¹ Copenhagen, N.K.S. 513b 4°, Lydgate's Life of St Edmund, apparently copied in the first half of 1788, not 1787 as stated in a late addition to its title.

² The principal earlier work is the Introduction to Malone's, Kemp facsimile (The Thorkelin Transcripts of Beowulf in Facsimile, EEMF 1 (Copenhagen, 1951))Google Scholar, where several of the points now made afresh can already be found. Kiernan's, KevinThe Thorkelin Transcripts of Beoivulf, Anglistica 25 (Copenhagen, 1986)Google Scholar should not be used without expert verification at source. Such footnotes as have been added in support of this observation cite the work as TTB. For the codicology, see now my ‘The Thorkelin Transcripts of Beowulf, a Codicological Description, with Notes on their Genesis and History’, The Library, ns 12 (1990), 1–22Google Scholar [cited as TCD]. The account presented on p. 13 of my ‘Beowulf: the Foundations of the Text’ (N[etherlands] S[ociety for] English] S[tudies] Bulletin 1.1 (04 1991), 1–22Google Scholar), when the A transcript had not yet been fully analysed, now needs correction, as does, on some points, the paper delivered to the International Society of Anglo-Saxonists at Oxford in 1993. Reference to A is by page and line (the same for transcript and facsimile), to B by present page followed by slash and facsimile page (prefixed M). Reference to the Vitellius manuscript is by folio and line according to the current foliation of 1884 (also used in Malone's, Kemp facsimile edition, The Nowell Codex: British Museum Cotton Vitellius A. XV Second MS, EEMF 12 (Copenhagen, 1963)Google Scholar) which is correct for Beowulf zs, it originally was, as it is now, and as it was when Thorkelin knew it. The historically interesting but now misleading ‘old’ foliation was unknown to Thorkelin and is irrelevant to these studies. It should perhaps be added that though the work of others is occasionally mentioned, the whole analysis has, for very good reason, been done entirely afresh.

³ TTB wholly ignores the evidence which the state of the quill can provide; and its one reference (p. 92: ‘new ink and pen begin’, B66r'M131a:19) is wrong as to the ink and guesswork as to the pen. When the pen has been mended the ink may look different to the naked eye, and any such judgement should therefore rest (and for the present study does rest) on examination under magnification and with proper illumination; and one can tell that a quill has been mended, but hardly if it is the same or a new one. It is not evident from TTB that its author was aware that steel nibs did not become common until their perfection and popularization by James Perry in the 1830s, so after Thorkelin's death; no steel nib would have worn so quickly that a new one was needed for virtually every page.

⁴ Kiernan, K. S., Beowulf and the Beowulf Manuscript (New Brunswick, NJ, 1981)Google Scholar. A revised edition, reprinting the original text with a new introductory essay, adding the text of his ‘The State of the Beowulf Manuscript 1882–1983’ (ASE 13 (1984), 23–42Google Scholar), and with a new Foreword by O'Brien, K., O'Keeffe, , was published at Ann Arbor, MI, in 1996.Google Scholar

⁵ It should be emphasized that no technical enhancement should be accepted that does not, as a check, include an adequate portion of unambiguous surrounding text. If the enhancement alters this text in any significant way, it is suspect.

⁶ See, for instance, n. 12.

⁷ The record in TTB is inaccurate. All four names on 2r/M3a have capital H written over original b, not Heorogar alone. Scyldinga 4v/M8a:11 (line 170) is the first name, Hroðgar 8v/M16a:1 (line 339) the first actor in the poem to be given a first-time capital where no stop precedes, so presumably because it has been recognized as a name. For the one this was the fourth occurrence in the manuscript text, for the other the seventh.

⁸ Exceptions are that Thorkelin, like his amanuensis, generally gives the fitt numbers a line to themselves, irrespective of the manuscript situation, that on the crowded final page he finds it impossible in some lines to get in as much as his copy-text does, and that he occasionally transposes a (part of a) word from one line to the next (or vice versa), presumably because he did not memorize the text in manuscript lines. That this occasional transposition is unintentional is suggested by the general practice of his transcription, and evidenced more particularly by the cases he noticed and corrected. Where an original reading gets onto the wrong page in this way it is always within the same opening; no argument for the use of A can therefore depend on it. It is to be noted that, tailed letters apart, Thorkelin's general faith in A was such that in his edition he repeatedly adopts A's error for his own correct reading, even when the form of that reading is supported elsewhere in his text.

⁹ The usual criteria of ink, state of the quill, etc. are of course also available; in fact only a single case has been observed where these betray the filling of an original lacuna and no dot has been identified with certainty. It should be added that examination of such letters as c and o in places where no dots are suspected shows that it is all too easy to discover imaginary dots in their bottom curve.

