[In the following essay, Macrae-Gibson discusses whether the skalds meant for their compositions to be interpreted in a larger cultural context and also analyzes Snorri's stance on the subject.]

Among Guðbrandur Vigfússon's scholarly concerns was the proper interpretation of Snorri's account of Old Norse poetry. He was particularly troubled that contemporary critics were misapplying it to Eddaic verse: “[Snorri's] system of terminology … is neither historically nor actually correct when applied to the older metres. Snorri, indeed, knows this, and does not attempt to analyse … the old poetry, but his modern followers have not shown such self-control …”.¹ The present paper has a related concern, though its business is rather the uses that should be made of Snorri's critique in the study of its proper subject, scaldic poetry—here, too, however, Vigfússon's thought had already identified one of the important points, as will appear. This paper is a development of the one of the same name circulated to the Second Saga Conference in Reykjavík in 1973,² and the follow-up paper which I had prepared for delivery and discussion at that conference, though in the upshot there was no opportunity for this.

Since 1973, fortunately, the view of scaldic verse as apt for aesthetic study has gained further ground. Such hostile views of its quality, as compared with the saga prose in which so much of it is set, as those of Hollander, who speaks of “the glaring contrast between its turgidity and conceits and the lucidity … of saga style”³ are less usual, and there have been valuable examinations of the nature of its poetic excellence, though none that I have seen even considers whether Snorri might have anything useful to say in the matter (with the single obvious exception of Margaret Clunies Ross's study of Skáldskaparmál,⁴ which in its nature concerns itself little with the sort of aesthetic considerations which are my main theme). It was clear, too, from other papers at the conference,⁵ that there is a greater willingness than formerly to consider at least some sagas, with their verses, as Bjarni Einarsson urged us to, “sem eina listræna heild”⁶ [as one artistic whole]. Yet still little attention is given to context in the study of scaldic verse. Even E. O. G. Turville-Petre, to whom English-speaking readers owe so much in this field, does not in the almost perfunctory section “Appreciation” in the Introduction to his Scaldic poetry⁷ even enquire whether one might identify an appropriateness of particular forms and figures to the particular context of a verse, and hence to the effect that it may have been intended to make, and he entertains only to reject the possibility that scaldic poetry might best be approached by reading it carefully in its saga context.⁸ When he observes that scaldic poets have repeatedly to say that sword struck against shield, but “if the sword is called ‘Oðinn's grey rainbow’ and the shield the ‘stormy sky of the valkyrie’, the bald statement may be worth making”,⁹ he does not consider what the test of worth should be. Is it merely a matter of ingenious and well-wrought decoration, or do we credit Ólafr hvítaskáld with an awareness that he is presenting an alarming image of a rainbow which does not celebrate the end of a storm but strikes lightning-like at the height of it, a new gloss on the heroic convention of battle as a way to heaven in which the bridge the warrior must cross is no bright Bifrost? The Lexicon poeticum¹⁰ records no other use of regnbogi in a sword-kenning, or indeed at all. When he praises the rich pictorial quality of scaldic diction, whereby “when Egill called the sea ‘the wounds or blood of the giant's neck’ (jotuns hals undir), we see vividly a stage in the formation of the world, the dying giant Ymir, blood gushing from his neck to form the roaring sea”,¹¹ is it not strange to ignore that Egill found this particular image of the sea dark as blood when his mind was on death, that it was an image in which the sea issues out of a personal death when his poem was to present the sea as a personalised killer, and yet an image in which that death is part of creation when the poem was to rise from despair to a re-creation? Many English-speaking readers first meet Old Norse in Gordon's Introduction;¹² his view of the importance of the verses in a saga is sufficiently shown by his omission from his Egils saga extract of the verse “Kominn emk á jó Íva …”,¹³ thus ascribing to Egill a wholly uncharacteristic mute appeal to Eiríkr. In what Frank¹⁴ describes as “one of the first studies devoted entirely to skaldic aesthetics”,¹⁵ Lie takes no account of saga context. Frank's own introduction to scaldic verse¹⁶ certainly gives the context of each verse presented, but this hardly affects her analyses. Von See in his introduction¹⁷ deals with the saga context of the verses (p. 86) only to point out that they certainly could not all have actually been composed in the stated situations. Thus he quotes approvingly de Vries's dry observation¹⁸ about the credulity one would need to believe that Hallfreðr really composed his “final verses” while mortally wounded, but it does not occur to him to discuss what the saga-writer's artistic purpose might have been in incorporating this final manifestation of his hero's character.

In considering the literary merits of scaldic verses as placed in a saga the question of their “authenticity” matters little. If we allow that a saga as we have it had in some sense “an author”, then whether he treated the verses as a source and built the saga round them, or structured the saga fitting in or writing appropriate verses where they seemed called for, the fact is that he left a work of which they formed part, and can be presumed to have intended them to contribute to the total effect the work was to make. Sometimes, to be sure, the contribution is no more than decorative, as with a number of the verses in Fóstbrœðra saga, for instance that in which Þormóðr records the deeds of Þorgeirr,¹⁹ adding nothing of consequence to the preceding prose. Occasionally a verse has an important function in the plot, as with Glúmr's “virkis spyrr at verkum …”.²⁰ But the function is more often complementary to the prose, to bring out things which the objectivity of saga prose style was not apt for. “Whatever … the Icelanders generally avoided saying in their laconic prose might find expression in verse”, as Foote and Wilson put it.²¹ This view affects one's criticism of both the verses and the prose.

