Hugvsindasvi

We need to talk about Gefjun

Toward a new etymology of an Old Icelandic theonym

Ritger til MA-prfs Viking and Medieval Norse Studies

Kevin French
September 2014

Hskli slands
Hugvsindasvi
Viking and Medieval Norse Studies

We need to talk about Gefjun

Toward a new etymology of an Old Icelandic theonym

Ritger til MA-prfs Viking and Medieval Norse Studies

Kevin French
Kt.: 260188-4299

Leibeinandi: Haraldur Bernharsson

September 2014

Abstract
The name of a deity often reveals something of their character, and can shed light on
obscure elements in the mythology. Unfortunately the prehistory of a word is itself
often obscure, and care must be taken not to project a meaning onto a word that was
never truly there. The goddess name Gefjun has long been considered to mean the
giving one, and the goddess interpreted as a generous deity of vegetation, but the
superficial similarity of the name to the word gefa v. to give is not reason enough to
come to such a conclusion. As Sturtevant (1952, 1667) pointed out, the root-final j
in Gefjun would have caused i-umlaut of the root vowel, indicating an earlier *a.
There is much evidence to suggest a connection to OI gfugr adj. noble and Goth.
gabei f. riches but the nature of that connection is unclear. One possibility is that the
name Gefjun is a deverbal from an unattested *gefja, pret. *gefjai. Another is that it is
a Hoffmann formation derived by the same manner as inn, jann, and possibly
some goddess names as well. This essay consideres the etymology of Gefjun through
comparative linguistics and investigation of Icelandic manuscript sources. In the end it
is concluded that the word is most likely a Hoffmann formation meaning she who
rules/pertains to *ga, and possible meanings of *gai are considered.

grip
Nfn goum lsir oftast einhverju r skapger goanna og geta skrt gosguleg
efni sem annars vru margbroti og villandi. v miur getur fruming ors
stundum veri hulin sjlf og a er htta a s merking sem finnst hafi aldrei veri
til alvrunni. Gyjunafni Gefjun hefur lengi veri tali a s sem gefur og ess
vegna er Gefjun talin gjafmild grurgyja, en yfirborssamanburur orsins Gefjun
og sagnorsins gefa er ekki sta til a halda eirri kenningu til streitu. Eins og
Sturtevant (1952, 1667) benti , bkstafurinn j sem stendur enda stofnsins hefi
valdi i-hljvarpi rtarinnar og vsar til frumnorrnnar *a. Til er mrg rk fyrir a
Gefjun s skyld fornslenska orinu gfugr og gotneska orinu gabei f. auur en s
tenging er ekki skr. a m vera a nafni Gefjun s leitt af sagnori sem hefur ekki
haldist fornslensku en hefi veri *gefja, t. *gefjai. Annar mguleiki er s a
Gefjun s Hoffmann-myndun og forma me lkum htti og inn, jann og
kannski nnur gyjunfn. essi ritgeri ltur orsifjar nafnsins Gefjun me v a
nota samanburarmlfri og rannsknir handritum. lok ritgerar er a
niurstaan a ori s lklegast Hoffmann-myndun og ir s sem rur/tengist
*ga, og mgulegar ingar orsins *ga eru grundaar.

Table of Contents
Acknowledgements ............................................................................................ 1
List of Abbreviations .......................................................................................... 2
1. Introduction ....................................................................................................... 3
2. Gefjun in Mythology ......................................................................................... 9
2.1. Old Icelandic Sources ................................................................................. 9
2.2. Glosses of Classical Goddesses ................................................................. 18
2.3. A Brief History of Research ...................................................................... 20
2.4. Conclusion ............................................................................................... 24
3. Feminine n-final substantive formation ........................................................... 25
3.1. Deverbal abstracts in *-ni- ....................................................................... 25
3.2. The Hoffmann Formation ........................................................................ 37
3.3. Other suffixes ........................................................................................... 45
4. Goddess names in Germanic Languages .......................................................... 47
4.1. Old Icelandic ............................................................................................. 47
4.1.1. Names ending -un ............................................................................ 47
4.1.2. Names ending -yn ............................................................................ 54
4.1.3. Names ending -n .............................................................................. 62
4.2. Matronae and Latin Devotional Epigraphy ............................................... 75
4.3. Conclusion ................................................................................................ 85
5. Old Icelandic Manuscript Evidence ................................................................. 86
5.1. Attestations of Gefjun in Icelandic Manuscripts........................................ 86
5.2. Interpreting the data ................................................................................. 89
5.3. Summary .................................................................................................. 92
6. Reconstructing Gefjun ...................................................................................... 94
6.1. The first component: Gefj- ....................................................................... 94
6.2. The ending -un on .............................................................................. 111
7. Conclusion ..................................................................................................... 114
Works Cited ................................................................................................... 120
Appendix A: Collected Manuscript attestations of Gefjun ............................. 133

Acknowledgements
Any attempt to express the depth of my gratitute to the uncountable
people who have made my education possible is destined to inadequacy. My
friends and family both old and new have been unbelievably supportive. I never
would have set foot in Iceland without the help of my grandfather, Paul Flynn,
and my parents Kevin and Colleen French. I would like to thank Bergur
Ketilsson, Gunnur Gunnarsdttir, Svands Bergsdttir, Ragnar Gslason, Danel,
var, and Sigds for making me feel like an honorary Icelander. I am extremely
grateful to Lars and Anne rlund for going above and beyond for me during
my time in Denmark. I owe a debt of gratitude to the faculty of both the
University of Iceland and the University of Copenhagen, and I want to single
out my advisor Haraldur Bernharsson, not only for his unwavering patience,
dedication, and hard work, but also for awakening a love of language in me that
I never knew I had. I dare not try to enumerate all of the friends who have
helped me some cannot go unsaid. To Anna Katrn Jnsdttir, Brendan
Driscoll, Yoav Tirosh, Annie Humphrey, and most importantly Heirn
Bergsdttir, I am very lucky to have your friendship.

List of Abbreviations
acc.

accusative

ODan.

Old Danish

adj.

adjective

OE

Old English

adv.

adverb

OF

Old Frisian

c.

common gender

OGut.

Old Gutnish

cf.

confer

OHG

Old High German

Dan.

Danish

OI

Old Icelandic

dat.

dative

OIr.

Old Irish

f.

feminine

OS

Old Saxon

Far.

Faroese

OSw.

Old Swedish

gen.

genitive

PGmc.

Proto-Germanic

Goth.

Gothic

PIE

Proto-Indo-European

Hitt.

Hittite

pl.

plural

Lat.

Latin

PN

Proto-Norse

Lith.

Lithuanian

sing.

singular

m.

masculine

Sw.

Swedish

MLG

Middle Low German

v.

verb

n.

neuter

pret.

preterite

NNo.

Nynorsk

pres.

present

No.

Norwegian

nom.

nominative

1. Introduction
In Gylfaginning chapter thirty-five,1 the reader is introduced to fourteen Old
Norse goddesses, many of whom are not otherwise described in the mythology.
It is clear that the author, ostensibly Snorri Sturluson, knows much more about
some of these figures than others, and for eight of these goddesses he supports
his claims about who they were by connecting their name to a word that was
contemporary to his own language. For example, the name of the goddess Lofn,
who he says arranges marriages that were previously forbidden, is said to be the
origin of Icelandic lof praise, because she is held in such high esteem by the
benefactors of her devine intervention (ed. Faulkes 1998, 29). Snorris
examples of folk-etymology are not taken seriously these days, but the study of
linguistics has come a long way since his time, and the careful application of
historical linguistics may be able to reveal lost information about the names of
deities and in turn, the deities to whom the names were given.
The relationship, or at least possibility of a relationship, between the
conception of a deity and the name by which the god is known (its theonym)
has a long history in the study of Norse deities. Perhaps the most famous
example is Adam of Bremans definition of the word Wodan, id est furor
(that is, furor; ed. Impensis Bibliopolii Hahniani 1876, 1747), corresponding
to Icelandic inn and supported by the gods associations with battle, magic,

All references to Gylfaginning and Skldskaparml (including the Nafnaulur) utilize the
editions by Anthony Faulkes (1998a and 1998b) unless otherwise specified.

and poetic inspiration. The name Freyr, cognate to Old English fre, meaning
lord (sgeir Blndal Magnsson 1989, 208 [Freyr]; Bosworth 1921, 331) and
indeed Freyr is portrayed in close mythological proximity to powerful human
rulers, specifically the Ynglingar dynasty in Heimskringla (ed. Bjarni
Aalbjarnarson 1941). These are examples where a connection between
theonym and conception of the deity is demonstrated with narrative evidence,
and it follows that if a scholar can interpret the names of other figures whose
possible roles in pre-Christian religion are poorly documented, such as Tr, it
might be possible to discern attributes which are not recorded in the primary
sources. Indeed, rr, who is widely recognized as the thunder god and whose
name is cognate to English thunder, is hardly discernible as related to thunder
by Old Icelandic mythology alone (Liberman 2012, 8). Some very enigmatic
deities such as Loki and Heimdallr have inspired a great variety of different
theories and interpretations due to the obscurity of their names etymologies.2
Though the goddess Gefjun might justifiably be called enigmatic, the
etymology of her name is not generally taken to contribute to the confusion. A
definition of exactly or nearly the giving one seems to be taken for granted by
the majority of scholarship.3 This proposal holds that the word Gefjun is derived
from the verb gefa to give, and reflects a function related to the fertility of the

A comprehensive overview of etymological research into the name Loki is presented in

Snorri and Saxo on tgaraloki, with Notes on Loki Laufeyjarson's Character, Career, and
Name by Anatoly Liberman ([1992] 1994); for Heimdallr see for example: Dronke 1997,
107.

See for example Olrik 1910, Clunies Ross 1978, Simek 1993, Davidson 1996, and others.

earth (Clunies Ross 1978, 153; Simek 1993, 1012 [Gefjon]). Additionally, the
goddess Freyja can also be referred to by the name Gefn, often taken to be
identical in meaning to Gefjun and possibly revealing a common origin of the
two goddesses which was artificially split by later traditions. There is a problem,
however. In 1952, Albert Morey Sturtevant wrote a short article demonstrating
that unlike the word gefa, the root vowel e in the word Gefjun can best be
explained as deriving from an earlier *a, which arrived at its present state by
means of i-umlaut, triggered by the following j (Sturtevant 1952, 1667. A
Proto-Norse vowel *e in this position would be expected to yield OI *Gifjun.
Compare for example ilja v. to cover with boards, which shares a root with
el n. inner wool and OE el n. plank, and can be reconstructed to a Proto
Germanic. *eljan-. As a result, the first portion of the name is best considered
descended from a PN, and ultimately PGmc. *gaj-. This appears to be related
to names appearing in Romano-Germanic devotional inscriptions dedicated to
the Gabiae, Alagabiae, and others in West Germanic-speaking areas, primarily
in the Rhine region of modern Germany (Beck 2009, 668). These names or
name components featuring Gabi have traditionally been interpreted the
same way as Gefjun, to mean givers (Beck 2009, 668; Neumann [1987] 2008,
263), even in this environment of greater phonological transparency. While it is
possible that the words Gefjun and gefa are related in some way, the nature of
that connection is far from transparent, and a direct derivation must be
considered untenable.

The objective of the present work is a reanalysis of the name Gefjun.

Considering the relationship that tends to be observable between deities and
their names, a description of Gefjun in Norse mythology and its implications
for Germanic pre-Christian religion will be discussed in chapter two, though a
thorough analysis of the actual goddess herself is a subject deserving of its own
dedicated study and will herein be done only insofar as it is useful to explaining
the name.
Chapter three is an investigation of Germanic word-formation relating to
feminine substantives ending in -n. Two methods of derivations in particular
are investigated in close detail, as awareness of their formation and function is
necessary throughout the rest of the paper. The first of these is the derivation of
abstract substantives from verbs, especially weak verbs, such as skipun f. order,
arrangement from skipa v. to arrange. Additionally, those substantives which
derive from -verbs are analyzed for the orthographic representation of their
unstressed vowel. The second method of word-formation is the thematicized
Hoffmann suffix, an Indo-European morpheme that was productive in the
formation of both male and female deity names, among other things, which has
reflexes in many Indo-European languages. The chapters which follow will
make frequent use of information presented in chapter three.
In chapter four, other goddess names in Old Icelandic and other Germanic
sources will also be analyzed, and again, mythological information will only be
utilized as a means to support a general understanding of the formation of their
names. The first half of the chapter categorizes Old Icelandic goddess names by
6

the endings -un, -yn, and -n, and the second half examines West Germanic
votive epigraphy, particularly names of goddesses to which inscriptions were
dedicated in the first few centuries, A.D., which may be linguistically related to
the goddess names analyzed in the first half of the chapter.
In chapter five, the word Gefjun and its attestations in Old Icelandic
manuscripts will be examined in close detail. Attention is paid specifically to
the representation of the vowels and to the inflectional morphology. It is found
that a disproportionate amount of attestations represent the second-syllable
vowel with the letter o when compared to other instances of second-syllable
unstressed u such as deverbal abstracts observed in section 3.1.3.
In the sixth chapter an attempt is made to describe the etymology of
Gefjun. It begins with an analysis of the first component of the word, gefj-, by
considering evidence within Old Icelandic and other Germanic languages, and
then also within a broader Indo-European perspective. Then all evidence
gathered in previous chapters is utilized in attempt a reconstruction of the fullyformed name that would later develop into Old Icelandic Gefjun. It is
concluded that the most likely Proto-Norse form is *gajanu, and that it was
formed by means of a Germanic reflex of the thematicized Hoffmann suffix.
Possible meanings are considered but full confidence in a precise definition
remains elusive.
In the concluding chapter, the implications of the reconstruction
proposed in the previous chapter are considered in relation to the mythological
figure Gefjun, as well as the possible relationship between the names Gefjun and
7

Gefn. A situation is proposed whereby the two names could share a common
origin, but uncertainty regarding the etymology of Gefn prevents a final
decision from being made.
Before proceeding, it is necessary to explain that, for lack of certainty
regarding the origin of the name, Gefjun is herein rendered following standard
normalization of Old Icelandic, identically to the form found in the slenzk
Fornrit edition of Heimskringla (ed. Bjarni Aalbjarnarson 1941). The name
appears to be made up of two parts which will herein be referred to as the
root and the suffix, which unless otherwise specified are considered to
correspond to Gefj- and -un, respectively. It is important to be aware that this is
a matter of convenience rather than an expression of belief concerning the
formation of the name. Although Gefj|un does seem to be the most likely
segmentation of the word, to analyze the boundary as Gef|jun is not impossible,
nor is it certain that the final part of the word is indeed a suffix.

2. Gefjun in Mythology
2.1. Gefjun in Old Icelandic Sources
Before analyzing the word Gefjun, it is productive to discuss the
mythological figure to whom it is attached.

2.1.1. The Plowing of Sjlland

Gefjun is attested in a verse attributed to Bragi inn gamli Boddason,
quoted both in Gylfaginning and Ynglinga saga. The verse as normalized and
translated by Margaret Clunies Ross (1978, 1556) is as follows:
Gefiun dr fr Gylfa

Gefjon, rejoicing in her patrimony,

gl, diprul, la,

deeply

svt af rennirauknum

increase from Gylfi, so that the

rauk Danmarkar auka.

hauling beasts of burden steamed.

ru xn ok tta

wise,

drew

Denmarks

The oxen had four heads and eight

ennitingl, ars gingu

fir vineyiar vri

forehead-ornaments (tingl), where

they went before the extensive

valrauf, fiogur haufu.

plunder of the meadow-island.

