Chapter 5
Filiation and transmission

[Skip to Content]


General Navigation

This chapter

[Beginning of Content]


§ 5.1    Since the beginning of the twentieth century, manuscripts of Cædmon’s Hymn have been divided into four main textual recensions, identified by dialect and their reading for the first word of 5b (sigla in brackets are from Dobbie 1937):

  1. The Northumbrian aelda recension (*β)
  2. The Northumbrian eordu recension (*γ)
  3. The West-Saxon ylda recension (*Z)
  4. The West-Saxon eorðan recension (*Æ)

In recent years, a fifth, late, and imperfect version of the poem has been identified in three manuscripts of the late eleventh or twelfth century:

  1. The West-Saxon eorðe recension

Originally thought to be a simple ad hoc corruption of the West-Saxon eorðan recension (e.g. Dobbie 1937, 43), this version of the poem has since been shown to have a much closer connection to the Northumbrian eordu text (Schwab 1972, 15-17; Orton 1998, 158-159).

§ 5.2    The evidence for this traditional filiation is compelling. The five recensions show very little internal variation but can be distinguished from each other on the basis of thirteen inter-recensional significant substantive variants[1] (variant readings unique to one recension are printed in reverse field; readings used to group different recensions together are printed with different backgrounds):

Table 1: Cædmon’s Hymn recensional variation

1aNuNu ƿeNu ƿueNu[2]Nu ƿe
4borordoror[4]4b missing
9afirumfirum onfirum onfirumfyrum on
9b-10--------6b added after 9

§ 5.3    Most of these variants involve the introduction of obvious (and usually minor) errors or the substitution of roughly equivalent forms. Thus early witnesses to the West-Saxon eorðan recension are distinguished by the reading onstealde in 4b for astelidæ (and orthographic/dialectal variants), an etymologically related and synonymous form in all other recensions. Manuscripts of the West-Saxon ylda recension, for their part, can be identified by a number of unique errors and trivial substitutions: the sensible ord, “origin,” 4b, and gescop, “made,” 5a, for synonyms or and scop in all other recensions[7]; the probable errors gehƿilc, middangearde, tida, and foldum, in 3b, 7a, 8b, and 9a, for expected gihuaes, tha middungeard, tiadæ, and foldu in most other witnesses. The West-Saxon eorðe recension is distinguished by a number of serious errors and alterations, the most significant of which include the omission of an equivalent to 4b from all witnesses and the movement of the equivalent to 6b to the end of the poem. The Northumbrian aelda and Northumbrian eordu recensions, finally, do not show any unique substantive variants. They can, however, be distinguished by the readings they share with other versions of the poem: the Northumbrian aelda recension agrees with the West-Saxon ylda recension against all other versions in its reading for the first word of 5b: aelda / ylda versus eordu / eorðan / eorðe; it agrees with the West-Saxon eorðan recension against the Northumbrian eordu, and West-Saxon eorðe and West-Saxon ylda recensions, however, in its readings for 1a and 9a: Nu scylun hergan and firum foldu versus Nu ƿue sciulun herga and firum on foldu. The Northumbrian eordu recension shows the opposite pattern of agreement and disagreement: it agrees with the West-Saxon eorðan and West-Saxon eorðe recensions against the Northumbrian aelda and West-Saxon ylda texts in 5b, and with the West-Saxon ylda and West-Saxon eorðe recensions against the Northumbrian aelda and West-Saxon eorðan texts in 1a and 9a.


§ 5.4    Traditionally, the Hymn’s recensional division has been understood as implying a particular textual history. This understanding is based on the recognition that the five variants in 5b used to name the major recensions can themselves be divided into two groups of dialectal cognates: aelda / ylda, “of men,” versus eordu / eorðan / eorðe, “(of) earth.”[8] The fact that aelda and ylda are etymologically related has been assumed, since the earliest days of modern Cædmon’s Hymn textual scholarship, to indicate that the recensions named for these variants must be closer to each other than other versions of the poem.[9] The fact that the Northumbrian eordu, West-Saxon eorðan, and West-Saxon eorðe recensions all share a cognate meaning “earth,” likewise, has been understood, since witnesses to the Northumbrian eordu recension were first discovered in 1906, to indicate that these three versions of the poem must also share some distant common ancestor, distinct from that responsible for the Northumbrian aelda tradition.[10] This view of the poem’s textual history was first suggested by Wuest in his pioneering article on the witnesses to the Northumbrian eordu recension (Wuest 1906; esp. 225). It was endorsed, with some disagreement as to the priority of the two branches, by Frampton (Frampton 1924, esp. 8), and given its most developed form by Dobbie in his influential study of the manuscript tradition (Dobbie 1937, 48; see figure 1, below). Apart from the suggestion that witnesses to what is now known as the West-Saxon eorðe recension are descended from the Northumbrian eordu recension directly rather than through the West-Saxon eorðan text (Schwab 1972, 15-17; Orton 1998, 158-159), and, most recently, a revival of Zupitza’s suggestion that the Northumbrian eordu : Northumbrian aelda split may have arisen orally in the course of the poem’s early history (Cavill 2000; Cavill 2002; cf. Zupitza 1878, 216), Wuest’s understanding of the poem’s recensional development does not appear to have been questioned in any significant way for almost a century.

Figure 1: Traditional stemma (as in Dobbie 1937, 48)

Stemma showing the traditional understanding of the development of Cædmon’s Hymn along two main aelda and eordu branches.

§ 5.5    In this case, however, the traditional view of the Hymn’s development is far less compelling. As we shall see, the assumption that the poem broke into separate aelda and eordu traditions early on in its history is supported by no linguistic, textual, or contextual evidence other than the eponymous readings in 5b. It is also contradicted by evidence of considerable “cross contamination” among individual members of its two supposed textual “branches.” In some cases, indeed, this “cross contamination” is regular enough to call the entire traditional stemma into question. The following section explores the evidence for and against the traditional analysis of the Hymn’s textual development. This is followed by a section in which an alternative stemma is proposed to account for the poem’s textual history. Throughout this chapter, the priority of the vernacular Hymn over Bede’s Latin paraphrase in Historia ecclesiastica IV. 24 is assumed; the evidence underlying this assumption is discussed below, Note B: Cædmon’s Hymn and Bede’s paraphrase.

