Chapter 3 The Norse anthropo-toponyms of Lewis
3.2 Extensive analysis of Norse data .1 Background .1 Background
This section will form the ‘extensive’ analysis of the Norse Lewis material as outlined in 1.3.2. It will also further expand on the methodology by discussing the principles that have been used to tackle the uncertainties involved in studying complex onomastic data. The acknowledgement of uncertainty is an integral part of studying the Norse data presented here and forms a significant part of the methodology used. This is primarily reflected by the use of the scale of certainty introduced in 1.3.2.3 which has been used in 3.1 and will be used here. Although Norse Lewis toponyms have been investigated previously, personal names, despite providing a crucial element to the study of these names, have been confined to the periphery of these discussions. It is likely that this is at least partly due to the
difficulties of firmly identifying the etymology of the specific elements in these toponyms.
Because of this, there is no clear overview of a potential stock of personal names found in the toponyms, and this section attempts to fill that gap. It should be pointed out that, in some instances, some of the personal names listed may look suspiciously similar to each other and the usefulness of trying to distinguish between names that are ultimately variants of the same personal name may seem unnecessary. However, minor differences in name formations can have major implications in terms of the context of naming. For example, are the Þór- specifics found in a number of Lewis toponyms a reflection of a masculine name such as Þórir, a feminine name such as Þóra, or even a theonym, Þórr, referring to the Norse pre-Christian god? Based on this, this section will provide an overview of the potential name-stock in the Norse data to form a basis for discussions in subsequent chapters.
3.2.2 The data
As a starting point, it should be noted that because many of the interpretations remain problematic, the scale of certainty will, out of necessity, underpin the presentation and analysis of the data. This can be emphasised by the fact that only five of the listings are given as ‘Certain’ according to the primary scale and, even then, the question of what the personal name is, according to the secondary scale, is often uncertain (see table 3.1). For
155 example, *Þolfstaðir (North Tolsta) has been categorised as ‘Certain’ on the primary scale and, although it most likely contains the personal name Þólfr, there are some additional possibilities that need to be included, resulting in: pn Þólfr, m. (b.) or pn Hólmr, m. (d.) or pn Hǫlgi, m. (d.) or pn Holfr, m. (d.) or *Talga ‘the cutting one’ (d.) + ON staðr ‘a stead, place, abode’ (a.). Finally, the syntax of the original Norse coinings is worth investigating briefly. In the material available to us, the structure is almost entirely uniform and consists of toponyms coined according to the typical syntax of Germanic languages with specific +
generic and the entries listed in Table 3.1 all demonstrate this pattern.
Thus, in *Karlavágr we get the specific element pn Karli and the generic element ON vágr, m., ‘a creek, bay’. The only possible exception to this pattern is Torsuigabac, where we might see an existing Norse coining being incorporated into a new one, giving ON *Torsvík + ON bakki ‘a bank of a river, water, chasm, etc.’ The distribution of toponyms, as outlined in Fig. 3.5 is, as is to be expected, highly concentrated along the coastline. It is also worth noting that it largely corresponds with Oftedal’s (2009 [1954], 28-9) map, with a concentration of names around Loch Eireasort, northern Ness, Bernera and Carloway.
3.2.3 The name-stock 3.2.3.1 Initial observations
The first observations to be made about the names presented in Table 3.2 must include an overview of the certainty of the interpretations. Although unlikely and rejected toponyms are not included here, the material is also analysed according to the secondary scale, where the likelihood of a certain personal name being present is reflected, as in the case of
*Þolfstaðir above. This means that the number of personal names listed will be
considerably higher than the total number of toponyms, since a single toponym may have three or four alternative possible personal names present. It will be noted that only one personal name has been placed in category a; Karli. The name in question is a well-attested one, found in a toponym applied to a major feature. There is,
Table 3.1 Reconstructed Old Norse formations classified as a. Certain on the primary scale of certainty
*Þolfstaðir (North Tolsta) NB535473
*Karlavágr (Carlabhagh) NB208423
*Þórissǽtr (Beinn Thòrshader ) NB246427
*Þolfstaðir (Tolastadh a’ Chaolais) NB194386
*Þórey (Eilean Thoraidh) NB422201
156 however, a considerable portion of probable names (19.0%). These names primarily
consist of instances where the toponym in question has been categorised as a. Certain on the primary scale, but where there is more than one option as to what the personal name in question is, as in *Þolfstaðir. The bulk of the data is formed by category c., with 65.3% of the entries. They mainly consist of names where a personal name is a viable option, but where a common noun would be just as likely, or where contextual evidence is too sparse to effectively apply the criteria used in the scale. For example, Giurshadir, is given as: pn Gyrðr, m. (c.) or ON gjǫ́ f. ‘ravine, gully’ (c.) + ON setr, n. ‘a seat, residence, mountain pastures, dairy lands’ (a.). Either interpretation of the specific element appears to fit, but because of the lack of contextual evidence and early forms, it is difficult to draw any firm conclusions.
