page 5 note 1 BMO Gr. xxviii.

page 5 note 2 Stockholm, 1818.

page 5 note 3 Sveinbjörn Egilsson, Edda Snorra Sturlusonar 160–9 (Reykjavig, 1848); Edda Snorra Sturlusonar 2.10–43 (Sumptibus Legati Arnamagnæani, Copenhagen, 1852).

page 5 note 4 Here referred to as DJ; see the list of bibliographical abbreviations.

page 6 note 5 Codex Wormianus (Levin and Munksgaard, Copenhagen, 1931; Corpus Codicum Islandicorum Medii Aevi II).

page 6 note 6 Cf. bibliographical abbreviations. For further titles see bibliography in AH 116–20, and H. Hermannsson, Catalogue of the Icelandic Collection (Ithaca, N. Y., 1914, 1927, 1943). Miss Holtsmark's excellent study forms the natural point of departure for further research; even though this writer has been impelled to differ from her conclusions on occasion, he is deeply conscious of his indebtedness to her.

page 6 note 7 Gustav Neckel and F. Niedner, Die jüngere Edda mit dem sogenannten ersten grammatischen Traktat 334–48 (Jena, 1925; Thule Series).

page 7 note 8 DJ x.

page 7 note 9 Axel Kock, Svensk ljudhistoria 4.467; AH 30.

page 7 note 10 AH 89.

page 7 note 11 BMO Gr. xxxv ff.

page 8 note 12 ANF 53.119–20 (1937).

page 8 note 13 DJ xxix.

page 8 note 14 Independently suggested by Hallfrid Christiansen in a paper read at the Oslo Videnskaps-Akademi in 1946, which she was kind enough to let me read.

page 9 note 1 Cf. introduction to phototypic edition by SigurÐur Nordal.

page 9 note 2 Op. cit. 16.

page 9 note 3 A partial listing of the orthographical peculiarities of Codex Wormianus was made by Finnur Jónsson in AM Ed. 3.2.51 ff. One should also consult his diplomatic reprint of the text in Edda Snorra Sturlusonar. Codex Wormianus. AM 242, fol. (Copenhagen and Kristiania, 1924).

page 9 note 4 The numbers refer to page and line as numbered in the manuscript and in this edition. If two or more successive references are to the same page, they are combined, as in 86.1.2 = page 86, lines 1 and 2.

page 9 note 5 Cf. DJ xi.

page 9 note 6 Barring the fact that they chose not to mark nasality outside of the examples illustrating this phenomenon; AH has overlooked this statement in saying that DJ marked ‘all nasal vowels’ (99).

page 10 note 7 AH 99.

page 10 note 8 The use of the symbol φ is noticeably more frequent in the FGT than in the rest of the Codex, where it occurs only twice in the Edda (78.17; 167.27) and four times in the Third Grammatical Treatise (94.10; 97.26; 98.4; 99.4). This may be due to its occurrence in the list of symbols; but it is used quite indiscriminately for historical œ, æ, α, φ, thus reflecting the confusion of these sounds in later Icelandic.

page 10 note 9 In case of doubt the spellings of Leiv Heggstad in Gamalnorsk Ordbok² (Oslo, 1930) have been followed, except that -sk has been restored for -z in the reflexive.

page 11 note 10 In case of doubt, DJ's practice has been followed. In the present edition of the text, typographical exigencies have made it necessary to substitute somewhat different characters for some of the FG's letters, particularly for his kappa and crossed g. The reader is directed to the facsimile of the alphabet, 90.15–32, on p. 30 below.

page 14 note 11 See section 4.10.

page 14 note 12 Commonly used in the sense of ‘Scandinavian’.

page 16 note 13 The sense is not clear here.

page 18 note 14 þódólfr Arnórsson; cf. section 5.7.

page 19 note 15 Or, possibly, ‘pouring’ or ‘bilgewater’.

page 20 note 16 Óttarr the Black; cf. section 5.7.

page 23 note 17 Apparently meaning the Irish here, as often in the Middle Ages; see section 5.3.

