This interface will soon cease to be publicly available. Use the new interface instead. Click here to switch over now.
Anonymous Þulur (Þul)
The þulur (Anon Þulur) are versified catalogues of poetic terms (heiti) for the main subjects of skaldic verse. These enumerations, in fornyrðislag metre and varying in length from a half-stanza (e.g. Þul Boga) to twelve stanzas (Þul Sverða), were evidently composed primarily for mnemonic and educational purposes and are likely to have been intended as reference lists for poets. At the same time, certain peculiarities of their content, such as the abundance of terms which were never used in poetry and a substantial learned component contained in a number of the þulur, indicate that at least in part these enumerations might have been compiled not so much for utilitarian but rather for encyclopedic purposes (see below).
Composed presumably in the second half of the twelfth century or at the beginning of the thirteenth (though some lists of heiti may have been compiled after 1250), the þulur are found in two main redactions, both transmitted in the mss of SnE, where they are added to the end of Skáldskaparmál (Skm). A sequence of thirty-five þulur, which consists of 106 stanzas enumerating a total of 1,500 heiti, is appended to Skm in mss R, Tˣ and C (cf. SnE 1931, xxxviii; Guðrún Nordal 2001, 45). The same collection of þulur, with certain minor changes in the order of subject matter and augmented by twenty-four additional lists comprising another sixty-eight stanzas and 1,083 heiti, is preserved in ms. A and partially in B (also represented selectively by 744ˣ in the present edition). Mss R, Tˣ and C contain exactly the same sequence of þulur, which indicates that all three mss ultimately derive from the same archetype, but in terms of variants, ms. C occupies an intermediate position between the two main groups of mss, R and Tˣ, on the one hand, and A and B on the other. The readings contained in C sometimes agree with those found in R and Tˣ against the readings of A and B or vice versa, and quite often the C variants are caused by omission, addition or rearrangement of heiti. Because the þulur are missing in mss U and W of Skm, it is generally believed that these versified lists were not composed by Snorri himself but were added at a later stage (cf. LH II, 179; SnE 1998, I, xviii).
The lexical material contained in the þulur is partially included in the prose enumerations of heiti in around forty articles (articuli) of Magnús Ólafsson’s early seventeenth-century LaufE, most of which were also copied in RE 1665. The content of these prose lists and the forms of names given in LaufE show that Magnús probably used several mss related to both of the extant medieval redactions of the þulur (R, Tˣ, C and A, B), but not derived solely from either of them (Faulkes, LaufE 1979, 166-72). Variants from LaufE and RE 1665 have usually not been incorporated in the present edition, though information from both is included in the Notes when pertinent.
In mss A and B, with a few exceptions, the scribes announce each new subject with a rubric (e.g. Sverða heiti ‘Names for swords’), in ms. A written in bright red colour throughout. Mss R, Tˣ and C originally had no rubrics, though it is possible that spaces left in R indicate that this ms. once had rubrics, which are now illegible. In Tˣ (and in some lists in C) rubrics were subsequently added in the margins in a later hand (SnE 1998, I, 227). None of the mss containing the so-called þulur introduces these sets of stanzas with a general title that could shed light on what the Old Norse word for such enumerations of heiti could have been. The term þulur (sg. þula), however, has adhered to these versified lists firmly and for a long time without being questioned at all, as if this term were actually found in the extant medieval mss, although Smirnitskaya (2003, 184, 193-5) argues that such lists of names in metrical form were most likely called tal ‘enumeration’, as is the case in the last stanza of the so-called Dvergatal, the list of dwarfs in Vsp 16/7 (NK 4): langniðia tal ‘list of lineages’. Since the Old Norse word designating these lists cannot be ascertained and the application of term þulur to these lists is a later scholarly construct, this term and its relevance for the Old Norse poetic catalogues of names will be examined in more detail here.
The term þula
The word þula does not appear in the Poetic Edda, but is used several times for poems composed in eddic metres, e.g. Rígsþula (Rþ), an eddic poem preserved in ms. W of SnE. Although some of its stanzas (sts 12-13, 24-5, 41) resemble the þulur of heiti in that they also list the names of persons representing different strata of Old Norse society, Rþ is a narrative poem and its content is by no means confined to mere enumeration. Moreover, whatever sense the term þula might have in the title of this poem, it certainly does not refer to the lists of names in question, and one may rather think of recurrent situations described in various parts of Rþ (Smirnitskaya 2003, 184; on Rþ see Dronke 1997, 174-214). Hence Finnur Jónsson (LH II, 174) denies this poem the status of ‘a real þula’ (en virkelig þula), by which he evidently understands enumerations of names. It is uncertain, however, whether Finnur’s postulated meaning can be supported by the title of another poem, Þorgrímsþula ‘Þorgrímr’s þula’, known only from the SnE mss, which Finnur Jónsson (LH II, 175) regards as ‘a real þula’. The very fact that the two different fragments of the latter poem (see Introductions to Anon Þorgþ I, II), enumerating mythical and legendary names, are cited in the sections of Skm which discuss heiti for ‘horse’ and ‘oxen’, seems to suggest that they are taken from a more extended lost poem whose subject matter we can only speculate about.
