Chapter 7
Editorial introductions

[Skip to Content]


General Navigation

This chapter

[Beginning of Content]


§ 7.1    Cædmon’s Hymn: a multimedia edition and archive is a critical edition and witness archive.[1] As a “critical edition,” the project presents readers with a series of explicitly evaluative representations of the poem at significant (though usually hypothetical) points in its transmission history.[2] As a “witness archive,” the project also provides readers with more lightly mediated access, in the form of transcriptions and digital colour facsimiles, to the basic material from which the critical editions are constructed.[3] In the print volume, these texts are presented in traditional form. On the accompanying CD-ROM, extensive use of cross-linking and other features peculiar to the electronic medium provide readers with a more interactive experience. References to specific language from the poem in the textual apparatus, introductory chapters, and glossary are accompanied by hyperlinks that allow the reader to navigate directly to the corresponding witness transcription. Similar links connect witness transcriptions to facsimiles of the relevant page and copy of the poem. Editions and transcriptions can be displayed in different ways depending on the nature of the information sought. Transcriptions of individual witnesses can be arranged in diplomatic form (i.e. with manuscript lineation, word-division, punctuation, and symbols recording other information about the appearance of the text in the manuscript) or with normalised spacing, and word- and line-division. Critical editions of the poem can be displayed with different types of critical apparatus and, in some cases, before or after emendation or correction.

About the critical editions (Print and CD-ROM)

§ 7.2    The edition side of Cædmon’s Hymn: a multimedia edition and archive contains eight distinct critical texts. There is one critical text for each of the poem’s five main recensions, a critical reconstruction of the poem’s likely Northumbrian archetype, and editions of two “scribal performances” from manuscripts of the West-Saxon eorðan recension (B1 and O).

§ 7.3    In the print volume, these texts are presented with an apparatus showing substantive textual variation (see below, § 7.8); on the CD-ROM, users can alter these texts to display different types of information. The default view, the Reading text (CD-ROM only), is an almost pure “clear text” eclectic edition of the relevant recension or witness.[4] In this view, the focus is on the poem as a literary text. The Old English is presented with modern punctuation, lineation, and word division. Links within the text take the reader to the edition’s glossary.

§ 7.4    The remaining views emphasise the poem’s textual development and are “inclusive” in intent (see Greetham 1994, 368). The Old English is presented with explicit indication of editorial activity and accompanied by a textual apparatus with readings from the relevant witnesses. Links within the text and apparatus take the reader to the individual witnesses from whose testimony the edition is constructed.

§ 7.5    These views are classified according to the level of abstraction used in grouping and displaying readings from the supporting witnesses. The least abstract view is the All witnesses view (CD-ROM only). In this view, the textual apparatus provides a diplomatic transcription of every reading in all relevant witnesses regardless of relative similarity or difference. By reproducing the readings of all relevant witnesses, this view encourages readers to compare editorial forms against those of the underlying tradition; by reproducing these forms diplomatically, the apparatus also invites readers to consider the extent to which features of manuscript layout, such as line boundaries, word-division, and other graphic conventions may have affected the poem’s development. Readings in this view can be displayed in Analytic or Parallel display. In the Parallel Display, witness readings are presented in the same order in every apparatus entry regardless of relative similarity or difference. In the Analytic Display, readings are grouped according to their relative similarity. The Parallel Display view provides readers with what is in essence a parallel text transcription of the Hymn’s manuscript tradition; the Analytic Display view provides readers with an explicit and synoptic account of relationships among witnesses for every reading in the edited text.[5]

§ 7.6    A second slightly more abstract view displays Orthographic variants (CD-ROM only). This view ignores all differences that do not affect spelling (such as differences in spacing, manuscript line division, or variant forms of graphemes). Manuscript readings are displayed using normalised word spacing and division, and forms showing identical spellings are grouped together (e.g. T1 heofon|rıces B1 heofonrıces Ca heofon rıces and Hr hel{o}fon|rıces are grouped together under heofonrices, 1b, in the apparatus to the Orthographic witness variants view of the Northumbrian archteype). Sigla are used to indicate agreement among individual witnesses. Manuscripts not mentioned explicitly in a given apparatus entry show the same spelling as the editorial lemma.

§ 7.7    A third level of abstraction is represented in the Substantive variants view (Print and CD-ROM). This view groups witnesses according to their impact on sense, metre, or syntax. “Impact” in this case is mechanically determined: for the purposes of this edition, substantive variants include all forms that affect sense, metre, or syntax in any way, regardless of origin or intention. Examples thus include contextually appropriate alternatives (e.g. T1 ƿuldor fæder : B1 ƿuldor godes|; M tıl : P to; T1 sceop : O gesceop|; T1 Nu sculon herıgean : B1 Nu ƿe herıgan sculon|), contextually inappropriate alternatives and nonsense forms (e.g. H ord : W ƿord : To ær|; H frea : Tr1 euca), alternative spellings that can be construed as affecting metre (e.g. M astelıdæ : Di astalde), and minor and obvious scribal errors (e.g. H ƿe : Di pue). Orthographic variants that do not affect sense, metre, or syntax are silently grouped together under a single headword (e.g. H astealde· and Di astalde are grouped together under astalde in the Substantive witness variants view of the Northumbrian archteype). Sigla are used to indicate substantive agreement among individual witnesses. Manuscripts not mentioned explicitly in a given apparatus entry are substantively identical to the editorial lemma.

§ 7.8    The fourth level of abstraction is represented in the (Potentially) significant variants view (CD-ROM only). In this view, apparatus entries include only those forms that might be understood as involving a change in metrical, lexical or syntactic “significance” from the editorial lemma: i.e. variation involving the substitution of one lexical form, metrical pattern,[6] or syntactic construction for another, or the irreversible destruction of sense, metre, or syntax. Such substitutions include variation between contextually appropriate alternatives (e.g. T1 ƿuldor fæder : B1 ƿuldor godes|; M tıl : P to; T1 sceop : O gesceop|; T1 Nu sculon herıgean : B1 Nu ƿe herıgan sculon|) and contextually inappropriate alternatives and nonsense forms that cannot easily be restored to the archetypal form (e.g. H ord : W ƿord : To ær|; H frea : Tr1 euca). They exclude, however, minor and obvious errors or alternative spellings of what can reasonably be considered as variant forms of the lemma, even if these have a minor effect on sense or metre (e.g. M astelıdæ : Di astalde; H ƿe : Di pue). Variants that are not (potentially) significant are grouped together silently under a single headword (e.g. M astelıdæ : Di astalde are grouped together in the Significant witness variants view of the apparatus to the Northumbrian archteype). Sigla are used to indicate agreement among individual witnesses. Manuscripts not mentioned explicitly in a given apparatus entry are not significantly different from the editorial lemma.

§ 7.9    In classifying variants as Orthographic, Substantive, or (Potentially) significant, I have applied the above criteria more rigorously in the case of early witnesses than late, and in the case of Anglo-Saxon scribes than continental. Thus M scepen· (for expected sceppend) is treated as potentially significant, even though it can be explained relatively easily as a scribal error, because it could be understood as involving a change in meaning (see below, § 7.24) and because the variant is in the hand of an eighth-century scribe (Note A: The dating of M and P). SanM æ for expected ond, on the other hand, is treated as substantive rather than (potentially) significant, because the error involves a fairly obvious misconstruction of an <e+t> ligature as <æ> and because it is in the hand of a fifteenth-century scribe prone to making such mistakes. The Northumbrian eordu recension, finally, represents a distinct case. All three witnesses to this recension are in continental hands, and, as I argue below (§§ 7.29-7.33), are probably derived from the same insular exemplar. No scribe shows any evidence to suggest that he understood what he was copying. In the apparatus associated with the edition of this recension, therefore, variants are classified as either “accidental” or “substantive” on the basis of their significance for the filiation of the surviving witnesses.

About the transcriptions (Print and CD-ROM)

§ 7.10    The archive side of Cædmon’s Hymn: a multimedia edition and archive contains transcriptions (print and CD-ROM) and where possible full colour facsimiles (CD-ROM only) of all known medieval witnesses to Cædmon’s Hymn. In the print text, transcriptions are diplomatic: the arrangement of the text reflects manuscript spacing, word- and line-division, and all corrections, deletions, gaps, and unclear passages are indicated explicitly. The accompanying CD-ROM also allows readers to display the texts of the witnesses in semi-diplomatic form (i.e. with editorial word- and line-division), and, in the case of O, before and after scribal correction.

About the facsimiles (CD-ROM only)

§ 7.11    Facsimiles of the surviving manuscripts are in colour on the CD-ROM wherever this has proved feasible. To, which was destroyed in an air raid during the Second World War, is represented by a digital facsimile of its sole surviving black and white photograph; C, which was badly burnt in the Ashburnham house fire of 1731, is not represented, although a photo of its sixteenth-century copy N is. For conservation reasons, Di and P1 are represented by scans of black and white photos. Because of the large number of manuscripts and manuscript libraries involved, the facsimiles have been made using various techniques: most have been scanned from photographic slides or prints; others have been scanned directly using equipment available on site. Except in the case of To, which only survives in a partial facsimile, users can choose between views of the Hymn as a detail or in the context of the entire page on which it is found.

§ 7.12    The remainder of this chapter is devoted to discussing the specific editorial decisions made in producing the project’s critical texts. Each section begins with a brief survey of the relevant manuscripts, examines the filiation and transmission of the text in question, discusses specific textual problems, and provides a justification for the editorial approach used in constructing the relevant critical text.

Introduction to the critical texts

The reconstructed Northumbrian archteype

Manuscripts and filiation

§ 7.13    Cædmon’s Hymn is known from five recensions and twenty-one medieval manuscripts. Descriptions of individual manuscripts can be found above, Chapter 4: Manuscripts. A detailed account of the filiation and transmission of these witnesses can be found in Chapter 5: Filiation and transmission.

Editorial approach

§ 7.14    This edition attempts to reconstruct the original text of Cædmon’s Hymn as it is implied by the evidence of the surviving manuscripts. This text may or may not be the same as that sung by Cædmon in response to his dream visitor (assuming Bede’s version of the story is at least broadly correct in its discussion of the circumstances under which the Whitby herdsman became a poet, see Chapter 1: Bede and Cædmon); it should, however, correspond in its substantive details with what seems to have been the single written archetype that ultimately underlies all known copies of the Hymn (see Chapter 5: Filiation and transmission, above; cf. Cavill 2000, esp. 525-530).[7] It is probably impossible to recover the exact spellings used in this archetype with certainty; for reasons discussed below (§ 7.26), accidental details of orthography are for the most part based on M and, when the relevant form is not found in that manuscript, the reconstructed form of the archetype of the Northumbrian eordu recension.

