by Dr. John Tucker

Margaret and Richard Beck Lectures
Icelandic Symposium
University of Victoria
November 20, 2004

When we began to plan this symposium, it struck me as an opportunity to return to a topic that I had investigated some years ago and then set aside, namely the matter of chapter divisions in the sagas. Since this is intended to be a relatively informal exchange of ideas relating to work in progress, the mix of scholars present encouraged the choice of a topic in which literature and language overlap. The study of saga chapters relates to the attempt to develop a grammar of saga narrative. Just as the study of language requires the analysis of continuous utterances into their constitutive parts (sentences, phrases, words, morphemes, phonemes, etc.), so the study of storytelling requires the analysis of whole stories into the units out of which they are constructed. Furthermore in the case of the sagas, isolating these building blocks should cast light on the irresolvable and abiding issue of saga scholarship: the conflicting claims of orality and literariness. So I am hoping that—starting from your various positions—you will have suggestions as to where I have gone wrong and how I ought to proceed with this inquiry.

A number of scholars have already attempted to identify the constitutive units of saga structures. The oldest of these attempts is þáttr theory, which postulates the piecemeal assembly of sagas out of smaller, pre-existing story units. Since a þáttr typically records an episode, it is storytelling conforming itself to remembered reality—arguably the form in which the oral record first expressed itself. More recently Jesse Byock, for one, borrowing from linguistics has argued for the feudeme. Others have advanced the importance of the scene.[1]

How are we to choose among these models? According to Jonathan Culler the only way to demonstrate the superiority of a theory of plot structure is to show that the descriptions of particular stories which it permits correspond with our intuitive sense of plot and that it is sufficiently precise to prohibit descriptions which are manifestly wrong.[2] But how reliable is our intuitive sense when it comes to narratives hundreds of years old? And what saga reading is manifestly wrong? The question then arises: Is there any direct, extra-textual evidence for what saga culture took to be the constitutive unit or units of saga narrative?

As far as I can see the only such evidence is the manuscript indications of chapter divisions. Yet these chapter divisions seem to have interested very few. The prevailing attitude was well summarized by Lee M. Hollander when he dismissed Norman Blake’s edition/translation of Jomsvikinga Saga in the following somewhat contradictory terms: The chapter divisions and headings of Icelandic saga manuscripts are notoriously inconsequential. Still, it is hardly permissible arbitrarily to cut up the text….[3]

No one seems to have taken Hollander up on his statement of notorious inconsequence, at least not directly. Ralph Allen cites the comment, but only to deny that it applies to the chapter divisions of Njáls saga. And Lars Lönnroth does not explicitly take issue with it when he notes Allenʼs citation in order to insist on the same exception.[4] But what are the grounds for treating the chapter usage in Njáls saga as exceptional? And considering the almost universal silence among scholars on the subject of chapter divisions, how can their inconsequence be notorious?

Yet I am sure that many modern readers, including scholars, have wondered at the logic of chapter divisions in the sagas. Even in printed editions, where the chapters tend to be combined and rationalized, the chaptering may strike us as strange or naive. At best chapter divisions may appear merely to emphasize the momentary pauses or transitions in the stories themselves, telling us nothing important or new. Nevertheless what can be said about chapter divisions is that they offer us the best clues we are ever likely to find as to the way in which contemporaries read the sagas. They record the informed understanding of a period that took the physical disposition of texts very seriously. For just when presently surviving sagas came to be composed and written down, there developed throughout Europe an intense interest in the technology of book-learning. As Malcolm Parkes has observed,

The late medieval book differs more from its early medieval predecessors than it does from the printed books of our own day. The scholarly apparatus which we take for granted—analytical table of contents, text disposed into books, chapters, and paragraphs, and accompanied by footnotes and index—originated in the application of the ideas of ordinatio (arranging) and compilatio (assembling) by the writers, scribes, and rubricators of the thirteenth, fourteenth, and fifteenth centuries.[5]

The most familiar and direct evidence of this new orientation is probably to be found in the books of the Bible, for though the Bible had been divided into chapters before the thirteenth century these division were unsystematic and local. Our present chapters are believed to descend from the Paris Bible and Stephen Langton, who died in 1228 and who recognized that scholarship would be facilitated by a single, universal standard. Biblical chapter divisions in any case deserve mention in this context because they provide a better model for the proper dimensions of saga chapters than do the chapter usages of modern novels.

Modern chapter breaks assume single readings, and a plethora of reading possibilities. But the sagas, like the Bible, were texts for rereading. As Lars Lönnroth reminds us, initials and rubrics seem to have functioned as a kind of signpost for readers leafing through the manuscript in search of some particular story or incident. Initials then are in the first instance a visual phenomenon. Their illumination or historiation arose out of mnemonic necessity, just as the highlining of a modern student does.

