P D 23Í3 64 UC-NRLF B M SD7 4fiö THE ORDER OF WORDS IN OLD NORSE PROSE WITH OCCASIONAL REFERENCES TO THE OTHER GERMANIC DIALECTS BY LUDWIG BERNSTEIN, A.M. SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN THE FACULTY OF PHILOSOPHY, COLUMBIA UNIVERSITY ■CF N / THE ORDER OF WORDS IN OLD NORSE PROSE WITH OCCASIONAL REFERENCES TO THE OTHER GERMANIC DIALECTS BY LUDWIG BERNSTEIN, A.M. »1 SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN THE FACULTY OF PHILOSOPHY, COLUMBIA UNIVERSITY Ube "Knicbecbociter pcese, Dew IQocfc V ■9 TO THE FOLLOWING GENTLEMEN : Prof. WILLIAM HENRY CARPENTER, Ph.D. Prof. A. V. WILLIAMS JACKSON, A.M., Ph.D. Prof. RICHARD J. H. GOTTHEIL, Ph.D. Prof. EDWIN R. A. SELIGMAN, A.M., Ph.D. this thesis is inscribed IN gratitude and esteem BY The Author m33G()39 'r'o I 'V '^ TABLE OF CONTENTS. Introduction . . . . . . - . I. INDEPENDENT DECLARATIVE CLAUSES A. Normal Order ..... ( 1 ) Subject and Predicate .... (a) Subject and Predicate verb (b) Subject, Auxiliary verb and Predicate Nomen (c) Subject and Compound Predicate . (2) Object (3) Adverb (Negation) B. Inversion Proper ...... (i) After an Adverb, Adverbial Phrase, Adverbial Clause ...... (2) After an Object, Logical Object and Noun Clause (3) After Predicate Substantive, Adjective, Infinitive and Participle ..... C. Rhetorical Inversion ..... II. INDEPENDENT INTERROGATIVE CLAUSES III. INDEPENDENT IMPERATIVE CLAUSES A. Of the First Person B. Of the Second Person C. Of the Third Person IV. CO-ORDINATE CLAUSES A. ''ok" .... B. "enn" .... C.">ó" .... D. "heldr" .... PAGB vii 1-32 I-I9 I-IO 1-4 4-7 7-10 10-12 12-19 19-31 22-27 27-29 29-31 3^-32 32-34 34-35 34 34-35 35 35-39 35-38 38 39 39 VI TABLE OF CONTENTS. E. " eigi at eins . . . F. "bæði ... ok" G. "hvárki . . . né " heldr" V. DEPENDENT CLAUSES A. Dependent Normal and Inverted Order . (i) Substantive Clauses .... (2) Adjective Clauses .... (3) Adverbial Clauses and list for dependent sion ...... B. Tratisposed Order ..... (i) Substantive Clauses .... (2) Adjective Clauses .... (3) Adverbial Clauses .... Author's Biographical Note ( Vita^ inver- 39 39 39 39-61 40-54 43-46 46 47-54 54-61 56 56-60 60-61 62 INTRODUCTION. Of all the Teutonic dialects, only the Anglo-Saxon and the Old Low German have been examined thoroughly from the standpoint of word-order, the Anglo-Saxon in elaborate treatises by Kube,' Smith," and Gorrell,* the Old Low German in an excellent dis- sertation by Ries/ Of the other great Germanic dialects, neither Old High German nor Old Norse is adequately represented, although Erdmann in his Otfried Syntax, as well as Holthausen and Kahle in their Icelandic grammars, could not help touching the subject slightly, while Mogk ° has discussed briefly the Scandi- navian inversion after the co-ordinate particle " ok." As regards the Gothic, I would say that an attempt on my part to investigate its order of words has led me to believe that, on account of its being almost a word-for-word translation, it is a dangerous thing to found upon it principles and theories. On the other hand, the occasional deviations of the Gothic from the Greek concern minor points and will be embodied in a separate paper. It is hardly necessary to point out that the order of words of a language reveals its genius to no inconsiderable degree and thus constitutes an important chapter in its syntax. Due attention has been called to it by eminent scholars, not only in the line of Germanics, but also in Sanskrit, Greek, Latin, and the Romance languages. It is strange that the Slavic dialects, with the excep- ' Emil Kube : Die Wortstelluug in der Sachsenchronik, Jena, 1886. In- ferior in merit to : ' C. Alfonso Smith's lucid treatment of "The Order of Words in Anglo- Saxon Prose," in the Publications of the Modern Language Association, 1893. ^ J. H. Gorrell : " Indirect Discourse in Anglo-Saxon," in the Publications of the Modern Language Association, 1895. * John Ries : " Die Stellung von Subject und Predicatsverbum im Heliand," in Quellen und Forschungen, vol. xli, * E. Mogk: "Die Inversion von Subject und Predicat in den Nordischen Sprachen," Lndogermanische Forschungett, vol. iv. 8 viii INTRODUCTION, tion perhaps of Russian in Gretsch's chapter on the order of words, have been so obstinately neglected even by so great a scholar as Miklosich, although the subject would be rich in interesting re- sults. As regards a comparative treatment of the Germanic order of words, we still lack it sorely, and if the present paper on the Old Norse should be considered as a modest contribution to that question, I shall be amply rewarded for the time and energy I have spent upon it. In taking up the question of the Old Norse order of words I have confined myself to the syntactic or grammatical aspect of the question — that is, to the prose, thus ignoring the poetic- rhetorical arrangement of words, and within the rich province of prose about 80 pages of the Younger Edda (Gylfaginning and BragarœSur') were chosen, further about 100 pages of the Olafssaga Tryggvasonar * (chapters 59-84, and chapters 124-150), and finally over 50 pages of the Eyrbyggja Saga ' (chapters 1-18, and chapters 41-57), altogether about 240 pages. This has been thought a sufficiently large amount of prose upon which to base a discussion of the order of words ; in fact, 100 pages might have done as well. Confining myself strictly to the immediate facts in Old Norse and making but occasional use of the other Teutonic dialects, I have ignored the pre-Germanic order of words entirely. ^ Quoted simply as Gylf. Ernst Wilken's Die Prosaische Edda, Paderborn, 1877, has been used. * Fornmanna Sögur, Fyrsta Bindi, Kaupmannahöfn, 1825. Quoted as Olafs. ^ Eyrbyggia Saga, herausgegeben von G. Vigfusson, Leipzig, 1864, quoted as Eyrb. ORDER OF WORDS IN OLD NORSE PROSE. 1. INDEPENDENT DECLARATIVE CLAUSES. A. NORMAL ORDER. ( I ) Subject and Predicate. (a) Subject and Predicate Verb. In accordance with the Germanic and Indo-European methods, the predominant mode of expressing the relation between agency and action, stripped of any modifiers, is in the simple affirmative clause : Subject + Predicate, which, for the sake of convenience, may be styled the ^''Normal Order." Whether this is due to the principle of placing the more important, the Subject, before the less important, the Predicate (the so-called " Princip der Voran- stellung des Wichtigen")/ is a matter of speculation which I dare not decide, especially in view of the fact that the Semitic norm for expressing the same relation is : Predicate + Subject, a fact which — if the above-mentioned principle be true — would natu- rally suggest that to the Semitic mode of thinking the action is more important than the agency. On the other hand, Abel Ber- gaigne " and Delbruck' advance the well-known ingenious theory of the Indo-European pre-historic order of words, which — to Ries* it is unquestionable — underwent the change into the Subject -f- ' Quellen und Forsckungen, vol. xli., page 2. ' " Sur la construction grammaticale consideree dans son developpement his- torique," etc., in the Mémoires de la socie'te de linguistique, Paris, 1875, "i- ^ " Die Altindische Wortfolge " {Syntactische Forsckungen von Delbriick und Windisch, Band iii. * Quellen und Forsckungen, vol. xli., page 9. I 2 ORDER OF WORDS IN OLD NORSE PROSE. Predicate order, also " vor der historischen Periode." The per- plexing contrast between the two great families of the Aryans and Semites is, of course, thus not accounted for, by any means, but it is, after all, a question that belongs in the province of the Philos- ophy of Philology. As far as the Old Norse is concerned, the normal order is, as sug- gested : Subject + Predicate, as : Gylf. 5, 15 ; hann svarar : Gylf. 5, 16 ; Hár segir, and so in innumerable cases. In the Gothic " Urkunde zu Neapel " and in the "Urkunde von Arezzo " we find a sufficient number of illustrations to justify the statement that the Gothic had the same order, as the formula " ik ufmelida " indicates, which occurs 4 times in the document of Naples. In Old High German we find in Tatian for the Latin " sciebat enim," her uuesta.' For the Anglo-Saxon Smith^ claims the normal order, and so does Ries ' for the Heliand. If I Subject is put in relation to 2 actions, the latter are in the overwhelming majority of cases connected by conjunctions, such as: "ok," "enn,""ne": Olafs. 269, 7: hann hljóp ok mælti ; Gylf. 99, 9-10 : hann bauS enn mælti ; Olafs. 106, 13 : hann lék né hló ; sometimes, however, the connecting particle is omitted, as Gylf. 67, 15 ; UtgarSa-Loki fylgir, gengr. 3 Predicates depending on i Subject must be connected by "ok," presenting in 23 out of 25 cases the scheme of Gylf. 11, 3, 4 : ]jeir tóku ok fluttu ok gerSu ; in i case " enn " is used in- stead of the second " ok " ; Gylf. 99, 7-8 : Baugi kallaði ok sagðí enn taldiz ; instead of "enn" "eSa"is used once: Olafs. 273, 26 : sumir fellu ok brutu eða fengu ; not a single instance has been found of the arrangement so frequent in German and Eng- lish of: "er kam, sah und siegte," if the predicates are simple verbs * ; nor is there any case in which the connecting particles are entirely done away with. If 4 Predicates depend on i Subject (there are 15 such cases) the arrangement is free, although there is a tendency (6 cases) to connect each predicate by "ok," as : Eyrb. 15, 25 : hón laut ok tók ok bra ok lagði. A very peculiar arrangement is Olafs. 113, 20 : vis forum ok bárum ok lögSum, gengum (4 cases) ; the ar- ' Tatian, Lateinisch und Altdeutsch, Sievers, Paderborn, 1872, page 319, line r. ^Public. Modern Lang. Association, 1893, p. 216, etc. ' Quellen und Forschungen, vol. xli., pages 11, 12, etc. * Compare, however, page 5. INDEPENDENT DECLARATIVE CLAUSES. 3 rangement in 2 pairs also occurs 4 times, as : Eyrb. 96, 31, etc : J»eir tóku ok báru, gengu ok fóru, while in i instance only the first 2 predicates are connected by " ok " : Olafs. 266, 2-3 : bis- kup skrydist, ok vigði, gekk, hafði. If I action refers to 2 or more Subjects, it is just as usual to place the several subjects before the verb, as to place the verb between the various subjects, as Gylf. 45, 8-9 : Guðr ok Rosta ok norn ríða ; but Eyrb. 16, i : Eyjolfr hljóp upp ok hans menn ; if, however, one of the subjects is a person already spoken of in a previous sentence and, therefore, supposed to be well known to the reader, the author employs the familiar construction of hint- ing at such a subject by the demonstrative pronouns : " J^eir, >aer, ])au" placed immediately before the second subject, a construc- tion comparatively rare in the Younger Edda and in the Olafssaga, but frequent in the Eyrbyggja Saga; as : gerSu þeir Gylfi (that is Gylfi and OSinn) sæt sina.' The " þeir," etc., in such cases may also mean " the band, the folk, the host," of the following subject, as J^eir póroddr sátu J>ar (póroddr and his host), Eyrb. 100, 11. In a combination of 2 Subjects and 2 Predicates the most fre- quent arrangement is that of: Olafs. 104, 21-22 : konungr ok hans menn tóku ok vörðust ; another peculiar order is that of Gylf. 10, 16 : hann for ok kona ok helzt, so further in Gylf. 83, 11-12, etc. ; further noteworthy is the combination of Olafs. 153, 17-19: raigjafar ok höfðingjar lautu, sögðu ; natural is, of course, the occurrence of the following combination : Eyrb. 85, i- 2 : þeir Steinþorr fóru ok drógu. The result of our discussion may be conveniently presented in the following table : A. I Subject and i Predicate verb. (i) Subject + predicate. B. I Subject and 2 Predicate verbs. (i) Subject + predicate -f- ok predicate (the majority). (2) Subject -|- predicate + predicate (not very frequent). C. I Subject and 3 Predicate verbs. (i) Subject + predicate -|- ok predicate + ok predicate (23 cases). ' Poestion : Einleitung in das Studium des Altnordischen, § 169. 4 ORDER OF WORDS IN OLD NORSE PROSE. (2) Subject -j- predicate + ok predicate + enn predicate (i case). (3) Subject + predicate + ok predicate + eSa predicate (i case). D. I Subject and 4 Predicate verbs. (i) Subject + predicate + ok predicate + ok predicate + ok predicate (6 cases). (2) Subject -\- predicate + ok predicate + ok predicate + predicate (4 cases). (3) Subject + predicate + ok predicate + predicate + ok predicate (4 cases). (4) Subject + predicate + ok predicate + predicate + predi- cate (i case). £. 2 or more Subjects, i Predicate verb. (i) Subject + ok subject + predicate ) (2) Subject -\- predicate + ok subject j ^ > ^ (3) I'cir, (J'ser, J)au) Subject + predicate (familiar and not in- frequent). F. 2 or more Subjects, 2 or more Predicate verbs. (i) Subject + ok subject + predicate + ok predicate (majority). (2) Subject 4" predicate + ok subject + ok predicate (less frequent) (3) Subject + ok subject + predicate+ predicate (rare). (4) J>eir, (]jær, ])au) Subject -|- predicate + ok predicate (rare). ( I ) Subject and Predicate. (b) Subject -\- Auxiliary Verb -f- Predicate Noun or Adjective. Heyse regards as the " Grundform der deutschen Wortfolge- lehre " the type : Subject + Copula + Predicativum, and re- duces to it in a very interesting way all the phenomena of the normal order in German.* Convenient as it may be to apply it also to Old Norse, it does not seem advisable to do so, mainly for the reason that there are languages within the Indo-European group, not to speak of the Semitic dialects, which do not employ the copula ; for instance, the Modern Russian, in which the ' Heyse, Job. Chr. Aug: Deutsche Grammatik, Hannover, 1886, p. 390. INDEPENDENT DECLARATIVE CLAUSES. 5 auxiliary " to be " is practically not expressed in the present tense, but circumscribed — if an Adjective is the predicativum — by the so-called predicative ending of the adjective. I Subject and i Predicate Noun or Adjective are combined in Old Norse in the familiar way, the medium being the copula which regularly precedes the predicate nomen ; the latter ap- pears either as a substantive, as Gylf. 20,2 : J^essi eru nöfn ; or as an adjective, as Gylf. 20,15 • hann er fullr ; or in the com- parative degree : Gylf. 24, 16-17 '• Ljósálfar eru fegri ; or in the superlative form, as Gylf. 31, 5 : hann er beztr. Very peculiar is the adjective use of "svá" followed by a genitive plural in Eyrb. 94, 3-4 : Kjartan var svá manna (such a man). As copulas are also treated the verbs : verSa, >ykkja, heita,' etc., as Gylf. 35,15 : Hestr heitr Gulltopr. If 2 Predicate Nouns or Adjectives refer to i Subject, the co- ordinate particle is rarely omitted, as in Olafs. 288, ir-12 : hann var rikr maSr, hermaSr mikill. In the majority of cases " ok " is the connecting particle, as Gylf. 14, 2 : hón var svört ok dökk ; sometimes " eSa," as Gylf TyT,, 2-3 : hón heitir öndurguð eSa öndurdís ; "ok >ó " occurs in Eyrb. 14, 18-19: Berg>órr var yngstr ok J»ó enn efniligsti ; " enn " in Eyrb. 11, 11-12 : hann var frændi en námagr .... finally "bæSi ... ok " is found, as Olafs. Ill, 27 : skógrinn var bæíi ])röngr ok myrkr. If 3 Predicate Nouns or Adjectives belong to i Subject, the ma- jority of cases shows a connection of only the last two nomina, as Olafs. 288, 12-13 '• hann var rikr maðr, hermaSr mikill ok újafnaSarmaSr ; so further Olafs. 259, 20 ; 300, 26-27, etc., Gylf. 10, 3, etc., a construction which hardly occurred, if the predicates were simple verbs ; the combination of Gylf. 31, 9-10 : hann er vitrastr ok fegrst-talaðr ok liknsamastr is rare. If 4 or more Predicate nou?is or Adjectives depend on i Subject, the treatment is free, although the arrangement in pairs appears comparatively more frequently, as Gylf. 40, 9-10 : fjöturinn varð slettr ok blautr, traustr ok sterkr ; sometimes out of four and even five predicate adjectives only the last two are connected : Olafs. 155, 7-8 : Sigvaldi var hárr, lángleitr, bjugnefjaðr, fölleitr ok eygSr vel. Occasionally only the first two predicate nomina are connected : Gylf. 37, 7-8 : Loki er friSr ok fagr, illr, fjöU- ' Cf. Heyse, page 390, a. 6 ORDER OF WORDS IN OLD NORSE PROSE. breitinn ; further: Olafs. 155, 15-16: hann var bæði mikill ok styrkr, fríðr, údæll. . . . If 2 Subjects have i Predicate Noun or Adjective in common, there is only one arrangement, that of Eyrb. 21, 7-8 : porsteinn ok Hallr vóru synir þeirra. If 2 Subjects are connected with 2 Predicate Nouns or Adjectives, the arrangement is equally divided between that of Olafs. 287, 5-6 : einn var porleifr, annar Ogmundr ; and that of Olafs. 267, 17-18 : brój'ir hét Arngeirr ok annar f>ór5r ; or the " ok " is sub- stituted by " enn," as Olafs. 255, 19-20 : sonr >eirra hét Ormr, enn annar Jjorvaldr. All the combinations that fall under the heading of " Subject + Auxiliary -\- Predicate Noun or Adjective " may conveniently be summed up in the following table : A. I Subject, I Predicate Noun {or Adj.). (i) Subject + copula + predicate substantive. (2) " " " adjective. (3) " " + svá followed by the genitive. P. I Subject, 2 Predicate Nouns {or Adj.). (i) Subject 4- copula + predicate nomen + no conjunction predicate nomen (rare). (2) Subject + copula + predicate nomen + ok predicate nomen (majority). (3) Subject + copula + predicate nomen + eSa predicate nomen (sometimes). (4) Subject + copula + predicate nomen + ok J)ó predicate nomen (sometimes). (5) Subject + copula -\- predicate nomen + enn predicate nomen (sometimes). (6) Subject -{- copula -|- bæði predicate nomen + ok predicate nomen (sometimes). C. I Subject, 3 Predicate Nouns {or Adj.) (i) Subject -|- copula + predicate nomen + predicate nomen -|- ok predicate nomen (majority). (2) Subject + copula -\- predicate nomen + ok predicate nomen + ok predicate nomen (rare). INDEPENDENT DECLARATIVE CLAUSES. 