On the verses, one will tend to see precisely the high contrast between their richly-wrought style and the prose's skilful impression of simplicity as the feature making them effective in their “pointing” function. When scaldic verse is treated as a thing in itself, it is possible to feel that the complexities of classical dróttkvætt are something of a poetic mistake. The less different poetic style is from prose the more it tends to appeal. Sighvatr Þórðarson has had, at least until recently, more attention from literary critics than most scalds, being felt to have “lasted” better because perceptive enough to simplify dróttkvætt, bringing into it something of the more “popular” Eddaic style (Lie's article on Sighvatr in the Kulturhistorisk leksikon,²² for instance, takes this line). He is praised for broad effects, such as could be made just as well in other poetic styles, for example the contrast in “Hooþótti mér hlæja …”²³ between the laughing hills while Ólafr lived and their gloom after his death, or that in “Fúss læzk maðr …”²⁴ between the extravagant grief of the southerner in the first half stanza, loudly declaring that he wants to die because his wife has died, and the restrained northman in the second, from whom indeed tears are appropriate on a loved king's battle-death, but only vígtár. Finnbogi Guðmundsson draws attention to this second contrast in his article on Sighvatr in Andvari;²⁵ it is notable that he says nothing about how particular devices of the craft of skáldskapr are used in its service. Sighvatr is not in fact rich in these, and his work can be usefully discussed without reference to them, but he well uses here the device of stress by suspension to emphasise “at deyja”, “vígtár” and “konungs orum”. To ignore them, and to prefer verse in which they can be ignored to verse which makes richer use of them, has been an aberration of modern taste, which study of the verses as contrasted with the prose in which they are set should help to correct.

On the prose, such study may modify one's view on what, when it is taken by itself, may seem to be limitations. Even so perceptive a critic as Foote, whose treatment of Gísli's dream verses²⁶ as “laying bare the sensibilities of a man tormented” by “inner conflict” is quite admirable, follows his introduction of the topic by observing that although “the poet of the dream-verses” was thus concerned, “it is difficult to see how much weight the author of the saga … attached to this idea.” He showed his concern, and indicated how much weight he attached, precisely by incorporating the verses; he would have weakened the effect by doubling it in prose. He can indeed be reproached with having here and there unduly repeated in prose the verse content, and he may not always have fitted in the verses with perfect skill, but that is another question, one that will be better answered if we approach the work with a prejudice in favour of its unity.

These dream verses represent the most fully developed example in the whole saga field of what is probably the most important function of the verses, the revelation of character. It is a truism of saga criticism that the characters' inward feelings are seldom explicitly stated, either by the author or by themselves; anything the reader is to gather of them he must infer. Gunnarr's celebrated speech as he turns back to Hlíðarendi is famous in large part just because of its rarity. But the verses which characters speak provide a major exception to this principle. In these, direct expression of inward feelings is wholly acceptable. Gísli's grief for Vésteinn²⁷ or Kári's for the burning of Njáll;²⁸ Ólof's determined love for Þorgrímr even though she has been pledged to Ketill;²⁹ Glúmr's brooding awareness of the gloom at Þverá since the days of his father, split by the sudden flash of anger of the “munat enn …” parenthesis³⁰—the whole essence of his character encapsulated. This type of character revelation has a function approaching that of a Shakespearean soliloquy, and the convention of the saga allows it only in verse. In prose, Sámr can be allowed to refuse gifts, saying “þá vera litla í skapi”,³¹ but we should be greatly surprised if he were made to add “glaumr verðr mér þorrinn”. In prose, we can be told that Glúmr “unði illa við málalok”, but for a direct presentation of how he felt the author must offer us a verse.³² Discussion of saga characterisation has given far too little attention to this. Hallberg, for instance, does not in his chapter “Stil och personskildring”³³ [style and character delineation] feel the need for a single mention of scaldic verse except as contrasted with the sagas. But if I am at all right one cannot respond properly to many sagas without being able to respond, as an instructed medieval Icelander might have responded, to the inset verses; they cannot safely be neglected by the literary critic.

“As a medieval … might have responded”. It is an impossible ideal, of course, and for most medieval literature we can get no nearer to it than what we can infer from the works themselves. In Old English verse we clutch at and dispute over such straws as “word ōþer fand, sōðe gebunden”;³⁴ we argue over how many stresses there should be in a line and where they should fall—sometimes, some of us guess, not on a syllable at all but on a rest, marked only by a harp-stroke. Norse scholars do not, I think, sufficiently value their good fortune in having available the actual critical views of a medieval Icelander. No scaldic scholar could be in doubt where the stresses fall, for Snorri's well-known account of the “stafir” and “stuðlar” of a verse puts the matter beyond question, but Snorri's guidance will in fact take us a great deal further than the technical analysis which is his chief concern, if we allow it to. Scholars have been too willing to assume that he was simply imperceptive of things which he did not make it his business to discuss, but on which he may nevertheless, perhaps in the by-going, have useful things to say if we look for them. That, as Vigfússon put it, “his object … was not the study of the old poets' minds and feelings, but simply the production of a handy ‘Gradus’”³⁵ is no reason to suppose that he was unresponsive to “minds and feelings”, or that his contemporaries were.

Kuhn, for instance, is severe on Snorri's poetic sensibilities.³⁶ He notes, justly, the artistic placing, interlacing and linking of clauses as among the merits of good dróttkvætt construction, but concludes that Snorri had no comprehension of this—“fur die kunstvolle Ordnung und Gliederung der Sätze in den alten Dróttkvættstrophen hat Snorri offenkundig auch kein Verständnis mehr gehabt”—because of his introduction into Háttatál of such strophes as no. 9 and those following, in which a series of parallel clauses appear, two to a line, one to a line, one to a pair of lines, a dully obvious arrangement which a good scald would have thought very little of: “Er zieht das einfache Aneinanderreihen von Sätzen vor, das die meisten Skalden der Frühzeit gering geachtet haben müssen.”