Details of the verse are subject to some debate. Manuscript variants make it
difficult to determine the identity of the word here given as la and interpreted
as patrimony. The meaning of djprull is also uncertain; Hilda Ellis
Davidson (1999, 54) translated it deep circle of land. Though some details are
debateable, the most important parts are easily discernible. Gefjun took
something valuable, most likely fruitful land, from Gylfi by means of four oxen,
who greatly exerted themselves. This was an explicitly aggressive act, indicated

by the use of the word valrauf spoils, robbing the slain (in battle), and
Denmark benefited or literally grew as a result.
Gylfaginning and Ynglinga saga add context, differing slightly in their
accounts but not conflicting. Gylfaginning begins as follows:
Gylfi konungr r ar lndum er n heitir Svj. Fr honum
er at sagt at hann gaf einni farandi konu at launum skemtunar
sinnar eitt plgsland rki snu at er fjrir xn drgi upp dag ok
ntt. En s kona var ein af sa tt. Hon er nefnd Gefjun. Hon
tk fjra xn noran r Jtunheimum, en at v ru synir jtuns
ok hennar, ok setti fyrir plg. En plgrinn gekk sv hart ok
djpt at upp leysti landit, ok drgu xninir at land t hafit ok
vestr ok nmu staar sundi nokkvoru. ar setti Gefjun landit ok
gaf nafn ok kallai Selund. Ok ar sem landit hafi upp gengit
var ar eptir vatn; at er n Lgrinn kallar Svj. Ok liggja
sv vkr Leginum sem nes Selundi. (ed. Faulkes 2005, 7)
King Gylfi ruled that land which is now called Sweden. Of him it
is said that he gave a wandering woman, in exchange for her
entertainment, a plowland in his kingdom which four oxen could
drag up in a day and a night. But that woman was one of the
sir. She is called Gefjun. She took four oxen from the north
out of Jtunheimar, and they were her sons with a giant, and set
them before the plow. And the plow went so hard and deep that
the land came free, and the oxen dragged the land out into the
sea and westward, and claimed a place in some sound. Gefjun set
the land there and named it, and called it Selund (Sjlland). And
there where the land had come up was left a lake; that is now
called Lgrinn (The lake) in Sweden. And the inlets in Lgrinn
lay as the headlands on Selund. (authors translation)
The story as it stands in Ynglinga saga chapter is as follows:

10

 sendi hann Gefjun norr yfir sundit landaleitan. kom hon

til Gylfa, ok gaf hann henni eitt plgsland. fr hon
Jtunheima ok gat ar fjra sonu vi jtni nkkurum. Hon br
eim yxnalki ok fri fyrir plginn ok dr landit t hafit
ok vestr gegnt insey, ok er at kllu Selund. ar byggi hon
san. Hennar fekk Skjldr, sonr ins. au bjoggu at Hleiru.
ar er vatn ea sj r eptir. at er kallat Lgrinn. Sv liggja firir
Leginum sem nes Selundi. (ed. Bjarni Aalbjarnarson 2002, 1415)
Then he [inn] sent Gefjun north over the sound to search for
land. She came then to Gylfi, and he gave her one plowland.
Then she went to Jtunheimar and begat four sons with some
jtunn; she changed them into the shape of oxen, and brought
them before the plow and dragged the land out into the sea and
west across from insey, and it is called Selund; she settled
there after that. She was married by Skjldr, son of inn; they
lived at Hleira (Lejre). There is a lake or sea left behind. It is
called Lgrinn (the lake). The fjords in Lgrinn lay as the
headlands in Selund. (authors translation)
The events described in Gylfaginning take place within the frame narrative; that
is, treated as taking place within human history. At this point in the story,
Gefjun is a mortal human woman albeit with great skill in magic rather
than an actual goddess such as how she is later described by Hrr, Jafnhrr, and
rii. Gylfaginning explains that Gefjun came to Gylfi, the king of Sweden,
disguised as a beggar woman (farandi kona) and offered skemtun entertainment
in exchange for one plowland, the amount of land which she could plow in one
day and one night. She used four oxen, who were actually her sons with an
anonymous giant. When they pulled the plow, it dug so wide and deep that the

11

land detached and they dragged it into the sea, where it became the island of
Sjlland in Denmark. Ynglinga saga presents the additional details that this was
done under inns instructions, and that Gefjun went to Jtunheimar after
making this arrangement with Gylfi specifically to beget children for this
purpose. It also adds that she then married inns son Skjldr and they settled
at Lejre, where they became the progenitors of the Skjldungar dynasty. This
story is also told in Vlsungsrmur (ed. Finnur Jnsson 1896, 4445 [strophes
2937]), the beginning of which is built on Prologus, Gylfaginning, and
Heimskringla. There is little variation from Snorris versions of the story and the
rma is not a reliable source for pre-Christian mythology, as it contains
innovations or mistakes such as the assertion that Gefjun eventually marries
Baldr (strophe 36)4 and conflating Njrr and gir (strophe 53).
As Davidson (1999, 53) points out, Gefjuns ploughing bears striking
resemblence to the land-claim of orgerr in the Hauksbk, AM 371 4to
version of Landnmabk, chapter 276:
sbjrn ht mar, son Heyangrs-Bjarnar. Hann andaisk
slands hafi er hann vildi t fara, en orgerr, kona hans, fr
t ok synir eirra. En a er mlt, at kona skyldi eigi vara nema
land en leia mtti kvgu tvvetra vorlangan dag slsetra
millum, hlfstalit5 naut ok haft vel. (normalized by the author
from the text of ed. Finnur Jnsson, 1892, 98-99)

4
5

Perhaps the authors interpretation of the sveinn inn hvti line from Lokasenna 20; see below.
The meaning of hlfstalit is not clear; it appears to be a hapax, but may be an error for

hlfstalpat which would mean half-adolescent (Jn Sigursson, ed. 1943, 264 f. 7).

12

There was a man named sbjrn, the son of Heyangr- jrn. He

died in Icelands sea while traveling [to Iceland], but his wife
orgerr went with their sons. But it is said that a woman should
not claim more land than that around which she could lead a
well-burdened two-winter-old heifer in an autumn-long day.
(authors translation)
Exactly what to make of the resemblance is difficult to say, but it seems
that the myth of Gefjun may make use of symbolic actions relating to a broader
concept of land-ownership, especially with regard to women (see Clunies Ross
1998, 123 on gender distinction in land-claims; Mundal 1990, 309 on Gefjun
doing things typical of male deities).

2.1.2. Gylfaginning list of synjur

Later in Gylfaginning, Gefjun is described in a list of goddesses. All that is
said is: Fjra er Gefjun. Hon er mr, ok henni jna r, er meyjar andask. The
fourth is Gefjun. She is a maiden, and those who die as maidens serve her (ed.
Faulkes 2005, 29; authors translation). The word mr, here translated maiden,
is often translated as virgin and seen as conflicting with her behavior (Lindow
2001, 135-6) but this is misleading; the word does not necessarily indicate that
the mr has never had sex. The goddess Freyja, whose coital tendencies are
well-known, is called s mr in Vlusp (ed. Neckel and Kuhn 1983 [strophe
25]), and it often identifies women by reference to their father with no
additional implications.6 The specificity with which Snorri singles out Gefjun as

Frigg, who is married and has at least one son with inn, is Fjrgyns mr (Lokasenna 26);
Hel is Loka mr (Ynglingatal strophe 13); and many more examples can be found.

13

a mr, and as relevant to other meyjar, and without reference to a father or

husband, probably indicates that she is unmarried, but it is unreasonable to take
it as a comment on her sexual status. Furthermore, this may explain why she is
able to have children with a giant; a mythological restriction on relations
between giants and synjur may not apply if the goddess is exploiting the giant,
similarly to how inns frequent sexual escapades are frequently prompted by
ulterior motives.7 Karin Olsen (2001, 125-126) notes that she seems to share
many characteristics of giants.

2.1.3. Lokasenna
Gefjun also plays a part in the eddic poem Lokasenna. There are three
relevant verses from her verbal exchange with Loki:
19.

[Gefion] kva:

Gefion said:

Hv it sir tveir

Why, you two sir,

skolo inni hr

must you here indoors

sryrom sakaz?

rail at each other with rending words?

Lopzki at veit,

Of Loptr is it not a characteristic well

at hann leikinn er
ok hann firgvall fri .

known
that he is whimsical
and all the deities dote on him?

20.

[Loki] kva:

Loki said:

egi , Gefion,

Hold your tongue, Gefion,

ess mun ek n geta,

er ik glapi at gei:

now I will tell

of the one who seduced your senses

sveinn inn hvti,

that blond boy

See for example: the theft of the mead of poetry by seducing Gunnl in Skldskaparml
(ed. Faulkes 1998, 4).

14

21.

er r sigli gaf

who gave you a trinket

ok lagir lr yfir.

and you put your thigh over.

[]inn:
rr ertu, Loki,

inn said:
You are a lunatic, Loki,

ok rviti,

and have lost your wits,

er fr r Gefion at gremi,

to get Gefion in rage against you,

vat aldar rlg

for all the fate of the world

hygg ek at hn ll um viti

I think she is aware of

iafngrla sem ek.

as accurately as I.

(Jn Helgason, ed. 1952, II:50)

(Dronke 1997, 337)

Once again Gefjun is said to exchange sex for a precious object, this time a
necklace. Her counterpart in this act, called sveinn inn hvti the white boy
(translated by Dronke as that blond boy) has been tentatively identified as
Heimdallr, who is called hvtastr sa the whitest of the sir in rymskvia and
inn hvti ss the white ss in Skaldskaparml (Dronke 1997, 360). As Margaret
Clunies Ross (1978, 153) pointed out, Lokis choice of words, lagir lr yfir
you laid your thigh over, seems to imply that Gefjun took the sexually
dominant role, although it is interesting that this time Gefjun is the one being
tricked.
inn warns Loki that Gefjun is a dangerous enemy because she knows
the future, which does not seem especially threatening, but in Norse
mythology, knowing the future seems to imply an ability to influence it
(Bek-Pedersen 2011, 193). Alternatively, it may be that knowledge of the
future itself is not the threat, but implies other magical abilities of a more
explicitly threatening nature. The Norse magic known as seir is described as
15

having prophetic as well as more aggressive functions (Price 2002, 94). By

either of these means the same conclusion can be reached, that Gefjun has
magical abilities which she can use against those who fall out of her favor, and
likely also for the benefit of those toward whom she feels positively. Though
here Gefjun is an actual goddess and not a human, this is easily comparable to
the Gylfaginning and Ynglinga saga accounts, regarding both her sexuality and
use of magic.

2.1.4. Vlsa ttr

Gefjun is also mentioned in Vlsa ttr, which is of questionable
relevance to genuine pre-Christian religion. In the northern parts of Norway,
Saint lafr came in disguise to a heathen household who had preserved a
severed horse penis which they passed around each evening and worshiped by
reciting a verse over it. Along with a few other members of the household, the
farmers daughter, though heathen, was reluctant to participate in this ritual,
and when it was her turn to recite a verse she said (normalization based on
Flateyjarbk, GKS 1005 fol dated to the end of the fourteenth century, eds.
Gubrandr Vigfsson and Unger 1862, 2:334, authors translation):
ess sver ek vi Gefjun

I swear this by Gefjun

ok vi guin nnur

and by the other gods

at ek nauig tek
vi nosa rauum.

that under duress do I take

the red phallus.

iggi mrnir etta blti

May mrnir accept sacred object,

en rll hjna

but servant of the household,

rf vi Vlsa.

take hold of Vlsi.

16

That the young woman swears by Gefjun specifically might be seen as support
for her association with young, unmarried women as described in Gylfaginning,
though there are other ways to explain the verse. One might object that Gefjun
was selected for alliteration with guin n.pl. the gods in the next line, but this
is not compelling because there are many ways the poet may have phrased it to
feature a better-known deity considering that there are many synonyms for
god. A more serious consideration is that since Vlsa ttr is attested no earlier
than Flateyjarbk at the end of the fourteenth century, the poet may have been
influenced by Snorris Edda and fashioned the verse after a then-contemporary
image of heathenry which was several hundred years removed from those it
sought to understand, rather than a faithful transmission of beliefs from earlier
times. While this must be kept in mind, Neil Price (2002, 168) has
demonstrated a remarkable correspondence between an event in Vlsa ttr and
a practice reported in the account of the Rus by Ahmad ibn Fadlan, wherein a
woman is lifted over a door frame, apparently to access a supernatural world not
occupied by normal living people. Price considers that this incredibly specific
detail, not attested anywhere else, cannot be shared by the two stories by
coincidence, and most likely reflects actual heathen practice, so that Vlsa ttr
may be more useful as a source of pre-Christian religion than previously
considered. This is in no way to say that its accuracy is absolute, but merely
that it should not be dismissed.

17

2.1.5. Additional mentions

There are two remaining sources of possible but uncertain relevance. In
Droplaugarsona saga (ed. Jakobsen 19021908, 1745), Grmr goes to
hlmganga against Gauss, a viking whom iron will not bite. Though Grmr
defeats him in the battle, Gauss manages to injure Grmrs leg, which became
infected. A woman pretending to be a doctor came to dress his wound and then
disappeared, and afterward Grmrs condition escalated rapidly so that he died.
The woman was in reality Gauss mistress Gefjun, in fjlkunnga the skilled in
magic (ed. Jakobsen 19021908, 175).
In Historia Norvegiae, Ails is said to fall off his horse and die in front of
the temple of Diana (trans. Kunin 2001, 13), which in Ynglinga saga is said to
have happened at a dsarsalr ds hall (ed. Bjarni Aalbjarnarson 1941, 58). As
will be discussed further in the following section, Gefjun is frequently
considered equivalent to Diana in translations into Old Icelandic and this may
be a reason, if tenuous, to suggest an association between Gefjun and the dsir.8

2.2. Glosses of Classical Goddesses

This concludes the Old Icelandic sources on Gefjun, but additional
evidence may be derived from the Old Icelandic tradition of glossing classical
deities with names of Nordic deities in translations of foreign material, or

Exactly what the value of this connection would be for an understanding of Gefjun is
difficult to say; given the range of activities and attributes of dsir it would be more
surprising if Gefjun were entirely differentiated from them. For a thorough discussion of
dsir see Luke John Murphy 2013, especially chapter 2, 4397.

18

interpretatio norroena. In medieval translations into Old Icelandic, Gefjun is used

very frequently to gloss Greco-Roman goddesses, such as Vesta, Minerva,
Diana, and Venus.9 The equivalence with Venus is found only once, in Stjrn;
Venus is usually considered equivalent to Freyja.
It is worth discussing Breta sgur in greater detail, as Diana/Gefjun plays
an active role in the story and it may have influenced future developments of
the tradition regarding Gefjun. The saga is a translation of Geoffrey of
Monmouths Historia Regum Britanniae, the history of the kings of Britain. In
the first book the protagonist, the exiled Trojan Brutus, makes a sacrifice to
Diana/Gefjun, who then appears to him in a dream and directs him to establish
a new kingdom in Britain (Hauksbk, ed. Finnur Jnsson 1896, 241). The
relevant passage reads as follows (normalization based on Hauksbk, ed. Finnur
Jnsson 1896, 241, authors translation):
San gekk Brtus fyri stalla Gefjunar ok hafi ker hendi ok
vn ok bl hvtrar hjartkollu ok mlti: , er vei[zt] himins
tindi ok setning allrar veraldar ok kannt helvtis deili, seg mr
mn forlg ok hvar ek skal byggja at yru ri ok hvar ek skal
ik guleg meyja lta drka at eilfu. Sv mlti hann nu
sinnum ok sofnai. Hann ttisk sj Gefjun hj sr ok mli:
vestrhalfu heimsins vi Gallarki liggr ein ey t hafit

Gefjun glosses Vesta several times in Fdesar saga, Spesar ok Kartasar; Minerva in
Trjumanna saga though it switches mid-text to Frigg; Venus in the Stjrn version of 1
Msebk. By far the most common gloss is for Diana, and occurs in Agnesar saga, Breta
sgur, Jns saga postula, Nikolauss saga erkibyskups, and Pls saga postula. The name Gefjun
also appears in Klements saga and Katerine saga but it is unclear if it is meant to gloss a
particular goddess.

19

bygg. ar byggu fyr meir risar. ar hfir r at byggja ok

nu lii at eilfu ok itt kyn m[u]n hafa vald yfir llum heimi.
[Then Brutus went before the altar of Gefjun and had in his hand
a vessel filled with wine and blood of a white deer and said:
You, who knows the tidings of heaven and arrangement of all
the world and have knowledge of hell, tell me my future and
where I shall build at your counsel, and where I shall have you,
godly maiden, worshiped forever. He said this nine times and
fell asleep. He thought that he (dreamed that he) saw Gefjun
next to him and speak: In the western half of the world by Gaul
lies an unsettled island in the sea. Giants once settled there. It
befits you and your company to settle there forever, and your
kindred will have authority over the whole world.
Although Breta sgur in the form in which it is preserved is from after
the time of Snorri Sturluson, Margaret Clunies Ross (1978, 155) has suggested
that Snorri was aware of the story, even if in a different form.

2.3. A Brief History of Research

2.3.1. Plow Processions
Gefjun has been the subject of considerable scholarly discussion. An
important early work is Axel Olrik (1910, 131), who considered Gefjuns
plowing of Sjlland to be related to a folk custom widespread across Western
Europe wherein a plow is led in a procession around town, usually by a man
dressed in womans clothing, at a time ranging from New Year to Shrovetide,
depending on the area (the potential value of plowing processions to the study
of pagan religion was discussed already in Grimm [1835] 1880, 25865, but is
connected to the Isis mentioned by Tacius and not to Gefjun). There are
20

many regional variations on the plowing custom, and plays are often
incorporated. One common element is that after its procession the plow is led
into a body of water, which Olrik considered related to the procession of
Nerthus in a wagon described by Tacitus which ended with human sacrifice in a
lake. Olrik believed that Gefjun was the local reflex of a ubiquitous Germanic
earth goddess (and therefore more or less identical to other earth goddesses,
including for example Nerthus, Sif, and Iunn), who was wed to the god of the
sky, rr in Olriks opinion10, with the result of fertile land. For support he
collected numerous place-names which he believed to be derived from Gefjun
and some from Iunn, and showed them to be close to places named for rr
(Olrik 1910, 218). While his interpretations are incredibly dated, the
thoroughness of his collection of folk practices is commendable, and a few
other points stand out as insightful. For example, he considered the possibility
that the Gylfi, who is listed under sea-kings in the Nafnaulur and whose
name is used in ocean-related kennings, was once a sea-giant rather than the
human king that Snorri presents (Olrik 1910, 114; note also that in both of
Snorris explanations of ragis verse, Gefjun is also human).