7 D. I Subject, 4 Predicate JVouns {or Adj.). (i) Subject + copula + predicate nomen + ok predicate nomen + predicate nomen + ok predicate nomen. (2) Subject -(- copula + predicate nomen + predicate nomen + predicate nomen + ok predicate nomen. (3) Subject + copula + predicate nomen + ok predicate nomen -f predicate nomen + predicate nomen. E. 2 Subjects, I Predicate Noun {or Adj.). (i) Subject + ok subject + copula + predicate nomen. P. 2 Subjects, 2 Predicate Nouns {or Adj.). (i) Subject + copula + predicate nomen + subject + predi- cate nomen. (2) Subject + copula + predicate nomen + ok subject + predi- cate nomen, (3) Subject + copula + predicate nomen + enn subject -|- predicate nomen. ( I ) Subject and Predicate. (c) Subject + compound predicate. If the predicate is a compound tense of the verb, the normal order is, " subject -|- auxiliary + verbal noun," the auxiliary thus occupying the position of the copula, the verbal noun that of the predicate noun of the above-treated type : subject + copula + predicate noun or adjective. Illustrations are : sá er nefndr, Olafs. no, 28 ; further : hann skal standa, Gylf. 25, 12, or : þak hennar var lagt, Gylf. 4, n ; sometimes the auxiliary verb itself is a compound tense, and the dependent verbal noun is, differ- ently from the German, usually placed at the end, or, better, after the compound form of the auxiliary, as : Hárekr let gera kirkju, Olafs. 107, 24 ; but just one line after the author says ; Haroldr hafSi látit gera (compare the Modern German : Har. hatte raachen lassen), and with the dependent participle : nökkurir höfSu verit sendir, Olafs. in, 19 ; on the other hand, there are also very few instances of the German treatment of that combina- tion, as : Olafs. 149, 5 : hún hafSi gipt verit = sie war anvermahlt worden. The position of the simple auxiliary after the verbal noun 8 ORDER OF WORDS IN OLD NORSE PROSE. hardly occurs in Old Norse Prose, is used very frequently, how- ever, for rhetorical and rhythmical reasons in the poetic language, so in the Elder Edda, Vafjjrúðnismál 45, line 2 : \z.\x leynaz munu etc. Smith finds it *' in the Orosius most frequently in the so- called progressive forms of the verb," in which cases "the auxil- iary follows the verb proper, and thus occupies the extreme end- position, thus exhibiting both marks of complete transposition " (in the above-mentioned dissertation, p. 231). Nor is the trans- posed order rare in the Old High Gerinan independent clause.' In Tatian, page 69, § 11, we read : inti gefulte uurdun tho taga sines ambachtes," and in Latin : et factum est ut impleti sunt dies ofificii ejus. As to the Heliand, compare Ries, page 12 : ik fullón skal, line 4767. Here perhaps also belongs the peculiar extreme end position of the verbs, although not of the auxiliary verb, in the Gothic "Urkunde von Arezzo " (Stamm-Heyne's Ulfilas 230): ik, Gudilaib, \o frabauhtaboka fram mis gavaurhta \w% Alamoda fidvor unkjane jah skilliggans .rig. (133) andnam jah ufmelida. In a combination of i Subject and 2 Predicate co?nJ)02ind verbs the auxiliary is generally employed but once, provided, of course, if it is the same employed by the 2 verbs, as : ek skal stySja ok styrkja, Olafs. 280, 14-15, and with the participle : dvergarnir höfSu skipaz ok tekit. . . . Gylf. 18, 7 ; but for the sake of clear- ness the auxiliary is repeated before the second verbal noun, if a clause or two or more phrases stand between the two verba in- finita, as : konúngr let taka hauk, er Astríðr átti, ok let plokka. , , . Olafs. 298, 22-23, and : sauSamaðr Snorra hafði verit á Óxna- brekkum um daginn ok hafSi set. . . . Eyrb. 87, 4-5. Instances of j Predicate compound verbs being attached to i Sub- ject are very rare ; the auxiliary is in such cases employed only once, and is placed before the first verbum infinitum, as : likit var sveipat enn saumat eigi um ok lagt, Eyrb. 96, 21-22. If I Predicate compound depends on 2 Subjects, the latter always precede, there being no instance of one of the 2 subjects ap- pearing after the predicate, as was the case with " 2 Subjects -j- ^ simple verb predicate." The illustrations are numerous : Sol ok Bil eru taldar. . . . Gylf. 45, 2 ; 45, lo-ii, etc. A combitiation of 2 Subjects and 2 Predicate compound verbs has not been found. ' Becker, p. 436, § 285. INDEPENDENT DECLARATIVE CLAUSES. 9 If a Predicate Noun is connected with a Predicate compound, we find in 22 cases out of 28 the predicate noun at the end, and the 6 cases of the predicate nouns being placed before the verbal noun occur exclusively in the Gy If ag inning, as : sá maðr er nefndr Mundilfari, Gylf . 14, 16 ; but : sá er Surtr nefndr, Gylf. 7, 9. Summing up all the possible modifications of the combinations : subject + predicate compound verb, we get the following table : A. I Subject, I Predicate compound verb. (i) Subject + auxiliary + infinitive ) ., , •,, , ) ' ^ ' . . -^ ' . . , f the only possible order. (2) Subject + auxiliary + participle ) (3) Subject + auxiliary + auxiliary participle + verb infinitive (most frequent). (4) Subject + auxiliary + auxiliary participle + verb parti- ciple (most frequent). (5) Subject 4" auxiliary + verb participle + auxiliary parti- ciple (very rare). B. I Subject, 2 Predicate compoimds. (i) Subject + ^ auxiliary + oc infinitive + ok /? infinitive (regular). (2) Subject + a auxiliary + a participle + ok /? participle (regular). (3) Subject + oc auxiliary -\- a infinitive + ok « auxiliary + /J infinitive (for the sake of clearness). (4) Subject + ^ auxiliary + a participle + ok « auxiliary -\- ok /? participle ( for the sake of clearness). (5) Subject + a auxiliary -\- a infinitive -f" ok /? auxiliary + /J infinitive (naturally). (6) Subject + a auxiliary + a participle + ok /? auxiliary + /^ participle (naturally). C. I Subject, 3 Predicate compound verbs. (i) Subject + auxiliary + oc infinitive + ok /? infinitive + ok y infinitive. (2) Subject + auxiliary -|- a participle + ok /? participle + ok Y participle. D. 2 Subjects, I Predicate compound verb. (i) Subject + ok subject + auxiliary + infinitive, (2) Subject + ok subject + auxiliary -f participle. 10 ORDER OF WORDS IN OLD NORSE PROSE. E. Subject + Predicate compound verb + Predicate Noun. (i) Subject + auxiliary + participle + predicate substantive. (2) Subject + auxiliary + participle + predicate adjective. (3) Subject + auxiliary + predicate substantive + participle (only in Gylf.). (4) Subject + auxiliary + predicate adjective + participle (only in Gylf.). (5) Subject + auxiliary + predicate adjective + infinitive (only in Gylf.). A. NORMAL ORDER. (2) Object. Ries, in discussing the position of the verb in regard to the other parts of the sentence, remarks : " Im Anschluss an Ber- gaigne scheint es mir absolut mehr keinem Zweifel zu unterliegen, dass die Stellung des Subjects an der Spitze, des Verbs am Ende des Satzes, aller ubrigen Satzglieder in ihrer Mitte, — wie es als das allgemein indogermanische Wortstellungsschema zu betrachten ist, auch die Grundlage der germanischen Wortfolge gebildet hat.'" He finds that in 11 cases out of 296 the object follows immediately after the subject and thus precedes the predicate. Smith,'" and so also Becker for Old and Middle High Gerfnaii,^ points out that in Anglo-Saxon the pronominal dative and accusa- tive are regularly placed before the predicate,* a phenomenon well known in French and extensively used in Russian, in which it is not restricted to the pronominal object. It was mentioned above in connection with the transposed order that the Gothic, in the fragment of Arezzo, has also the predicate following the object : ik . . . gavaurhta . . . jah skillingans .rig. adnam jah ufmelida." Becker says that Old High German " lasst auch zuweilen das — (Noun) — Object dem Predicate vorangehen," and quotes Otfried and for Middle High German the Nibelungenlied.'' No matter whether one believes that the poetic order of words, or, better, ' Pages 88 and 91, Quellen und Forschungen, vol. xli. * Pages 219 and 220 of his dissertation. ^ Becker's Deutsche Grat?imatik, p. 462. * Compare also Kube's dissertation, Die Wortstellung in der Sachsenchronik, page 15. * Becker, Detitsche Grammatik, pages 460, 461. INDEPENDENT DECLARATIVE CLAUSES. II the poetic order, in rare exceptional cases is able to throw true light upon the order of the language, — the quotations from the Heliand and Otfrted, few as they are, are interesting, especially since the type : subject + object + predicate is also the order of the language of the deaf and dumb.^ As regards the Old Norse prose such order could not be discovered in the loo pages of the Olafssaga that were carefully studied for that purpose, while, on the other hand, the Elder Edda employs it occasionally for reasons of alliteration and rhythm. The only order possible in Old Norse prose is thus : subject -\- predicate + object^ no matter whether the object be a pronoun or a noun. Illustrations follow : (a) Accusative, pronoun or noun : Olafs. 102, i ek segi fat ; Olafs. 106, 12-13 : hann tók ríki. Olafs. 108, 21 etc.: þeir brendu Kölni ok allar borgir (2 accus.) ; Olafs. 145, 20 : hann kal- laði sik Ola (2 accus.), (b) Dative, pronoun or noun : Olafs. 135, 4-5. f»angbrandr sagði honum ; Olafs. 112, 14 : hann unni Kniiti ; Olafs. 104, 13-14 : Olafr hét J^eim makligri ömbun (2 datives). (c) Genitive pronomi or noun : Olafs. 114, 18 : Jieirhefndu föSur sins etc. The combination of Dative and Accusative shows, at a ratio of 3 to I, that the indirect object precedes the direct, as Olafs. 103, 17 : hann bauS borgarmönnum griS ; so further in Olafssaga : 118, 12-13 ; 145» 24-25 ; 147, 3 ; 264, 20 ; Eyrbyggja : 97, 15 etc. In Anglo-Saxon the same tendency prevails.^ The Dative follows the Accusative, for instance, in Olafs. 129, 19-20 : Haroldr hafSi sent orS Hákoni jarli, etc. The accusative personal pronoun precedes a getiitive noun, as : hann eggjaSi fik hins versta verks,' or a dative noun, as Olafs. 264, 8-9 : >essi vandr svikari hefir afsett mik allri minni eign ; and the prono7ninal dative object precedes a nominal genitive object, as : viljum vér unna hanum tignar.^ This marked tendency of the Old Norse pronominal object to precede a substantival object occurs — according to Becker — both in Old and Middle High German : " Unter den erganzenden Objecten folgt immer der Sachcasus, als Object der Art, dem Personencasus als Object des Individuums." * ' Ries, page 2, footnote. ' Smith, Publications of the Modern Language Association, 1893, p. 218. ^ Poestion, Einl. in das Studiurn des Altnord., pp. 129, 130. * Becker, Deutsche Grammatik, p. 464. 12 ORDER OF WORDS IN OLD NORSE PROSE. A. NORMAL ORDER. (3) Adverb. Just as the object never precedes its predicate in Old Norse prose so also are the adverb and the adverbial cotnbination invariably bound to the end position in a combination of subject + simple verb. Constructions like the Anglo-Saxon : þa hvælhuntan fyrrast go/ are monstrous in Old Norse. Ries, on the other hand, quotes 20 instances which show a similar structure : it undar iro handun wohs 2869.' A few illustrations will do for the Old Norse : úlfrinn gapti ákafliga, Gylf. 42, 1-2 ; Jieir hestar heita svá, Gylf. 15, i ; j'riSja rot stendr á himni, Gylf. 21, 3 ; mörg dœmi finnaz til Jiess, Gylf. 52, i. Of two or more adverbs of the same character, the adverb more general in meaning precedes the one of a more vivid and detailed description : Jiat er niðr í inn níunda heim, Gylf. 6, 18 ; annar endi hornsins var út i hafi, Gylf. 68, 22 ; gengu fau J'ann dag allan til myrks, Gylf. 58, 8 ; enn opt um daga . . . Gylf. 78, 12 ; excep- tions to that rule are rare : viSir losna or jröSu upp, Gylf. 82, 3, compare also Gylf. 62, 18 ; 41, T1-12 ; 97, 20. If the modifiers of the verb are adverbial factors of different character, there is a slight preponderance of the order : adverb temporis + adverb loci ' over that of local adverb first and tem- poral second ; the ratio is 55 to 45 : þórr veSr \k eptir miSri ánni, Gylf. 79, 20 ; hann hljóp >egar út 1 ana, Gylf. 79, 4 ; stjornur hverfa \k af himni, Gylf. 82, 2. On the other hand we read Gylf. 57, 9 : J>órr dvaldiz ])ar of nóttina, and Gylf. 78, 13 : . . . hann falz J'á Jiar. In combinations of external adverbs, that is Adverbs loci and temporis, with qualitative adverbs (causæ, modi, gradus etc.), the genius of the language permits the greatest freedom, there being no absolute norm : as " eigi er nú fróSliga spyrt," Gylf. 16, 6, but : ok liggja svá víkr í Leginum, Gylf. 3, 12, and so forth." As regards the position of the negative, which in Gothic almost invariably, and — according to March — in Anglo-Saxon regularly, precedes the predicate, * it is distinguished by a considerable in- ' Cf. March, Anglo-Saxon Grammar, p. 220. ^ Que lien und Forschungen, p. 91. ^ Cf. March, Anglo-Saxon Grammar, p. 220. ^ Cf. Heyse, Deutsche Grammatik, p. 395. * So also always in Russian. INDEPENDENT DECLARATIVE CLAUSES. 1 3 clination toward the present German norm, although the tendency is no binding law and disappears to a large extent in the dependent clause, as will be discussed later on. Parallel forms of the German and Old Norse in the independent clause are as follows : The negative follows the simple predicate and shows a " finaliz- ing" tendency. (a) Simple predicate. German : du weisst es nicht and Old Norse : \\x fannt eigi, Gylf. 68, 7, 69, 6, 73, 17. German : damals wusstest du nicht and Old Norse : ]?at sáttu eigi, Gylf. 69, I : \i. trúði hann þeim eigi, 34, 17 ; 40, 21 ; 56, 5 ; 90, 4- German : warum weisst du nicht ? and Old Norse : hvi spyrr >ú eigi? Gylf. 50, 14-15. 74, 11-12. German : schreien Sie nicht ! and Old Norse : látit ]>éT eigi, Gylf. 61, 8. (b) Compound predicate verb. The negative follows the auxiliary and stands as near as pos- sible to the verbal noun : Ger?nan : ich werde es nicht wissen and Old Norse : Jiessi man mik ekki skaSa, Olafs. 286, 7 ; Gylf. 95, 13 etc. German: . . . auch hatte ich es nicht gewusst and Old Norse : ok er pjalfi eigi . . . kominn, Gylf. 6;^, 21 ; 75, 19 ; 83, 15 etc. German : das kann ich nicht wissen and Old Norse : þeira má hann eigi missa, Gylf. 30, 16 ; etc. German: wenn ... so wirst du nicht wissen können and Old Norse : ef . . . fa muntu eigi hræða mega, Gylf. 40, 25, but different in Gylf. 29, ir. German : hast du es nicht gewusst ? and Old Norse : hvart hefir Jjórr ekki þess hefnt ? Gylf. 70, 7-8. If the predicate is modified by i Object and i Adverb or ad- verbial combination, Old Norse employs on the first 100 pages of Úi&Edda out of 250 cases x8o times the order Object -\- adverb, and in 70 instances that of Adverb -\- object, the ratio thus being 3-^ to I. The other old Gennanic dialects are considerably freer in that respect. It should be noticed that the object which precedes the adverb is not only pronominal, but also almost as frequently substantival. Illustrations of the former charactei are : drepa 14 ORDER OF WORDS IN OLD NORSE PROSE. mátti Freyr hann með hendi sinni, Gylf. 47, 13 ; ok báru hann til iilfsins, Gylf. 38, 20 ; svá drap hann sik or Dróma, Gylf. 39, 13, and so forth ; illustrations of the substantival object before the adverb are : Gylf. 33, 9 : hón á fann bœ á himni, er Fólk- vangr heitir. This example is especially interesting. We should expect that the relative clause which refers to " bœ " and not to "himni" should follow immediately after the " bœ." Other il- lustrations are : Gylf. 44, 5, hón gætir dura í höllinni, 38, 2-3 ; etc. A few examples will do to illustrate the secondary tendency of adverb + object : Gylf. 30, 2 : hann á >ar riki ; 33, 4 : Njörðr gat sííSan tvau born ; 40, 5 : >á sá hann á sinum bœ mikil hús, Gylf. 3, I : Gylfi réS J^ar löndum ; 4, 10; 11, 4; 11, 12; etc. Out of the 70 cases of the type : verb + adverb + object there are 26 in which the dative-object would be expressed by a geni- tive possessivus in many other languages, as : hrafnar tveir sitja á öxlum hanum = hans, Gylf. 48, 16 ; . . .ok lagiSi á bak sér, Gylf. 59, II ; ... ok bitr í sporS sér, 38, 4-5, etc. The adverb which modifies an adjective generally precedes it, no matter whether the latter is placed before or after the substantive : svá margir ormar eru . . . Gylf. 23, 7 ; furSu ilia barnaeign gat Loki, Gylf. 42, 7 ; geysi mörg heiti . . . Gylf. 29, i ; ... at gera borg . . . svá trausta, Gylf. 52, 9 ; etc.; irregular is Gylf. 83, 5 ; ok er sú björt mjök ; . . . hann verða mundu ágætan mjök, Gylf. 39, 5, so also Gylf. 39, 5 ; 38, 13 ; 82, 20. If an adverb modifies another adverb, it precedes it, as Gylf. 70, 1 1 : hann bjóz svá skyndiliga, etc., there being no case in 100 pages of the Edda of a modifying adverb which follows the modified one. While the predicate simple, as was shown at some length above, never in the normal order of its predicative combination admits a modifier (Object or Adverb) before it, the question naturally arises, whether the compound predicate verb also always throws its modifier or modifiers to the end, or better, whether in such cases the verbal noun is tied to the auxiliary, showing thus the type : subject + auxiliary -j- predicate + modifiers. Smith asserts that in Anglo-Saxon the " final position of the second member is the most common if the modifiers are few," and quotes inverted and normal sentences to prove it.' Becker remarks," on the other ' Smith, Order of Words, p. 217. ° Becker's Detitsche Grammatik, p. 469. INDEPENDENT DECLARATIVE CLAUSES. 1 5 hand, that not only in Old High Genua?!, but even with Luther the construction of subject + auxiliary + predicate + modifiers (Object) " ist noch sehr gelaufig." Ries did not separate the com- pound tense from the simple predicate.' As regards Old Norse, 240 pages have been examined (normal order only as being the safest criterion) with the following result. Out of the 156 cases with an auxiliary and verbal noun there are 6 that have no verbal modifiers at all, and are, thus, neutral : Olafs. 