ú-wa-a-tar a-pi-yu i-ya-nu-un, 'The army had reached me at Harranas, and I there took command of the army.'

KBo. 3.4.2.15 =

Hrozný, BoSt. 3.182.15: nu

GIM-an I.NA HAR. SAG La-wa-šá a-ar-hu-un, 'And when I was come to Mt. Lawasa..

KBo. 4.10.1.5: nu-ut-ta UDUM ŠI.IM.TI.KA a-ri, ‘And for you the day of your fate is (will be?) here.'

Law-Code §31 (Hr.): ták-ku LŮ-áš EL.LUM GÍM-áš-šá wa (?ši?)e-li-eš na-at an-da a-ra-an-zi na-an-za A.NA DAM. ŠU da-a-i, 'If a free man and a slave woman are in love (?) and they have come together, and he takes her for his wife

Friedrich (loc. cit.) cites arškizzi as the iterative from ari 'he arrives'; but in fact ari retains its original perfect meaning 'he has arrived', while arškizzi is the corresponding present 'he arrives'17. Typical examples follow:

Hattusilis 2.10-13 (pp. 14-17 Götze): "The enemy from Turmittas began continually to assail the country Tuhhuppiyas; and because Ippassanas was devastated, he reached (a-ar-áš-ki-it) Suwataras.'

KBo. 3.4.3.70: 'He reached (a-ar-áš-ki-it) Zazzisas, and took the Upper Country'.

The IE languages present a series of forms parallel to those just cited from Hittite. The causative arnuzi corresponds to Skt. rnóti, Gk. opvvo; artari, with its preterit artat, comes as close as a Hittite verb could to the Skt. aorist ärta, and Gk. @pro; the sk-formation arškizzi is parallel with Skt. rccháti; and ari finds its analogue in the Skt. perfect āra, plural ārur. Hittite arai alone remains without an IE formation to match it, unless this is furnished by the jo-stem of Latin orior. That is a possibility which I plan to discuss at length on another occasion.

The meanings also of the Hittite root are nearly all found in the IE languages, but they are not so neatly parcelled out among the several formations. Hittite has evidently preserved several distinctions which have faded out in the IE languages, although it is possible to find passages which illustrate the original values. Thus rnóti is causative

17 Sommer (BoSt. 4.132, 10.21f.) holds that the Hittite sk-formations have an 'iterative-durative' function. Many of them certainly have; but it is a mistake to attempt to force them all into that category. The verb cited above is one of a considerable group whose use cannot be distinguished from the corresponding presents in the Indo-European languages. Apparently the Hittite 'iterativeduratives' represent a secondary development analogous to the Latin inceptives in -scō.

in Rigveda 9.10.6: ápa dvā'rā matīnā'm pratnā' ṛṇvanti kārávaḥ the ancient poets open the doors of worship'; and Greek ŏpvvμ is regularly causative in the active voice. The meaning of artari appears in Gk. PTO 'he started up'. The Old Persian sk-imperfect arasam corresponds in use precisely with Hittite arškit; e.g. Behistan 2.6: yātā adam arasam Mādam, 'until I reached Media'. The Skt. perfect shows the force of Hittite ari in Rigveda 2.9.3: yásmād yóner uda'ritha, from what womb thou art sprung'. Latin orior, whether or not it is connected. in form, has the same meaning as Hittite arai, 'rise up'; even the suggestion of hostility is contained in the compound adorior.

The IE languages nowhere present clear evidence for the vowel ē in the perfect of the root *er-; but in view of the Hittite forms, Skt. āra, ārur probably contain IE e; the Greek perfect opwpa is analogical in any case.