page 28 note 18 Thorgerd was worshipped in northern Norway, and according to the Heimskringla Earl Hákon sacrificed his son to her in a battle with the Jomsvikings, which won him the day.

page 29 note 19 Thor received the great brewing kettle from Hýmir's hall, according to the HýmiskviÐa, a lay in the Poetic Edda.

page 30 note 20 DJ (82) omit this final alphabet, which appears in the MS., on the plea that it is a superfluous addition by the scribe, since FG says he has already written the alphabet. While the parts of the alphabet have been listed previously, this is the first time all the letters are set up in their new sequence. Tittle is here included for the first time, and there are none of the scribal errors (e.g. inclusion of H and X) which mar the earlier lists.

page 31 note 1 Cf. Noreen 10 ff.; AH 10–11.

page 31 note 2 Runerne i den oldislandske literatur 94 (Copenhagen, 1883).

page 33 note 3 Neckel and Niedner, Die jüngere Edda 45.

page 33 note 4 K. Gíslason, Annaler f. nordisk Oldkyndighed 1863.400 ff.; BMO Gr. xxviii, note 1.

page 33 note 5 K. J. Lyngby was the first to show the validity of his analysis in Tidskrift f. Philologie og Pædagogik 2.317 (Copenhagen, 1861); further confirmation was offered by Sophus Bugge in ANF 2.230–4 (1885); and the final vindication was made by Adolf Noreen in ANF 3.1–41 (1886).

page 34 note 5a Hallfrid Christiansen has pointed out the absence of contrast between short orals and short nasals: the latter occur only before nasal consonants. Nasality, in other words, was distinctive only for long vowels. Mrs. Christiansen regards his failure to state this as a weakness in his analysis; but it is hard to see how he could have handled it otherwise with his equipment. Once he had abstracted nasality as distinctive, he marked it wherever he found it, a procedure which is defensible even in modern terms.

page 34 note 6 Axel Kock explained the form as due to analogy with the other forms of the word sýja ‘sew’, Svensk ljudhistoria 4.474 (Lund, 1921); Sophus Bugge ascribed it to earlier unsyncopated forms where the n did not follow the vowel, ANF 2.230–4 (1885).

page 34 note 7 Cf. D. A. Seip, Norsk Språkhistorie til omkring 1370, p. 82 (Oslo, 1931); AH 27 ff.

page 35 note 8 AH 92–3.

page 35 note 9 Cf. von Friesen in ANF 43.110–30 (1927).

page 35 note 10 Noreen 116; Marstrander, Bidrag 79.

page 35 note 11 Cf. Noreen 87 ff.; for a new formulation of this process see John Svensson, Diftongering med palatalt förslag i de nordiska språken (Lund, 1944).

page 35 note 12 It is clear that the accent had shifted to the second vowel in the FG's time; otherwise there would have been no basis for argument concerning its first element. This raises the question of whether his fourth example should be written eor or ior (MS ivr). DJ restore to eo, while Marius Kristensen contends that the author was illustrating all the possible vowels that could be used as glides and therefore had i in this word (Nordiska Studier till. A. Noreen 21). Although this view is attractive, it has not been accepted here for the following reasons: (1) the derivation of the word from ehwa- gives us no warrant for supposing that its quality was more i-like than that of earn; (2) in the absence of a phonemically valid distinction in this position, it is most unlikely that he would have distinguished the first vowel of earn from that of eor; (3) by writing eo in this word the list of examples becomes perfectly alphabetic (which Kristensen noticed, but changed ey to φy, a quite unlikely emendation, in order to preserve the io). Changing the ms. reading here is justified by the fact that earn is also spelled with i in spite of the author's express statement to the contrary. The sound in both words must have been a glide of indeterminate phonetic quality, which could be written e or i as one pleased since it did not occur before either of these vowels. The FG shows by his somewhat ill-tempered tone that he lacks a good criterion for settling the problem; and so he has to resort to the far-fetched appeal to skaldic authority and the judgment of older men, in whose speech the glide may still have preserved a little more of the traditional e-quality.

