Runes and Verse: The Medialities of Early Scandinavian Poetry

Abstract The paper discusses a number of versified runic inscriptions, mainly from Scandinavia, and from ca. 400 to 1400 AD, to explore what they reveal about the forms and functions of early Scandinavian poetry outside the manuscript tradition. With a particular focus on ‘authors’ and ‘audiences’, as defined by Bredehoft in his work on Anglo-Saxon inscriptions, the paper elucidates the potential oral contexts of Scandinavian runic verse and concludes that, although runic writing is a form of literacy, the examples show that for most of its history it is associated with various kinds of oral context. Runic verse shows that inscriptions provide one of the best ways into understanding the Scandinavian oral tradition, not only before the arrival of manuscript literacy, but also during its infancy.

It has long been recognised that there are many similarities between Old English and Old Norse literary culture and especially poetry, despite their chronological disparities. While many scholars nowadays prefer to stress these chronological and other disparities, or simply to ignore the similarities and concentrate on just the one tradition, there is still room for a nuanced comparison of the two bodies of poetry, as in for example recent work by Matthew Townend. Having examined some similarities in poetic diction he argues that these derive in part from the common roots of Old English and Old Norse. Such a shared specialised poetic diction suggests to him "that there existed a well-developed North-West Germanic poetic culture ... the reflexes of which can be observed in our extant Old English and Old Norse verse" (Townend 2015, 18). As well as this similarity of poetic vocabulary, and of course their well-known common metrical structures, these two corpora also share certain structural similarities which relate not only to patterns of transmission but also to the wider role of verse in their respective cultures. Thus, it is worthy of note that both corpora include anonymous as well as non-anonymous verse, and both include verse that is transmitted in runic inscriptions as well as in manuscripts in the roman alphabet. But despite these similarities, the poetical cultures of Anglo-Saxon England and early Scandinavia display significant differences, in particular when these medialities of roman and runic are considered more closely. A brief consideration of these differences will highlight those aspects of the early Scandinavian corpus which set it apart from the Old English corpus, before moving on to focus on the Scandinavian corpus, with a particular consideration of the potential oral contexts of runic verse.
In Authors, Audiences, and Old English Verse, Thomas Bredehoft argued for the integration of epigraphical texts into the much-studied corpus of Old English poetry, an argument that was certainly needed given the relative neglect of runology by literary Anglo-Saxonists. Bredehoft's overall aim was to assert "the importance of literate practice in the composition and appreciation of Old English verse" (2009, xiii), which he did by identifying a non-anonymous literate strand that can be distinguished from the oral-formulaic tradition that has been the main focus of past scholarship. Bredehoft also wanted to upset "our comfortable associations of Old English verse with orality and anonymity" (2009,200). He identified an "inscriptional tradition of Old English verse", consisting of epigraphical texts in both the roman and the runic alphabets, namely some ten inscribed stones on which "Old English verse was used ... to memorialize the dead ... and to identify a commissioner for the stone and its inscription..." (2009,54). One example is the ninth-century memorial from Great Urswick in Lancashire (now Cumbria), north-west England: +tunwinisetae aeftertoroȝ tredaebeku naefterhisb aeurnaegebidaesþe rs au lae Tunwini settae aefter Torhtredae becun aefter his baeurnae; gebiddaes þer saulae.
Tunwine put up (this) cross in memory of his lord (son?) Torhtred; pray for the (his) soul. (CASSS II, 148-50; Fig. 1) Bredehoft (2009, 63) argued that these inscriptions function "by figuring the commissioners [sc. of the monument] as originators of these verse texts" (in this example Tunwine) and compared this "association between specific verse texts and particular individuals" with the manuscript tradition of verse associated with Caedmon, Bede and Alfred. He argued that this strand of Old English verse attributed to named individuals forms a distinct, literate genre in contrast to the anonymous, oral tradition thought to be typical of poems like Beowulf.