The case against the traditional view of the textual development of Cædmon’s Hymn

Dialect and material context

§ 5.6    As argued above, the traditional understanding of the textual development of Cædmon’s Hymn rests entirely upon the fact that the variants most commonly used to identify the poem’s individual recensions can be divided into two groups of etymological cognates. The problem, however, is that these two, supposedly ancient, branches then develop no other distinguishing textual, linguistic, or material characteristics in the course of a textual history spanning approximately seven hundred years and two dialects. The traditional aelda and eordu “branches” both survive in West-Saxon and Northumbrian dialects and both are found in copies of Bede’s Historia ecclesiastica or (West-Saxon eorðan recension only) its Old English translation. Unlike Bede’s Death Song, moreover, where, with one partial exception, members of the two main recensions of the Old English poem can be identified by their association with particular versions of Cuthbert’s Epistola as well as by dialect and the presence of several characteristic textual variants (Dobbie 1942, c-cvii, esp. ci), no such direct correlation exists between the traditional “branches” of Cædmon’s Hymn and the associated text of Bede’s Historia: while certain recensions of the Hymn show a strong affinity with certain recensions of the Historia ecclesiastica or its Old English translation, neither “branch” is found exclusively in any one recension of the Historia ecclesiastica, and there is considerable evidence to suggest that the poem was often added to existing manuscripts of the Historia ecclesiastica (see O’Keeffe 1990, 37).


§ 5.7    A more significant problem with the traditional division of the poem into aelda and eordu “branches,” however, involves the large amount of apparent “cross-contamination.” Witnesses to the Northumbrian aelda recension, for example, are far closer to members of the West-Saxon eorðan recension than to their supposed descendents in the West-Saxon ylda recension: they share but a single unique substantive variant with the West-Saxon ylda recension (the eponymous readings in 5b), and differ from it in nine:

Table 2: Substantive variation between the Northumbrian aelda recension and West-Saxon ylda recension.

 Northumbrian aelda recensionWest-Saxon ylda recension
1aNuNu ƿe

For its part, the Northumbrian aelda recension shares two unique readings (the absence of ƿue/ƿe and to in 1a and 9a) with the West-Saxon eorðan recension, a descendant of the other supposed “branch,” and differs from it, in its earliest manuscripts at least, in only two substantively different forms: astelidæ : onstealde, 4b, and aelda : eorðan, 5b.[11]

§ 5.8    The Northumbrian eordu recension, similarly, shows a greater affinity with the corrupt West-Saxon ylda and West-Saxon eorðe recensions than it does with its supposed nearest “relative,” the West-Saxon eorðan recension. Like the West-Saxon ylda and West-Saxon eorðe recensions, it reads ƿue/ƿe and on in 1a and 9a[12] and, with all recensions except the West-Saxon eorðan recension, astalde in 4b.

§ 5.9    Of these various agreements and disagreements across the two main traditional “branches,” it is the variants in 1a and 9a that are the most troubling. This is because they can be used to construct an alternative textual stemma:

Figure 2: Alternative stemma

Alternative to the traditional stemma in which the Hymn is broken into +ƿue (1a) and on (9a) and -ƿue (1a) and on branches.

If, as the traditional view suggests, the survival of a single shared reading in one Northumbrian and one West-Saxon recension can be understood as indicating the existence of a genetic relationship, then the variants in 1a and 9a would appear, at first glance at least, to offer at least as great a claim to textual significance, particularly in as much as these other variants are almost invariably correlated: with the exception of three late manuscripts in the West-Saxon eorðan recension (B1 To Ca),[13] witnesses that read ƿue/ƿe in 1a always also have on in 9a.[14] While a random pattern of variation might produce some agreement among individual recensions or witnesses, this consistency suggests the possibility of a more significant connection.

The nominal variation in 5b

§ 5.10    The traditional preference for the nominal variation in 5b over the coordinated pronominal and prepositional variation in 1a and 9a as a genetic indicator has rarely if ever been defended explicitly. To the extent that it has been discussed at all, its defence appears to rest on the assumption that identical nominal substitutions are less likely to occur coincidentally in transmission than substitutions in the use of pronouns and prepositions.[15] As Orton, who inexplicably ignores the pronominal variation in 1a, suggests, it is “[o]n the whole... easier to believe that on was added twice independently in 9 than that aelda was substituted for eordu twice in 5” (Orton 1998, 157).

§ 5.11    This is a reasonable assumption in principle. Nouns are a larger and more open word class than pronouns and prepositions, and, as a result, are far less subject to identical but independent variation in unmarked environments. In the absence of any strong indication as to the correct reading, we might expect that scribes would have difficulty making identical but independent substitutions for difficult or corrupt readings in a common exemplar.