2.3.2. Gefjun as a sea-goddess

A frequently recurring but flawed theory about Gefjun is that she can be
identified as a sea-goddess by the Old English term geofon (geofen, gifen, gyfen)

10

He went as far as to suggest that Iunns marriage to

ragi in Norse mythology is a later

development from her marriage to sabragr, which is a name for rr (Olrik 1910, 24)

21

m. or n. ocean and Old Saxon gean m. or n. id.. Jacob Grimm ([1835] 1880,
23940; 311) believed there had been a god referred to in such sources as
Heliand and Beowulf, his assertion that this was a being rather than just a poetic
term for the ocean supported by nothing other than identification with Gefjun;
he subsequently listed Gefjun with the sea goddess Rn. This was taken up by
Battaglia (1991, 41546) who argued that Gefjun was a figure in Beowulf who
mourned the death of Scyld (4178) and had hostile relations with Beowulf,
based on lines such as Geofon um wol (ed. Heaney 2000, line 515) which he
translates Gefion welled up in waves (Battaglia 1991, 428). He interprets this
to mean that the Goddess' realm apparently rose up against him. Davidsons
(1996, 519) treatment contained some valuable insights awkwardly worked
around the idea that Gefjun is a sea-goddess. This was again elaborated to its
most exaggerated degree by North (1997, 22126) who, after criticizing
previous scholars failure to explain the morphology of Gefjun largely fails to do
so himself. He derives the word from an unattested *gefja which he defines to
give without explanation for why it would mean that (223), and makes no
comment on his proposed etymology of geofon; he also notes the discrepency
between grammatical genders of Gefjun and geofon without explaining why he
thinks it unimportant. He goes as far as placing the Norse goddess Gefjun into
the Old English version of the story of Noahs ark, and argues that the story of
Gefjuns land claim from Gylfi is metaphorical for the ocean eroding the soil
from one place and depositing it in another (2246).

22

De Vries (1962, 160 [Gefjon]) rejected the connection, although he did

not specify why. One major obstacle to the identification other than the
differing grammatical gender is the sound change known as West Germanic
gemination. After a light syllable ending in a consonant other than r, a
following *j caused gemination of the preceding consonant (Hogg 2011, 712
[4.11-4.14]). Compare Goth hafjan v. to lift and OI hefja v. to heave, lift,
raise with OE hebban v. id., OS hebbian v. to lift up, exhalt; to have, to
consider. Since the j in Gefjun follows a light syllable, it would be expected to
cause gemination to the PGmc * so that an Old English or Old Saxon form
would not have an intervocalic f or respectively, but -bb-. It is more likely
that if an Old Icelandic word can be identified with geofon, it might be geimi m.
sea (poetic) as de Vries (1962, 161 [geimi]) cautiously mentions (cf. OI himinn
m. heaven, sky, OE heofon m. heaven), and although this connection is
somewhat speculative an identification with Gefjun should be ruled out as
highly unlikely. Authors who have argued both for this proposed cognate and a
deverbal derivation from a verb *gefja (North 1998, 223) have also not
accounted for the fact that in West Germanic deverbals from -verbs,
equivalent to verbs ending -ns and -an/un in Gothic and Old Icelandic, use a
suffix -ung (Krahe/Meid 1969, III:117 [98]; 209-211 [152]).
That Gefjun may have had an association with the sea or water in general
is definitely concievable, and considering the frequent association of water and
female divinity (Gunnell 2007) it would almost be surprising if there were not
traces of such an association. Davidsons (1996, 5159) article provides
23

significant evidence that Gefjun had some association with water, but not
necessarily more so than other mythological figures like the Norns. However, if
such a connection is to be made, it cannot be on etymological grounds.

2.4. Conclusion
Etymology has been a dominant force in the study of Gefjun, so it is
unfortunate that the theories regarding her name which seem to have the most
widespread influence are wholly inadequate. She is not the giving one in
fact the mythology mostly depicts her taking; nor is she the ocean. Perhaps time
and further study will reveal that she is a goddess of vegetation after all, but
probably not for the reasons currently maintained.
Some examples of further reading on Gefjun which are influenced by
Gefjun as a literary and mythological figure without much influence from the
theonym itself, see The myth of Gefjon and Gylfi and its function in Snorra
Edda

and

Heimskringla

(Clunies

Ross

1978,

14965),

Metamorphosen einer Gttin (Heizmann 2002, 197255), ragi

Gefjon:
oddasons

Ragnarsdrpa: A Monstrous Poem (Olsen 2001, 123139) and very briefly in

The Position of the Individual Gods and Goddesses in Various Types of
Sources with Special Reference to the Female Divinities (Mundal 1990,
294315, specifically 30810).

24

3. Feminine n-final substantive formation

In the coming chapters, it will be of great importance to be aware of two
methods of word-formation which will prove vital for an understanding of the
development of the word Gefjun and other theonyms. The chapter will also
conclude with a brief discussion of some other suffixes that bear more
superficial resemblence to the ending of Gefjun, but which can be easily
rejected as sources for the name.

3.1. Deverbal substantives in *-ni3.1.1. *-ni- abstracts formed from roots of strong verbs
A type of deverbal substantive and adjective inherited into Germanic was
formed by attaching *-ni- to the verbal root. Germanic substantives derived by
this method are exclusively feminine, and formation from strong verbs was no
longer productive in PGmc, but there are a number of them which have been
inherited from PIE (Krahe/Meid 1969, III:115-116 [98]). An example of such
a formation is *ni- (OI bn f. a petition : bn f. prayer, request, OE bn f.
prayer; cf. Old Armenian ban word, speech; manner, thing) < *bh-ni(Krahe/Meid 1969, III:115-118 [98] without incorporation of laryngeals;
Kroonen 2013, 72 [*bni-]).
In Old Icelandic, inherited feminines in (root) + *-ni- were subject to iumlaut, frequently resulting in doublets such as bn petition : bn prayer,
kvn : kvn f. woman, wife, and sjn : sn vision. This is due to paradigmatic
root vowel alternation which subsequently split the paradigm into separate
25

words, each with one of the two vowels generalized throughout their own
paradigm, for example nominative singular *niz became Old Icelandic bn,
but which would have an unumlauted vowel in the genitive by regular
development, OI bnar11. In the following table, the paradigm as it would
appear by phonological development alone (a), which cannot be demonstrated
to have actually existed in Old Icelandic, is contrasted with the paradigms of
bn (b) and bn (c) gathered from Old Icelandic manuscripts.12
Table 3.1: Hypothetical regular development of the paradigm of OI bn (a) and split
paradigms of bn (b) and bn (c) as they appear in Old Icelandic.

a.
sg.

pl.

11

b.

c.

nom.

bn

bn

bn

acc.

bn

bn

bn

dat.

bn

bn

bn

gen.

bnar

bnar

bnar

nom.

bnir

*bnir

bnir

acc.

bnir

*bnir

bnir

dat.

bnum

bnum

bnum

gen.

bna

bna

bna

The development of the i-stem genitive singular ending is debated; either an ending *-z
was replaced analogically by the -stem ending *-z > *-, or a PGmc ending *-aiz
developed into PN *- or *- (Krahe and Meid 1969, II:26-31 [13, 15], Syrett 2012,
99-104). Boutkan (1995, 244-246 [3.2.7.6.]) disagrees that this came by analogical
change.

12

Plural nominative and accusative bnir are not attested in ONP, but are provided from the
word's Modern Icelandic paradigm and Old Icelandic attested plural nominative and
accusative sjnir.

26

The earliest attestation of bn or bn in the ONP is a dative plural bnum

written benum from AM 237 a fol. dated to 1150, suggesting that the
paradigmatic split was already complete, as the dative plural would probably not
be expected to show i-umlaut13. The lexical drift that separated the two forms
slightly in meaning was most likely complete already at this time, as they are
kept discernibly separate in early attestations.

3.1.2. *-ni- abstracts formed from roots and stem suffixes

of weak verbs
While new feminine substantives in *-ni- did not form from strong verbs
in Germanic, they could be produced from weak verbs, in which case the *-nisuffix attached to a vowel determined by the verbs suffix vowel. That is, jaand ija-verbs produced deverbals in *-ni- (e.g. *hauzni- f. hearing, Goth.
hauseins, OI heyrn from PGmc. *hauzijan- v. to hear, Goth. hauzjan, OI
heyra), -verbs in *-ni- (in North and East Germanic only, e.g. PGmc.
*lani-, Goth. lans f. invitation, OI laan/lun, invitation, enticement
from *lan- v. to invite, Goth. lan, OI laa), and e-verbs in *-ni- (e.g.
*luni- (Goth. lubains hope, OE lufen hope) from *lun- (OHG lobn to
praise).

13

Feminine i-stems, like their masculine counterparts, likely had a PGmc dative plural
ending *-imaz, but a reflex of the masculine i-stem gastiz (attested in the nom. sg. on the
Gellehus horn 2, DR 12) appears in the dative plural on the elder fuark Stentoften
runestone DR 357 as gestumz. The ending is analogical from the a-stems, although the
cause of the umlauted vowel is debated. On one hand it could itself be analogical leveling
from the other cases, but on the other hand the original ending could have caused umlaut
before being replaced. See Schulte 1998, 76-82 with references.

27

The suffix vowels and i-stem ending is preserved in Gothic, but in Old
Icelandic the suffix vowels of both *-ni- and *-ni- abstracts have been lost and
cannot be distinguished from each other except by conferring with other
languages or the paradigms of the verbs from which they derive (see Johnsen
2012, 3351 on the loss of unstressed *). Like many other feminine i-stems,
the ending was replaced in the singular with -stem endings, and in Old
Icelandic they universally show u-umlaut when applicable in the singular
nominative, accusative, and dative, and plural dative, with no trace of i-umlaut
even in words formed with *-ni- where the verb from which it was derived
does show umlaut. For example, cf. lausn f. liberation from *lausni-, derived
from leysa v. to loosen; to free from PGmc. *lausijan-; lgn f. net from
*lagni-, derived from leggja v. to lay, PGmc. *lagjan-. The umlauted
diphthong in OI heyrn f. hearing from *hauzni- v. to hear is fronted as a
result of the PN palatal sibilant < *z rather than i-umlaut (see Noreen 1923,
71.8). However, i-umlaut is shown in East Norse forms (Johnsen 2012, 357),
such as Old Swedish and Old Danish vrn defense from *warni-, derived
from *warjan- to defend (OSw. vria, OD vri, OI verja), cf. OI vrn f.
defense, though they are sparsely attested in East Norse. The one example
from Old Gutnish, lausn liberation (< *lausijan-, OGut loysa v. to free) also
fails to show i-umlaut. Because examples of unambiguous *-ni- and *-niderivations with an underlying root vowel e are lacking, nothing definitive can
be said about breaking.

28

3.1.3. *-ni- abstracts formed from -verbs

Abstracts in *-n- do not appear in West Germanic, which formed
feminine deverbals with the suffix *-ung- (Krahe/Meid 1969, III:117 [98];
20911 [152]), but are very numerous in North and East Germanic. They are
easily identified in Old Icelandic as they appear as the verbal root with -an
added to the end, sometimes with umlaut of the ending -an to -un, so that the
derivative of skipa from PGmc *skipn- v. to arrange, place in order comes
skipan/skipun ultimately from PGmc *skipni- f. order, arrangement.
Like the other weak verb-derived abstracts discussed above, the original
feminine i-stem endings were replaced with -stem endings at some point, and
this appears to be the origin of the alternation in the suffix vowel between a and
u (see Noreen 1923, 137.2 Anm. 3). The development from *-niz to -un is
given by sgeir Blndal Magnsson (1989, 167 [-an]): *-n-i > *-n- >
*-anu > *-n > -on > -un. The umlauted a surfaces as u rather than because it
is in an unstressed position, wherein only three vowel sounds were distinctive
(see below), a process which Hreinn Benediktsson ([1962] 2002, 78, f.7) calls
structurally conditioned analogical sound change.
The suffix vowel is sometimes written o, which is understood to be an
orthographic variation on u, representing the same sound. The Old Icelandic
system of unstressed vowels had only three distinctive phonemes: a low vowel
a, a non-low front vowel written e or i, and a non-low back rounded vowel
written o or u, the latter two respectively marked [] and [] by Hreinn
Benediktsson ([1962] 2002, 7881). The orthographic variation is most likely
29

caused not by actual phonetic differences, but rather by changes in the stressed
vowel system resulting in different patterns of identifying these unstressed
vowels with stressed equivalents (7681). Therefore an abstract derived from
skipa may appear written skipan, skipon, or skipun, but it is only safe to
assume a phonological distinction between skipan skipan on one hand, and
skipon/skipun skipun on the other.
By regular development the genitive would not be expected to show uumlaut, where the -stem ending *-z developed into Old Icelandic -ar. The
same applies to the plural nominative, accusative, and genitive; because they are
abstracts, plurals of nouns of this type are uncommon but when they do appear
they have normal i-stem plural endings without umlaut (except in the plural
dative where it is expected). However, there was a strong tendency to generalize
one suffix vowel or the other throughout the singular, with a significantly more
common in manuscript sources, but with u ultimately becoming the standard in
Modern Icelandic. The paradigms in Table 3.2 represent (a) the expected
development of the declension following only phonological change, contrasted
with (b) a recurring medieval and (c) early modern paradigm which has
generalized a low suffix vowel, and the paradigm in Modern Icelandic. It is
important to remember that because unlike bn f. petition and bn f. prayer,
which differ slightly in meaning, there is no discernible lexical difference
between skipan and skipun, so that paradigm (b) cannot be said to have existed
to the exclusion of paradigm (c); they seem to have always coexisted, even
during times when paradigm (b) dominated. A plural dative *skipanum is not
30

attested, but cf. eggjanum, written aeggianum in Flateyjarbk (ed.

Gubrandur Vigfsson & Unger 1860, I:511), though this could reflect
influence from Norwegian. The plural genitive is not attested in manuscript
sources.
Table 3.2: Paradigms of OI skipun with regular development (a)
and generalized suffix vowels a (b) and u (c).

sg.

pl.

a.

b.

c.

nom.

skipun

skipan

skipun

acc.

skipun

skipan

skipun

dat.

skipun

skipan

skipun

gen.

skipanar

skipanar

skipunar

nom.

skipanir

skipanir

skipunir

acc.

skipanir

skipanir

skipunir

dat.

skipunum

skipunum

gen.

skipana

skipuna

Careful examination of the actual distribution of suffix vowels in early

manuscripts is necessary in order to consider the likelihood of identifying it
with the word Gefjun. While a survey of the suffix vowel representation of all
such words would be a worthy subject for future research, the volume of
material necessitates that only a small selection be observed for now. Four
words were selected for closer observation using attestations found in the ONP,
which are skipun order, arrangement, skemmtun entertainment, eggjun an
egging on, and vejun a wagering, betting. Of the *-ni- abstracts, skipun is
the most frequently-occurring, and skemmtun also has a high number of
attestations. The words eggjun and vejun were chosen for phonological

31

similarity to the word Gefjun, in particular a root vowel e and a semivowel j

immediately preceding the suffix. Attestations from the earliest available until
those dated to 1500 in the ONP were selected.
The results are seen in table 3.3. Each word is broken down by case, and
for suffix vowel representation among attestations in that case. Total number of
attestations is shown to the left in each cell, and to the right the percentage
frequency of that vowels representation within that case for that word.
Table 3.3: Totals and frequency of orthographic representation of suffix vowels in *-nideverbal feminines substantives. A vowel v is considered the same as u.
skipun

nom.

acc.

dat.

gen.

Total

skemmtun

eggjun

vejun

Total

38

79.17

14

82.35

75.00

50.00

58

77.33

14.58

16.67

10.67

6.25

17.65

25.00

33.33

12.00

50

96.15

25

83.33

10

71.43

81.82

94

87.85

1.92

0.93

1.92

16.67

28.57

18.18

12

11.21

49

85.96

17

89.47

32

88.89

50.00

99

86.84

5.26

5.26

2.78

50.00

5.26

8.77

5.26

8.33

7.89

75.00

19

67.86

100.00

28

71.79

12.50

17.86

15.38

12.50

14.29

12.82

143

86.67

75

79.79

47

82.46

13

68.42

278 82.99

12

7.27

6.38

1.75

10.53

21

6.27

10

6.06

13

13.83

15.79

21.05

36

10.75

32

Among these words, it is clear that the representation of the suffix vowel
is not predicted by paradigm (a) in Table 2 above. The plurals, which are almost
always rendered as expected with the exception of one instance of eggjanum
noted above, are not shown. It can also be seen that the relative frequency of
suffix vowels is fairly consistent throughout the paradigm, not strongly
corresponding to case. In fact it is overall less likely that a genitive will have a
suffix vowel a as expected. Though vejun has a lower precentage of a than the
other words, this is probably due to the relatively few attestations. Having
established that the likelihood of any one vowel being written does not seem
heavily influenced by the word nor the case, the rate of each is shown by time
period, with intervals of half-centuries, in Figure 3.4. In the case of a dating
range, the average was used for placement in the graph.