115, 28, 134, 11 ; 303, 16 ; Eyrb. 88, 20 ; Gylf. 25, 12 ; 99, 4 ; and in 26 the only verbal modifier is a predicate noun (Olafs. no, 28; in, 12; III, 17 ; 112, 17-18 ; 115, 15-16; 116, 1-2 ; 116, 2-3 ; 116, 10 ; 116, 29 ; 128, 9 ; 250, 20 ; 255, 7 ; 276, 6 ; 276, 27 ; 280, 6 ; 288, 18; Eyrb. 11, 7; 14, 10; 18, i; Gylf. 14, 16; 35, 12; 69, 3; 75, 8 ; 92, i) which in the overwhelming majority of cases, at a ratio of 5 to I, stands after the verbum infinitum, as has been said before. We have thus only 124 valid cases left, 47 of which have all the verb modifiers after the verbum infinitum. Of those 47 cases : (i) An adverb has been found only once, in Gylf. 94, 18 : enn J'at var set siSarst. (2) I or more adverbial combinations occur 22 times : nökkurir höfSu verit sendir til Holdsetu, Olafs. in, 19-20; so also Olafs. 119, 45; 122, 9-10; 122, 14-15; 253, 3; 266, 25; 293, 5-6; 29s. 12 ; Eyrb. 10, i ; 84, i ; 84, 29 ; 86, 19-20 ; 87, 23 ; Gylf. 4, II ; 21, II ; 32, 4 ; 32, 10 ; 44, 6-7 ; 44, 9 ; 76, 14 ; 76, 17 ; 77, I- (3) An adverb + adverbial combination occurs twice, Olafs. 118, 22 : hinn hviti haukrinn hefir flogit lángt á eySimerk ; so also Olafs. 112, 3-4. (4) An adverbial combination + adverb once : Eyrb. 84, 23 ; iss var lagSr á HofstaSavág mjök, etc. (5) An object occurs 10 times : Kniitr konúngr let stefna l^ing, Olafs. 112, 24, so further : 114, 22-23 '■> 129, 20 ; 149, 9 ; 150, 2 ; 262, 21 ; 264, 8-9 ; 264, 20 ; 293, 3-4, Gylf. 32, 8. (6) An object -\- adverb is found once in Gylf. 74, 15 : ek mun visa J^ér til. (7) An abverb + objects also once : Olafs. 290, 27-28 : Rögn- valdr let hirSa vandliga bæ^i afhögg ok spónu. ' Quellen und Eorsckungen, pp. 91-93. 1 6 ORDER OF WORDS IN OLD NORSE PROSE. (8) An object -\- adverbial coftibination 9 times : hiin hafði gipt verit einum jarli á Englandi, Olafs. 149, 5, so also : 155, 14 ; 270, 22 ; 277, 2 ; 291, 22 ; Eyrb. 9, 5-6 ; 97, 20-20 ; 104, 2 ; Gylf. 30, 12. It has been noticed already that there is a peculiar lack of illus- trations with one or two adverbs after the verbal noun. This accounts for the fact that out of the 77 cases that are left for con- sideration, 43 instances (over one half) are found in which the auxiliary is separated from the verbal noun by i or 2 adverbs. (i ) I or 2 adverbs : (a) Zoct : her: Gylf. 62, 11-12; aptr : Olafs. 118, 25-26; J>ar : Gylf. 93, 3, Olafs. 119, 16 : Olafr var þar kominn. (b) Temporis : áðr : Olafs. 107, 24-25 ; siSan : Olafs. 114, 14; Eyrb. 6, 7 ; Gylf. 39, 13 ; \k: Olafs. 118, 11-12, 124, 15 ; Eyrb. 18, 31 ; 84, 2 ; Gylf. 12, 11 ; nú : Olafs. 126, 12 ; 130, 10 ; fyrst : Eyrb. 96, 4 ; Gylf. 14, 3 ; hón var fyrst gipt, etc. (c) Modi : svá : Olafs. 123, 14, Gylf. 25, 8 ; ekki : Olafs. 124, II ; 127, 29 ; eigi : Olafs. 138, 4 ; 353, 12 ; 286, 4 ; Eyrb. 9'hi 5 ; Gylf. 95, 13 ; enn örninn mátti eigi stöðva sik. (d) Other adverbs : sannliga : Olafs. 113, 11-12 ; jöfn : Eyr- byggja 89, 10; ymist : Gylf. 27, 6 ; ok : Olafs. 136, 12; 295, 26; Gylf. 43, lo-ii ; enn: Gylf. 37, 3 ; 50, 9-10; J>essar eru enn nefndar. (e) Prepositional adverbs : frá : Eyrb. 89, 5 ; við : Gylf. 55, 13 : allmikil fjölkyngi mun w'v6 vera höfS. (f) 2 adverbs : eigi lengi : Olafs. 257, 20 ; hálfu sí5r : 284, 21 ; svá dauða : 292, 12 ; nú her 292, 20 ; Jiá ok : Eyrb: 21, 5 ; hvergi víSar : 95, 3 ; svá nær : Gylf. 68, 3-4 : }>ú hafðir svá nær haft oss, etc. (2) An object separates the auxiliary from the verbal noun in 10 cases : (a) Genitive : þess : Olafs. 118, 13-14 : Gormr konúngr hafði ])ess heitstrengt. (b) Dative : \é.x : Gylf. 52, 6 ; J^vi : Olafs. 296, 4 ; engum mat : Eyrb. 95, 4 ; mörgum mönnum : Olafs. 146, 7-8 : ok Jjú munt mörgum mönnum koma, etc. (c) Accusative : engan greiSa: Eyrb. 96, 29; J-á ferS: Gylf. 54^ 17 ; engi tiSindi : Olafs. 10 1, 26 ; \\k : Olafs. 264, 21 ; and finally the two accusatives in Olafs. 105, lo-ii : enn hann hafSi allar borgir ok hero]? lagt, etc. INDEPENDENT DECLARATIVE CLAUSES. ly (3) Adverb + object stand between auxiliary and verb twice : ok bae : Eyrb. 12, 20; ok kirkju : Eyrb. 92, 15-16: f»ó- roddr let ok kirkju gjöra. . . . (4) Object atid adverb separate the auxiliary from the verb 6 times : mik eigi : Olafs. 286, 6-7 ; }>at upp : 289, 7-8; ])ér kunnigt : 296, 22 ; hofi upp : Eyrb. 6, 24-25 ; sér fátt : Eyrb, 15, 18-19 ; I'^t vel : Gylf. 99, 19 : enn Baugi lætr Jjat vel vera. (5) An adverbial combination is used for the same purpose 4 times : lengi æfi : Olafs. 288, 9 ; á land : Olafs. 147, 20-21 ; enn annat sinn : Gylf. 69, 16 ; til ]jess : Eyrb. 14, 15-16 ; Steinjiórr er til Jjcss tekinn. (6) An adverbial combination + adverb occurs once only : af öllum vel : Olafs. 288, 4-5 : ]>etta ráS var af öllum vel rómat. (7) Adverbs -\- adverbial combination is also found only once : far opt fyrir dyrum : Gylf. 50, 13-14: alljjröngt mun J^ar opt fyrir dyrum vera. (8) An object + adverbial combination separates auxiliary from verb once : ]>éx at Oddi : Eyrb. 18, 28-29 '■> ^^^i g^g^^ radLXi ]>éx at Oddi verSa. (9) The predicate noun separates the parts of the compound predicate 5 times (compare above) : (a) As a sifnple predicate noun 3 times only in Gylf. 7, 9 ; 42, 5, and 74, 7 : sá er mistilteinn kallaSr. (b) In combination with an adverb once in Gylf. 69, 3 : J>át eru mi fjörur kallaSar. (c) In combination with an object once in Gylf. 90, i : ok hitt mun ]>ér undarligt ])ykkja. (10) Object + object + Relative-Clause (the finest illustration) is found once : Olafs. 126, 7-8 : enn Danir hafa kvikfé sitt ok alt annat gods, j^at er }?eir eiga, flutt þángat etc. (11) Apposition -\- adverbial combination is found once in Olafs. 152, 12 : j^au naut voru oil einn veg mörkut. Here must also be mentioned 2 peculiar cases in which an ad- verbial phrase of time separates : (a) the copula from the predicate noun : Olafs. 107, 3-4 : hann var 5 vetr konúngr ; and (b) the co?Jipound verb frotn the predicate adjective : Olafs. 128, 9 : veSrit hafSi verit um daginn kyrt ok bjart. The discussion here would, however, lose its value without a 1 8 ORDER OF WORDS IN OLD NORSE PROSE. further examination as to how many of the instances in which the auxiliary is separated from the verbal noun by a modifier show complete final position of the verb. Out of the 77 cases which belong here, in 48 instances we find verbal modifiers not only be- fore, but also after, the verbum infinitum, while 29 cases, that is, about 39 per cent., show the complete final position of the verbal noun, whereby the following modifiers are thrown between the two constituent parts of the compound predicate. (a) Object 4 times. (i) Accusative in Olafs. loi, 26 ; 264, 21 : ek hefi þik dýrSkat. (2) Dative in Olafs. 296, 4 ; J>ú munt \n\ ráíSa. (3) Genitive in Olafs. 118, 13-14: Gormr konfmgr hafSi J>ess heitstrengt. (b) Adverb 10 times. (i) Loci : Olafs. 118, 25-26 ; 149, 16 : Olafr var þar kominn. (2) Temý : Eyrb. 84, 2 : hann hafSi Jiá beSit. (3) Modi: Olafs. 123, 14; Gylf. 25, 8; Eyrb. 93, 5 : enn pórgunna vildi eigi selja, (4) Other adverbs and adverbially used prepositions : Olafs. 136, 12; 295, 26; Eyrb. 89, ic ; Gylf. 55, 13; allmikil fjölkyngi mun við vera höfS. (c) Adverb + adverb 3 times: Olafs. 292, 12 ; 292, 20; 294, 4 : Rögnvaldr hafSi sva um búit. (d) Object + adverb 3 times : Olafs, 286, 6-7 ; 296, 22 ; Gylf. 99, 19 : Baugi lætr fat vel vera. (e) Adverbial combination once : Eyrb. 14, 15-16 ; Stein])órr er til Jiess tekinn. (f) Adverb -\- adverbial combination : Gylf. 50, 13-14: allj^röngt mun ])ar opt fyrir dyrum vera. (g) Adverbial combinatiofi + adverb: Olafs. 288, 4: j^etta ráð var af öllum vel rómat. (h) Object + adverbial combination: Eyrb. 18, 28-29: ekki gagn man Jiér at Oddi verSa. (i) Predicate noun 3 times : Gylf. 7, 8 ; 42, 5-6 ; 74, 7 : sá er mistilteinn kallaSr. (j) Object -\- predicate noun: Gylf. 90, i : ok hitt mun J>ér undarligt fykkja. (k) Apposition -\- adverbial combination : Olafs. 152, 12: J>au naut vöru oil einn veg mörkut. INDEPENDENT DECLARATIVE CLAUSES. 1 9 It thus seems that the Old Norse employs a very interesting compromise between the Modern English and Modern German, with a most decided sympathy for the latter order. B. INVERSION PROPER.' It has been emphasized more than once that there is no excep- tion to the Old Norse law^ according to which the verb which is modified is bound to precede its modifier (object and adverb) and it has also been remarked that Old Saxon and Anglo-Saxon had various — the latter even numerous — illustrations in which said law is broken. This is a point of the utmost importance for our immediate discussion. For, if it be true that the verb-modifier (object and adverb) must folloiv the verb which it modifies, thus showing the strictest possible local and logical coherence with it, will not that very modifier, if — for some reason to be discussed later — it is torn out of the normal order and placed at the begin- ning of the sentence, show the same strictness of logical and local coherence to the verb ? The question cannot but l^e answered in the positive. And, again, which order could better typify said closeness than the type : Verb-modifier + verb + subject, which is the retro-perspective, as it were, of the normal order (subjects- verb + verb-modifier), and which type, for lack of a better term, we may call Inversion proper. This argument — to my mind — also accounts for the not infrequent exceptions to the rule in Old Saxon and Anglo-Saxofi, in which dialects, as has been remarked before, the verb-modifier is by no means always bound to follow the modified verb." Ries regards " das Frincip der Ideenassocia- tion " as a minor element in the making up of the inverted order ; but, on the other hand, I do not believe that his arguments as to the " stilistisch-rhetorisch-syntactische Griinde " ' are valid, since they only explain why the verb-modifier is torn out of its normal order, without accounting also for the main point, why — the verbal modifier being given in an initial position — the predicate should immediately follow, and not the subject. It would be monotonous, and therefore, from a rhetorical-syntactical standpoint (which in- 1 Interrogative and imperative clauses show inversion in almost all of the Indo-European languages. The inversion in such cases is subject to common Indo-Germanic feeling and does not properly belong under this heading. * Cf. also Ries, Quellen und Forschungen, vol. xli., pp. 46, 47, § 13. * The same, § 5-g. 20 ORDER OF WORDS IN OLD NORSE PROSE. eludes Stilistics), blamable to say : he came home, he ate then, he read afterwards ; so in order to make the style more vivid we place the verbal modifiers of the second and third clauses at the beginning, and say : he came home, then he ate, and after- wards he read. The parataxis is thus closer. But while this ex- plains why the verbal modifiers are placed at the beginning, it does, of course, not explain why we should say : he came home, then ate he and afterwards read he. The initial position of the verb-modifier, especially the object and its substitutes (verbal nouns, object-clauses), is, finally, brought about for the sake of emphasis and stress ; that the emphatic position could not be that of subject -[- verb-modifier -\- verb has been shown above, consequently the only emphatic place of abverb and object had to be that of the beginning, as it is con- tained in type : verb-modifier -{- verb + subject. Before taking up the Old Norse inversion in detail, a few words about the Teutonic and other dialects will perhaps not be out of place. Inversion is a common Germanic feature, and all the dif- ferent dialects partake of it. Aside from Old Norse, which hardly chronicles any exceptions in prose, Old High German is more affected by it than either Anglo-Saxon or Old Saxon. Ries found in the Heliand material enough to write a separate chapter about the exceptions under the heading of ''^ Irreguldr-gerade Folge." ' But in the case of Heliand, instances brought about by rhythmical and alliterative considerations have crept in, which the Old Saxon prose would hardly furnish. As to the Anglo-Saxon inversion. Smith remarks" that "inversion is a means consistently em- ployed," and Matzner, speaking about Modern English, says^ that "it has preserved enduringly echoes of Germanic connections of words," and quotes copious illustrations of inversion. Old High German, in its development through Middle to Modern High German, presents a great many interesting facts that have — to my knowledge — not been recorded as yet, in spite of Becker's elaborate treatise on the German " Wortfolge," a chapter of 56 pages." Gothic, finally, is not a reliable source as to the word- order. Of the non-Germanic languages the Romance dialects ' Quellen und Forschungen, vol. xli., pp. 47-56. * Publications of the Modern Language Association, 1893, p. 222. ^ Matzner's English Grammar^ vol. iii., pp. 535. * Becker's Deutsche Grammatik, pp. 423-47S. INDEPENDENT DECLARATIVE CLAUSES. 21 know it,' while the Slavic (especially Modern Russian and Polish) use inversion very frequently, perhaps as a result of the Western influence (German and Scandinavian).' As regards the Old Norse law of inversion it may be thus briefly formulated : If at the beginning of the sentence there is a word or words, a phrase or phrases, a clause or clauses adverbial or objective in character, the predicate, of which these elements are locally and logically a part, follows immediately and in turn is followed by the subject. That does not mean that all the modi- fiers of a verb must necessarily stand at the beginning. On the contrary, if out of 3 or 4 modifiers only i is torn from its normal order, the other modifiers may retain, in spite of the inversion, the place to which they were entitled in the normal order, as : Normal : )ieir fóru eptir J'at suSr meS landi til NorSimbraland. Inverted : (a) eptir }>at fóru J^eir suðr meS landi til NorSim- braland ; or (b) suSr fóru J-eir eptir J'at meS landi til Norðim- braland ; or (c) meS landi fóru J^eir eptir \zX. suSr til NorSimbra- land ; or, finally (d) til NorSimbraland fóru \t\x eptir J'at su^r meS landi. ^ Where a co-ordinate conjunction precedes a modifier, the inver- sion is due not to the conjunction, but to the modifier: for in- stance, in a clause like : " and there were many hardships awaiting them," the inversion is, of course, not due to the " and," or due to the fact that the " and " is coupled with the adverb " there," but simply to the adverb " there," which has the immediate in- verting power. That would appear as a matter of course in the English inversion, but in Old Norse the co-ordinate conjunctions themselves, as will be discussed later, require inversion to a con- siderably greater extent than the German "und," " doch," etc. Inversion proper in Old Norse prose always and in all possible modifications of Subject and Predicate takes place : (i) After an Adverb, Adverbial Phrase and Adverbial Clause. (2) After an Object, Logical Object and Noun Clause. ' Dietz, Romanische Granirnatik, vol. iii., p. 463, 3. * To Professor Leo Wiener I am indebted for this suggestion. Compare also the comprehensive work : Buslayev, Historical Grammar of the Russian Lan- guage, who, on page 372, quotes illustrations. I shall make use of the Rus- sian and Polish material which I have collected in a separate paper. ^ But compare, on the other hand, a clause like, Olafs. 149, i : enn þar for um landit þíngbo'S nökkut + subst. clause, explaining the " þíngbo'S." 22 ORDER OF WORDS IN OLD NORSE PROSE. (3) After a Predicate Substantive, Adjective, Infinitive and Par- ticiple. (i) After an Adverb, Adverbial Phrase, and Adverbial Clause. A. Adverb. (i) Local : her or enn her (4),' Olafs. 281, 23 : her eru nil saman komnir margir höfSingjar ; þar (43), Olafs. 125, 4: J'ar tók hann kaupskip ; ok >ar ( i ), Olafs. 117,26: ok þar höfSu Danir sigr ; enn Jar (i), Olafs. 149, 1 : enn >ar for >ingbo^ nökkut ; J^aSan, or "ok" or "enn " Jaðan (5), Olafs. 144, 9 : J>a5an sigldi hann vestr ; enn eptir (i) as a prepositional adverb : Olafs. 296, 4 : enn eptir mun ek vera. Only in the poetic passages quoted in the Younger Edda from the Elder Edda are there exceptions to be found: as Grimn. 14 : enn J^ar Freyja ræSr, or Grimn. 15 : enn þar Forseti byggir, etc., etc. (2) Two or more local adverbs : enn niSr ok norör (i), Gylf. 77, 3 : enn niðr ok norSr liggr Helvegr. Ok jjar útan um (i), Gylf. 12, 4 : ok >ar útan um liggr inn djúpi sjár. Altogether 57 cases, (3) Temporal adverbs : J>á (212), Olafs. 107, 2, etc. ; nú or enn nii (219), Olafs. 130, 14; 280, 10, etc.; %\\2S\. or enn sij^an (30), Olafs. 131, 11-12; 106, 7-8, etc.; síðazt (i), Olafs. 253, 28-29 ; ok fiegar (i), Olafs. 269, 4-5 ; jafnan or ok jafnan (2), Olafs. 280, 17 ; 288, 25-26 ; enn héSan (i), Olafs. 286, 7-8 ; ok aldri (i), Olafs. 306, 14-15 ; snimma (i), Olafs. 288, 15-16 ; fyrr (i), Gylf. 7, i ; fyrst (i), Gylf. 7, 6 ; næst (i), Gylf. 9, 12 ; opt or " ok " opt (2), Gylf. 59, 9 ; enn meSan (i), Gylf. 80, 18 ; enn áðr (i), Gylf. 8x, 7. Again the only exceptions were to be found in poetry : Grim- nism. II ; enn nú SkaSi byggvir, etc. (4) Temporal -\- temporal : >ar næst (3), Gylf. 13, 5 ; eptir fa (i), Olafs. 109, 25-26. Altogether 285 cases. (5) The negative : eigi, ekki, ok eigi, ok ekki, enn eigi, enn ekki : Gylf. 74, 4 ; Olafs. 304, 28, etc. 35 cases. (6) Causal : >vi, ok ]>vi, enn \v\ (19), Olafs. 120, 19; Gylf. 49, 3 ; >áSan (i), Gylf. 16, 3. 20 cases. (7) Miscellaneous adverbs : ok enn (i), Gylf. 36, i ; litt (i), Olafs. 298, 17 ; allmiklu (i), Gylf. 78, 5 ; gjarna or enn ' The numbers in parentheses indicate how frequently the word occurs. INDEPENDENT DECLARATIVE CLAUSES. 