Fig. 3.5 Distribution of reconstructed Norse anthropo-toponyms classified as Certain-Maybe on the primary scale. © Map data 2017 Google
Category d., which shows personal names that are unlikely to be found in the place-names in question, but where the possibility cannot be entirely excluded, form 14.9% of the total
157 number of potential personal names. Nevertheless, the overall picture presented is one of considerably greater variety than one might have expected. In total, fifty-seven potential personal names have been identified. As discussed, many of these reflect different possible interpretations for the same toponym and this number should not necessarily be regarded as representative of any real Norse stock of personal names in Lewis. However,
considering only the certain and probable names, at least fifteen different names can be identified. Considering that the full corpus consists of sixty-seven toponyms, this is not an insignificant number.
Table 3.2 Possible personal names found in Norse Lewis toponyms and their certainty
pn a b c d pn a b c d pn a b c d
158 3.2.3.2 The masculine names
Masculine personal names form the majority of the name-stock presented here. It should be noted that potential surnames, nicknames and theonyms such as Kárinn (?surname) and Gasi (nickname) have also been included here. Looking at categories a-c., excluding the unlikely interpretations, we can note that variants of Þór- form a significant number of the stock. The large number will partly be a result of several of these personal names being given as alternatives for the same toponym, but they nevertheless appear to form an important part of the name-stock. In addition to this, two of the five toponyms categorised as ‘Certain’ contain Þór- (table 3.1). If the interpretation of the six toponyms containing Ull- as the pn Ulli or the theonym Ullr is correct, this would also make this a frequent name. However, the wider context of these names needs to be further investigated and they will be discussed in 5.5. In terms of the remaining name-stock, a considerable variety in the personal names can be found, even when the fact that several different personal names are often listed under the same toponym is taken into account, represented by frequently attested names such as Eiríkr, Egill, Gunnarr, Sveinn, and Þólfr.
3.2.3.3 Feminine names
Although most of the personal names listed are (not surprisingly) masculine names, there are some notable exceptions. Although feminine names form a significantly smaller number than the masculine ones (18.2% of the total number), there are several interesting factors to consider here. The most notable feminine name is perhaps Iórunn, which has already been discussed in 1.2.3.5.138 However, Guðrún is the only feminine name which is categorised as b. Probable and is the only entry where the feminine personal name is given as the only option. Generally, when there is a possible feminine present in a toponym there is also a masculine (and perhaps more likely) alternative. For example, in Torastaidh, the feminine name Þóra is given as a possibility for the specific element, but the other masculine names would appear to be just as likely. These names will be further discussed in 5.5.
138 See 3.1 Eòrapaidh, Eòradal, Eòranish Mhor, Eorshader.
159 3.2.3.4 Theonyms
There are two potential theonyms recorded in the name-stock139: Þórr: ‘The god Thor, the god of thunder’ (Cl.-Vig.) and Ullr: ‘the name of one of the gods, the step-son of Thor’
(Cl.-Vig.). These have been included alongside the other personal names (Table 3.2).
Perhaps the most significant of these names is Þórr, particularly since the exact form of this name appears to only be found as a theonym. However, the identification of this name remains problematic since a different variant of Þór- could in many cases easily be used to explain the specific element in a toponym. Þórðr, which, in a Norwegian context, appears to develop into Tor, ‘af Mandsnavnet Tor (det gamle Þórðr) [from the masculine name Tor (the old Þórðr)]’ (NG, vol.4, 45), is a particularly likely candidate. There are also several issues involved in the case of Ullr. Firstly, the interpretation of the specific element in these toponyms as a personal name is far from certain, as outlined in 3.1. Secondly, even if a personal name is present, it may represent the non-theonym Ulli. The possibility and significance of theonyms being present in the Norse material will be further discussed in Chapter 5.
3.2.4 Generic elements 3.2.4.1 Background
As previously discussed (see 1.2), scholars such as Nicolaisen have previously made a close association between permanent settlement and habitative generics. Along with the assumption of the close relationship between ownership and anthropo-toponyms (see 1.2.1.2), we would expect the highest numbers of personal names being attached to habitative generics such as ON bólstaðr ‘a farm’, ON staðir ‘a stead’, ON bý(r) ‘a farm’, and potentially ON setr ‘a seat, a residence, mountain pastures, dairy lands’. However, as will become evident, this is often not the case. Additionally, one might expect certain topographical generics to be relatively common – some features are more likely to have a greater abundance of recorded forms and are generally more frequently mentioned in sources. This is particularly the case of major features in the landscape, such as islands, that are commonly used as topographical and navigational descriptors and we would therefore expect a generic such as ON ey ‘an island’ to be relatively common. Here, the commemoration may not be as straightforward as in habitative generics. Some potential
139 Excluding the highly doubtful entries with Skalli, who according to Lind (NID) is the name of a iǫtunn ‘a giant’, see appendix 2 Scaladale.