¹⁰ His notebook in Rigsarkivet (6431.5.E3, notebook Biblioth. Cotton.) contains two Old English passages copied out with accustomed rapidity on 2 October 1786, with corrections currente calamo:gl struck through and followed immediately by geleornis; and neatlh: t struck through and continued as neahterne. Note that the second one already shows the problem with Old English h that TTB, p. 142 mistakenly wishes to link with his visit to Scodand of a year later (cf. below, p. 41.ri.35).

¹¹ Like his suppletion of A's gio‥ntas to gigantas at 134r.20 (A4:16; the facts are misrepresented by TTB pp. 48–9 and 124Google Scholar, but correcdy stated by Zupitza, who, however, did not recognize the correcting hand as Thorkelin's; the correction from e to a in ntas of the next manuscript page appears to be done in A's ink). The amanuensis apparently only wrote gio at first (the pointed o is not really that, but apparently imitates what he thought he saw in the manuscript; in view of the spacing, the ga of gantas must have been lost on the verso), added ‥ntes at the correcting stage, then changed e to a; the rest is shown by the ink and hand to be late Thorkelin. As to the erased Latin translation, which TTB, p. 121Google Scholar, claims was written before B, such traces as remain show Danish script.

¹² The number of original ð, however, proves on proper examination to be rather larger than sug gested by TTB. Thus of the eighteen cross-strokes on ð on 181v (50v/M100a), one of which does duty for two adjoining characters, one (in eðel, 12) is clearly late. Its ink is lighter, it is written with a quill scraped thinner and so making a thinner line when drawn sideways than the pen which wrote the original page, it is not the firm stroke that all the others are, and it is written at a different angle. These others all have the colour of their surroundings, have the same width and direction, and in ten cases show ink running from the stem into the cross-stroke, either one way or both ways. It is thus evident that all seventeen are original, and that eðel provides a guide for judging other alterations on this page. According to TTB, however, eight of the seventeen (including six of those with ink running from the stem into the cross-stroke) are, without argument, later changes. The recto page tells a similar story; further checking on this point seems supererogatory.

¹³ To call this a mistake, and knowingly changing the manuscript (TTB, p. 119), though helping to bolster up claims of Thorkelin's inaccuracy, is merely silly. The whole purpose of transcription is to reproduce the manuscript in a (to us) more legible form. On the graphemic level what is done here is unexceptionable, and no different from transcribing the Old English form of r by a modern one (simple one-to-one correspondence). If this sort of argument is to be used, the first candidates for complaint should be the different forms of s in the manuscript, which Thorkelin (be it said with full justification) indiscriminately transcribes as long and short s, while the amanuensis is obviously at pains to keep them. The matter of the ð, to be sure, is much more disturbing, though (see previous note) less so than TTB suggests.

¹⁴ TTB's query (p 140) – how do we ‘confidently date the inks, not to mention the four hands’ is perfectly capable of an answer, but it requires the sort of careful palaeographical scrutiny that TTB does not give (cf. above, p. 24, n. 3). This does not, of course, mean that no doubt remains, or that calendar dates can be appended. The only real question is what was done while Thorkelin was still in England.

¹⁵ The only evidence found for the date of his return is in his manuscript autobiography at Copenhagen (Rigsarkivet, Personal File 6431.5, D: Selvbiografiske optegnelser), written a very long time after the event. He does not appear in the British Museum records again until January 1788.

¹⁶ In contrast to the use of the knife, which generally will do so, erasure with pumice need not, in good writing paper (which this is), show any obvious thin spot against the light; though thorough erasure, like that of a premature fitt capital (B[eowulf] 54:08, 172v.14), will of course show one. It is accordingly characteristic that the amanuensis touches up the last character (or more) before the erasure when beginning to write the correct text, making it quite plain what is text and what is imperfect erasure. There are also a very few instances (e.g. hiora 85:11) where, for some reason, he appears to have begun erasing a word and then stopped.

¹⁷ For further examples of these and others, see Malone, 's Introduction to the facsimile, pp. 5–21.Google Scholar

¹⁸ An examination of B's treatment of this word elsewhere in the Transcript suggests that an editor should read wedera here.

¹⁹ þæzim 140v.05, A15:02; swa me 141v.12, A16:12; full line 142v.05–06, A17:13–14.

