studium linguarumque antiquarum necnon Germanicanmi per 
 annos tres incubui. Anno MCMXI et ad gradum Magistri in 
 Artibus sum provectus et Universitatis in philologia Germanica 
 discipulus nominatus. Duobus post annis ad Universitatem lUi- 
 noisensem sum vocatus, ubi nunc lectiones Germanicas habeo. 
 
 Audivi inter alios w. ill. W. H. Carpenter, Calvinum Thomas, 
 A. F. J. Remy et R. Tombo in litteris linguisque Germanicis, N. 
 G. McCrea et J. C. Egbert in litteris Latinis atque titulis, A. V. W. 
 Jackson in Indo-Iranicis, J. L. Gerig in Celticis, deinde J. Schick 
 Monacensem in rebus philologicis. Omnibus quibus praeceptori- 
 bus, nee minus illis, qui antea in gymnasio collegioque me consilio 
 et auctoritate semper adiuverunt, — quorum in numero de me 
 optime meriti Patricius Stuhlmann Praemonstratensis atque Caro- 
 lus G. Herbermann Neo-Eboracensis hie mihi imprimis commem- 
 orandi videntur, — sincere animo gratias ago quam maximas. 
 Thomas autem praecipue atque Remy, qui studia mea inde a 
 principio insigni erga me benevolentia atque cura sunt moderati, 
 pia grataque memoria me semper persecuturum esse poUiceor. 
 
 124 
 
i 
 
RETURN CIRCULATION DEPARTMENT ' 
 
 TO— ► 202 Main Library 
 
 LOAN PERIOD 1 
 - HOME USE 
 
 2 ; 
 
 3 
 
 4 
 
 5 < 
 
 b 
 
 ALL BOOKS MAY BE RECALLED AFTER 7 DAYS 
 
 Renewals and Recharges may be made 4 days prior to the due date. 
 
 Books may be Renewed by calling 642-3405. 
 
 DUE AS STAMPED BELOW 
 
 FPRi9 R loni 
 
 
 
 •^•^Wft^jSaixi 
 
 '91 
 
 
 AOifl Disc i-ViK i 
 
 ^'91 
 
 
 JAN 06 It 
 
 9? 
 
 1 
 
 
 *' , # 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 UNIVERSITY OF CALIFORNIA, BERKELEY 
 FORM NO. DD6 BERKELEY, CA 94720 
 
liiiH 
 
 <^D0fl377Sflfl 
 
m 
 
 ,.>^4.l