2) (h)nysa 'scrutinize, examine'. The h- is here spurious; cf. njōsn 'spying, scouting' and Goth. bi-niuhsjan, OE neos(i)an, OS niusian:niusōn, OHG niusen 'to investigate, trace, try'.

Associative verb is h-njōsa 'to sniff, scent, sneeze' (cf. Falk-Torp 1. 776); semantic point of contact 'to scent, get onto the trail (like a dog)'. Indeed, this latter sense may possibly represent the basic idea of Germanic *neuhs-> ON nýsa:njōsn (cf. Fick 299).

III. (h)r

1) (h)rasa 'to rush headlong, stumble, stagger'. The h- is here spurious; cf. OE ræ san 'to rush', OE ræ's ON rās 'race, rushing'.

=

Associative verbs are h-rapa 'to tumble down' and h-rata 'to totter, stagger'. The latter verb (see below) also appeared without initial h(rata), which fact intensified the association with rasa; cf. rasa:rata and h-rasa:h-rata.

2) (h)rata 'to totter, fall'. The h- is here most probably organic; cf. Grk. κpadów 'shake, swing', Skr. kurdati 'springt, hüpft' (see FalkTorp 2. 869, s.v. radd).

Associative verb is rata (earlier vrata < Goth. wraton) 'to wander, journey'. The identity of form (aside from the initial h-) and the similarity in meaning between hrata and rata brought about an early association between the two verbs, as is shown by the fact that the verb hrata appears in the Elder Edda with and without initial h-; cf. hrata Vsp. 52.3 and ratar Grp. 36.2.

3) (h)rīfa 'to grasp, clutch; pull, pluck; scratch, pick'. The h- here is organic; cf. OE ge-hrifnian 'to grasp, tear', IE root *(s)k(e)ribh, cf. Lat. scribo 'to scratch, write' (see Falk-Torp 2. 906 s.v. rive 'harke, rechen').

In the sense of 'to grasp, clutch' the verb hrifa always appears with initial h-. Only in the derived senses of 'to pull, pluck, scratch, pick', etc., does the form rifa without initial h- occur and then too, less frequently than does the form hrifa with initial h-.

The associative verb with (h)rīfa in these derived senses is rīfa 'to tear (apart)' (cf. OFris. rīva 'to tear', MLG riven 'to rub', etc.; see FalkTorp 2. 906, s.v. rive 'reissen').

4) (h)rjōta 'rebound, fall; growl, roar, snore'. The h- here is organic. Evidently we have to do here with two verbs identical in form, hrjōta 'to fall, plunge', etc., connected with OE hreosan 'to fall, plunge', MHG rūzen 'to move hurriedly (cf. Falk-Torp 2. 924, s.v. rutte; Fick 107 s.v. hrut 2) and hrjóta 'to growl, roar, snore', etc., connected with OE hrutan,

OFris. hrūta, OHG rūzan 'to roar, growl', etc. (cf. Falk-Torp 2. 1033, 8.v. skryde; Fick 107, s.v. hrut 1).

Only hrjōta 2 appears without initial h- (rjōta). As the associative verb I suggest a lost verb *rjōta 'to utter loud tones of distress' = OE reotan, OHG riozan < *reutan; cf. ON rīta (< *reut-jan) 'to squeal', rauta 'to roar'. Either *rjōta was blended with hrjōta, resulting in historical rjōta alongside hrjōta with a slight semantic change (cf. Goth. wōpjan 'Poav': OE wēpan 'to lament'), or we may discard an original *rjōta and postulate rýta: rauta as the associative group.

5) (h)ryđja 'to clear out'. The h- is here spurious; PG *ruđ-jan > ON ryđ-ja = OE a-ryddan 'to rob, plunder'; *reud-an > MHG rieten 'to clear out, destroy'; *reud-jan > OHG, MHG riuten > NHG reuten 'to clear out'.

Associative verb is h-rjōđa 'to clear out, strip, unload (a vessel); belch, vomit'; semantic point of contact probably 'to clean' (cf. Fick 108, s.v. hrud 2). Germanic *hreuđan meant 'to load, adorn'; cf. the adjectival past participle ON hrodinn 'painted, stained' = OE gehroden 'loaded, adorned', cf. also this root in extended form OE hyrst 'adornment,' hyrstan 'to adorn' OHG hrust:hrusten (NHG rüsten). From the idea of 'adorn' was developed the sense of 'to clean (out)', whence contact between ON hrjōđa and ryđja 'to clean out'; cf. OE ge-hroden and h-ryding 'cleared land' = ON h-ryđ-n-ing ‘a clearing out of the court (dōmr, kviđr), challenge (to the neighbors)'.

=

=

Probably ON (h)rođa 'to throw together' also belongs to the (h)ryđja group (*rud-ōn > ON rođa3 OFris. tō-rotha, MLG roden, MHG roten 'to clear out'), which accounts for the initial h-; cf. hrođa with hrudning, a by-form of hrydning.

'ON rjōða is not derived from *reuð-an as Falk-Torp (2. 925, s.v. rydde) and Fick (351, s.v. rud 2) maintain.

In the first place, ON rjōða never occurs in the meaning 'reuten, räumen' which these scholars attribute to it. It is not recorded in any of our ON dictionaries in this sense but only in the sense of 'to redden (with blood)' and therefore should be connected with the stem *rauð- 'red'; cf. ON rjöðr, rauðr ‘ruddy, red.'

In the second place, ON rjōða never appears with initial h-, which would most likely have been the case if it had belonged to the (h)ryðja group.

Again, the fact that the verb hrjōða never appears without initial h- indicates that between these two verbs, rjōða and hrjōða, there was no semantic contact, i.e. that rjōða never meant 'to clear out'; but compare (h)ryðja to clear out' :hrjōða.