§ 5.12    That this is so can be demonstrated from the extant poetic corpus. The mid tenth-century copy of the Battle of Brunanburh in Cambridge, Corpus Christi College, 173 (“Parker Chronicle”) (ChronA), for example, contains the nonsense reading cul bod ge hna des in 49b, a line we know from other witnesses should have read cumbolgehnastes, “of banner clash.” A subsequent corrector, usually identified with the early eleventh-century scribe of the now largely destroyed † London, British Library, Cotton Otho B. xi (ChronG) (see Lutz 1981, 221), attempted to fix this obvious nonsense by adding the interlinear gloss ł cumbel, “vel [i.e. or] banner” above the first half of the compound while leaving the second, equally nonsensical, element uncorrected.[16] Similarly, the scribe of the early-eleventh-century O witness to the West-Saxon eorðan recension of Cædmon’s Hymn corrects an exemplar that apparently read wero (cf. N [for C] ƿeoroda) for expected uerc (West-Saxon ƿeorc), “work(s),” to ƿera, “of men.”[17]

§ 5.13    The problem, however, is that both these examples involve the attempted corrections of a nonsensical exemplar for which the correct reading was itself highly unusual, and, hence, unpredictable. The correct form in Brunanburh, 49b, cumbolgehnastes, is a nonce compound in Old English (the simplices are attested somewhat more frequently); in Cædmon’s Hymn, 3a, it is the collocation of uerc with uuldurfadur that is unusual: while uerc itself, with over sixty occurrences in the uninflected form alone, is not rare, the combination with uuldurfadur is otherwise unattested in Old English.[18] This means in both cases that scribes attempting to fix the difficult nonsense forms in their exemplars would have had very little guidance in guessing the correct form. A scribe facing cul bod ge hna des in Brunanburh, 49a, with no indication as to the original reading, would not be able to rely on either his innate knowledge of Old English vocabulary[19] nor, in this case, his familiarity with similar lines from elsewhere in the poetic corpus in guessing the correct form. A scribe facing an exemplar in wero in 3a of Cædmon’s Hymn, similarly, while perhaps less handicapped by the unusualness of the poem’s vocabulary, would be equally unlikely to guess the correct reading by recognising any underlying formula. In both cases, indeed, a scribe intent on making “formulaic” substitutions if anything would be likely to take his copy farther away from what we are fairly certain must have been the original reading.[20]

§ 5.14    The nominal variation eordu : aelda in Cædmon’s Hymn, 5b, is different, however, in that it involves neither a corrupt exemplar nor a choice between two unusual or otherwise unattested forms: eordu barnum, while not otherwise attested in Old English, makes perfect sense and metre; the alternative, aelda barnum, with eleven occurrences in various cases in Old English (Healey et al. 1994), similar collocations in other Germanic languages (Dobbie 1937, 48), and a very large number of close parallels following the pattern [ noun referring to a class of people] + barn-, is metrically, semantically, and syntactically appropriate and well attested; it is also, after israhela barnum, the only repeated collocation in the corpus suitable for vocalic alliteration in the off-verse (Healey et al. 1994). If we assume for the sake of argument and on the principle of difficilior lectio that the unique Northumbrian eordu reading was original, it is difficult to see why a scribe facing an exemplar in eordu bearnum would feel the need to change his text for anything other than formulaic or (as we shall see below) doctrinal reasons; if he decided on these grounds to do so, it is equally difficult to see what variant he would be tempted to introduce other than the sensible, metrical, formulaically appropriate, and well-attested aelda barnum.

The pronominal and prepositional variation in 1a and 9a

§ 5.15    A similar difference in relative “difficulty” is also characteristic of the pronominal and prepositional variation in 1a and 9a. In these examples, the “less difficult” and more commonly attested readings are those that include the pronoun or preposition.

firum foldu : firum on foldu, 9a

§ 5.16    In 9a, the difficilior reading is that of the Northumbrian aelda and West-Saxon eorðan recensions. The Northumbrian eordu, West-Saxon ylda, and West-Saxon eorðe collocation, firum on foldu, is both syntactically less ambiguous, and, with twenty-one occurrences in the Anglo-Saxon corpus, far better attested. In the Northumbrian eordu, West-Saxon eorðe, and, more corruptly, West-Saxon ylda version of the poem, 9a is a clear adverbial phrase serving a single syntactic function (it supplies an indirect object for tiade, “appointed,” in 8b):

5    He aerist scoop      eordu bearnum

    hefen to hrofe,      halig sceppend;

    ða middungeard,      moncinnes ƿeard,

    eci drichtin,      aefter tiade

    firum on foldu,      frea allmechtig.

He first made, for the children of earth [West-Saxon ylda recension: ylda, “of men”],[21] heaven as a roof. Then afterwards the guardian of mankind, the eternal lord, the almighty king, appointed the middle-earth, for men on the land.

(Text: Northumbrian eordu recension[22]).

The reading of the Northumbrian aelda and West-Saxon eorðan recensions, on the other hand, is syntactically more ambiguous. In this version of the text, foldu can be understood as either genitive singular, in which case it is a complement to firum, “for men of the earth,” or, perhaps more likely,[23] accusative singular, in which case it functions as a direct object for tiadæ, parallel to middungeard in 7a:

5    He aerist scop      aelda barnum

    heben til hrofe,      haleg sceppend;

    tha middungeard,      moncynnæs uard,

    eci dryctin,      æfter tiadæ

    firum foldu,      frea allmectig.

He first made, for the children of men [West-Saxon eorðan recension: eorðan, “(of) earth”], heaven as a roof. Then afterwards the guardian of mankind, the eternal lord, the almighty king, appointed the middle-earth, the land, for men (or: for men of the earth).

(Text: Northumbrian aelda recension[24]).

§ 5.17    As with the nominal substitution in 5b, it is difficult to see why a scribe, faced with an exemplar showing the syntactically, semantically, and metrically appropriate facilior reading firum on foldu would change his copy to the syntactically more difficult and otherwise unattested collocation firum foldu—and very easy to see how the same scribe, faced with the non-formulaic and syntactically more ambiguous (and hence difficilior) reading of the Northumbrian aelda and West-Saxon eorðan recensions, might be tempted to substitute the less-ambiguous, better attested, and formulaic reading of the Northumbrian eordu, West-Saxon ylda, and West-Saxon eorðe recensions.

Nu scylun : Nu ƿue sciulun, 1a

§ 5.18    In the case of 1a, the reading with ƿue/ƿe is better attested and less “difficult.” The use of a nominative pronoun before plural indicative verbs is very common in Old English and can be paralleled by hundreds of examples from the corpus. The reading without ƿue/ƿe, on the other hand, like that with eordu in 5b and without on in 9a, is both far less common and syntactically more difficult.