As already noted, the Scandinavian tradition also incorporates both anonymous and non-anonymous verse, and verse in both runic inscriptions and manuscripts. However, despite this structural similarity, it is not possible simply to transfer Bredehoft's argument to the Scandinavian tradition. His model, which is limited to the Old English corpus, rather serves to highlight an important difference between that and the Scandinavian corpus. In Old English, the oldest surviving verse texts in both inscriptions and manuscripts are from around the eighth century. In the Scandinavian corpus, by contrast, there is a big gap between the oldest inscriptional verse, which is from around 400 AD, and the oldest manuscripts recording verse, or indeed any manuscripts, which are from around 800 years later. In Scandinavia, therefore, runic verse cannot have been dependent on a pre-existing, literate, manuscript practice, as Bredehoft suggests was the case in Anglo-Saxon England.
The purpose of this paper is to think around that gap between runic and roman in the Scandinavian corpus and thus to explore the relationship of runic verse to both orality and literacy. While runic writing is undoubtedly a form of literacy, it is clear that there are many important differences between runic/epigraphic literacy and manuscript literacy. The runologist Terje Spurkland memorably, if somewhat facetiously, advocated the term 'runacy' for runic literacy precisely in order to emphasise the differences between the ways in which runic and roman script were used, and their "dissimilar conceptual relationships between the oral and the written" (2004,; see also Bianchi 2010, 25-8). While the term 'runacy' has not gained much currency, it remains important to keep these differences in mind. As implied in Bredehoft's argument, the identification of an 'author' automatically creates an 'audience', even if who this audience was is not always obvious any longer. This paper will also focus on 'authors' and 'audiences' of runic verse, both of them in the widest possible sense.

Oral into Written
The examples discussed below illustrate how Scandinavian runic verse actually works. Most are well known, and a lot more could be said about these monuments or objects and their inscriptions than there is space for in this short paper. Previous discussions of this material have particularly focused on aspects like metre and diction, especially in order to identify and classify what is verse and what is not, and to pin down the metrical details. Definitions are particularly difficult as Scandinavian runic verse is very varied (in contrast to the more limited Old English corpus), and as both Heinrich Beck (2001) and Edith Marold (2010) have noted, the corpus stretches from single lines of simple alliterative verse to complex stanzas in dróttkvaett. Here, no attempt is made to define what constitutes verse in a runic context, a complex task that is well beyond the scope of this paper. 1 Rather, a pragmatic approach will be taken, in that, unless otherwise noted, the examples discussed below are chosen from texts that most scholars who have dealt with them have understood to be formulated as verse. There is no doubt that there is a continuum between what we can classify as verse and what is at best alliterating prose but any disagreements at the edges of these definitions do not invalidate the general points about authorship, audience and how verse or verse-like texts are used, which is what this paper is concerned with.
The earliest 'author' of verse (in Bredehoft's sense) on an object found in Scandinavia is the goldsmith Hlewagastiz who made the smaller Gallehus horn in around the year 400. 2 ekhlewagastiz:holtijaz:horna:tawido: ek Hlewagastiz Holtijaz horna tawido I Hlewagastiz Holtijaz made the horn (DR 12; Fig. 2

)
The inscription goes around the rim of the horn, beginning with the word ek 'I'. The last word tawido is inscribed in thinner runes than the rest, perhaps because whoever inscribed them realised they were running out of space and had to squeeze them in, or to signal that this is the end of the inscription. This first-person construction appears to be marked, as it would have been possible, given the space available, to leave out the pronoun and instead have a third-person statement that Hlewagastiz made the horn, with the appropriate thirdperson verb form tawide. The third-person statement can be found in other maker's inscriptions such as the box from Garbølle (Stenmagle), Denmark (dated to sometime before 400):

Hagiradaz tawide.
Hagiradaz made. (DR EM85;88) In contrast to this the Gallehus inscription is in verse and foregrounds the first-person pronoun, invoking the craftsman's voice, and these facts imply a performance and an audience. But who is the audience for this simple verse? One could imagine the craftsman declaiming the verse during the act of giving the horn to a patron, but presumably the real audience is those who can read the inscription. That audience might be, as we are, reading it long after Hlewagastiz is dead, yet he is still in some sense speaking to us directly through that first-person pronoun and the text that ostensibly represents his speech. In this way the text hovers uncertainly between orality and literacy.