§ 7.15    As far as I am aware, this is the first modern edition of Cædmon’s Hymn to be based on a critical evaluation of the entire surviving manuscript tradition. Although the Hymn’s filiation and textual transmission has been studied frequently since the discovery of the Northumbrian eordu recension by Wuest in 1906, no scholar appears to have attempted the critical reconstruction of the archetypal Hymn that much of this research implies. Most scholarly and student-oriented editions of the poem are “best-text” or single-witness (e.g. Pope and Fulk 2001; Mitchell and Robinson 2001); a few other editions, most notably Wuest 1906, Smith 1978, and O’Donnell 1996b, provide critical texts of individual recensions. Even Dobbie, who undertook his seminal investigation into the Hymn’s transmission as a prolegomenon to his edition of the poem in the Anglo-Saxon Poetic Records (Dobbie 1942) and whose work on the text is among the most cited of modern discussions, appears to have ignored several of his own most significant conclusions about the poem in his subsequent edition. The decision to group witnesses of the poem by dialect, for example, contradicts Dobbie’s earlier conclusion that dialect was an insignificant factor in the poem’s textual development[8] and causes him to group together manuscripts that he had previously argued came from different textual branches.[9]

Conventions and apparatus

§ 7.16    The reconstructed text of the archetypal Cædmon’s Hymn in this edition follows the conventions outlined above, §§ 7.2-7.9. Unlike editions of the five main recensions, however, the archetypal text comes on the CD-ROM with two sets of textual apparatus (the print text uses the default apparatus). The default (recensional) apparatus collates the text of the Hymn against that of the five main recensions. This view emphasises the textual development of the Hymn as a whole while ignoring minor, recension-internal differences among witnesses to any given version of the poem; it corresponds in this regard largely with the approach to the Hymn’s textual development taken in Chapter 5: Filiation and transmission, above. The alternative (witness) apparatus compares the reconstructed text against transcriptions of all known witnesses without regard for their recensional affiliation. This view emphasises local agreement and disagreement among the Hymn’s many witnesses over the poem’s recensional development. It should be of interest to readers who are interested in tracing the history and popularity of specific forms, who are looking for information about the copying habits of individual scribes, or who wish to test the editor’s conclusions concerning the most likely reconstruction. Both apparatus can be further altered on the CD-ROM to display the four standard levels of abstraction discussed above: an all witnesses view, to display readings from all relevant witnesses or recensions; an accidental variants view, to show non-orthographic differences; a substantive variants view, to reveal variation that affects sense, syntax, or metre; and a significant variants view, to show variants that have a (potentially) significant effect on sense, metre, or syntax. The parallel all witness view, finally, can be used to construct a parallel text edition of all surviving witnesses or recensions of the poem.

Northumbrian aelda recension


§ 7.17    The Northumbrian aelda recension is known from two manuscripts usually dated to the early- to mid- eighth centuries:

The vernacular Hymn appears as a gloss to the m-text of the Historia ecclesiastica in both manuscripts. In P the Old English text has been copied at the foot of the page containing Bede’s Latin paraphrase; in M, it appears at the top of the last page in the manuscript among other glosses and notes.


Latin text

§ 7.18    M and P contain the two earliest known copies of the Historia ecclesiastica and two of only three known copies of the m-text of Bede’s work produced in English scriptoria (Colgrave and Mynors 1969, xlvi-xlvii). One other eighth-century English manuscript of the m-text is known, the partially destroyed London, British Library, Cotton Tiberius A. xiv (Tib. A. xiv). A fourth, now lost, copy can be deduced from the evidence of three ninth-century German and French manuscripts: Wolfenbüttel, Herzog-August Bibliothek, Weissenburg 34 (Wbl), Würzburg, Universitätsbibliothek, M. p. th. f. 118 (Wbg), and Namur, Public Library, Fonds de la ville II (Namur) and a fifth perhaps from the evidence of a dozen twelfth-century continental manuscripts (Colgrave and Mynors 1969, xliv-xlv and lxviii-lxix). All other English copies of the Historia contain the c-text. Among the early m-type recension manuscripts, only M and P contain a copy of the Old English Hymn.

§ 7.19    The precise relationship of these early English manuscripts of the Historia ecclesiastica to each other and their common archetype is unknown. P shows a particularly close affinity in several details to Tib. A. xiv, suggesting either direct filial descent or a close fraternal relationship (see Fulk 1992, 426). M, for its part, is close to P and Tib. A. xiv in substance, but shows numerous minor differences usually attributed to the haste with which the scribe appears to have worked. In the textual introduction to the EEMF facsimile of the manuscript (Hunter Blair 1959), Mynors concludes that M and P are independent copies of “the same very early and authentic parent text” (33; also Colgrave and Mynors 1969, xxxix-xl and xliv). Several scholars, more controversially, have suggested that P may be in part in Bede’s own hand (see below, § A.6).

§ 7.20    Perhaps the best we can say with confidence is that all three copies appear to belong to the earliest dissemination of Bede’s work. P and Tib. A. xiv belong to a single early branch of this common tradition; M, with its many small differences from P and Tib. A. xiv, either belongs to a second early branch of the Historia’s first dissemination or shows the effects of a hasty but competent performance on the text represented by P and Tib. A. xiv.

Old English text

§ 7.21    As is true of the host manuscripts’ Latin text, the witnesses to the Old English poem in M and P appear to be fraternally rather than filially related (as noted above, § 7.18, neither Tib. A. xiv nor the exemplars implied by Wbl, Wbg, Namur, or the twelfth-century continental texts appear to have contained a copy of the vernacular Hymn). The two witnesses differ thoroughly in minor details of orthography and spelling but show only four substantive variants.

§ 7.22    Although the Hymn appears to have been copied by a scribe of the main Latin text of the Historia ecclesiastica in both M and P, this does not necessarily imply that the scribes responsible found their texts of the Hymn and the Historia in the same exemplar. This is particularly true in the case of M where the Hymn has been copied alongside “memoranda” and other glosses on the last page of the manuscript. As mentioned below, § A.3, this location suggests that the scribe may have collected this material from the margins of a different exemplar. As Anderson has demonstrated, moreover, the orthography of the Old English Hymn in M differs in several regards from the scribe’s usual practice in recording Old English place and personal names in the main text of the Historia ecclesiastica, again suggesting that the vernacular poem and main Latin text of the Historia may have come from different exemplars (Anderson 1941, 143 [§ 7.iii] and 144 [§ 2], 145).

Editorial approach

§ 7.23    The principal problem facing the editor of the Northumbrian aelda recension of Cædmon’s Hymn is that of deciding which version of the text to print. The two witnesses are at the same time too different to be understood as mere copies of a common exemplar and too similar to warrant a parallel text edition. With neither version having an obvious claim to priority, an editor has little textual reason for preferring the readings of one version above those of the other.

§ 7.24    The variants themselves do not help decide the issue. Each manuscript shows one relatively certain mistake. P aeldu for expected aelda, 5b, shows what is probably a confusion of <u> and open-headed <a> in the inflectional ending. Similar errors are found throughout the main text of the Latin Historia ecclesiastica in P, though they are relatively less common in the stint of Scribe D (the scribe responsible for copying Cædmon’s Hymn) than in those of the other three main-text scribes (Parkes 1982, 7). M scepen·, 6b, shows what is probably a mistake for sceppend with assimilation of the final -/nd/ and graphic simplification of the medial -<pp>- (O’Donnell 1996a, 53-54). A partial parallel can be found in Psalm 50, 46, sceppen, and Beowulf, 106a, where scyppen has been corrected to scyppend with the addition of a final <d> above the line. The other possibility, that the M reading is the only recorded survival of an Old English cognate for WGmc skapinaz (cf. OHG scaffin, sceffin, “judge,” OFris skeppena, “jury man”) seems unlikely on metrical and semantic grounds (see Dobbie 1937, 13-15).

§ 7.25    M and P do differ in two sensible readings that are also syntactically and metrically appropriate: P and : M end, 2b, and P to : M til, 6a. Here too, however, there is little reason to prefer one form over the other. In the case of and : end, it is probably impossible to determine priority: the two readings are exact synonyms and cognates (see above, § 6.14): either could be the archetypal form. In that of to : til, neither reading is obviously more or less likely to represent an innovation. On the one hand, the P reading to is common to all other copies of Cædmon’s Hymn and more common in the Old English corpus at large—it is easy to see how a scribe, faced with the statistically rare form til, might consciously or unconsciously substitute the more common to; on the other hand, however, til is a Northumbrian word and M and P are both Northumbrian manuscripts: it is also easy to see how a Northumbrian scribe working in haste might have substituted a familiar dialectal form for its exact synonym.

§ 7.26    All previous editions of the Northumbrian aelda recension have taken M as their base (Pope and Fulk 2001; Smith 1978; Dobbie 1942).[11] In doing so, these editors appear to have been swayed by two main criteria: a now untenable belief, based on the so-called “Moore Memoranda,” that M was certainly the older of the two manuscripts (see Note A: The dating of M and P, below), and the observation that M contains a slightly greater preponderance of rare dialectal spellings in its copy of the Hymn (see Anderson 1941, 143-145). While contemporary scholarship is now less certain that M represents the earlier manuscript, the linguistic interest of the forms in this manuscript remains a legitimate ground for preferring the M text. This edition, like its predecessors, therefore bases its editorial text on M with corrections and collations from P.

Conventions and apparatus

§ 7.27    The editorial text of the Northumbrian aelda recension follows the conventions outlined above, §§ 7.5-7.9. M and P show only two potentially significant variants, of which only the substitution to : til is likely to involve truly appropriate alternatives. The second potentially significant variant, M scepen for expected sceppend is included only because it has a potential effect on sense, syntax and metre. The other two substantive variants, and : end, 2b, and aeldu : aelda, 5b, are not treated as significant in this edition. In the case of and : end this is because the two forms are such close cognates; in that of aeldu : aelda, it is because of the frequency with which <a> and <u> are confused in P.