It is important to emphasize the visual character of chapter-marking initials, in order to remind ourselves that not only narrative imperatives but also mechanical factors and decorative considerations can affect their placement. If a scribe forgot to allow space for the later insertion of an initial in a different coloured ink, he might end up changing the location of a chapter opening. And the physical size and layout of the manuscript might affect the frequency of occurrence of decorative initials. But it would be a mistake to over-emphasize the physical or decorative aspect of initial-making, if by doing so we are tempted to dismiss them as simply the aping of a prestigious but alien technology, the product of scribes chiefly concerned that Icelandic books look real.

Whatever their source, the paradoxical effect of introducing divisions is, of course, to create unities. Chapters have their own dynamics, contained as they are between openings and closings. And openings and closings enjoy a rhetorical foregrounding which can be exploited in various ways. Take beginnings, for example. What are we to make of the tendency to open chapters by referring to rulers and reigns? Is it a borrowed historiographical dating convention or an act of flattery? For whom should the honour be reserved? for kings and earls? or for rulers who play important parts in the stories? And if their introductions involve the repeated use of the same formula, should a translator reproduce or mask the repetition?

That these questions should arise depends on the fact that the introduction of chapter divisions, in Roman Jakobsons oft-cited phrase, projects the principle of equivalence from the axis of selection into the axis of combination.[6] Jakobson was talking about the sonic equivalents characteristic of poetry, but equivalence can operate at a higher level. Chapter openings are all formally equivalent, which means that they relate not only syntagmatically (the axis of combination) to the closures they succeed and the chapters they introduce, but paradigmatically (the axis of selection) to all the other openings. They enhance the possibilities and effects of choice and comparison.

But the discussion has become too vague. The time has come to turn to a concrete example. I have chosen Gunnlaugs saga, a text of suitable brevity from my point of view. Conveniently it exists in two relatively early manuscripts which do not entirely agree on the matter of chaptering. This disagreement has allowed editors and translators to exercise their judgment in the matter of chapter divisions. Your handout includes a rather simple schematization of some of the differences exhibited by the two manuscripts together with those of one edition (that of Peter Foote) and one translation (that of Gwyn Jones). This handout allows us to make some observations: first, the chapters in the A manuscript vary considerably in size, but some are by our standards very short. Second, the B manuscript prefers longer chapters, but its chapter divisions almost always line up with those of A. Third, Peter Foote prefers the chaptering of B, though he does not feel himself bound by it.

Now let us turn to the second handout. A moment ago I referred to the frequent habit of starting chapters with references to rulers. Notice that three of the chapters (as these are set out in the Foote edition) begin this way: Chapter 6 with Earl Eirik and his brother Svein in Norway, Chapter 7 with Ethelred in England, Chapter 9 with Olaf in Sweden. But when get to Chapter 8 something different happens. This chapter begins Síðan siglir Gunnlaugr af Englandi með kaupmönnum norðr til Dyflinnar. Þá réð fyrir Irlandi Sigtryggr konung silkisegg… (Gunnlaug then sailed with some merchants northwards from England to Dublin. At that time Ireland was ruled by King Sigtrygg Silken Beard…) Why is it that Sigtrygg is denied his place in the chapter opening sentence? The pattern of the two sentences that begin Chapter 8—Gunnlaug sails to a place and lands; the ruler of the place is identified—is the same as that which is spread across the chapter break that introduce the two preceding chapters: Chapters 5 and 6 close with Gunnlaug sailing somewhere; Chapters 6 and 7 begin, as I have mentioned, with the identification of the relevant ruler. Are we to understand that the change in Chapter 8 is carefully calculated, perhaps to avoid repetition, or is it a simple scribal mistake of the kind referred to above? Is it that Sigtrygg is less important?

Most opening formulae, like those just mentioned, give pride of place to temporal adverbials. The notable exception is the x hét maðr formula (a man was named) which with minor variants begins most sagas and many chapters within them. About three-quarters of the twenty-eight sagas I surveyed in preparing this paper, including Gunnlaugs saga, begin in precisely this way.[7] As an starting strategy, x hét maðr strikes me as remarkable, for it so uncompromisingly rejects that most conventional and useful of opening gambits, the phrase in illo tempore or once upon a time, which may be used either to anchor a story in time or cast it thoroughly adrift. Itʼs as if the act of naming is what sets the clock ticking, asserting social context, family responsibility, psychology, story. Hávarƒar saga Ísfirƒings merely makes the phenomenon explicit: Þat er upphaf þessarar sögu, at Þorbjörn hét maðr (It is the beginning of this saga that Thorbjorn was the name of a man…). Is character naming as an opening device an exclusively male, freeborn phenomenon? Almost, it would seem. Two chapters of Kormáks saga, as well as one of Bjarnar saga Hítdælakappa, the opening chapter of post-Classical Fljótsdæla saga open x hét kona or something similar. And the occasional thrall is also given pride of place. Jófríðr is allowed to open one chapter in the A version of Gunnlaugs saga with the anomalous Jófríðr var átján vetra er Þorsteinn fekk hana, (Jófríd was eighteen when Thorstein married her [my trans.]) but this clause occurs in a passage considered suspect by editors and relegated to footnote status.