The three known skaldic occurrences of the word shed even less light on the meaning of þula in Old Norse. The earliest is found in a stanza by Sneglu-Halli (c. 1060) and refers to a deficient praise-poem that this skald allegedly composed at the English court in honour of King Harold Godwineson (Haraldr Guðinason): Ortak eina | of jarl þulu; | verðrat drápa | með Dǫnum verri ‘I composed a þula about an earl; a drápa cannot be worse among the Danes’ (SnH Lv 7/1-4II; see Note there). The second occurrence is found in the anonymous Málsháttakvæði (c. 1200; Anon Mhkv 11/1, 3): Stefjum verðr at stæla brag, … ella mun þat þykkja þula ‘Poetry has to be fitted with refrains … else it shall seem a rigamarole’. As has been justly pointed out (Kreutzer 1977, 88; Clunies Ross 1987, 81 n.), in both cases the term þula is used ironically, even pejoratively, and an unsophisticated poem assigned to the þula-type is juxtaposed to the most valued and well-structured type of skaldic composition, the drápa. However, no precise information is provided by either skald concerning the form or other traditional properties of a þula, except that it has no stef ‘refrain’. A later poetic instance is found in Bósa saga, where the term þula refers to Buslubæn ‘Busla’s Curse’, a poetic invocation uttered by Bósi’s foster-mother, Busla (Busl 7/10VIII (Bós 7): eða vilt þú þulu lengri? ‘or do you want a longer þula?’), and is then repeated in the accompanying prose (cf. Bós ch. 5, FSN III, 206: er sú þula var úti ‘when this þula was over’). Here again, the sense of the term is undoubtedly broader than a mere enumeration, though its immediate implication is uncertain.
Hence, as is evident from the examples cited so far, in the extant Old Norse sources the word þula is used in a wider sense than ‘a versified list of names’, the meaning now assigned to it in scholarly works. There is, nevertheless, one more – and the only known – Old Norse example which fully justifies the scholarly definition of the term þula. This is Allra flagða þula ‘The þula of all trolls’, an enumeration of ninety giants’ and giantesses’ names in the form of an alliterative poem found in Vilhjálms saga sjóðs (Loth 1962-5, IV, 66-8) from the late fourteenth or early fifteenth century. What is now considered the title of this poem, but probably is just an introductory statement, prefaces the enumeration proper (Loth 1962-5, IV, 66): aull skulu þier standa sem stiaki bundinn unzs at eg hefi vt kuedit allra flagda þulu ‘you must all stand as if tied to a stake until I have finished reciting the þula of all trolls’. As far as the names contained in Allra flagða þula are concerned, this poem has very few parallels in the respective lists of Þul Jǫtna I, II and Þul Trollkvenna found in the sets of þulur in Skm (see also Introductions and Notes to Þul Jǫtna I, II and Þul Trollkvenna), and this leads to the conclusion that the compiler of the þula in Vilhjálms saga most probably did not consult these lists. But from the ways in which the names enumerated in this þula are put together and arranged into pairs, including the apparent tendency to use, in addition to alliteration, such mnemonic devices as internal or end rhyme, it is evident that Allra flagða þula is composed within the same traditional genre of list-poems as the so-called þulur in Skm (cf. also the arrangement of names according to alliteration on the initial sound in Þul Jǫtna I 1/5-8, 2/1-6 (h-); Þul Trollkvenna 1 (g-), etc.).
Traditions of þulur composition outside skaldic poetry
The tradition of þulur composition might be very old and, as has been argued by Vogt (1927a; 1927b; 1942; see also de Vries 1934b, 56-8; 1964-7, I, 31-2), could go back to ancient Germanic ritual practices. Based on certain instances of usage of the cognate ON þulr, apparently ‘(old) wise man’, ‘speaker’ (however, ‘poet’ by the skalds, and cf. OE þyle glossing inter alia Lat. orator) and þylja ‘chant’ or ‘murmur’, Vogt (1927b) advanced the hypothesis that a þula might originally have been cultic speech (Kultrede) uttered by a þulr, i.e. a priest and a cultic orator (Kultredner), whose duties probably were to preserve old sacral lore and to recite it in public on ritual occasions. According to that theory, the diverse enumerations of names of mythical beings and objects found in a number of eddic poems, such as the dwarf-names in Vsp 11-16, the river-, horse-, serpent-, valkyrie- and Óðinn-names in Grí 27-30, 34, 36, 46-54 (to which may also be added some lists of the names of heroes; e.g. Hyndl 22-4), may be seen as direct successors to the ancient þulur.