Northumbrian eordu recension


§ 7.28    The Northumbrian eordu recension is found in three continental manuscripts of the late twelfth, and early and late fifteenth centuries:

All three manuscripts contain copies of the Latin Historia ecclesiastica. The Old English text of the Hymn has been copied within the main text of the Historia in each manuscript immediately preceding the beginning of Bede’s paraphrase in IV. 24 (i.e. between iste est sensus and Nunc debemus in IV. 24). As this location suggests, the poem is in the hand of the main Latin text in all three witnesses.


Latin text

§ 7.29    The precise filiation of the Historia ecclesiastica in these manuscripts is unknown, although it seems likely that all three belong to the same textual tradition. Colgrave and Mynors, who do not appear to have been aware of the presence of the Old English Hymn in any of these manuscripts, assigned them to different branches of the Historia’s transmission. Br and P1, for which they collated selected passages of the Latin text, they assigned to the “German” m-type recension; Di, which they did not examine, they tentatively assigned to the “English” c-type text, presumably on the basis of the strong English focus of its contents (Colgrave and Mynors 1969, lxv-lxvi, lxi).

§ 7.30    This last filiation, however, is almost certainly wrong (O’Donnell 1996b, 151-159). Even leaving aside the similarities in the location and text of the vernacular Hymn in all three manuscripts,[12] Di shows a number of features that suggest that it is closer to P1 and Br than members of the “English” tradition. The manuscript is written in a mixture of English and continental hands, has a first person plural reference to the monastery at Cîteaux, and includes a version of Bede’s prose Vita Cuthberti that is otherwise found only in continental, indeed, primarily Cistercian, manuscripts (see Colgrave 1940, 49, where the manuscript is assigned the siglum Dj1).[13] As Wuest first pointed out, moreover, Di shares at least one peculiarity with the Latin text of P1 and Br: the omission of miraculorum from its proper place between omnium and auctor in Bede’s paraphrase, quomodo ille... omnium miraculorum auctor extitit ... (Wuest 1906, 218-219). In Di, the word is omitted entirely; in P1 and Br, the form has been incorrectly inserted between primo and filiis. Although given the necessarily selective nature of the collations used by editors of the Historia ecclesiastica, it is impossible to say how many other copies of the “German” recension contained this error, the mistake does not appear to go back to the earliest exemplars—neither Colgrave and Mynors nor Plummer (who used a full collation of Namur and partial collations of other members of the “German” family in establishing his text) mention the variant in their critical apparatus (ed. Colgrave and Mynors 1969, 416, and ed. Plummer [1896] 1969, 1: 259-260).

Old English text

§ 7.31    The text of the Old English Hymn reflects the evidence of the Latin paraphrase. On the one hand, the three copies of the vernacular text all agree in the three key recensional readings that characterise the Northumbrian eordu recension: ƿue (Br P1 puc Di pue), 1a; eordu (correctly reported in all three manuscripts), 5b; and firum on foldu (Br P1 firum ol foldu), 9a. On the other hand, the three witnesses show enough independent variation to suggest that none is a direct copy of any other. The earliest manuscript, Di, has several errors which are not reflected in the later (and generally more corrupt) P1 and Br: modgedeanc| for expected modgedanc (as in P1), 2b; puldur fudur for expected ƿuldurfadur (cf. P1 Br fadur), 3a; drıntın for expected drichtin (as in Br), 8a; c ef|ter for aefter (as in P1 and Br), 8b. Br, likewise, although the latest and the most corrupt of the three surviving manuscripts, has several forms which are closer to the expected Old English forms than those found in either Di or P1: aerıst vs. P1 aeırst and Di u\a/erst, 5a; he |fen vs. P1 Di efen, 6a; drichtin| vs. P1 drıchtıııı Di drıntın, 8a. Unique readings in P1 include aeırst (Br aerıst Di u\a/erst), 5a and her (Br Di he), 5a (O’Donnell 1996b, 146-150).

§ 7.32    Taken together, the evidence of the Latin and Old English texts in these manuscripts suggests that the two fifteenth-century manuscripts Br and P1 are closer to each other than to the twelfth-century Di. The two witnesses agree against Di in seven readings, and share another five in which their respective variants are more similar to each other than to the reading of the earlier manuscript. In many cases, the shared Br and P1 reading involves an obvious, and possibly coincidental, example of the confusion of insular letter forms, e.g. <e> and <c>: Br P1 puc for ƿue (Di pue), 1a; Br P1 puere for ƿuerc (Di puerc), 3a; Br P1 moneınnes for moncinnes (as in Di), 7b; Br P1 eei for eci (as in Di), 8a; <f> and <s>: Br hesım|ruica es P1 hesıııȷ rıııca es for hefunricaes (as in Di), 1b; or mistakes in the interpretation of minims: Br hesım|ruica es P1 hesıııȷ rıııca es for hefunricaes (as in Di), 1b; Br metundaes P1 ıııetııııdaes for metudaes (as in Di), 2a. In a few cases, however, Br and P1 show mistakes that seem more idiosyncratic and less likely to be the result of coincidental error. These include the omission of expected ƿuldur- from Br P1 fadur (Di puldur fudur), 3a; the confusion of <p> and <b>, and <l> and <n> in Br peannum P1 pearıııım (Di bearnum), 5b and Br P1 ol (Di on), 9a; and finally, probably, the mark over <a> in Br da̽ P1 | (Di da·), 7a (see below §§ 7.37-7.40).

§ 7.33    These similarities and differences suggest the following stemma for the transmission of the Hymn, and to the extent that the anomalies in the placement of miraculorum and the insertion of the Old English poem in the main Latin text are evidence of a common exemplar, the Latin Historia ecclesiastica:

Transmission of Northumbrian eordu recension.

Image showing the likely transmission of Northumbrian eordu recension.

Since none of the variant readings that divide witnesses to the Northumbrian eordu recension involves sensible, metrically acceptable, and syntactically appropriate alternatives, it seems likely that the text of the Hymn in all three manuscripts goes back to a single copy, imported to the continent sometime before the twelfth century (O’Donnell 1996b, 155-159). This copy may or may not have been found in the same insular manuscript as that responsible for the Latin text of the Historia ecclesiastica in all three manuscripts. The vernacular Hymn often travels independently of Bede’s Latin, and, as the use of a copy of the West-Saxon eorðan recension to gloss the Latin Historia in the perhaps continental To demonstrates, scribes could freely borrow texts of the poem from manuscripts containing other recensions of the Historia ecclesiastica. At the same time, however, the poor knowledge of Old English shown by these scribes suggests that none were responsible for adding the vernacular poem to the main text of the Historia ecclesiastica in this tradition.

Editorial approach

§ 7.34    The Northumbrian eordu recension of Cædmon’s Hymn is the most frequently edited text of the poem. It is, indeed, the only version to have been the subject of more than one critical edition of the manuscript tradition (O’Donnell 1996b; Dobbie 1937, 21; Wuest 1906). In each case, the editorial approach has been explicitly reconstructive. This edition follows the lead of these earlier editions in that its focus remains the reconstruction of the Northumbrian eordu recension’s archetypal insular text (scholars interested in the recension as a case study in the transmission of Old English texts by continental scribes or as raw evidence for linguistic study of Northumbrian will almost certainly prefer to work directly with the facsimiles and diplomatic transcriptions collected in the witness archive). The main difference between this edition and previous critical texts lies in the amount of evidence available. With the identification of Br as a third witness to the Northumbrian eordu recension (O’Donnell 1996b), scholars are now in a better position to assess the evidence of the two better known witnesses, P1 and Di.

§ 7.35    In most cases, discovering the underlying form is a relatively straightforward task. Most of the mistakes the scribes of the surviving witnesses make have their roots in obvious misinterpretations of individual letters or minims. There remain, however, five forms in which recovering the reading of the archetypal insular exemplar is more problematic:

While the substantive readings these forms reflect is beyond doubt in each case, the precise spellings are somewhat more open to debate. Since in a few cases one or more of the possible solutions represents our principal evidence for the existence of certain dialectal or orthographic forms in Old English, it is worthwhile considering their reconstruction in some detail.


§ 7.36    Wuest reconstructed this word as scuilun in his editio princeps, which was based on Di and P1. Smith and Dobbie for their parts both propose sciulun, an intermediate form showing the influence of initial palatals on /u/ (see Hogg 1992, § 5.68). That this second suggestion is correct is indicated by the spelling in Br. Brother Anthony, the scribe of Br, generally dots his <i>’s (cf. Cavill 2000, 512); his reading for this word, scinlun with <n> for expected <u> but a dotted <i> for the first minim suggests that the archetype read <iu>.

modgedanc, eordu, and ða

§ 7.37    The principal problem with these three words involves the representation of the dental spirants. This is indicated using the characters <ð> and <þ> in later Old English, and by <d> in earlier manuscripts (see Hogg 1992, §§ 2.59, n. 1 and 2.77), including M and P. In the case of the Northumbrian eordu recension, modern editors have differed significantly in their reconstruction of the underlying forms: Wuest assumes <ð> in the case of modgedanc and da, but <d> in the case of eordu (Wuest 1906, 219); Dobbie reconstructs <ð> in all three forms (Dobbie 1937, 21, also n. 27); Smith reads <d> in the three words (see Smith 1978, 38-42 [textual apparatus]).

§ 7.38    The manuscript forms for these words are clear enough: the three surviving witnesses read <d> in each case. That the underlying Insular manuscript of this version of the poem may have had a different character, however, is suggested by two features of the surviving copies: the reading Di modgedeanc| for expected modgedanc, 2b, and marks above the <a> in Br and P1 (Br da̽ P1 |, 7a). In the case of Di modgedeanc|, Wuest suggests that the extra <e> in -deanc may point back to the original <ð> of an early exemplar (216):

[Di] z. 20 weiß ich für das e in -deanc keine sprachliche deutung. dass die vorlage dem schreiber unklar war, beweist die rasur. Y [the presumed ancestor of Di and P1] bot hier wol vereinzeltes ð: der schreiber hield das durchstrichene d vermutlich für die ligatur d + e und gelangte so zu dea-. der schreiber von [P1] hingegen übersieht den strich im ð und schreibt richtiger modgedanc z. 3-4.

In the case of ða, in 7a, the possibility of an original in <ð> is suggested by the marks found above the <a> in P1 and Br. As Wuest, who did not know of Br, notes of P1 (217-218):

Dass dem da ([Di] z. 23) in [P1] z. 5 entspricht, lässt auf ein ða in Y (vielleicht aber auch blos auf einen über dem a stehnden dehnungsstrich) schließen.