2$ gjarna (3), Olafs. 284, 30 ; nær = almost (r), Gylf. 35, 9-10; alt = throughout (i), Gylf. 17,18; allvel (i), Olafs. 301, 4 ; sváor ok svá (26), Gylf. 3, 13 ; skjótt (i), Gylf. 15, 10 ; un- darliga (i), Gylf. 49, 7 ; ok jafnskjótt (i), Gylf. 54, 12 ; vel or ok vel (2), Gylf. 63, 17 ; ilia (i), Olafs. 263, i ; áheyriligt (i), Olafs. 304, 27-28; sannliga (i), Olafs. 271, 23. Alto- gether 43 cases. (8) Several adverbs of different character : litlu síðarr (3), Olafs. 151, 10; nökkuru síðarr (2), Olafs. 117, 24-25 ; enn litlu siSarr (i), Olafs. 109, 1-2 ; enn J^egar eptir (i), Gylf. 84, 5 ; svá mikils (adverbial genitive) (i), Gylf. 42, 9 ; ok svá langt (i), Gylf. 69, 8-9 ; ok eigi siSr (i), Olafs. 262, 17 ; ok ]>vi siSr (i), Olafs. 269, 19. II cases. Altogether 451 illustra- tions. B. Adverbial phrase. (i) Local : ok í hánum miðjum, Gylf. 7, 2-3 ; enn í AsgarSi inum forna, Gylf. 6, 3 ; ok undir hvert horn, Gylf. 11, 13— 14; ok meS sjávarströndu, Gylf. 12, 5; í J>eirri kirkju, Olafs. 272, 6 ; á Náströndum, Gylf. 88, 4-5 ; í þeim staS, Gylf. 31, 12 ; frá Scotlandi, Olafs. 143, 24-25, etc. On 150 pages' 31 cases were counted. (2) Temporal : í J>ann tima, Olafs. 119, 13-14 ; á einu sumri, Olafs. 255, 28 ; ok at skilnaSi, Olafs. 299, 16-17 ; ok í enda veraldar, Gylf. 7, 11 ; enn at vetri, Gylf. 99, 14 ; enn í J>ví bili, Gylf. 58, 24. Enn um kveldit, Gylf. 51, 12, etc. The temporal phrases are very frequent. On 150 pages 112 cases were counted. (3) Te77iporal phrase -\- temporal phrase : a ]>vi sama sumri eptir aljiingi, Olafs. 273, 16-17 '■> ^itt sumar á alj^íngi, Olafs. 269, 15 ; and finally Gylf. 57, 9-10 : enn í óttu fyrir dag stóS hann upp. 3 cases. (4) Causal {source and instrument^ : fyrir Jivi, Olafs. 150, 10— II ; fyrir J'at, Olafs. 270, 7 ; af J>essum mönnum, Gylf. 89, 14 ; af þessu, Gylf. 98, 15 ; >ar af, Gylf. 13, 19 ; ok af hen- nar nafni, Gylf. 34, 3 ; or Ymis holdi, Gylf. 12, 11, and so on. Illustration : enn af atkvæSi guíanna urSu j^eir vitandi, Gylf. 18, 9. 47 cases. ' That has been regarded as sufficient to illustrate "Inversion proper." Younger Edda and Olafssaga. 24 ORDER OF WORDS IN OLD NORSE PROSE. (5) Miscellaneous adverbial phrases : enn til J'eirar borgar, Gylf. 12, 7-8 ; ok J'ar til, Gylf. 17, 22 ; enn á moti J^eim, Olafs. 108 ; 13-14 ; fyr öngan mun, Gylf. 9-4. Illustration : allramæst var hann tignaSr urn Austrveg. Olafs. 275, 18- 19. Altogether 10 cases. (6) Several adverbial phrases o/ different character : ok fyrsta sinn á skóginum, Gylf. 68, 5 ; and 148, 11 Olafs : enn ]?vi nærst meSr guj>s lofi skirSi hann Olaf. Altogether 175 cases. C. Adverbial Clauses. (i) Loci: enn hvar sem hann áttí J^ing bauð hann öUum mönnum at lata, Olafs. 119, 20. i instance. (2) Temforis : here the cases are very numerous. Gylf. 46, II : ok er hann kom heim, mælti hann ekki ; Gylf. 69, 17- 18 : enn er þórr heyrði }>essa tölu, greip hann til hamarsins; Olafs. 1 15, 24-25 : ]>á. er Gormr var roskinn fékk hann konu feirar ; Olafs. 138, 16-17 : sem hann hafSi Jjessa luti séð ok heyrt, aetlaSi hann. . . . Gylf. 4, 10 : ok er hann kom inn Í borgina, sá hann J'ar. 76 cases. Here the two cases belong which are apparently exceptions: Gylf. 12, 15, and 68, 18. The former reads in Wilkens edition of the Younger Edda : ]>á er j^eir gengu með sævarströndu, Börs synir fundu tré. Such an order is monstrous in Old Norse prose, and is diametrically opposed to the genius of the language. It is a glaring mistake, and certainly no im- provement upon y^onsson's reading, p. 19, 16 : ]>a er Ipeir Börs synir gengu meS sævar ströndu, fundu j^eir tré, etc., which is natural and in accordance with the spirit of the language ; the page G. 68, in which the other exception occurs, is full of anacolutha, and that special instance is an anacoluthon itself : enn er pjálfi ])reytti rás viS ])ann er Hugi hét, J^at var hugr minn. (3) Causal : Olafs. 272, 2-4 : ok fyrir \\\ at hann hélt helga trú, var hann kallaSr Máni enn kristni. i case (4) Final negative: Olafs. 132, 8 : ef landsherrinn slægist í móti J^eim (lest perhaps) flýSu \€\x \k brott, etc. i case. (5) Conditional : Olafs. 258, 4-5 : enn ef honum hlotnaSust herteknir menn, sendi hann \i. aptr til feSra sinna. i case. (6) Te7nporal -)- temporal : Olafs. 259, 9 ; 268, 25 ; 139, 16 ; and 147, 17 : enn er ábóti var búinn ok allir vóru skrýddir gekk ábóti til strandar, etc. i case. INDEPENDENT DECLARATIVE CLAUSES. 25 (7) Several adverbial clauses of different character : {a) tem- poral -\- local : Olafs. 149, 21 : enn erhiin kom J^ar er Olafr stóð, leit hun á hann öllum megin. {b) temporal -{- substan- tive clause : Olafs. 122, 8 : J^á er keisarinn spurSi, at Hákon var Í Danmerk ok ætlaði at berjast í nióti honum, sendi hann jarla sina tvá ; and according to Jonsson's reading : Gylf. 71, 8-9, which Wilken, p. 93, 12, changes, {c) tem- poral -\- modal : Gylf. 78, 10 : J'á er guSin vdru orðin svá rei5, sem van var, hljóp hann á braut, etc. (ð') teínporal -\- relative : Olafs. in, 23 and 152, 27: ok er Haraldr kom með herinn, er Hákon jarl hafði til forráSa, tók hann at herja. 6 cases. Altogether 87 cases. D. Adverb + adverbial phrase : {a) one adverb + one adverbial phrase : siSan at kveldi, Gylf. 59, 12 ; ok litlu fyrir dagan, Gylf. 60, 17 ; enn opt um daga, Gylf. 78, 12 ; enn eptir um daginn, Gylf. 54, 8 ; upp á himin, Gylf. 22, 2-3 ; snim- ma um várit bjó hann skip sin.' (^) several adverbs -\- one adverbial phrase : enn fyrir innan á jörSunni gerSu J^eir borg, Gylf. 12, 6. (<:) one adverbial phrase + one adverb : enn at morni J^á, Gylf. 77, 10 ; enn at miSri nótt Jíá, Gylf. 60, 6 ; á hans dögum ofarliga, Olafs. 117, 15 ; enn um várit eptir, Olafs. 121, 29 ; and so forth, 16 cases ; altogether 22 instances. E. Adverb + adverbial clause : {a) one adverb + one adverbial clause : Gylf. 3, 9-10 ; enn J'ar sem landit hafSdi upp gengit, var far eptir vatn ; so also Gylf. 89, 12; {b^ one ad- verbial clause + one adverb. The Old Norse is very fond of using the particle "J>á" as a sort of convenient and reinforcing resume after any dependent clause, especially frequently after : (i) llie temporal clause^ of which there are no less than 75 cases, as Gylf. 10, 13 : enn er hann fell, J>á hljóp svá mikit blóS etc., 13, 9-10, 74, 20 ; 6Z, 6-7 ; 34, i ; 58, 5 etc. It will be remembered that we had 76, or almost just as many cases of temporal clauses without a following "])á." Considerably less frequently — 9 times only — the reinforcing adverb is used after : (2) The conditional clause : Gylf. 22, 13 ; 40, 24-25 ; 50, 16 ; 55. 8, 9 ; 77, 14- Olafs. 149, 25-26, 262, 8, 281, 18 ; 297, 30 : ef \\x, koniingr, vill leggja . . . ]>á matt I'll j^at svá vel gera. 'Olafs. 124, 27. 26 ORDER OF WORDS IN OLD NORSE PROSE. (3) After a causal clause 4 times : Olafs. 149, 25-26 ; 151, 26-28 ; 169, 1-2 ; 265, 13 : enn J'viat hvárigir vildu öSrum samneyta kristnir men, Jjá var J?at ráS tekit ; once in Olafs. 137, 29, the "J^a " is substituted by " Jivi " : J^viat \i\ tignaSir aldri . . . )>vi man nafn Y\\X víðfrægjast. (4) After a concessive clause "Jiá" follows twice, in Gylf., 27, 14 ; 17, I : enn svd sterk sem hón er, \k mun hón brotna. (5) After a local clause "J'á" follows once in Gylf. 33, 10 : hvar sem hón ríSr, þá á hón hálfan val. After the modal- comparative "svá " follows once in Olafs. 272, 21-22 : . . . svá sem hann var fjarlægr, svá vildi hann. {c) Several adverbial clauses of the same or of different char- acter. Any combination of clauses is possible after which, in what may be called the majority of cases, 41, \é. + inversion follows. If compared with the cases under 7, b, we find 85 per cent, with a following >á, Gylf. 40, 20-21 ; 46, 6-7 ; 54, 8-9; 54, lo-ii ; 63, 15-16 ; 66, 7-8 ; 96, 14-15 ; 38, 16 ; 54, 4. Olafs. 10, 9 ; 126, 25 ; 255, 31 ; 260, 28 etc. One illustration : Olafs. 284, 12-16 : meS Jvi at ek hefi svá upphafvit (causal), at ek hefvir fullkommliga stadt upp (consec.) . . . enn alt landsfólk er her rétt trúat (causal) . . . J)á skal ek eigi siSr ástunda etc. . . . F. Adverbial phrase -\- adverbial clause : in the majority of cases both the phrase and the clause are temporal : Gylf. 39, 1-2 ; 67, II ; Olafs. iii, 24; in, 25 ; 138, 27 ; 265, 5-6 ; 273, 22-23 ; 290, 9-1 1 ; 259, 28-29 ; 280, 3-4; 260, 17, and finally Gylf. 9, 16: ok inn fyrsta dag, er hón sleikti steina, kom . . . manns hár, 12 cases ; in the follow- ing 2 instances the clause is a relative : Olafs. 294, 9-10, and 151, 6-7 : enn meS fé J>ví, er Olafr konúngr gaf ]?áng- brandi, keypti hann mey. G. Adverb -\- adverbial phrase + adverbial clause : {a) the latter order plus inversiofi occurs only once in Olafs. 267, 13-14 ; enn norSr í sveitum, er J>eir fóru Jiar yfir, tóku trú göfgir menn ; and {b) in Olafs. 133, 13-14 a variation of that combination is found in a case of adverbial phrase -|- adverb -\- temporal clause : ok hinn næsta vetr eptir, er þeir Otto keisari skil5u,sat Olafr í ríki sínu á Vindlandi. (ít) Ad- verbial phrase + adverbial clause + adverb is represented by 10 instances: Gylf. 11, 7-8; 20, 13-14; 51, 7-8; 52, INDEPENDENT DECLARATIVE CLAUSES. 2/ 15-16 ; 72, 15 ; 92, 6. Olafs. 286, 12-13, 283, 13 etc. and 146, II : enn til J^ess at \i\ efist eigi um ]?essi andsvör, \-x máttu >at , . . hafa. (2) Inversion follows after an Object, Logical Object and Noun-Clause : A. Object which may be either : (i) An Accusative : {a) pronoun: hann, Gylf. 15, 17 ; 21, 7 ; 31, 5; 75, 13-14; Olafs. 250, 22; 298, 7-8; hit, Olafs. 299, 14; hana, Gylf. 16, 8; 44, 13; >ann, Gylf. 10, 15 ; 25, 7 ; Olafs. 1x5, 13-14; i'at, Gylf. 10, 8; 13, 6-7; 29, 2 ; 4c, 4-5 etc., altogether 20 times ; þá, Gylf. 9, 5-6 ; 49, I ; >ær, Gylf. 22, 8 ; 45, 7 ; >enna, Gylf. 96, i ; Olafs. 266, 23 ; ])etta, Gylf. 40, 7 ; Olafs. 271, ii : >etta veitti sauSa- maSrinn honum. Very interesting is the inversion after the adjective-pronoun j'essi, the modified noun following after the subject : in Gylf. 18, 12 : ok >essi segir hón nöfn j'eira. . . , if)) no?ninal accusative with or without a preced- ing adjective is found 24 times, as in Gylf. 15, 14 ; 17, 19 ; 24, II ; 68, 4; 100, 9 and so forth. Illustration : Olafs. 256, 18-19 • J'essa penningha hefir ]^i'i samandregit. . . . (2) Or a Dative : {a) of a pronoun : honum (6), Gylf. 76, 12 ; 13, 19 ; Olafs. 137, 12 ; 124, 23-24 ; 145, 4 ; 300, 2 ; henni (2), Gylf. 34, 4 ; 42, 17 ; >eini (4), Gylf. 8, 2 ; 99, i ; Olafs. 274, I ; 294, 14; >vi (1), Gylf. 53, 4; and once >ér (i), Olafs. 278, 4-5 : Jiér er nú kunnig ætt okkur ; {b) nominal dative with or without an attribute occurs in Gylf. 38, 21 ; 83, 4; 79, 12; Olafs. 264, 10; 130, 7-8; 145, 11-12 : konúngi gjörSist forvitni mikil ; 6 times. (3) Or a Genitive {a) pronominal : fess, Gylf. 78, 3 ; 81, 2 ; Olafs. 143, 19 ; ]iessa, Olafs. 131, 25 ; þeirra, Gylf. 30, 16 : ]>eirra ma hann eigi missa. ... {fi) nomifial genitive moáx- fied by an adjective-pronoun in Olafs. 130, 25 : J^essa striSs getr HallfreySr í Olafs drápu {c) the genitive of the definite article himi used absolutely once in Olafs. 268, 19 : enn bins vil ek eggja, at \\x brennir kirkjuna. 7 cases. (4) Or two accusatives : Gylf. 11,6: grjót ok urSir gerSu Jeir af tönnum. 28 ORDER OF WORDS IN OLD NORSE PROSE. (5) Or tivo datives: Olafs. 264, 12 etc.: enn mér ok minu hyski hefir hann veitt bruna. . . . (6) Or NomÍ7iative and Accusative^ instead of Accusative + accusative in apposition in the interesting anacoluthon of Gylf. 9, 9-10 : inn gamli hrimj'uss hann kolkim vér Ymi. Both Wilken and Jonsson agree in the reading. Other- wise one might be tempted to read " inn gamia hrímþuss, hann . . ." The only cases in which Inversion does not follow after an object are again to be found in the poetic passages, as in Grimnis- mál 14 : hálfan vál hón kyss hverjan dag, enn hálfan Óðinn á," or in Grímn. 18 "enn þat fair vitu," or in VafþrúSnismál 41 : vál þeir kjosa etc. In connection with this it is proper not to overlook the peculiar constructions in Gylf. 12, 11 : mikit ^ótú. mér ])t\x hafa J>á snúit til leiSar and, again, in Gylf. 35, 8 : allmikit Jjykki-mér guSin eiga undir gæzlu eSa trúnaði Iðunnar etc. Are "])ykki mér" and "J^otti mér " verbs in the singular and thus im- personal : " methinks, tnethought " ? Wilken, in opposition to Jonsson, as if to show how closely coupled the two words are, connects them by a dash, not only 38, 8, but also 96, 11 and 98, 20. That would leave little doubt as to his conception of the passages just quoted. But if the above-mentioned verbs are sin- gulars and impersonal, the Old Norse order is wrong, according to what we said before, that after an object : mikit = Grosses and allmikit = uberaus Wichtiges inversion should follow, and, thus, the order should be : mikit ])ykki mér hafa J)eir ]>á, as in German : Grosses-diinkt mich-haben sie verrichtet etc. In view of the strictness with which the law of inversion is carried through in Old Norse prose, I do not hesitate to regard the forms " >ykki and J>ótti " as Subjunctives, in the sense of the potentialis, and as Plurals ' at that, and consequently I look upon the above-men- tioned cases as personal constructions : magna videntur mihi ei fecisse. . . . B. Inversion follows after a logical object in the inserted clause without any exception, which is the rule not only in the Gerfnanic, but also in the Romance, Slavic, and many other ^ Poestion, on p. 348 of his Einkitung in das Studiuin des Altnordischen, tries to get out of the dilemma, explaining that in the 3d person singular þykki is used instead of þykkir etc. What he says about the subjunctive refers to þykki instead of þykkir. INDEPENDENT DECLARATIVE CLAUSES. 29 dialects. In the Germanic languages the words quoted in the oratio recta, no matter whether they reproduce the complete statement or thought or whether they are broken by the inserted clause, stand to the verb of the latter clause in the relation of object + predicate, after which the sub- ject must follow if it is expressed, as : Gylf. 59, i, eigi þarf ek — segir hann — at spyrja \\\i at nafni. " Eigi \z.xi ek " is logically as well an object of "segir hann " as the whole : " eigi \2lXÍ ek at spyrja j^ik at nafni." Not only the verbs of saying, of which there are numerous cases (segja, kveSa), but also the verbs of thinking are used in such a construc- tion : Gylf. 56, i: fair — vænti ek — . . . kunni segja , . . Gylf. 26, 5 : enn Ijósálfar einir — hyggjum vér — at nú byggvi ]>á staði ; also Gylf. 94, 4. C. Inversion after Noiin-Clauses practically does not occur, there being no instances of the substantive (or adjective) clauses preceding an independent clause to which it stands in relation. The only case that belongs here is Gylf. 68, 16 : enn sá er Logi heitir, J^at var villieldr. But here, after the adjective clause, the subject is for the sake of emphasis used twice, sá and J^at denoting one subject. The regular order is natural here and is also employed in such cases in German : " und der, welcher Logi hiess, — der or das war das Feuer." Inversion after the Adjective-clause in a combination of : D. Object + Relative Clause is found in Olafs. 279, 3-5, and in 264, 1-3 : Alia J)essa luti, er sá fjandi hafSi talat, sag^i hann syni sinum. The object is given twice in Olafs. 262, i8-2o : ok alia \k luti, sem Jiit segit af honum, slikt hit sama flytr hann af ykkr. " Here, perhaps, also belongs the peculiar anacoluthon in Gylf. 24, 7 : sú dögg (instead of \k dogg) er J'aSan af fellr á jörSina, \3X kalla menn hunangs- fall," Compare also Gylf. 9, 10. (3) Inversion proper, finally, follows after a predicate noun, the latter being either : (a) A Substantive (42 cases) as : Gylf. 47, 9-10 : geysi mi- kit mein var hánum \zX, so also Gylf. 14, i ; i4> 15 I 27, 19 ; 50» 3 ; 50, 15 etc., etc. 30 ORDER OF WORDS IN OLD NORSE PROSE. (b) Or an Adjective (17), as : mikill er Óðinn fyrir sér, Gylf. 51, 12; enn meiri muntu vera, Gylf. 62, 10; so also 48, 2 ; 70, 4 ; 96, I ; Olafs. 102, 5 ; 118, 29 ; 152, 14-15 etc. Here also belong Gylf. 42, 15 : önnur er Saga, and Gylf. 42, 16 : JiriSja er Eir etc., etc., in Gylf. 27, n the numeral tólf : tólf eru æsir góðkunnigir ; and also 87, 6, where the adjective is attributive : margar eru J^ar vistir góSar. (c) Or an Infinitive (15 times) : enn fylgja ma ek ]?ér, Gylf. 5, 1-2 ; 47, 43 ; 52, 12 ; 61, 7 ; 82, 4 ; Olafs. 27, 2 ; 259, 13 etc., etc. (d) Or a Participle (12 times) : set muntu hafa, Gylf. 40, 2 ; goldit var hánum þetta, Gylf. 78, 7 ; 61,5; 56, 10 ; 36, 3 ; Olafs. 130, 13 ; 305, 9-10 etc Inversion proper in Old Norse prose is thus not a phenomenon that may or may not appear, as in Anglo-Saxon, and, to a large extent, in Old Saxon and even in Old High German, but it is a law which, as in Modern German, is carried through to the letter allowing no exceptions. But however logical and precise, sharp and compact the " Inversion proper " be, the language which raises such a construction to a supreme law must suffer consider- ably under the disadvantage of working like a machine ; and if, as in the case especially of Old Norse, the poetry breaks so often the fetter of an order so philosophic, it is not only and exclusively due to alliterative considerations, for the rhetorically most beauti- ful languages are not bound by such a law. Before closing this chapter on Inversion it must be said that if the predicate is a compound verb, the subject — generally speak- ing — stands between the two constituent parts of the predicate, as Gylf. 20, 5 : J^ar skulu guSin eiga dóma sína, and so forth. But, on the other hand, there are a great many cases — at a rough estimate about 30 per cent. — in which the subject stands after the compound form, as Gylf. 15, 24 : J^aSan eru komnir Jjessir úlfar ; Olafs. 