NOTE 1. I have found no illustrations of de used as a temporal connective in the poetry.

NOTE 2. The ON. use of the relative particle er as a temporal conjunction offers a parallel to de temporal1. I quote one example, but the number might be increased indefinitely Cleasby's Icelandic Dictionary, Egils Saga: næst er ver komum.

6 a. swa oft swa.

Though the cognates of the elements of this connective occur in about all the Germanic dialects, I have been able to discover the corresponding temporal connective only in Gothic, swa ufte swa. It corresponds etymologically, and also respecting its use, to the Modern English as often as. It occurs but seldom, and I have found only about eighty instances of its use in OE. prose, while I have not noted it at all in the poetry, nor in Middle English.

In this example the verb is omitted: Lch. 1. 398. 15 cweþe... pater noster swa oft swa þæt oder . .

Occasionally there is a balancing or correlative swa in the main clause, thus: LS. 2. 292. 1200 and swa oft swa he pyder ferde, swa forhtodon þa deobla on ge-wit-seocum mannum. Much more frequently do we find a demonstrative donne at the head of the main clause, as in this example: Wulf. 106. 28 and swa oft swa hy fyrdedon oððe to gefeohte woldon, ponne offrodon hy heora lac. However, in the larger number of cases we have no such balancing adverb, but the main clause follows directly, thus: O. 218. 15 swa oft swa hiene Romane mid gefeohte gesohton, he hie simle gefliemde.

NOTE 1. The two parallels in Gothic occur in the same chapter and in consecutive verses: 1. Corinth. 11.25 sa stikls so niujo triggwa ist in meinamma bloba; þata waurkjaiþ,

1 See under đa đe, p. 23.

swa ufta swe drigkaiþ, du meinai gamundai; 1. Corinth. 11.26 Swa ufta auk swe matjaip pana hlaif jah þana stikl drigkaiþ daubau fraujins gakannjaiþ, unte qimai.

That swa oft swa had not become altogether a closed compound seems to be shown by these two examples: O. 274. 24 7 siþþan he was Sapan þæm cyninge to don geset, op his lifes ende, þæt he swa oft sceolde stupian swa he to his horse wolde: CP. 273.12 forðæm hit gewitt sua oft fram us sua us unnytte geðohtas to cumað. Here we may see the origin of the particle. The swa oft belongs properly to the main clause, the swa to the subordinate member of the sentence. The temporal character was originally only incidental, and the idea of comparison predominated. But for the historical period of OE. the temporal force is the only one felt, and the whole expression is a compound conjunction used to introduce a clause of repeated action.

It is interesting to note that one of the two examples (1. Corinth. 11. 26) of swa ufta swe in Gothic is of this type.

The two examples quoted above are the only ones I have found in the prose, and I have noted none in the poetry.

This connective is very rare, only five examples having come under my notice in all OE. prose; but these occur in texts ranging from Ælfred to Wulfstan. Probably in origin the construction is modal, but in none of the cases that I have noted could this be considered the primary signification. It is noticeable that this construction is used only with

the optative present of mugan; and that in every case the subject of the subordinate clause is placed between the swa and the oftost, while nothing else ever is. CP. 5. 3 & fordon ic de bebiode dæt du do swæ ic geliefe dæt du wille, dæt du de dissa woruldðinga to dæm geametige swæ du oftost mæge, dæt, &c. Since the number of examples is so few, I list them here for convenience of reference: E. Th. 445. 23; Inst. 415.7; Wulf. 73. 21; 290.22.

NOTE 1. Grein's Sprachschatz gives only one reference for this connective in the poetry: Metra des Boëtius1, 22.9 sece þæt sidðan on his sefan innan 7 forlæte an swa he oftost mæge ælcne ymbhogan.

6 d. swa oft swa... oftost.

This curious construction seems to arise from a confusion of the common swa oft swa construction and the rare swa...oftost. I have found it only in OE. prose, and have been able to discover no parallels in either the poetry or the other Germanic dialects. Both examples are from Wulf: 143. 11 and þæt we swuteljað swa oft swa we oftost pater noster singað; 234.11 donne lærað hy us eac swa oft swa we oftost magon, þæt we secan ure cyrican.