With the identification of Br as a third witness to the recension, Wuest’s suggestion acquires additional weight. In this manuscript, the equivalent form in Br has a small cross-like smudge above the vowel. Taken with the macron in P1, this suggests that at least one common ancestor of these two manuscripts had some kind of elevated mark above the <a>. That the mark was not a suspension or accent mark (indicating, for example, an original in dam) is suggested by the form in Br. Suspension marks are the most common mark of abbreviation in Latin manuscripts: a continental scribe faced with an unknown form ending in <ā> or <á> would almost certainly understand the accent as a Latin suspension. If the original version of the poem had an insular <ð> with rounded back and a prominent cross stroke, however, the same scribe might well understand the top of this character as looking something like a cross (Br) or a badly formed suspension mark (P1; see O’Donnell 1996b, 157-159; Cavill 2000, 513).

§ 7.39    Wuest’s suggestion concerning the significance of the additional <e> in Di modgedeanc| is equally possible, but by no means conclusive. If the form does point back to an original in <ð>, however, then it seems more likely that the continental scribe responsible for the current form has misunderstood the cross stroke of the original <ð> as an <e+a> ligature rather than the <d+e> combination Wuest suggests. At the same time, the scribes of all three surviving witnesses (or their exemplars) show themselves remarkably willing to add extra letters to the original text. Examples include P1 Br ıııetııııdaes (expected metudaes), 2a; P1 her (expected He), 5a; and Di u\a/erst (expected aerist), 5a.

§ 7.40    Dobbie’s reading eorðu for Di P1 Br eordu, 5b, is a normalisation (see Dobbie 1937, 21, n. 27). Nothing in the surviving witnesses suggests that the underlying text read eorðu.


§ 7.41    The final editorial problem in the poem involves the underlying form for hefen (Br he |fen P1 Di efen), 6a. Before the identification of Br, the common reading in Di and P1 was assumed to be evidence of the sporadic disappearance of initial /h/ before vowels. As Br he |fen makes clear, however, the original text of this recension read hefen. The initial <h> appears to have been dropped independently in both P1 and Di.[14]

Conventions and apparatus

§ 7.42    The critical text and apparatus of the Northumbrian eordu recension follow the conventions outlined above, §§ 7.5-7.9. The only exception to the usual practice in this project lies in the treatment of significant, substantive, and accidental variants. Since none of the scribes of the surviving witnesses appear to have understood Old English and since all three were working in a tradition apparently derived from the same archetype, the variants they show are in one sense all “accidental”—none of the variants affect metre, sense, or syntax as the scribes understood it. At the same time, however, some of the textual variants are more important than others in helping to work out the textual relationships among the witnesses. In order to help the reader sort this information, a two-part division has been used in the apparatus to this edition[15]: differences that appear either to support or contradict the proposed reconstruction are treated as “substantive variants.” These include forms such as P1 Br fadur Di puldur fudur, 3a, but also the mark above the <a> in P1 | and Br da̽, 7a. Variants that seem more likely to be the result of incidental or coincidental variation are treated as accidentals. Minim errors are treated as accidental variants if the variant forms show the same number of minims; variants showing different amounts of minims, or forms in which scribes have clearly interpreted the minim strokes as different letters are treated as substantive variants. As in all editions in Cædmon’s Hymn: a multimedia edition and archive, two final views present a straight collation of all readings from all witnesses.

West-Saxon ylda recension


§ 7.43    The West-Saxon ylda recension of Cædmon’s Hymn is known from seven manuscript copies ranging in date from the beginning of the eleventh to the mid-fifteenth centuries:

All seven manuscripts also contain copies of the Latin Historia ecclesiastica. The vernacular text of the Hymn is found alongside Bede’s Latin paraphrase of the poem in IV. 24 in each. In six witnesses (Bd H Ln Mg SanM W), the Old English poem has been copied in the margins of the page on which the paraphrase is found; in a seventh (Tr1), it has been inserted into the main Latin text immediately before the translation (i.e. between iste est sensus and Nunc debemus in IV. 24).

§ 7.44    Most witnesses to the West-Saxon ylda recension have been copied by scribes other than those responsible for the main Latin text of the page upon which the poem is found. In three witnesses—the mid-twelfth-century Mg, fourteenth-century Tr1, and mid-fifteenth-century SanM—the Hymn appears to be the work of the main scribe of the manuscript in question.


Latin text

§ 7.45    It is impossible to reconstruct a reliable filiation for the West-Saxon ylda recension of the Hymn. The fact that most witnesses to this version of the poem have been copied by scribes other than those responsible for the main Latin text of the manuscripts in which they appear suggests that this version of the Old English text was not transmitted as an integral part of the Historia ecclesiastica in its earliest manuscripts. That this was so is confirmed by the main Latin text of the manuscripts in which the poems are found: while all seven witnesses to the West-Saxon ylda recension transmit a version of the “English” or c-text version of the Latin Historia, the individual manuscripts belong to several different branches of this large family: one witness, H, reflects the earliest known version of the c-text (Colgrave and Mynors 1969, xlii-xliii and xlvi-xlviii); two other witnesses, Bd and W, are drawn from the so-called “Winchester” family (Colgrave and Mynors 1969, l-li); Ln belongs to the defective “Burney” or “Yorkshire” tradition (Colgrave and Mynors 1969, liii-lv); and Mg SanM Tr1 all contain copies of the innovative “Digby” or Southern common text (Colgrave and Mynors 1969, lv-lvii). Not all manuscripts in these various traditions of the Historia ecclesiastica contain a copy of the Old English poem; not all that do contain the West-Saxon ylda recension.[17]

Old English text

§ 7.46    Even if it were possible to construct a filiation based on the evidence of the Latin text in the witnesses to this recension, it is doubtful that this filiation would tell us very much about the transmission of the poem. This is because the West-Saxon ylda recension has been transmitted with a remarkably high degree of substantive accuracy. In the course of four centuries, the scribes responsible for copying this version of the poem introduce only one potentially sensible textual variant: the substitution of the marginally sensible and metrically problematic ƿord in the mid-eleventh-century W for ord, 4b, in all other witness. All other variants in the poem are either nonsensical or the result of obvious error or accidental differences in orthography. These variants are distributed apparently at random, moreover: every copy of the text shows at least one unique reading and no pre-fourteenth-century manuscripts appear to share diverging forms on any regular basis. While a more regular pattern can be seen in the innovations found in the recension’s fourteenth- and fifteenth-century witnesses, Tr1 and SanM, the common variants here are also better understood as coincidental variation rather than evidence of a close genetic relationship: both witnesses contain mistakes not found in the other text and the variants they share, like the variants in which they differ, tend to belong to the most common sort of errors introduced by scribes of late manuscripts: confusion of <ƿ> and <p>, of <t> and <c>, and, in Tr1 in particular, <f> and <p>.

§ 7.47    The substantive accuracy with which the West-Saxon ylda recension has been transmitted is all the more remarkable considering the state of its text. As Dobbie has pointed out, this version of the poem suffers from a number of serious grammatical, syntactical, and lexical difficulties: an accusative gehƿilc, “each,” for expected genitive gihuaes (West-Saxon gehƿæs), “of each,” 3b[18]; the omission of a metrically necessary þa, “then, when,” and substitution of a dative or instrumental middangearde, “to/with the earth,” for expected accusative middangeard, “earth,” 7a; the substitution of an apparently genitive plural noun tida, “of times,” for an expected and syntactically necessary third person singular preterite verb teode, “appoint, establish,” 8b, and the use of the dative plural foldum, “to the earths,” for an expected genitive or accusative singular foldan, “(of) earth,” 9a (Dobbie 1937, 39-40).

§ 7.48    Of these, the most serious corruption is the use of tida in place of the expected verb. In most versions of the Old English poem, 7 through 9 form a separate clause:

5    He aerist scop      eordu barnum

    heben til hrofe,      haleg sceppend;

    tha middungeard,      moncynnæs uard,

    eci dryctin,      æfter tiadæ

    firum foldu,      frea allmectig.

While it is possible in the absence of teode to understand 7 through 8a as ornamental variation of material in the preceding clause, the sense of the poem breaks down entirely in 8b—there is, as far as I am aware, no way of forming a sensible clause from the material in the last three half-lines of this recension. Interestingly, this fact appears to have been recognised by the scribes themselves: in the three twelfth-century and earlier copies of the West-Saxon ylda recension for which punctuation at this point can be recovered, scribes isolate the final lines of the poem by placing points before the missing verb in 8b: either at the end of the preceding clause (6a), or following the “ornamental variation” in 6b or 8a (see O’Donnell 1996a, 25 and 27-30; cf. O’Keeffe 1990, 44-46 and fig. 3):

5    He ærest gescop      ylda bearnum

    heofon to hrofe.(Mg)      halig scyppend.Ln

     middangearde,      mancynnes ƿeard,

    ece drihten.H      æfter tida

    firum on foldum,      frea ælmihtig

§ 7.49    As Dobbie suggests, most of the errors in this recension can be attributed to the apparent misapprehension of Northern spellings by the poem’s original West-Saxon translator. The reading tida, “of times,” for expected teode, “adorned,” has been explained by Dobbie as the result of a misunderstanding of the medial -<ia>- and final -<æ> of an original tiadæ as -<i>- and -<a> respectively (Dobbie 1937, 39-40). A similar mistake may lie behind the reading in 9a in all witnesses to the West-Saxon ylda recension, foldum for expected foldan. As Orton 1998 has suggested, this reading is probably best explained as a misunderstanding of a Northumbrian weak accusative singular in -<a> or -<u> (156), although dittography of the dative plural ending on firum cannot be ruled out. The reading middangearde (for expected þa middangeard), in 7a, finally, appears to represent an attempt at limiting the syntactical damage caused by the substitution of tida for teode in 8b: as O’Donnell 1996a suggests, the change in case (from accusative to dative) and the metrically necessary omission of the adverb ða allows 7a to be construed as part of the preceding clause: in the West-Saxon ylda recension, middangearde is a dative of advantage or reference approximately parallel to ylda bearnum in 5b: “He, the holy creator, first made heaven as a roof for the children of men, for the earth...” (21-32, esp. 29-30; trans. Orton 1998, 156, where a similar point is made). In all other versions of the text, middungeard is an accusative singular object for tiadæ in 8b, “then the guardian of mankind, the eternal lord, the holy architect, the almighty ruler, appointed the middle earth, the nations, for men of earth” (trans. based on the readings of the West-Saxon eorðan recension; the Northumbrian eordu and West-Saxon eorðe recensions show other differences unrelated to the variant in question).