109, 15, 119, 5-6, 145, 7 etc., etc. As regards the final position of the compound verb in the inverted order,' it must suffice to state that it is found considerably less frequently here than in the normal order. Old Norse, strict as it is, preserving thus within the province of inversion, a comparatively greater freedom than Modern German. ' I refrain from presenting the material for the latter point as well as for the former, because it is too voluminous, and the truth may be verified by looking up some 3 or 4 pages. INDEPENDENT DECLARATIVE CLAUSES. 3 I C. RHETORICAL INVERSION. In a treatise like this, which deals exclusively with the syntac- tic norm of the Old Norse order of words, the rhetorical inversion can receive but little attention, and a few remarks must suffice. On almost every page instances of inversion are found in which no verb-modifier precedes the subject. They are, indeed, so fre- quent as to form a secondary mode, and by no means a very unimportant one, of expressing the relation of Subject and Predi- cate. Ries ' calls it "die ungerade Folge in freier Anwendung," and devotes fully 30 pages to its discussion. The rhetorical in- version extensively used in Anglo-Saxon and Old High Gerfnan, and especially in the Slavic dialects, and appearing also in Modern German and Moderti English : " Sagte der Königsson," " Said the prince," — may in the majority of cases be reduced to the law of analogy ; in others to the omission of an understood adverbial expression of inverting power ; in others, again, to the pictur- esqueness and vividness of the style, which is fond of an occa- sional change in order to prevent monotony and to heighten the effect. The adverbial modifier of a preceding clause is so strong that its effect extends itself to the following sentences, which is without an introducing adverb : Gylf. 58, 13-14 : enn of miSja nótt varS landskjálfti mikill, gekk jörðin undir ]ieim skykkjum ok skalf húsit ; Gylf. 80, 9-10 : Ipá váru teknir synir Loka . . . brugSu æsir Vala ... ok reif hann í sundr. In both instances there is a group of 3 sentences : the first sentence has an adverb- ial expression, then comes the second with the rhetorically in- verted order, and is followed by ok -\- sentence. So further Gylf. 74, 16-17 ; 79, 11-12. In the rapid narrative and brilliant description of the events connected with " Ragnarökr " the author has no time, as it were, to connect the actions by the calm "ok": Gylf. 83, 11 etc.: æsirnir hervæða sik, ok sœkja fram á völluna ; ríSr fyrst ÓSinn, . . . stefnir hann móti etc. etc. — In the simple enumera- tion as : Gylf. 10, 6 : hét einn OSinn, annar Vili, J>riSiVé ; Gylf. 13, 3 : hét karlmaSrinn Askr enn konan Embla ; so also Gylf. 13, 1-2 : gaf inn fyrsti önd, annar vit, j^riSi ásjónu etc., the predicate is once, at the beginning of the first sentence, empha- sized and then simply omitted. ' Quellen und Forschungen, vol. xli., pp. 12-42. 32 ORDER OF WORDS IN OLD NORSE PROSE. Very frequently the adverb, instead of preceding the predicate, is found immediately after it and in front of the subject. In such cases we may almost speak of the retroactive effect of the adverb ; as Gylf. 57, 12 : var þá annarr haltr eptra fœti ; Gylf. 79, 16-17 • ferr \á. Loki fyrir netinu ; Olafs. in, 1-2 : for >á Ríngr konúngr aptr. . . . Olafs. 121, 9-10: gengu síðan saman fylkingar. . . . In Gylfag. 58, 14-17 : ^á stóð J)órr upp ok hét ok leituSuz ok fundu, settiz pórr. . . . further Gylf. 59, 3 ; 67, 9 etc., the " >á " is omitted. In a few cases inversion seems to be used to indicate a causal relation : Gylf. 57, 12-14. Thor saw it and said that the peasant or one of his family had not handled carefully the goat's bones, for he recognized : " kennir hann," at brotinn var lærleggrinn ; so also Gylf. 59, 2 etc. The rhetorical inversion is not so frequent in the Younger Edda as in the Olafssaga and in the Eyrbyggjasaga. Its rhetorical effect if used for the purpose of parallelism and chiasmus is considerable. The Old Norse independent clause, even in spite of the strictness of its "inversion proper," com- mands a certain freedom with which the Modern Germanic dia- lects cannot compare. II. INDEPENDENT INTERROGATIVE CLAUSES. The rules and laws which in the declarative (positive or nega- tive) clause required an " Inversion proper," are just as valid in the interrogative sentence. Accordingly, we should expect and we also find inversion in all cases where the interrogative word is either : A. An Object^ as : hvat leik vilit >ér nú bjóSa mér? Gylf. 65, 21 ; or without a following noun : hvat hafSiz Allföör ? Gylf. 17, II ; 29, 14-15 : hvat hafa J'eir gert ? Olafs. 134, 23 : hvern dyrSkit þér ? Further illustrations are : Gylf. (>,^\ 8. 9 ; 29, 14 ; Olafs. 135, 3 ; 256, 17-18 ; 298, 2-3. B. Or an Adverb or an adverbial expression, as : hvar er sá guS ? Gylf. 6, 8 ; 9, 11 ; 20, 3 ; Olafs. 261, 19 ; Gylf. 12, 13 : hvaðan kómu menninir J^eir ? 26, 6 ; 77, i ; Gylf. 9, 2 : hvernig ÓXU ættir >a5an ? 1 2, 3 ; 40, 8-9 ; 98, 21 ; Gylf. 14, 15 : hversu styrir hann gang solar? Gylf. 50, 14-15 : hvi spyrr \ú dregit á braut hanzka minn ? So further : Gylf. 60, 21 ; 61, 3 ; 70, 7-8 ; 77, 1-2 ; 78, 7 ; 89, i etc., or without 34 ORDER OF WORDS IN OLD NORSE PROSE. an interrogative particle : Gylf. 22, i ; brennr eldr yfir Bifröst ? further : Gylf. 16, 7 ; 62, 9-10 ; 65, 5-6 ; 73, 3-4 ; 74, 5-7 etc. etc. The independent double question is treated the same way as the simple question, as Gylf. 89, 1-2 : hvárt Ufa nökkur guSin J'á, eí5a er \z. nökkur jörí5 eða himinn ? so also Gylf. 49, 6. The " eSa " is not necessarily preceded by a question : as Gylf. 65, 17 ; auð- sætt er nú at máttr ]:>ínn er ekki svá mikill. . . . eða viltu freista um fleiri leika ? " so further : Gylf. 34, 9 etc. etc. etc. III. INDEPENDENT IMPERATIVE CLAUSES. As in the interrogative clause so also in the case of the impera- tive clause, the main interest concentrates itself about the verb. The subject is again well known, and for that reason not infre- quently omitted, as Gylf. 61, 11 : " hverfit aptr " instead of: " hverfit Jér aptr " ; Eyrb. 107, 23 : " enda flyjum nú allir " instead of " enda flyjum vér nú allir." What Ries says about the Heliand is also, to a large extent, true in regard to the other Germanic dia- lects : " Die mit der fortschreitenden Entwickelung der Sprache zunehmende Neigung zu differenzieren, verschiedenem Inhalt auch verschiedene Form zu geben, musste die . . . abweichende Stel- lung begiinstigen, welche Heische-und- Fragesatz von der breiten Masse einfacher Aussagesatze auch der ausseren Form nach abhob." ' We shall have to divide the imperative clauses into those of the first, second, and third persons : A. Imperative clauses of the first persofi (Adhortativus) have the inversion without any exception, Olafs. 136, 14 : biðjum vér nú allir þann sama himna gu5 ; Olafs. 136, 16 : löggjumst vér niðr, etc. Here perhaps also belongs Olafs. 127, 2 : heyra viljum vér ! B. Imperative clauses of the second person (Command and re- quest) have always inversion, no matter whether they are or are not introduced by an adverb. Illustrations of the former charac- ter are : Olafs. 126, 14 : nii gefit þér ráS ; also Olafs. 263, 27. A conditional clause precedes + >á : Olafs. 256, 6-7 : . . . J^á fa fú ok lát hann lausan. In connection with " skulu " the impera- tive occurs : Gylf. 5, 2 : skaltii Jiá spyrja hann nafns sjálfr, so also Gylf. 46, 21: ok nú skaltú fara ok biSja hennar mér til hanSa. ' Qtiellen unci For sch., v. xli., p. 57. INDEPENDENT IMPERATIVE CLAUSES. 35 Without an introducing adverb : Olafs. 104,25 : gjöritj'ér allir samt >vílíkt ; Olafs. 130, 17 : hafit J'ér >ökk ; Olafs. 137,24: heyr }?ú, Ólafr ! further : Olafs. 138, 7 ; 152, 2 ; 264, 17 ; 302, 30 ; Eyrb. 93, 21 ; Gylf. 61, 8. If the subject of the imperative clause is strongly contrasted with another subject contained in a previous clause and is thus relatively weightier than the predicate, its emphatic position will naturally also determine its irregular position, and as the unem- phatic subject follows the verb, it will now, as an emphatic sub- ject, have to precede it : as Gylf. 59, 13, 14 : \z. mælti Skrýmir til J>órr, at hann vill leggjazt niðr ok sofa : enn \éx takit nestbaggann ok búit til náttverðar ySr. The contrasted subject of the predicate is, as far as can be determined, in such cases always introduced by the contrasting or adversative conjunction " enn." Gylf. 34, 21-22 (in Jonssons edition of the Prose Eddd) : kunna mun ek j^ar af at segja, enn "J^ú " skalt nú fyrst heyra ' ; further : Olafs. 280, 13 ; and finally Olafs. 305, 20-21 : nú mun ek gera bál mikit, enn " ])it " J?órr gángit J^ar at sinum megin, etc. etc C. Imperative clauses of the third person (optative) do not occur very frequently and always show inversion : Eyrb. 19, i: fári hann \i.. The " ok " precedes, but has no influence upon the order in Gylf. 66, 18 : ok fáiz JJÓrr vi5 hana, and also Olafs. 297, 28 : ok fáe \ér kallit mik, gangi til einhverr ok fáiz viS mik. . . . IV. CO-ORDINATE CLAUSES. The co-ordinate conjunctions which occur on the 200 pages that were examined for that purpose (100 pages of iho. Prose Edda and 100 of the Olafssaga) are: '' ok," " enn," " >ó," " enn >ó," " heldr " ; "eigi at eins . . . heldr"; "bæSi . . . ok," " hvárki . . . né." Their influence upon the order in the co-ordinate clause is not uniform. A) "ok." Heyse is indignant at and, in fact, attacks and con- demns the Modern German inversion after " und," calling it ' Wilken's reading is on page 35, lo-ii. 36 ORDER OF WORDS IN OLD NORSE PROSE. "veralterer Kanzleistil " and stating that in the better prose it affects unpleasantly the German " Sprachgefiihl, und unser aes- thetisches Gefiihl." ' This view has been generally adopted by Modern Grammarians.' Poeschel, on the other hand, after a thorough scientific investigation comes to the result that the inver- sion after "und " is a phenomenon well known in the oldest stage of the language and extensively used in its other periods.' Mogk discussing the same subject'' in regard to the Scandinavian lan- guages says : " In der altesten Zeit ist die Umstellung von Zeit- wort und Nomen — after " ok " — durchweg neben der gewöhn- lichen Wortfolge herrschend." In regard to the " Gylfaginning" and " Bragarœður " as well as to the " Olafssaga" the statement of the distinguished scholar will need to be modified, and as 200 pages of ordinary Old Norse prose may well serve as a criterion for the whole Old Norse prose literature in a question like the one under discussion, I would differ with Mogk and venture to state that inversion after "ok" in Old Norse prose is not only "neben der gewöhnlichen Wortfolge herrschend," but, on the contrary, that inversion is most decidedly the rule, the normal order appear- ing only in exceptional cases. Out of the 138 cases in the Younger Edda, in which " ok " introduces a co-ordinate clause containing subject and predicate, inversion is found 128 times, leaving thus only 10 illustrations of the normal arrangement ; while in the Olafssaga out of 109 cases 98 show inversion and 11 the normal order. To quote here all instances of inversion seems almost impossible. Gylf. 3, 8 : ok drógu öxninir J>at land ; further illustrations for 25 pages of Gylf. are : 3, 12 ; 4,3 ; 5, 3 ; 5» ^ ; 5, 8 ; 6, 15 ; 8, 6-7 ; 8, 17 ; 8, 18-19 ; 10, 5 ; 10, 17 ; 1 1, 10 ; 13, 4 ; 13,7-8; i3>i4; 14,6; 14,13; 14, 20 ; 15, 18 19; 17, 2; 17, 5; 17,20; 18,2-3; 20,2; 20,14; 22,5; 22,10; 23,2; 24, 15 ; 25, 2-3 ; 25, 13 ; 26, 2 ; 26, 3-4 ; 27, 4 ; 27, 7 ; 27, 8 ; 27, 16 ; 28, 2-3 ; 29, ID ; 29, 11 ; and so forth on almost every page. In Olafs. on 25 pages : 102, 28-29 : ok J>ókknaSist hvárt öSru vel." " ' Heyse : Deutsche Grammatik ; Hannover, 1886, p. 274, in the elaborate Anmerkung. ^ Also by Whitney : Germatt Grammar, Revised (compendium), p. 258, special rules. ^ In the Einladungsschrift der Fiirstcn- und Landesschule in Grim7na, 1891. ^ Indogernianische Forschungen, volume iv., p. 388 etc. "Inversion von Subject und Predikat in den nordischen Sprachen," deals almost exclusively with inversion after " ok." CO-ORDINATE CLAUSES. 3/ 104, 19-20 ; 105, 6-7 ; 106, 26 ; 107, 14-15 ; 107, 22 ; 109, 27 ; 110,21; 113,15; "4,7; Í16, 15 ; 116,17-18; 117,16-17; 118, I ; iiS, 25 ; 118, 26 ; 119, 6 ; 121, 10 ; 121, 11 ; 121, 13 ; 121, 25-26 ; 124, 24 ; 125, 28 ; 126, 13 ; 127, 17 ; 127, 20 ; 130, ^ ; 13°, 15 ; 130, 18 ; ^3°, 21 and so forth. It must be remembered that "ok" has not only the simple conjunctional force. It is, in fact, like the German *' auch," with which it is etymologically related, much more an adverb than a copulative conjunction, the Old High German " unte " and the Anglo-Saxon " ond " not being truly represented at all in Old Norse, but merely substituted by the "ok." The purely ad- verbial force of the " ok " is, of course, best exemplified in all those cases in which it is not the introductory word, as Gylf. 21, 6 : hón heitir ok Asbrii ; Gylf. 28, i : hann heitir ok Val- föðr ; Olafs. 109, 17 : ])á var ok gjör kirkja í Arósi etc. In all these cases the " ok " could well be translated by the German " auch," but never, of course, by the " und." It is thus not un- safe to say that the inversion after " ok " is in by far the majority of cases due to the universal law of Old Norse inversion after a preceding adverbial expression, which the " ok " was also felt to be. Before going over to the normal order after "ok," a few words about Anglo-Saxon might not be without interest. When com- pared with the inversion and lack of inversion after many ad- verbial expressions in Anglo-Saxon, it seems that the inversion after " ond " is not so very rare : Anglo-Sax. Chronicle : 709 : ond wæs totæled . . . Westseaxna land. 836 : Ond feng AeJ^el- wulf Ecgbrehting to Westseaxna rice. 875 : Ond for Godruns. . . . Kube calls the just quoted illustrations " einige schwer zu erklarende Falle der. Inversion." * Normal order after " ok " in the Gylf. is found in : 10, 6-7 ; 16, I ; 16, 3 ; 31, 7 ; 34, 7 ; 44, 6 ; 68, 3 ; 90, i ; 97, 9 ; 98, 13 ; and in Olafs. 102, 5-6 ; 115, 28; 119, 25 ; 125, 27 ; 127, 9-10; '^ZZ, 8-9 ; 138, 4 ; 141, lo-ii ; 142, 8 ; 146, 7-8 ; 295, 20. To account for the normal order after " ok " in a dogmatic way is hardly a safe thing to do. But would not the mere state- ment of two or more simultaneous actions coupled together by a co-ordinate conjunction more likely be expressed by the normal order ? In such a case the " ok " would have a much more con- ' Emil Kube : Die Wortstellung in der Sachsenchronik, Jena, 1886, § 6. 38 ORDER OF WORDS IN OLD NORSE PROSE. junctional force than if it were used as a particle to express pro- gression and succession of action, in which latter case its logical force is that of an adverb. For instance, in Gylf. 5, 1-3 : hann svarar ok segir ... ok sneriz sá maSr . . . the action is pro- gressing and instead of "ok " we might as well have used : }>á, or " eptir ])at." On the other hand, in Gylf. 44, 5-6 : hón gætir dura ok lykr aptr ... ok hón er sett til varnar á I'ingum . . . etc. Syn's watching the door, her closing it and her being placed there to prevent falsehoods are, of course, not meant to be suc- cessive actions. The same is further true of Gylf. 16, i ; 16, 3 ; 31, 6-7 : hann er svá fagr ok bjartr, at ... ok eitt gras er svá hvítt, at . . . (perhaps here also influence of parallelism) 34, 10 : hann er djarfastr ... ok hann ræSr ; 97, 8-9. In other cases, again, the subject is emphatic, and having the greater stress it requires the prominent, irregular and emphatic position, as Gylf. 98, 11-13: J'eir biSja Suttung ok bjóSa hánum ... til sættar i föSurgjöld mjöSinn dyra ok ">at " — namely the offer of the important mead — verðr at sætt meS >eim ; so further in Gylf. 10, 6 : ok J)at er min trúa, at. . . . 90, i : ok hitt mun >ér undarligt þykkja, er. . . . etc. In both cases the subject in the prominent position heightens our interest for the following de- pendent clause which is logically a subject. Gylf. 68, 3 is per- haps a dependent clause. What has been said about the normal order after " ok " in Gylfag. is also true in regard to the Olafs- saga. B) " enn " in 118 out of 122 cases in the Younger Edda requires the normal order, as Gylf. 3, 4 : enn sú kona var ein af ása-ætt, so further : 3, 7 ; 6, 17 ; 8, 18 ; 9, 8 ; 9, 13 ; 15, 17 ; 17, 7 ; 20, i ; 20, 12 ; 20, 17 ; 24, 15 ; 30, 3 ; 30, 8 and so forth. In the Olafssaga all the 90 instances show normal arrangement, as 102, i : enn ek segi >at, further iii, 6 ; 113, 11-12 ; 115, 1-2 ; 121, 24-25 ; 123, 20 ; 126, 16 ; 127, 5 ; 130, 17 ; 133, 9-10 etc. In regard to the inversion after " enn," which occurs 4 times in Gylf. 23, 11: enn er >at sagt ; 28, 4 : ok enn hefir hann nefnz á fleiri vega ; 66, 16 and 69, 3 : ok enn mælti hann, it must be said that the " enn " seems to mean, " moreover, further, besides," being thus strictly adverbial in force. In the latter meaning it occurs frequently in the middle of the sentence, as in Gylf. 23, ID : svá er enn sagt ; Olafs. 130, 8 : ok spurSi enn Ola, and so forth. DEPENDENT CLAUSES. 39 C) "l'ó," generally found in connection with " ok " or "enn," is a conjunctional adverb in character and thus always takes in- version, as Gylf. 67, 20 : enn ]>ó veit ek ; 29, 5 : enn J'ó er )>er l>at skjótast at segja ; further : 40, 18 ; 64, 19 ; — In Olafs. " Jjó " in- troduces more frequently a clause, as 305, 30 : ok fó geng ek til fessa profs. . . . 