Diana Whaley 2012, ‘Editorial methodology’ in Diana Whaley (ed.), Poetry from the Kings’ Sagas 1: From Mythical Times to c. 1035. Skaldic Poetry of the Scandinavian Middle Ages 1. Turnhout: Brepols, pp. xxx-xxxv.
The ultimate goal of the present edition is to present and interpret texts which approximate as closely as possible to the poetry of the skalds in the form they originally delivered it. Although the oral genesis of much skaldic poetry and the existence in some cases of significantly divergent variants might lead one to question the notion of recovering a single, fixed text even as an ideal, much of the corpus, and the respect accorded to it by medieval prose authors and scribes, suggests a relatively stable tradition in which such an ideal is not inappropriate. The present edition, therefore, seeks to recover the skald’s original, but in doing so it recognizes that the manuscripts are all we have, and that in practice it is the archetype or earliest written version that we aim to reconstruct. We cannot reach beyond that to the skald’s original without entering the realm of conjecture, as exemplified on a grand scale, for instance, by Guðbrandur Vigfússon (CPB II, 28 et passim) in his theory that much of the court poetry had been transformed by twelfth-century revisers who replaced place-names and other factual detail by vapid praise. As in other aspects of the editorial project, notably the division of the corpus (§1.4 above) and the reconstruction of poetic structures (§3.1 below), respect for the manuscripts has been at the heart of our approach, and an essentially conservative editorial stance has been adopted.
The first step in the editorial process, then, has been to identify the relevant manuscripts, whether vellum or paper, medieval or modern, and here and elsewhere in the discussion ‘manuscripts’ includes other types of witness: runic artefacts with metrical texts, and early printed editions of key manuscripts now lost. The rich diversity of the skaldic corpus is accompanied by a wide range of textual situations. The simplest case, if not necessarily the easiest editorially, is that of a complete, continuous poem preserved in a single medieval manuscript, but this is a rarity, and the majority case is of poetic fragments cited within multiple manuscripts of prose (or strictly prosimetrum) works, narrative or pedagogical. Most of the manuscripts are not autograph texts, but rather copies (often at several removes) of the prose works in question and separated by several centuries from the putative date of composition of the poetry.
Normally all manuscripts with independent texts (i.e. texts that are not copies of extant manuscripts) have been consulted and are presented in the edition, but where a very large number of such manuscripts exist (e.g. over seventy for SólarljóðVII), principled selectivity in the presentation has been necessary. Later copies of manuscripts still extant have not been used, unless believed to be of particular value, for instance because they supply readings in a badly damaged exemplar, or contain readings from an additional source. In cases where various copies of a lost exemplar exist, some selectivity has often been necessary, editors routinely citing the best copies but adding readings from subsidiary copies in cases of particular textual interest or difficulty.
Once the relevant manuscript texts for the poetry in question have been located (and here the project has benefited greatly from access to Jón Helgason’s card index (see §2.2 above)), editors seek to become acquainted with their orthography, dating and probable relationships, and to collate their readings for each skaldic stanza. Through the generous cooperation of numerous institutes and libraries, but above all of the sister Arnamagnæan Institutes in Reykjavík and Copenhagen, it has been possible for editors to examine manuscripts ‘in the flesh’, or using high-quality photographs or the thousands of digital images that are available on the project’s website.
The editing of each stanza is based on the complete paradosis, i.e. the textual tradition comprised of the full range of significant readings. Out of the available manuscripts a main or base manuscript has been identified and used as the basis for the text presented in the edition, always subject to correction from the other manuscripts where they appear to have a more original reading. The main manuscript is chosen, and the remaining manuscripts grouped and ordered, primarily on the basis of a stemma, where it has been possible to create one (see further below), and the stemma is a key factor in assigning weight to the evidence of particular manuscripts or combinations of manuscripts. For example, in the case of a two-branched stemma a variant reading preserved not in the main manuscript but in a sister manuscript within the same branch and in reliable manuscripts belonging to the other branch will normally be preferred to a reading in the main manuscript alone. In many cases the choice of main manuscript is the same as that adopted in Skj, but in all cases the evidence has been reassessed, partly in the light of subsequent scholarly views, for instance the widely accepted view that Finnur Jónsson went too far in privileging certain redactions (notably Heimskringla) or certain manuscripts (e.g. the Kˣ transcript of Heimskringla or the R manuscript of Snorra Edda).