§ 5.19    As traditionally understood by editors and translators of Cædmon’s Hymn, the subject of scylun is assumed to be implicit in versions of the poem without an expressed pronominal subject (e.g. Dobbie 1942, 198; Smith 1978, 3-4; Frampton 1924, 7). As Mitchell points out, however, there is very little evidence to suggest that the “1st pers. nom. pl. pron. we could be unexpressed at the beginning of a poem in which it does not occur and in which there was therefore no first person grammatical referent” (Mitchell 1985b, 192; cf. Pope and Fulk 2001, 57). The alternative reading, suggested variously by Mitchell, Ball, and Howlett, is to take uerc, “work(s),” 3a, a neuter noun which can be construed as nominative or accusative singular or plural, as the nominative plural subject of the poem’s opening verb (Mitchell 1985b; Ball 1985, 39-41; Howlett 1974, 6). This is itself not without difficulty, however, since, as Mitchell notes, the resulting “Adverb-Verb-triple Object-Subject pattern” does not appear to be attested anywhere else in the Old English corpus (Mitchell 1985b, 193).

§ 5.20    Once again, it is easy to see how the facilior reading might have arisen. While the fact that 1a appears without ƿue/ƿe in different recensions, in different dialects, and in manuscripts copied more than a century apart indicates that the pronounless construction was acceptable Old English, the lack of precise parallels elsewhere in the corpus, together with the fact that it has been replaced in all witnesses copied after the end of the tenth century, suggests that it was considered difficult or unusual by later scribes at the very least. As with the variation in 5b, moreover, Bede’s paraphrase may have had an effect on later scribal practice. The syntactic ambiguity of vernacular recensions without ƿue/ƿe is impossible to reproduce in Latin: by translating the first verb of the poem using explicit first person plural morphology, Bede may have influenced subsequent scribes to introduce a similarly explicit first person plural marker in their Old English, helping ensure the success of the “easier” form (see further, § 5.28).

An alternative hypothesis for the recensional development of Cædmon’s Hymn

§ 5.21    Taken together, the evidence of §§ 5.6-5.20 suggests that it is in fact not “easier to believe that on was added twice independently in 9 than that aelda was substituted for eordu twice in 5.” While nominal variation is, in unmarked contexts, less likely to be the result of identical independent substitution than variation in the use of pronouns or prepositions, the variants in 1a, 5b, and 9a of Cædmon’s Hymn are all strongly marked for relative difficulty. In each case, the variation in question is between an unusual or nonce formulation and a formulation found more frequently in the Anglo-Saxon poetic corpus. In 5b and 9a, the “less difficult” formulation belongs to a common formula or system; in 1a and 9a, the “less difficult” variants represent noticeably less complex syntax; in 1a and 5b, the “less difficult” variants are supported by the reading of Bede’s Latin paraphrase. In each case it is relatively easy to see how a redactor, faced with a version of the text showing any of these more difficult variants, might be led into substituting the easier reading—and next to impossible to see how scribes facing the easier reading could be led independently into substituting the less commonly attested, non-formulaic, and hence “more difficult” form. Although the traditional view of the Hymn’s textual development depends on the assumption that the nominal variants in 5b used to identify the poem’s five main recensions are more significant than the pronominal and prepositional variation that appears to link recensions across the traditional eordu : aelda branches, there is in fact no reason why this must be true. Both the nominal variation in 5b and the pronominal and prepositional variation in 1a and 9a have equal claim to our attention as possible recensional variants.

§ 5.22    This is significant because it suggests a way in which the apparently contradictory patterns of variation and contamination separating the five main recensions can be resolved. If we assume for the sake of argument that Cædmon’s original text contained all three difficilior readings, (i.e. Nu scylun hergan, eordu barnum, and firum foldu) and that all three facilior alternatives arose in the course of transmission, then it is possible to construct a model of the poem’s transmission that avoids the assumption that various branches of the poem’s text could be transmitted for centuries without developing any supporting features other than their eponymous variants:

Figure 3: New stemma

Proposal for a stemma in which the West-Saxon eorðan recension is assumed to represent the closest surviving text to the original written text of Cædmon’s Hymn.

In this view, the West-Saxon eorðan recension, the only version to show all three difficilior readings in its earliest witnesses, is assumed to be closest to Cædmon’s original song, despite the relatively late date of its surviving witnesses and the fact that its text was presumably translated from Northumbrian in preparation for its incorporation into the Historia’s Old English translation. All other texts of the poem show one or more facilior readings, and, as a result, are assumed to represent secondary developments in the poem’s text. In the case of the Northumbrian aelda recension, this development involves the introduction of the Psalm-like tag, aelda barnum, for what was presumably Caedmon’s original and far more unusual eordu barnum; in the case of the Northumbrian eordu, and West-Saxon eorðe Recensions, the development involves the introduction of ƿue/ƿe to 1a and on to 9a; in the case of the West-Saxon ylda recension, it involves the adoption all three facilior readings: ƿue/ƿe and on in 1a and 9a, and ylda, the West Saxon cognate of aelda, in 5b.

§ 5.23    There are, of course, a number of obvious problems with this view. The first involves the date, provenance, and dialect of the witnesses to the recension it assumes most closely represents the original text of Cædmon’s poem. Cædmon was, if Bede’s account is to be believed, a seventh-century Northumbrian. The West-Saxon eorðan recension is therefore in the wrong dialect, appears in manuscripts copied in the wrong part of England, and, in as much as its earliest witness was not copied before the beginning of the tenth century, enters the surviving manuscript record more than two centuries after Cædmon’s most likely floruit. In selecting this version, moreover, the alternative hypothesis appears to ignore the evidence of the two earliest witnesses to Cædmon’s Hymn, M and P, both of which are in Northumbrian manuscripts copied, at the very latest, before the end of the eighth century and, in at least one case, arguably in Bede’s own monastery (see below, Note B: Cædmon’s Hymn and Bede’s paraphrase).