133, 11-12 : enn \ó hugfesti hann, etc.; further 135, 11 ; 138, 1-2 ; 146, 18 ; 288, 23 ; 298, 11 ; 303, 5 ; 305, 27 ; 306, 17-18. . . . D) "heldr," being also essentially conjunctional adverbial, re- quires inversion in Gylf. 54, 14 : heldr synjaSi hann hánum at byggva. . . . The clause preceding it is negative. As a second member of the combination : E) " eigi at eins . . . heldr " it also requires inversion in Olafs : 267, r : eigi at eins )'ar í næstum sveitum, heldr fóru J'eir viSa um island; and 281, 20-21 : eigi at eins stundligu riki, heldr gerir hann sina )'ræla bræSr sins, etc. F) "bæSi ... ok " is found in Olafs. 155, 19-20 : bæSi var hann hárr ok digr, in which the " bæSi " is followed by inversion while the " ok" connects only a predicate adjective. G) " hvárki . . . né " is used in J^onssorCs Edition of the Edda : 41, 10 : hvárki svaf hann né drakk. Wilken reads : 46, 11 : ekki svaf hann, ekki drakk hann. . . . Summing up the order of words after the co-ordinate particles, we find that only " enn " requires the normal arrangement, while the other particles, with a few exceptions after "ok," are followed by inversion. In Old Norse, then, it would seem that the only co-ordinate conjunction pure and simple, without any admixture of adverbial force, is "enn," while the remaining particles, as is frequently the case with the German " doch, jedoch, also, auch," * are regarded as adverbial conjunctions or conjunctional ad- verbs. V. DEPENDENT CLAUSES. As regards the dependent clauses, Old Norse is comparatively free, much more so than any other Early Teutonic dialect. In Anglo-Saxon, in by far the majority of cases, 82 per cent if the verb is simple," the predicate holds an extreme end position as in Modern German, an order which is better known as ^^ transpo- ' Heyse : Deutsche Grammatik, page 275, 3. Anmerkung. ' Modern Lang. Associat. Publications, 1893, p. 229. 40 ORDER OF WORDS IN OLD NORSE PROSE. sition." In Old-Saxon has "die verschiedene Behandlung, die die Stellung des Verbs zu den iibrigen Satzgliedern erfahrt . . . sich allmalig zum wesentlichen, fiir die beiden Satztypen charac- teristischen Artunterschied herausgebildet." ' Gothic occasionally shows independently of the Greek original the extreme end- position of the auxiliary verb ; for instance, almost always in the phrase : hausidedu)? ]?atei quij^an ist, Matth. 5, 43 etc., for which the Greek has the Aorist Passive," but also in the independent clause the auxiliary is, in opposition to the original, at liberty to follow the verbal noun : ni nauhjianuh galagij's vas, John 3, 24 ; etc. Becker^ remarks that the transposed order appears as a law all through Old High Gerinan literature and that exceptions are very rare. In view of a law which may be styled typically Ger- manic, it is rather strange and interesting to observe that the Old Norse by no means shares, generally speaking, the fondness for the transposed order of the dialects just mentioned. On the contrary, it treats its dependent clauses very much like its inde- pendent clauses, preserving for the transposed order a propor- tionately insignificant amount of dependent clauses. The most important orders are thus the " normal " and the " inverted " ar- rangements. In the latter order, however, we have to discrimi- nate between (i) inversion caused by a subordinate conjuntion only, an order which is employed very rarely and only for rhetori- cal considerations — being bound and subject to no laws and rules — as a mere variety and deviation from the normal monotony ; and (2) inversion caused by some word modifying the predicate of the dependent clause, a phenomenon so familiar in the Old Norse independent clause, as an order to be strictly observed and followed. The most convenient way of discussing the order of words in dependent clauses is to consider the normal and in- verted arrangements, with all the peculiarities and exceptions, in one chapter, and the transposed order in another. A. Dependent Normal and Inverted Orders. If a subordinate conjunction introduces the dependent clause, the order is, generally speaking, normal : x -|- Subject -j- predi- cate ; all dependent clauses sharing equally in that arrangement. ■ Quellen und Forschuiigen, v. 41, p. 67. 2 -^KovoazB OVH éfjpTJQ/f. '' Becker's Deutsche Grammaiik, p. 436. DEPENDENT CLAUSES. 4I If the predicative combination is expressed hy coýula plus predi- cate noun, the latter follows the copula, as in the independent clauses, no matter whether it is a substantive or an adjective. If the predicate verb is modified by an adverb or by an adverbial expres- sion, the latter is bound to the position after the modified verb, as Gylf. 15, 13 : J^o at hón fari ákafliga ; 40, i : J'óttu vitir eigi áðr. . . . Olafs. 131, 12 : ok herjaði hvar sem hann kom vi5 land. The object is no less strictly bound to that law than is the adver- bial modifier : Gylf. 38, 6 : at hón skyldi skipta öllum vistum ; Gylf. 40, 20: Jiótt ek slita í sundr svá mjótt band; Olafs, no, 19 : enn er íngjaldr spurSi J^at. Olafs. in, 6 : til er \€\\ fundu Gorm konúng. If the predicate is a compound tense of the verb and is modified by an adverb or an adverbial expression, the same tendency pre- vails as in the independent clause to place them frequently between auxiliary and verbal noun : as Gylf. 79, i : svá sem net er síðan gört ; 79, 8 : at J^at mundi vél vera ; 89, 10 : er fyrrum höfðu verit ; 94, 6 : ... at Loki skal aldri lauss verða ; Olafs. 127, 6 : . . . at ek man her leggja. ... 139, 8-9 : ... at hann mætti þar grundvalla guðs kristni. The intermediate position is, of course, not obligatory, and, as in the independent clause, the adverbial modifiers are not infrequently placed after the modified verb : Olafs. 135, 18 : þá er Olafr hafSi verit 3 vetr á Vindlandi ; 139, 6 : at hann mundi fara í Garða ríki ; Gylf. 68, 2-3 : ef ek hefSa vitat áSr, ... 73, 14-15 : ... at hann skyldi standa upp á Jjingum. . . . If the dependent coinpound verb is modified by an object, the latter shows the same inclination as the adverbial modifier to be placed between the constituent parts of the predicate : Olafs. 138, 16 : sem hann hafði ]7essa luti séð ok heyrt . . . 139, 26 : J)au er mega öSrum nökkura hjálp veita ; 140, 3-4 : furr enn ek hefir yðr, koniingr, ok allan Jienna lyS leidt . . . 140, 28 : J'egar hann mátti nökkut atfærast. Olafs. 141, 8-9 : at móðir ySur man ])enna mann hafa fyrirsjeth forSum daga ; Olafs. 146, 24 : at >essi maSr mundi honum sanna luti sagt hafa, etc. The object is, of course, not bound to precede the verbal noun, but may just as well, although less frequently, follow it : Olafs. 139, 14 : . . . at konúngr let stefna fjölmennt þíng ; Olafs. 141, 13 : sá er . . . mundi prySa j^etta n'ki. . . . Olafs. 141, 19-20 : at údygS mundi undir búa okkrum kærleikum. Olafs. 142, 11 : at hann mætti njóta ]7eirra góðgernínga etc. . . . 43 ORDER OF WORDS IN OLD NORSE PROSE. In regard to the ^legation " eigi " ' in the dependent clause it must be said that, as a rule, it partakes of the most decided tendency of the adverb to follow the word it modifies, if the dependent verb is in a simple tense : Gylf. 8, 5 : ]>á er sd iss . . . rann eigi ; 40, I : ]5Óttu vitir eigi. . . . Gylf. 51, 7 : er Jieir drekka eigi ; 65, 1-2 : at stikillinn vill eigi . . . 65, 10 : ef ]>ú gerir eigi . . . 69, 15. . . . at J^ér komit eigi, . . . 74, 12 : )>vi at ek sé eigi. The negation in front of the dependent verb is, in the normal order, rare : Gylf. 47, 9 : at hann eigi átti ... 62, 12 : sá er eigi kunni . . . and finally, perhaps^ 93, 17 : at eigi soSnaði. If the dependent predicate is a verb in a compound tense, the negation is in by far the majority of cases, in 21 out of 22, placed between auxiliary and verbal noun, never after the compound verb : Gylf. 29, 11- 12 : ef jjú skalt eigi kunna frásegja ; 34, 19 : enn \k er æsir vildu eigi leysa hann ; 39, 6 : ef slik stórsmíSi mætti eigi halda hánum ; 39 16 : at j'eir mundu eigi fa bundit úlfinn ; 40, 6 : er ]5Ú matt eigi reyna ; 40, 24 : enn ef J^ú fær eigi ])etta band slitit ; further : 41, 1-2 ; 46, 21 ; 56, 5 ; 56, 13 ; 56, 16 ; 57, 13 ; 59,19 ; 64, 19 ; 66, 9 ; 67, 7 ; 68, 21 ; 77, 15-16 ; 78, 6-7 ; 96, 3. Only once " eigi " stands in the normal order between subject and auxiliary : Gylf. 44, 6 : >eim er eigi skulu inn ganga. It hardly need be said that, for the sake of emphasis, any modifier may be taken out of its normal place, and we shall see later on what changes the irregular position of said verb-modifiers brings about in dependent clauses. Summing up all possible combinations in the normal order of dependent clauses, of i subject, i verb, verb-modifier, we receive the following table : A. Dependefii Sifnple Predicate Verb. (i) Conjunction or x + subject + verb. (2) Conjunction or x + subject + verb^- verb-modifier. (3) Conjunction or x -[- subject + copula + predicate substan- tive. (4) Conjunction or x + subject + copula -f- predicate adjec- tive. (5) Conjunction or x + subject + copula -\- predicate substan- tive + verb-modifier. (6) Conjunction or x -j- subject + copula + predicate adjec- tive + verb-modifier. ' The remarks on " eigi " are based upon Gylfaginning. DEPENDENT CLAUSES. 43 B. Dependent Compound Predicate Verb. (i) Conjunction or x + subject + auxiliary + verbal noun. (2) Conjunction or x + subject + auxiliary + auxiliary infini- tive + verbal noun. (3) Conjunction or x + subject -)- auxiliary + verbal noun + auxiliary infinitive. (4) Conjunction or x + subject + auxiliary -\- verb-modifier + verbal noun. (5) Conjunction or x -j- subject -\- auxiliary -|- verbal noun -|- verb-modifier. (6) Conjunction or x + subject + auxiliary -\- verb-modifier -|- verbal noun -|- verb-modifier. (7) Conjunction or x + subject + auxiliary -f- verbal noun + predicate noun or adjective. (8) Conjunction or x + subject + auxiliary + predicate noun or adjective + verbal noun. There is no deviation from the normal order, if the dependent clause has two or more predicates and one subject ; or two or more subjects and one predicate j or, finally, two or more subjects and two or more predicates. The tables for the independent clauses may be consulted for that purpose, with the understanding, of course, that in each case the conjunction or x precedes the subject, (i) Substantive Clauses. {a) indirect declarative : after verbs of saying, thinking, be- lieving, after verbs of direct perception and after simple introductory expressions have all the normal order : svá er sagt, at LoSbrókar synir hafi rekit mestan hernaS í fornes- kiu . . . Olafs. 114, 24-25; ]?ótti heiSingjum . . ., at goSin voru reiS Stefni . . Olafs. 286, 13-14 ; ek veit, at sá guS ermáttigr . . . Olafs. 136, lo-ii ; þá bar, at J^eir vildu riSa til várjííngs í Hegranes . . . Olafs. 273, 3-4. Only once is there an inversion without anything to account for it in Eyrb. 5, 25-26 : ok sá J^eir, at skárust í landit inn firSir stórir ; while on the other hand, in all cases where by em- phasis the modifier of the dependent predicate verb or copula is torn from its regular position after the verb, and placed at the beginning of the clause immediately after the subordinating conjunction, Inversion appears as a strict rule with almost no exceptions. As a list of all the verb- 44 ORDER OF WORDS IN OLD NORSE PROSE. modifiers requiring inversion in the dependent clause, irre- spective of the character of such a clause, will be added to this chapter, a few illustrations must do in each case, and as many illustrations will be given as are essential for the establishment of the principle in all its bearings. Olafs. 125, 19-20 : hann spurði, at \z.x var fyrir Otto ; Gylf. 5, 15 : hann svarar, at fyrst vil hann vita . . . Gylf. 39, 7-8 : úlfrinn hugsaSi, at hánum hafSi afi vaxit. Gylf. 64, 13-14 ... ok hyggr, at eigi skal hann Jiurfa optar at lúta í hornit. Olafs. 295, 18: nil hefi ek heyrt, at viSa um heiminn sé haldinn annar siSr. Gylf. 68, 21 : enn \zX. veit triia min, at J>á varS J^atundr, etc., and finally Gylf. 16, 9 : kann vera at ]7at kallir J>ú regnboga. Nothing could be more precise and definite. All the various modifiers which require in- version in the independent clause have the same inverting power in the dependetit one, no matter whether it is an object or an adverbial expression, or a predicate noun, adjec- tive, infinitive or participle. (b) indirect interrogative clauses after verbs of inquiry, report, thinking, believing have all the normal order : hvi spyrr ]>ú eigi pess hversu margar dyrr eru á höllini, Gylf. 50, 14-15 ; . • . fur skömmu var mér sagt, hverr J>ú vart ok hverr J^ú munt verSa, Olafs. 148, 2-3, which as the direct question would be: hverr ertu ok hverr muntu verSa? JJorvaldr segir, hvat hann hafSi gert, Olafs. 270, 17-18; direct : hvat hefir j^ii gert ? hún vildi kjör af hafa, hvern hún skyldi eiga af J^eim mönnum, Olafs. 149, 10 : direct : hvern skal ek eiga ? enn j^at Jiótti ásum mikit undr, hversu stór björg sá hestr dró, Gylf. 53, 7 ; hann hl'ddi, ef hann næmi nökkur orSaskil, Eyrb. 13, i. While in the latter instance, which is introduced by a conjunction only, the normal order would be natural and expected, according to the rule that a mere subordinate conjunction requires normal order, we should expect, on the other hand, that in dependent interrogative clauses introduced by an object, or a predicate noun or an adverbial expression as the in- terrogative words, inversion would follow, as it does in almost every such case. But if there were no such exter- nal differentiation, we could not possibly discriminate be- tween an independent and dependent interrogative clause, DEPENDENT CLAUSES. 45 and this is, perhaps, the only, although logically the most natural, case where Old Norse prose becomes inconsistent in the strictness of its use of inversion. The instance which might be called an exception is found in Gylf. 4, 4-5, in the dependent double question : ok hugsaði hann hvárt J>at mundi vera af eðli sjálfra }>eirra, eSa mundi \yi valda guðmögn, after which a relative clause describing the "guSmögn" follows, the subject being thus forced to the end position for stilistic reasons. On the other hand, the several instances of Inversion in the dependent inter- rogative clauses all show that the verb-modifier which caused the inverted order in each case is tiot the interroga- tive word, but, on the contrary, that the interrogative word precedes such modifiers: Gylf. 67, 17 : ok spyrr, hvernig hánum J)ykkir ferS sin orSin ; "hvernig" is the inter- rogative word which precedes the object "hánum," the modifier which requires inversion ; but immediately after that we read : eða hvárt hann hefir hitt rikara mann etc. ; so also Olafs. 146, 28 : spurði hann, hvaSan honum kom sú speki ; Olafs. 284, 5 : konungr frétti, ef honum væri hugr . . . further: Eyrb. 21, 25-26 : Oddr spyrr, hvárt hrossum höfSu stolit útlendir menn ; Eyrb. 97, 4-5 ; and Gylf. 5, 17, and finally Gylf. 90, 4-5 : veit ek eigi, hvaSan >ér kemr >at. (c) indirect ifnperative-optative clauses, after verbs of com- mand, volition, desire, etc., are always expressed by the normal order : Olafs. 139, 3 : siSan baS hann biskup þann .... at hann færi í GarSaríki, . . . 136, 14-15 : biðjum vér nú, at hann skýli oss fyrir sitt krossmark. Eyrb. 93, 12 : enn f»urí5r vildi, at hón færi þangat. There is no case of inversion where the subordinate clause is introduced by a dependent particle only, but, again, if the particle is fol- lowed by some modifier of the dependent predicate, Inver- sion appears, as always : Olafs. 279, 27-26 : it was agreed, at ]'essu boSi skulu J»eir 1 mxóti standa . . . Olafs. 297, 25, Eyrb. 16, 14-15 : and the stipulation was, at >á skyldi >e- gar upp gjalda féit ; further Olafs. 296, 25-26 : it is my request, at j^enna mann lokkit J^er meS . . . fögrum orðum, and finally Gylf. 27,2: I prefer, at heldr spyrir >ii eitt sinn ófróðliga. 46 ORDER OF WORDS IN OLD NORSE PROSE. (2) Adjective clauses. Although the relative clauses have con- siderably more cases of transposition than any other kind of dependent clauses, it is nevertheless safe to say that the normal order is the general mode of expressing it, over 70 per cent of the relative clauses being treated that way. If the relative word is simply an object case of " er " inversion is not used, unless there be some modifier of inverting power after the relative object. Not even after the relative in- troducer "sem" or " er " depending on a post-positive preposition does inversion take place: "and she invites you" 1 höfuðborg j>essa rikis, "sem" hún séalf sitr "í," Olafs. 102, 18 ; sá gu5, er alia luti hefir skapat, Olafs. 127, 13-14; and 127, 15 : sinu eyrendi, J^vi "sem" l'ér farit " með." Olafs. 259, 18-19: . . . J^essi maSr, " er " hann sagði svá mikla frægS " af." Olafs. 112, 8-9 : ok kallaSi Knút af knúti })eim, " er " (Obj.) barnit hafSi meS sér, so further Olafs. loi, 27 : nema j'au " er " (Obj.) >ú hefir frétt, etc. If " er " is object, the predicate precedes the subject only if the latter is forced to the end position by an apposition or by adverbial expressions, appositive in character, as Olafs. 255, 16 : feirra dóttir var ]7orlaug, er átti Guðmundr hinn ríki á MöSruvöllum ; or Eyrb. 8, 19 : GerSr var dóttir, er átti ]?ormóðr goSi, son Odds hins rakka ; so further Eyrb. 9, 21 ; 9, 23 etc. Neither the " er " nor the "sem," although logically objects, have, as remarked above, strange as it may seem, in themselves inverting power. The constant use in Old Norse of " er " and " sem " as pure subordinate conjunctions after which no inversion follows, is perhaps the reason why the same words used relatively as objects were not felt to be true objects with the power of inversion, especially since as relative pronouns they lacked all inflection and external indication of the oblique case. But if the relative pronouns are followed by some word modifying the dependent predicate, the latter, of course, precedes the subject, that is, Inversion takes place, as Gylf. 9, 3 : trúi-])ér pann guð, er nú sagSir fú frá ? Gylf. 97, 8 : spyrr hann feirra hluta, er eigi kann hann orlausn, or Olafs. 