The prehistory of these eddic lists is, of course, obscure; moreover, both their origin and their actual relation to the poems they are embedded in are debatable, and at least some of these stanzaic lists may be later interpolations (cf. Dronke 1997, 67 and de Vries 1934b, 42; for an alternative opinion, see e.g. Ralph 1972, 106-8, who argues for the unity of Grí). But irrespective of their status and immediate purpose in particular texts, the very fact that these lists were readily included in the eddic poems is the most convincing evidence that such versified lists of names were regarded as an important part of eddic gnomic poetry (cf. de Vries 1934b) and served as an efficient means of preserving and transmitting mythical and legendary information. There are good reasons to suspect, however, that at least some of the names included in the mythological þulur could have originated from these very lists (e.g. Grí 27/1-2, 6 (NK 62): Síð oc Víð, | Sœkin oc Eikin … Gipul oc Gǫpul). In fact, these names do not appear in independent sources and are not supported by any known myth, but, as is clear from their form and the ways they are combined into pairs with other names in the list, they fit so perfectly into the connected patterns of rhythm, rhyme and/or alliteration that they were most likely invented by the compilers of the mythological þulur embedded in the eddic poems (see Gurevich 1992b, 68-76). In the context of the present discussion it is important to note that the skaldic þulur have retained and even developed this property of the older mythological þulur and prove to be not just depositories of heiti but also generators of skaldic synonyms.
Þulur lists associated with skaldic poetry
Although many of the skaldic þulur appended to Skm obviously follow the enumerations of mythical names of the type found in the eddic poems, from which the compiler(s) of the skaldic þulur adopted both the genre itself and a great deal of their lexical material, and from which they sometimes even copied ready-made combinations of names (LH II, 179-80; SnE 1998, I, xvi), these þulur are appreciably different from their forerunners. The skaldic þulur arose from other needs than those which underlay the composition of mythological þulur. While the latter were intended to preserve and transmit traditional lore, the skaldic þulur were evidently compiled for the purposes of preserving poetic vocabulary and training young poets to whom they most likely served as aids to memory or versified dictionaries (LH II, 175; Turville-Petre 1976, xli; Clunies Ross 1987, 81; SnE 2005, xxi). Even when the contents of the skaldic þulur are names of mythical beings and legendary heroes, they are not enumerated primarily for their mythological content, i.e. as the names of unique individuals remarkable for their personal destinies or deeds, but rather as interchangeable poetic synonyms meant for variation of traditional kenning-types. There can be little doubt, for example, that the vast quantity of names for sea-kings (Viking chieftains) in the opening þula of Skm (Þul Sækonunga) was above all meant to supply the lexical material for such sea-, ship- and battle-kennings as vǫllr Meita ‘field of Meiti [SEA]’, hrafn Heita ‘horse of Heiti [SHIP]’ and mót Meita ‘meeting of Meiti [BATTLE]’ (LP: Heiti; Meiti). For that purpose, any specific information, e.g. that Meiti was a brother of Heiti, was virtually irrelevant. But at the same time, closer examination of this and some other þulur containing mythical and legendary material shows that many, if not most, of the names the þulur list never actually appear in skaldic poetry. This suggests that the þulur poets were evidently striving to exhaust their subject matter and to compile lengthier lists than required for skaldic needs, and this reveals the antiquarian interest in old lore characteristic of learned activity in twelfth- and thirteenth-century Iceland.
The categories of the þulur in Skm are, however, much broader than mythical and legendary subjects, and aside from the large stock of names related to Old Norse myth and legend, they contain lists of heiti for a great variety of other subjects. These comprise above all the main heroic concepts central to skaldic poetry (women, men, battle, weapons and armour), followed by what seems to be intended to serve as an all-embracing description of the world with its diverse inhabitants of sea, land and air, the heavenly bodies and elements, to which are added several lists describing the microcosm of the human body and other subjects. This sequence includes heiti for the sea, rivers, fish, whales, earth, animals (the latter in two sets), heaven, sun, moon, day and night, wind, fire, birds, mind and heart, hand, islands, fjords and grain (for the full list, see the Table below). Thus the very range and ordering of referents whose designations are listed in the þulur go beyond the boundaries of a mere catalogue of skaldic heiti, and some of these lists must have been of little or no value to the skalds; e.g. none of the heiti for ‘rooster’ recorded in Þul Hana is used in kennings, and of the total of fifteen words enumerated in this þula only the most common term for ‘rooster’, hani, appears in skaldic poetry.
The obviously encyclopedic dimension of the þulur and their apparent intention to give a full-scale account of the world allow us to speculate about the possible influence on these lists of Latin encyclopedic works, of which Isidore of Seville’s Etymologiae was the most popular in Western Europe in the Middle Ages (Clunies Ross 1987, 88). It is likely that some of the important medieval compendia were known in Iceland already in the twelfth century and may have been used by Icelandic authors as sources for various kinds of scholarly knowledge and as models for their own writings on geography, natural history and cosmology (cf. AÍ; Clunies Ross and Gade 2012). That the compilers of the þulur were bringing together native traditional lore and learning adopted from medieval scholarly writings, whether Latin or Old Norse, is evident from the content of these catalogues of words. Thus in Þul Á (st. 3), side by side with the names of the mythical rivers Vǫnd and Strǫnd (cf. Grí 28/9), certain geographical names are recorded, including Tifr ‘the Tiber’, Vína ‘the Dvina’ and Eufrátes ‘the Euphrates’, all of which are also included in a small Old Norse treatise on the great rivers of the world based on Isidore (Hb 1892-6, cxvi, 150). It seems that in their efforts to fill the þulur with as much lexical material as possible, the compilers often did not care whether the heiti they incorporated in these lists would ever be used by the skalds. In a number of þulur, Latin words (e.g. laurus, lúna, díes, nox, corvus, gallus, gallína, aquila, scorpion), and even a few Byzantine Greek ones (cf. nis ‘night’ = classical Greek νύξ) are given along with the native terms. Clunies Ross (1987, 86) argues that ‘some of the Latin and Greek-derived words in the þulur lists may owe their presence there to school glosses’, and she finds certain similarities, both in subject-matter and in structure, between the þulur and medieval vernacular lexical glosses of difficult Latin words used for educational purposes (e.g. in the grouping of words into subject sets and sub-sets). And indeed, it seems that for at least one of the Latin words contained in the þulur it is possible to identify exactly such a source (see Note to Þul Dœgra l. 8).