§ 7.50    The only unique form common to all manuscripts of the West-Saxon ylda recension that cannot be explained easily as the result of an original error in the interpretation of Northern spellings when the recension was first translated into West-Saxon is the substitution gehƿilc for expected gehƿæs in 3b. It is possible that the form is a performance variant—gehƿilc is syntactically appropriate, does not damage metre, and, understood as an appositive variation on ord, makes reasonable sense in context: instead of a genitive governed by ord in 5b as in most other versions of the poem (“[he] appointed the beginning of each of wonders”), the presumably accusative gehƿilc must instead be read as being syntactically parallel to ord (“[he] appointed the beginning, each of wonders”). On the other hand, however, the collocation ƿundra gehƿilc does not appear to be formulaic and may have its origins in a graphic mistake: in insular script, the two forms show a slight resemblance to each other. If the first part of the <æ> were slightly open, the loop of the second part of the digraph slightly higher and closer to the <s> than normal, a careless scribe could perhaps mistake <æs> for a minim followed by <l> and <c>.

Editorial approach

§ 7.51    The main problem facing an editor of the West-Saxon ylda recension of Cædmon’s Hymn lies in the recension’s serious and obvious errors. On the one hand, few of this recension’s innovative forms can be regarded reasonably as anything but mistakes: in all four places in which all witnesses to the West-Saxon ylda recension preserve a unique form found in no other version of the text, the ylda reading is clearly inferior and can be explained by an appeal to relatively straightforward translation errors or the misapprehension of specific letters and forms. Even the sole potentially sensible variant in the recension’s internal history, W ƿord, 4b, for ord in all other manuscripts, is better understood as a result of scribal carelessness than formulaic acuity on the part of a scribe who makes a number of similar errors (see above, § 7.46). If this is a scribal “performance,” it is of a markedly inferior type to that produced in the highly innovative, but usually metrically and syntactically appropriate, B1 witness to the West-Saxon eorðan recension (see below, § 7.61).

§ 7.52    On the other hand, however, it seems equally clear that the surviving witnesses by-and-large represent both the original state of the recension’s text, and a version that was read and apparently accepted without significant correction by scribes working over a period of four hundred years. Since most of the shared errors in this recension appear to go back to the original translation from Northumbrian into West-Saxon, an idealised, corrected, edition of the poem would produce a text that may never have existed in Anglo-Saxon England, and one that differs significantly from that copied by scribes who, in the eleventh and early twelfth centuries, at least, might still be assumed to have understood their native poetry. As flawed as it is, the unemended consensus text implied by the ylda witnesses is, in this sense, a scribally authorised and accurately preserved copy of an Old English poem.

§ 7.53    This edition attempts to negotiate these competing interests by allowing readers to choose between emended and unemended views of the poem. The default view (print and CD-ROM) is a critically reconstructed version of the text the original translator would have produced if he had correctly translated his Northumbrian original. This version replaces the nonsensical forms middangearde, tida, and foldum in 7a, 8b, and 9a with middangeard, teode, and foldan respectively, but leaves gehƿilc uncorrected, both because the variant is less obviously the result of a direct translation error, and because it does make some sense, syntax, and metre. This corrected version of the poem should be of interest to all who are interested in the recension as an Old English literary text (i.e. for literary enjoyment, learning Old English, metrical studies, and even, to the extent that the recension provides evidence useful in the reconstruction of the Hymn’s textual transmission, textual criticism).

§ 7.54    The alternative view (CD-ROM only) attempts to reconstruct the original (apparently corrupt) text that seems to have been responsible for all subsequent copies. This version is less likely to be of intrinsic interest to general readers, though it may prove of interest to those studying Anglo-Saxon scribal practice.

Conventions and apparatus

§ 7.55    The critical text and apparatus of the West-Saxon ylda recension follows the conventions outlined above, §§ 7.5-7.9. As mentioned above (§§ 7.48-7.49), the West-Saxon ylda recension, in its unemended form, implies a different understanding of the syntax of 5 through 9. For this reason, the two critical texts have different punctuation. The unemended view of the text shows the punctuation implied by the syntax of the surviving witnesses; the altered punctuation required by the emended text is itself treated as an emendation and indicated using square brackets. For orthographic and substantive variants, the divisions are straightforward and relatively mechanical. Accidental variants involve all differences in orthography that do not appear to affect metre, sense, or syntax. Substantive variants, here as elsewhere in this edition, are defined as all verbal differences between witnesses that do have an effect on sense, metre, or syntax, no matter how small. Significant variants need to be defined on a case by case basis. In this case, they include all substantive differences that cannot be explained as the result of an obvious scribal error. Thus the W variant ƿord in 4a is included as a potentially significant variant because it is found in a relatively early manuscript and makes (some) sense. The SanM variant æ for in 2b, on the other hand, is excluded from this list because it makes no sense and has an obvious scribal origin in the substitution <æ> for the <e+t> ligature. The Tr1 variant euca for expected frea, 9b, finally, is significant. Although it is found in a late manuscript and, as far as I am aware, is nonsensical, the connection between the expected and variant forms seems surprising enough to warrant inclusion. Readers who disagree with these decisions may prefer to work from the more mechanically determined “substantive” view.

West-Saxon eorðan recension


§ 7.56    The West-Saxon eorðan recension of Cædmon’s Hymn is known from six manuscripts ranging in date from the first half of the tenth to the end of the twelfth or beginning of the thirteenth centuries. Two of these manuscripts, C and To, have been seriously damaged or destroyed in the modern era (see above, Chapter 4: Manuscripts). In the case of C, our knowledge of the text comes from a sixteenth-century transcription of the manuscript by Laurence Nowell (N). The text of the Hymn in To is preserved in a pre-1940 photographic facsimile:

Five of the six manuscripts contain copies of the Old English translation of the Historia ecclesiastica. The sixth, To, contained a copy of what may have been the c-text of the Latin Historia (Colgrave and Mynors 1969, lxi). In manuscripts of the Old English Historia, the vernacular Hymn replaces Bede’s Latin paraphrase in the main text of IV. 24. In To, as in all other witnesses to the Hymn in manuscripts of the Latin Historia except M, the text of the vernacular poem appears on the same page as Bede’s Latin paraphrase. In manuscripts of the Old English Historia, the text of the vernacular Hymn of necessity has been copied by a hand of the manuscript’s principal text. In To, the Hymn was copied in a hand that may or may not be the same as that of the main Latin text: while the Latin and Old English show some similarities, differences in style of script and the reduced size of the facsimile make final conclusions difficult to draw (see Chapter 4: Manuscripts (To)).


Framing text

§ 7.57    Since the early eighteenth century, manuscripts of the Old English Historia have been divided into two main textual groups: one consisting of T1 and B1; the second containing C (as represented by N), O and Ca.[20] The two oldest manuscripts in each tradition, T1 and C, appear to have contained relatively similar texts.[21] Among the descendants, the members of the C/N-O-Ca group show the least internal variation, especially in the case of O and Ca which are particularly close and probably linearly related.[22] T1 and B1, on the other hand, show far more internal variation. While they share a number of common errors and omissions, the text of B1 in particular has been freely handled and contains many unique readings not found in any other manuscript of the Old English Historia.

T1 and C/N

§ 7.58    As one might expect given the location of the poem in the main text of these manuscripts, the Hymn shows a similar textual division. As is true of the framing text in the manuscripts, the Old English Hymn in T1 and C (as represented by N) appears to have shown little variation. Although the two copies of the Hymn are separated by five substantive variants, four of these involve obvious scribal errors in C/N: Ne for expected Nu, 1a; eorþū for expected eorðan, 5b; eode for expected teode, 8b, and fınū for expected firum, 9a. Since none of these errors are found in O or Ca, manuscripts that are otherwise quite conservative in the text they transmit, they are perhaps best attributed to Nowell, N’s sixteenth-century copyist (see also O’Keeffe 1990, 39). The fifth innovative form in this manuscript, ƿeoroda for expected ƿeorc, 3a, however, is reflected in readings of O and Ca, and, as a result, can be used to characterise the entire C/N-O-Ca branch.


§ 7.59    As is once again true of the framing text, the individual members of the C/N-O-Ca branch show a closer affinity with each other than do T1 and B1. While copies of the Hymn at the farthest reaches of this branch differ from each other considerably, the variants which separate them can, with only two exceptions, be traced to either the nonsensical and possibly sixteenth-century errors in C/N or corrections in O. In its uncorrected state, O has only three readings (apart from expected forms for the possible transcription errors in C/N) which are not found in the earlier manuscript: a substitution of the approximate synonym ƿera (corrected from ƿero) for C/N ƿeoroda, 3a; the addition of the prefix ge- to C/N scop (O gesceop|), 5a; and the inflectional difference, folda (corrected to foldan) for C/N foldan, 9a. In its corrected state, O supplies all but one of the readings in Ca. The only exception is the inflectional difference and substitution of synonyms Ca ƿuldres for O ƿundra, 3b.