In this use of the first-person pronoun the Gallehus horn can be compared and contrasted with another maker's inscription from around 700 years later, the Bridekirk font from Cumbria in England: Here and elsewhere this is considered as a Scandinavian inscription since it is mostly in Scandinavian runes, and in an area with other inscriptions in Scandinavian runes, though the language of the text is early Middle English. Again, the maker's inscription, here in rough rhyming rather than alliterative verse, names the craftsman (there is even a self-portrait of him below the inscription). But this time it is not the craftsman himself speaking. If the use of verse implies a spoken performance, the first-person oblique pronoun me, the object of the verb of making, must indicate that any speaking is done by an inanimate object, the font itself, rather than any human voice. Or perhaps we should conceptualise the font as writing rather than speaking? The use of several bookhand characters from the English tradition, the eth, the yogh and the tirrhonian symbol for 'and' (all marked in curly brackets above), suggests the literate world of written verse rather than the oral world of spoken verse. Even the wynn, although originally a runic character, should here be considered as a bookhand character, since it is otherwise foreign to the Scandinavian runic alphabet of this period but common in Old English manuscripts. Overall the text conforms to Scandinavian runic practice, as indicated by the cross at the beginning, the use of dividers, and the bind-runes, but it is now a runic text interacting with the literate tradition rather than the oral. If we take the text at its word, then what we have is a baptismal font speaking Middle English and writing both Scandinavian runes and English bookhand, telling us about the craftsman, who is mute, but who seems to have illustrated himself in the act of making. In such a situation of multiple modalities, the question of authorship is complex: was the 'author' of the inscription the craftsman who made the monument, or someone else? The simplest solution is that Richard was responsible. If someone else had written it, that would entail that someone wanting, not only to proclaim that Richard had made the monument, but also to present this information as if spoken by the monument itself, giving us three potential 'authors'. Richard, however, has already inscribed his portrait into the font, and it makes sense that he reciprocally allows the font to speak of him. In this way, Richard is using the distancing capabilities of literacy. He is no longer using literacy simply to fix his own oral communication in writing, as Hlewagastiz did, but is aware that readers will be reading, not hearing, his voice, and that the inscription will become a medial text, rather than a representation of his speech. This distancing both requires and allows the font to speak, when Richard will be long dead but the font remains.
The inscriptions discussed so far are entirely in verse and, in all three cases, the written text is visually delimited from the rest of the object, though the fact that they are in verse is not otherwise obvious. The Great Urswick inscription, consisting metrically of two long lines, gives no visual clues to the fact that it is in verse. As with Gallehus, the runecarver has struggled to fit his text into the prepared space, in this case unsuccessfully. Neither do Gallehus nor Bridekirk, being short, offer much opportunity for visual clues that they are in verse. However, in the Scandinavian tradition there are examples of runic verse that do offer such visual clues. Such clues are usually possible because the verse is part of an inscription that also includes text in prose and this underlying difference can be exploited to demonstrate that visually.
The Rök stone, from Östergötland in Sweden, and conventionally dated to around 800, famously includes a verse that could easily have come from the Poetic Edda, in its reference to a hero of the Migration period (often assumed to be Theoderic the Ostrogoth), and in its form which is a reasonably regular stanza of fornyrðislag: Þjóðríkr the bold, chief of sea-warriors, ruled over the shores of the Hreiðsea. Now he sits armed on his Goth(ic horse), his shield strapped, the prince of the Maerings. (Ög 136; Fig. 4) 3 The inscription, which covers both broad faces, and the three narrow faces, of the stone, is long, complex and fascinating, and it is not possible to do it justice (or even to quote it in its entirety) here, but several aspects of it relevant to the topic at hand need to be mentioned. It presents a complex situation in which the 'author' of the text, a certain Varinn commemorating his dead son, is at first spoken of in the third person. There follows a series of first-person statements which may or may not be the words of Varinn himself. The third layer consists of the stanza quoted above. As it is explicitly a memorial inscription, so in Bredehoft's sense it has an 'author', namely the bereaved father Varinn. Throughout the inscription there are first-person forms (which might be singular or plural), but if these represent Varinn, there has been a switch from third to first person after the initial thirdperson statement of commemoration.