²⁰ 132v.19/2:06 þady; 135v.01/6:10 æglaca ehtende; 162v.01/41:06 swy…scolde; 178v.01/61:02 figende, 181v.02/64:09 don; 182r.15,16/65:10 rege þ nat; 182r.18/65:11 & ðær; 182v.09/65:18 leoda; 182v.17/ 66:05 þhu hruce … [preceding cwæð has r corrected to wynn in A's ink], 183r.01/66:08 f; 183r.02/66:09 dug; 192r.20/77:06 nægling…

²¹ It is TTB's decision not to note ‘corrections of “p” to thorn or wynn in the first 918 lines of A’ (p. 46)Google Scholar that effectively barred Kiernan from finding out the truth. By no means all other correc tions have been noted by him, particularly not those of p to r, but also e.g. on the first page aldon (or what may have looked like it to the corrector) corrected to aldor (14), or similarly on the verso ante corrected to ahte (9), both done by means of over-heavy dark downstrokes. Nor has the original reading always been correctly identified. In addition, TTB eventually develops a highly idiosyncratic view of how some of A's graphs should be interpreted, a view that has the effect of enhancing A's apparent accuracy. Thus the objective p's of pœron and pado (A19:14, 144^v.1,2) are silently accepted as w's.

²² The miscorrection of scoldon, 151r.05, A27:7, reported by TTB, is in fact (as the letter spacing confirms) the opposite, correction of scoldan to scoldon in a manner wholly uncharacteristic of the amanuensis – which (and the failure to use magnification?) probably explains TTB's error. Malone, (p. 17) seems to have similarly mistaken the correction.Google Scholar

²³ It should be said that a very dark ink was also used at what appears to have been a later stage, when Thorkelin, again haphazardly, retopped a few characters that had suffered in his erasure of the interlined Latin. But it seems more likely that the only changes still made at that late date consisted in the suppletion, from B and in Insular script, of a few of the remaining gaps, rather than that all the corrections in dark ink would have been made so late. See also below, n. 24.

²⁴ India-rubber was not yet available; as a soft pencil-eraser it was just coming in at this time.

²⁵ There can of course be no certainty, but the initial use of erasure suggests a certain continuity from the earlier correction; its quick abandonment, and the premature abandonment of the whole process suggest a young man in a hurry, rather than the older Thorkelin or the amanuensis. If against this there is the fact that none of the early corrected wynns have the straight top, neither does that in fitela (cf. next note), which can in no case be due to the amanuensis. Correction in the early pages would no longer be for the latter's benefit.

²⁶ It should be remarked that the miscorrection of fitela (A28:13, 152r.6) has been done with the wynn used in correcting the early part of the manuscript, not that with the straight top. Its ink also looks darker than that of the other corrections on this page.

²⁷ See p. 26 and n. 7.

²⁸ He might indeed have turned back to capitalize these after reaching line 339 (at the earliest), but this would seem quite out of character.

²⁹ This is probably also the reason why in his correction of A he ignored the manuscript.

³⁰ The amount of special pleading in TTB (pt 3) is remarkable, and the hares it starts are pursued with such vigour that it rarely finds breath to ask itself questions. P. 129 begins with half a page of speculation based on the notion that Thorkelin wrote holme in 230/B6 = T7 = M11a:1, when the naked eye can tell (and a lens confirms) that there is no form of e there, but only a small downward hook terminating the offstroke of m, while of the claimed erasure there is no trace. The conclusions drawn and explanations offered in the other half of the page thus lack any support. The ‘many times’ Thorkelin ‘began or ended a line with the wrong word’ [ibid.; in fact fifteen for nearly 2900 lines] are easily explained by his working rapidly and not keeping discrete manuscript lines in his head, and the whole discussion is futile, starting from the mis taken notion that as we have it B was bound in England. The odd coincidences quoted to show B's dependence on A are not balanced by an equal consideration of the numerous cases where A is right and B wrong or absent, and too many of those cited do not survive investigation. Thus the cases on 182v listed on p. 131 can merely show that the page was then as bad as now: that A, though using a slower copying process, got them wrong was to be expected; that Thorkelin, trying to keep up some speed, should have done the same, even more so. On the (much poorer) recto he does not even bother to transcribe some still perfectly legible letters that would never have served him to produce any text.

³¹ In the present case, though, there is a single word (three letters) at the beginning of the line in which the number stands.