But that is quite unfair. First of all, Snorri did not invent these structures. Rognvaldr's Háttalykill, for instance, contains two sextánmælt stanzas, and Snorri's survey of forms could not ignore them. More importantly, Snorri offers these parallel-clause structures as a means of varying the usual dróttkvætt structure, “at breyta háttum”.³⁷ The implication is that the interlaced style is the normal one, and in his own verses before he introduces this variety he certainly employs it—one could rather charge his Háttatál 1 with extravagant excess of interlace than suppose that its writer did not recognise the propriety of interlace. Further, we need not suppose that he expected a poet necessarily to adopt in toto in a stanza forms that would give sextánmælt, or áttmælt, or fjórðungalok. He may be taken to be saying: “Here is a valuable way of modifying the usual clause interlace for a different effect. It is not difficult to write a whole stanza like this if you wish.” We may compare his presentation of the “sannkenning”. This is one of the devices with which a poet who wishes to cultivate an enriched vocabulary (“heyja sér orðfjolða”)³⁸ should be familiar, and Snorri demonstrates it by constructing a verse containing sixteen “appropriate” epithets in the eight lines. On this occasion, for once, he actually points out that though he has done it as a demonstration he is not recommending it as a practice: “en þó fegra þær mjok i kveðandi, at eigi sé svá vandliga eptir þeim farit.”³⁹ Generally he does not, because his purpose is to draw poets' attention to the resources available to them, not advise them when to use these. His presentation of “refhvorf” construction, in what is explicitly a demonstration piece (“Síks glóðar verr sœkir …”)⁴⁰ represents his method. He introduces the device with the comment that though it may seem hard to find suitable words this is in fact quite possible. “Look,” he seems to say, “I can get two opposed pairs into every line if I want to, creating what can certainly be described as ‘in mestu refhvorf’.” As for simple clause structure as a variation on interlace within the stanza, Snorri must have been aware that poets regularly use it. Gísli's “Hugðak Sjofn í svefni …”⁴¹ (one of the dream-verses) contains major and significant interlace in its second half stanza, but it opens in fjórðungalok, with single clauses in lines 1-2 and 2-3, emphasising by contrast the complexities which are to follow. A famous verse from Kormáks saga⁴² opens with a clause consisting of a pair of monosyllables, “brim gnýr”, as though it were going to be sextánmælt, stating concisely a theme which is then elaborately developed in the rest of the half stanza. Here then we have a case in which Snorri makes plain to us the nature of a device which we can then look for as employed by poets, and certainly not a basis for distrusting his poetic sensibility.

Much the same can be said of his prescription that metaphors, once established, should be maintained consistently through a verse: “þá þykkja nýgørvingar vel kveðnar, ef þat mál, er upp er tekit, haldi of alla vísulengð”, and the verse “Sviðr lætr sóknar naðra …”⁴³ with which he illustrates this. It develops elaborately the figure of snake standing for sword, with the movements of the one consistently representing those of the other to the extent that the verse becomes a tiny allegory. In his translation⁴⁴ Faulkes actually renders nýgørvingar by “allegory”, evidently convinced that this structure, with a single image systematically developed, is the only one Snorri is approving. Foote and Wilson also appear to accept that Snorri regards anything else as nykrat, a poetic monster.⁴⁵ If he did, they are certainly right that plenty of scalds wrote nykrat verse,⁴⁶ and he is a very narrow-minded guide to our appreciation of it. But there is no need for this; he can properly be seen as simply once more using the good teaching technique of making his point by a heavily obvious example, uncomplicated by anything else. He is in fact quite explicit about what makes a verse nykrat: “ef sverð er ormr kallaðr, en síðan fiskr eða vondr eða annan veg breytt, þat kalla menn nykrat.”⁴⁷ The fault resides in a change of metaphor for the same referent. In Egill's verse “Þverra nú, þeirs þverrðu …”⁴⁸ there is a shift of metaphor from “daylight” (“mjaðveitar dagr”) to “snow” (“digulsnjár”), but this is no breach of Snorri's principle, for the one represents “gold” and the other “silver”. They are elegantly related metaphors for related referents, a pleasing exemplification of the principle. A good scald can move smoothly from metaphor to metaphor like this in various ways. There is a fine specimen (which Foote and Wilson point out) in the Gunnlaugs saga verse “Munat háðvorum …”,⁴⁹ managed by using the word jorð simultaneously as proper and as common noun, and Egill's famous sea-storm verse⁵⁰ succeeds in seeing the ship first as a bull and then as a swan, without a crude jump of metaphor such as Snorri would deplore, by making “bull” merely an element in a kenning whose referent is “sea” (“stafnkvígs veg”), while “swan” is the head of its kenning (“Gestils olpt”). And so on; Snorri points out, in bold primary colours, the principle, and it is then for us to see how it is applied, in subtle tones, by poets.⁵¹ We can, of course, choose to proceed without his guidance, and perhaps reach the same destination. Frank's examination of how the kennings of a stanza can “interanimate each other”⁵² is pursuing a line of criticism very similar to some of the above, and with no mention of Snorri. But grant that his guidance is there, and is valid, and besides doing more justice to his memory we may find ourselves directed to more things than we might have found out for ourselves.

Even on so unprofound a level as his comment that vowel alliteration should preferably not be on identical vowels (“er þá fegra at sinn hljóðstafr sé hverr þeira”), and additional casual initial vowels are to be avoided even in such minor unstressed words as “er” or “um” (“er þat leyfi, en eigi rétt setning”)⁵³ he opens my ears to an elegance I should otherwise have missed; and often the level is much deeper. Consider his discussion of hjástælt construction—the device of closing a half stanza with a five-syllable sentence which by its isolation attracts importance to itself: “þær fimm samstofur … lúka heilu máli.” This “hjástál” ought to be, he says, “forn minni”.⁵⁴ If he really means that only proverbs are licit in the hjástál he would seem to have over-generalised from certain particular instances, but the observation of the special significance of a sentence thus placed is important, deserving more attention than, for instance, Hollander gives it:

Snorri … adds a stanza to exemplify the (rare) dróttkvætt variety he calls hjástælt (having a stál as an extra) in which the last five syllables of lines 4 and 8 make up the words of an orðtak (proverb). His own, made-up ones are vapid enough … It is to be regretted that Snorri does not otherwise expatiate on the nature of the stál …⁵⁵

I take it that in Snorri's own specimen⁵⁶ the poetic intention is an implied comment on the magnitude of Hákon's fame and the importance of his dwelling by placing them in the context of the great cosmic events, the hjástál acting as a pregnant comment on the matter of the preceding verse. That, then, is how one should expect to find this syntactic position used in other cases. Now verse no. 7 of Víga-Glúms saga, already mentioned, contains what is syntactically a pure hjástál, “váru þau forðum”, and Snorri's statements suggest that this should be viewed as having much more weight, as a concluding comment to the half stanza, than one might otherwise have given it. The stress would seem to be one of regret for the great days done, a personal variant on the forn minni. If this is the emotion presented as uppermost in Glúmr's mind, we can well then accept his continuing into a wish that nothing should be lost of his fame, a wish that evidently appeared in the lost two lines, presumably converting the statement “liggr … talit gorva” from its apparent meaning that there will now be no more of such deeds to signify instead that the tale has been reckoned, but misreckoned. Without this presentation of Glúmr's mind, his giving away in this verse of the truth about Þorvaldr krókr is abrupt and unmotivated, inserted after “sat nú Glúmr í virðingu” as a mere piece of contrived plot-mechanics. It is interesting to note that beside this suggestion of a stress on “váru þau forðum”, nothing that Snorri says suggests, as one might otherwise have supposed, that the isolated word preceding the stál receives special stress. No emphasis is laid on “morð”; if the word is significant it is slipped in quietly.