§ 7.60    The T1-B1-To branch of the West-Saxon eorðan recension shows a less conservative pattern of development. While all three manuscripts in the group agree in reading ƿeorc against the problematic ƿeoroda/ƿera of C/N-O-Ca, they show a large number of other internal differences:

Table 1: Internal differences among members of the T1-B1-To branch of the West-Saxon eorðan recension (reverse field is used for readings in which B1 or To differ from T1)

  T1 B1 To
1a Nu Nu ƿe Nu ƿe
  sculon herigean herigan sculon sceolon herian
3a ƿuldorfæder ƿuldorgodes ƿulderfæder
3b gehƿæs fela gehƿæs
4b or ord ær
  onstealde astealde astealde
6b scyppend scyppend drihten
7a þa þe þa

§ 7.61    These differences can be explained by an appeal to the circumstances in which each text of the vernacular poem was copied. B1, which differs from T1 in seven readings and from To in six, is, in keeping with its framing text, by far the most innovative copy of any witness to the Old English Hymn. The seven variants it shows with T1 are all metrically appropriate and make at least some sense and syntax. In three cases, its “innovations” introduce readings shared with the West-Saxon ylda recension of the Hymn. They are, as a result, perhaps to be attributed to conscious or unconscious contamination (that this is likely is suggested by O, which also shows all three of these innovations, in two cases by correction). The remaining four variants, however, are both unique to B1 and metrically, syntactically, lexically, or visually far more striking.[23] The inversion in B1 of herıgan and sculon| in 1a, for example, has no significant effect on sense and little on syntax, but changes the metre to a Type B-1 from the Type A-3 line found in all other manuscripts of the Hymn. The substitution of the relative marker þe in B1 for the temporal adverb þa, 7a, in the other manuscripts of the West-Saxon eorðan recension, in contrast, has no effect on metre, but a significant effect on syntax. B1 ƿuldor godes| (for ƿuldorfæder and variants in all other manuscripts), 3a, involves the substitution of visually dissimilar synonyms. B1 fela (for gehƿæs and variants in all other witnesses), 3b, is equally striking graphically, and has an effect on both metre and syntax. In these cases, the innovation is almost certainly to be ascribed to the same source as the more general innovation found in the B1 copy of the Old English Historia ecclesiastica as a whole. As the following collation of a single page from B1 against T1 demonstrates, similar substitutions, inflectional differences, and additions and omissions of words are found throughout the B1 copy of the framing prose text (the following extracts have been transcribed from the facsimiles in this edition; punctuation as in Miller 1890-1898. Variant forms are reproduced in bold face; superscript numbers indicate relationships between readings in the two witnesses):

1    T1 singan? Cwæð he1: Sing me frumsceaft. Þa

    B1 singan? ða cwæð he1: Sing me frumsceaft. Þa

2    T1 he ða þas andsware onfeng, þa ongon he sona singan in herenesse2

    B1 he ða þas andsware onfeng, þa ongan he sona singan on herunge2

3    T1 Godes Scyppendes þa fers ⁊ þa word3 þe he næfre gehyrde4,

    B1 Godes Scyppendes ða uers ⁊ þa word godes3 þe he næfre ær ne gehyrde4,

4    T1 þære5 endebyrdnesse þis is6

    B1 ne heora5 endebyrdnesse6

5    T1 Nu7 sculon herigean8 heofonrices weard,

    B1 Nu we7 herigan sculon8 heofonrices weard

6    T1 meotodes meahte ⁊ his modgeþanc,

    B1 metodes mihte ⁊ his modgeþanc

7    T1 weorc wuldorfæder9, swa he wundra gehwæs10

    B1 weorc wuldorgodes9 swa he wundra fela10

8    T1 ece Drihten or11 onstealde12.

    B1 éce drihten ord11 astealde12

9    T1 he ærest sceop eorðan bearnum

    B1 he ærest sceop eorðan bearnum

10    T1 heofon to hrofe halig scyppend;

    B1 heofon to hrofe halig scyppend

11    T1 þa13 middangeard monncynnes weard,

    B1 þe13 middangeard manncynnes weard

12    T1 ece Drihten, æfter teode

    B1 éce drihten æfter teode

13    T1 firum foldan, frea ælmihtig.

    B1 fyrum foldan frea ælmihtig

14    T1 Þa aras he from þæm slæpe, ⁊ eal, þa þe14 he slæpende song, fæste15

    B1 Þa aras he fram þam slæpe, ⁊ eall ðæt14 he slæpende sang he hyt fæste15

15    T1 in16 gemynde hæfde. ⁊ þæm wordum sona monig word in þæt ilce

    B1 on16 gemynde hæfde. ⁊ þam wordum sona monig word in ylce

16    T1 gemet Godes wyrdes17 songes togeþeodde18. Þa com he on morgenne19

    B1 gemet gode wyrðes17 sanges þær togeþeodde18. Þa cóm he on morgen19

17    T1 to þæm tungerefan, þe20 his ealdormon wæs: sægde him hwylce gife

    B1 to ðam túngerefan, se ðe20 his ealdorman wæs: sæde him hwylce gyfe

18    T1 he onfeng21; ⁊ he hine sona to þære abbudissan gelædde ⁊ hire þa22

    B1 he onfangen hæfde21; ⁊ he hyne sona to þære abbodessan gelædde ⁊ hyre 22

19    T1 cyðde ⁊ sægde. Þa heht heo gesomnian ealle þa...

    B1 cyþde ⁊ sæde. Þa het heo gesamnian ealle þa...

Further confirmation of this tendency comes from Grant’s exhaustive survey of innovation in the B1 copy of Bede (Grant 1989). The addition of ƿe to 1a of Cædmon’s Hymn, for example, is paralleled by “83” examples in B1 of the addition of a “noun or pronoun as the subject or object” for verbs which appear without an explicit subject or object in T1 (331-2, 336-7).[24] Substitutions of stressed elements such as B1 -godes (Cædmon’s Hymn, 3b), ord (Cædmon’s Hymn, 5b), or, from the prose cited above, B1 herunge T1 herenesse, 3; B1 wyrðes T1 wordes, 17, likewise are, with over 360 occurrences, among the most frequent variants cited by Grant from the B1 text.[25] Variation in the choice of adjectives is also frequent (approximately 150 examples), although Cædmon’s Hymn 3a is the only example Grant cites of a substitution involving fela or gehwa (98-108). The substitution astealde for on|stealde· is but one example of hundreds of similar variants in the use of prefixes with nouns, adjectives, adverbs and verbs (84-9 [nouns]; 109-110 [adjectives]; 127 [adverbs]; and 197-218 [verbs]).[26] The substitution of the relative pronoun for þa in Cædmon’s Hymn, 7a, is one of numerous examples of the (correct and incorrect) substitution or addition of the relative particle (131-132 and 143-4).

§ 7.62    As was the case with B1, the unique variation in To is probably best attributed to the context in which it is found: in this case as perhaps memorial error introduced when the text was first placed in the marginal context. Its two unique forms, drıhten· for expected scyppend, 6b, and ær| for expected or (as in T1, N and the pre-correction O) or ord| (as in B1, the corrected form of O, and Ca), 4b, involve the substitution of a synonym and approximate homograph and homophone respectively. The two forms it shares with B1, the addition of ƿe to 1a and the substitution astealde· for T1 on|stealde·, 4b, involve changes that bring the text closer in line with the West-Saxon ylda recension—a motive which, as we shall see, may also influence the scribes of B1 and O (§§ 7.67 and 7.68). This in turn suggests that these particular variants may have arisen independently. While it is relatively certain that To belongs to the recension’s T1-B1 tradition, its precise place in the stemma remains indeterminate:

Transmission of the West-Saxon eorðan recension.

Image showing the likely transmission of the West-Saxon eorðan recension.

Editorial approach

§ 7.63    The principal textual problem facing an editor of the West-Saxon eorðan recension of Cædmon’s Hymn is that of the recension’s significant textual variation. On the one hand, as argued above (§§ 5.22-5.37), the eorðan text is probably, in its earliest witnesses at least, the most substantively conservative of all surviving versions of the poem. In particular, it shows no evidence of the influence of Bede’s Latin paraphrase in 1a, 5b, or 9a in T1, C/N, or (the precorrection form of) O. On the other hand, the recension also shows by far the greatest amount of significant textual variation among its surviving witnesses. B1, the most innovative witness in any recension of the poem, differs from its presumed ancestor T1 in seven significant readings. Ca, the latest member of the C/N-O-Ca sub-recension, contains five significant readings that differ from the pre-correction text of O, its immediate (or very close) ancestor.

§ 7.64    This variation is all the more striking when it is compared to that found among the witnesses to the three other comparable recensions of Cædmon’s Hymn (as argued above, Editorial introduction: Northumbrian eordu recension, the evidence suggests that all witnesses to the Northumbrian eordu recension are continental descendants of a single Insular witness). The seven surviving witnesses to the West-Saxon ylda recension show a total of four potentially significant textual variants, of which three are obvious errors preserved in fourteenth- and fifteenth- century manuscripts and the fourth a metrically inappropriate substitution of marginal sense in the hand of a demonstrably careless scribe. The two eighth-century witnesses to the Northumbrian aelda recension differ in two potentially significant forms: a dialectal and metrically transparent substitution of the synonyms and a probable scribal error. Even the relatively corrupt West-Saxon eorðe recension shows only two sets of significant variants among its three witnesses in the equivalent to 3b and 8b (see below, Editorial introduction: West-Saxon eorðe recension).

§ 7.65    As important as the sheer amount of textual variation found among the witnesses to the West-Saxon eorðan recension, however, is its significance. Where, among the witnesses to the other recensions, sensible, metrical, and syntactically appropriate variation is the exception, among witnesses to the West-Saxon eorðan recension, such variation is the norm. Of the approximately eighteen substantive variants exhibited by the six surviving witnesses of the West-Saxon eorðan recension, only five involve clearly inappropriate forms: the four probably sixteenth-century errors in C/N, and the substitution ær| for expected or in 4b of To. While the remaining variants are not always aesthetically preferable to those found in the other recensions, they are on the whole reasonably sensible and syntactically and metrically appropriate—and certainly make much better sense than the omission of drihten from 8b of CArms, or the substitutions scepen· and ƿord in M 6b, and W 4b, respectively.

§ 7.66    A final important aspect of the variation exhibited by the witnesses to the West-Saxon eorðan recension of the Hymn is its distribution. For the variants themselves are not found equally in all six manuscripts. While all witnesses except T1 introduce unique readings, none shows the same pattern of variation or introduces its variation for the same reasons. The five unique readings in C/N are all errors explainable by the misapprehension of individual letters and forms by the scribe of C or its sixteenth-century transcriber. The four substantive differences between To and T1 can be explained as a result of contamination by the West-Saxon ylda recension or scribal lapses. Ca, as noted above, shows only one variant not found in the corrected text of O, ƿuldres for ƿundra in 3b.