The þulur preserve a large number of heiti which are not found in other sources, and many of these words are ‘characterising heiti’ (on these, see Gurevich 1992a and 1992c), probably mostly coined by the þulur compiler(s) according to certain semantic patterns borrowed from mythological name-giving (e.g. ‘noise-maker’, ‘glittering one’). Such heiti can be applied to an unlimited number of subjects rather than denote a particular referent; e.g. anything or anybody can be called a ‘noise-maker’ (the sea, the wind, birds, animals, weapons, etc.), and for that reason this type of poetic term seldom appears in skaldic verse. They do occur in most of the lists, however, and in the þulur enumerating names for animals, birds and weapons they constitute more than fifty percent of the entire word-stock. Moreover, in the þulur these heiti often have different referents, i.e. they are given in several lists, and such a polysemy must have made it difficult for the skalds to use them in poetic composition (e.g. olgr, lit. ‘noise-maker’, which is found in Þul Óðins 6/7, Þul Øxna 3/5, Þul Elds 3/1 and Þul Hauks 2/7 but not otherwise attested in poetry). It is unlikely that the invention of such poetic synonyms, which were virtually impossible to use in poetry, could have been common practice, and it is at odds with the principle formulated in Skm, according to which only heiti found in the works of the ‘main’ skalds were said to be suitable for poetic composition. Thus, judging from the actual contents of the þulur, their purpose as reference lists intended for use by the skalds often seems to be less apparent than their encyclopedic character (cf. also Bugge 1875, 245-6; SnE 1998, I, xvii). These lists nevertheless must have functioned as poetic dictionaries, and it cannot be accidental that some of the heiti, which are not attested in the early poetry but are recorded in the þulur, suddenly appear in fourteenth-century poems (e.g. by Einarr Gilsson and Árni Jónsson; see Gurevich 1992c, 40).
Relations between the two main þulur redactions
What was said above about the learned elements in these lists leads directly to the most controversial problem in þulur research, namely, the question of the interrelation between the two main groups of þulur mss. As already mentioned, mss R, Tˣ and C contain a set of thirty-five þulur in fornyrðislag, while mss A and B add to this sequence a rather long string of other þulur, four of which are inserted into the first part of the list, i.e. into the section also present in R, Tˣ and C, while the remaining þulur form a direct continuation of the set found in R, Tˣ, C. A comparison of the texts contained in A and B, even though they evince minor differences, makes it obvious that they go back to the same redaction of þulur (SnE 1931, xxxii). The most extensive collection, however, is preserved only in ms. A and contains fifty-nine þulur in fornyrðislag. Moreover, this ms. includes several þulur and þulur-like stanzas in dróttkvætt metre incorporated into the final part of the þulur sequence, which is otherwise composed in fornyrðislag (see Þul Women, Þul Islands and Þul Waves). It is impossible to say whether the lost archetype of the A, B redaction ended in exactly the same way as the version known from A, because the sequence in ms. B breaks off at the very beginning of Þul Fugla (1/4). Hence the text in ms. B is incomplete, and it cannot be ascertained whether ms. A contains independent later additions made to the final section of the þulur sequence. The following table gives the entire list of the þulur and their order in mss R, Tˣ, C, A and B. (Since original rubrics are found in only some of the mss, all the names of the þulur given below correspond to the titles used in the present edition.)
The order of the þulur in the mss of Skm
The problem of the relationship between the two main redactions of the þulur will probably never be solved. Bugge (1875, 214-16) believed that the entire list of þulur preserved only in ms. A must be a single whole and thus ought to be regarded as the original list. Bugge’s reasoning was based on the general observation that the þulur sequence contained in the R, Tˣ, C redaction looks incomplete, especially when compared to the corresponding prose sections of Skm dealing with poetic heiti. He emphasised that the very choice of subject matter included in the þulur and the logic of enumeration itself lead us to expect a continuation of the list along the lines we actually find in mss A and B. This assumption, Bugge thought, was corroborated by other indications that Snorri could have known and used the complete version of the þulur when compiling Skm (Bugge 1875, 211-15; cf. also Müllenhoff 1870-99, V, 227-8; SnE 1998, I, xv-xvi; Clunies Ross 1987, 87; Guðrún Nordal 2001, 207). In Skm (SnE 1998, I, 85), Snorri prefaces his prose listing of terms for ‘heaven’ by saying that all the words are written down (rituð), but that he has not found all of them in poetry; he adds his view that these and other heiti should only be used in poetry if they occur in the works of the hǫfuðskáld ‘chief poets’. The prose list of terms for ‘heaven’ that follows coincides almost completely with those found in Þul Himins I, II, while one of the heiti, leiptrhrjóðr ‘lightning-coverer’ (or perhaps two separate words, see Note to Þul Himins II l. 7), occurs only in the second þula, which is missing in R, Tˣ and C but recorded in A and B. It is also noteworthy that most of the terms for ‘heaven’ mentioned both in Þul Himins I, II and in the corresponding prose list in Skm do not appear in any other source and that, at least partially, the same applies to the lists of heiti for ‘sun’ and ‘moon’ following the heiti for ‘heaven’. This, as well as a striking similarity in content, makes it likely that Snorri is referring here to these very poetic lists, or probably to some earlier redaction of the þulur from which the two main extant redactions ultimately derive.