Related to true hjástælt form I take to be all others where the half stanza is concluded by a more or less syntactically isolated phrase occupying most of the last line, as in Þormóðr's verse before Stiklastaðir, “Ála þryngr at éli …”,⁵⁷ whose concluding “eða hér liggjum” ironically comments on the assurance of victory despite all odds which has occupied the rest of the half stanza, by supplying the more likely alternative outcome. The degree of separation, and consequent stress, is rather less, but the poetic tendency is the same, and gives a clear light on Þormóðr's character. Again, no special importance attaches to “viggruðr”, the word immediately preceding the stál in this case.

Once more, what is superficially merely a mechanical statement by Snorri of a rule of composition has pointed the way to a quite extensive line of criticism. In passing, on the actual “rule” in this case, Snorri himself uses what is syntactically a hjástál for what is certainly not a proverb in his “orðskviðuháttr” and “álagsháttr” verses.⁵⁸

Taking my departure, next, from Snorri's advice, already mentioned, that while it is possible to compose a stanza based on a single type of enriched diction, a variety of types produces more attractive verse, I examine this aspect of Egill's sea-storm verse (also already mentioned). It is notable for five metaphoric kennings, in Snorri's terms five kennings containing nýgørvingar. If they changed the metaphor for the same referent they would be objectionable, but they do not: we have seen one way in which this is elegantly avoided; the shift from “jotunn” to “gandr” might superficially seem to offend, but the very fact that we expect continuity of metaphor will ensure that we understand “gandr” as implying the “jormungandr”, who is of course of jotunn kind. If they were all of the same type the verse would be inferior, as too single-mindedly formed on the one pattern, but they are not; they are most carefully varied. One (“stafnkvígs veg”) is “tvíkennt”, one supported by a simple sannkenning,⁵⁹ “andærr”, one doubly supported (probably what Snorri means by “tvíriðit”), by “svalbúinn” and “eirar vanr”, the first of which, incidentally, is very rich, for the “sval-” element goes with the kenning's referent, “wind”, but the conversion to a cold disposition brought about by the addition of “-búinn” transfers it to a fit with the kenning's head, “(jormun)gandr”. The remaining two kennings are then of quite different logical structure: in “éla meitli” the squalls constitute the chisel, while in “Gestils olpt” the ship appertains to Gestill as if his swan. Besides all this there is one nýgørving which is not “kennt”, namely “þél”, and one straight sannkenning, “stórt”. Finally, to make a further variation, the biting squalls, kenned in the first half stanza, appear simply described, as “gustum”, in the second, while the ship's prow is not kenned at all, variety being achieved by using three different words to describe it “sem heitir” (Snorri's first “grein skáldskaparmáls”⁶⁰): stál, stafn, and brandr.

This remarkable richness and variety of diction is disciplined by having the first and last lines linked together by word and structure echo (“stáli”/“stál”, “fyr stáli”/“fyr brandi”)—compare Snorri's recommendation of “langloka” as one poetic device, though the method here is quite different—and containing chiefly literal diction, the simple physical prow of the ship buffeted fiercely, with the more ornate formulations enclosed between these lines. Turville-Petre reports an impression of “controlled energy” from this verse;⁶¹ my analysis, which attempts to see how that effect is created, is much indebted to Snorri for its line of investigation. In passing, if such features are indeed the significant ones for how the verse creates its power, one may judge that they should also be the ones to be kept, as far as may be, or found equivalents for, in a translation. It is clear from Turville-Petre's that he did not. The enclosure is quite altered, with the fourth line of the original removed to the first and the verse now concluded with two unkenned lines, making a quite changed rhetorical progression; the variation of kennings is lost, so that for instance the first lines of the two half stanzas now have identical structures, importing a degree of “tilsagt” construction foreign to the original; the unkenned nýgørving “þél” becomes another kenning, “a file of breakers”; and so on. I have attempted a rendering which does try to reflect the features of diction; the striking differences from Turville-Petre's may perhaps illustrate the sort of apprehension of how the verse works to which Snorri has pointed me—I make no claims for its poetic quality:

TURVILLE-PETRE

The angry troll of tree-trunks
the tempest's chisel wieldeth,
around the bull of bow-sprits
beats a file of breakers;
the freezing wolf of forests
files the swan of the sea-god,
grinds the beak of the galley,
grimly batters the forecastle.

MACRAE-GIBSON

By prow a rasp, savagely
(as bulls boat ever on)
recut ever with squall's flying
knives, forestkiller-flung;
and cold as cruelly he willow
lays low, so files fierce
with gusts the proud storm-swan
his foe, on prow, by bowsprit.

In the twofold approach to scaldic verse which I am urging, this attempt to use Snorri's guidance on what features of structure are important represents one arm. The other is an examination of relationship with context, where we have this. The importance to the present verse of its placing in the saga as a whole is less directly obvious than to some, but important it is, and accurate poetic response to it enriches response to the saga. Egill has angrily left the land, after raising his “níðstong” against Eiríkr and Gunnhildr. It is most proper, according to the literary convention usually, though oddly, labelled “the pathetic fallacy”, that the sea should catch his mood with a storm. The verse describes the storm as seen through the mood, with the abrasive image of the file, and the fierce one of the wind as a hostile giant, appropriate and powerful—the effect could not be got in prose. To imply Egill's control of the situation arising from his anger by ascribing to him a verse imposing such discipline on its ferocity is a stroke just as effective, and subtler; even less could anything like it be achieved in prose. The verse, then, in its place, is far more than just an example of Egill's ability to “depict nature as no other scald could do”,⁶² though it is that, of course.