In the majority case of skaldic poetry preserved in prose works, the stemmata of those works provide preliminary guidance as to the stemma for the skaldic texts, with the proviso that some compilations such as Flateyjarbók depend on a range of sources so that their component texts belong to a variety of different stemmata. Stemmata produced, mainly on the principle of patterns of common error, by the editors of reliable critical editions have therefore been invaluable, and the advice of scholars currently working on major editions has also been sought and gratefully received. However, stemmata – as their makers invariably point out – are hypothetical representations of relationships between texts, which often contain uncertainties and missing links. Further, it cannot be automatically assumed that the stemmata for prose works and the poetry they cite will correspond exactly. Medieval or early modern scribes may have had access to poetic texts, oral or written, other than those embedded in the prose works they were copying, added to which poetic texts may have been corrupted by scribes who did not fully understand how skaldic poetry works, or ‘improved’ by scribes who did (and who are in effect the first traceable editors of the poetry), and these things have been borne in mind in the analysis of the readings. Other, non-stemmatic, factors also enter into the processes of selecting the main manuscript and determining priorities among the other manuscripts, especially the age and physical condition of manuscripts, and the overall quantity and quality of the skaldic text preserved there (quality reflecting both the accuracy of the scribe and the reliability of his exemplar).
For some of the corpus (including much of the poetry of the kings’ sagas edited in SkP I and II), preservation is within manuscripts representing more than one prose work, and the same principles for prioritisation apply there, with the added difficulty that even where individual stemmata can be constructed, an overarching stemma suggesting textual priorities between prose works is extremely elusive.
Moving to the editing of individual stanzas, questions there focus on prioritisation of variant readings (where variation exists), and the interpretation of the text. Here again the stemma is indispensable, but a stemma is in effect a working hypothesis which cannot be operated mechanically in order to reconstruct the archetype of the stanza, both for the reasons already mentioned and because rival readings may have approximately equal support from the stemma, and a wide range of other considerations therefore comes into play. Fundamental are linguistic considerations, and in general it would appear reasonable to assume general conformity or comparability to known phenomena of early Norse-Icelandic orthography, phonology, grammar, syntax, lexis, word-formation, and semantics, with the caveat that our knowledge of these phenomena is incomplete and imperfect, especially for the period of the earliest poetry, so that it can be difficult to distinguish between corruptions and otherwise unrecorded usages. In cases of uncertainty comparanda are often suggestive. At the lexical level, for instance, factors in favour of postulating otherwise unrecorded words include: compounding from known elements according to known patterns; the existence of cognate or related words in Old Norse-Icelandic or other Germanic languages; a linguistic formation that has parallels elsewhere. Similar considerations can be applied in the case of grammatical features such as nouns with unparalleled plural forms or alternative genders. A further general problem is that, while we know that skaldic usage differs somewhat from that of prose (e.g., at the syntactic level, in the more frequent occurrence of subjectless verbs, lack of number concord between nouns/pronouns and verbs, or use of plural nouns with singular meaning), we do not know exactly where the boundaries of acceptable difference lie. Again, comparative evidence is of especial use, particularly from the poetry of the Poetic Edda and the skaldic corpus itself, in earlier editions and (increasingly, as the editorial project has progressed) the present one. It should be mentioned here that some of the classic – and still indispensable – resources for skaldic study are highly interdependent, and share a quite heavily interventionist editorial approach, with the result that some authentic words, by-forms and usages may have been lost by emendation; this applies especially to Skj B, Meissner and LP.