page 36 note 13 Bernard Bloch and George Trager, Outline of Linguistic Analysis 23 (Baltimore, Linguistic Society of America, 1942).

page 36 note 14 This is also the view of Marius Kristensen in Nordiska Studier tillegnade Adolf Noreen 16–24 (Uppsala, 1904), though arrived at by a slightly different process of reasoning. While AH accepts his view, she does not appear to have drawn the full conclusions from it.

page 37 note 15 As shown by Geo. T. Flom in Festskrift Pipping 100–13 (Helsingfors, 1924), he may have gotten the idea from the occasional writing of N for nn in Norwegian manuscripts of his day; AH, however, does not accept this theory (34).

page 37 note 16 AH 40.

page 37 note 17 AH 98.

page 37 note 18 Noreen 40 ff.

page 38 note 19 S. Einarsson, Icelandic (Baltimore, 1945).

page 40 note 1 AH 87.

page 40 note 2 For excellent statements on the skalds and their importance as linguistic arbiters, see Lee M. Hollander, The Skalds (New York, 1945).

page 40 note 3 87.4–5.

page 40 note 4 ANF 53.113 (1937); originally intimated by Björn M. Ólsen in Runerne i den oldislandske literatur 103–4 (1883).

page 41 note 5 Runerne 13.

page 41 note 6 Also emphasized by Finnur Jónsson, Litt. hist. II.1.254 (1898); G. Storm in ANF 2.172–6 (1885) criticises BMO's theory concerning the runic influences in FGT.

page 41 note 7 Heinrich Keil, Grammatici latini (Leipzig, 1857–78). The quotation from Donatus is in vol. 4, p. 367.

page 42 note 8 Keil 4.368.

page 42 note 9 Keil, Gr. Lat., Suppl. 225.

page 42 note 10 Ælfrics Grammatik und Glossar 5, ed. Julius Zupitza (Berlin, 1880).

page 42 note 11 The Third Gram. Treat. quotes Priscianus as saying that ‘each vowel can have ten or more sounds’; BMO Gr. 40.

page 42 note 12 Vincent de Beauvais, Speculum, cited by AH (29) from Ch. Thurot, Notices et extraits 23 f.

page 42 note 13 E.g. N. S. Trubetzkoy, Grundzüge der Phonologie 36 (Prague, 1939): ‘Daraus folgt, dass ein Phonem durch mehrere verschiedene Sprachlaute realisiert werden kann’. Since the above discussion of the litera was written, an illuminating article by David Abercrombie has appeared in Lingua 2.54–63 (1949), entitled What is a “Letter”? His study of English orthoepists of the 17th century leads him to substantially the same conclusion as that of the present writer. He does not directly equate the litera with the phoneme, but suggests it when he questions whether ‘if letter had been retained in something like its traditional functional sense, the need for a phoneme theory would ever have arisen …‘ When the phoneticians abandoned the litera, the reason was of course the very ambiguity in meaning which Abercrombie points out: the word could mean figura and potestas and nomen, as well as all three. A distinction was imperative if scholars were to get away from unrealistic speculation on the relations of the attributes.

page 43 note 14 Cf. 84.16.

page 44 note 15 AH 87.

page 44 note 16 Ælfric 5.

page 44 note 17 BMO Gr. xxvii.

page 44 note 18 AH 17–8.

page 44 note 19 AH 26.

page 45 note 20 W. Keller, Über die Akzente in den ags. Handschriften, Prager deutsche Studien 8.1–120 (1908); Wm. H. Hulme, Quantity Marks in OE Mass., MLN 11.17–25 (1896).

page 45 note 21 AH 30.

page 45 note 22 Axel Kock, Svensk ljudhistoria 4.467; AH 30.

page 45 note 23 The frequent use of φ in the text of FGT may constitute a confirmation of this (cf. above, paragraph 2.3). The last scribe probably copied the φ's from a preceding manuscript, but the fact that they are used erroneously (for , etc.) points to a still earlier scribe.