Obviously, not every saga verse has diction so rich and varied. There are some very simple ones, and Professor R. I. Page has drawn my attention (in private correspondence) to cases “where the terms of a poem are framed to reduce variety of diction, as in Hallfreðar saga no. 14.”⁶³ This one, however, is a very special case: a poetic tour de force in answer to a challenge to work a sword reference into every half-line of a verse, quite different from verses such as I am discussing, whose content is organic to the saga in which they stand.

The storm verse just discussed illustrated only in moderate degree the importance of saga context to a verse; only in moderate degree, too, the principle of continuity of metaphor. Both can be more amply demonstrated in Egill's “dunhenda” verse “Olvar mik …”.⁶⁴ This exhibits real virtuosity in the management of metaphor, going far beyond mere observation of Snorri's rule, and yet it is only making richer use of the same principle. It establishes a weather metaphor “dew” (“ýring”) for “ale” in the first half stanza (associated with a weapon kenning “atgeirr ýrar” for “horn”), which continues in the second with “rain” for “battle” (“regn(bjóðr)”), associated with weapon-and-weather kennings for both “shield” and “poetry” (“oddský”, “regn Hávars þegna”). This sustained metaphoric line has much to do with the close structure of the stanza. Egill, who asks for ale in the first line,⁶⁵ is linked with Bárðr, who, as “regnbjóðr”, is represented as asking for “oddskýs regn”⁶⁶ in the last; the weapon content in the first kenning anticipates the weapon threat developed in the second; and while the overt reference of the third is doubtless to the verse now being uttered, the implication that the rain which Bárðr is inviting is about to set in (which it does as soon as the verse is over) is surely also present. Here the rhetorical progression is quite different from that in the storm verse: as the implied threat develops so does the complexity of the diction. And in this case the verse is absolutely central to the presentation of the character and thought of Egill, as indeed is the whole series of verses at this point of his saga. His cold anger at the lying trick played by Bárðr in the ill entertainment of his party; his penetration of and contempt for the attempt at poisoning; the controlled development of his vengeful intention in the verse I have been discussing; his derision for those who try to stop his escape, and die for it: all these things are presented in the verses, with the prose as restrained and external as ever.

Even more obviously central, of course, to presentation of character and thought is Sonatorrek, and this also strikingly illustrates continuity of metaphoric line, over a longer span than can be accommodated in a dróttkvætt verse. The familiar tree figure for “man” not only keeps reappearing for Egill's sons, in stanzas 4 (“sem hlynir marka”), 11 (“randviðr”), 21 (“ættar ask … kynvið”), and perhaps by implication 20, where the fire figure for “sickness” (“sóttar brími”) may suggest the common figure of fire as the enemy of trees (thus “sviga læ” in Voluspá 52); but it is also used for the poem itself in stanza 5 (“mærðar timbr máli laufgat”), a powerful figure of living trees in contrast to the dead trees of the previous stanza, a contrast sustained by that between the bearing out of the material body from the physical hall (“koggla berr frænda hrørs af fletjum niðr”, st. 4) and the bearing out of the “materials of verse” from the metaphorical hall (“berk út ór orðhofi mærðar timbr”, st. 5). Not only a contrast, however; because scaldic practice is to continue a metaphor for the same referent, an implied link is established in which the trees of stanza 5 in effect are the sons, brought alive again in their father's “mærð” (one thinks of the famous line from Hávamál “en orðstírr deyr aldregi”, p. 54, st.76 in the edition of D. A. H. Evans, London, 1986). This progress, of course, parallels the return of Egill from despair which constitutes the “action” of the saga at this point; one cannot sanely, in criticism, divorce the one from the other. That is not the only extended line of imagery in Sonatorrek; I have already mentioned the “sea” line, opening in stanza 3.

In the final section of this paper I want to take up a question to which I have been tending to assume an answer, a question important also in other areas of medieval literature, namely whether we are to suppose that scalds felt imagery as a purely self-contained thing, to be valued for its internal accuracy and elegance; or whether we can suppose them responsive also to possible resonances outside the image, to implications which would relate to the larger context in which the image occurs. Scalds certainly were concerned to form internally artful image structures, as when they deliberately play with different levels of imagery within a kenning, setting up an animate image and then subjecting it to inanimate treatment,⁶⁷ and the view that they did not look outside is often taken. As Anne Holtsmark puts it, for most poets “er kjenningene blitt sjablon og valgt tilsynelatende tilfeldig” [the kennings have become stereotype, chosen it would seem at random], and even exceptional ones achieve no more than “å bruke teknikken så de får fortalt to ting på en gang” [to use the technique to get two things said at the same time].⁶⁸ It will already be clear that I do not see them as limited to this inward view, but here Snorri might at first appear to give me no support. When he tells poets that “brjóst skal svá kenna at kalla hús eða garð eða skip hjarta, anda eða lifrar, eljunar land, hugar ok minnis,”⁶⁹ a modern sensibility would assume that there was a large poetic difference between “the house of the liver”, “the ship of the spirit” and “the land of memory”, but Snorri seems quite indifferent about where a poet might use which. The inference of indifference, however, is only sound on the premise that if Snorri had been concerned he would necessarily have expressed that concern here, and this premise is to be rejected; it must be emphasised again that his purpose was to remind poets of the resources available to them, not to tell them where to use them. The negative evidence here does not establish that he saw no imaginative difference between “hús lifrar” and “minnis land”, and elsewhere there is positive evidence that he was quite well aware of such differences. He records that kennings for men commonly have as base word a god- or elf-name. A giant-name can also be used, but “er þat flest háð eða lastmæli”.⁷⁰ The difference here is not logical, but emotional. He might well have passed it over as obvious, as obvious as that “hús lifrar” differs from “minnis land”, but fortunately for us he happened to mention it—one can almost hear a “but of course” introducing his comment. Thus if modern readers find themselves assuming that in Þormóðr's verse “Á sér at vér voorum …”⁷¹ the kenning “Hildr hvítings” was chosen for the ironic comparison between the woman with the ketill of hot water and a noble lady with a drinking horn,⁷² they have Snorri rather for than against them.