Figure 3.4: Number of representations of suffix vowels in *-ni- deverbals skipun,

skemmtun, eggjun, and vejun by half-century, by total number of attestations
100
90
80
70
60
50
40
30
20
10
0

a
o
u

33

Figure 3.5: Frequency of representations of suffix vowels in *-nideverbals skipun, skemmtun, eggjun, and vejun by half-century,
by percentage of all attestations within the given half-century.
100.00
90.00
80.00
70.00
60.00
50.00
40.00
30.00
20.00
10.00
0.00

a
o
u

Runic evidence on these words is unfortunately sparse. The Greenlandic

runic

inscription

GR

(Olsen

1949,

5171)

contains

the

word

(t)=yr=ku=n(a)(r) drkunar f. worship but is a medieval inscription and it

cannot be assumed that it testifies to an earlier form of the language than early
Icelandic manuscripts. Uppland runic inscription U 349 contains the word
skibin skipun. The carver marks all of his other vowels as expected. Other
words

with

unstressed

syllables

from

the

same

inscription

include

(normalizations after Upplands runinskrifter, eds. Wessn and Jannson 1943

1946:II, 95):
Unstressed a: risa risa, ina enna, ala alla
Unstressed i: uikitil Viktill, ifti fti, uti uti, ialbi hialpi
Unstressed u: usur Ossurr

34

It may well be that his language did indeed contain a word skipin with front
vowel in the suffix. The relevance of this is, however, limited, as the language
of the carver was certainly a variety of East Norse, and so does not have an
immediate bearing on Old Icelandic. See 3.2 for a possible explanation.
Already by the time of the Icelandic Homily Book, Holm perg 15 fol,
dated to around 1200, it is clear that the relationship between phonological
development and distribution of suffix vowels had broken down, demonstrated
by forms like frtonar freistanar (46r26; de Leeuw van Weenen 2004, 47
[freistun]) and ktonar skemmtanar (75v7; de Leeuw van Weenen 2004, 146
[skemmtun]).
When *-ni- abstracts appear with the rounded suffix vowel written o
or u, the root vowel is susceptible to u-umlaut, but it is only applied
consistently when the root vowel in the verb from which it derives is a; for
example, both dagan and dgun f. dawn can be found, but not *dagun or
*dgan. The situation is more complicated when it comes to breaking, which is
shown in words like fordjrfun (perhaps under the influence of djrfung?) and
jfnun/jafnan,

but

not

in

words

like

efan/*efun14,

blezan/blezun,

opinberan/opinberun, and others. It is probably significant that blezan/-un and

opinberan/-un are both late additions to the Old Icelandic vocabulary,
respectively formed to loanwords derived from OE bletsian v. to bless and
MLG openbre adv. offenbar, ffentlich (sgeir Blndal Magnsson 1989, 64
[blessa, bleza], 691 [opinber]). Root vowels other than a or e are unaffected by
14

Attested in the genitive efvnar in Holm perg 3 fol 140v (ed. Loth 19691970, 330).

35

the suffix; both vsan and vsun f. pointing/direction can be found, but no
*sun with u-umlaut. Note that there are many forms which show umlaut and
breaking that are not caused by the suffix, such as fjlgan f. a multiplying from
the verb fjlga , most likely from earlier *felugn- to multiply, cf. the prefix fjl
much, manifold, corresponding to Gothic filu adv. very.
It is not surprising that of all possible root vowel alternations only a/ is
represented consistently, as it is extremely common in the language and is even
still productive in Modern Icelandic (now between a and ), albeit in a
morphologically- rather than phonologically-conditioned manner (Kristjn
rnason 2011, 246 [12.3.4]).

3.1.4. Conclusion
Both names Gefjun and Gefn resemble deverbal substantives described in
the immediately preceding sections. It can be concluded from the evidence
presented here that if that is how the names were formed, it was most likely to
different verbs. The name Gefjun resembles the construction from a weak verb, which would have been *gajn- as also noted by Much (1891, 317) and
Sturtevant (1952, 166). On the other hand, Gefn actually could be formed
directly from *gean-, so that at an earlier stage it was *geniz, and perhaps at
one point having stem alternation *Gifn ~ Gefn due to umlaut in cases other
than the genitive. On the possibility of an OI *gefja, see 6.1.2.

36

3.2. The Hoffmann Formation

In 1955 Karl Hoffmann identified a suffix in Indo-European languages
that he describes as identifying possession of the object or quality of the word
to which it was attached. Olsen (2004 22944) cites numerous examples, such
as *hap-hn- (Welsh afon river) from *hap- water, where the literal
meaning of the composed form is having a lot of water. In recent times it has
frequently been suggested that the suffix is actually the second component of
a compound, perhaps identifiable with Latin onus m. load, rather than a suffix
(Pinault 2000, 9596; Olsen 2004, 232), and here the term Hoffmann
formation will be used in a PIE context, although it is clear that it functioned
as a suffix in Germanic.
Olsen (2004, 237-44) details a variety of uses of Hoffmann formations.
The basic formation from *-honh with accent on the root creates
substantivized adjectives, indicating possession of or authority over the root to
which it is attached, sometimes with a specifically pejorative meaning,
something like burdened with (the first component of the formation). When
the thematic vowel is added to make it *-hnh-o-, it indicates an individual
with some relationship to the possession of the first component of the
formation, perhaps but not necessarily the one who is in charge of it.
According to Olsen, the oldest meaning of this formation was something or
someone having or being connected with one who has the *honh-, i.e. load,
charge or authority, expressed by the first member (Olsen 2004, 242).

37

Without reference to Hoffmann, Meid (1957, 72-126) wrote a highly

detailed account of deity names (and a few words which are not deity names)
composed with a suffix that he identified as *-no- in Indo-European languages,
frequently called the Herrschersuffix, which can be identified as the
thematicized version of a Hoffmann formation (Pinault 2000, 6667 [6]).
Among them are fourteen Germanic names (counting both Fjrgynn and
Fjrgyn separately), including some known only in Latin inscriptions or, in the
case of Tamfana, a reference made by Tacitus. According to Meid (1957, 77
80) the meaning of the suffix is somewhat too general to give a very precise
definition that applies equally well to all cases, but it is possible to draw some
conclusions. A deity with a name of such composition is representative of a
certain element, natural phenomenon, group of people, and even psychological
or cosmological principle (802). The diversity of applications of the suffix
may make interpretation difficult. For example, the word inn (from PGmc.
*Wanaz) can be interpreted as a god of r m. spirit, mind; poetry, as a god
who is r adj. mad, frantic, furious, or possibly as a characteristic member of
or leader of the Wild Hunt. Indeed, Meid (1957, 867) sees all of these
characteristics coexisting in the god, beginning with a basic meaning of ecstasy,
madness, inspiration with the others resulting from this15.

15

Alternative explanations of the name inn are manifold; the inclusion of Meids is not
meant to imply their finality but rather as a demonstration and example of his own
theories; for a concise summary of scholarship regarding the name inn along with his
own interpretation, that an association with death is primary and the other attributes
secondary, see Liberman 2012, 825.

38

A deity with an *-no- suffix name usually seems to have some power
over the object or concept to which the suffix is attached, but is better
described as representative of it, with the connotation of authority derived
therefrom (834). Meid suggested that in a much earlier level of development,
the beliefs which would eventually inform Indo-European religion were such
that the phenomenon itself was worshiped, but that over time as the religious
worldview developed the deity became more distinct from the object, at which
point the suffix would be used to distinguish the being from the phenomenon
in name (90). He considers that deity name pairs such as inn and r (the
husband of Freyja in Snorris Edda), and Ullr and Ullinn (a name inferred from
Scandinavian place-names; see Brink 2007, 116) are remnants of this
development (86). Meid gives an example of the Lithuanian thunder god
erk nas, which he derives from PIE *perkuus m. oak (Lat. quercus). He
suggests that at one time, oak trees themselves were worshiped, but over time a
Herr der Eiche developed, and the oak tree became a conduit for communion
between the god and his worshipers rather than the intended recipient of
worship itself within their worldview (90).
A distinguishing characteristic of Hoffmann formations in many IndoEuropean languages is the elongation of the medial vowel, which is derived
from the stem suffix, for example trib nus m. chief from tribus f. tribe. In an
article discussing that specifically, Meid (1956, 260) explained that this
elongation does not appear to be productive in the Celtic or Germanic
languages, with few remaining traces in the languages. Thanks to Hoffmann and
39

later elaborations such as those by Pinault (2000, 61117) and Olsen (2004,
22944), it can be shown that the elongation is due to a laryngeal in the suffix
which was later lost along with compensatory lengthening of the vowel. An
example given by Olsen is Greek , corresponding to a PIE *diuo-h-nhh (Meid 1956, 277 reconstructs *- as an earlier Greek form); here the
lengthening of o to Greek rather than is a result of the loss of the laryngeal
h at the beginning of the suffix. Olsen does not appear to be suggesting that
*diuo-h-nh-h was already a composed word during the PIE period, but
rather choosing a hypothetical example for demonstration. The lengthening
itself must have become productive in many daughter languages on the model
of earlier words (see Pinault 2000, 667 [6]).
In North Germanic, masculine Hoffmann-formations of the thematicized
type often have an ending -inn even when not derived from i-stems, such as in
inn from *wuz and *Ullinn from *wuluz (OI. Ullr, got. wulus m.
splendor). The name inn is usually reconstructed to PGmc *wanaz on the
basis of its other Germanic cognates (OHG Wuotan, OE Wden, Old Saxon
Uuden). If inn had instead developed from *winaz, i-umlaut would be
expected resulting in *inn such as in OI i m. rage; madness; The English
Wednesday has traditionally been seen as evidence for an Old English *Wden
(< *W den; Meid 1957), but Bammesberger (1999, 16) has demonstrated that
this is not likely, and only Wden can be assumed for Old English. Noreen
(1923 [173.5]) refers to a Middle Norwegian on, and writing in Old
English, fric said that in Danish, the god Mercurius is called Oon (Lassen
40

2011, 98); these suggest a variant with a medial vowel *-u-, which presumably
derived from the stem vowel of *wuz if indeed it existed; since Noreen did
not mention the source of Oon it was not possible to confirm for the present
work, and lfrics rendering of a word from a language foreign to him should
be treated with caution.
Although adjectives and substantives in *-ina- are likely to have
occurred in Proto-Germanic (cf. Got. fulgins adj. hidden; kindins m.
governor), they are far less common than equivalents in *-ana- and cannot
serve as a base for analogical change (Boutkan 1995, 789). The situation is
made even more complicated by a number of runic examples that show a suffix
-ina- already in the Elder Futhark period, such as N KJ60 faikinaz (perhaps
related to OI feikn f.adj. awful, mostrous; Syrett 1994, 203) and N KJ71 azinn
(stone according to Antonsen 2002, 191, cf. OI arinn m. hearth, dated 375570). A number of theories have been proposed which attempt to explain how
the -inn ending could come about without causing umlaut. Kock (1898, 484
554) proposed a complicated sequence of sound changes wherein first PGmc *an- becomes *-in- when following an *i in the first syllable (which is also to
say, words wherein i-umlaut could not possibly show regardless of the timing,
as pointed out by Syrett 1994, 193) or immediately following a velar consonant,
and then later another change occurred wherein a second-syllable *-an(C)
becomes *-in(C). This was accepted and elaborated by Boutkan (1995, 7882)
who believed furthermore that the nominative ending *- > n in a word like

41

inn would be lost by regular development, and then reinstated analogically,16

proposing the development *wanaz > *w-anz > *w-an *w-an-r >
inn, where Kocks theory applies only between inn and the previous step,
but not until after i-umlaut ceases to be productive.
Syrett (1994, 187-204 [7.6]) offered a solution that is vastly more
simple.17 The idea is that the ending *-na-, already used to form adjectives
denoting the material from which something is made (e.g. *gulnaz adj. of
gold, OI gullinn id.; Ringe 2006, 294) and also does not typically show iumlaut, was generalized throughout participles and adjectives previously in
*-ana- or *-ina- (Syrett 1994, 2034 [7.6.5]). He says that this could have
then spread to some substantives as well. Perhaps this is the source of the word
skibin skipun on Uppland runic inscription 349 mentioned in 3.1.3.
Meid (1957, 11926) identified a number of Germanic feminine names
which he believed to be formed with this suffix, including Lofn, Hrn, and
Fjrgyn from Old Icelandic and Vercana, Hludana, and Tamfana from Latin
renderings of Germanic goddess names. If Syretts (1994, 187204 [7.6])
theory is correct, the ld Icelandic examples, Lofn and Hrn, might resemble

16

outkans reasoning is rather bewildering; he says that a PN *-an should become *-an
because the PGmc. a-stem accusative plural *-anz and PGmc. masculine n-stem genitive
singular *-anas, from earlier *-enas, both surface as -a in Old Icelandic. Not only does the
evidence seem unrelated to the conclusion, but placing ones faith in the regular
development of North Germanic n-stems seems highly inadvisable. Note that rejection of
outkans additions does not imply that Kocks is wrong.

17

Boutkan (1995) made no mention of Syrett (1994) as he almost certainly did not have the
benefit of consulting The Unaccented Vowels of Proto-Norse before the completion of his
own work, released only the following year.

42

*-ni- deverbals because this remodeling, combined with the analogical

replacement of the i-stem ending with -stem endings in the singular, actually
made them indistinguishable from the deverbals, both consisting of a root and
*-n. The unstressed *-- has been lost in *-ni- and *-ni- deverbals by the
time of Old Icelandic (see 3.1.2) and so it would be expected for other words
which also consist of a root followed by *-n-.
Meids analysis of Lofn is faulty as is detailed in section 4.1.3.3 below,
although this does not entirely preclude the possibility of Lofn as formed with
the Herrschersuffix. It is also difficult to accept Meids admittedly hesitant
interpretation of Hludana as related to holda, a sort of spirit, because the
metathesis that would be required to allow a connection seems unlikely.
Despite these words of caution, it is still likely that feminine deity names could
be formed in Germanic languages by means of the Herrschersuffix. If feminine
Hoffmann-formations of the thematic type can be found in Germanic, although
thematicization in itself would no longer be a meaningful concept, words
created on the model of older Hoffmann formations would be -stems
(meaning that a PGmc feminine Herrschersuffix *-n can be proposed), as
names like Vercana, Hludana, and Tamfana most likely are; they are generally
attested in a dative ending -ae as in Latins first declension, as opposed to istems like Lubaini, showing a Latin third declension dative ending.
Likely examples of these words are more difficult to discern in Old
Icelandic than their masculine counterparts, but some candidates can be found
especially in poetic terminology, such as borkn or brkn f. female wolf, possibly
43

related to berkja v. to bark according to de Vries (1977, 51 [borkn]) and sgeir

Blndal Magnsson (1989, 73 [borkn, brkn]) or barki m. windpipe;
unfortunately the word only appears in the Nafnaulur in the nominative and
its declension is unknown. Another possible example is a troll-woman name
Leikn, which de Vries (1977, 351) compares to OE scnlce f. sorceress and
reconstructs *laikn but *laikan does not seem less possible, nor is *laikniz
impossible should it derive directly from the strong verb leika to play although
that seems less likely on semantic grounds to be a troll-womans name. A very
closely related formation designates collectives and place-names based on an
object or substance found in that location, which Olsen (2004, 243244)
reconstructs for PIE on the basis of words like Latin porrna f. field of leek
from porrum n. leek and Lithuanian bernas m. birch wood from beras m.
birch to *-i-hnho-; several possible compounds of this type may be
discernible in place names, especially for islands, such as Bkn f., and Hven f.,
possibly derived respectfully from bk f. beech tree (sgeir Blndal
Magnsson 1989, 69 [Bkn]) and a word cognate to Gothic *a (attested in
the dative hwan) f. foam (sgeir Blndal Magnsson 1989, 394 [Hven]).
Although a deverbal derivation from a cognate of Gothic ajan v. to foam
cannot be ruled out entirely for Hven, it is less likely because ajan is a weak
verb, and *-ni- deverbals from weak verbs typically show u-umlaut, although it
is not clear from where the e comes. If Bkn and Hven are Bokn, Norway, and
Ven, Sweden, the respective absence and (possible) presence of i-umlaut

44

corresponds to the deverbals discussed in 3.1.2 and accords well with the
theory of Syrett (1994, 187-204 [7.6]).

3.3. Other suffixes

Feminine names frequently end with -unn, such as Iunn, Ingunn,
Steinunn, and many others. The etymology of this ending is disputable. It may
be related to the verb unna to allow, bestow; to love, and unnr f. wave has
also been suggested (de Vries 1962, 365 [unnr]; sgeir Blndal Magnsson
1989, 1088 [1 -unn]). The advantage of such a suffix is that the accusative and
dative endings are -i, as can also be seen in the Modern Icelandic paradigm of
Gefjun, acc./dat. Gefjuni. However as chapter 5 will show, this declension is not
borne out in manuscripts, and should be considered analogical, most likely
from -unn names.
It is certain that the suffix is not part of the name Gefjun. Not only does
Gefjun lack the geminate nn, but representation of the suffix vowel is drastically
different as well. The names Gefjun and Iunn appear frequently in close
proximity to each other and so can be easily compared, for example GKS 2365
4to [15r7] and [15v12], both iv alongside [15v20-21] gef|ion and
[15v18] gefion.
There was also a feminizing suffix *-unj- which in OI becomes an *nstem *unjn-, and creates feminines from other nouns, such as synja f.
goddess from ss m. god (Johnsen 2005, 143145). The suffix is rare, and
the suffix vowel is affected by i-umlaut, so this is also rejected as a candidate for
contributing to the name Gefjun.
45

3.4. Conclusion
At this point, both deverbals from weak -verbs and Hoffmann
formations provide viable means of derivation of the name Gefjun. Further
investigation is necessary. If Gefjun is a deverbal, fluctuation between umlauted
and non-umlauted forms of the suffix vowel is expected, with a vowel a
appearing alongside u and o, which is investigated thoroughly in chapter 5.
There is unfortuantely little in Old Icelandic to compare Gefjun to in order to
determine the possibility of a Hoffmann formation, but an explanation as to
how it would have formed and what word it used as a base will be needed in
order to evaluate that theory. A Hoffmann formation would much more easily
lend to an interpretation wherein the names Gefn and Gefjun have a common
origin than the deverbal theory.