§ 7.67    The two manuscripts responsible for the bulk of the variants exhibited by the witnesses to the eorðan text, O and B1, similarly, vary in different ways and for different reasons. Where B1 is the product of a highly innovative scribe who shows considerable freedom in copying both the Hymn and the framing text of the prose Old English Historia ecclesiastica, the scribe of O is a relatively conservative copyist who introduces the majority of his innovations by correction (as Miller 1890-1898 notes, such corrections are found throughout the O text of the Old English Historia [xviii-xx]). Apart from the expected forms for the transcription errors, the uncorrected text of O introduces only three forms not found in its earlier relative (see above, § 7.58). In its corrected state, O “corrects” folda to foldan and ƿero to ƿera and adds another two potentially significant substantive variants: \ƿe/, 1a; and oór\d/ (corrected from oór, 4b). Since these sensible and syntactically and metrically appropriate variants are all made by correction and since all but one (ƿero{a}) are found in other recensions of the Hymn, it seems likely that the O scribe was either collating his already-copied text against a witness to the West-Saxon ylda recension,[27] or knew a similar text by heart and corrected his exemplar to match the version with which he was more familiar (both possibilities are discussed briefly by Jabbour 1968, 197). That this second possibility may be the more likely is suggested by the fact that he neither adopts any of the West-Saxon ylda recension’s nonsensical readings nor corrects ƿero to ƿeorc as in the West-Saxon ylda recension and all other versions of the poem: ƿera, “of men,” looks very much like what we might describe as a memorial conflation or misguided attempt at correction. It is, on the one hand, syntactically and semantically similar to the reading preserved in C/N ƿeoroda, “of hosts”: the fact that the final <a> in the corrected O form is written over what looks like an original <o> may even suggest the scribe began to write weroda before changing his mind. On the other hand, however, ƿero{a} is also graphically similar to ƿeorc, “work(s),” and fills an equivalent slot in the metrical line: where the C/N reading is Type D*2 or D*4, the O text is Type D-2 or D-4 (as in all other recensions of the Hymn). Recognising that the C/N form was incorrect, the scribe of O appears to have corrected his original ƿero by supplying a form which is semantically and grammatically similar to the form in C/N, but metrically equivalent to that in all other versions of the Hymn.

§ 7.68    In contrast to the relatively pedestrian O, the B1 text is far more spontaneous and innovative. None of the seven differences it exhibits from the text of its closest surviving ancestor are by correction. On the one hand, of course, this may suggest simply that the variants have been inherited from an earlier copy. B1 was copied by two scribes and, in as much as Grant reports no real difference in the amount or type of innovation in each scribe’s stint, it appears that the manuscript may owe much of its originality to some earlier exemplar. On the other hand, however, the variants themselves are also in most cases of a different type: where all but one of the five variants introduced by the corrections to O are found in other versions of the Hymn, four of the seven variants attested in B1 are unparalleled among the Hymn’s other witnesses, and, in one case, indeed, anywhere else in the poetic corpus.[28]

§ 7.69    A critical text of the West-Saxon eorðan recension has then, ideally, two competing goals. On the one hand, the evidence of the surviving manuscripts suggests that, with the exception of B1 and, possibly, O, most scribes attempted to reproduce their exemplars with a reasonable degree of accuracy. While the amount of innovation among witnesses to this recension is less than that found among witnesses to the ylda tradition, the variants introduced tend on the whole towards what looks very much like memorial contamination and trivialisation. On the other hand, the evidence of O and B1 shows that not all scribes of this recension shared this desire for accuracy.

§ 7.70    Taken together this evidence suggests both that it is possible to reconstruct a text closely resembling the original exemplar of the West-Saxon eorðan recension, and that the reproduction of this text was a primary goal of most of the recension’s scribes. That this text resembled that of T1 in almost every respect is suggested by the distribution of variants among the surviving witnesses, and, in particular, by the relative lack of significant substantive textual variation between T1, C (as represented by N), and the uncorrected O.

§ 7.71    The second goal for any new edition of the West-Saxon eorðan recension involves providing readers with a sense of the text’s development. The changes introduced in O and B1 represent very different types of scribal performances. Where the majority of innovations in B1 are found in no other copy of Cædmon’s Hymn, four out of the five introduced in O are found in other versions of the poem. Where the scribes responsible for the innovations in B1 appear to have worked relatively fluently, the scribe responsible for O “performed” primarily by correction. In presenting the B1 and O performances as editions—rather than simply part of the witness archive supporting all recensions of the HymnCædmon’s Hymn: a multimedia edition and archive draws readers’ attention to both the nature of the performances and their significance to the development of this recension: the B1 edition presents an editorially lineated version of the poem with an apparatus emphasising the significant differences this text shows to T1 and To and members of the C/N-O-Ca sub-recension; the O text focuses on the scribe’s performance-as-process by recording the corrections, additions, and deletions in the editorial text. By comparing these various editorial texts with each other and the variants presented in the accompanying textual apparatus, readers can gain a sense of the decisions these redactors made in altering their original text.

Apparatus and conventions

§ 7.72    The critical text of the West-Saxon eorðan recension follows the conventions outlined above, §§ 7.5-7.9.

West-Saxon eorðe recension


§ 7.73    The West-Saxon eorðe recension is known from three manuscripts from the late eleventh or first half of the twelfth century:

All three manuscripts contain copies of the c-text of Bede’s Historia ecclesiastica. In Hr and Ld, the Old English Hymn has been copied as a marginal gloss to Bede’s paraphrase of the poem, with signes de renvoi (added in a later hand in the case of Hr) linking the Old English text to the equivalent Latin in both manuscripts. In CArms, the poem has been copied in alternating lines with the paraphrase as part of the main text. The only indication that the Old English is not part of Bede’s Latin comes at the beginning and end, where the first and last two or three words have been copied interlineally. In Ld and CArms, the scribe of the Old English is the same as that of the Latin text. In Hr, the vernacular poem is the work of a contemporary corrector.


Latin text

§ 7.74    The precise filiation of the Latin text of the Historia ecclesiastica in these witnesses is unknown. Colgrave and Mynors 1969 identify Hr as belonging to the “Digby” or “Southern Common Text” of the Historia (lvi). Ld contains a conflated text that Colgrave and Mynors associate with the “Gloucester” tradition (liii). CArms remains unclassified but appears to be associated with the “English” tradition.

Old English text

§ 7.75    Relations among the three witnesses to the Old English text are easier to reconstruct. As the following diagram illustrates, the three witnesses can be broken into two main branches, with Ld and Hr in one, and CArms in the other. This division is suggested by a series of mutually exclusive errors in the two branches: a scribe in the tradition represented by CArms has had trouble with the insular forms of <g>, <ƿ>, <f>, and <r> (e.g. modre þanc; seorc, sysū); Ld and Hr, on the other hand, both show a confusion of <n> and <o>, not found in CArms, in scyppeod:

Transmission of West-Saxon eorðe recension.

Image showing the likely transmission of the West-Saxon eorðe recension.

Of the three witnesses, finally, Hr contains by far the more innovative and recent version of the text. In contrast to Ld and CArms, which contain the text in slightly varying forms, Hr omits most of 3b and 4a.

Editorial approach

§ 7.76    The West-Saxon eorðe recension poses two main problems for the editor. The first involves determining the recension’s relationship to other versions of the poem; the second finding an appropriate medium for the presentation of the poem’s corrupt text.

Relationship of the West-Saxon eorðe recension to other versions of the Hymn.

§ 7.77    Until recently, the eorðe text was assumed to be a corruption of the West-Saxon eorðan recension (e.g. Dobbie 1937, 43; O’Donnell 1996b, Appendix). This assumption, which appears to have been questioned first by Schwab (Schwab 1972, 15-17), was based primarily on the similarities of dialect between the eorðe and eorðan texts and the fact that both versions contained an inflected form of eorðe, 5b. The assumption ignores, however, a number of other features that would seem to ally the West-Saxon eorðe recension more closely with the West-Saxon ylda or Northumbrian eordu texts: like the eordu and ylda recensions, the eorðe version is found exclusively in copies of the Latin Historia ecclesiastica; like eordu and ylda, it has an explicit pronominal subject in 1a and adds on to 9a; the recension’s eponymous reading in 5b, eorðe can perhaps be more easily explained as a corruption of an original eordu showing Northumbrian loss of inflectional /n/ than as a late weakening of West-Saxon weak endings, especially since the surviving witnesses agree for the most part in preserving the full form of similar endings throughout the rest of the poem (e.g. heofonrices, heofon and middangeard; cf. Cavill 2000, 506 and 510). The only problem with this assumed derivation from the Northumbrian eordu recension, indeed, lies in the textual history of the Latin Historia ecclesiastica: Hr, like Ld, CArms, and all other surviving witnesses to Bede’s Latin Historia except for M and P, belongs to the “English” or c-text of the Historia ecclesiastica; witnesses to the Northumbrian eordu recension all appear to belong to the “German” or m-text version. If the West-Saxon eorðe recension has indeed been translated from the Northumbrian eordu recension, the scribe responsible for the initial copy of this version would presumably also have had to have access to a “German” text of the Historia ecclesiastica.

Textual corruptions

§ 7.78    The second major problem facing editors of the West-Saxon eorðe recension involves its many obvious corruptions. On the one hand, the three surviving witnesses clearly are descended from an extremely corrupt original. While they show some minor differences and one major variant (discussed below), all three manuscripts agree in omitting all or some of 3b-4a, dropping all of 4b, and moving the equivalent of 6b to the end of the poem. On the other hand, however, this corrupt text appears to have undergone some development itself. In Ld and CArms, 3b and 4a show signs of what may be an attempt to repair the damage caused by the omission of 4b: by adding þa (an adverb found at this point of the poem in no other recension) to the end of 4a, the two manuscripts make what could be understood as an unsuccessful attempt at linking the poem’s first and second halves. In Hr, this editorial intervention may have been carried further: by dropping 4a, and most of 3b as well as 4b, Hr turns the equivalent of 3a-6a into a single clause—making, with the exception of eorðe, marginally acceptable sense if usual metre:

3b         ....         swa he

    ærest sceop      eorðe bearnum

    heofon to hrofe.

As he first made heaven as a roof for men earth.

§ 7.79    The problem with this variation, however, is that it may or may not be deliberate. While the change makes some (albeit still problematic) sense out of a passage that is irredeemably garbled in Ld and CArms, the “correction” itself looks suspiciously like a scribal error: the omitted text from Ld begins and ends with the pronoun he—suggesting eyeskip might have played as important a role as a desire to fix the manuscript’s nonsensical original. In addition, the Hr text shows no signs of having attempted repairs to other corrupt sections of the text (including, indeed, the problematic ending on eorðe).