One can also call in support cases in which it is reasonable to say that res ipse loquitur, cases, particularly, where formations have unique features. A well-known example is the kenning “hlátra ham” in Egill's Hofuðlausn st. 20.⁷³ The poet needed a kenning for “breast”. To fit the immediate structural context he needed a rhyme on “gram”, for which “ham” was the obvious choice—hamr is not in fact elsewhere recorded in, nor is it given by Snorri for, breast kennings, but it is as natural as the hús and garðr which are. He then needed to complete the kenning with a word fitting the h-alliteration. So “hjarta ham”, or perhaps “hugar ham” (to suggest words which Snorri gives) would do very well. When instead he produces the unique “hlátra ham”, it is hard to suppose that it did not occur to him that anyone might look beyond immediate logicality and be reminded of the phrase hafa at hlátri. It seems much more likely that the implication was well present in his mind, the phrase then reflecting bravado on Egill's part as he composed, trusting that on delivery the point would not be taken critically against him—as we are told that it was not, though well might Eiríkr have looked at him sharply, and commented only “bezta er kvæðit fram flutt”. This is not the only example of possible irony in Hofuðlausn, of course, but it is the most striking, and the one that most sharply illuminates the character of Egill. That is to say, of Egill the hero of our saga; his connection, if any, with an historical poet Egill is beside the present point.

Less well known is the verse “Aldr hefk frétt …”⁷⁴ in Orkneyinga saga. Earl Rognvaldr is not going to mince his words, he says, about the noble and formidable lady Ragna, and he uses of himself the kenning “menja myrðir”. Kennings of this type for a generous man, who is the enemy of his stores of valuables because constantly giving them away, are well established, but “myrðir”, from myrða “to commit morð—dishonourable killing—on”, is unique. Can we suppose this a mere conventional self-compliment, with the pejorative resonances of morð unconsidered? Surely they rather go with the whole tone of the verse, in which Rognvaldr is clearly attacking the forward and persistent Ragna. The like can be said, though the tone is quite different, of the kenning which Hallr in the verse preceding the above in the same saga⁷⁵ is made to use of himself as a man of Iceland, “grúpans granni”. Since sausages were common in Iceland, Icelanders and sausages might commonly be found close to each other, and so it is logically sound to call an Icelander “sausage's neighbour”, but the mocking implication, that Icelanders and sausages belong on the same level, cannot but have been intended (the saga context does not make it clear whether we are to take it as self-mockery by Hallr or indignant ascription to Rognvaldr).

Not every verse calls for this sort of investigation into the appropriateness of its imagery. There is certainly a type whose sole or main function is to build up intellectual tension by suspending the solution of a piece of diction until the end, so as to reward the percipient reader with the satisfaction and release felt when the pattern completes itself.⁷⁶ Hallberg⁷⁷ applies Lie's description to Hofgarða-Refr's quatrain “Opt kom jarðar leiptra …”,⁷⁸ justly identifying the whole function of the long suspension before “stafna” as being to leave the oddly conjoined “jarðar leiptr” (as it stands “lightning of the earth”, apparently a “refhvorf”) jostling in the perplexed reader's mind, with “Baldr” in the next line doing nothing to explain, until “stafna” finally makes it instead “flash of the earth of prows”, hence “gold”, and the whole kenning, including “Baldr”, reveals itself as “man”. Such purely intellectual verse is commoner outside the saga setting, but quite legitimate within, simply to illustrate the poetic skill of a poet. Also, some saga verses are less well adapted to their contexts than others. But far more than the few I have mentioned are well adapted,⁷⁹ and will stand this sort of investigation.

When those two unique medieval literary forms, the scaldic verse and the saga, come together, then, they are not to be seen as simply coexisting in the same manuscript, but as able to form a further, joint art-form, in which they cross-illuminate each other. This goes back to the earliest stages of the preserved saga tradition, if Heiðarvíga saga is indeed the oldest family saga preserved. Here the characterisation by verse is of no great subtlety, but the direct bravery of Eiríkr viðsjá, wishing no truck with cautious withdrawal to a better defensive position, is well brought out in his two very direct verses in ch. 29, “Flykkjask frægir rekkar …” and “Fast holdu vér foldu …”.⁸⁰ In each case the structure builds up with a simple directness, in clauses successively one, one, two and four lines long, and the only syntactic suspension throws into stress his resolved course of action: “skalk … ór stað bíða”. We would have a less clear perception of his character without them (though one may feel some doubt about the third verse ascribed to him, “Hlotit hofum …”).⁸¹ It reaches its highest point in Gísli's dream-verses. Any extended treatment of these would require at least a monograph to itself; I have already mentioned the opening of “Hugðak Sjofn í svefni …”; in its conclusion Gísli's perplexed inability to understand the fact or nature of the comfort that is being offered to him is stressed by the isolated concluding position given to “und því váru”. In the later sagas, to be sure, it fades or vanishes. It is hard to attach much importance to many of the verses in Njáls saga; Hrafnkels saga has none at all. It is not clear that this change to a purely prose style represents an aesthetic gain. Foolishness, perhaps, to reflect on the verses we might have had in Laxdæla saga, expressing as it might be Unnr djúpúðga's proud anger, in her age, if anyone asked after her strength, or providing ipsissima verba for the “morg hæðilig orð” which Vigdís addressed to Ingjaldr in ch. 15. Wisdom, certainly, to value the verses we do have, and use the utmost of our aesthetic sympathies to respond to them, supported on the one hand by our understanding of the saga situations in which they appear, and on the other—as Vigfússon in his day urged—by a correct perception of what we can learn from Snorri Sturluson.

Notes

1. Corpvs Poeticvm Boreale, eds. Gudbrand Vigfusson and F. York Powell, 2 vols. (Oxford, 1883), I, 432.

2. Alþjóðlegt fornsagnaþing Reykjavík 2-8 ágúst 1973. Fyrirlestrar, 2 vols. (Reykjavík, 1973), I.

3. Lee M. Hollander, The skalds: a selection of their poems (Princeton, 1945), pp. 1-2.

4. Skáldskaparmál: Snorri Sturluson's ars poetica and medieval theories of language (Odense, 1987). Hereafter cited simply as “Clunies Ross”.