Perhaps the most interesting of the opening formulae is the phrase nú er at segja which occurs in Chapters 10, 11, and 12 of Gunnlaugs saga. One must accept that for the scribe certainly and saga-writer possibly the repetitions of this opening is a self-conscious stylistic device. A quick check of the Randolph Quirk translation reveals that the modern scholar, creature of a different aesthetic, cannot bring himself to adopt identical phrasing—thus the three different renderings: first, To return now to Gunnlaug…, second, Now there is Hrafnʼs story to tell…, third, Now it is to be told of Hrafn that…. Having imposed structural equivalence on consecutive moments in a narrative, the Icelandic text chooses to underline their identity, whereas the English translator has accepted the equivalence imposed by chapter placement but has chosen to mask it stylistically.

Quirkʼs choice of different translations, it should be noted, is not simply a matter of trying to avoid repetition. The narrative gaps which are being closed by the formula are not all quite the same. The first, which he renders to return involves a kind of loop back in time (three years to be exact)—a heterodiagetic retroversion, as the narratologists would term it. This kind of chapter break should have a familiar feel to it, for novelists often avail themselves of the gap between chapters to gain just this kind of disjunctive freedom. In a tale such as the present saga, about two men competing for the same woman, the action necessarily bifurcates periodically, posing the saga-writer some nice narrative dilemmas as he moves back and forth between his lovers and their stories.

The nú er at segja formula seems at first sight to resemble very closely all the other openings that lead off with an adverbial of time, which is to say all but two of the thirteen chapters of Gunnlaugs saga. But there is a difference, for this now can belong to the story-telling present, unlike all the other adverbs which locate actions in relation to one another and to events, seasons, and reigns of the world of the story. It is not, I think, an accident that the narrative present should be foregrounded in this way as the story reaches its climax and the urgency of the telling begins to match the urgency of the action.

Though I spoke earlier of chapter divisions as providing occasions both for opening and for closing, we need not, I think, delay long over examples of the latter, though Gunnlaugs saga, like all sagas, provides instances aplenty of authorial concern for the dynamics and decorums of closure. Closing formulae bid characters adieu (Chapter 2) or leave children to grow up (Chapter 3), or as we have seen, they send heroes on voyages, sometimes allowing them to reach their proximate destinations (Chapters 5 and 6).

Sometimes, more suggestively, they strike an ominous note (Chapter 9) or they reflect on death (Chapter 12) before they utter the final amen (Chapter 13): Ok lýkr þar nú sögunni) (And there the saga ends).

Endings are not simply the incidental by-products of openings, but rather significant parts of the same structural moment, yet they do no seem equally have engaged the saga-writerʼs attention. In this respect, saga narrative differs from that of the modern novel; in novels chapter endings matter a good deal because they serve the important task of whetting the appetite. Sagas do not deal in suspense in our modern sense, since they are texts for rereading as we have noted. Saga chapters present rather entry points, as Lars Lönnroth, in the passage cited above, has remarked.

To conclude, Lee Hollander is quite right to rail against the arbitrary cutting up of saga texts, though mistaken in accusing Norman Blake of having done so, at least according to Norman Blake himself who in conversation with me denied having done any such thing. But Hollander is surely wrong to speak of the notorious inconsequence of the chapter headings and divisions of saga manuscripts. Chaptering may or may not be original to their texts and they are surely preserved with varying degrees of accuracy, but they tell us something non-trivial. And they do more than facilitate the comparative analysis of manuscript versions of the same story, for they help us understand the principles of segmental equivalence on which the construction of saga narrative depends.

1. Andersson. The Icelandic Family Saga, pp. 3–30; Lönnroth, Njáls Saga, pp. 68–82; Byock, Feud, pp. 47-62; Clover, Scene in Saga Composition, ANF vol. 89, 1974, pp. 57–83.

2. Culler, Jonathan. Defining Narrative Units, Style and Structure in Literature. Edited by Roger Fowler, p. 127.

3. Review in Speculum vol. xx, 1963, p. 328. (Actually Norman Blakeʼs thirty-eight chapter edition and translation of Jomsvikinga saga, the target of Hollanderʼs criticism here, must be truer to the manuscript divisions than Hollanderʼs own thirty-eight chapter translation of 1955, since Blake adds no chapters and evidently suppresses fewer than does his critic.)

4. Allen, Fire and Iron, p. 66; Lönnroth, Structural Divisions in the Njála Manuscripts, ANF, vol. 90, 1975, p. 79.

5. Parkes, The Influence of the Concepts of Ordinatio and Compilatio on the Development of the Book, Medieval Learning and Literature. Edited by Alexander & Gibson, p. 135.

6. Jakobson, Closing Statement: Linguistics and Poetics, Style in Language. Edited by Thomas Sebeok, p. 358.

7. Svarfdœla saga, on the other hand, never uses the phrase introductorily, though twenty-two of Njáls sagaʼs 159 chapters begin with the same formula.