722. MS. defective at edge. Zupitza's transliteration of the facsimile of the MS. has '[gehr] an.' There is room for two letters beforehrān, but there is no evidence forge -. On the contrary, whilsthrinānusually governs the dat.,gehrinānmore commonly takes the accus. (paceGrein).

723. MS. faded at edge. Kemble, Grein-Wülcker, and Heyne '[hē] ābolgen.' Zupitza says: "Nowbolgenis still distinct, and before it I think I see traces of two letters of which the first seems to have beeng; but what preceded this is entirely faded. "

752. Many editors normalise to 'scēata.' See Sievers § 230.

762. MS. defective at corner. Ettmüller, Wülcker, Heyne 'þǣr.' Zupitza's transliteration 'hwǣr,' as if there were no doubt as to the reading, but his foot-note runs: "hwǣr(hwwith another ink, and crossed out in pencil) B, ...ǣrA; now only the lower part ofrleft. " Cf. l. 797.

765. MS. 'þæt he wæs.' Grein suggested the accepted emendation — the omission ofhē.

780. MS. 'hetlic'; Grundtvig 'betlīc.' Cf. l. 1925.

788. Zupitza and others 'helle-hæfton,' but nothing is gained by making them a compound. For- anof the weak declension,- onis not uncommon.

Almost all editors insert 'tō' before 'fæste.'

811. Kemble first inserted 'wæs' after 'hē.' Heyne has: '(hewæsfāg wið god), 'which appears to me a distinct enfeeblement of the MS. reading.Fāgcomes at the beginning of a line in the MS., and Heyne says it cannot be settled whether of notwæsstood before it. This is very misleading. "There was no room forwæsbeforefag "(Zupitza), as a glance at the facsimile suffices to show.

836. MS. defective at edge. Cf. l. 926

849. MS. 'hat on heolfre,' and so Wülcker. Grein 'hātan' =hātonin the text. The reading in the text is much easier than that of the MS., and l. 1423 turns the probability in its favour.

870-1. Rieger and Bugge put 'word ... gebunden' in a parenthesis

875. MS. 'Sigemunde.' Grein's emendation 'Sigemundes' is good in itself, and is more probably in that the next word begins withs.

880. Heyne normalises to 'swylces.'

895. Many editors normalise to 'gehlōd.' Sievers § 392, N. 3.

897. Earle adopts Scherer's emendation 'hāt [e],'with heat.

900. Cosijn's emendation 'āron ðāh,'with honours throve, is adopted by Heyne and by Earle. Forāron = āruncf.scyponl. 1154, andhēafdonl. 1242, and for the phrase cf.weorð-myndum þāhl. 8. Nevertheless I cannot bring myself to abandon the clear reading of the MS., which makes at least as good sense as many another passage.

902. MS. 'earfoð,' retained by Wülcker; cf. l. 534. On the other hand, see ll. 602, 2349.

906. 'æþellingum.'

911. There appears to be no sufficient reason for making a compound,fæder-æþelum, here, as the editors do. Cf. ll. 21, 1479.

915. Some editors mark the close of this episode by a space between this line and the next. There is nothing more than a dot in the MS., not a fresh line, nor even a capital tohwīlum.

945. Heyne 'eald-metod.' See note on l. 1776.

949. MS. 'ǣnigre.'

954. Kemble's emendation. No gap in MS.

963. MS. 'him.'

965. MS. 'hand gripe.' Kemble's emendation is required for the sake of the alliteration.

976. MS. 'mid gripe'; Thorpe 'nið-gripe'; Bugge 'nȳd-gripe.'

984-5. Suggestions too numerous to mention have been made for the emendation of these lines. Heyne adopts a fresh one with each new edition. Sievers considers the second half of l. 984 metrically deficient, and proposes:

fēondes fingras, ⁠ foran ǣghwylc wæs

stiðra nægla stȳle gelīcost.

986. MS. 'hilde hilde rinces,' the firsthildebeing the last word on the page, the second the first word overleaf. In such cases italics in the text seem needless. Forhand-sporusee Sievers § 279.

1000. MS. 'þe.'

1002-5. These lines, as given in Holder's edition, show the principal emendations that have been suggested:

Nō þæt ȳðe byð

tō beflēonne (fremme sē þe wille!),

ac gesēcan sceal ⁠ sāwl-berendra [gehwā],

nȳde genȳded⁠ niþða bearna.

1013. Thorkelin A 'blæd agande,' B 'blædagande.' The MS. now has onlyblædleft, anddeon the next line.

1014-5. Bugge proposed to put these two lines in parentheses, because of "the difficulty of finding an antecedent forþāra. " Heyne (5th edition) and Earle adopt the suggestion. This can only be on the principle — of two difficulties choose the greater. What a master of parenthesis-style the" scop "must have been, to keep his hearers waiting for the subject ofbugon, past two other finite verbs with a different subject, until four lines lower down! And what is to hinder the antecedent ofþārabeing implied in⁠ blǣd-āgende, in speaking of a court where everyone was doubtless related to everyone else, as in a Scotch clan?

1020. MS. 'brand.'

1026. MS. 'scotenum'; Grein 2 'scoterum'; Wülcker 'scēotendum,' for which cf. ll. 703, 1154. Heyne quotesoxenum, nefenum, as examples of weak dat. pls.

1030-1. The MS. has 'heafod beorge wirum be wunden walan utan heold.' Ettmüller 'wala,' adopted by Grein. If we leave the MS. reading unaltered, there is a choice of difficulties. Either we must takewalanas subject andhēafod-beorgeas object, with a striking violation of grammatical concorn in the verbhēold; or we must (with Heyne and Socin) takehēafod-beorgeas a weak fem. noun in the nom. andwalanas object, with considerable loss to the sense. The nom. pl.scūr-beorge( "Ruin" 5) also tells against the latter view, which has no support from analogy.

1032. Thorkelin 'laf' (now gone in the MS.). On account of this reading, Bugge ( "Beiträge" xii. 92) supports Thorpe's emendationmeahte, confirming it by the formscūr-heardin the next line, and by a reference to Sievers: "der erste halbvers ist nach den untersuchungen Sievers '(" Beiträge "x. 455) metrisch unrichtig." It is a curious commentary on this last reason, that Sievers himself quotes the line, with the formlāfe, among the examples of his type A ( "Beit. " x. 273).