[215] ll. 2570 etc.

[216] intrepidum mentis habitum retinere memento.

[217] ll. 2663 etc.

[218] Cf. Beowulf, 2705: forwrāt Wedra helm wyrm on middan.

[219] Cf. Cotton. Gnomic verses, ll. 26-7: Draca sceal on hlǣwe: frōd, frætwum wlanc.

[220] virusque profundens: wearp wæl-fȳre, 2582.

Cf. Beowulf, 2567-8, 2569, 2561 (hring-boga), 2827 (wōhbogen).

[222] Volospá, 172-3 in Corpus Poeticum Boreale. I, 200.

[223] Cf. on this Olrik, Heltedigtning, I, 305-15.

[224] Panzer, Beowulf, 313.

[225] A further and more specific parallel between Lotherus and Heremod has been pointed out by Sarrazin (Anglia, XIX, 392). It seems from Beowulf that Heremod went into exile (ll. 1714-15), and apparently mid Eotenum (l. 902) which (in view of the use of the word Eotena, Eotenum, in the Finnsburg episode) very probably means "among the Jutes." A late Scandinavian document tells us that Lotherus ... superatus in Jutiam profugit (Messenius, Scondia illustrata, printed 1700, but written about 1620).

[226] Pointed out by Panzer. A possible parallel to the old man who hides his treasure is discussed by Bugge and Olrik in Dania, I, 233-245 (1890-92).

[227] Cf. Ettmüller, Scopas and Boceras, 1850, p. ix; Carmen de Beovvulfi rebus gestis, 1875, p. iii.

[228] P.B.B. XI, 167-170.

[229] Sarrazin, Der Schauplatz des ersten Beowulfliedes (P.B.B. XI, 170 etc.); Sievers, Die Heimat des Beowulfdichters (P.B.B. XI, 354 etc.); Sarrazin, Altnordisches im Beowulfliede (P.B.B. XI, 528 etc.); Sievers, Altnordisches im Beowulf? (P.B.B. XII, 168 etc.)

[230] Beovulf-Studien, 68.

[231] Sarrazin has countered this argument by urging that since the present day Swedes and Danes have better manners than the English, they therefore presumably had better manners already in the eighth century. I admit the premises, but deny the deduction.

[232] Sedgefield, Beowulf (1st ed.), p. 27.

[233] Schück, Studier i Beovulfsagan, 41.

[234] The brief Fata Apostolorum is doubted by Sievers (Anglia, XIII, 24).

[235] Two of these occur twice: hātan heolfre, 1423, 849; nīowan stefne, 1789, 2594; the rest once only, 141, 561, 963, 977, 1104, 1502, 1505, 1542, 1746, 2102, 2290, 2347, 2440, 2482, 2492, 2692. See Barnouw, 51.

[236] 74, 99, 122, 257, 390, 412.

[237] Christ, 510.

[238] Lichtenheld omits 2011, se mǣra mago Healfdenes, inserting instead 1474, where the same phrase occurs, but with a vocative force.

[239] 758, 813, 2011, 2587, 2928, 2971, 2977, 3120.

[240] 1199.

[241] 102, 713, 919, 997, 1016, 1448, 1984, 2255, 2264, 2675, 3024, 3028, 3097.

[242] Saintsbury in Short History of English Literature, I. 3.

[243] Morsbach, 270.

[244] Morsbach, 271.

[245] Chadwick, Heroic Age, 4.

[246] "Thus in place of the expression to widan feore we find occasionally widan feore in the same sense, and even in Beowulf we meet with widan feorh, which is not improbably the oldest form of the phrase. Before the loss of the final -u it [widan feorhu] would be a perfectly regular half verse, but the operation of this change would render it impossible and necessitate the substitution of a synonymous expression. In principle, it should be observed, the assumption of such substitutions seems to be absolutely necessary, unless we are prepared to deny that any old poems or even verses survived the period of apocope." Chadwick, Heroic Age, pp. 46-7.