Further evidence from Þul Tungls supports Bugge’s contention that the written records Snorri had at his disposal supplied heiti for more concepts than the sequence of þulur found in the R, Tˣ, C redaction and that, in terms of subject matter, these records must have agreed with the ‘complete’ set of þulur preserved in A (and partially in B). The prose list in Skm that corresponds to Þul Tungls contains nine heiti for ‘moon’ also recorded in that þula, and two of them, skjalgr ‘squinter’ and fengari (apparently a foreign word, cf. Byzantine Greek φεγγάριον ‘moon’) are known only from these poetic and prose lists. Other prose enumerations of heiti in Skm that seem to use the þulur as their sources, besides the heiti for ‘sun’, are the lists of names for ‘bear’ and ‘stag’ and probably for ‘heart’, ‘mind’ and á hendi ‘parts of the arm’ (SnE 1998, I, 88, 108), though the relations of the latter two prose lists to the respective þulur, recorded only in ms. A, are less clear (see Introductions to Þul Hugar ok hjarta and Þul á hendi; cf. Bugge 1875, 212; SnE 1998, I, xvi). Furthermore, none of the above-mentioned prose lists of heiti in Skm is accompanied by poetic examples, although we find such examples in most other cases, which speaks in favour of the assumption that these terms were taken not from skaldic verse, but rather from already existing collections of skaldic synonyms. All this seems to support Bugge’s (1875, 212) opinion that Snorri, while composing his lists of heiti, was using one or more þulur compilations as one of his sources. Hence it is unlikely that he could have compiled these mnemonic stanzas himself or could have included them at the end of Skm; this was most probably done by somebody else and at a later stage (cf. also SnE 1998, I, xviii).
As to the origin of the skaldic þulur, Bugge (1875, 236-42) believed that, with a few exceptions (e.g. Þul Himins I), the entire list was composed by one and the same poet. He even surmised that it might be the work of Bjarni Kolbeinsson (1150-1223), bishop of Orkney (1188-1223) and the skald who may have composed Jómsvíkingadrápa (Bjbp JómsI) and perhaps also Málsháttakvæði (see Introduction to Anon Mhkv), a poem in a similar antiquarian and learned spirit as the þulur. Certain indications that the þulur could have originated in the Northern Isles rather than in Iceland Bugge found first of all among the geographical names contained in the lists of heiti for ‘island’ and ‘river’ (although the identifications of some of these rivers are dubious; cf. LH II, 182 and Notes to Þul Á). According to Bugge (1875, 219-44; cf. also CPB II, 423), such place names as Apardjón (Aberdeen), Tvedda (the Tweed), Humra (the Humber), Tems (the Thames), along with certain loanwords from Old English and Old Irish (e.g. barlak ‘barley’, korki ‘oats’; see Þul Sáðs 1/8), demonstrate a good knowledge of Scotland and northern England. This evidence, however, is hardly sufficient to support Bugge’s hypothesis about the origin of the þulur.
Bugge’s attempt to attribute the þulur to one poet and to prove their Orcadian provenance was heavily criticised by Finnur Jónsson (1893c; LH II, 176-83). From Finnur’s standpoint, these lists of heiti were probably composed by different poets and only later compiled into sequences, and according to him, Snorri is not likely to have used them because the prose enumerations in Skm to which Bugge referred contain certain heiti not recorded in the þulur and vice versa. As to the relationship between the two þulur redactions, he argued that all þulur missing in the R, Tˣ, C redaction were later additions. According to Finnur Jónsson (1893c, 495-8; LH II, 176-8), the most convincing evidence for this is the fact that the A, B redaction contains more regular stanzas than the R redaction of the þulur. Moreover, his comparison of the two main redactions of the first part of the sequence, which is preserved in all the mss (R, Tˣ, C, and A, B), showed notable changes in A and B both in terms of verse structure and in the order in which certain þulur appear in the list. This, along with several new þulur incorporated into the first part of the sequence, made it obvious to Finnur Jónsson that, in A and B, the ‘old’ list had undergone a number of revisions made by later editors, who, as he believed, probably also added the twenty-four þulur which were not present in the initial version of the þulur. Hence, the first and the second redactions may have been composed at different times. Whereas the first sequence, according to Finnur Jónsson (1893c, 505; LH II, 179), could have been composed at the latest around 1200, the second redaction could not be later than the second half of the thirteenth century.