The origin of the construction from a confusion (contamination) of the two constructions treated above is especially clear from a study of the latter example. Here we have the characteristic position of the subject, and the characteristic verb of the swa... oftost type, as well as the full swa oft swa formula of the more common variety. It seems clear also that in the first example the oftost is dragged in by association with the swa immediately preceding.2

1 Grein, Bibliothek, vol. 3.

For an example of a similar mixed construction, see swa lange... oð dat p. 133.

7a. prep. + object (noun of time) + de.

Occasionally what seems to be a real temporal clause is introduced by a prepositional formula of the nature indicated above. In general, it may be said that such clauses are with difficulty distinguished from mere relative adjective clauses qualifying the noun of time in the quasi-stereotyped phrase introducing the clause. No hard and fast line can be drawn, for the temporal nature of the clause is incidental, and its primary function is to characterize the noun of time. But often the phrase (prep. + object+de) may be translated when, or is logically equivalent to when, and the time of the main verb is fixed by the relative clause.

I have noted clauses of this kind, denoting time when, introduced by on, in, to, and ymbe. The nouns used as objects are dæg, niht, afen, gear, tid, tima, first, gefeoht, and fleam. The most common case is the dative, but the accusative and instrumental are also found. De, of course, is merely the relative particle, and may represent any number, gender, or case. The noun of time may or may not be modified.

The following examples will illustrate the normal form of this clause, as well as the cases, prepositions, &c., which are found: L. 17. 29 Soolice on þam dæge þe Loo eode of sodoma hyt rinde fyr 7 swef of heofone; Dial. 29. 30 7 sona ongæt þæt seo Godes fæmne wæs gehæled in þa ylcan tide, pe se Godes peow cyrde; O. 168. 36 Ymbe done timan pe piss wæs, ..., he besierede pæt folc; Gen. 21. 8 and Abraham worhte, swa swa heora gewuna wæs, mycelne gebeorscipe to blysse his mannum on pone dæg, þe man þæt cild fram soce Sara ateah.

Sometimes the relative particle is separated from its antecedent by the main clause, as in this example:

Dial. 172. 18 he pa acsiende smeaðancollice ongeat, þæt on da ylcan tid wæs þæs biscopes forofore, þe se drihnes wer oncneow his upstige to heofonum. So also in this: LS. 1, 48. 410 On pam dæge pu scealt cuman to me be ic com to mannum. This example,

I think, can hardly be considered a true temporal clause. Its difference from the following example will perhaps show the distinction, as well as give some idea as to the difficulty of distinguishing: ÆH. 1. 286. 22 Efne du gesihst done mannan beforan de, ac on Jære tide pe du his neb gesihst, þu ne gesihst na his hricg; ÆH. 2. 356. 8 and wæron forði þa gebytlu on dam dæge swiðost geworhte, de he ða ælmessan gewunelice dælde; Dial. 30. 12 sona swa þæs cnihtes fæder him fram cyrde, on pa ylcan tid he oncneow, þæt him wæs eft lif seald, be he ær gehyrde of þæs hælendes sylfes mude, þæt him lif gehaten wæs.

The following is the only example I have noted in which to is used in these phrases: Lch. 1. 256. 12 7 to dam timan de se fefor to dam men genealæcean wylle, smyre hyne þærmid. The word-order is peculiar in the following sentence, but is due to the Latin: OET. Vesp. Psalms 19. 10 geher us in dege in dam we gecegað dec. Latin: exaudi nos in die qua invocaverimus te. pa is probably for be in this example: BH. 168. 2 þa gelomp in seolfan tid, þa mon þone cyning fulwade, dæt þær wæs, &c. This view is supported by the fact that MSS. B. and Ca, have pe.

O. 206. 13 pa on dam gefeohte pe hie pæt fæsten brecan woldon, was Romana fela mid flanum ofscotod. Gefeoht is not strictly a noun of time, but the construction is identical with that of the other examples, and the word does represent the time in which the action of the main verb occurs.

d