page 45 note 24 No mention of this list appears in the literature; yet it shows certain interesting features, e.g. that it is copied in the wrong order, but with substantially the same contents from FGT 88.13. It gives the impression that the scribe's eye followed the double line (e.g.^ebbb), slipping baek and forth from the upper to the lower lines. It includes the vowels from FGT 85.27, but skips from the upper to the lower line before the former is concluded; it copies the errors in the consonant list (H X), but it includes one correct item which is not in FGT 88.13, namely the name the for þ. Hence I conclude that it is copied not from this Codex, but from an earlier manuscript. Here, also, the d has a straight stave.

page 46 note 25 Cf. the customary types of the latter at 77.10, 82.31 (N); 75.24, 77.19.31 (L); 75.27, 77.32 (D).

page 46 note 26 Harald Spehr, Der ursprung der isländischen schrift 120 (Halle/Saale, 1929).

page 46 note 27 AH 18 ff.

page 46 note 28 NTSpr 9.371.

page 46 note 29 DJ 69.

page 46 note 30 AH 20.

page 47 note 31 Note that the Arnamagnæan edition of FGT translates both atkvæÐi and jartein into Latin as potestas. AM Ed. 2.28, etc.

page 47 note 32 Since in FGT it everywhere refers to the sound in general, specifically the mass of it, while English ‘pronunciation’ specifies rather the quality of sound (distinction from other sounds), it seemed best to use ‘sound’ rather than ‘pronunciation’ in translating it. DJ have used lydmasse ‘sound mass’ quite often, which would be awkward in English and also seemed a little too specific and modern. In 84.21 they have even translated it stavelse ‘syllable’, although the text has a regular word samstafa for the syllable.

page 47 note 33 The term ‘meaning’ (corresponding to DJ's usual translation betydning) seemed too closely associated with semantic functions; hence jartein has here usually been translated with ‘value’.

page 47 note 34 Ælfric's renderings show no parallel: clypjendlice for vowels, samod swegende for consonants; op.cit. 5.

page 48 note 35 Our term ‘spell’, which is one of its meanings, refers too narrowly to the symbol itself, so the term ‘combine’ has most often been used to render it.

page 48 note 36 Cf. Ælfric, who translates it as stæfgefeg (op.cit. 7).

page 48 note 37 AH 15; on p. 89 the omissions are erroneously listed as k, z, y, q, x.

page 48 note 38 89.16.

page 49 note 39 Cf. the lists at 85.27, 88.13, 90.31.

page 49 note 40 The errors referred to are the inclusion of capitalized H and X in the consonant list.

page 49 note 41 84.24.

page 49 note 42 85.29–31.

page 49 note 43 89.2.

page 50 note 44 Cf. Seip in NTSpr. 9.364–5.

page 50 note 45 90.5.

page 50 note 46 90.5.

page 50 note 47 89.15.

page 50 note 48 88.32; 89.20; 90.5.

page 50 note 49 89.16 ff.

page 50 note 50 Byrhtferth's Manual 1.196 (ed. S. J. Crawford, London, 1929).

page 51 note 51 AH 67.

page 51 note 52 AH 40.

page 51 note 53 AH 42.

page 51 note 54 Cf. C. D. Buck, The letter “Y”, Manly Anniversary Studies 340–50 (Chicago, 1923).

page 52 note 55 89.8.

page 52 note 56 Cf. Spehr 37 ff.

page 52 note 57 Spehr 32.

page 52 note 58 Kr. Kaalund, Palæografisk Atlas, Oldnorsk-islandsk Afdeling (Copenhagen, 1905).

page 52 note 59 Der ursprung 54.

page 52 note 60 Der ursprung 39–43.

page 53 note 61 Der ursprung 54.

page 53 note 62 Buck, loc.cit. 349.

page 53 note 63 Mémoires de la Société de linguistique de Paris 8.188 (1894).

page 53 note 64 E. S. Sheldon, The Origin of the English Names of the Letters of the Alphabet, Studies and Notes in Philology and Literature 1.66–87 (1892).

page 53 note 65 Buck, loc.cit.; cf. E. S. Sheldon, Further Notes on the Names of the Letters, Studies and Notes in Philology and Literature 2.115–71 (1893).