The first-person statements are interrupted by the stanza, which is visually separated from the rest of the text. It is written first on two horizontal lines at the bottom of Face A, and then continues onto the narrow Face B of the stone. Visually, therefore, it is not necessarily a given that this verse is an integral part of the text. 4 Also it is not clear whether it is supposed to have been articulated by Varinn, by some other speaker, or whether it is a purely literate quotation as indicated by its visual separateness. Is this verse to be understood as a part of the statements made by the first-person speaker of the prose, or is it an addition, something apart because it is in verse, as well as being set off visually? Another oddity is that the stanza begins in the past tense but concludes in the present, linking the past of Þjóðríkr with the present of the inscription. This further complicates our sense of who is speaking, to whom and when. It is tempting to see the verse as a quotation from a pre-existing poem, whereas the rest of the inscription merely alludes to, rather than actually quoting, well-known narratives, what Stephen Mitchell has described (2013, 283; see also Harris 2010, 131) as "memories that are at once cultural and communal".

S[t]e[inn] [þe]ss[i] er settr eptir Sibba Góða/Goða, son Foldars, en hans liði setti at ...
This stone is placed in memory of Sibbi the good, Fuldarr's son, and his retinue placed on ... (Öl 1;Fig. 5) and the stanza itself: +: fulkin : likr : hins : fulkþu : flaistr (:)· uisi · þat · maistar · taiþir : tulka · þruþar : traukr : i : þaimsi · huki · munat : raiþ:uiþur : raþa : ruk:starkr · i · tanmarku : --ntils : iarmun··kruntar : urkrontari : lonti He lies concealed, he who was followed by the greatest deeds (most men knew that), a chieftain (battle-tree of [the goddess] Þrúðr) in this howe; never again shall such a battle-hardened sea-warrior (Viðurr-of-the-carriage of [the sea-king] Endill's mighty dominion ( = god of the vessels of the sea) ), rule unsurpassed over land in Denmark. (Öl 1;Fig. 5) Both texts begin at the same place, near the bottom of the stone. Both are introduced by a cross or cross-like shape, but then go off in different directions, the memorial formula to the right and the stanza to the left. Both are arranged boustrophedon 'as the ox turns in ploughing'. The stanza, moreover, is arranged in such a way that the end of the first helmingr coincides with the end of a line of runes at the top of the stone, suggesting that the rune carver was aware of and wished to indicate the stanza's metrical structure.
Like Rök this is a memorial stone, with the memorial formula providing a putative author of the inscription, in the nameless commissioner of the monument (Marold 1998, 670-72). Whereas in the case of Rök, this commissioner/author of the memorial inscription may not have been the 'author' of the quoted stanza, the situation is less clear in the case of Karlevi. However, in both inscriptions the separation of prose and verse is reminiscent of the ways in which skaldic verse is cited in Old Icelandic prosimetrum, which could imply that the verse had a separate author from the prose. There are indications that the Karlevi stanza was composed by someone schooled in the West Norse poetic tradition, whereas the monument's location in the Baltic and the fact that it commemorates someone who ruled in Denmark implies an East Scandinavian commissioner. But even if the prose formula and the stanza were composed by different people, the Karlevi stanza is not, like Rök, an allusion to a well-known oral poem. It is instead a verse produced for this particular location, the spot where the stone still stands 1000 years later, as is made clear by the deictic reference to 'this mound', an aspect which makes it fairly easy to reconstruct an oral context for it. The stanza praises the dead man in conventional terms familiar from other skaldic panegyrics and the reference to the mound suggests it was recited during Sibbi's burial. The runic version of the stanza is thus most likely a citation of an originally oral text, since presumably the stone was erected some time after the funeral, and one of its functions is precisely to perpetuate the verse declaimed on that occasion.
There is other evidence of an awareness of literacy on this monument, in the form of the roman-alphabet and presumably Latin-language text on the back: It is not clear if this is contemporary with the runic inscription, though it is possible. The dating of the monument is a little uncertain, but scholars agree on the late tenth century to around 1000. Overall, Karlevi reveals one way in which oral texts can make their way into writing. Here there are no speaking monuments, but an act of writing that is fully literate, though emerging from an oral predecessor, and designed to record the oral communication that took place at the burial, that is the verse, while the memorial formula did not necessarily have any oral pre-existence.
The fifth-century stone from Tune, Østfold, Norway, may also have a relationship to funerary rituals (for a metrical analysis, see Marold 2011, 75-8). Mentioned on Face A is an 'author', Wiwaz, who commissioned or made the monument, speaking in the first person, and in verse, about that act.