§ 5.24    A second problem with the alternative model is that it appears to ignore the evidence of Bede’s own paraphrase, which seems to have been based on a vernacular text showing the facilior readings in 1a (debemus), 5b (filiis hominum), and, perhaps, 9a (dehinc terram Custos humani generis omnipotens creauit).[25] While it is relatively easy to see how a scribe, working in haste or reading formulaically in the manner proposed in O’Keeffe 1990 might be led into trivialising difficult readings from his exemplar by introducing syntactically less complex or formulaically more common collocations, it is more difficult to see how a translator of Bede’s learning and literacy could make similarly inadvertent substitutions in the production of a Latin version of Cædmon’s text in his careful account of the poet’s life.

§ 5.25    A third problem with the proposed model is that it does not, at first glance at least, appear to solve the problem of conflicting distribution of recensional variation across the different textual traditions. If the traditional view of the Hymn’s development requires us to assume that the variants in 1a and 9a coincidentally and identically entered both textual branches, the alternative view appears to solve this problem by requiring us to believe that the nominal variation in 5b did something similar instead.

§ 5.26    None of these objections is as serious as it sounds, however. The dates of the surviving witnesses to the West-Saxon eorðan recension are less important than the date of the source from which the translator of the Old English Historia acquired his text. As Plummer, Whitelock, and Colgrave have demonstrated, this source appears to have been a very high quality, and hence apparently quite early, version of the Historia’s c-text: a “good and pure” manuscript that avoids the “unauthorised additions of later MSS.” characteristic of its tradition (Plummer [1896] 1969, 2: cxxix; cf. Colgrave and Mynors 1969, xliii, n. 2; Whitelock 1962, 70 and n. 123). It appears at times, indeed, to have been superior to even the best of the tradition’s surviving witnesses: the late eighth-century manuscripts London, British Library, Cotton Tiberius C. ii (Tib. C. ii) and Kassel, Landesbibliothek, 4to MS. theol. 2, and the eleventh-century H, a manuscript showing a close connection to Tib. C. ii (see Whitelock 1962, 70, n. 123). If the original translator of the Old English Bede found his copy of the Hymn in the same authoritative manuscript from which he acquired the text of the Historia as a whole, then the presumably Northumbrian text upon which the recension is based may have been copied within a few years of the eighth-century M and P, the earliest known witnesses to the Historia and the vernacular Hymn.

§ 5.27    The possibility that the translator of the Old English Bede acquired his text of the Hymn from a very high quality and presumably early Northumbrian manuscript of the Latin Historia also helps explain how this relatively late text could have preserved all three difficilior readings. As O’Donnell 1996a demonstrated, the process by which scribes copied vernacular texts as glosses or translations in predominately Latin manuscripts encouraged the introduction of minor, Latin influenced, modifications (see esp. O’Donnell 1996a, 56-57). While most variants found among witnesses to Old English poems copied as glosses or translations can be ascribed to mechanical error or insignificant differences in dialect or orthography, a sizeable minority are caused by the sporadic influence of the accompanying Latin text. In the case of the common text of the Paris and Eadwine Psalters (Paris, Bibliothèque Nationale, lat. 8824 [PPs]; Cambridge, Trinity College, R. 17. 1 [EPs]), for example, at least six and probably eight of the twenty-one potentially significant substantive variants separating the two witnesses can be attributed to the influence of the accompanying Latin text (O’Donnell 1996a, 36-49; Baker 1984, 265). In several cases this influence involves the addition of supernumerary words from the Latin original (e.g. Psalms 90:16.3b, 91:1.1b, and 92:7.1a). In others, it concerns changes in inflection, syntax, or wording apparently prompted by the underlying Latin (e.g. Psalms 93:8.1a [change in mood]; 91:8.1a [a possibly Latin-influenced variation between a pronoun and adverb]). Similar examples can be found in Psalm 102:4.1a (found in PPs and Oxford, Bodleian Library, Junius 121 [Jun 121]), and Gloria I, 13b (Jun 121 and Cambridge, Corpus Christi College, 201). In several of these cases, the influenced material is nonsensical or results in poor metre; in others, however, the readings provide reasonably sound, syntactically and metrically appropriate equivalents.[26]

§ 5.28    What this suggests is that the translator of the West-Saxon eorðan tradition may have acquired his text of the Hymn from a manuscript copied before the Hymn developed much internal variation or began to show any strong influence from the surrounding Latin. As the following table demonstrates, indeed, there is a strong correlation between a preference for facilior readings 1a, 5b, and 9a and transmission in manuscripts of the Latin text of Historia ecclesiastica (in the following table, facilior readings are reproduced in reverse field, difficilior in normal type. Grey is used for a reading in the paraphrase not precisely paralleled in the Old English text):

Table 3: Correlation between manuscript context and preference for facilior and difficilior readings

Recensions Copied in the Latin Historia ecclesiasticaLatin Historia ecclesiasticaRecension Copied in the Old English Translation of the Historia ecclesiastica
Northumbrian aelda recensionNorthumbrian eordu recensionWest-Saxon eorðe recensionWest-Saxon ylda recension West-Saxon eorðan recension
scylunƿue sciulunƿe sceolonƿe sculondebemussculon
aelda barnumeordu bearnumeorðe bearnumylda bearnumfiliis hominumeorðan bearnum
folduon folduon foldenon foldumterramfoldan

The only recension of the Hymn to show all three difficilior readings is the only version of the text characteristically not found in manuscripts of the Latin Historia ecclesiastica. All other recensions of the text are characterised by their location and their use of one or more facilior readings: aelda in 5b, and ƿue/ƿe and on in 1a and 9a.