160 factors to consider include association (secular and religious), discovery, usage, or a
particular event giving rise to the place-name in question. This will be further discussed in Chapter 5. Finally, in the few cases where the generic element is not certain (5.9%), they have been excluded from the current table, but can be found with their respective entries in 3.1.
Table 3.3 Generic elements in the Norse anthropo-toponyms of Lewis
element No. feature
fjall, n. ‘a fell, a mountain’ 8 (top.)
staðir ‘a stead, a place, an abode’ 8 (hab.)
dalr ‘a dale’, m. 7 (top.)
setr, n., ‘a seat, a residence, mountain pastures, dairy
lands’ 7 (hab.) or
(agric.) vík, f. ‘a small creek, an inlet, a bay’ 6 (top.)
ey, f. ‘an island’ 4 (top.)
nes, n. ‘a ness’ 4 (top.)
vatn, n. ‘water’ 3 (top.)
klettr, m. ‘a rock, a cliff’ 2 (top.)
vágr, m., ‘a creek, bay’ 2 (top.)
bólstaðr ‘a farm’ 1 (hab.)
bý(r) (bær), m., ‘a farm’ 1 (hab.)
fjörðr, m., ‘a firth’ 1 (top.)
gerði, n. ‘a place girded round, a hedged or fenced field, a
garth’ 1 (hab.)
hamarr, m., ‘a hammer, a crag’ 1 (top.)
holt, n., ‘a wood, a copsewood, a coppice’ 1 (top.) möl, f., ‘pebbles, worn stones, i.e. the bed of pebbles on the
beach or in a river’ 1 (top.)
pollr, m. ‘a pool’ 1 (top.)
sker, n., ‘a skerry, an isolated rock in the sea’ 1 (top.)
steinn ‘a stone’ 1 (top.)
161 topt, f. ‘a green tuft or knoll, a green, grassy place; a place
marked out for a house or building, a toft’ 1 (hab.) varða, f., ‘a beacon, a pile of stones or wood to “warn” a
wayfarer’ 1
(top.) (but human-made)
3.2.4.2 The generic elements
Habitative generics that can be identified with relative certainty make up 30.2% of the total number of generic elements found in the Norse Lewis data if setr is included (Table 3.3).
The overall numbers for habitative generics appear to be surprisingly low. Particularly in the case of bý(r) ‘a farm’ and bólstaðr ‘a farm’, with one entry each, one might have expected larger numbers, but this pattern has been previously noted for other Norse Lewis toponyms by scholars such as MacAulay (1971-2, 335). Additionally, the use of bólstaðr in our area overall is comparatively sparse, as previously noted by Gammeltoft (2001a, 82) who records a total of sixteen names in bólstaðr for Lewis. It might also be suspected that the interpretations would favour habitative generics anyway, since the presence of a habitative generic element might lend strength to the certainty of a personal name being present. These numbers show that this may be an inaccurate assumption, since
topographical generics are more frequent. The key point here is that the presence of a topographical generic does not exclude the possibility that a feature is used to denote a settlement, as argued by Kruse (2004, 105) and it certainly does not exclude the possibility of a personal name being present. On the contrary, the Scandinavian evidence would point to the opposite being true, especially if we consider toponyms coined in early phases of settlement and land-claiming, considering major early settlements such as Dal and Nes (Kruse 2004, 105). Nevertheless, the importance of habitative generic elements should not be ignored and two of the habitative generic elements, staðir ‘a stead, place, abode’ and setr ‘a seat, residence, mountain pastures, dairy lands’ provide two of the most frequently used elements in the data (23.8%). The topographical generics that can be identified with certainty form the majority of the generic elements found in the Norse data (69.4%). In fact, the three most frequently occurring topographical generics, ON fjall, ON dalr, and ON vík form 33.3% of the total number. However, it is also worth noting that out of the five entries given as a. Certain in Table 3.1, three contain habitative generics. As stated in the hypothesis, we would expect prominent features in the landscape to be more frequently identifiable, particularly islands. Therefore it is essential to remember that especially
162 interpretations with ON ey categorised as certain can to some extent be accounted for by their very nature as major landscape features. Additionally, as discussed above, it is just as likely that several of the features associated with topographical generics represent
settlements. In light of the patterns of Scandinavian settlement mentioned above, we might be particularly likely to find settlement names represented by elements such as ON dalr and ON vík, several of which may have formed appropriate locations for important settlements.
163