§ 7.80    Despite these doubts about the intentions of the Hr “editor,” Cædmon’s Hymn: a multimedia edition and archive bases its critical text of the recension on this late version of the poem. In doing so, it departs from its practice in editing the West-Saxon ylda, Northumbrian eordu, and West-Saxon eorðan recensions. In these editions, the editorial goal has been the reconstruction of either an idealised, pre-corruption form of the text represented by the surviving witnesses, or in the case of the “uncorrected” text of the West-Saxon ylda recension, the recension’s earliest form, corruptions and all.

§ 7.81    The critical text of the West-Saxon eorðe recension is different, however, in that its primary interest lies in the text’s transmission history. A critical reconstruction of the recension’s ideal (i.e. non-corrupt) original would produce something that differed in no substantive way from a West-Saxon translation of the reconstructed Northumbrian eordu recension. The original, corrupt, exemplar of the recensions would presumably look much like Ld. By selecting the latest manuscript of this recension for its critical text, this edition shows what happened to a recension as it was transmitted by scribes who were neither as careful as the ylda scribes, nor as interested in preserving sense and metre as those responsible for B1 and O. The changes introduced in the version represented by Hr improve the sense of the poem without bringing it any closer to its original state. To the extent this is deliberate, it represents a style of transmission seen nowhere else in the Old English poetic corpus (O’Donnell 1996a). To the extent these changes are the result of a scribal error, they demonstrate the way in which scribal mistakes can compound to change a text’s form and meaning irreparably (see Moffat 1992). Either way, they are evidence of the fate of texts in a late scribal tradition.

Conventions and apparatus

§ 7.82    The critical text of the West-Saxon eorðe recension follows the conventions outlined above, §§ 7.5-7.9.


[1]This combination of approaches is a well-established desideratum in electronic editing. See McGillivray 1994, McGann 1995, Vanhoutte 1999, and Vanhoutte 2000.

[2]See McGillivray 1994 for a well-balanced defence of the continuing relevance of evaluative approaches to editing in the “post-critical” age.

[3]The extent to which digital representations of evidence necessarily interpret the underlying material from which they have been derived is discussed in Solopova 2000, “Guidelines for Transcription of the Manuscripts of The Wife of Bath’s Prologue,” especially § [2.1], “The interpretative character of transcription of manuscripts for the computer.”

[4]Terminology used in this discussion is based on Greetham 1994, especially Appendix II. The “Reading texts” are not pure clear text editions in that they include explicit indication of emendations supported by no manuscript authority (cf. Greetham 1994, 368). The editions are eclectic in that they are based on a copy-text but supplemented with readings from other manuscripts and modern editorial emendations when necessary.

[5]The reconstructed text of the Northumbrian archteype has two sets of apparatus (on the CD-ROM), one collating readings from the edited recensional texts, another showing readings from all surviving witnesses. See below, § 7.16.

[6]Metrical patterns are assumed to be “significantly” different when the variation alters the type or subtype to which a pattern belongs in Siever’s system. In most cases, this excludes variation between resolved and unresolved lifts, but includes variation between short and long lifts. This is an imperfect distinction, since resolution does at times appear to be statistically significant (see in particular, Hutcheson 1995, 396). Variation between resolved and unresolved forms is, however, quite common in the general corpus of multiply attested Old English poetry. See O’Donnell 1996a for a discussion of variation in the corpus.

[7]In arguing that versions of the Hymn with “eordu-eorðan... might... reflect an oral text which has undergone slight garbling in the process of transmission” (525), Cavill 2000 revives a suggestion first made in Zupitza 1878, who was writing before the discovery of the Northumbrian eordu recension and the development of contemporary oral theory. Cavill’s claim that this makes the question of “lectio difficilior... redundant, because there is no lectio” is disingenuous, however. Whether we assume the transmission was “oral” or “literate,” we are still left with the problem of explaining how a significantly difficilior variant might arise in the course of the Hymn’s dissemination. All current theories of “oral” transmission would predict the opposite pattern of substitution, i.e. from the nonce—and hence difficilior—collocation eordu barnum to the relatively well attested—and hence facilior—formula aelda barnum. See, for example, the discussion of the “disappointed scribal expectations” (O’Keeffe 1990, 95) which lead to the substitution of the formulaic eorðbugendum for nonce iegbugendum in the Tr1 copy of Metrical Preface to the Pastoral Care (O’Keeffe 1990, 93). Memorial transmission, likewise, Jabbour notes, has a documented tendency “to convert the specific to the general, the individualized to the stylized; aberrations in either style or conception... into forms more nearly central to the tradition” (Jabbour 1968, 43).

[8]Dobbie 1937, 45-48, esp. 46, 47 and the illustration on 48; this same view is reflected in the introduction to Dobbie 1942, despite its lack of realisation in the accompanying text. See Dobbie 1942, xcvii-xcviii.

[9]There are strong reasons for questioning Dobbie’s original view of the Hymn’s textual transmission; an alternative explanation is proposed above, Chapter 5: Filiation and transmission.

[10]The traditional dating of these manuscripts depends on a number of controversial assumptions about the copying of the so-called “memoranda” in each manuscript. For a discussion of the issues involved, see Note A: The dating of M and P, below.

[11]Pope and Fulk 2001 and Smith 1978 are best-text editions. Dobbie 1942 takes M as the base for his composite text of the “Northumbrian Version.” See above, § 7.14.

[12]Tr1 has an unrelated copy of the Hymn in a similar position, indicating that the Hymn could be placed independently in this position. With the exception of the Old English translation of the Historia ecclesiastica (which always contains the West-Saxon eorðan recension), no single recension of the Historia ecclesiastica is characterised by the presence of a particular text of the vernacular Hymn.

[13]Di is unique in combining a complete copy of the Historia with the prose Life. In addition to the complete Historia, Di also contains the excerpts from IV. 31-32 found in all other manuscripts of the prose Life. See Laistner 1943, 107, and cf. 89-90 with 94-102.

[14]Cavill 2000 suggests that the “odds against the scribes of Dijon [Di] and Paris [P1] independently dropping the h” must be “huge.” Coincidental errors, even in unrelated manuscripts of the Hymn, are not unheard of, however. See, for example, the coincidental errors in Di and Hr, line 8b/7b, Br and CArms, line 9b, and especially the coincidental error and correction in W and Ld, line 1a. Given that no scribe of Northumbrian eordu recension appears to have understood the text, it seems safer to assume the scribes of P1 and Di made a coincidental error than that the scribe of Br made a spontaneous correction.

[15]In the apparatus to the edition of the Hymn’s hypothetical archetype, variants from this edition are divided into three categories as if they belonged to a living tradition.

[16]Dates in brackets represent the consensus for the manuscript and vernacular Hymn. Dates separated by a forward slash indicate date ranges. Semi-colons distinguish between the date of the manuscript, and that of the Old English Hymn when these differ. When a single date or date-range is given, the Hymn has been copied at the same time as the main Latin text.

[17]Hr, another member of the “Digby” tradition, contains a copy of the West-Saxon eorðe recension; Ld, CArms, and To, all of which Colgrave and Mynors assign to unspecified branches of the c-text (Colgrave and Mynors 1969, liii and lxi), likewise contain copies of the vernacular poem’s eorðe or eorðan recensions. Although Colgrave and Mynors assign Di to the English tradition, the manuscript is more likely a member of the same “German” or m-text tradition as P1 and Br. See above, §§ 7.29-7.30.

[18]Dobbie describes this substitution as being “quite ungrammatical” without explaining what he means (Dobbie 1937, 39). Grammatically, gehƿilc can be understood as an accusative object of astealde, “established,” parallel to ord, “beginning,” in 4b, a pattern of (syntactic) variation and repetition which is extremely common in Old English poetry, although it is largely confined in Cædmon’s Hymn to the last five lines of the poem. See above, §§ 3.16-3.19.

[19]The text of the Hymn in this manuscript was destroyed in 1731; a sixteenth-century transcription is preserved in N London, British Library, Additional 43703.

[20]See Grant 1989, 5-7; Whitelock 1962, 81, n. 22; Miller 1890-1898, 1: xxiv-xxvi; Schipper 1897, 2: xi-xxxv. For a modified view of this traditional division, see Grant 1989, 6. This modified stemma does not affect the following discussion: it still assumes that C O and Ca belong to a common tradition and that T1 and B1 are closer to each other than to the other three copies of the Old English Historia ecclesiastica.

[21]See Miller 1890-1898, 1: xxv.

[22]Schipper 1897, 2: xix. Both Dobbie and Schipper cite Julius Zupitza Altenglisches Übungsbuch, 2nd edition [Vienna: 1881] iv) as the first to notice this relationship (Dobbie 1937, 213; Schipper 1898-1899, xix). I have been unable to find a record of this text.

[23]See Jabbour 1968, 70, for a discussion of such “striking” variants.

[24]The figure “83” is given on 331. Although Grant does not break his count down into separate figures for nouns and pronouns, all but one of the examples he cites involve the addition of a pronoun.

[25]Cædmon’s Hymn, 3b, is the only example of variation between -god and -fæder listed by Grant; variation between B1 god and T1 drihten (and, less frequently, vice versa), however, is relatively common. In Grant’s citations, B1 substitutes god(-) for T1 driht(e)n(-) five times, B1 driht(e)n(-) for T1 god- twice. B1 and T1 agree in god(-) twice for driht(e)n(-) in other manuscripts of the Historia. See Grant 1989, 51-2.

[26]Grant does not include the variant from Cædmon’s Hymn in his list of variants in the use of prefixes.

[27]The O scribe does not adopt any of the West-Saxon ylda recension’s most innovative readings: gehƿilc, middangearde, and tida (see above, §§ 7.47-7.50). This may indicate that a second, corrected copy of the West-Saxon ylda recension was in circulation, or it may be further evidence to suggest that the West-Saxon ylda recension’s remarkable stability was the result of a deliberate scribal policy which the O scribe, who was copying his text in a different context, felt free to ignore.

[28]herıgan sculon|, 1a, is also found in Seasons for Fasting, 39a; ƿund ra fela appears four other times with wundra as a simplex (Exodus, 10b; Order of the World, 7b; Riddle 21, 8b; Riddle 84, 10b) and another three with wundra as the second element of a compound; ƿuldorgodes (indeed any compound of wuldor with god) is unparalleled in the poetic corpus (Healey et al. 1994); while the substitution þe for þa is more difficult to check against the corpus, it seems to be syntactically permissible.