<...z> (= Wiwaz)(?) (erected) the stone for Woduridaz. Three daughters shared the inheritance, the closest family heirs. (N KJ72; Fig. 8) While many scholars have found reference to a funeral feast in this part of the inscription, this is not explicit in this particular interpretation, which follows a recent article by Thórhallur Eythórsson (2012), although even he admits that a funeral feast is likely to have taken place. An alternative interpretation has the dead man himself speaking in the first person, and in verse, declaring that three daughters prepared the monument for him:

[Me]z(?) Woduride staina þrijoz dohtriz dalidun(?) arbija arjostez(?) arbijano.
For me, Woduridaz, three daughters, the most distinguished of the heirs, prepared the stone. (N KJ72; Fig. 8) While the first interpretation seems unnecessarily repetitive, this alternative interpretation seems even more unlikely. While it is possible to find examples of the speaking dead in poetry, there are no parallels in the runic corpus, which is very much about the living remembering the dead. The first-person statement is restricted to the 'maker' of the runes, who like other makers already discussed records this act in verse.

In Praise of Writing
The examples discussed so far represent various ways in which oral texts are transformed into written ones, but the next examples focus on writing. They celebrate the skills of the rune-carver, but do this in verse and so introduce an oral element into a text celebrating literacy. The inscription on the early eleventh-century stone from Fyrby in Södermanland, Sweden, is unusually entirely in verse: iak · uait : hastain : þa : hulmstain : bryþr · menr : rynasta : a : miþkarþi : setu : stain : auk : stafa : marga eftiR · fraystain · faþur · sin · Once again, it is not clear who the foregrounded first-person speaker is supposed to be. Fred Wulf (2003, 995) took this as an 'epische Formel' indicating the start of a verse and not meant to be taken literally, but this is to ignore the broader context of the ways in which the first-person pronoun is used in a range of runic inscriptions, as explored here. Since the inscription mentions only one stone, the question arises as to which runes the brothers carved. They may, for instance, have carved runes on the many staves they set up, but unfortunately such things, being made of wood, do not survive. Or the stone mentioned in the inscription may refer to another rune-stone with a more conventional memorial formula. If so, the speaker of the verse on this stone could be another person entirely, commenting on the brothers' runic skills demonstrated elsewhere. Or this stone, with its unconventional memorial formula, was the only stone they set up, in which case the firstperson speaker could be one of the brothers, boasting of their achievements. Or this could be another case of an inanimate object, the stone, bearing witness to its creation, like the Bridekirk font. The 'authorship' of this particular inscription remains a mystery.
That man who is most rune-skilled west of the sea carved these runes with that axe which Gaukr Trandill's son owned in the south of the country (= Iceland). (Or Barnes20) 5 There is no first-person speaker here, the verse is a simple statement, whether of fact or not is debatable, except that the inscription was clearly not written with an axe, let alone one once owned by that tenth-century hero of a lost saga, the Icelander Gaukr Trandilsson. As in some of the other Maeshowe graffiti, part of the inscription (the first two words and the first letter of maðr) is in coded runes. This highlights the claim that the carver is skilled in runes, and suggests that he himself is speaking. This inscription conforms to a common pattern in Maeshowe: eleven of its 33 inscriptions are third-person statements that a named person has carved the runes, presumably carved by those same persons. There are, however, also two first-person inscriptions in Maeshowe (Or Barnes4 and Or Barnes8). There is a certain rhythmical, alliterative quality to these, even if it is a stretch to call them verse. The first sentence of one of these is indeed reminiscent of ljóðaháttr though Krüger and Busch (forthcoming) conclude that it is alliterating prose: þat · man · sat · er (·) ek · saehe · at fe · uar · ført · a brot · þrim · notom · uar fe · brǫt · ført · haeltr · aen þaeir br(e) hǫh þ(e)na Þat mun satt, er ek segi, at fé var foert á brott.

Þrim nóttum var fé brott foert, heldr en þeir bryti haug þenna.
That which I say will be true, that wealth was brought away. Wealth was brought away three nights before they broke this mound. (Or Barnes4) Both texts refer to the non-existent treasures of Maeshowe, a running joke throughout the graffiti, but also suggest traditional stories about mounds, mound-breaking and the finding of treasure. Like Rök, these inscriptions allude to a common narrative tradition.