The order of the þulur in Skj
Finnur Jónsson’s ideas about the relationships between the two main sets of the þulur served as the basis for his edition of these lists in Skj. In Skj (followed by Skald) the þulur consist of two parts, the main sequence, which is given in accordance with ms. R, and the ‘Addition’ (Tillæg) from mss A and B. In the first part of the edition, the þulur are presented exactly in the order in which they are found in ms. R, but the ‘addition’, at least at the beginning, does not reflect any extant ms. because the þulur are not given in the actual sequence in which they are recorded in the A, B redaction, but rather in the order in which each of the ‘new’ þulur appears in that list. Thus their actual position in the sequence is no longer transparent. In Skj, the four þulur (Þul Konunga, Þul Dverga, Þul Óðins and Þul Viðar) inserted in the initial list are given at the beginning of the ‘addition’ as though they belonged to a separate set of þulur, and they are immediately followed by Þul Tungls; only then follows the rest of the list as it is recorded in A and partially in B. This artificial order clearly violates the logic of the listing of the subject matter in the A, B redaction. There Þul Konunga is either at the head of the entire list (so B) or immediately after Þul Sækonunga (so A); Þul Dverga precedes the list of the names of giants and thus comes first in the set of þulur with the names of mythical beings; Þul Óðins heads the enumerations of the gods and the goddesses; Þul Viðar is given right after the list of heiti for ‘earth’; and, finally, Þul Tungls follows Þul Sólar.
The rationale behind the ordering of the þulur in the present edition
In the present edition the þulur are given throughout in the order they are recorded in A, the most complete ms., because this is the only possible way to avoid unnecessary changes and to preserve the medieval arrangement of the entire list, even though this arrangement may not be the original one and differs somewhat from the earlier extant redaction represented by ms. R. Furthermore, and as argued in more detail below, there are good reasons to believe that ms. A (and partially B) preserves a version of the þulur which is closer to the original than the sequence recorded in R, Tˣ and C, while the list in R, Tˣ, C may be incomplete and derive from a defective source ms.
The last þula recorded in the R, Tˣ, C redaction contains heiti for ‘sun’ (Þul Sólar). In mss A and B and, similarly, in the prose list of heiti in Skm, the terms for ‘sun’ are followed by the heiti for ‘moon’ (Þul Tungls). However, none of the heiti contained in the first two lines of Þul Tungls, i.e. Alskír, geisli | ok eyglóa ‘All-bright, beam and ever-glow’, actually seems to belong in the list of terms for ‘moon’. Alskír and eyglóa are also placed adjacently in the prose list in Skm (SnE 1998, I, 85 (ms. R); SnE 1848-87, II, 460 (ms. A)), which contains almost the same list of heiti, although here they are given among the poetic terms for ‘sun’ rather than for ‘moon’. None of these heiti is used in skaldic poetry and they are most likely taken from Alv 16/4, 6 (NK 126: eygló; alscír) in which both words are names for ‘sun’. As is also the case with the third heiti, geisli ‘beam’ (see Introduction to Þul Tungls and Notes to ll. 1-2), the presence of all these words in Þul Tungls must have been caused by confusion with the preceding þula of heiti for ‘sun’, since all these terms are obviously heiti for ‘sun’ and not for ‘moon’ (cf. Meissner 138). The list of Tungls heiti proper opens with the word máni ‘moon’ (l. 3; see Notes to Þul Tungls ll. 1-3; Gurevich 2008, 353-5).
There are also other arguments in favour of the assumption that the first two lines of Þul Tungls actually belong to the preceding þula. In each of the surrounding þulur the heiti are arranged in such a way that the lists begin with the most important (usually the most common) term for the subject of the respective þula. Hence the preceding Þul Sólar starts with sól ‘sun’ (R) and sunna ‘sunlight’ (A, B), Þul Dœgra puts at the very beginning the word dœgr ‘half-day’, the following Þul Himins II starts with heimr ‘world’ and the list of Veðra heiti opens with veðr ‘wind’. Beginnings of this kind are quite common throughout the entire sequence and found in more than twenty þulur. In contrast, the þulur starting with a rare poetic word are exceptional and usually have an introductory formula, which helps identify the referent of the list (e.g. Þul Øxna and Þul Orma). It is evident, therefore, that the heiti alskír ‘all-bright’ (Þul Tungls l. 1) is unusual as the opening word of a þula, and there is every reason to attribute the lines in question to Þul Sólar. This means that the last þula in the first part of the sequence does not end at the point where the list breaks off in the R, Tˣ, C redaction, since the list continues directly in the second part of the þulur preserved in mss A and B. This is important evidence in support of the unity of the entire text.