§ 5.29    This correlation suggests in turn the solution to the second problem with the alternative stemma proposed above. Rather than ignoring the evidence of Bede’s Latin translation of the Hymn, this account of the evidence of the surviving recensions suggests instead that Bede’s paraphrase may itself be responsible for the introduction of at least one, and perhaps two or three, of the three facilior forms found in the Northumbrian aelda, Northumbrian eordu, West-Saxon ylda, and West-Saxon eorðe recensions: debemus (i.e. ƿue/ƿe scylun), and filiis hominum (i.e. aelda barnum).

§ 5.30    This is not as unlikely as it sounds. As several scholars have noted, albeit to different ends, Bede’s paraphrase of Cædmon’s Hymn is not always very close to the surviving vernacular text (see among others, Orchard 1996; Kiernan 1990a; Schwab 1984). While his version remains quite close to the surviving Old English recensions for the equivalent of most of the poem’s first five lines, the translation strays considerably from these texts in the equivalent to last four. Eight of the seventeen Old English words in the last four lines of the vernacular Hymn have no equivalent in Bede’s Latin paraphrase, which also adds a word not found in any Old English version, culmine, “gable,” apparently for its Latin metrical qualities (see Orchard 1996, 413). Even when the two texts are more closely parallel, moreover, they often differ in the precise language or constructions used. Apart from the use of Custos for uard in the equivalent to 7a, for example, Bede uses none of the usual Latin translations for Cædmon’s epithets for God: he translates dryctin as Deus in the equivalent to 4b (the usual Latin equivalent is dominus), ignores Cædmon’s use of frea in 9b, and translates the first appearance of uard in 1a with auctorem, a term that contains no sense of guardianship found in the vernacular text (see §§ 3.31-3.35, above). His translation of Cædmon’s syntactic constructions, likewise, is at times very loose. In the last four lines of the poem, Bede eliminates all trace of the heavy ornamental variation so characteristic of Cædmon’s original text. He translates the verbs of 5a and 8a (scop and tiadæ, respectively) using a single Latin form, creauit. He reproduces only one of Cædmon’s two temporal adverbs (dehinc versus tha and æfter, 7a and 8b). He ignores the second reference to men as the beneficiaries of his creation (firum) in 8a (see above, §§ 2.40-2.41), and collapses the two references to the earth in the poem’s closing lines (middungeard, 7a, and foldu, 9a) into a single Latin form, terram. Even in the first five lines of Cædmon’s text, where Bede does follow the text of the vernacular poem more closely, he nevertheless still permits himself to alter the syntax of 3b-4b, changing the appositive reference to eci dryctin in the clause sue he uundra gihuaes, / eci dryctin, or astelidæ into an adverbial clause in its own right: quomodo ille, cum sit aeternus Deus, omnium miraculorum auctor extitit.

§ 5.31    These changes suggest that Bede was not overly concerned with reproducing the precise wording of Cædmon’s text in his Latin paraphrase. Given his willingness to change the syntax of even relatively straightforward passages in the original Old English, it requires no stretch of the imagination to assume that he would not necessarily have sought a formulation that reproduced the syntactic ambiguity inherent in the difficilior reading of Cædmon’s opening line.[27] In the case of his translation of 5b, Bede may also have been influenced by a general similarity between Cædmon’s poem and various Psalms and Canticles of the Anglo-Saxon liturgy (see §§ 2.37-2.45, above). While the difficilior reading for the passage, eordu barnum, is attested nowhere else in the Old English corpus, the facilior reading, aelda barnum, is regularly found as a translation of the common Christian Latin liturgical tag, filiis hominum. If Bede did not make the substitution himself in the process of translating the poem from Old English to Latin, the variation may represent a subtle and learned improvement to the Old English text originating in the course of the poem’s transmission from Whitby to Jarrow.

§ 5.32    The close correlation between facilior readings in 1a, 5b, and 9a and the appearance of the vernacular Hymn in the margins of Bede’s Latin Historia ecclesiastica suggests an answer to the third objection to the alternative stemma proposed above: namely that the proposal still requires us to assume that one or more recensions of the poem introduced identical textual variation at one or more points coincidentally. In the case of the variation in 1a and 9a, the appearance of correlated variants in most surviving witnesses to the poem suggests that the three recensions showing these variants may be genetically related—or in other words, that the Northumbrian eordu, West-Saxon eorðe, and West-Saxon ylda recensions may stem at some point from a single common exemplar in which ƿue/ƿe was added to 1a and on to 9a (cf. Cavill 2000, esp. 526-529). In the case of 5b, the appearance of the facilior reading aelda / ylda can be attributed to the coincidental influence of Bede’s Latin translation: while the reading of the Northumbrian aelda recension may reflect the vernacular text known to Bede, the adoption of ylda in later West-Saxon manuscripts can be explained plausibly as a result of the influence of Bede’s paraphrase on the scribe of the recensional archetype in a manner similar to that found in copies of the Old English metrical Psalms (see § 5.27, above).

Conclusion: The transmission of Cædmon’s Hymn

§ 5.33    If these arguments are correct then the recensional development of Cædmon’s Hymn can be summarised as a competition between three, for the most part quite similar, versions of the original poem. The first branch, showing the difficilior readings in 1a, 5b, and 9a and presumably reflecting a text very close to Cædmon’s original song, was copied in the apparently early, highly authoritative (but now lost) manuscript from which the translator of the Old English Historia acquired his Latin text. It is most fully represented only in the West-Saxon eorðan recension. This version of the poem, while technically accomplished and highly formulaic (see above, Chapter 3: Cædmon’s Hymn and Germanic convention), nevertheless contained a number of unusual constructions and rare or perhaps even unique collocations. As the evidence of the three surviving recensions with a cognate to eordu in 5b attests, moreover, this form of the poem was apparently widely known in Anglo-Saxon England at the time the first manuscripts of Bede’s Historia ecclesiastica were copied and disseminated.