The multiple authorship of the inscriptions of Maeshowe is clearer than on Rök. The difference lies in the specific act of carving, rather than in the more general sense of 'authorship' of the texts. The rune-carvers of Maeshowe frequently name themselves and it is clear that the inscriptions are carved by different people, though certain themes that run through the texts also make it clear that they are a group of some sort, with shared traditions and expectations. In the case of Rök, by contrast, even though the inscriptions either allude to or cite a variety of shared traditions, it is most likely that all of them were carved by one person, who may or may not have been the 'author' of the monument, its commissioner, Varinn. Any oral context for Rök lies either in the common traditions of the society in which it was carved, not necessarily linked to the carving process, or in discussions among readers of the inscription, subsequent to its carving. In Maeshowe, on the other hand, each individual inscription seems to have its own 'author', two of whom chose to write in the first person (and one in some sort of verse). The group quality of the Maeshowe texts emerges in the way some of the inscriptions echo each other, or even respond to each other, suggesting a dialogue within the group prior to or during the carving of the runes. Furthermore, some of the inscriptions might well allude to stories told within the group in the mound.
Oddr Skald raised this stone and made the stone-bridge in memory of Þorsteinn, his son. He owned three estates in 'hamarr'-partition and thirty marks (deposit) with Eiríkr. (Vg 4) Or it can be used of the rune-carver, as at Roslags-Bro kyrka in Uppland, Sweden:
Sigþrúðr had the stone raised in memory of Kári, her husbandman. May God help his spirit. Þorbjǫrn Skald cut the runes. (U 532) Both of these usages conform to Bredehoft's idea of 'authorship', despite the absence of verse. Mats Malm has argued (2010, 137) that the Old Norse word skáld "derives from words for 'resound', that is 'say', 'pronounce'", and concludes that in these examples the word refers to the fact that the rune carvers "convey statements of a sort which in an oral culture were a priori understood as carried by a voice" (2010,139). In this form of writing, runes are equivalent to, or replace, oral communication.
A further example of the use of the word skáld in a runic inscription from Bryggen in Bergen, Norway, mentioned but not discussed by Malm, shows how things have changed by 1300:  N B548; Fig. 9) Like much of the medieval runic poetry from Bergen and elsewhere, the reading presented above is sometimes conjectural and the stanza, if stanza it is, is not yet fully interpreted, as we await eagerly the publication of this material in a forthcoming volume of Skaldic Poetry of the Scandinavian Middle Ages (SkP VI). Despite the uncertainties, however, and even without a definitive interpretation, it is possible to identify some interesting aspects of this inscription relevant to the questions of authorship being discussed here.

÷ o^l ber·ek · ypisaeliu · a^rmg(l)(a)sek faerþ a^þ^u(s)u aerþa ygi(r) -: uil^ldae^k ÷ grimnis ÷ gil^ldi : (f)(a)^r(u)n^ ^n-n^u ha^ua ska^ld af staelda^r (s)-
Despite the damage to the end of the quadrilateral stick, it is clear that each of its four sides is meant to contain a complete couplet, giving a full dróttkvaett stanza overall. However, these four sides appear to have been inscribed by two different hands. 6 The first and third couplets, as presented above, appear to be by one hand, and the second and fourth by another, as can be deduced from some of the letter forms and the use of dividers by the first hand. Despite the use of runes and conventions of runic writing (including extensive use of bind-runes by both hands), it is clear that the carvers are also familiar with the conventions of writing in the roman alphabet, as shown by two examples of double l in the third couplet (both bind-runes), and the spelling of haua in the fourth, which would normally be spelled with an f in a runic inscription and seems to represent a manuscript spelling in the roman alphabet tradition.