'The ON verb (h)roða is not recorded under the head of *ruð-õn either by Falk-Torp (2.925, s.v. rydde) or by Fick (352, s.v. ruda).

6) (h)ræll 'a weaver's rod'. The h- is here organic; *hrah-il-a-> hræll, *hreh-ul-a > OE hreol > Eng. reel; Grk. Kрéкw 'I beat the loom', KepKis 'a staff for beating the loom' (see Fick 101, s.v. hreh).

As the associative word I suggest ræl-ni 'sport, play'; semantic point of contact 'play' = 'quick motions to and fro' as in plying the loom; cf. MHG spiln 'to dart, move quickly' > 'to play' = MHG spielen, and Eng. 'play (of weapons)' etc. = 'quick motion'.

This assumption, however, may not be valid because of the comparatively rare occurrence of the word rælni (recorded only by CleasbyVigfússon). If the word rælni was of such rare occurrence that it was not associated with hræll, then we must explain the loss of the initial h- in hræll (ræll) as due to the general confusion which obtained between initial hr- and r-.

It will be seen from the above analysis that the leveling process between the given groups of initial consonants has not produced a uniform result; analogy has resulted now in the accretion, now in the loss of initial h-. This means that the associative process between these groups of consonants in question was of a double or reciprocal nature, even though in any one given case leveling resulted in only one direction, not in both.

When in the Late OIcel. period kn- and gn- passed over into hn- (cf., e.g., knīfr > hnīfr; gneggja > hneggja) the frequency of initial hn- was thereby increased to such an extent as to affect the leveling between initial hn- and n- in favor of hn-. The resultant victory of initial hnover n- then had its effect upon the two other groups of initial consonants (hl-:l- and hr-:r-), resulting in the leveling in favor of hl- and hrover l- and r-. That is to say that in Mod. Icel. the original reciprocal analogy has finally resulted almost entirely in one direction, i.e. in favor of initial h-.

In the OIcel. period, on the other hand, the reciprocal nature of the "Fritzner records a verb ræla but with uncertain meaning, so that this verb cannot be utilized in our discussion.

"In WGerm. this reciprocal relation is not so much in evidence because here at the very earliest period a tendency existed for initial h- to disappear before l, n, r, whereas in OIcel. initial h- regularly remained (and still remains in Mod. Icel.) before l, n, r.

4

In OHG (cf. Braune, Ahd. Gramm. §1531) when this tendency first appeared, the reciprocal nature of the analogy was still in evidence; cf., e.g., in the Hildebrandslied, wer, welihhes, werdar, ringa without initial h- (which still existed before w and r, as the alliteration shows) and gi-h-ueit, bi-h-rahanen, h-rūmen with

associative process was still strongly in evidence. Of the nine cases which I have recorded, five show an accretion and four a loss of original initial h. Those showing accretion of initial h- are h-lykka, h-neisa, h-nýsa, h-rasa; those showing a loss of original h- are (h)rata, (h)rīfa, (h)rjōta and (h)ræll. This does not, however, take into account the far greater number of cases where initial h- was either lost or added because of the general confusion between the consonant groups in question.

spurious initial h-. Later when initial h- before these consonants disappeared, the tendency to add an initial h- before these consonants correspondingly decreased.

For conditions in OE, see Siever's Angs. Gramm.3 §2171-2; in OS, see Holthausen, As. Gramm.2 §217; in both these dialects the reciprocal relation was about the same as in OHG.

As an example of spurious initial h- due to association with a definite word (parallel to the cases which I have pointed out in OIcel.) I may mention MHG (h)eischen (> NHG heischen) 'to demand'; associative word heizen 'to command'.

THE ITALIC IMPERFECT

WALTER PETERSEN

UNIVERSITY OF FLORIDA

The explanation of the Italic imperfects in *-fām Lat. -bam which is commonly accepted in the hand books1 is not without its difficulties, although it has many points in its favor. It is true that the derivation of e.g. legebam from a participial or infinitival form of lego plus *bhyā-m, the old injunctive of IE *bheu-, is unobjectionable from the phonetic point of view, and that both the second syllable of Osc. fu-fans and Lat. -bam -bās, etc., may without the slightest hesitation be derived from *bhya-m. It is also true that the derivation of the formation from one of the most common of IE roots and the analogy of periphrastic imperfects of the type of Engl. was reading are additional favoring considerations. Nevertheless there are two very serious difficulties which make it hard to agree with this hypothesis in its received form.

In the first place the part of the verb preceding the imperfect suffix has the appearance of the stem of the verb both in Osc. fu-fans and in most Latin verbs, sc. those of the first and second conjugations and some of the fourth, e.g. ama-bam, monē-bam, sci-bam, while those of the third and the remaining verbs of the fourth conjugation look as though they were made by analogy to the second, e.g. legē-bam : legère after monē-bam: monēre. Now composition of a complete word *fa-m in Italic times, with no similarly constructed pattern to follow, could take place only with another complete word, and not with a stem. We cannot of course refer to the fact that stems functioned as words in the early IE period, since there is not the slightest doubt that our formation is not older than Italic times, when stems had long since ceased to function as words. If e.g. amā- of amā-bam is a stem, there is no possibility of the latter being a real periphrastic formation, but it must at least have been made over by analogy. Cf. Streitberg, Urgerman. Gram. 341.

1 So e.g. Brugmann K. Vgl. Gram. 550, Gr. 2.3.2506, Lindsay Lat. Lang. 489 f., Sommer, Handb. d. lat. Laut- u. Formenl. 568 f., Stolz, Hist. Gram.1 287 ff.

* I ignore, of course, the use of verbal stems in a narrowly circumscribed function like the second person singular imperative, which can have nothing to do with the first part of the type amā-bam.