[247] Heroic Age, 46.

[248] Birch, Cart. Sax. No. 81. See Morsbach, 260.

[249] The most important examples being breguntford (Birch, Cart. Sax. No. 115, dating between 693 and 731; perhaps 705): heffled in the life of St Gregory written by a Whitby monk apparently before 713: -gar on the Bewcastle Column, earlier than the end of the first quarter of the eighth century and perhaps much earlier: and many names in ford and feld in the Moore MS of Bede's Ecclesiastical History (a MS written about 737).

[250] An English Miscellany presented to Dr Furnivall, 370.

[251] Grienberger, Anglia, XXVII, 448.

[252] i.e. flodu ahof might stand for flōd u[p] ăhōf, as is suggested by Chadwick, Heroic Age, 69.

[253] In the Franks casket b already appears as f, and the n of sefu, "seven," has been lost.

[254] Birch, Cart. Sax. No. 45.

[255] Chadwick, Heroic Age, 67: "In personal names we must clearly allow for traditional orthography." Morsbach admits this in another connection (p. 259).

[256] Lübke's preface to Müllenhoff's Beovulf. Both the tendencies specially associated with Müllenhoff's name—the "mythologizing" and the "dissecting"—are due to the influence of Lachmann. It must be frankly admitted that on these subjects Müllenhoff did not begin his studies with an open mind.

[257] "Es ist einfach genug"—Beovulf, 110.

[258] Möller, V.E. 140: cf. Schücking, B.R. 14.

[259] Earle, Deeds of Beowulf, xlix (an excellent criticism of Müllenhoff).

[260] Heusler, Lied u. Epos, 26.

[261] Epic and Romance, Chap. II, § 2.

[262] Ballad and Epic, 311-12.

[263] Beowulfs Rückkehr, 1905.

[264] e.g. Genesis.

[265] Chap. IV, pp. 29-33.

[266] Chap. V, pp. 34-41.

[267] Chap. VI, cf. esp. p. 50.

[268] In the portion which Schücking excludes, we twice have gǣð = gāið (2034, 2055). Elsewhere in the Return we have dōn = dōan (2166) whilst frēa (1934), Hondsciō (2076) need to be considered.

[269] 2069.

[270] 2093.

[271] Satzverknüpfung im Beowulf, 139.

[272] Þȳlǣs = "lest" (1918); ac in direct question (1990); þā occurring unsupported late in the sentence (2192); forþām (1957) [see Sievers in P.B.B. XXIX, 313]; swā = "since," "because" (2184). But Schücking admits in his edition two other instances of forþām (146 and 2645), so this can hardly count.

[273] hȳrde ic as introducing a statement, 62, 2163, 2172; sið ðan ǣrest, 6, 1947.

[274] A similar use of þā, 1078, 1988; cf. 1114, 1125, 2135.

[275] hæbbe, 1928; gēong, 2019.

[276] þurfe, 2495.

[277] Schücking, Chap. VIII.

[278] Cf. Brandl in Herrigs Archiv, CXV, 421 (1905).

[279] e.g. Blackburn in Pub. Mod. Lang. Assoc. Amer. XII, 204-225; Bradley in the Encyc. Brit. III, 760; Chadwick, H.A. 49; Clarke, Sidelights, 10.

[280] Chadwick, in Cambridge History, I, 30.

[281] We may refer especially to the account of Attila's funeral given by Jordanes. [Mr Chadwick's note.]

[282] Chadwick in The Heroic Age, 53.

[283] It is adopted, e.g., by Clarke, Sidelights, 8.

[284] Yet this is very doubtful: see Leeds, Archæology, 27, 74.

[285] Notably in Book VIII (ed. Holder, 264) and Book III (ed. Holder, 74).

[286] 'Fasta fornlämningar i Beowulf,' in Ant. Tidskrift för Sverige, XVIII, 4, 64.