There can be little doubt that the confusion of the lists of heiti for ‘sun’ and ‘moon’ was only in part owing to the fact that these names were not used in poetry. First and foremost, this confusion can be explained by certain structural changes in the A, B redaction aimed at attaining a more regular stanzaic division. It is likely that the original division of subject matter in the þulur sequence did not necessarily coincide with an eight-line stanzaic division (this characteristic of the þula genre remains up to the present; cf. CVC: þula ‘now used of strings of rhymes running on without strophic division’; see also Halvorsen 1976b, 403-5). Many þulur contain additional lines which do not belong in regular eight-line stanzas, and some have fewer than eight lines (e.g. Þul Øxna, Þul Hrúts, Þul Hafrs and Þul Vargs). In the R, Tˣ, C redaction, Þul Bjarnar is silently followed by Þul Hjartar, which is treated as a mere continuation of the preceding list. Apparently, in the case of the heiti for ‘sun’ and ‘moon’ the lack of strophic regularity, on the one hand, and the similar subject matter of the two lists, on the other, could have led to their subsequent confusion (it is worth noting in this connection that the respective prose lists in Skm also confuse terms for ‘sun’ and ‘moon’, though that confusion involves other heiti than those in the þulur; see Note to mulinn ‘crescent’, Þul Tungls l. 4). It seems that the compiler of the archetype of mss A and B did not recognise the topical boundary between these two þulur, and, because he was used to a more regular strophic construction, he reproduced them as two eight-line stanzas rather than as a ten-line stanza followed by a six-line stanza. In fact, in the A, B redaction there is a tendency to bring the divisions between subject matter and stanzas into proper correlation. Thus, the verse structure of the A, B versions of Þul Øxna differs throughout from the R, Tˣ, C redaction, and the originally non-uniform stanzas in ms. R have been replaced by four regular eight-line stanzas in A and B (see Introduction to Þul Øxna). In all likelihood, the same type of structural change is found in Þul Elds, which is recorded only in mss A and B (see Notes to Þul Elds 1/1-4, 4/7-8; Gurevich 2008, 356-9). A similar trend can also be detected in ms. C (see Introduction to Þul Jǫtna II), though on a lesser scale.
At a higher level of composition the attempt to achieve structural regularity in the A, B redaction manifests itself in a series of noticeable changes in the order of the þulur with the names of mythical beings recorded in the initial part of the sequence. Here the compiler of the A, B redaction apparently tried to eliminate what he must have perceived as defects in the arrangement of entire sets of þulur. In R, Tˣ, C the two þulur with Jǫtna heiti ‘Names for giants’ (Þul Jǫtna I-II) are separated from one another by no less than three other lists (Þul Trollkvenna, Þul Þórs and Þul Ása I). Only after having enumerated the names of Óðinn’s sons does the compiler of the R, Tˣ, C redaction return to the list of giant-heiti. He introduces the second þula with the clause Enn eru eptir | jǫtna heiti ‘Still there are names of giants left’, and he does not continue to list the names of the gods until he has completed Þul Jǫtna II. In the A, B redaction, Þul Jǫtna I and II are merged into one list (for details, see Introductions to Þul Jǫtna I, II), and the entire set is rearranged in such a way that neither the names of the gods nor the names of the giants are separated by other heiti. The þulur containing the names of the gods have undergone visible changes as well, all aimed at attaining a more reasonable order and structure. In mss A and B they are introduced by Óðins nǫfn ‘Óðinn’s names’, a þula that is missing in the R, Tˣ, C group, and this þula is followed by the list of Heiti sona Óðins ‘Names of Óðinn’s sons’ (Þul Ása I), whereas in R, Tˣ and C the first list of heiti in this sub-set is Þul Þórs (in ms. R marked with a large initial capital). In A and B, however, the list of Þórs heiti is attached to the list of Heiti sona Óðins so that there is no division between the two þulur (presumably because Þórr was one of Óðinn’s sons), and this joint list of the names of Óðinn’s sons is immediately followed by Þul Ása II.
All of the rearrangements and structural changes found in the A, B redaction of the first part of the þulur sequence clearly indicate that a later compiler (or compilers) was diligently working on already existing lists of heiti. As argued above, there are also good reasons for interpreting in the same way certain traces of formal innovation or even of confusion of adjacent þulur detected in the second part of the sequence missing in R, Tˣ and C. On the basis of all this evidence it seems warranted to regard at least some of the four þulur which have been inserted into the first part of the list in mss A and B as a result of rearrangements in the ‘old’ list rather than as an invention of ‘new’ þulur by a later compiler or compilers. If so, Þul Konunga, Þul Dverga, Þul Óðins and Þul Viðar could well have been transferred from the second part of the þulur sequence now known only from mss A and B to new positions in the first part. As regards Þul Óðins and Þul Dverga, however, the possibility cannot be excluded that they were originally present among the þulur with mythical names and that they were then removed in a ms. higher on the stemma in the R, Tˣ, C redaction because similar lists of names for the dwarfs and for Óðinn had already been recorded in Gylfaginning (see Gylf, SnE 2005, 16-17, 21-2). The fact that other þulur in the second part of the sequence in mss A and B did not undergo such rearrangements does not necessarily imply that they were composed later. The changes in the A, B redaction represent a tendency towards regularity and not a full-scale attempt at achieving structural uniformity; e.g. only one of many þulur with animal-names (Þul Øxna) which display non-uniform stanzaic structures underwent changes in the A, B redaction while all the rest preserved the form they have in mss R, Tˣ and C.