5. Notably Edith Marold, “The presentation of the skalds in Íslendingasögur”, Alþjóðlegt fornsagnaþing (1973: as at n. 2, above), II.

6. Skáldasögur (Reykjavík, 1961), p. 54.

7. (Oxford, 1976), pp. lxxiv-vii.

8. “Scaldic poetry and the English-speaking reader”, in Einarsbók: afmæliskveðja til Einars Ól. Sveinssonar, ed. Bjarni Guðnason et al. (Reykjavík, 1969), pp. 357-73, at 371.

9. Origins of Icelandic literature (Oxford, 1953), p. 31.

10. Sveinbjörn Egilsson, Lexicon poeticum antiquæ linguæ septentrionalis, second edition, ed. Finnur Jónsson (Copenhagen, 1931).

11. “Scaldic poetry: history and literature”, Bibliography of Old Norse-Icelandic studies, 1969 (Copenhagen, 1970), pp. 7-21, at 18.

12. E. V. Gordon, An introduction to Old Norse (Oxford, 1927). No change on this point in the second edition, revised by A. R. Taylor (Oxford, 1957).

13. Egils saga Skalla-Grímssonar, ed. Sigurður Nordal, Íslenzk fornrit, II (Reykjavík, 1933), ch. 59, p. 180, st. 33. I normally cite all saga verses as from the Íslenzk fornrit series.

14. Roberta Frank, “Skaldic poetry”, in Old Norse-Icelandic literature: a critical guide, eds. Carol J. Clover and John Lindow, Islandica, XLV (Ithaca and London, 1985), pp. 157-96, at 183.

15. Hallvard Lie, “‘Natur’ og ‘unatur’ i skaldekunsten”, Avhandlinger utg. av Det Norske Vidensk.-Akad i Oslo. ii. Hist.-filos. kl. 1957, no. 1; repr. in his Om sagakunst og skaldskap (Øvre Ervik, 1982), pp. 201-315.

16. Old Norse court poetry: the dróttkvætt stanza, Islandica, XLII (Ithaca and London, 1978).

17. Klaus von See, Skaldendichtung: eine Einführung (Munich and Zurich, 1980).

18. Jan de Vries, Altnordische Literaturgeschichte, second ed., 2 vols. (Berlin, 1964-67), I, 191.

19. Fóstbræðra saga, ed. Guðni Jónsson, in Vestfirðinga sogur, eds. Björn K. Þórólfsson and Guðni Jónsson, Íslenzk fornrit, VI (Reykjavík, 1943), ch. 8, p. 160, st. 7.

20. Víga-Glúms saga in Eyfirðinga sogur, ed. Jónas Kristjánsson, Íslenzk fornrit, IX (Reykjavík, 1956), ch. 23, p. 81, st. 7.

21. P. G. Foote and D. M. Wilson, The Viking achievement (London, 1970), p. 362.

22. Kulturhistorisk leksikon for nordisk middelader, 22 vols. (Copenhagen and elsewhere, 1956-78), XV, cols. 231-38, s.v. Sigvatr Þórðarson.

23. Den norsk-islandske skjaldedigtning, B. Rettet tekst, ed. Finnur Jónsson, 2 vols. (Copenhagen and Christiania, 1912-15), I, 252, Lausavísur, st. 26.

24. Ibid., I, 251, st. 22.

25. “Frá Sighvati skáldi Þórðarsyni”, Andvari, new series, 12 (1970), 85-102, at 99.

26. “An essay on the saga of Gisli”, in The saga of Gisli, transl. George Johnston, with Notes by Peter Foote (London, 1963), pp. 109 and 119-23.

27. Gísla saga, ed. Bjórn K. Þórólfsson, in Vestfirðinga sogur (1943; as at n. 19, above), ch. 14, p. 46, st. 5.

28. Brennu-Njáls saga, ed. Einar Ól. Sveinsson, Íslenzk fornrit, XII (Reykjavík, 1954), ch. 132, p. 346, st. 15.

29. Víglundar saga, in Kjalnesinga saga, ed. Jóhannes Halldórsson, Íslenzk fornrit, XIV (Reykjavík, 1959), ch. 6, pp. 71-72, st. 1.

30. Víga-Glúms saga (1956; as at n. 20, above), ch. 7, p. 20, st. 1.

31. Hrafnkels saga, in Austfirðinga sogur, ed. Jón Jóhannesson, Íslenzk fornrit, XI (Reykjavík, 1950), ch. 10, p. 133.

32. “Illt's á jorð of orðit …”; Víga-Glúms saga (1956; as at n. 20, above), ch. 27, p. 96, n. 13.

33. Den isländska sagan (Stockholm, 1956), pp. 63-72.

34. Beowulf, ll.870-71.

35. Corpvs Poeticvm Boreale, II, 447. The editorial material is formally by the editors jointly, but the reference on p. 450 to “the Editor's Dictionary” puts the authorship of this part at least out of doubt.

36. Hans Kuhn, “Von Bragi bis Snorri: zur Geschichte des Dróttkvætts”, in Einarsbók (1969; as at n. 8, above), pp. 211-32, at 230.

37. Háttatal, in Edda Snorra Sturlusonar ed. Magnús Finnbogason (Reykjavík, 1952), p. 261 (and so all further refs. to Snorra Edda).

38. Skáldskaparmál, p. 103.

39. Háttatal, p. 257.

40. Háttatal no. 17, p. 265.

41. Gísla saga, (1943; as at n. 27, above), ch. 33, p. 109, st. 38.

42. Kormáks saga in Vatnsdœla saga, ed. Einar Ól. Sveinsson, Íslenzk fornrit, VIII (Reykjavík, 1939), ch. 19, pp. 269-70, st. 56.

43. Háttatal no. 6, p. 258.

44. Anthony Faulkes (transl.), Snorri Sturluson Edda (London, 1987).

45. Viking achievement, p. 332. To illustrate the “monstrous” style they choose (pp.365-66) a sequence from Vellekla in which, for all the hardiness of the imagery, the only change of metaphor which might fall under Snorri's ban is from “blood” to various terms for “sea” to signify “poetry”, and since the sea was first formed from (Ymir's) blood even that could be defended.

