The Dating of the Eddic Poems

The gap between oral composition and manuscript recording makes the dating of the Eddic poems one of the hardest problems in Old Norse philology.

Manuscript versus composition

The Poetic Edda survives in Codex Regius (GKS 2365 4to), written down around 1270. Most of the poems are almost certainly older than the manuscript, but by how much remains contested. Finnur Jónsson placed the bulk of the poems in the Viking Age, roughly 900 to 1050, while Sigurður Nordal, in his influential study of Völuspá, argued for a much more precise dating to around 1000.

The difficulty is that the poems circulated as living oral tradition rather than fixed texts. Each performance could involve minor reformulations, inserted stanzas, or lost material. What we read in Codex Regius is therefore not necessarily identical to what a skald performed three centuries earlier. Ursula Dronke emphasized that traces of oral compositional technique, above all formulaic language, point to a long tradition history behind the recorded versions.

A further complication is the heterogeneity of the poems: mythological poems, heroic poems, and gnomic poems may have different origins and different transmission histories. Treating them as a uniform corpus with a single date is methodologically questionable.

Internal evidence and external parallels

Internal criteria such as historical allusions, linguistic forms, and metrical features have been used to date individual poems. Hávamál and Hymiskviða display archaic traits suggesting considerable age, while certain parts of the narrative frame in Völuspá may reflect Christian eschatology, hinting at late composition or reworking.

Comparison with skaldic poetry provides additional fixed points. Kennings and mythological allusions in datable skaldic verse show that stories of Thor and Odin were well known by at least the tenth century. This sets a terminus ante quem for at least part of the tradition history of the mythological poetry.

The survival of similar stanzas in other manuscripts, such as Hauksbók, and parallels in the Prose Edda illustrate how fluid the texts were. Speaking of a single 'original version' is probably misleading; each manuscript represents a freezing of an ongoing tradition process.

Current state of scholarship

Contemporary scholarship is more cautious about precise datings than earlier generations. The prevailing view is that most mythological poems have roots in pre-Christian times, probably the eighth to eleventh centuries, but that the texts we possess are products of a complex transmission in which oral and written stages are intertwined.

John McKinnell and Rudolf Simek have stressed the importance of distinguishing between the subject matter of a poem, its form, and its recorded shape. These three aspects may have very different ages. A story may be ancient, the metrical form may be younger, and the specific wording may reflect the choices of a late redactor.

The question of dating is more than academic: it determines the extent to which the Eddic poems can be regarded as evidence of actual pre-Christian belief and cult, rather than as literary constructions created in, or shaped by, the medieval Icelandic literate environment.

Sapp's quantitative breakthrough (2022)

In 2022, Christopher D. Sapp published Dating the Old Norse Poetic Edda: A multifactorial analysis of linguistic features, which represented a methodological breakthrough in the dating question. Sapp works from the principle that the Eddic poems can be compared to datable skaldic poems whose authors and dates are known, and examines three linguistic features that change diachronically in Old West Norse: the frequency of the expletive particle of/um, the type of negation (clitic -at versus the newer independent eigi), and the type of relative clause (er alone versus the newer sá er construction).

Using a Naïve Bayes Classifier, a statistical model trained on the known dates of skaldic poetry, Sapp assigns each Eddic poem a probability score for composition in seven time periods from the ninth to the thirteenth century. The model achieves 80.4 percent classification accuracy on the known skaldic poems. The results show that the great majority of Eddic poems are dated to the tenth or eleventh century, that is, the pre-Christian period or the decades around the conversion. None of the 29 poems in Codex Regius is classified as late as the thirteenth century.

Among the most notable results are that Þrymskviða is dated to the ninth century (the only poem that early), that Alvíssmál, Oddrúnargrátr, and Hymiskviða are dated to the tenth century despite being widely considered late compositions, and that Helgakviða Hundingsbana I is placed as late as the late twelfth century. Hávamál is divided into three parts with different dates: stanzas 1–110 and 138–164 to the tenth century, while Loddfáfnismál (111–137) is dated to the early eleventh century. The results broadly support Finnur Jónsson's early datings and argue against Klaus von See's theory that the Eddic poems are late, learned products of twelfth- and thirteenth-century Iceland.

Sapp's study does not replace the datings of earlier scholars but gives them an empirical, quantitative foundation. The linguistic dating concerns the age of the surviving text, not of the underlying narrative, and does not rule out the possibility that individual stanzas may be older or younger than the poem's main compositional layer.