[287] See Schücking, Das angelsächsische Totenklaglied, in Engl. Stud. XXXIX, 1-13.

[288] Blackburn, in Pub. Mod. Lang. Assoc. Amer. Cf. Hart, Ballad and Epic, 175.

[289] Clark Hall, xlvii.

[290] Blackburn, as above, p. 126.

[291] Chadwick, in Cambridge History, I, 30.

[292] Clark Hall, xlvii. See, to the contrary, Klaeber in Anglia, XXXVI, 196.

[293] This point is fully developed by Brandl, 1002-3. As Brandl points out, if we want to find a parallel to the hero Beowulf, saving his people from their temporal and ghostly foes, we must look, not to the other heroes of Old English heroic poetry, such as Waldhere or Hengest, but to Moses in the Old English Exodus. [Since this was written the essentially Christian character of Beowulf has been further, and I think finally, demonstrated by Klaeber, in the last section of his article on Die Christlichen Elemente im Beowulf, in Anglia, XXXVI; see especially 194-199.]

[294] Cf. Beowulf, ll. 180 etc.

[295] Bradley, in Encyc. Brit.

[296] Bradley, in Encyc. Brit. III, 760-1.

[297] Blackburn, 218.

[298] See Finnur Jónsson, Den Norsk-Islandske Skjaldedigtning, B. ii. 473-4.

[299] MS A, followed by Magnússon, makes Glam bláeygðr, "blue-eyed": Boer reads gráeygðr, considering grey a more uncanny colour.

[300] MS A has fon^m or fen^m, it is difficult to tell which. Magnússon reads fenum, "morasses."

[301] Immediately inside the door of the Icelandic dwelling was the anddyri or vestibule. For want of a better word, I translate anddyri by "porch": but it is a porch inside the building. Opening out of this 'porch' were a number of rooms. Chief among which were the skáli or "hall," and the stufa or "sitting room," the latter reached by a passage (gǫng). These were separated from the "porch" by panelling. In the struggle with Glam, Grettir is lying in the hall (skáli), but the panelling has all been broken away from the great cross-beam to which it was fixed. Grettir consequently sees Glam enter the outer door; Glam turns to the skáli, and glares down it, leaning over the cross-beam; then enters the hall, and the struggle begins. See Guðmundssen (V.), Privatbolegen på Island i Sagatiden, 1889.

[302] The partition beams (set-stokkar) stood between the middle of the skáli or hall and the planked daïs which ran down each side. The strength of the combatants is such that the stokkar give way. Grettir gets no footing to withstand Glam till they reach the outer-door. Here there is a stone set in the ground, which apparently gives a better footing for a push than for a pull. So Grettir changes his tactics, gets a purchase on the stone, and at the same time pushes against Glam's breast, and so dashes Glam's head and shoulders against the lintel of the outer-door.

[303] So MS 551 a. Magnússon reads dvaldist þar "he stayed there."

[304] Meaning that an attack by the evil beings would at least break the monotony.

[305] A passage (gǫng) had to be traversed between the door of the room (stufa) and the porch (anddyri).

[306] MSS bælt. Boer reads bolat "hewn down."

[307] A night troll, if caught by the sunrise, was supposed to turn into stone.

[308] Skúta may be acc. of the noun skúti, "overhanging precipice, cave"; or it may be the verb, "hang over." Grettir and his companion see that the sides of the ravine are precipitous (skúta upp) and so clean-cut (meitil-berg: meitill, "a chisel") that they give no hold to the climber. Hence the need for the rope. The translators all take skúta as acc. of skúti, which is quite possible: but they are surely wrong when they proceed to identify the skúti with the hellir behind the waterfall. For this cave behind the waterfall is introduced in the saga as something which Grettir discovers after he has dived beneath the fall, the fall in front naturally hiding it till then.

The verb skúta occurs elsewhere in Grettis saga, of the glaciers overhanging a valley. Boer's attempt to reconstruct the scene appears to me wrong: cf. Ranisch in A.f.d.A. XXVIII, 217.