§ 5.34    Very early on in the Hymn’s transmission history, however, a second, perhaps “learned,” variant of the poem, in which Cædmon’s original eordu barnum was altered to the more common (and liturgical) formula aelda barnum, appears to have developed. This is the version of the text translated by Bede’s paraphrase and found in surviving witnesses to the Northumbrian aelda recension. It is impossible to say whether Bede introduced the variant himself (and hence influenced the subsequent Old English glossators) or translated a vernacular text in which eordu bearnum had already been replaced by the more liturgical phrase. Either way, this recension of the text appears, at first at least, to have enjoyed a relatively limited circulation: its key variant does not show up in three of the four other recensions; its appearance in the West-Saxon ylda recension may have been influenced directly by its appearance in the Latin paraphrase.

§ 5.35    The Hymn’s third tradition also developed relatively early on in the poem’s transmission history. This tradition was characterised by the correlated addition (probably trivialisations) of ƿue/ƿe and on in 1a and 9a and, initially at least, the preservation of the difficilior reading eordu in 5b. This was a very widely disseminated tradition: it is the version responsible for the Northumbrian eordu, West-Saxon eorðe and West-Saxon ylda texts. Since both West-Saxon recensions in this group show evidence to suggest that they are the result of independent translations from different Northumbrian exemplars (see §§ 6.52-6.54), it also appears to have been a relatively early addition to the text.

§ 5.36    This leaves, finally, the nominal substitution in 5b of the West-Saxon ylda recension. Although the appearance of the correlated variants ƿue/ƿe and on in 1a and 9a suggest that this text is related to the tradition responsible for the Northumbrian eordu and West-Saxon eorðe recensions, its agreement with the Northumbrian aelda in 5b suggests some influence of the Jarrow tradition. There are two possible explanations for this: the first is that the scribe of the exemplar to this tradition knew something of the Northumbrian aelda tradition; the second, perhaps more likely given the text’s many serious errors, is that the scribe was consciously or unconsciously influenced directly from Bede’s Latin.

§ 5.37    When this suggested history is applied to the stemma proposed in § 5.22 above, the following manuscript history is produced:

Figure 4: The transmission of Cædmon’s Hymn

A detailed stemma showing the transmission of Cædmon’s Hymn as outlined in Figure 3


[1]Variants are difficult to count. The figure “thirteen” represents the number of potentially significant contrasts among the surviving recensions. Thus the variation in 1a, Nu : Nu ƿe : Nu ƿue : Nu : Nu ƿe, shows one potentially significant “variant”: the addition or omission of the first person plural nominative pronoun. For the purposes of this chapter, references are to the edited text (before emendation except in the case of Northumbrian eordu recension) of the recension indicated; if no recension is indicated, references are from the edited text of the Hymn’s assumed archetype. Accidental differences among witnesses to a given recension are ignored unless they are the focus of discussion; substantive differences among witnesses are noted where relevant. Definitions for “significant,” “substantive,” and “accidental” as they are used in this edition are given below, §§ 7.6-7.9; some individual recensions present problems that affect these definitions; these are discussed in the editorial introductions to the individual recensions (see Chapter 7: Editorial introductions, below).

[2]Certain late witnesses to this recension show the West-Saxon ylda reading in this position.

[3]Not all witnesses to this recension show this reading.

[4]Certain late witnesses to this recension show the West-Saxon ylda reading in this position.

[5]Certain late witnesses to this recension show the West-Saxon ylda reading in this position.

[6]Not all witnesses to this recension show this reading.

[7]gescop also appears in two later witnesses to the West-Saxon eorðan recension (O and Ca). See below, § 7.67.

[8]The reading of the West-Saxon eorðan recension (eorðan) and especially the Northumbrian eordu recension (eordu) can be construed as either accusative or genitive singular. See below, § 5.16 and note. The West-Saxon eorðe recension reading eorðe is, formally speaking, nominative singular, a case which makes little sense in context.

[9]This is the assumption behind the exchange between Wülker and Zupitza (Wülker 1876; Zupitza 1878; Wülker 1885).

[10]The significance of the discovery of the Northumbrian eordu recension is often underestimated, particularly in debates about the priority of aelda and eordu in 5b, or whether the Old English poem might be a back translation from Bede’s paraphrase. Before 1906, the West-Saxon eorðan recension was understood almost universally to be a late West-Saxon corruption of the original Northumbrian aelda recension (see, for example, the exchange between Wülker and Zupitza in Wülker 1876; Zupitza 1878; Wülker 1885). By showing that a version of the poem containing eordu had also circulated in Northumbria, Wuest demonstrated that the West-Saxon eorðan recension in fact belonged to a much older (and indeed more dominant) tradition than had been suspected previously.

[11]Later witnesses to the West-Saxon eorðan recension show more differences with the Northumbrian aelda recension, and, in some cases, a greater affinity with the West-Saxon ylda recension. See § 7.67, below.

[12]Some late witnesses to the West-Saxon eorðan recension also add ƿe in 1a, see below, § 7.67.

[13]This list excludes O, which adds by correction ƿe and a number of other readings apparently influenced by an ylda-type text.