As we have more than once assumed, the third pl. perf. of the IE languages (Skt. ārur, Lat. ēgère, Tocharian weñare 'they have said') appears in Hittite as the third pl. pret. (šekkir, ekir, erir), although the corresponding third sing. perfect (Skt. āra, Greek xe, Lat. ēgit) must be identified with the Hittite third sing. present (šakki, aki, ari). This distribution of forms undoubtedly stands in some relation to the fact that in the mi-conjugation also the third pl. preterit ends in ir (e.g. eszi 'he is', ešir 'they were'). In view of Skt. third pl. aorists like ádur 'they gave', it may be that the ending ir was present from the beginning in the mi-conjugation, but on the whole it is more likely that in prehistoric Hittite, as in Vedic Sanskrit, the perfect came to be used sometimes as a preterit, so that it was in a position to contribute an ending to the preterit tense (which, on any theory, is of composite origin).

Once the ending ir had established itself as a preterit, its place was taken in the present (originally perfect) by the ending anzi of the miconjugation. This alien origin explains the fact that in our verbs the third pl. present differed in vocalism from the other plural forms. (akanzi, aranzi). The third pl. imperative also came from the miconjugation, and differed in vocalism from most of the plural forms (arantu). Another form from the mi-conjugation is clearly the participle, and this also follows the vocalism of the singular.

The contrast thus established between the vocalism of the third pl. preterit on the one hand, and the third pl. present and imperative, and the participle, on the other, reacted upon certain verbs which originally had the stem-vowel e throughout. Thus from eš- 'inhabit,

=

dwell; be' we have the singular forms ešmi, ešzi, ešun, ešta, eš, ešdu, as well as the plural forms ešten and ešir; but the analogical proportion, erir: aranzi ešir: x, yielded ašanzi 'they are,' and similarly we get ašandu 'let them be', and ašanza 'being'. Just so epmi 'I take' (:Lat. adipiscor, coēpi, etc.) always shows the stem-vowel e, except in appanzi 'they take', appandu 'let them take', appanza 'taking', and the verbal noun appatar 'taking, dwelling'. All the forms of ed- 'eat' have radical e, except adanzi 'they eat', adandu 'let them eat', adanna 'to eat' (infin.), adanzi 'to eat' (supine), and the iterative stem azk- (e.g. azzi-ik-kán-zi 'they eat').

That these forms with secondary a are analogical and not due to a phonetic development (say, anticipation of the a of the following syllable), is shown by the fairly numerous third pl. presents which retain radical e; e.g. eššanzi 'they use, treat', šešanzi 'they sleep', memanzi 'they say', wekanzi 'they ask, demand'.

Since two, at least, of the verbs which changed e to a in the third pl. present (eš- and ed-) must have had short radical vowels, it seems necessary to conclude that the inducing forms (šekkir, egir, erir) also had short vowels in the radical syllable. This is an additional reason for thinking that Hittite had lost the original distinction between long and short vowels. The frequent double writing of a vowel (e-eš-zi, a-ar-áš, ar-ha-a-ri) should therefore not be interpreted as evidence of long quantity.

The variation between e and a in the consonantal verbs is therefore to be traced to the same source as the variation between o and ē in IE perfect. We have found three verbs which show traces of the peculiar formation both in Hittite and in the IE languages. Very likely others will be discovered. Many Hittite verbs, however, owe to analogy an a in the radical syllable of the third plural present and imperative, the participle, and certain other forms.

Scarcely less important than our main conclusion is the demonstration than the Hittite hi-conjugation is of composite origin. While some of its salient features are of the same origin as the IE perfect, it usually carries the meaning of the IE present (or aorist), and some of its forms (i.e. third plural present and imperative, and the participle) come from the original inflection of the present. It will not be surprising if traces of aorist inflection are found in the hi-conjugation.