[309] The old editions read fimm tigir faðma "fifty fathoms": but according to Boer's collation the best MS (A) read X, whilst four of the five others collated give XV (fimtán). The editors seem dissatisfied with this: yet sixty to ninety feet seems a good enough height for a dive.

[310] ok sat þar hjá, not in MS A, nor in Boer's edition.

[311] The two poems are given according to the version of William Morris.

[312] On his first arrival at Leire, Bjarki had been attacked by, and had slain, the watch-dogs (Rímur, IV, 41): this naturally brings him now into disfavour, and he has to dispute with men.

[313] Reading kappana.

[314] The MSS have either Sandeyar or Saudeyar (Sauðeyar). But that Sandeyar is the correct form is shown by the name Sandø, which is given still to the island of Dollsey, where Orm's fight is localized (Panzer, 403).

[315] Literally "she-cat," ketta; but the word may mean "giantess." It is used in some MSS of the Grettis saga of the giantess who attacks Grettir at Sandhaugar.

[316] See Sweet, Oldest English Texts, 1885, p. 170.

[317] See Catalogue of MSS. in the Library of Corpus Christi College, Cambridge by Montague Rhodes James, Camb., 1912, p. 437.

[318] See Publications of the Palæographical Society, 1880, where a facsimile of part of the Vespasian MS is given. (Pt. 10, Plate 165: subsequently Ser. I, Vol. II.)

[319] So Zimmer, Nennius Vindicatus, Berlin, 1893, pp. 78 etc., and Duchesne (Revue Celtique, XV, 196). Duchesne sums up these genealogies as "un recueil constitué, vers la fin du VII^e siècle, dans le royaume de Strathcluyd, mais complété par diverses retouches, dont la dernière est de 796."

[320] This is shown by one of the supplementary Mercian pedigrees being made to end, both in the Vespasian genealogy and the Historia Brittonum, in Ecgfrith, who reigned for a few months in 796. See Thurneysen (Z.f.d.Ph. XXVIII, 101).

[321] Ed. Mommsen, p. 203.

[322] Anno 626: a similar genealogy will be found in these MSS and in the Parker MS, anno 755 (accession of Offa II).

[323] Zimmer (Nennius Vindicatus, p. 84) argues that this Geta-Woden pedigree belongs to a portion of the Historia Brittonum written down A.D. 685. Thurneysen (Z.f.d.Ph. XXVIII, 103-4) dates the section in which it occurs 679; Duchesne (Revue Celtique, XV, 196) places it more vaguely between the end of the sixth and the beginning of the eighth century; van Hamel (Hoops Reallexikon s.v. Nennius) between much the same limits, and clearly before 705.

[324] Zimmer (p. 275) says A.D. 796; Duchesne (p. 196) A.D. 800; Thurneysen (Zeitschr. f. Celtische Philologie, I, 166) A.D. 826; Skene (Four Ancient Books of Wales, 1868, I, 38) A.D. 858; van Hamel (p. 304) A.D. 820-859. See also Chadwick, Origin, 38.

[325] Bradshaw, Investigations among Early Welsh, Breton and Cornish MSS. in Collected Papers, 466.

[326] See above, p. 196.

[327] Cf. Bretwalda.

[328] The genealogies have recently been dealt with by E. Hackenberg, Die Stammtafeln der angelsächsischen Königreiche, Berlin, 1918; and by Brandl, (Herrig's Archiv, CXXXVII, 1-24). Most of Brandl's derivations seem to me to depend upon very perilous conjectures. Thus he derives Scēfing from the Gr.-Lat. scapha, "a skiff": a word which was not adopted into Old English. This seems to be sacrificing all probability to the desire to find a new interpretation: and, even so, it is not quite successful. For Riley in the Gentleman's Magazine, August, 1857, p. 126, suggested the derivation of the name of Scef from the schiff or skiff in which he came.