46

4. Goddess Names in Germanic Languages

4.1. Old Icelandic
4.1.1. Goddess names ending -un
4.1.1.1. Introduction
As Sturtevant (1952, 167) noted, only one other Old Icelandic goddess
name other than Gefjun can be definitely identified as having the same ending,
which is Njrun. Another name may be discernible in the two-word phrase
Ingunar Freyr, referring to the god Freyr.
4.1.1.2. Njrun
The name Njrun is found in the Nafnaulur entry for synja heiti
names of goddesses. It occurs frequently in kennings meaning woman,
sometimes compounded with beii-, eld-, or hl- (Lexicon Poeticum 1931, 429
[Njrun]; Skj BI:78[36], 80[44], 103[35], 126[4], 188189[1], 27980[12],
487[34]; BII:216[21], 47677[1]). Njrun also appears in the compound
draumnjrun in the poem Alvssml as a poetic term meaning night (eds.
Neckel and Kuhn 1983 [strophe 30]). Unfortunately, no mythological
information can be found regarding the goddess, which makes it more difficult
to discern the most likely of several different interpretations of the name.
It has been suggested that Njrun could be a descendent of the goddess
Nerthus described by Tacitus around 100 A.D. (Nsstrm 1995, 60; Hopkins
2012, 39-44), but this seems highly unlikely. The argument in favor of this
interpretation is as follows: Nerthus is described by Tacitus as a female earth
47

deity, whom he refers to as Mother Earth (Tacitus, ed. Hutton 1914, 320321) and while there is no clear reflex in Norse mythology recorded in Eddic
poetry or other North Germanic sources, the word itself has been taken to
represent a Germanc *neruz, which, given standard phonological development
of North Germanic, would yield Njrr in Old Icelandic (de Vries 1931, 3637).18 Njrr is, of course, a male deity of the Vanir family, but many mythical
themes and ritual elements which are described as related to the Vanir in North
Germanic sources are reminiscent of Tacitus' description of the cult of Nerthus,
such as processions of the deitys idol in a wagon, ritually enforced peace, and
fertility of the land (see McKinnell 2005, 5062, 74); it could be either that the
name ceased to be suitable for a goddess or Tacitus misunderstood and reported
the name of not the female, but the male deity (de Vries 1931, 37). In stanza 36
of Lokasenna (ed. Jn Helgason 1952, 2:52) the god Loki accuses Njrr of
procreating with his sister, who is unnamed. The Vanir deities Freyr and Freyja
have names which are nearly the same except for their grammatical gender 19
and if the first component of the compound Ingunarfreyr, a name for Freyr, is a
feminine personal name *Ingun (see 4.1.1.3.) this may be a feminine equivalent
to Yngvi, another name of Freyr. The evidence regarding Freyr and Freyja has
been taken to imply that Vanir deities tend to come in sibling pairs with
equivalent names (McKinnell 2005, 52, 55), so it would not be unexpected if

18
19

*Neru- to *[Njru] by u-breaking, to Njr- by and syncopy of the unstressed vowel *u.
It would be wrong to say they are quite the same other than the gender difference,
because Freyr is a strong substantive and Freyja a weak substantive.

48

there were once two deities named *neruz, one who was male, kept his name,
and retained prominence nearly a millennium after Tacitus, the Njrr who
appears in Norse mythology, and another who was female and either lost her
name, became obscure, or never actually had that name to begin with. As
*neruz is a u-stem substantive, which is extremely rare in the feminine gender
and highly susceptible to analogical change, it is believed by some that the
name was unsustainable for a goddess and was changed to a more clearly
grammatically feminine equivalent (Nsstrm 1994, 60).
This is not an illogical argument but the evidence against it is
substantial. There is some concern that the spelling Nerthus is not an
accurate representation of the name of the goddess observed by Tacitus,
although John McKinnell (2005, 501) has demonstrated that it most likely is.
The goddess name Skai is a typically masculine word is an acceptable name for
a goddess, and indeed is even used for a man in Vlsunga saga without any
apparent conflict (ed. Finch 1965, 1). It is possible that the word *Neruz >
Njrr is the result of a root *ner- and suffix *-uz, but other possible
explanations require that the dental be part of the stem (see de Vries 1977,
4101 [Njrr] with seven different suggestions, apparently changing his
position, or at least his confidence in it, from the one he expressed in de Vries
1931, 367), and even if this is not the case it is unclear for how long the name
would have been analyzable as *Ner-uz rather than *Ner-uz. A name
*Njrun would be indisputably related to Njrr, but clearly no such word
exists. When de Vries (1931, 367) compared Njrr : Njrun to r : inn
49

and Ullr : Ullinn, he neglected that the dental of the suffix *-u- which he
describes as important in the creation of the names r, Ullr, and Njrr is itself
still present in both the longer forms, so the comparison is not equal.
One difficulty in determining the etymology of Njrun is in its
overabundance of possibilities. The simplest is that it is a feminine deverbal
abstract from the weak -verb njrva 'to bind, lash, tie up' and therefore the
same derivation as the modern Icelandic njrvun a fixing in place, the only
difference being that in the modern language v and u are no longer in
allophonic distribution. In Norse mythology, gods and valkyrjur are commonly
associated with the placing and removing of bonds and fetters, and the word
bnd bonds is even a poetic term for gods, so this is an entirely normal name
for a goddess.
In Alvssml, the term draumnjrun is used in close proximity to the
giant Nrr (in the dative, Nrvi), the father of the night goddess Ntt. If the
name Njrun is related to that of another deity, this Nrr is highly likely to be
the one. It is true that the underlying root vowel of the former must be *e and
the latter *a to account for the respective breaking and umlaut, but sgeir
Blndal Magnsson (1989, 671 [nrva]) notes that this alternation is attested in
words related to nrr, for example Nrvasund/Nrvasund.
A third possibility is that Njrun is related to the matronae name
Nervinae, and therefore also possibly to the Nervii (Neumann [1983] 2008, 46),
a Belgic tribe who were of Germanic origin according to Strabo (ed. Jones
1917, 194) and Tacitus (1914, 3023). Given the possibility of a relationship
50

between Gefjun and Hlyn and the Gabiae-related matronae names and the
name Hludana, this explanation is especially interesting, and will be explored in
much greater detail in section 4.2.2., but for now it should also be pointed out
that a connection between Njrun and Nrr can easily coexist with a purly
linguistic association between Njrun and the Nervinae. This is to say nothing of
an actual historical continuity; it is by no means beyond reason that the name
might occur twice.
It should also be considered that Njrun may never have been a goddess
at all, and the appearance of the name in the Nafnaulur could be a mistake
resulting from the use of the term draumnjrun in Alvssml. It is true that it is
used in woman-kennings in skaldic poetry in a manner similar to those which
make use of goddess names, but the same is true of the word Nauma (Lexicon
Poeticum 422 [Nauma]), which never occurs in reference to a goddess, even in
the ulur.
For a final point on Njrun it should be pointed out that while works
which connect the name to Njrr and to Nerthus tend to regard Njrun as an
earth deity (de Vries 1931, 367; Nsstrm 1995, 60; Hopkins 2012, 3944),
it is by no means necessary to maintain the connection to these other names in
order to interpret the name as one of an earth deity. In describing several
Matronae name components, Neumann ([2003] 2008, 231) said [...] *nerwndie Engstelle im Gelnde usw. Das sind ezeichnungen, die kleine Siedlungen
benennen mgen, vielleicht auch speziell den Kultplatz der Matronen, where

51

*nerwn- is represented in the name Nervinae, but theoritically could contribute

to the name Njrun as well, see 4.2.2.4 for more on this possibility.
4.1.1.3. *Ingun?
The god Freyr, who is also called Yngvi, is called Ingunar Freyr by
Byggvir in Lokasenna strophe 43 (GKS 3265 4to [16r29] ingunar freyr).
Taken at face-value, the first component resembles a genitive of an unattested
feminine personal name *Ingun. While the feminine personal name Ingunn,
genitive Ingunnar, with geminate -nn is very common, it is never attested with
only a single -n unless in this compound. As Ynguni is a word referring to an
Ynglingr king, this has occasionally been interpreted to mean Lord of the
Ynglings like the phrase fra Ingwina in line 1319 of Beowulf (ed. Heaney 2000,
923), but Ingunar is clearly a singular genitive, not plural. Kock believed it to
be a contraction from Inguna rfreyr but as Schck (1940, 291) points out, the
two-word phrase is always written with a space between Ingunar and Freyr. The
idea that *Ingun could be the name of a deity or of the earth preserved in this
compound has been addressed by Schck (1940), although he considered it a
mistake for Ingunn.
The declension of Ingunn reveals a means by which Ingunar might be
explained as a relic form preserved in the compound Ingunar-Freyr. The words
declension resembles that of an i-stem, the endings of which are often replaced
by those of an -stem, but is often preserved in personal names, perhaps due to
the influence of ij-stems (see Johnsen 2005, 629 [2.4.2]). Compare the

52

following paradigms of Hildr and Gunnr, respectively an ij- and i-stem

(Johnsen 2005, 65):
nom.

Ingunn

Gunnr

Hildr

acc.

Ingunni

Gunni

Hildi

dat.

Ingunni

Gunni

Hildi

gen.

Ingunnar

Gunnar

Hildar

Table 4.1: Declension of feminine personal names Ingunn, Gunnr, and Hildr

If Ingunn was also an ij-stem, then the second final n may have come about by
assimilation of the nominative ending: *-n > -nn. At this point, the geminate nn may have spread analogically to the rest of the paradigm on the basis of
common feminine names ending -unn such as Steinunn (gen. Steinunnar) and
runn (gen. runnar). In this case, the development would be as follows:
nom.

*Inguni

>

acc.

*Inguni

*Inguni

Ingunni

dat.

*Inguni

*Inguni

Ingunni

gen.

*Inguna

Ingunar

Ingunnar

>

Ingunn

Ingunn

Table 4.2: a concievable, though speculative, development of the name Ingunn

This is a rather complicated and speculative solution, and the situation

regarding Freyr and Yng-/Ing- names is more complicated than just this. In his
study of the *-no- suffix in deity names, Meid (1957, 1234) argued for a
name Yngvin on the basis of Ynglinga saga chapter 17, offering a text
normalized en Yngvi ea Yngvin var kallar hverr eira tmanna alla vi en
53

Ynglingar allir saman and each of their kinsmen were called Yngvi or Yngvin
for their lifetime, and Ynglingar altogether. The word here normalized Yngvin
is found in manuscript sources written yng ni (AM 35 fol., dated 1685-1700)
and ngvin (Holm papp 18 fol., dated 1681-82); it is omitted from AM 45
fol., and interestingly, in AM 38 fol. yngune is crossed out and corrected to
Ingi in the margin. It is normalized to Ynguni by Bjarni Aalbjarnarson
(ed. 1941, 34). Clearly the identity of this word is far from certain, but its mere
existence means that great caution should be taken in interpreting Ingunar.
While the suggestion offered here is possible there is no compelling evidence
for it having happened.
4.1.1.4. Conclusions
Ultimately neither name Njrun or *Ingun can contribute significantly to
the study of the name Gefjun. In the case of the former, this is due to an
overabundance of possible derivations, and the latter, because of great
uncertainty surrounding the word. A single formational suffix cannot be
identified on the basis of these words which results in an ending -un. A strong
case can be made for Njrun being a deverbal, but this does not preclude other
possibilities.

4.1.2. Names in -yn

4.1.2.1. Introduction
The following names all end with -yn, genitive -ynjar. The names
Fjrgyn and Hlyn have recieved frequent commentary with little concensus,
54

but the suggestion has been made that the Herrschersuffix may have played a
part, especially in Fjrgyn. The name Sigyn has not been associated with the
Herrschersuffix, nor is there reason to believe it should be, but it may be
exemplary of a different name component, perhaps from earlier *-vin, genitive
*-vinjar.
4.1.2.2. Fjrgyn
Fjrgyn is a name well-attested in the Old Norse corpus as a name for
the earth and the mother of rr. In Vlusp strophe 56 rr is called
Fjrgynjar burr the son of Fjrgyn and rr is also called the son of Fjrgyn in
Hrbarslj strophe 56. It is listed as a heiti of the earth in Skldskaparml
chapter 57 and an anonymous drttkvtt verse is quoted in support of this, and
it appears in the accusative fjrgynju (fiogynio in AM 2365 4to Codex Regius,
39r5) in Oddrnargrtr. Meanwhile the masculine name Fjrgynn appears in
Lokasenna strophe 26 when Loki calls Frigg Fjrgyns mr and also the kenning
dttir Fjrgyns listed for Frigg in chapter 19 of Skldskaparml.
Within Germanic languages Fjrgyn can be compared to Goth fairguni f.
mountain, mountain range, OE firgen, fyrgen n. mountain, mountain
woodland, OHG Fergunna, Firgunnea, Virgunnia a mountain range in central
Germany (de Vries 1962, 126 [fjrgyn]; Lehmann 1989, 1045 [fairguni]).
Of all the names under observation in the present work, Fjrgyn and its
masculine counterpart Fjrgynn may be the most discussed and debated due the
relationship it may or may not have to supposed cognates in other Indo-

55

European languages. In particular the Lithuanian name of the thunder god

erk nas has inspired comparison (see: Meid 1957, 125-126; Friedrich 1970,
133-140; Maher [1973] 1977, 457-458; a terse summary of the scholarship can
be found in Lehman 1989, 104-105 [fairguni]). The common origin of the two
words is said to be PIE *perkus m. oak tree (Lat. quercus m. id.). It is
suggested that the oak tree was associated with thunder
Meid (1957, 125-126) accepts the connection but not without hesitation.
He explains that in order for the comparison to work, the masculine Fjrgynn,
apparently a ja-stem due to the umlaut in the second syllable, would have to
have been reformed based on the feminine Fjrgyn, taking the place of the true
cognate which in Proto-Germanic would have been *Fergunaz, but which has
no attested reflexes.
In order for Fjrgyn to be related to fjrr m. tree or fjr n. life (cf.
Goth. fairus m. world), its formation must predate Verners Law, with an
accent not fixed on the first syllable. There is no reason to believe this could
not have happened, in fact Goth. fairus m. world seems highly applicable as
a name of the earth. The word fjrn, discussed in 4.1.3.6 could have resulted
from paradigmatic split if the pre-Verners Law form had an alternating stress
pattern, causing voicing only in come cases. There may have been a stress
pattern that would cause voicing by Verners Law if inn was formed from a
*wuz before the PGmc. accent became fixed, resulting in PGmc. *wanaz
(to explain OE Wden, not *Wen). However without demonstrating such a

56

stress pattern to associate with such a suffix, this remains ad hoc and
speculative.
4.1.2.3. Hlyn
Like Fjrgyn, the name Hlyn is listed as a name of the earth and the
mother of rr. In verse 56 of Vlusp rr is called mgr Hlynjar son of
Hlyn. Skldskaparml chapter 57 lists it as a heiti for the earth and cites a
verse from Vlu-Steinns gmundardrpa which uses a kenning Hlynjar beinir
Hlyns bones, meaning rocks. Strophe 26 of Vellekla by Einarr sklaglamm
uses the name in the genitive in a construction interpreted by Finnur Jnsson as
myrk- Hlvinjar -markar and to mean Jylland (Skj BI:122[27]).
The name is unambiguously a j-stem, with the genitive almost always
attested with an ending -jar even when abbreviated (cf. Bergsbk, Holm perg 1
fol 19va12: hlodyni), though there is a rare exception in AM 54 fol 9rb14
hlodynar, a copy of lfs saga Tryggvasonar en mesta dated to the end of the
fourteenth century. There is a small amount of variation in the suffix vowel y;
in AM 748 II 4to 6v1, dated 1390-1410, it appears hlovin and in Codex
Trajectinus Traj 1374 fol [46r10] it is written hlodun, but these rare variants
are not enough to challenge the standard -yn, genitive -ynjar.
In Vellekla, Hlynjar is sometimes replaced with Foldynjar in
manuscripts of Heimskringla (AM 35 fol [147r23] foldyniar; AM 45 fol 24
folldyni, 11; AM 38 fol [81r19] foldyni) though manuscripts of lafs saga
Tryggvasonar en mesta have Hlyn consistently (AM 61 fol [14v25]

57

hlodyniar; AM 54 fol [9r14] hlodynar; Holm perg 1 fol [19va12]

hlodyn). The mistake may have been introduced by a form with the h
unwritten, as in AM 37 fol [86v14] myrklodyniar if a scribe then attempted to
correct lodyniar and recognized that it mean earth but chose the word
incorrectly. The word fold f. earth; grassy field has the same basic meaning,
and because the syllable bears neither rhyme nor alliteration the change is
inconsequential to the metre.
The vowel of the first syllable is uncertain; it is often given Hl- (Skj BI:
122[27]) but also Hl- (de Vries 1962, 239 [Hlyn]) with all parties involved
largely uncertain and usually noting the uncertainty. Manuscript attestations
with the vowel written any differently from o were not able to be produced,
and this holds for manuscripts that do tend to differentiate orthographically
between o and umlauted vowels (GKS 2365 4to [2v15] reads a k i mi
mg hloyniar kjmr inn mri mgr Hlynjar for example) and so Hlyn
seems the more likely of the two. This is also supported by Johnsens (2005,
138145) articulation of Dahls Law, wherein an ending *-(i)j, when preceded
by two light syllables, manifests the Sievers variant *-ij as if following a heavy
syllable. If the name is Hlyn it seems that the syllablic structure would be ,
and would end up as *Hlynr, genitive *Hlynar.
Even narrowing down the root vowel there are several possibilities for
the origin of the root. Some that come to mind are OI hlir f.pl. hearth and
OE hl f. spoils, booty; troop, gang although neither these is particularly
relevant mythologically. sgeir Blndal Magnsson (1989, 341 [Hlyn]) refers
58

to a dialectal Sw. ld meaning hay- or cornstack which stands more to reason.