Each of the first three couplets contains a first-person pronoun. The first two of these pronouns collocate with present-tense verbs which might represent the very moment of speaking -in the first couplet the speaker announces that he is presenting a poem (apparently to a woman), while in the second couplet the (second) speaker describes what he is seeing (perhaps the same woman). However, if the above interpretation is correct, the poem-kenning Yggjar ǫl is not complete within the first couplet as the element Yggjar is postponed until the second couplet. This means either that the two 'authors' of the text are so well-versed in skaldic technique that they can compose jointly in this way or, perhaps more likely, that they are alternating in writing down a pre-existing stanza. In the third couplet, the first-person pronoun collocates with a past-tense verb, presumably the first speaker again, though we do not know what the intended statement is, except that it again involves a poem-kenning. The fourth couplet however contains a third-person subject, in the plural, as shown by the verb form hafa. It is not too speculative to see this as referring to the two speakers/rune-carvers of this poetic text, coming together in the final couplet, though written by the second hand. But what exactly have these skalds done? If we assume that staela refers to the act of making an intercalated refrain for the poem, then we have something like 'the skalds have fitted [the poem] with a refrain'. Does this imply that the two carvers composed the whole stanza jointly, and that this is the refrain? Or have they added a refrain to the pre-existing poem which they have just cited three couplets of? There is some evidence of the citation of pre-existing stanzas in the medieval runic corpus, though not always exactly as they survive in the manuscript tradition (Marold 1998, 690).
Incomplete preservation makes it difficult to reconstruct a context of authorship and use for this inscription, though its literary context is fairly clear. Just one example is in Bjarni Kolbeinsson's Jómsvíkingadrápa which also begins with a first-person speaker presenting his poem as Óðinn's drink (though in an ironic way): Jómsvíkingdrápa is also one of the few long poems for which its refrains survive in context. One of these concerns the poet's supposed unrequited love for a woman, and the theme of love is prominent in the poem more generally. It is hard to determine if Jómsvíkingadrápa, perhaps a century earlier than the Bergen rune-stick, is the product of an oral or a literate culture, though it is likely that both played their part (Jesch 2014). Certainly by the time of the rune-stick, around 1300, runic verse is interacting with written texts and is perhaps dependent on the literate tradition, even when, as in this case, it involves the 'voice' of one or more first-person speakers.

Conclusion
Much more could be said about these inscriptions, or indeed many other verse inscriptions from the fifth to the fourteenth century. Those discussed above were chosen to explore the particular question of possible oral contexts of runic verse. Although runic writing is a form of literacy, the examples show that for most of its history it is associated with various kinds of oral context. In some cases the runic verse is no more than a representation of its oral equivalent, as probably in the Gallehus inscription. In other cases, the runic verse is a quotation, either of a well-known verse in the common tradition (Rök), or of one less wellknown, but still composed earlier, for a specific occasion (Karlevi) -the Bergen stick must also belong in one or the other of these categories. Other verses seem to have been composed specifically to be carved on the runic object, but allude to and were probably associated with certain oral contexts in which it was normal to declaim verse, such as a funeral feast (Tune). Yet other verses, such as the one from Maeshowe, arose out of the dialogues of a group engaging in both oral and runic discourse, as also perhaps the one from Bergen.
Much of this is also true of runic inscriptions that are not in verse, as shown by the example of Maeshowe, in which only one or two of the thirty-three inscriptions can be described as in verse or verse-like. But more so than prose, verse at least suggests the speaking voice, particularly when it includes the attention-grabbing first-person pronoun as in many of the examples above. For this reason verse inscriptions are better suited to reconstructing the oral contexts in which these texts arose. Inscriptions not in verse may never have existed in oral form other than perhaps a commissioner dictating the text to the rune carver, whereas verse inscriptions suggest, in various ways, that they had an existence independent of their written form. In a study restricted to Viking Age rune-stone inscriptions, Fred Wulf argued (2003), largely on metrical grounds, that the majority of runic verses were composed by the rune-carvers themselves. This does not entirely contradict the argument presented here -as Wulf himself shows, many of this particular category of verses are highly formulaic, suggesting that the rune-carvers were familiar with similar verses from an oral context, perhaps that of the funeral rites for the dead being commemorated. But in the full corpus of Scandinavian runic verse, extending over a millennium, and preserved in a variety of physical contexts, it is clear that many of the verses did have an oral existence independent of their surviving written form.
In Old Norse studies, there has been much discussion over the years of the possible oral background to the literature that survives in Icelandic manuscripts, whether prose or verse, though this oral context is difficult to reconstruct. The runic verse discussed above shows that inscriptions provide one of the best ways into understanding the Scandinavian oral tradition, not only before the arrival of manuscript literacy, but also during its infancy. The written medium of runes, paradoxically, gives us some of the most useful insights into the uses and contexts of the lost medium of orality.