Linguistic considerations shade into stylistic ones, and again a major general difficulty is distinguishing between the unusual and the impossible. That skaldic clauses can be fragmented and their arrangement in the half-stanza or helmingr convoluted is beyond dispute (e.g. Reichardt 1928; Edwards 1983), but how extreme this dislocation can be is contested, as famously illustrated by E. A. Kock’s objections in his NN to many of the editorial decisions embedded in Skj. Other phenomena certainly occur but are rather rare, so that in encountering possible examples of these where the paradosis is not unanimous editors may be faced with difficult decisions. Examples include tautology of various sorts (e.g. repetition of the same concept within a single noun phrase), apposition (e.g. parallel subjects to the same verb), and half-kennings (where the base-word of a familiar kenning-type is unaccompanied by a determinant). The kenning, as the leading feature of skaldic poetry, is frequently a cause for editorial hesitation, since alongside hundreds of unproblematic examples there are many which do not entirely conform to the patterns described in Snorra Edda or Meissner, prompting the question of whether a text is corrupt or the expectations raised by Snorri, Meissner and others too normative (see §5.1.2 below). Again, this contested issue features repeatedly in Kock’s comments on Skj, since he will frequently reject emendations designed to ‘correct’ kennings, envisaging instead unusual kenning-types or a looser, often adjectival, use of genitives in noun phrases. The possibility of a more widespread adverbial use of genitives than hitherto recognized has been suggested in recent scholarship (e.g. Poole 2004). In all these cases and others of a similar sort, contributors to this edition have made case-by-case decisions in the light of the paradosis and of relevant parallels. In doing so they may also invoke the textual-critical concept of the lectio difficilior ‘more difficult reading’ and lectio facilior ‘easier reading’, which predicts that if corruption occurs it is more likely to replace a rare linguistic or stylistic feature by an obvious one than the reverse, so that if the totality of the evidence permits, an editor may favour the lectio difficilior. In postulating corruption, editors may also invoke commonly recognised, cross-cultural, types of scribal error, such as dittography, haplography, misreading or miswriting of certain graphs or abbreviations, or anticipation or repetition of adjacent words.
The intricate metrical patterns of syllable-counting, alliteration and internal rhyme that characterise regular dróttkvætt and kindred metres (described in, e.g., Sievers 1878; Kuhn 1983; Gade 1995a) are considered to have acted as a mnemonic and a partial protection against corruption, especially during the period of oral transmission, and they, together with the recurrent patterns of syntactic filler which occupy particular metrical positions, are another important aid to establishing the text where the overall metrical practice of a skald or poem is regular.
Editorial decisions most often involve arbitrating between manuscript readings, but emendation, here defined narrowly as the adoption into the text of a reading which appears in no manuscript, is available as a last resort – an unwilling step into the unknown undertaken only where the manuscript readings are not viable, where a possible route of corruption can be surmised, and where the specific emendation proposed can be justified. The present edition adopts an essentially conservative editorial approach to the text, and this is one of the features that differentiate it most strongly from Skj. This means that there is at least a predisposition to considering the merits of unfamiliar linguistic or stylistic usages rather than emending them away, and similarly emendations are not normally made for metrical reasons if they go against the evidence of all manuscripts. Something of a special case is that of poetry preserved in only one manuscript, especially if it is one such as Flateyjarbók whose skaldic texts, where others are available for comparison, often compare poorly. In these cases emendation may be slightly more readily considered, though it still needs careful justification.
The production of a single critical text which aspires as far as possible to recapture the archetype and through that the skald’s own words is the only practicable option in an edition of so large a corpus as the skaldic one. Unavoidably, it runs counter to the ‘new philological’ approach which foregrounds individual manuscripts and the reception history of texts, yet measures have been adopted which go some way to promoting new philological values: the ‘Readings’ section for each stanza gives all significant (non-orthographic) variants in the manuscripts listed so that the text of each can be reconstructed, and in the extreme situation of a stanza preserved in manuscript traditions so discrepant that they cannot be reconciled or conflated, the full text of both versions is printed, one as the main text and the other in the Notes. Further, the electronic version of the edition contains a digital archive of manuscript images so that for almost every stanza the relevant folio(s) can be easily accessed; in most cases a transcription of at least the base manuscript is also available.
At the level of editorial interpretation, too, the edition seeks as far as is practicable to represent the range of possibilities. Even where no variant readings are present, a skaldic text often presents a multiplicity of interpretative options, and the solution printed in the Text is that judged by the editor to be the best, but in many instances it is the best by only a narrow margin, and the inclusion of explanatory Notes enables the main alternatives to be outlined. Since many or most of the textual problems have already been addressed in print, such Notes give present-day editors an opportunity to acknowledge their deep and continuous indebtedness to those of previous generations.