An interesting suggestion (de Vries 1962, 239 [Hlyn]) is that it is derived
from the verb hlaa to pile, load, in which case the root comes from the
preterite singular stem. This is relevant to Gefjun in that *ga- is the preterite
singular stem of the verb *gean-, although the j would still not be accounted
for. However, the word ld which sgeir Blndal Magnsson (1989, 341
[Hlyn]) mentions is itself almost definitely derived from hlaa and so there is
not significant reason to say that Hlyn derived directly from the verb.
4.1.2.4. Sigyn
Sigyn is the name of Lokis wife. Gylfaginning ch. 50 and the epilogue to
Lokasenna explain that the sir bound Loki to a rock in a cave by his sons'
intestines, and hung a venomous snake above his head so that the venom would
drip on his face. Sigyn holds a bowl above Loki to catch the venom before it
hits him, though when the bowl fills and she has to empty it, Loki writhes in
pain, causing earthquakes.
The name itself is usually believed to be composed of sig- victory and
*vin f. (female) friend, (Simek 1993, 284 [Sigyn]), or rather sig n. rope which
is let down; rope with weights (de Vries 1962, 474 [Sigyn]; O 1989, 813
[Sigyn]), with the second component surfacing as -yn in the same manner as the
alternation between Bjrgvin and Bjrgyn f.sb. Bergen, Norway. A word *vin
as a simplex with the meaning (female) friend is not attested, but a feminine
n-stem vina with that meaning does occur, as well as a j-stem vin, gen. vinjar

59

but with the meaning meadow. A genitive of Sigyn occurs in Haustlng

strophe 7 and in the list of kennings for Loki listed in Skldskaparml chapter
16. While readings of the nominative Sigyn are common in manuscripts (cf.
Codex Regius GKS 2365 4to [2r9] igyn; Hauksbk AM 544 4to [20v13]
igyn), the genitive varies between Sigynjar (igyniar AM 242 fol, [55:24]
and Traj 1374 fol [26r30]); and Signjar (ignar GKS 3267 4to [25v16]), the
latter most likely incorrectly chosen over the former because the name Sign is
a common femine personal name.
As *vin f. (female) friend is unattested, it is difficult to conjecture what
its genitive might have been if it existed. Alongside the Old Icelandic n-stem
vina there is also the Old High German winia Freundin, Gelibte, also an nstem according to Braune (2004, 1:211 [226]), but an -stem substantive is
lacking, and perhaps it is best to look elsewhere for an explanation of the name.
The simplest explanation is not vastly different from the traditional one
above. As there are few indications that Sigyn was any kind of earth goddess
unless by association with the earthquakes that Loki causes, vin f. meadow is
probably not correct, but OE wyn, wynn f. joy (cf. OS wunnia f. id.) is very
common in Old English compounded personal names. It can be reconstructed
to PGmc. *wunj, which would most likely result in an Old Icelandic *yn. It
actually may be that de Vries (1962, 474 [Sigyn]; 664 [vin]) meant to imply this
when he wrote *Sig-vin but underspecified which of his definitions for vin he
meant, as he includes OE wyn in his discussion of it.

60

A name composed of compounds meaning victory and joy does not

seem to fit with the mythological character of Sigyn, but this has more to do
with the first component rather than the second. The etymology endorsed by
de Vries (1962, 474 [Sigyn]) and sgeir Blndal Magnsson (1989, 813 [Sigyn])
from sig n. rope which is let down; rope with weights is more relevant to
extant myths, since Sigyn is helpful to Loki when he is tied down. Considering
the frequency of sig- victory in personal names, the possibility that it is a
generic name without correlation to mythological function remains.
4.1.2.5. Discussion
More endings -yn, genitive -ynja have been proposed for these three
goddess names than goddesses themselves. One from vin f. meadow is known
from Bjrgvin ~ Bjrgyn f. ergen, Norway (sgeir Blndal Magnsson 1989,
61 [Bjrgvin, Bjrgvin, Bjrgyn]) and could apply to Fjrgyn or Hlyn because
they are both names for the earth.
None of these solutions seem useful for describing Sigyn. She is not an
earth-goddess as far as can be discerned, nor does it seem fitting to identify her
with victory; a word *vin, genitive *vinjar f. (female) friend poses etymological
difficulties. The simplest explanation is that -yn descends from the same source
as OE wyn(n) f. joy, a common name component, that being a PGmc. *wunj.
It is of course not certain that the simplest explanation is best; perhaps
mythological information has been lost that would reveal that another ending is
more fitting, and on the other hand perhaps *yn f. joy is a component of

61

Fjrgyn and Hlyn. These names have been debated for ages and it does not
seem likely that the debate will end in the near future.
Whatever the situation, it is clear that none of these names are
particularly useful in explaining the construction of Gefjun. They hardly vary
from a suffix vowel y, and there is no situation wherein Gefjun could have been
formed with a like suffix without umlaut. If Johnsen (2005, 138145) is right
about the effects of Dahls Law in Old Icelandic, Gefjun would probably have
been analyzed as two short syllables and recieved the Sievers variant *-ij,
which would result in an Old Icelandic *Gefynr or *Gefynn. The names in -yn
appear to be wholly separate from the formation of Gefjun.

4.1.3. Names in -n
4.1.3.1. Introduction
Goddess names which, like Gefn, end with -n and no suffix vowel, can
be found in Snorri Sturlusons Gylfaginning as well as a few other sources.
Gylfaginning mentions Sjfn, a goddess who turns peoples' minds to love, and
Lofn, another love goddess who can intervene to overturn prohibited marriages
(ed. Faulkes 2005, 2930 [Ch. 35]). To these can possibly be added Hrn or
Hrn, a name for Freyja according to Nafnaulur (ed. Faulkes 1998, 130 [501]);
mrn, which is usually interpreted to mean giant-woman (Lexicon Poeicum 420
[2. mrn]; deVries 1962, 26 [Mrn]); Fjrn, a name for the earth listed in
Nafnaulur (ed. Faulkes 1998, 130 [501]); and Njrn, a name found in kvenna
heiti kennd (Skj AI:698 [3]).

62

4.1.3.2. Sjfn
Sjfn is described in Gylfaginning chapter 35 (ed. Faulkes 2005, 29-30),
listed in the Nafnaulur (ed. Faulkes 1998, 114-115 [verse 435]), and used in
three woman kennings; one in Plctsdrpa (Skj AI: 610[14]) and two in
lausavsur from Gsla saga Srssonar (Skj BI: 99[17], 103[35]). Unfortunately in
none of these attestations is there a genitive, as it would be extremely valuable
to know whether it would be Sjafnar as in Modern Icelandic or *Sjfnar. There
are a number of compounds incorporating Sjafnar- as a component, such as the
plant name Sjafnargras Thalictrum minus, but these cannot be demonstrated
to have been constructed during the Old Icelandic period, and the root vowel a
may be analogical.
On the goddess, Snorri says: Sjaunda Sjfn: hon gtir mjk til at sna
hugum manna til sta, kvenna ok karla. Af hennar nafni er elskuginn kallar sjafni.
(Seventh [is] Sjfn, she looks after turning peoples' minds to love, women and
men, and from her name is love called sjafni). The word sjafni appears in
Skldskaparml chapter 70 (ed. Faulkes 1998, 108) and Nafnaulur under
Hugar heiti ok hjarta (heiti of the mind and heart; Skj AI:688[zz]). Snorris folk
etymology is not likely to be correct, but the words do appear to be related. It
seems unlikely that Sjfn is derived directly from the verb sefa to soothe,
which was already a class-two -verb in the earliest manuscripts (cf.
Morkinskinna, GKS 1009 fol) and has no cognates in other Germanic
languages. More likely is a derivation from sefi m. mind, sense, a masculine nstem cognate to Old English sefa understanding, mind, heart, which would
63

require that the word Sjfn not be a deverbal abstract. The words sefi and sjafni
are considered to derive from the same source, a single word with an ablauting
n-stem paradigm *se-an- ~ *se-n- (Hreinn Benediktsson [1986] 2002, 325).
There is therefore more reason to believe that Sjfn is related to sjafni and sefi
than the verb sefa.
4.1.3.3. Lofn
Of Lofn, Gylfaginning chapter 35 says:
tta Lofn: hon er sv mild ok g til heita at hon fr leyfi af
Alfr ea Frigg til manna samgangs, kvenna ok karla, tt r
s bannat ea vertekit. Fyrir v er af hennar nafni lof kallat, ok
sv at er lofat er mjk af mnnum. (ed. Faulkes 2005, 29)
Eighth [is] Lofn, she is so gentle and good to call on that she
gets permission from Allfather or Frigg to arrange relationships
between people, women and men, though they were previously
prohibited or denied. It is from her name that lof (praise) is
known, and that she is lofu ('praised') much by men.
(authors translation)
The word Lofn is only attested as a personal name, and the genitive Lofnar
appears in a drttkvtt verse from Gunnlaugs saga ormstungu (eds. Sigurur
Nordal and Guni Jnsson 1938, 107 [verse 25]). The name is attested in
various skaldic poems as a part of woman kennings (Skj BI: 277[1], 385[5],
523[4], 600[32]; BII: 427[33], 5734[18], 575[23]). In Old Icelandic, the -verb
verb lofa means to promise; to praise; to permit. It also makes up the
compound name Lofnheir. It stands alongside lofun permission and lof, n.
praise; permission, which Snorri connects to the goddess and accounts for
64

both of the meanings permit and praise: the goddess acquires permission, for
which she is praised. Also related are the ija-verb leyfa to permit; to praise,
leyfi sb.n. leave, permission, and the adjective ljfr mild, gentle.
The rather broad semantic range of these words, and the place of Lofn
among them, can be clarified somewhat with cognates in other Germanic
languages. The -verb lofa to promise; praise; permit also has a reflex in OE
lufian to praise, highly value and derivative lufung, sb.f. the act of loving,
which corresponds by means of its construction to OI lofun sb.f. permission.20
Cognate to the OI leyfa to permit; to praise and leyfi leave, permission are
Goth. galaubjan v. to believe; to permit and galaubeins belief, OE (ge)leafa
sb.m. permission and (ge)leaf sb.n. permission. In Gothic, *lubains sb.f.
hope, attested in the genitive singular lubainais answers to OE lufen sb.f.
hope. These seem to be reconstructable to *luniz, itself derived from an verb which almost certainly meant to hope. The verb itself is attested in OHG
lobn, but it has merged with the -stem lobn in meaning (von Steinmeyer, et
al. 1968, 5:1235 [lobn]). Most of the words here analyzed have a great deal of
semantic overlap in the daughter languages and therefore it would not be
surprising if they did already in the proto-language, so it is significant that the
-stem and its derivatives are identical in meaning in both Gothic and Old
English.

20

Unlike both North and East Germanic, West Germanic formed deverbal feminine abstracts
from -verb *-ung- rather than *-ni- (Krahe and Meid 1969, III:117 [98]; 209-211
[152]).

65

Though the -verb does not occur in Old Icelandic, it is from this source
that Lofn is most easily explained as deriving, as the name is exactly the
expected cognate of Gothic *lubains and Old English lufen. If this is the case,
the name means hope rather than relating directly to praise or permission, but
because the verb meaning hope no longer survived in Snorris language, he
was unable to identify the goddess with it. If this is true, Snorri probably did
not have any mythological information about her, but still needed to explain
the name because of its use in skaldic poetry. This conclusion is furthermore
supported by the female personal name Lubaini (dative) attested in a Latin
inscription in modern-day Belgium (CIL XIII 3622), which Neumann ([1983]
2008, 38) agrees corresponds to Gothic lubains as discussed above.
Meid (1957, 121) considered Lofn a feminine Hoffmann formation
equivalent to a masculine name Lobbon(n)us appearing in a devotional
inscription from Utrecht, and reconstructed *Luan, but Simek (1993, 189
[(Lobbon(n)us)]) dismisses this as imaginative interpretation of an illegible
inscription. Meids reconstruction is feasible, but his evidence should be
rejected. It is also not clear that the second-syllable *a in *Luan would have
syncopated in OI, see 6.2.
4.1.3.4. Hrn, H rn
Hrn or Hrn is listed as one of Freyjas several alternative names listed
in Gylfaginning chapter 35, along with Gefn (ed. Faulkes 2005, 29) and in
Nafnaulur appears both in a list of names of Freyja and names of troll-women.

66

The genitive occurs in the third strophe of Einarr Sklasons drttkvtt poem
xarflokkr, which the manuscript evidence suggests is Hrnar or Hrnar with
the same umlauted vowel as the nominative (honar GKS 3267 4to 28r32;
honar AM 242 fol 73,7; hona Traj 1347 fol 29r32). The identity of the
root vowel is uncertain; it is sometimes given as Hrn (deVries 1962, 277
[Hrn]; sgeir Blndal Magnsson 1989, 413 [1 Hrn, H rn]) though also to
Hrn (Skj B:II, 234[15], 436[37]; rhallur Vilmundarson and Bjarni
Vilhjlmsson, eds. 2009, 481). Verse 11 of the drttkvtt poem Geirviardrpa
by Stjrnu-Oddi Helgason rhymes it with stjrnu, though it is not clear that the
poem can be trusted to distinguish and due to the possibility that it was
composed to too late a date (Stjrnu-Odda draumr, in eds. rhallur
Vilmundarson and Bjarni Vilhjlmsson 2009, ccxxii-ccxxiii [35]).
Because of the umlauted genitive, the word has been connected to the
masculine wa-stem hrr, genitive hrvar flax; linen (sgeir Blndal
Magnsson 1989, 413 [1 Hrn, H rn]). The reason for this is that the rootfinal *w of *harwaz or *harzwaz, from which hrr is likely derived, is a possible
explanation for umlaut where the inflectional ending is not responsible for it.
A vowel would have arisen by means of both i- and u-umlaut, cf. OI
skkva v. to sink, from *sankwijan- (Hreinn Benediktsson 2002c [1963], 151;
given as PN *sankwian). sgeir Blndal Magnsson (1989, 413 [1 Hrn,
H rn]) suggests a reconstruction *harw-n, meaning flax-goddess, agreeing
with de Vries (1962, 277 [Hrn]), which is possible but would likely mean the
OI form is Hrn, as it has already been demonstrated that the ending *-n
67

regularly fails to cause i-umlaut (see 3.1.2). On the other hand, the name is
possibly reflected in Swedish place names Hrnevi and Jrnevi (de Vries (1962,
277 [Hrn]), and East Norse, unlike its western counterpart, regularly does show
i-umlaut in these constructions (cf. OS, OD vrn, OI vrn f. defense). The
question then becomes whether or not the u-umlaut is regular, and why it is
not shown in the Swedish place-names. This possibility can be demonstrated by
Dan. gre, Sw. gra v. to do, prepare, OI gra v. id. from *garwijan- (Ringe
2006, 222), alongside Dan. snke, Sw. snka v. to sink (transitive), OI skkva
id. from *sankwijan- (Hreinn Benediktsson [1963] 2002, 151). It may be that
Hrn represents the regular development in Old Icelandic, while Hrn is a
loanword from an East Norse language, which was subsequently unrounded in
its language of origin, giving Hrn- in compounds.
4.1.3.5. Mrn
The word mrn is somewhat obscure, both in terms of its phonological
history and its meaning. It occurs twice in kennings for the jtunn jazi in
Haustlng (Skj BI: 15[6], 16[12]), wherein he is called fair mrna and fair
mrnar. It also occurs repeatedly in Vlsa ttr (eds. Gubrandr Vigfsson and
Unger 1862, 2:331336), the participants in a pagan ritual recite verses in turn
and each end them with iggi mrnir etta blti (may mrnir accept this
sacrifice), although it is not certain that the word is a plural of mrn f. trollwoman or a singular mrnir m. sword (name for Vlsi?). It also occurs in
Nafnaulur (ed. Faulkes 1998, 112 [425]), two verses by Einarr Gilsson,

68

rsdrpa, and an anonymous fornyrislag verse from the 13th century (Skj BI:
141[7], BII: 152[22], 422[16], 438[14]). It is generally interpreted to mean a
type of ogress or other female supernatural being. jazi is a giant and the father
of Skai, who is probably meant when mrn is used in the singular in
Haustlng.21 The root vowel rather than is confirmed by rsdrpa 7, line 6:
s barna ser mrnar (Finnur Jnsson 1912, AI:149 [verse 7]), where is
necessary for mrnar to achieve full-rhyme with barna.
The manuscript evidence suggests that the root contained an umlauted
vowel even in the singular genitive (mrnar), plural nominative (mrnir), and
possibly plural genitive (mrna). Flateyjarbk GKS 1005 fol [121vb37] reads
maurn (maurnir) although as this is from Vlsa ttr it is uncertain whether
this is a plural of a feminine word mrn or a singular of a masculine word
mrnir, as described above (facsimile Finnur Jnsson 1930). Examples of more
secure identifications from with the feminine substantive from Haustlng
include GKS 2367 4to [25v14] morna and [25v32] mornar, Traj 1375 fol
[26v10] mornar, and AM 262 fol. [55:22] morna.
It seems unlikely that an analogical change would have generalized the
vowel, as alternation between and a is extremely common in feminine words,
including feminine deverbal abstracts in -n < *ni-, such as vrn, gen. varnar
defense (from verja to defend) and gn, gen. agnar silence (from egja to

21

The plural form in the poem is sometimes corrected to singular mrnar, but whether jazi
was father to one or many, he is himself a giant so the definition of mrn must at least
include, if not exclusively, 'offspring of a giant'.