[14]The significance of this correlation is obscured in the stemma in Cavill 2000, 528, fig. 5. Although Cavill’s figure implies that the “We-pronoun” entered the textual tradition first, and that correlation with on developed later, most evidence suggests the opposite is true. With the exception of B1, To, O (after correction), and O’s descendent, Ca, all surviving manuscripts of Cædmon’s Hymn have either ƿue/ƿe and on or neither form. All four exceptions are s. xiin-s. xii copies of the West-Saxon eorðan recension. As Cavill’s figure correctly suggests, there is no evidence to suggest that the presence of ƿe in these manuscripts reflects the early form of their tradition. The two earliest witnesses to the recension, T1 (s. x1) and C (s. xmed), omit pronoun and preposition from 1a and 9a. In O (s. xiin), the pronoun is introduced by correction along with a number of other changes that seem to be designed to bring the text into closer alignment with the West-Saxon ylda recension. Ca appears to be a direct descendent of O and adopts most of its corrections: it reads ƿe because its exemplar does. Only in the case of B1 and To (dated to s. xi1 and s. xii, respectively) is there any evidence of the pronoun entering the textual record spontaneously without a corresponding on in 9a. Although Cavill points to Bede’s paraphrase as evidence that ƿue/ƿe might have entered the Old English textual record first, this is far from conclusive: since Latin requires an explicit indication of person on plural verbs, Bede could not reproduce the syntactic ambiguity of the pronounless Old English reading even if his source text omitted the pronoun; in the case of 9a, his translation recasts the Old English syntax. There is no way of telling what Bede’s source read in these lines. See also Cavill 2002, esp. 6-9, where it is suggested that “the difference between mnauscripts that have we and those that do not is one of date, not of manuscript grouping.” While the presence or absence of ƿue/ƿe may be related to date—especially if one accepts Cavill’s argument for a late date for the original insular examplar responsible for Br, P1, and Di (see Cavill 2000)—the evidence of witnesses to the West-Saxon eorðan recension also demonstrates a strong correlation with “manuscript grouping.” See also below, §§ 5.21-5.37.

[15]For the possibility that the variants in 1a and 9a may be the result of coincidental variation: see Smith 1978, 3-4; Dobbie 1937, 46. The only scholar to specifically address the question of whether the nouns in 5b might also be coincidental appears to be Orton 1998, 157. Cavill 2000 has recently drawn new attention to the importance of the variants in 1a and 9a; his account obscures several key points, however. See above, § 5.9 and note.

[16]The ChronG text of the poem, to the extent that this can be recovered from early modern transcriptions, appears to incorporate this correction into the main text: 49b is reconstructed as the nonsensical cumbelgehnades in both Campbell 1938 and Lutz 1981.

[17]Examples of unambiguously incorrect variation in the use of prepositions and pronouns are more rare in the corpus, but do exist. In the metrical translation of Psalm 93:7.1b, the mid-twelfth-century scribe of Cambridge, Trinity College, R. 17. 1 [EPs] appears to misinterpret an original ge- prefix as a second person plural pronoun, reading ge ne sæwe for expected ne gesæwe as in the mid-eleventh-century Paris, Bibliothèque Nationale, lat. 8824 (“The Paris Psalter”) (PPs) (O’Donnell 1996a, 48-49; Baker 1984, 281). Likewise, the Vercelli text of Soul and Body has a number of examples of the addition of superfluous—or the misinterpretation or omission of syntactically necessary—pronouns: e.g. 57b, 117b, and 123b (O’Donnell 1996a, 312, 355 and 351-352). That pronouns and prepositions were slightly easier to vary coincidentally, however, is suggested both by the number of substitutions of roughly equivalent forms in the corpus of multiply attested poems, and by examples of successful corrections, such as that in Brunanburh, 35a, where the ChronG scribe correctly repairs the nonsense form cnearen flot in ChronA to the expected cnear on flot (see O’Donnell 1996a, 173; Lutz 1981, 85).

[18]uerc is found in the same line with uuldurfadur on only one other occasion (mid wuldorfæder weorca to leane, Menologium, 147). It is collocated in the same half-line with a compound beginning with wuldor- or wundor- on only two other occasions, neither of which is morphologically, metrically, or syntactically parallel to 3a: wuldorlean weorca (Christ C, 1079a) and weorca wuldorlean (Guthlac B, 1373a).

[19]This is only one of a number of archaic or otherwise unusual forms in the poem; see Campbell 1938, esp. 41; the difficulty these forms cause the scribe of ChronA are discussed in O’Donnell 1996a, 150-151; also O’Keeffe 1990, 118-121; Bately 1986, xciii.

[20]For formulaic reading, see O’Keeffe 1990; certain problems with O’Keeffe’s analysis of Cædmon’s Hymn are discussed below: Note D: Memorial performance, transitional literacy, and the transmission of Cædmon’s Hymn.

[21]eordu is ambiguous in context: another possible translation is “He first made the earth for children.”

[22]The West-Saxon ylda recension and West-Saxon eorðe recension show a number of corruptions unrelated to the variation in 9a.

[23]The Northumbrian aelda recension reading foldu and Northumbrian eordu recension form eordu in Cædmon’s Hymn, 9a and 5b, are usually described on philological grounds as accusative singular (Campbell 1991, §§ 472, 617, and esp. 331.6; Hogg 1992, §§ 6.56, 7.98, and esp. 3.34; Smith 1978, 35, § 18 note). In the case specifically of eordu, there is considerable evidence of the use of -<u> in oblique cases, including the genitive singular, in early Northumbrian (see Ross 1937, 64-65; Brunner and Sievers 1965, § 276 Anm. 6; cf. Smith 1978, 35, § 18 note). There appears to be less evidence for the operation of a similar analogy in the case of other feminine n-stem nouns.

[24]The West-Saxon eorðan recension shows a nearly identical text at this point.

[25]The evidence of Bede’s paraphrase is more ambiguous in the case of 1a and 9a than 5b. Since Latin syntax requires the explicit indication of person on present plural indicative verbs, Bede’s use of debemus does not necessarily indicate he was working from an original reading ƿue/ƿe in 1a—although it does indicate that he understood the subject of scylun to be first person rather than third person plural (i.e. ƿue/ƿe rather than uerc). His equivalent to 9a recasts the syntax of the Old English, changing the dative plural indirect object firum to a genitive singular humani generis, complement of Custos. See also § 2.40, above.

[26]These variants are discussed in greater detail in O’Donnell 1996a, chapter 2; the Latin influence in witnesses to the translation of the Psalms is discussed in Baker 1984, 265.

[27]See Cavill 2002, 14, for further examples of this emphasis on unambiguous syntax in Bede’s paraphrase.

Another possibility in the case of 1a, of course, is that Bede’s paraphrase is simply evidence that there was no ambiguity about the subject of scylun for contemporary speakers of Old English. See above, § 5.19.