114, 20-21 : . . . Englands, er áðr höfSu átt bans a^ttmenn ok frændr, etc. Other cases are : Gylf. 12, 2; 32,9; Olafs. no, 25 ; 115,23; 117,20; 120,5; 279, 6; 279, 8 ; Eyrb. 91, 18. DEPENDENT CLAUSES. 47 (3) Adverbial clauses : (a) Local adverbial clauses all have normal order : Gylf. 3, 1 1 ; par sem landit hafði upp gengit. Gylf. 71, 6-7 : pa sneriz J>órr á braut ]>angat, er hann sá öxnaflokk nökkurn. Gylf. 98, 23 : ok kam J'ar, er prælar slógu hey. Olafs. 131, 12 : ok herjaði hvar sem hann kom við land, etc. What has been said about the " er " and " sem " in the adjective-clause used objectively is also true if the same words are used adverbially. As introducers of a dependent clause they have lost the inverting power, and Inversion is used only if the local adverb of the relative clause is followed by some modifier of the predicate of the dependent clause, as Eyrb. 104, 20, Gylf. 89, 6 : ok byggja ]>eir á iSavelli, far sem fyrr var AsgarSr, further also Gylf. 20, 10 : enn önnur er með hrím-]'ussum, J^ar sem forí5um var Ginnunga-gap. Not counted here are the familiar clauses. . . . "j>ar, sem heitir Gimlé, or )>ar sem heitir á Kjallarst," in which cases the subject is omitted. (b) Temporal adverbial clauses have almost exclusively the normal order, if there is nothing to prevent it, as Gylf. 4, 9 : ok er hann kom inn í borgina, etc., or 13, 9 : ok ]'á er Óðinn settiz }>ar í hásæti ; 7, 14, etc., hversu skipaðiz, áðr enn ættirnar yröi eða aukaðiz mannfólkit. Here the second clause after " eSa " is inverted, so also after " ok " in Gylf. 17, 1-2 : \i. mun hón brotna ]'á er Múspells megir fara at herja ... ok svima hestar j^eira yfir storar ár. Other temporal conjunctions are "meSan " in Olafs. 305, 4-5 : . . . meÖan hann heldr trúlyndi við mik ; and " ]>egar", in Olafs. 306, 28 : J>egar er konúngr for út or firði. The temporal clauses are, of course, very numerous. Inversion occurs, again, in all cases where the temporal particle precedes a modifier of the dependent predicate verb or copula : enn er " petta " (Obj.) spur'ðu landsmenn, Olafs. n8, 5; enn er " bóndum (Obj.) kom þíngboð koniings, Olafs. 279, 20-21 ; áör enn " sagt " (pred. parte.) er allt )>at, Gylf. 31, i ; 79, 10; enn pa er " full" (pred. adj.) er munnlaugin, Gylf. 80, 17 ; and finally Eyrb. 18, 24: ok er mjök (adverb) leiö á kveldit, (c) Modal-comparative clauses (of manner and degree) have normal order, unless prevented by well-known reasons. 48 ORDER OF WORDS IN OLD NORSE PROSE. Gylf. 79, i: tók hann lingarn ok reiS á ræksna svá sem- net er siSan gört. Olafs. 138, 14-15 : ok muntu . , . öðlast meiri dyrS, enn J>ú hefir nú séð her. Olafs. 254, 9-1 1 : Ólafr heiti ek . . . sem |jú munt heyrt hafa getit. In the conditional-modal sense " sem " is used in Olafs. 290, 5: JJÓrólfr . . . elskaSi hann, sem hann væri hans son ; and so forth. As in all clauses Itiversion is caused by the fact that the modifier of the dependent predicate is removed from its normal place, so in Olafs. 118, 28 : svá driipir nil Danmerk, sem dauSr (pred. adj.) sé Knútr ; Olafs. 263, 18 : . . . sem furr (Adv.) var hann vanr ; Eyrb. 98, 2 : . . . sem fa (adv.) var siör ; Gylf. 40, 11 : ... sem nú (adv.) skaltu heyra ; Gylf. 40, 19-20 : " methinks, as if, sem önga frægö (object) muna ek af hljóta ; so further Gylf. 56, 9 etc. (d) Causal clauses are also expressed by the normal order, as Olafs. 102, 14 : fyrir J>vi at ]>éx farit meS friSi. . . . Gylf. 43, 9 ; er SÚ sök til J>ess, at hón gaf sér ýmis heiti, . . , sometimes " er " is used as causal particle: Gylf. 22, 14; 34, 8 ; 47, 6 ; Jjessi sök var til, er Freyr var svá vápnlauss, ... so also Gylf. 69, 10 ; sometimes " sem " is used in the same sense : Gylf. 75, 2-3 : OSinn bar j^eim mun verst J»enna skaSa, sem hann kunni mesta skyn. . . . Gylf. 78, 5-6 : almiklu kom Loki á leiS. . . . ]>\\ er hann varS eigi leystr frá Helju. — Here again, Inversion is used under the well known condition that the modifier of the dependent verb shall be placed directly after the causal particle, as Gylf. 26, 21. . . . ]:>vi at ]>etta (obj.) vitu allir menn ; Gylf. 28, I : hann heitir ok ValföSr \w\ at hans óskasynir (pred. noun) eru allir J'eir, etc. Gylf. 82, 5-Ó ; þá geysiz hafit. . . . fyrir J>vi at ])á (adv.) snyz Mi'SgarSsormr í jötunmóS, so also Eyrb. 84, 6-7 ; 96, 29-30 ; Eyrb. 95, 17-18: J>viat mér (Dative) segir svá hugr um ; Olafs. 258, 29. . . . J>viat litlu áSr hafSi hann hertekna sonu J'essa sama hertoga leyst. . . . etc., etc. (e) Final clauses {purpose) are expressed in the normal order, as Olaf. 127, 13 : I advise to fast in order that ... til )>ess at sá gu5 gefi oss at sigrast á Dönum. , . . Gylf. 78, II : gerði j'ar hús ok fjórar dyrr, at hann mátti sjá or húsinu í allar ættir, and one more illustration : Eyrb. 12, DEPENDENT CLAUSES. 49 1 : enn til )>ess, at )>eir vaeri vel sáttir ok vinir I'aían af, \k gjöríi hann, etc., etc. Inversion is, of course, used if the final particle is followed by a word modifying the dependent predicate, as : Gylf 24, 2-3 : ... til )'ess, at eigi (Neg.) skyli limar bans tréna eSa fiina. Gylf, 69, 7 : ok vannz hánum lengS til at jöröina (Obj.) tæki sporSr ok höfut5, and, finally, Gylf. 73, 10 : Frigg tók svardaga til l>ess at eira (pred. inf.) skyldu Baldri eldr ok vatn. (f) Consecutive- Residt-Clauses are treated like the other clauses. If there is nothing to prevent it, the normal order follows : as Olaf. 108, 7 : peir skiptu riki með sér svá at Lotharíus hafÖi Burgundíam ; Olafs. 138, 26: ok svá fékk honum mikils, at hans augu vóru full af tárum ; Eyrb. 84, 19 : fá vóru sár hans mjök gróinn, svá at hann var vel vápnfærr ; Gylf. 14, 17: þau váru svá fögr, at hann kallaÖi son sinn Mána. In Olafs. 154, 18 : J)á dreif lio til Haralds konúngs, svá at sveinn varö borinn ofrliði ok flyöi hann, inversion follows after the " ok " ; rather peculiar is the passage Olafs. 263, 27 : nii gjör \\x svá mannliga at \\x rek fa brottu, the order of which is normal, the verb "rek" being in the imperative form, and being, further, preceded by a subordinate conjunction. One is reminded here of the famous Gothic " saihvei ei saihvei," in which the second predicate is also preceded by an ap- parently dependent particle, although it will be remem- bered that both Braune and Douse treat the " ei " from an entirely different standpoint. As regards the Inverted order of the consecutive clause, it appears here under the same premises as everywhere else : namely, if the subordi- nate conjunction is followed by a modifier of the depend- ent predicate ; as Gylf 42, 10 : svá mikils virðu guðin vé sin, at eigi (Adv.) vildu fau saurga]>á með blóði úlfsins. Olafs. 105, 4-5 : hann veitti atsókn . . . svá haröa, at á skamri stundu (adv. phrase) brutu feir. . . ; Olafs. 118, 24-25. . . . komu margar krákur ok plokkaí^u hann svá, at af honum (prepos. + case) eru allar fjaðrar ; Eyrb. 96, 17 : sótti hón fá svá fast, at honum (object) gékzt hugr viö Olafs. 128, 16 : . . . svá at um morgininn eptir (Adverbial phrase -j- adverb) sá engi likindi Danavirkis. . . . Twice, however, in Gylf agimiing Inversion occurs without regard to the well- 50 ORDER OF WORDS IN OLD NORSE PROSE. established rule : 57, 20 : hann heröi hendrnar á hamar- skaptinu, svá at hvítnuðu knúarnir ; and 89, 15 : af }>essum mönnum kemr svá mikil kynslóS, at byggviz heimr allr ; they are very exceptional cases : in the former instance the rhetorical stress lies on the impressive verb ; in the latter on the subject, and the emphatic position — which is iden- tical with the uncommon and irregular one — removed the verb from its normal second to its irregular first place, and the subject from its normal first to its unusual second place, (g) Concessive clauses, also, generally have the normal order : Gylf. 15, 13. . . . )'ó at hón fári ákafliga ; Gylf. 40, i: })óttu vitir exgi áí5r J'essi tíöindi ; Gylf. 27, 14 : ok svá sem önnur gu'Sin eru máttug ; Olafs. 278, 5 : )>óat )'ú hafir eigi verit lengi Í Jiessu landi ; Olafs. 294, 4 : . . . l>óat sveinninn væri bun- dinn viS skipit ; Eyrb. 100, 7 : f oat menn væri skírSir. . . . Here, again, as in the other dependent clauses, Inversion is employed only under the condition that the verb modifier be torn from its regular position after the predicate ; as Gylf. 42, II : )>ótt svá segi spárnar ; Olafs. 291, 28 : fóat eigi furfti avlit inn at bera, and the illustration in Olafs, 104, 5-7 : . . . fóat eptir boði náttúrunnar (in the course of nature, of time) hafi farsælan oss lengi fylgjusavm verit ; so further Gylf. 40, 5-6 ; Eyrb. 11, 23 ; 95, lo-ii ; in Olafs. 294, 6, the inversion takes place contrary to the rule : Rögnvaldr hafSi svá um búit, at hann mátti eigi leysa sik, at \ó vóru lausar hendr hans. . . . ; it will be noticed that instead of ]'óat or fótt, the usual concessive particle, we have here : at J^ó ; but whether the inversion is due to that change is an entirely different question, which I dare not answer. Very peculiar and interesting is the inversion in Olafs. 300, 29-30 : enn sá guð er svá máttugr, at hann J)vær af manninum í skírninni allar syndir, ** aldri hefir hann áSr svá lit gört " = habe er auch noch so Böses je vorher gethan ; although this is, to my mind, the only way to translate it, one thing is especially perplexing : the strange indicative of " hafa " instead of the subjunctive which is used in the kindred dialects. The "aldri," too, at the be- ginning seems no less out of place. It may be an awkward attempt at a construction so familiar in English and espe- cially in German, DEPENDENT CLAUSES. 5 1 (h) Conditional clauses have almost always the normal order. They may be introduced by "ef." . . . Gylf. 52, 15 . . . ef hann fengi gert borginn á einum vetri ; or by " útan " : Olafs. 129, 12: I'ltan J'ér sýnit þar meS opinber tákn ; or by " nema ": Olafs. 259, 4 : nema Sveinn konúngr væri útleystr ok frelstr meÖ öllum mönnum ; or simply by " at " as in Gylf. 15, 12: eigimundi hón ^ámeir hvata göngunni, athónhræd- diz bana sinn ; sometimes by " at " following an expression of condition : Olafs. 305, 30 etc.: ok \ó geng ek . . . með J?eim skildaga at hvárgi . . . skal öÖrum hjálpa ; some- times, further, by "er": Eyrb. 22,30; Olafs. 271, 24: " especially if " = . . . allra helzt er hann hefir heilsamligri skilning á sínu ráí5i, enn frændr hans. In Gylf. 67, 23, etc.: ok ef ek lifi ok megak ráða, etc., the "ok" in the de- pendent clause requires inversion, while once, in Gylf. 93, 18, the condition itself is expressed by the inverted order without any particle : vilit ]>ér gefa mér fylli mina af uxanum \k mun soðna á seySinum. . . . On the other hand, In- version here, as in every other dependent clause, is bound to take place if a modifier of the dependent predicate ap- pears right after the subordinating particle : Gylf. 42, 9 : . . . ef J^eim er ills af hánum van ; Olafs. 258, 4 ; Eyrb. 90, 27 ; 90, 29 : ef honum væri nökkut gjört til úfriðar. Gylf. 49, II ; Eyrb. 95, 13: ef eigi eru rammar skorður viS reistar. . . . Summing up all the circumstances under which Inversion must take place in the dependent clauses in Old Norse prose, we receive the following list, which it may be interesting to compare with the one given under the heading of ** Inversion proper " in indepen- dent clauses. I. An Adverb, an Adverbial phrase or an Adverbial clause preceded by a dependent conjunction, relative pronoun or inter- rogative word, invariably causes inversion. A. Adverb. (i) loci : \?lx (ro times) : Olafs. 125, 19-20 ; 150, 7 ; 300, 12 ; Eyrb. 15, 11-12 ; 14, 25 ; 85, 9-10, 95, 19 ; Gylf. 9, 12 ; 48, i; 54, 11; her (i) : Olafs. 102, 11-12 ; upp (i) : Gylf. 3, 8; Eyrb. 88, 12-13; fj^rri (i) : Gylf. 39, 13; 52 ORDER OF WORDS IN OLD NORSE PROSE. fangat (i) : Eyrb. 6, 28. . . . (2) temporis : >á (12) Olafs. 139, 2; 265, II ; 297, 27; Gylf. 68, 21 ; 69, 24 70, 13 ; 82, 5-6 ; 81, 7-8 ; Eyrb. 16, 14-15 ; 96, 29-30 98, 2; 100, 4-5; nú (5): Gylf. 9, 3 ; 12, 2 ; 40, 11 Eyrb. 6, 24; 90, 13; æ (i): Olafs. 296, 18; áSr (i) Olafs. 114, 20-21, aldri (3) : Gylf. 60, 15 ; Olafs. 126, 25 281, 24 ; fyrr (2) : Gylf. 89, 6 ; Olafs. 263, 18 ; fyrir (i) Eyrb. 104, 20 ; fyrst (i) : Gylf. 5, 15 ; forSum (i) : Gylf 20, 10; (3) negation: ei (i) : Olafs. 259, 9-10; ekki (3) Gylf. 63, 14 ; 60, 5 ; 65, 18 ; eingi (i) : Gylf. 70, 2-3 engi (3) : Gylf. 72, 10 ; 31, 11 ; 46, 20 ; eigi (15) : Gylf, 67, 19 ; 72, 2 ; 64, 13-14 ; 69, 13 ; 97, 8 ; 24, 2-3 ; 49, II ; 29, 28 ; 42, 20 ; 48, 5 ; 89, 4 ; Eyrb. 21, 20 ; 97, 4-5 ; 95, 13 ; 18, 15 ; (4) miscellaneous adverbs : svá (2) : Gylf. 82, 2 ; 42, II ; heldr (2) : Gylf. 27, 2 ; Olafs. 296, 27 ; helzt (i) : Gylf. 64, 3; mjök (i) : Eyrb. 18, 24; meirr (i) : Eyrb. 22, 29 ; ákafiigast (i) : Gylf. 65, 12 ; víöar (i) : Eyrb. 22, 21 ; (5) prepositional adverbs : gögnum (i) : Gylf. 99, 22 ; á (2) : Gylf. 53, 12 ; Olafs. 121, 12. (6) several adverbs: litlu áí5r (i) : Olafs. 258, 29; nökkuru síSar (i) : Gylf. 54, 17-18. . . . B. Adverbial phrase : (i) loci: eptir sálinum (i) : Gylf. 88, 7 ; (2) temporis: á skamri stundu (i) : Olafs. 105, 415 ; á einu kveldi (i) : Olafs. 112, 29; Í furstu (i) : Olafs. 258, 20; í furstunni (i) : Olafs. 291, 19-20; (3) temporal adverbial accusatives : ina niundu hverja nótt (i) : Gylf. 76, 13 ; inn fyrra dag (i) : Gylf. 76, 19-20 ; (4) miscellaneous adverbial phrases : meiS \v\ (i) : Gylf. 10, 14 ; >aöan af (i) : Gylf. 11, 20 ; af ættinni (i) : Gylf. 15, 24 ; af bans heiti (i) : Gylf. 27, 5 ; af >essum systkinum (i) : Gylf. 37, 18 ; af nafni hennar (i) : Gylf. 43, 6 ; meö hánum (i) : Gylf. 76, 7 ; eptir boöi náttúrunnar (i): Olafs. 104, 6-7 ; af honum (i) : Olafs. 118, 24; viÖ sinn harm (1) : Olafs. 135, 27 ; meÖ guös miskun ok áeggjan drotníngar (i) : Olafs. 141, 22-23 ; fyrir mar- gra luta sakir (i) : Olafs. 256, 4; metJ kappi miklu (i) : Olafs. 262, 17-18 ; til litils (i) : Olafs. 285, 2. . . . C. Adverbial clause : svá er sagt at >á er hann svaf — Gylf. 9, 6 ; fetta for . . . at ])á er sinarnar knytti — Eyrb. 88, 15. DEPENDENT CLAUSES. 53 D. Adverb + adverbial p/n-ase : par Í eyjunum (i) : Olafs. 145, 9; eigi meí5 nökkuru móti (i) : Olafs. 266, 20 ; víða um heiminn (i) : 295, 18-19 > upp or hlaOunum (i) : Eyrb. loi, 6. E. Adverbial phrase -|- adverb : niu nóttum sfðarr (i) : Gylf. 47, 3 ; um morgininn eptir (i) : Olafs. 128, 16. F. Adverbial clause + adverbial phrase : at }>vi har'ðara er f>órr knúðiz ]>ví fastara (i) : Gylf. 67, 2-3. G. Adverbial phrase -\- adverbial clause + adverb : svá at um alt Jjat riki, er átt haföi Tryggvi . . . fá : Olafs. 280, 6. II. Object, Object + Adverbial Phrase j Object -\- Dependent Clause. A. Object. (i) accusative of the pronoun : j^at (3) : Olafs. 113, 6 ; Gylf. 16, 9; 59, 4; I'etta (4): Gylf. 26, 21; 73, 18; 98, 9; Olafs. 118, 5. (2) nominal accusative: at oil búsgögn ok oil reiðigögn (i) : Gylf. 18, 1-2 ; önga fræg'ð (i) : Gylf. 40, 19-20; engan kniit (i) : Gylf. 59, 17 ; jör^ina (i) : Gylf, 69, 7-8 ; ekki vald (i) : Gylf. 69, 17 ; samflott (i) : Olafs. 153, 20-21 ; meira (i) : Olafs. 296, 20 and finally 296, 26 : at f»enna mann. ^-^ profiominal dative : mér (4) : Olafs. 102, 2 ; 139, 30 ; Gylf. 210 ; Eyrb. 95, 17-18 ; okkr (1) : Olafs. 295, II ; \éx (3) : Gylf. 90, 4-5 ; Eyrb. 16, 17- 18 ; Olafs. 305, 24 ; ytir (2) : Olafs. 286, 8-9 ; Eyrb. 95, 11-12 ; honum or hánum (13) ; Gylf. 37, 7-8 : 56, i ; 67, 17; 90, 27; 90, 29; Eyrb. 89, 26; 90, 11-12 ; 96, 17, Olafs. 146, 28; 258,4; 268, 18; 284, 5; 201, 22; henni (i) : Olafs. 103, 3 ; fessu (i) : Gylf. 53, 18-19 ; and finally feim (i) : Gylf. 42, 9. (4) nominal dative : fessu boði (i) : Olafs. 279, 28 ; í>ór ok Óðni (i) : Olafs. 280, 8 ; bóndum (i) : Olafs. 279, 20; pórði (i) : Eyrb. 11, 13; hrossum porbjarnar (i) : Eyrb. 21, 25-26. (5) nominal genitive : Jjeirar gjafar (i) : Gylf. 47, 11; fess hlutar (i): Gylf. 56, 8-9. B. Object + adverbial phrase : at bátinn undir honum reiddi vindr, etc., Olafs. 294, 2. 54 ORDER OF WORDS IN OLD NORSE PROSE. C. Object + relative clause : sva at lutskipti >at alt, er hann fékk i hernat5i. . . . Olafs. 258, i. D. Adverbial Accusative + temporal clause : at annan tima, er ek kemr. . , . Olafs. 121, 19-20. III. Predicate ?ioun, if placed immediately after the dependent particle, invariably requires inversion : A. Predicate Substantive : sva er sagt, at Arnúlfus hét maör, Olafs. 105, 29 ; here also belong : Olafs. 116, 4: er Silfra- skalli var kalla'ðr ; 120, 5 : er hinn imgi var kallaSr, and finally Olafs. 142, 22 : er hinn 3 var keisari, the last three showing an emphatic position of the predicate noun in the relative clause. B. Predicate Adjective : sem dauSr sé Knútr, Olafs. 118, 28; at sátt mundi vera j^at, Olafs. 254, 14 ; at sundrleit, Olafs. 262, 16 ; er görr var, Gylf. 17, 12 ; at jafnsatt, Gyif. 40, 5 ; fa er full er, Gylf. 80, 17. Not infrequently the adjective in its attributive form is given the emphatic position, being separated from the subject which it modifies by the predi- cate : at 2 væri kosti til, Olafs. 126, 17 ; again : Gylf. 49, 10 : at margr kemr sá . , . ; and finally : Olafs. 262, 27 : at allr varö vátr steinninn, in which case the " steinninn " is separated from the modifier " allr " by copula and predicate adjective. C. Predicate infinitive : at dveljaz munu stundirnar, Gylf. 31, I, and : at eira skyldu eldr ok vatn, Gylf. 73, 11. D. Predicate participle : er átt hafSi, Gylf. 32, 9 ; Olafs. no, 25 ; 117, 20 ; 279, 6 ; 279, 8 ; sem verit höföu, Olafs. 115» 23 ; at verit hefir, Olafs. 