69

be silent; see also 3.1.2). Therefore it would be a very rare exception if the
word derived from the ja-verb merja to bruise, crush, past tense mari, for
which a derived noun mrn should have a genitive *marnar. If a phonological
explanation lies behind the genitive and plural nominative, it might be
compared to gjlnar f.pl. gills. The fact that it shows umlaut throughout the
paradigm means that the umlaut is independent of the feminine -stem
substantive suffix and suggests a construction similar to that of Gefn, which
does not show the expected u-umlaut or breaking, so in both cases there must
be a syncopated vowel that affected the root, blocking the effect of the
morphological ending.
One possibility is that the word is related to the word mrr, dative mrvi
suet. In the modern language there is a verb mrva to fatten an animal, but it
is rare, not found in ONP, and is a class-two -verb with a third-person
preterite mrvai, so an expected feminine deverbal abstract would be
*mrun/*mrvan or modern *mrvun, and would be very difficult to explain
why this would be an appropriate name for a being or group of beings. If, on
the other hand, the ending were a personalizing suffix describing, perhaps, a
type of spirit associated with hunting or herding animals, which receives
sacrifices of animal fat, are household kitchen spirits, or some other speculative
solution, it may not be necessary to derive the word from a verb22.

22

If Skai is meant by the singular mrn in Haustlng the former two suggestions seem more
applicable, though the occurrence in Vlsa ttr leaves room for the latter; the word is not

70

Another possibility, suggested by sgeir Blndal Magnsson (1989, 654

[Mrn]), is that the word is related to mara f. 'mare', a malicious spirit in
folklore which rides a sleeping person, causing them to have bad dreams, and
which is reflected in the second component of the English nightmare. sgeir
Blndal Magnsson (1989, 654 [Mrn]) gives the reconstruction *mar(u)n, but
the u which he gives in parentheses is necessary to explain the ubiquitous
umlaut throughout the paradigm. Although feminine deverbals from *-ni- are
polysyllabics which regularly fail to show i-umlaut (Johnsen 2012; see 3.1.2),
this is apparently not the case with polysyllabics with u as the vowel of the
second syllable. For example the datives of jkull m. glacier and mndull m.
handle are jkli and mndli, which show u-umlaut after the vowel which
triggered it has been syncopated. Though this can possibly be explained as
analogical leveling, the same is not seen among masculine polysyllabics with iumlaut, such as ketill m. kettle, dative katli. It seems that mrn allows for two
possible reconstructions, *marw-an- and *mar-un-, with the morphological
ending subsequently being changed to *- or -u depending on the time of the
change. It is not altogether clear how a derivation from mara f. mare (spirit)
would produce a *marun, although a relationship between mara and mrn on
mythological grounds is more intuitive.

metrically necessary in Vlsa ttr so the poet seems to have chosen it unmotivated by
external requirements.

71

4.1.3.6. Fjrn
Nafnaulur (ed. Faulkes 1998, 130 [501]) and kvenna heiti kennd (Skj
AI: 698[3]) list Fjrn, the former as a name for the earth, mentioning Fjrgyn in
the same verse. The word could be related to OI fjr n. life, OE feorh n. id.,
Got. fairus m. world. This is the interpretation supported by sgeir lndal
Magnsson (1989, 183 [fjrn]), who reconstructs *ferhwn, in which case it is
possible that it is related to Fjrgyn, (see 4.1.2.2 above for Fjrgyn).
In 3.2 a suffix was mentioned which is related to the Hoffmann suffix,
another type of formation denoting a place which is characterized by an
abundance of something. The examples given were Bkn and Hven, two names
for islands listed in the ulur, with names which might mean a place
with/characterized by beech trees and a place with/characterized by foam. It
would make sense for fjrn, as a name of the earth, to be interpreted as a place
where there is fjr (life). It is not unlikely that Fjrgyn and fjrn are formed
from the same root, although either, or both, could just as easily be formed
from fjrr m. tree rather than fjr n. life, but it is more likely that they were
formed separately, with Fjrgyn being the older form, composed before the
phonemicization of Verners Law in order for the change *h to *g to occur. The
evidence of the Matronae Nerhvinae (4.2.2.3) speaks in support of this theory.
4.1.3.7. Njrn
The name Njrn is listed in kvenna heiti kennd while Njrun is not,
which might be taken to mean they are alternates of each other, and therefore

72

also possibly that Gefjun and Gefn are as well. Though possible, it is also not
quite definitive, as names listed in kvenna heiti kennd include many which are
not goddess names; it is simply a list of words which can be used as
components of kennings for woman, and includes goddess names, trees, and
other things. Another possibility that should not be discounted is that Njrn is a
misreading of Mrn, given that i is not dotted in early orthography and capitals
are not common; it can be rather difficult to discern norn from morn.
Indeed, this very thing may have happened in trllkvenna heiti, where the word
which Faulkes normalizes to Mrn he reads as niavrn in the Codex Regius
GKS 2367 4to [42r22] (ed. Faulkes 1998, 150). Nonetheless, that they may be
alternates of each other cannot be ruled out. Again, the Herrschersuffix and the
remodelling according to Syrett (1994, 187204 [7.6]) may offer a solution;
Njrn could descend from a remodeled form *nerwn, with the *-- failing to
leave a trace as in the *-ni- deverbals (3.1.2).
It might be unwise to place a deity attested in a Roman-era West
Germanic inscription in a medieval Scandinavian text, but if either of these two
names is related to the Nervinae (4.2.2.4), Njrn is the more expected reflex.
Still, with so few attestations and no attested genitive, it is even more difficult
to make conjectures about Njrn than about Njrun.

73

4.1.3.8. Discussion
Of the names listed above, only one, Lofn, can be easily derived from a
verb23. It is possible that Sjfn derived from the verb sefa (to soothe), but this
is less likely than other possibilities. The others have major obstacles to such a
derivation. Mrn is the best attested of the words analyzed here and has the
advantage of frequent use in the plural, and curiously it shows umlaut
throughout the paradigm, probably due to either a stem-final w in the root or a
syncopated u in an earlier, polysyllabic form of the word.
Many of these names show strong signs of Hoffmann formations
according to the remodeling of *-an- in participles, adjectives, and possibly
substantives proposed by Syrett (1994, 2034 [7.6.5], for which see 3.2).
Hrn/H rn in particular might reflect East- and West-Norse variants exactly as
predicted if it comes from *harwn. The possible doublet Njrun : Njrn might
be explainable as respectively deriving from a pre-remodeling *nerwan and
post-remodeling *nerwn (see also 4.2.2.4). This may go a long way in
explaining Gefjun and Gefn, a possibility which will be explored in the coming
chapters as more evidence is gathered.

23

Note that, while a verbal derivation for Lofn is supportable by evidence and certainly the
simplest solution, it is also not entirely necessary; if an alternative solution must be sought
for the other names then it may apply to Lofn as well, and without a plural it is difficult to
know with certainty what class of substantive Lofn is.

74

4.2. Matronae and Latin Devotional Epigraphy

4.2.1. The Matronae cult
An important source for pagan beliefs of the first few centuries, A.D. is a
body of votive inscriptions concentrating in the areas of the Roman Empire
along the Rhine River. Hundreds of inscriptions and relief sculptures have been
found which have been dedicated to gods and goddesses worshiped by Celtic
and Germanic peoples (Beck 2009). The inscriptions are in Latin, and Roman
deity names appear frequently, though often accompanied by a name or epithet
which seems to reflect the language of the local culture (Simek 1993, 2078
[Matron names]; Beck 2009, 4950). For example Mars Thingsus, attested on a
fourth-century stone altar at Hadrians Wall, is taken to be a god presiding over
the legal assembly known in Old Icelandic as the ing. He is often identified by
modern scholars with the god Tr because they were equated though
interpretation germanica, resulting in modern Tuesday from Latin dies Marti,
although it is also possible that this was an independent deity unrelated to Tr
and unknown in Icelandic sources (Simek 1996, 203 [Mars Thingus]). In other
inscriptions, there are no identifiable Roman theonyms, but rather a title in
Latin, or sometimes simply a name adapted into Latin from the Germanic or
Celtic (e.g. Hludana, Nehelennia). An important subgroup of these divinities is
those which appear primarily in the plural, generally depicted in groups of three
when accompanied by an abbreviation, and are given the title mtrs or
mtrnae (Beck 2009, 39). Epithets other than mtrs or mtrnae are much less

75

common, but include I nns Junos, Parcae fates, Fatae fates, Nymphae
nymphs; brides, and others ( eck 2009, 82106).
There are no sources which elucidate the worship of the matronae
beyond the inscriptions themselves, and much remains obscure about its
nature. For example, it is difficult to understand why the matronae often appear
in groups of three, and several possible interpretations present themselves.
Triplism is a common element in Gaulish archaeology and insular Celtic
mythology including triplicate goddesses such as the Mrrgain, and even the
island of Ireland is represented by the three goddesses riu, Banba, and Ftla
(Beck 2009, 778). Triplism also occurs frequently in Germanic mythology,
notably in the case of the three norns Urr, Verandi, and Skuld, who,
according to Vlusp emerge from a body of water at the base of the world tree
(or a hall by the well according to Gylfaginning chapter 15; ed. Faulkes 2005,
18) and dispense fate (Vlusp, ed. Neckel and Kuhn [verse 20]); this might
also underlie the set of three creator gods inn, Vili, and V found in
Gylfaginning chapter 6. An important question is, since the manifestation of the
matronae cult in Latin devotional inscriptions clearly reflects a partial
syncretism of elements from different cultures, whether the designation
matronae or the triplicate form were actually native to the Germanic peoples, or
whether one or both spread along with the concept of devotional inscriptions in
general.
It is unclear how the word matrona should be interpreted. The basic
meaning is married woman but can also be applied to high-ranking young
76

girls (Glare 1968, 1084 [mtrna]). Some or all may have been considered
literal mothers, and their epithets may in some cases reflect tribal associations,
as Shaw (2011) has argued especially regarding the Matronae Austriahenae.
However Simek (1996, 204 [Matrons]) notes that the iconocraphy found in
relief illustrations include depictions of girls with loose hair, suggesting that
they are not married, in addition to others with the dress of married women.
The title matrona should, in this context, be regarded as indicating importance
and authority rather than meaning that matronae are necessarily married women
or mothers, even though that also occurs.

4.2.2. Comparison of Matronae names to Old Icelandic names

4.2.2.1. (-)Gabiae : Gefjun/Gefn
At least fourteen devotional inscriptions with theonyms including an
element gabi can be found in the epigraphic corpus.24 The name Gabiabus,
occurs three times on its own (CIL XIII 7856, 7939, 7940), three times with the
title Matronis (CIL XIII 7780, 7937, 7938), twice accompanied by Iunonibus
Junos; once as a title (CIL XIII 8192) and interestingly once reading Iunonibus
sive Gabiabus to the Junos or the Gabiae (CIL XIII 8612). It is also found in a
compound Matronis Alagabiabus (CIL XIII 8529) which is believed to be a
Germanic prefix *ala- all, and also Ollogabiabus (CIL XIII 7280) which is
CIL XIII 7280 Ollogabiabvs, 7780 Matronis Gabiabvs, 7856 Gabiabvs, 7867 Deae Idbans
Gabiae, 7937 Matronis Gabiabvs, 7938 Matronis Gabiabvs, 7939 Gabiabvs, 7940 Gabiabvs,
24

8192 Ivnoniibus Cabiabvs, 8529 Matronis Alagabiabvs, 8612 Ivnonibvs sive Gabiabvs; AE 1924
Deabus Alaisiagis Baudihillie et Friagabi, 94; AE 1981 Gabinis, 678; RIB 1071 Deae
Garmangabi.

77

taken to be the Celtic equivalent of the same (Beck 2009, 67). The form
Gabiabus is, according to Neumann ([2003] 2008, 246), is a nonstandard dative
plural also found for example in deabus f. goddess (dat.) indicating a
nominative plural Gabiae, and the name also appears in a singular dative Deae
Idbans Gabiae (CIL XIII 7867; Gutenbrunner 1936, 90 reads Idiangabiae
instead; Simek 1996, 170 [Idban(?)gabia] says it is difficult to be sure). A similar
Gabinis (AE 1981, 678), also dative plural (nominative given Gabinae by Beck
2009, 66), is found, and two compound names, Deae Garmangabi (RIB 1071)
and Friagabi (AE 1924, 94) also occur. All of these are found in Germany
except for Garmangabi and Friagabi, which are both located in Britain.
The etymology of Gabiae and the other similar names have been debated
thoroughly. Much (1891, 3167) considered it a deverbal from the same
source as the word gefjanda that appears in the Codex Regius (GKS 2365 4to
[22r25]), describing Njrr as gefjanda gu. He does not propose a meaning, but
says that the Gothic would be *gabjn. This explanation does not seem to have
enjoyed much currency, presumably because the reading gefjanda in GKS 2365
4to does not seem widely regarded as accurately reflecting a kenning for Njrr
(see section 6.1.2 for discussion of *gefja) although on strictly phonological
grounds this makes more sense than a derivation from *gean- v. to give.
The discussion of Kern (1870, 1567) is more interesting. He connects
the name to Dutch gave f. gift, talent as well as to OI gifta v. to marry
(transitive); to give away in marriage. With unfortunately little elaboration he
notes that in an inscription Matronibus is replaced by Junonibus and that this
78

may signify that they were considered to be in his words Junones pronubae25.
As discussed in section 6.1, the identification of gipta and Gabiae or Gefjun is
not etymologically straightforward, but there are other reasons to suppose a
connection to marriage. In Roman religion, Juno was a goddess of marriage,
although it is now believed that the direct association she was once considered
to have with the marriage itself is innaccurate (Hersch 2010, 2624). In Norse
mythology Gefjun is said to be a goddess of unmarried women, which is
contrasts sharply with Juno (and so perhaps also the Gabiae) but within the
same thematic sphere, relating respectively to the time before and the time after
the marriage.
The most common interpretation of Gabiae is similar to the most
common interpretation of Gefjun, that it is ultimately derived from a verb
*gean- to give and that it means givers (Neumann [1987] 2008, 263; [2003]
2008, 233; Beck 2009, 66; Simek 1996, 97 [Gabiae]). At one time Neumann
([2003] 2008, 233), who wrote about the word several times throughout his
career, compares its formation to fylgja f. fetch, a type of spirit in order to
account for the j, not noticing that fylgja is formed from a verb which also has a
j. He also compared OHG rt-gebo m. counselor and OE ring-geba m. ringgiver, but again, there is no explanation for the *j or *ij that he attempted to
explain by comparison to fylgja. He also noted that this explanation did not
account for the root vowel e rather than a.

25

For discussion of Roman pronubae see Hersch 2010, especially pp. 190212.

79

While not all of Neumanns argument holds up, one element of it is

particularly interesting. In his discussion of Garmangabis (Neumann [1998]
2008, 365366 [Garmangabis]), he wrote that in Gabiabus (dat.) and -gabi
(dat.), there is evidence of an */j-stem substantive with an inflectional
paradigm descendent from the same origin as Skt. dev f. goddess, nom.pl.
devya; Lithuanian mart f. girl, nom.pl. mar ios. Gutenbrunner (1936, 44)
made the same suggestion and offered a reconstruction *ga(z)26, supposing a
nominative singular Gabis. There are other reasons to suggest a PGmc. *ga ~
*gaj- as well, for which see 6.1.2. If Johnsen (2005, 1179 [2.11.4]) is
correct that *gan f. (Goth. gabei f. riches) resulted from an n-stem extension
to an original *ga, the n-stem may be what resulted in the name Gabinis
(dat.). It should be noted that alongside -gabi there is also a singular dative
Deae Idbans Gabiae, possibly indicating uncertainty regarding how to adapt the
word to the Latin declension system; Gutenbrunner (1936, 90) gives this a
nominative singular Idiangabis. The partial paradigm in Latin which can be
arranged from the works of Neumann ([1998] 2008, 3656 [Garmangabis]) and
Gutenbrunner (1936, 434, 90) is as follows:

26

The PGmc nominative ending of /j-stems was *-, with a Gothic reflex -i (cf. mawi, acc.

mauja f. girl) but reflexes in OI have a nominative ending -r such as mr, acc. mey (Johnsen
2005, 627 [4.2.2.1.]). West Germanic lost word-final *-z so it is unsure what happened in
the nominative of /j-stems (Johnsen 2005, 356 [2.2.1]), but Gutenbrunner seems to
think, or at least consider the possibility, that it was also added to this word class in West
Germanic before it was lost.

80

singular

plural

nom.

*Gabis

*Gabiae

dat.