[329] For a list of the Icelandic versions, see Heusler, Die gelehrte Urgeschichte im altisländischen Schrifttum, pp. 18-19, in the Abhandlungen d. preuss. Akad., Phil. Hist. Klasse, 1908, Berlin.

[330] The names are given as in the Trinity Roll (T), collated with Corpus (C) and Moseley (M). For Paris (P) I follow Kemble's report (Postscript to Preface, 1837, pp. vii, viii: Stammtafel der Westsachsen, pp. 18, 31). All seem to agree in writing t for c in Steph and Steldius, and in Boerinus, obviously, as Kemble pointed out, r is written by error for ƿ = Beowinus [or Beowius]; Cinrinicius T, Cinrinicus C, Cininicus P, Siuruncius M; Suethedus TCP, Suechedius M; Gethius T, Thecius M, Ehecius CP; Geate T, Geathe CM, Geathus P.

[331] I follow the spelling of the Moseley roll in this note.

[332] Dacia = "Denmark": Dacia and Dania were identified.

[333] uocabitur, Gertz; uocatur, all MSS.

[334] This account of the peaceful reign of Ro is simply false etymology from Danish ro, "rest."

[335] Note that Ro (Hrothgar), the son of Haldanus (Healfdene), is here represented as his father. Saxo Grammaticus, combining divergent accounts, as he often does, accordingly mentions two Roes—one the brother of Haldanus, the other his son. See above, pp. 131-2.

[336] cum piratica classe, Langebek; the MSS have cum pietate (!) with or without classe.

[337] post quem, Holder-Egger, Gertz; postquam, all MSS.

[338] Snyo: the viceroy whom Athisl had placed over the Danes.

[339] in added by Gertz; omitted in all MSS.

[340] A scribal error for transalbinas, "beyond the Elbe."

[341] Assembly.

[342] Island.

[343] I have substituted u for v, and have abandoned spellings like theutones, thezauro, orrifico, charitas, phas (for fas), atlethas, choercuit, iocundum, charum, fœlicissima, nanque, hæreditarii, exoluere.

The actual reading of the 1514 text is abandoned by substituting: p. 130, l. 3 ingeniti for ingenitis (1514); p. 132, l. 22, iacientis for iacentis; p. 134, l. 2, diutinæ for diutiuæ; p. 136, l. 11, fudit for fugit; p. 136, l. 20, ut for aut; p. 137, l. 8, ammirationi for ammirationis; p. 137, l. 16, offert for affert; p. 137, l. 17, Roluoni for Rouolni; p. 137, l. 27, ministerio for ministros; p. 137, l. 33 diuturnus for diuturnius; p. 206, l. 22, diutinam for diutina; p. 207, l. 3, ei for eique; p. 207, l. 5, destituat for deficiat; p. 209, l. 2, latere for latera; p. 209, l. 5, conscisci for concissi; p. 209, l. 14, defoderat for defodera.

[344] Above this heading B has Gesta Offe Regis merciorum.

[345] A repeats sibi after constitueret.

[346] Hic Riganus binomin[i]s fuit. Vocabatur enim alio nomine Aliel. Riganus uero a rigore. Huic erat filius Hildebrandus, miles strenuus, ab ense sic dictus. Hunc uoluit pater promouere: Contemporary rubric in A, inserted in the middle of the sketch representing Riganus demanding the kingdom from Warmundus.

[347] optat, B.

[348] celebri, B; celibri, A.

[349] hoc, B.

[350] ueheementer, A.

[351] ueheementi, A.

[352] eciam, B.

[353] Added in margin in A; not in B.

[354] hec omitted, B.

[355] Added in margin in A; not in B.

[356] dereliquerunt, B.

[357] precipue omitted, B.

[358] ei omitted, B.

[359] Qualmhul vel Qualmweld in margin, A.

[360] planies, A: planicies, perhaps corrected from planies, B.