³² Such a statement as that spręc, beginning 164r.1 (33r/M71a:1, line 1398), is a later addition by Thorkelin on the basis of A (TTB, p. 128; the dots printed there are not found in pt 2 or in the transcript; the late correction of v in va to þ should of course be discounted), is disproved by the condition of the quill point as well as the ink and the direction, which show that the whole line must have been written as a single effort. As so often (though not invariably, see e.g. 3r/M5a:13, 8r/M15a:3, 9r/M17a:12, 25r/M49a:5, 36r/M71a:15, 37v/M74a:8), the pen was mended before the new page, and the word was written with the newly mended pen. Other cases cited will similarly not support this claim. The ‘three occasions’ of p. 126Google Scholar are all later or indeed very much later than the original copying. Similarly, selest (25r/M49a:7) is misreported as being on erasure; its miswritten first e was retopped by Thorkelin with a broader pen and in a light ink, both features exactly like the correction of v to w on this page. (The whole word was indeed first written to begin the previous line and not too thoroughly erased there before the correct line was begun over it.) The claim that, when copying B, Thorkelin sometimes wrote line-initial words first, before writing the rest of a page, to facilitate consultation of A, cannot therefore rest on it, nor do we find any series or sequence, however small, of line-initial words written with the quill-point in similar condition and contrasting progressively with the rest of the page, which such a practice would occasion. On the present page the pen was mended after gumena in line 5, but all the words beginning the following lines, including the erased selest, are written with the mended pen. No other evidence is adduced.

³³ One of the prominent weaknesses of the record in TTB's pt 2 is its indiscriminate use of the term ‘later’ for alterations made at once, as shown by e.g. the (letter or word) spacing taking the alteration into account; alterations made clearly in the same ink and with the quill in the same state and therefore presumably still at the time of original writing; and alterations evidently made at some time (mostly long) after the completion of the whole transcript, such as the correction of v to w, made in the same light ink throughout most of the manuscript. The claim (p. 136) that certain errors prove that Thorkelin, while copying, was consciously emending the text ignores all evidence for correction currente calamo, plentiful though it is, by simply classifying it as ‘later’. Thus of the five instances cited there of for written for rof, one (146r.11) was so corrected, a second was never written (145v.20: f written and erased, then rof), vitiating the argument; of the list of seventeen then presented as examples, only the first two, and perhaps the last, were corrected afterwards. For word-spacing evidence see such a case as 138v.2 (7v/M14a:2) to, which has normal placement and word-spacing without the (later) alteration of v to w (TTB, p. 51: ‘to (later, when v changed to w in weder)’Google Scholar). At 191v.14 (60v/M120a:14: ‘man dryhten’) Thorkelin wrote some dots, and continued with dryhten (also with a re-topped e). He then mended his pen, added man, and continued the transcript with mœgenes (it is of course not demonstrable that man was not inserted a very little later, but it has the same ink and the same state of the newly mended quill as mœgenes, and the mending implies a pause). TTB, p. 87Google Scholar reads ‘man (later, with space left before m)’. Similarly 140v.1 (9r/M18a:1) tiges was not written ‘in a different ink from the rest of the line,’ (TTB, p. 126Google Scholar) but the pen was mended after it had been written, the ink remains the same, and the state of the pen is that of 9v/M17a:20. But one's judgement of the ink can apparently also differ: of eight cases cited TTB, p. 139Google Scholar as later additions, only the interlined and of 153v.1 (22v/M44a:1) appears genuine; the others look like their surroundings. The cross-strokes of ð (cf. above, p. 29, n. 12) tell a similar tale. Conversely, the term ‘later’ is frequently wanting where it would clearly have been in place, e.g. 176r.6 (45r/M89a:7) where TTB's ‘him (im x um)’ (i.e. um replaced by im) represents him, with im written above deleted original um in the same light ink that corrected d to ð the next word. Roughly, TTB's ‘later’ appears to extend from twenty years or more after the original inscription to the mere fact that writing is a linear process. A tantalizing instance, apparently with future reference, occurs at 201r.17 (70r/M139a:18), where TTB quotes B as ‘§17 wud[u] (MS? cf.A), rec (r later?)’. When last seen (October 1997), B still read wud … ec (w corrected from v). Cf my TCD, n. 8.

³⁴ He evidently tries to keep the characters grouped as they are in the manuscript, but for one reason or another often departs from this aim. Similar observations can be made about his linkages.

³⁵ TTB p. 142Google Scholar claims: ‘A favorite emendation was changing the Old English spelling burg to the Scottish spelling burgh, another sign that Thorkelin made his transcript after returning from Scotland in 1788’, but this is fanciful. If the Old English form had indeed been burg it could have been argued, but the instances quoted are all of burh, to which (in five out of its twelve instances) he adds the (to him) more usual g. Cf. such cases as leohg for leoht 166v. 1 (35v/M76:1), the weird and wonderful beohrhte originally written in the same line (after which the first h was expuncted in the same ink), or the end of n. 10 above.

³⁶ If in any case a difference should appear between the report in the two transcripts and that in “The State of the Beowulf Manuscript 1882–1983’, pp. 23–42Google Scholar, this would also have to be taken into account, but no such difference appears. The procedure suggested here allows the passing over of a vast number of cases where the present writer's judgement differs from that of TTB. This should not be taken to mean, however, that there are not a fair number of agreements also.