page 54 note 66 AH 69; BMO Gr. xxvii.

page 54 note 67 L. Havet, Mémoires de la Soc. de ling. de Paris 6.79–82 (1889).

page 54 note 68 AH 70.

page 54 note 69 For his biography see Dictionary of National Biography, s.v. Baret.

page 54 note 70 Baret's Alvearie has other interesting remarks as well about orthography, distributed under the respective letters. He is quite critical of the current orthography and bases his criticisms in part on some of the same traditional grammarian's doctrines that must have been familiar to the FG. Under C he refers to those ‘which haue deuised how to frame both C and H into one character naming it Che, as in olde time our auncesters the English Saxons haue done.’ Under T he refers to the pronunciation of t as th: ‘In times past within our memorie, T hath beene pronounced at the end of Latine wordes before a vowell, as though it were th. amath, caputh, vocath, &c.’ Under X he writes: ‘And as for X, I thinke, they might as well saie that & and 2 are letters, as it is. For surelie ech of them seemeth to be a syllable, rather than a single letter, and euerie one of them to be compacted of diuers and sundrie letters.'

page 55 note 71 G. Neckel, Die jüngere Edda 46 (Jena, 1925):

page 56 note 1 R-M. S. Heffner informs me that Notker (d. 1022) was the first OHG writer who systematically distinguished long and short accented vowels; he also marked length in inflectional morphemes and stress which was correlated with sentence patterns. Cf. E. H. Sehrt and T. Starck, Notkers des deutschen Werke l.ix (Halle/Saale, 1933); Sehrt in JEGP 35.331–6 (1936); O. Fleischer, in ZfdPh 14.128–72, 285–300 (1882).

page 56 note 2 The question of a possible connection between the Scandinavian Orm in England and the FG in Iceland is an alluring one. But a cursory study of the text and of the analyses made by the best students of Orm's orthography has not brought out any very promising parallels; cf. Erik Björkman, Orms doppelkonsonanten, Anglia 37.350–81, 494–6 (1913) and Joseph Hall, Selections from Early Middle English 2.479–492 (Oxford, 1920). Orm doubled consonants for length, even in unstressed syllables, and marked vowel length by means of single accents; but he also used double and triple accents for emphasis and breves for short vowels, and had no æ, φ, or y for umlauts.

page 56 note 3 Auraicept na n-Eces, ed. George Calder (Edinburgh, 1917).

page 56 note 4 AH 83.

page 56 note 5 Mildred K. Pope, The tractatus orthographiae of T. H., Parish studentis, MLR 5.185–193 (1910).

page 56 note 6 Björn M. Ólsen did the pioneer work in tracing his sources in his Runerne 83 ff. (1883), also in BMO Gr. xxv ff. (1884). His views have been supplemented and criticised by Anne Holtsmark (op.cit.), D. A. Seip in NTSpr 9.352–71 (1938), and Magnus Olsen in ANF 53.109–46 (1937).

page 57 note 7 W. J. Chase, ed., The Ars Minor of Donatus (Madison, Wis., 1926; U. of Wis. Studies in Soc. Sci. and Hist., No. 11).

page 57 note 8 Keil 4.368.

page 57 note 9 Priscian teaches the superiority of the vowels over the consonants, comparing the former to the spirit, the latter to the body (Keil 2.13); Remigius advances the derivation of titulus from Titan (Commentum in Artem primam Donati 1) and the doctrine that vowels can be long or short (Keil, Gr. Lat. Suppl. 224). Priscian even suggests that k is superfluous (Keil 2.12) and may be the source of FG's preference for c.

page 57 note 10 AH 42; she also notes that he substitutes pronuntiatio for potestas, with which she equates FG's atkvæÐi (but note that FG also uses jartein and that Remigius, too, uses pronuntiatio).

page 57 note 11 C. H. Haskins, List of Textbooks from the Close of the Twelfth Century (where we find Priscian, Donatus, and Remigius representing the field of grammar), Harvard Studies in Classical Philology 20.75–94 (1909); cf. AH 76.