INITIAL h BEFORE 1, n, r IN OLD ICELANDIC

A TYPE OF ASSOCIATIVE CONSONANT GROUPS1

ALBERT MOREY STURTEVANT

UNIVERSITY OF KANSAS

Initial h before l, n, r disappeared in ONorw. at a preliterary date but was regularly retained in OIcel. (cf. Noreen §289). In OIcel. this initial h- was often lost, or an initial h- was often added to l, n, r, through association between words having initial hl, hn, hr and words having initial l, n, r. This associative process was due to resemblance in meaning or in form aside from the initial consonants in question. After the loss or the accretion of initial h- had become established between certain words or word groups there developed a general feeling of uncertainty as to whether a word should begin with hl or l, hn or n, hr or r. Most cases of this type of general confusion are naturally found in the Late OIcel. period, yet several instances may be cited in the classical period; which shows that the process was even then well under way. For the classical period I may mention the following examples: Hniflungr (Elder Edda) for Niflungr, rār 'damp' (Elder Edda) for hrār, hrjā for rjā2 'wrestle'; for the Late OIcel. period hnezla (for nezla) 'button loop', hniđra (for niđra) 'to lower', ređr (for hreďr) 'genitals', hreifr (for reifr) ‘glad', hrifsa (for rifsa) 'to rob', ringja (for hringja) 'a round pail'.

The purpose of the following analysis is to determine the word (or word group) with which a given form may have been associated with the resultant loss or accretion of initial h-. Such words will be called

1 Works to which reference is made throughout this paper are: Cleasby-Vigfússon, An Icelandic-English Dictionary, Oxford, 1874. Falk and Torp, Norwegisches-Dänisches Etymologisches Wörterbuch, Heidelberg, 1910. Fick, August, Vergleichendes Etymologisches Wörterbuch der Indogermanischen Sprachen“, Göttingen, 1909. Fritzner, Johan, Ordbog over det gamle norske sprog, Kristiania, 1886. Larsson, Ludvig, Ordförrådet i de älsta islänska handskrifterna, Lund, 1891. Noreen, Adolf, Altisländische Grammatik, Halle, 1923.

It is not certain which of these two forms is primary but the greater frequency of the form rjā (without initial h-) favors this form as primary.

associative words. Derivatives of a given form will not be quoted (unless necessary), since the form in question will represent the whole word group.

Only forms belonging to the saga period or to an earlier date will be taken into consideration. For determining whether an initial h- is original or a later analogical accretion the etymology of the word is the first criterion, but if the etymology of the word is uncertain then the only criterion is the frequency of the initial h- especially in the oldest texts (Elder Edda, Larsson); the greater this frequency in the oldest texts the more likely it is that the h- is original, and vice versa. But even here we must be cautious. For example, it is almost an assured fact3 that Hāvamāl I is of West Norwegian origin, which may account, e.g., for the loss of initial h- in the form rās (for hrās), Hāv. 152, 2: ā rītum rās viþar. According to the nature of the case the associations pointed out in my analysis cannot be proved; association is a tendency, and whether this tendency seems plausible or not the merits of each case must decide.

I. (h)l

1) (h)lykkr 'bend, curve'. The h- is here spurious. The form lykkr <*luk-ja-R belongs to the group lūka, lauk : luk-um, lok-inn ‘to close'. For the associative word I suggest h-lekkr (cf. OE hlence > Eng. link) 'link, chain'.

II. (h)n

1) (h)neis-a 'shame, disgrace'. Etymology doubtful, but neis- may be from *nais- < *naiss- < *nait-p- with p-extension (as in *hvat-þ- > *hvass-> ON hvass, Goth. h-ass-(aba), cf. ON hvatr) and therefore connected with Goth. ga-naitjan ‘årμav', nait-eins 'Bλaoyŋuía', OE næ tan, OHG neizzen 'plagen, quälen'.

No evidence as to the original form of the word is offered by the older texts. The word is not recorded by Larsson and occurs only once in the Elder Edda and here in the Hav. 49. 4: neiss es nøkkvipr halr. The latter evidence, however, is not conclusive, since the Hav. I is undoubtedly of ONorw. origin.

Assuming the original form of our word group to be neis-, I suggest as the associative group h-neyk- 'disgrace'; cf. h-neyk-ja 'to put to shame', h-neyk-sla 'to offend', h-neyk-slan 'offence', h-neyk-sl(i) ‘disgrace', etc.

Cf. George T. Flom, 'A Group of Words from Hávamál I in the Light of Modern Norwegian and Icelandic Dialects', Scan. Studies and Notes 1. 251-73.