272, 23 ; enn sagt er, Gylf, 31, i ; at brotinn var, Gylf. 57, 14 ; ok er búit var, Gylf. 79, 10 ; er brendr er, Gylf. 87, 2-3 ; en fallnir vóru, Eyrb. 86, 28 ; at komit var, Eyrb. 98, 7-8 ; at rifin var, Eryb. 99, 8 ; while in all the instances just cited the participle is in the past tense, we have in Gylf. 7, 8, a present participle : svát logandi er hann ok brennandi, etc. etc. B. Transposed Order. Under " Transposition " we understand that arrangement of words in which the subject and the verbal modifiers precede the DEPENDENT CLAUSES. 55 predicate: subject -\- verb-modifiers -{• verb. In the partial trans- position the predicate may further be followed by verbal modi- fiers, while the complete transposition does not permit a verbal modifier to follow the predicate. The two possibilities of the transposed order in the case of a simple predicate are thus : (i) Subject + verbal modifiers + verb final. (2) Subject + verbal modifiers + verb + verbal modifiers. Of the latter type there are exceedingly few instances, so that it may almost be said that if transposition of the simple predicate verb is used, it is generally in its cotnplete final form. If the predicate is a compound tense of the verb, the arrange- ment becomes more complicated : it is a question first of the relative position of the constituent parts of the predicate, and secondly of the verbal modifiers. In the case of complete trans- position, the auxiliary holds the extreme final position in the sentence, no matter whether it is preceded by verbal modifiers or not : the two possible types being thus : (i) Subject + verbal noun -j- auxiliary final. (2) Subject + verbal nomx -\- y&x\)-rí\odí\^^x% A^ auxiliary final. In the partial transposition of the compound verb the auxili- ary must, of course, also follow the verbal noun, while the verb- modifiers are at liberty to follow the auxiliary, thus : (3) Subject + verbal noun + auxiliary -{- verb-modifiers. (4) Subject -|- verbal noun + verb-modifiers + auxiliary -(- verb-modifiers. In the independent clause of Old Norse prose transposition is almost entirely unknown, and we had to quote the Poetic Edda in order to show that it occurs at all in the Old Norse language. As regards the dependent clauses, unless prevented by reasons and phenomena requiring inversion, the normal order is by far the predominant mode of dependent expression, there being not a few groups of subordinate clauses without any genuine instance of transposition at all, at least in the 250 pages examined for the pur- pose. It is on account of the infrequency of the transposed order in Old Norse that no attempt can be made to formulate dogmatic principles and hypotheses as to why transposition is used, and it must suffice to state that it occurs only as an entirely insignificant secondary mode of expressing dependent relation ; with the ex- ception of the adjective clauses in which, as we shall see, it is employed more frequently. 56 ORDER OF WORDS IN OLD NORSE PROSE. ( 1 ) Substantive clauses. Neither the declarative nor the impera- tive sentence shows a genuine case of transposition. The subject is omitted, although the auxiliary has the final position, in Eyrb. 95, 6 : hón hyggr at eiga mundi ; Eyrb. 96, i : ek veit at verSa mun ; so also in Gylf. 68, 21-22 . . . ek niunda eigi trúa at vera mætti. Once, in Eyrb. 11, 27 : kvaiS ]?ar engi vig baeta skulu, the subordinate conjunction is omitted and instead of the inversion after " )>ar " we have transposition. In the indirect interrogative clauses, however, there are comparatively more cases of trans- posed order. Genuine cases are: Gylf. 4, 17 ... ok spurSi hverr höllina átti ; Gylf. 58, 22-23 • M j'óttiz pórr skilja hvat lætum verit heföi of nóttina ; Eyrb. 16, 10 : enn ek kjósa hvárr okkar leysa skal ; further, Eyrb. 22, 28-29 : ok sá hvat er titt var, (observe the "er" after " hvat "), and finally Olafs. 127, 18-19: ok seam hvat tiltækiligazt pikkir. Transposition of the predi- cate with the subject omitted and understood is found in : Olafs. 147, 23 : . . . Jjeir vissu varla . . . hvat vera mundi ; and finally Olafs. 293, 14-15 : vita, hvat valda mundi. Out of 334 normal dependent declarative clauses, 3 are what may be called trans- posed. Out of 96 normal dependent imperative clauses, i is peculiarly transposed. Out of 108 normal dependent interroga- tive clauses, 5 show genuine, and 2 so-called transposition. (2) Adjective clauses. While in the substantive clauses trans- position is of the utmost insignificance, the dependent relative sen- tence shows an entirely different proportion between the normal and transposed arrangements. If out of the 677 clauses, only 485 show normal order, while there are 194 transposed, trans- position already begins to play a by no means insignificant role. In fact, the dependent relative sentence is perhaps the only one which employs at all the transposed order, if we ignore the 5 cases of the dependent interrogative, because the remaining 4 of the substantive as well as the 32 instances in the adverbial clauses to be discussed later are practically incomplete in their transposi- tion. A more detailed treatment of the adjective clause is, there- fore, indispensable. The question why in the relative sentence transposition is so strikingly more frequent than in any other subordinate clause could hardly be answered satisfactorily in a general treatment like this, but would undoubtedly be a subject well worthy of special consideration. I would take the liberty, though, to suggest that such a strange phenomenon /í?/«/í, perhaps, DEPENDENT CLAUSES. 57 to an older siage in the development of the language, to a time when transposition had been employed with the same liberty, for in- stance, as in Anglo-Saxon or Old High German, and that in a comparatively later process of abandoning such order, it felt especially embarrassed in regard to the relative clause. The original parataxis of the pre-Germanic relative clause had yielded in the further developrnent of the language to the hypotaxis as expressed in the transposed arrangement. Now, when at a much later period the Old Norse abandoned the dependent transposed order without giving up at the same time the dependent hypo- tactical arrangement, the return to the normal order was espe- cially embarrassing in regard to the relative clause, which had undergone the change from parataxis to transposed hypotaxis and was to receive again the order of the parataxis, that is, the normal order, and the result was that while the new rule broke through in so many instances of the relative clauses, transposition stuck to others of them. The dependent final verb of the relative clause is separated from its subject by an adverbial expression, an object, and by a predicate noun. A. By an adverb : eptir : Olafs. 105, 14 : ]>at er eptir var ; aptr, Eyrb. loi, 31 : er aptr gengu ; nærr : Olafs. 282, 28 ; 283, 11 : er nærr var ; inn : Gylf. 38, 12 : er inn gengr inni ; Olafs. 118, 18 ; 119, I ; 292, 14 : feir er inni vóru ; her : Eyrb. 94, 30 : er her er ; far : Eyrb. 3, 19 ; 85, 12-13 ; Olafs. 302, 8 : er J^ar var ; J>ar til : Gylf. 13, 14 : er far til liggja ; }>£ : Olafs. 297, 7 : hverr fa var ; við : Eyrb. 99, 19 : feir er viÖ vóru ; and finally svá : Olafs. 108, 6 : 142, 20 ; Gylf. 7, 4 ; 15» 5 I 21, 12 ; 22, 7 ; 30, 7 ; 48, 13 ; 50, 7 and 89, 13 : er svá heita. B. Other adverbal expressions {^ preposition + case^ : i borginni : Olafs. 105, 8 : fat er í borginni var ; með feira : Olafs. 107, 20 : ermeS feim var ; Olafs. 113, 2-3 : er fyrir feim var ; Olafs. 125, 5 : er á skipinu vóru ; Olafs. 279, 3 ; 303, 11 : er í móti stóí ; Olafs. 282, II : er í móti mæla, but immediately after that : ok eigi vil- ja hlýða. Olafs. 292, ti, 12 : er i skálanum vóru ; Eyrb. 10, 23 : er til skersins la ; Eyrb. 98, 8 : feir er í húsinu vóru ; Eyrb. 102, 10 : er við eldinn sátu ; Olafs. 303, 23 : sem í eyjunni var ; Olafs. 304, 23 : er á fau triia. Gylf. 11, 8 : er or sárum rann ok laust for ; Gylf. 15, 16-17 • ^^ eptir henni ferr ; Gylf. 15, 18 : er fyrir henni hleypr ; Gylf. 20, 13 : er til hrimfussa horfir ; Gylf. 22, 17 : 58 ORDER OF WORDS IN OLD NORSE PROSE. er fyrir ósköpum veröa ; Gylf. 28, 1-2 : peir er í val falla ; Gylf. 48, 12 : er á hans boriSi stendr ; further in Gylf. 75, 9 : er til feirar sendifarar varð ; and finally in the combination of adverb + ad- verbial phrase : Olafs. 112, 12-13 5 er par í landi vóru. A closer observation of all the illustrations cited above will reveal the very interesting fact that with the exception of Olafs. 282, II (Í móti mæla) and Olafs. 304, 23 (er á ]>au trúa) in all cases of transposition the final verbs are intransitives : to be (20 times), to become (i) ; to lie (3) ; to go, to run, to ride, to fall (7) ; to sit (i) ; to stand (2) ; while the adverb " svá " is never found in any other connection but with "heita " (10). In other words, out of 46 cases in which an adverb or an adverbial ex- pression precedes the final predicate, 34 all show an intransitive verb, 10 have " heita " preceded by " svá," which almost seems to be a stereotyped phrase, and 2, only, show the variation of prepo- sition plus case + *' to speak," and " to believe." C. Object. (i) Pronominal object precedes the final verb : mér : Eyrb. 95, 24; Olafs. 261, 2-3 : sem mér skilja ; honum : Olafs. 131, 6 ; 268, 5 ; 260, 24 : er honum þjónaðu ; henni : Eyrb. 95, 26 : er henni likar ; hana : Eyrb. 97, 7 ; Gylf. 15, 14 : er hana soekir ; peim : Gylf. 8, 3 : sú er ]>eim fylgði. (2) A substantive object : Olafs. 136, 11 : er himnum ræðr ok skapat hefir alia luti ; Olafs. 150, 10 : er sigr fengi. Olafs- 281, 19-20 : er alia sér trúliga Jjjónandi elskar ok au'Sgar. Olafs. 284, 17, 18: er island byggvir. Eyrb. 85, 25 : er vápnin sóttu. Olafs. 295, 20 : er J'ann sið hafa ; Gylf. 12, 14 : er heim byggja ; Gylf. 53, 19-20 : sá er flestu illu ræí5r ; Gylf. 70, 8 : ]>á er mikinn matt hafa, and finally Eyrbyggja 92, 6 : er frælana átti ; altogether 19 cases. An attempt to classify the final verbs preceded by an object is more difficult, there being no such definite lines as in the case of adverbs and adverbial expressions. The auxiliary " to have " (to get, to be in possession of) occurs 4 times ; *' to seem, to deem, to please, to love " 6 cases ; " to be at the head " 2 cases ; " to obey, to serve " (2) ; " to inhabit " (2) ; " to seek " (2) and " to see " (i). D. Predicate noun is more frequently found before the final verb than all the above-mentioned groups counted together. It occurs : DEPENDENT CLAUSES. 59 (i) As a substantive 89 times, the majority of which, 58, is found in the Prose Edda, as : 7, 3 ; 7, 7 ; 13, 9 ; 14, 2 ; 14, 4 ; 14, 5 ; 14, 19 ; 15, 6 ; 15, 22 ; 18, i ; 24, 5 ; 24, 14-15; 25, 2; 25, 4; 25, 6; 25, 11; 26, II, 12; 27,4; 27, 18; 30, 2-3; 32, 6; 32,7-8; 33, 9-10; 34, 10; 35, 21 ; 37, 9 ; 39, 19 ; 4°, 13 ; 41, 10 ; 41, 12 ; 42, 5-6 ; 42, 14-15 ; 43, I ; 43, 6 ; 45, 9 ", 47, lo-n ; 48, 5-6 ; 49, 13- 14 ; 49, 15, etc. on almost every page. Olafs. 103, 4 ; 106, 15 ; 106, 22 ; 107, 3 ; 107, 21 ; 108, 20 ; 108, 24 ; 109, 26 ; 114, 28; 115, 25; 139, 3; 145, 5; 153, 4; 153, 14-15; 154, 28 ; 250, 21 ; 260, I ; 265, 7 ; Eyrb. 3, 7 ; 3, 8 ; 4, 27 ; 4, 28 ; 8, 27 ; 8, 29 ; 13, 10 ; 13, 10 ; 18, 6 ; 21, 16 ; 21, 23 ; 92, 24 ; 96, 28. Almost in each of these instances we find the formula : sá er + Proper name + heitir, or hét, or heita, or hétu. Of other cases : Gylf. 81, 11 : \k verðr fat er mikil tíí5indi ^ykkja ; Gylf. 43, i : henni )'jóna fær er meyjar andaz ; in Eyrb. 96, 28 the predicate sub- stantive is a preposition + proper noun : . . . peir kómust á bæ Jíann, er "í Nasi" heitir "hinu neðra" (observe the peculiar position of neöra). (2) As an adjective 16 times : Olafs. 105, 9 : pat er fémætt var ; Olafs. 117, 13 : NorÖimbraland, er heiðit var ; Olafs. 256, 25 : }'at er meira var ; Eyrb. 85, 3 : vápn fau er pyngst vóru ; Eyrb. 92, 29 : ]>ann er afbragðHgr væri ; Eyrb. 94, 23 : >at er flatt la. Eyrb. 97, 19 : klæði fau, er vat vóru ; and finally in Gylf. 9, 16 : er saltir váru ; 11, 8 : er . . . laust for ; 11, 15 : er lausir fóru ; 27, 3 : er skylt er 27, 5 : er blitt er ; 34, 15 : er vitrastr er ; 43, 6-7 : er fagrt er ; 44, 12-13 '• sá er hóflátr er ; 62, 13 : sá er siiSarst gekk. Reviewing the transposition in the relative clause, it will be found that out of 194 cases in the 170 instances quoted above the dependent predicate was a simple verb. Thus only 24 cases are left in which the dependent predicate is a compound tense of the verb : Olafs. no, 29 : >ess, er sagt er : 150, 15 : er berjast skyldu ; 258, 12 : er látit höfSu ; 258, 14 : er átt höföu ; in Eyrb. 9, 8 : er hafa vildi ; 22, 29 : J^at er mælt er ; 102, 3 : er látizt höfðu ; in Gylf. 6, 14 : er lifa skal ok aldri tynaz ; 6, 7 : er brotin váru ; 22, 4 : er fara vilja ; 22, 9-10 : er borit verSr ; 99, 2 : er kaupa vildi ; ICO, 19-20 : er yrkja kunnu, and in the five instances of Olafs. 302, 9-10, in which the auxiliary is also a compound tense : . . . 6o ORDER OF WORDS IN OLD NORSE PROSE. er Hákon jarl hafSi gera látit," — in all these instances, 14 in number, the auxiliary immediately follows the verbal noun, being separated by no verbal modifier and occupying the extreme end position. In other cases, however, (10) : Olafs. 136, 12 : er . . . skapat hefir alia luti ; Eyrb. 11, 27: >au er oröit höfíSu á Jjórsnesi ; in Gylf. 15, 25 : er kallaSr er Mánagarmr, so further : 24, 14 ; 29, 15 ; 61, 5 ; 63, 9 ; 81, 4 : er kallaSr er fimbulvetr ; further in 45, 4 : er ]?jóna skulu í Valhöll, and finally in 83,20 : er bundinn er fyrir Gnipahelli, — the auxiliary also follows immediately after the verbal noun, but precedes some modifier of the dependent verb. Among the relative clauses no instance was found showing the type : Subject + verbal noun + verb-modifier + auxiliary plus or minus verbal modifier, a type which is, in fact, rare in any de- pendent clause. (3) Adverbial clauses. Hardly two genuine instances of com- plete transposition are found among the different adverbial clauses. All the illustrations to be mentioned under this heading are in- complete. In most of them the subject is understood and omitted. We can speak only of a more or less close approach to transposi- tion : A. Local clause : Gylf. 55, 8 : hvert er — wherever — fara skal (Subject omitted). B. Temporal clause : Gylf. 57, 2 : en er soSit var ; Eyrb. 11, 8: meSan barizt var; Eyrb. 102, 14: meSan sætt er; and, finally, the only good instance in Olafs. 129, 23-24 : er konúngr hafSi skirast látit = ** als der König sich hatte taufen lassen." C. Modal-Manner clause. The phrase : " svá sem sagt er " in : Gylf. 49, 4 ; 77, 13 ; or "svá sem her segir" in : Gylf. 10, 17 ; 12, 10; 24, 6; 26, 13; 31, 13; "sem fyrr segir": Eyrb. 7, 12: Other incomplete cases are Gylf. 65, 11 : enn mér liz ; 66, 10 : sem mik varSi ; 69, 6-7 ; sem ]>ér sýndiz ; 70, 1-2 : sem siðar varS ; Eyrb. 22, 25-26 : sem ySr likar ; 104, 10 : sem komit var ; Olafs. 256, 20 : enn réttligt var ; 305, 6-7 : sem hverjum synist. D. Causal. The only and at the same time incomplete case : Olafs. 291, 7 : fviat mér leit^ist. E. FÍ7ial clause. Eyrb. 95, 23-24 : ... til fess, at henni liki." Incomplete and only case. F. Consecutive-Result clause. The only 2 cases : Olafs. 282, 26- 27 : svá at skilja mætti ; and Gylf. 97, 17 . . . svá at spyrja kunni hann fróSleiks, are incomplete. DEPENDENT CLAUSES. 6l G. Conditional clause. Olafs. 261, 5 : enn ef svá er, and Olafs. 296, 16 : ef yðr pikkir, are again no genuine cases. Peculiar is : Olafs. 295, 25 : ef hann vill kristinn gerast, which is neither normal nor transposed. 02 ORDER OF WORDS IN OLD NORSE PROSE. VITA. The author was born the 4th of May, 1870, in Mitau, Kurland, Russia. Having received his preliminary education in private schools and by private tutors, he entered the Classical Gym- nasium in Mitau, the former " Gymnasium Academicum," in January, 1882. December 19, 1890, he received from the same institution the " testimonium maturitatis," having passed the series of examinations known as the " Abiturienten examen," which latter involved — according to the regulations of said institution — an eight-years' course of study in the German, Russian and Latin languages and literatures, in mathematical and historical branches, and a six-years' course in the Greek language and literature. In February, 1893, he entered the School of Philosophy of Columbia University, received the degree of M.A. in June, 1894, and studied till May, 1895. He took the following courses : Arabic and Heb- rew, under Prof. R. J. H. Gottheil ; the German, Old Norse, and Scandinavian literatures, and " Faust," under the late Prof. H. H. Boyesen ; Germanic philology, Old High German, and Icelandic, under Prof. W. H. Carpenter ; Gothic and Middle High Ger- man, under Mr. E. H. Babbitt ; Anglo-Saxon language and Historical English (Early and Middle English), Anglo-Saxon poetry. Linguistics, and Zoroaster's teaching, under Prof. A. V. W. Jackson ; Anglo-Saxon Syntax and Chaucer, under Prof. Th. R. Price. Further : History of French literature in the i8th cen- tury (Voltaire), under Prof. A. Cohn ; and, finally, Philosophy of Kant and his successors — Fichte, Schelling, Hegel, etc., under Prof. N. M. Butler. He is at present engaged as Instructor of English to foreigners in Evening School No. 27 (under Dr. J. S. Taylor), and as In- structor of German in Grammar Schools No. 2 (under Dr. W. L. Ettinger), No. 12 (under Mr. J. F. Townley), No. 96 (under Mr. B. W. Purcell), and No. 18 (under Dr. B. C. Magie). He wishes to express his heartiest thanks to Prof. W. H. Car- penter for the painstaking correction of the proof sheets of the present thesis, a correction which required much time and energy. OíCAJ^"^ U^ .íiA^^ Q\«i^^ i.í> r.V-\OOW ,-ll-' 49CB^ 146S16V JVÍ23G039 -^ ^ / .:: <^-y THE UNIVERSITY OF CAUFORNIA LIBRARY m