At the same time, the supposed unity of the two parts of the þulur sequence does not preclude the possibility of later additions to the list. Eighteen of twenty-six species of whale whose names are listed in Þul Hvala are described in Konungs skuggsjá (Holm-Olsen 1983, 15-17; cf. SnE 1998, I, xviii), and that work also discusses one of the trees mentioned in Þul Viðar, namely, beinviðr ‘holly’ (see Note to Þul Viðar 1/8), which may suggest that the compiler of these particular þulur could have known Konungs skuggsjá and used it as one of his sources. If that were the case, some lists of heiti must have been composed after 1250. Even such a presumably old sub-set of the þulur as that of the names of mythical beings could perhaps contain later additions, and this might explain the somewhat chaotic composition discussed above. The introductory statement prefacing Þul Jǫtna II probably suggests that, after having finished copying Þul Jǫtna I and the next three þulur, the compiler or scribe of the source ms. for both extant redactions came across another list of giant-names and decided to include it in the þulur sequence. Similarly, there could have been subsequent additions to and rearrangements of the heiti in some þulur. Kock (NN §3140) calls attention to the fact that certain lines in Þul Fugla are hypermetrical, which may indicate that the original list of names was augmented by new heiti. It could well be that most of the Latin and Greek words which are occasionally found in the þulur also belong to a new lexical layer added to certain lists at a later stage. One of Finnur Jónsson’s (LH II, 177) arguments against the unity of the two parts of the þulur sequence is that the Latin and Greek words appear only in the so-called ‘additional’ list in mss A and B. It is difficult, however, to appraise the actual significance of this observation for the discussion of the relations between the two parts of the sequence, and Bugge (1875, 215 n. 1) is probably right when he interprets it as an accidental difference. In fact, these foreign words are found only in seven þulur (Þul Viðar, Þul Tungls, Þul Dœgra, Þul Hrafns, Þul Hana, Þul Ara and Þul Orma), and since the first part of the þulur sequence contains quite a few enumerations of mythical personal names as well as realia connected with the heroic activities of Old Norse warriors (battle, weapons, ship), this distribution should not come as a surprise. Furthermore, one of the Greek words in Þul Tungls (fengari, l. 8) is also recorded in the corresponding prose section of Skm enumerating heiti for ‘moon’, and in the prose list it appears in all mss including U (SnE 1848-87, II, 341, 460, 592), which suggests that the word could have been taken from an earlier version of the þulur used by Snorri himself. Moreover, the evidence provided by LaufE and RE 1665, in which many of the þulur are rendered as prose and often with noticeable differences from the medieval versions, shows that these lists were subject to change in subsequent periods as well.
The most difficult problems in connection with possible later additions and changes in the þulur arise in the final part of the list, preserved only in ms. A. The basis for the ordering of subject matter in those þulur which follow Þul Fugla (where B breaks off) is clearly different from that of the rest of the sequence, though it is hard to discern a unifying principle underlying the miscellaneous lists of heiti collected there. It has been suggested (Clunies Ross 1987, 88 n.) that this might have been a ‘female’ unifying principle, because the final part contains heiti for different groups of female beings, such as women, valkyries, giantesses or troll-women (in Grýlu heiti ‘Names for Grýla’, though this þula in fact enumerates more male than female names), as well as particular words and concepts that occur in kenning-types associated with these female beings. However that may be, it is evident that the structure of the final part of the þulur list in ms. A differs substantially from that of the preceding parts. Some of the þulur recorded there might have been expected earlier in the list (e.g. Þul Eyja, Þul Fjarða and Þul Grýlu, since Grýla is mentioned among the names of troll-women), and sometimes their content overlaps with that of certain þulur in the first part of the sequence (e.g. the eight names for valkyries enumerated in Þul Valkyrja are also recorded in Þul Ásynja 4). Hence one cannot rule out the possibility that at least some of the þulur in the final part could be later additions (on heiti for ‘mind’, ‘heart’, and ‘arm’ which have parallels in the prose list in the final part of Skm, see Guðrún Nordal 2001, 236, who argues that the popularity of body imagery was increasing after Snorri’s time). It is particularly striking that this section also contains a list of base-words that could be used in kennings for ‘woman’ (the so-called Kvenna heiti ókend ‘Women-names without a determinant’; see Introduction to Þul Kvenna II), in contrast to the enumeration of heiti for ‘women’ proper (Þul Kvenna I) given in the first part of the þulur sequence. Furthermore, several stanzas in dróttkvætt metre have been incorporated in the þulur list apparently to illustrate particular enumerations of heiti, such as two anonymous mnemonic stanzas exemplifying heiti for ‘woman’ (Þul Women), one anonymous stanza (Þul Islands) and two stanzas attributed to Einarr Skúlason (ESk Lv 8-9), all of which contain the names of various islands, and one stanza with names for waves, Nǫfn Ægis dœtra ‘Names of Ægir’s daughters’ (Þul Waves). Bugge (1875) was inclined to give preference to ms. A, while Müllenhoff (1870-99, V, 223-6) argued that the original set of the þulur is given in B and that the entire final part of the sequence in A is a later addition.
With the exception of the final set of þulur in A, whose provenance is unclear, and some possible later additions, the two parts of the þulur sequence as preserved in mss A and B appear to form a coherent whole. This follows not only from the structural evidence discussed above, but also from the apparent uniformity of content in many þulur regardless of where they are placed in the entire list, because both parts contain the same types of lexical material, some of which is peculiar to the þulur (see e.g. the discussion of ‘characterising’ heiti above).