[361] blodifeld, B.

[362] Gloria triumphi, in margin, A.

[363] tripudium, B; tripuduum, A.

[364] scis, A, B.

[365] menbra, A.

[366] gracias, B.

[367] hosstibus, A.

[368] romotis, A.

[369] congnouerunt, A.

[370] Warmandi, A.

[371] habenas repeated after regni above in A, but cancelled in B.

[372] exaggeret, B.

[373] pulcritudinis, B; pulchritudini, A.

[374] ingnota, A.

[375] euuangelii, B.

[376] consingnatas, A.

[377] from B, written over erasure.

[378] scribitur, B.

[379] Epistola, in margin, A.

[380] incongnita, A.

[381] dicebant, B.

[382] frustratim, A, B.

[383] ossium, B.

[384] congnouit, A.

[385] hoc omitted, B.

[386] congnicione, A.

[387] sui, A.

[388] obtemperare, B.

[389] menbra, A.

[390] qui, AB; quae, Wats.

[391] recongnosce, A.

[392] sancte et dulcissime, B.

[393] ut added above line, A, B.

[394] scenobium, A; the si s erased in B.

[395] deo, B

[396] tuinfreth, B.

[397] scenobio, A; s erased B.

[398] de tirannide Beormredi regis Mercie, B.

[399] fecerat, wanting in A; added in margin, B.

[400] Pinefredum, B; Penefredum, A, but with i above in first case.

[401] uariis repeated, A; second variis cancelled, B.

[402] considerans, B, inserted in margin; omitted, A.

[403] Marcelline, A; Marcell, B.

[404] vixisset, B, inserted in margin; omitted, A.

[405] Alberto, etc. passim, B.

[406] virtutibus, in margin, later hand, A; in B, over erasure.

[407] est in margin, A.

[408] et omitted, B.

[409] innotuerunt, B.

[410] in pietatis manu, B.

[411] premissimis, A.

[412] sinistrum, B.

[413] quam in margin, A; over erasure, B.

[414] Space for cap. left vacant, A.

[415] aucmentum, A.

[416] facinoris, B.

[417] congnouit, A.

[418] celeriter, B.

[419] cum in A is inserted after peruenisset, instead of before: and this was probably the original reading in B, although subsequently corrected.

[420] per, B.

[421] corrected to nullatenus dormire quasi suspectam permisit, B.

[422] Justa Vindicta, A, in margin.

[423] Mr Mackie, in an excellent article on the Fragment (J.E.G.Ph. XVI, 251) objects that my criticism of Hickes' accuracy "is not altogether judicial." Mackie urges that, since the MS is no longer extant, we cannot tell how far the errors are due to Hickes, and how far they already existed in the MS from which Hickes copied.

But we must not forget that there are other transcripts by Hickes, of MSS which are still extant, and from these we can estimate his accuracy. It is no disrespect to the memory of Hickes, a scholar to whom we are all indebted, to recognize frankly that his transcripts are not sufficiently accurate to make them at all a satisfactory substitute for the original MS. Hickes' transcript of the Cottonian Gnomic Verses (Thesaurus, I, 207) shows an average of one error in every four lines: about half these errors are mere matters of spelling, the others are serious. Hickes' transcript of the Calendar (Thesaurus, I, 203) shows an average of one error in every six lines. When, therefore, we find in the Finnsburg Fragment inaccuracies of exactly the type which Hickes often commits, it would be "hardly judicial" to attribute these to the MS which he copied, and to attribute to Hickes in this particular instance an accuracy to which he has really no claim.

Mr Mackie doubts the legitimacy of emending Garulf to Garulf[e]: but we must remember that Hickes (or his printer) was systematically careless as to the final e: cf. Calendar, 15, 23, 41, 141, 144, 171, 210; Gnomic Verses, 45. Other forms in the Finnsburg Fragment which can be easily paralleled by Hickes' miswritings in the Calendar and Gnomic Verses are