page 57 note 12 AH 58, from a commentary on the Ars minor of ca. a.d. 900.

page 57 note 13 AH 58.

page 57 note 14 AH 70 ff.

page 57 note 15 Cf. Remigius in Keil, Suppl. 224.

page 58 note 16 AH 34, 40.

page 58 note 17 H. Penzl, The Phonemic Split of Germanic k in Old English, Language 23.37 (1947), fn. 23; D. A. Seip, NTSpr 9.365 ff.

page 58 note 18 NTSpr 9.364.

page 58 note 19 Kr. Nyrop, Grammaire historique de la langue française³ 1.380 f. (Copenhagen, 1914); Schwan-Behrens-Bloch, Grammaire de l'ancien français⁴ §279 (Leipzig, 1932); W. von Wartburg, Évolution et structure de la langue française 112 (Leipzig, 1934).

page 58 note 20 AH 57, 61.

page 58 note 21 AH 65.

page 58 note 22 Above, §4.25.

page 58 note 23 AH 83. Cf. Charles Thurot, Notices et extraits (Paris, 1868).

page 58 note 24 AH 25.

page 59 note 25 See Geo. T. Flom, Studies in Scandinavian Paleography, JEGP 14.530–43 (1915); Seip, NTSpr 9.352 f.

page 59 note 26 Hallfrid Christiansen, Verdens første fonolog, unpublished essay.

page 59 note 27 Björn M. ólsen's theory that the ng was derived from the runic k for ng (BMO Gr. xxvii) has received no support from other scholars. AH (38) suggests and disproves a possibility of deriving it from the Irish Ogam. Spehr's theory that the nasal dot comes from Irish is discredited by AH (26); she thinks it was simply the only diacritic left, probably suggested by the dot over the y.

page 59 note 28 AH 28.

page 59 note 29 BMO Gr. xx; Konrad Maurer, Island 225 ff., 244 ff., 259 ff.

page 59 note 30 F. Jónsson, Litt. Hist. II.1.349 (1898); BMO Gr. xviii.

page 60 note 31 The quotation from þjódólfr consists of the two lines on 86.14, from Óttar on 87.6. Both are found in Snorri's Heimskringla in more complete forms. Cf. DJ 73–5.

page 60 note 32 Cf. Magnus Olsen's poetic and ingenious characterization in ANF 53.121–2 (1937). Neckel, in his Thule Series translation (48), suggests that the examples ‘lassen vermuten dass der gelehrte Mann Frau und Kinder hatte’.

page 60 note 33 The later date was proposed by Marius Hægstad and concurred in by Geo. T. Flom and Anne Holtsmark (AH 16). The early date was proposed by Sveinbjörn Egilsson in AM Ed. 2.10 (1852) and supported by Finnur Jónsson (DJ x and Litt. Hist. 2.928), Halldór Hermannsson (Islandica 22.42), Jón Helgason (Norrøn Literaturhistorie 104), Fredrik Paasche (Norsk Litt. Hist. 1.305 f.), D. A. Seip (NTSpr 9.352), and Magnus Olsen (ANF 53.109).

page 60 note 34 AH 13–6.

page 60 note 35 þóroddr by Gudbrand Vigfusson in Sturl. Saga 1.xxxviii (Oxford, 1878) and in Eyrb. Saga xli (Leipzig, 1864); followed by Maurer, Island 459; Bugge, Sæmundar Edda lxvi; Brenner, Altn. Handbuch 9.31. Eyjólfr by Halldór Hermannsson, loc. cit.

page 61 note 36 Runerne 96 (1883).

page 61 note 37 ANF 53.129 ff. (1937); Jón Helgason, Norrøn Literaturhistorie 104 (1934).

page 61 note 38 Hungrvaka, ch. 16, in Biskupasögur 1.80 (Copenhagen, 1858).

page 61 note 39 Sturlunga Saga 1.206 (Oxford, 1878).

page 61 note 40 Cf. Spehr, Der. Ursprung 90.

page 62 note 41 NTSpr 9.359.