46. This indeed is generally accepted, thus by Turville-Petre, Scaldic poetry, p. liii: “this practice … [is] found at all periods of scaldic poetry.”

47. Háttatal, p. 259.

48. Egils saga (1933; as at n. 13, above), ch. 78, p. 269, st. 51.

49. Gunnlaugs saga, in Borgfirðinga sogur, ed. Sigurðar Nordal and Guðni Jónsson, Íslenzk fornrit, III (Reykjavík, 1938), ch. 10, pp. 85-86, st. 11.

50. “Þél hoggr stórt …”; Egils saga (1933; as at n. 13, above), ch. 57, p. 172, st. 32.

51. Thus one need not see as objectionably nykrat verse which shifts between entirely conventional figures such as nobody would view as nýgørvingar, as from “tree” to “goddess” for “woman”, or in which the metaphors are cognate, as in Steinunn Refsdóttir's successive animal images for a ship in “Braut fyr bjollu gæti …” (Den norsk-islandske skjaldedigtning, B (as at n. 23, above), I, 128, st. 2. Even this is not common, and care is needed in interpreting evidence. Clunies Ross presents on p. 128 (as at n. 4, above) what seems a clear case; a ship has been a bear, then “in the following half-strophe” it becomes a horse. But the verses are known only from Snorri, and his introduction of the second with “ok enn kvað hann” (Skáldskaparmál, p. 139) does not assert that it immediately follows the first. Verse which would clearly fall under Snorri's ban is much rarer than has been supposed.

52. Court poetry (1978; as at n. 16, above), p. 52, “Skaldic poetry” (1985; as at n. 14, above), p. 169.

53. Háttatal, pp. 254-55.

54. Háttatal, p. 264.

55. “Observations on the nature and function of the parenthetic sentence in skaldic poetry”, Journal of English and Germanic philology, 64 (1965), 635-44, at 636.

56. “Manndýrðir fá mærðar …”; Háttatal, no. 13, p. 263.

57. Fóstbrœðra saga (1943; as at n. 19, above), ch. 24, p. 264, st 34.

58. Háttatal nos. 26 and 27, pp. 270-71.

59. Using the term as Snorri does in Háttatal, p. 257, though his application of it in Skáldskaparmál appears to be rather different; see Clunies Ross, pp. 69-75.

60. Skáldskaparmál, p. 102.

61. Origins, pp.42-43.

62. Ibid.

63. “Eitt es sverð …”: Hallfreðar saga in Vatnsdœla saga (1939; as at n. 42, above), ch. 6, pp. 161-62, st. 14.

64. Egils saga, (1933; as at n. 13, above), ch. 44, p. 110, st. 10.

65. Preferring here the editorial form of the opening “Ol ber mér” in Den norsk-islandske skjaldedigtning, B (1912-15; as at n. 23, above), I, 43, Egill Skallagrímsson, Lausavísur, st. 4. That in the Íslenzk fornrit ed. establishes a different linking. Egill is “drenched” with ale in the first line, as Bárðr is to be by a different sort of “drink” by the end.

66. Assuming the sense “invite” for “bjóða” to be present. The other sense “offer” is also there. In offering drink, Bárðr is also “offering” battle, though he does not know it. My analysis of the structure is little disturbed if “offer” is taken as the only sense.

67. See Clunies Ross (as at n. 4, above), pp. 127-28. A protest is to be registered, however, against presenting such effects as cruder than they are. If a horse has a “breast covered with red paint”, then certainly the phrase “has the effect of suddenly throwing the audience back to the inanimate referent, the ship, in the midst of … animate representation”. But “rauðum steini” need not define the colour as paint; the verb steina is not thus limited. When the ship is seen as a horse assailed by the sea its red-painted prow becomes a breast stained with wound-blood, an admirable combined image, not an abrupt shift from one to the other.

68. Kulturhistorisk leksikon for nordisk middelalder (as at n. 22, above), XV, col. 388, s.v. Skaldediktning.

69. Skáldskaparmál, p. 225.

70. Skáldskaparmál, p. 142.

71. Den norsk-islandske skjaldedigtning, B (1912-15; as at n. 23, above), I, 265, Þórmóðr Kolbrúnarskáld, Lausavísur, st. 21.

72. If the kenning is accepted as present at all. The Íslenzk fornrit ed. (Fóstbrœðra saga, 1943, as at n. 19, above, ch. 24, p. 270, no. 36) accepts “heldr”, not “hildr”, and takes “hvítings” with “sár” as giving “sword-wound”.

73. See discussion by Hollander, The skalds, p. 73, nn. 38, 39.

74. Orkneyinga saga, ed. Finnbogi Guðmundsson, Íslenzk fornrit, XXXIV (Reykjavík, 1965), ch. 81, p. 184, st. 43.

75. “Sendak son þinn …”, ibid., p. 183, st. 42.

76. Paraphrased from Lie, “‘Natur’ og ‘unatur’”, p. 33 (p. 229 in Om sagakunst): “Den ekte skaldestrofe vil spenningen, motstanden for dens egen skyld; den vil tvinge tilhøreren inn i psykiske tvangssituasjoner der han anspent leter etter en utvei, for til slutt å lønne ham med løsningens og avspenningens lyst, idet det brutte og forrevne endelig kan gjennomskues og for bevisstheten smelte sammen til den logiske helhet som er strofens nøkteme, reale gehalt.”

77. Den fornisländska poesien (Stockholm, 1962), pp. 16-18.

78. Den norsk-islandske skjaldedigtning, B (1912-15, as at n. 23, above), I, 295, st. 2, 1.2.

79. Thus at the conference where the present paper was first brought forward, Paul Bibire's “Verses in the Íslendingasogur” discussed the remarkable woman-kenning “Gerðr liðar hanga leygjar” of Eyrbyggja saga no. 21, ch. 28, pointing out how the death-resonance of hangi anticipates the speaker's own death, which is in fact about to occur; see Alþjóðlegt fornsagnaþing (1973; as at n. 5, above), I, p. 23 of Bibire's contribution.

80. See Borgfirðinga sogur (1938; as at n. 49, above), pp. 299-300, sts. 10 and 11.

81. Ibid., ch. 30, p. 304, st. 12.