Gabi/Gabiae

Gabiabus

Table 4.3: A partial paradigm of a Latinized Germanic gabiae

according to Neumann ([1998] 2008, 3656 [Garmangabis]) and
Gutenbrunner (1936, 434, 90)

It is perhaps unwise to put too much faith into the evidence provided by
Latin endings attached to Germanic root sas a reflection the inflectional
morphology of the Germanic word, but this is a compelling argument. A Latin
d.sg. Gabi would be expected to correspond to a nominative Gabis, but a d.sg.
Gabiae would be a closer approximation of a Germanic /j-stem, possibly
*gaji or gajai according to Johnsen (2005, 37 [2.2.3]).
4.2.2.2. Hludana/Hluena : Hloyn
Hludana is not a matrona, but is attested rather as a singular dea f.
goddess. Meid (1957, 120) considered the name to be composed from a word
holda, a type of spirit or demon, but that metathesis does not seem very likely.
Gutenbrunner (1936, 745) considered it related to Hlyn and to be a name
for the earth as a goddess. There may be precedent for the vowel * being
written u in Roman Germania with alternation between Romanehae and
Rumanehae (Neumann [1987] 2008, 255), although these could also reflect a
Latinized version (based on Rm) and a form based on Germanic *R m (cf.
Goth. Ruma). Gutenbrunner (1936, 837) notes that in one inscription it is
spelled Hluena and calls the suffix vowel e Schwachung (weakening). Both
81

the root and the suffix are difficult to explain, but perhaps rather than
weakening the variation in suffix reflects either ablaut or productivity of one
suffix. He accurately points out that even if the roots are the same in Hludana
and Hlyn, the suffixes are completely different. This might be due to active
processes of name formation, but unfortunately any relationship or lack thereof
between the two names is impossible to determine with confidence.
4.2.2.3. Alaferhviae : Fjrgyn
The Matronae or Nymphae Alaferhviae seem to have a name composed of
*ala- all attached to *ferhw-, either meaning tree (OI fjrr tree, Lat. quercus
m. oak tree) or life or world (OI fjr n. life, OE feorh n. life, Goth.
fairus m. world; this interpretation is supported by Gutenbrunner [2001]
2008, 404). If Fjrgyn is composed of one of these words, which seems likely,
the consonant *h has become voiced by Verners Law, which has clearly not
happened to Alaferhviae. What can be gathered from this is that the word
probably had currency as a naming component, with different examples
forming independently of each other.
4.2.2.4. Nervinae : Njrun
The Nervinae are attested in one inscription in Bavay, France (CIL XIII,
3569) in the dative plural Nervinis with a first declension ending. The name
seems to reflect a tribal affiliation with the Nervii described by Tacitus (ed.
Hutton 1914, 302-3) and Strabo (ed. Jones 1917, 194) as a tribe residing in
Gaul but claiming Germanic ancestry, and whose capital was Bavay according
82

to Gutenbrunner (1936, 153). According to Neuman ([2003] 2008, 230), the

goddesses were named for the tribe, which was itself named for the landscape in
which they lived or originated; that is, they may have lived near a land
formation or body of water which was named for the characteristic attribute of
being narrow, and cites Njrvasund for precedent of the use of the adjective in
place names. His proposed etymology is that the name was an adjectival suffix
*-n- attached to *narwa- adj. eingeschnrt, eng; though Neumann does not
mention it specifically there is precedent for the fluctuation between *narwaand *nerwa-, for which see 4.1.1.2.
The name could easily be built on the same root as OI Njrun, as a
PGmc. *nerw- would most likely develop into *njrv- by u-breaking, and the v
would then be lost before the round vowel u. If Nervinae is a Romanized plural
of a Germanic *nerwn, it could even be cognate with the name Njrn that
appears in Nafnaulur (for Njrn see 4.1.3.7; for syncopy of unstressed * see
3.1.2 and 3.2).
It is very reasonable to connect the Nervinae to Njrun and Njrn on
strictly phonological grounds, but there are even greater concerns related to the
possibe transmission of the name than there are with Gabiae and Gefjun. The
Nervii were a Belgic tribe, although allegedly they considered themselves to be
of Germanic ancestry. It would be difficult to explain why the name of a
goddess or set of goddesses with affiliation to a specific tribe that was prominent
in the first few centuries A.D. in Gaul would be preserved intact a thousand
years later in Iceland. A solution may lie in Neumanns ([2003] 2008, 231)
83

understanding. It need not be assumed that the presence of similar name on

two disparate fringes of the history of European polytheism indicates and actual
continuity between the Nervinae attested in Gaul and the names Njrun and
Njrn in Nafnaulur and skaldic poetry if instead it is supposed that *narwa- ~
*nerwa- had widespread currency as a possible component for compounding
and derivation of names. Note that accourding to sgeir Blndal Magnsson
(1989, 670 [Njarar]) the group name Njarar m.pl. appearing in Vlundarkvia
has been connected to the Nervii, though sgeir Blndal Magnsson does not
express acceptance of that etymology.
If it is true that Nervinae is composed with an adjectival suffix *-n-, it is
evidence in favor of Syretts (1994, 187204 [7.6]) suggestion that the
adjectival suffix *-n- could be used to form substantives in the vein of inn,
and simultaneously accords with Meids (1957, 7581) description of the *-nosuffix denoting representation of a group of people, though that might be
skipping a step. Even if Neumanns ([2003] 2008, 231) *nerwn- is accepted, the
medial vowel could result from the tribal name, which Gutenbrunner (1936,
153) reconstructs *Nerwiz (*Nerwijz by more current orthography), rather
than from a method of word formation which Syrett (1994, 187-204 [7.6])
proposed specifically for North Germanic. Neumanns ([2003] 2008, 231)
phonological reconstruction makes sense, but the actual process of derivation
he describes need not be accepted in order to arrive at it.

84

4.3. Conclusion
The Matronae Gabiae and Nervinae provide important evidence for the
interpretation of Gefjun : Gefn and Njrun : Njrn. While the names do not
quite allow for positive identification as the same deities who appear in Old
Icelandic sources, all of the names seem to point toward name-formation as an
active and ongoing process in the first few hundred years, A.D. Perhaps the
most valuable part of this chapter is the position of Gutenbrunner (1936, 44)
and Neumann ([1998] 2008, 3656 [Garmangabis]) that the name of the
Matronae Gabiae is composed from an /j-stem, which is an idea that will be
explored in greater detail in 6.1.3.3.
So far, a strong case can be made for both deverbals and Hoffmann
formations as sources of feminine names in Old Icelandic. It is true that Lofn
could theoretically also be a Hoffmann formation, the name Lubaini (dative) in
a Latin inscription (CIL XIII 3622) strongly suggests a deverbal from an -verb.
If the name in the inscription is not of parallel formation to OI Lofn, it is still
evidence for deverbal name-derivation. For other names such as Hrn and mrn
other explanations have needed to be saught and the Hoffmann formation
seems to better account for the words and their paradigms than other
explanations. No conclusion can be reached until the word Gefjun is examined
more closely, in order to determine with the most confidence possible its
paradigm as it existed in Old Icelandic.

85

5. Old Icelandic Manuscript Evidence

5.1 Attestations of Gefjun
In preparation for the present work, 150 attestations of the name Gefjun
were collected from manuscripts when available and editions of manuscripts if
they were available but the manuscript itself was not. An additional attestation
was rejected for illegibility. There was one instance where the scribe included
two variants, in which case both variants are counted. The attestations are listed
in Appendix 1 along with their grammatical case, manuscript, title of the text,
and approximate date. The page and line number are included when possible,
otherwise the page number in the edition is given.
There are no major surprises regarding the first syllable. Typically the
name begins gef and occasionally gief, the latter reflecting palatalization of
g by the following front vowel e (Kristjn rnason 1003 [6.2.2]). Only one,
LBS 3936 4to 248r24-25 Gif|in dated 18801883, definitely has a different
root vowel. AM 235 fol 37r26, dated c.1400 looks like gofion although it is
not certain; the scribes e has a large loop in general and it probably is
gefion. A handful of copies of Droplaugarsona saga write v for f (JS 450 4to
[34r12] Gievion; JS 630 4to [78v4] Gevion; L S 2116 4to [19r15]
Geveon). In Lbs 756, dated by the scribe in 1777, Gefjun appears four times,
twice (18r8 and 43rc24) written Geffion, and otherwise (8r2 and in another
hand from 1854, 7v17) Gefion. These isolated incidents of variance are
expected and normal.
86

The representation of the suffix vowel is more noteworthy. As discussed

in 3.1.3, Old Icelandic had only three unstressed vowels, and the non-low,
back, rounded vowel which Hreinn Benediktsson (2002 [1962], 789 [2.4])
writes [] could be written o, u, or v without any distinction or implied
difference in sound. With attestations of Gefjun there is fluctuation between o
and u (or v, an allograph of u in medieval Icelandic orthography), but
o dominates to an unexpected degree. Two instances of a were also found,
but it should be noted that both of these (rubric in AM 35 fol 9v and Holm.
papp. 18 fol 2v) are copies of the same, mostly lost, manuscript Kringla Lbs
fragm 82. AM 35 fol even has a note by the copyist correcting it to Gefion
(both are counted in the analysis). Finally, one instance of an ending -inn was
found in AM 750 4to [5r1-2], dated to the second half of the 17th century,
reading Giefion e gief in (Gefjun ea gefinn); only the former was counted.
Unlike the analysis of *ni- deverbals in 3.1.3, attestations of Gefjun
were gathered from any date, including many post-medieval manuscripts. This
decision was made for the purpose of observing the development of the word
into the modern paradigm.
Total

Percentage

1.33

126

84.00

u/v 22

14.66

Table 5.1: the representation of suffix vowel in attestation of Gefjun by total and percentage.

87

The prevelence of o is striking. This extends even to manuscripts

which typically render the unstressed non-low back vowel with u, for
example Codex Womianus AM 242 fol. wherein u clearly dominates (for
example on page 8, with three attestations of Gefjun, only one unstressed nonlow back vowel occurs written o, in troo Trju Troy, compared to at least
thirty written u), but Gefjun is spelled with a suffix vowel o in all of its
seven occurrences (see also AM 242 fol. [9:18] huu hverju, [10:9] miiu
mijum, [13:19] iarnviur jrnvijur for representation of unstressed sequence
ju in the same manuscript).
It is also noteworthy that among those that do use u, the Codex Regius
of the Prose Edda GKS 2367 4to uses u in five out of seven attestations in
the same text as Codex Womianus AM 242 fol. and it is inconcievable that
Codex Wormianus was the innovator in this regard. A u spelling is also
common in copies of Nafnaulur and Jns saga postula. AM 544 4to Hauksbk
uses both interchangeably. In thirty-five attestations from copies of
Droplaugarsona saga, only two, Lbs 1511 4to [111v11], dated 1888, and Lbs
5157 4to [15v18] dated to the second half of the 19th century, use u. Only
one u was found in a copy of Heimskringla (AM 38 fol [4v18]) making
Gefian slightly less rare within that text (out of 12 attestations found). In
many texs such as Lokasenna, Vlsa ttr, and most saints sagas, u is never
used at all.
This data can be compared to table 3 in 3.1.3 for a comparison to *-nideverbals. The difference is quite extreme; the prevelence of o in Gefjun,
88

84%, is almost the same as the 83.83% prevalence of a in such deverbals. As a

result, the argument that Gefjun is a deverbal is severely weakened, and
explanation should be saught elsewhere if possible.
No data was found whatsoever to support an accusative and dative
Gefjuni as in the modern paradigm, even with accusatives and datives from as
late as the 19th century, although they are all certainly copies of early
examplars. Interestingly, 50% of twelve genitives found use u, but this seems
to be a product of the genitive being attested mostly in texts which otherwise
mostly use u otherwise.

5.2. Interpreting the data

As previously noted, because the Old Icelandic unstressed non-low,
round, back vowel which Hreinn Benediktsson (2002 [1962], 78-79 [2.4])
denotes [] could generally be written with either grapheme. However, the
orthographic representation of the unstressed vowel in Gefjun with the letter
o in 84% of attestations gathered greatly exceeds what is expected, especially
in manuscripts where u dominates such as Codex Wormianus AM 242 fol, as
noted with examples in the preceding section. The strength of this tendency is
great enough that it forces the consideration that the vowel somehow is not the
unstressed vowel [].
There are certain conditions where a syllable that does not recieve
primary stress may contain a vowel other than the three which are considered to
make up the system of unstressed vowels (see: Hreinn Benediktsson ([1962]
2002, 7491). For example, there can be found such as heilg f. adj. holy,
89

masculine heilagr (but also gmul f. adj. old, masculine gamall), or a front
vowel which is marked consistently across manuscripts, such as manneskja f.
person, human, which is written with e in all attestations listed in the ONP.
In Modern Icelandic, disyllabic neuters ending -an have plurals in -n, such as
lkan n. model, n.pl. lkn. Unfortunately these are not well-attested in in the
plural nominative or accusative in early manuscripts. In AM 623 4to (ed.
Finnur Jnsson 1927, 39) dated to c.1325, is found malikon, though better
evidence would be valuable, because the manuscript uses o for o, , and the
unstressed round vowel (see: ollom for llum, adj.dat.pl. all) and a paradigm
of sg. lkan ~ pl. *lkn would be highly vulnerable to analogical change.
When o is used in Gefjun it is never marked in a significant way to
differentiate it from a normal o, but there is an orthographic feature of Old
Icelandic which may be the cause of this. Generally the in the sequence j
written io, with the umlauted vowel marked only rarely (Hreinn
Benediktsson 2002 [1963], 160). Icelandic j arose by u-breaking of *e, but
while examples are difficult or impossible to produce, it is not beyond
possibility that it could also occur by umlaut of a preserved Proto-Norse *ja
which did not itself arise by a-breaking. The tendency for Gefjun to be rendered
gefion is so strong that it is worth considering whether the form underlying
this representation is Gefjn, from PGmc. *gajan, with the second-syllable
vowel unmarked due to orthographic convention.
A PGmc. *gajan (or more realistically, PN *gajanu) would be the
expected form of a thematicized Hoffman formation, or the Herrschersuffix,
90

made to *gaj-, as compounds beginning with -stems generally use *a as the

compositional vowel (Syrett 1994, 745 [4.5]).
There is of course a severe difficulty in declaring Gefjun to have been
composed with this suffix, which is that no other word with the same
conditions can be produced for comparison. Njrun, which is consistently
written with u or v in the second syllable does not qualify as
counterevidence, because it is not certain that it is the same formation and even
if it is, an early form *njrvn would be highly susceptible to structurallyconditioned analogical change as Hreinn Benediktsson (2002 [1962], 78 f.7)
describes happening to the personal name *sigvrr > Sigurr. Njrun is also
attested neither as early nor often as Gefjun. However, Meid (1957, 72-126) has
made a strong case the the suffix once had greater currency in the Germanic
languages, and the possibility of the remodeling proposed by Syrett (1994, 203204 [7.6.5]) for masculine substantives also applying to feminine substantives
may have obscured evidence for more such formations in North Germanic. So
it is unfortunate that there is little to compare Gefjun to, but there are also
reasons to explain why this is so; there is a high chance that there were more
comparable examples at some point in the past, which are now lost.
A rather serious, but not insurmountable, weakness of this theory is that
if a form of the word with were in the language of the earliest scribes, a
genitive *Gefjanar would certainly be expected. However, the genitive mostly
appears in texts otherwise associated with representing the vowel u. Only one
genitive, in Nikolaus saga erkibiskups, is found in a text which does not use a
91

u spelling at least once. It may be an unfortunate coincidence that of the

genitive attestations found, all feature analogical change of to u. More
thorough research of the earliest manuscripts would be elucidating.

5.3. Summary
Manuscript evidence has revealed valuable evidence about the word
Gefjun. Two important points can be discerned. First of all, the name does not
have the distribution of suffix vowels a and u that is seen in *-ni- deverbals
such as skipan ~ skipun, with only two examples of a vowel a, and they are
copied from the same source. The other point is that the distribution of
spellings with the suffix vowel written o and u is overwhelmingly in favor
of o and would be difficult to explain as purly othographic. The solution
proposed was that the vowel was not u at all, but rather , which is usually
written o when following a j, as in jtunheimum iotvheim (GKS 2367 4to
1v14).
In Table 5.2 the proposed paradigm for Gefjun composed with the
Hoffmann suffix is contrasted with the paradigm that would occur by regular
development from an *-ni- deverbal (b) and by comparison to attestations of
*-ni- deverbals (c). Note that paradigm (b) would probably discernible in
manuscripts, as 3.1.3 has shown that the distribution of suffix vowels a and u
was not predictable by case.

92

sg.

a.

b.

c.

* Gefjn

Gefjun

Gefjun ~ *Gefjan

* Gefjn

Gefjun

Gefjun ~ *Gefjan

nom.

( )

acc.

( )

dat.

( )

* Gefjn

Gefjun

Gefjun ~ Gefjan

gen.

*Gefjanar

*Gefjanar

Gefjun ~ *Gefjanar

Table 5.2: Possible declensions of the name Gefjun in very early Old Icelandic.
Paradigm (a) is the declension proposed for a Hoffmann formation, while
paradigm (b) is the declension for a deverbal derivation (cf. 3.1.2).

It should be remembered that the *-ni- deverbals are themselves held

to have gone through a period where their realization was * (sgeir Blndal
Magnsson 1989, 1086 [-un]), but that this had changed to u due to its
unstressed position at some point before the earliest Old Icelandic manuscripts.
The difference proposed here is that there may have been secondary stress, the
same argument invoked by Syrett (1994, 196204 [7.6.46]) to explain why
the ending *-n- does not typically cause umlaut. By either derivation, an
original became u, the difference is a matter of timing and means. According
to the proposal for a Hoffmann formation here the change still occurred, but
not until after it had already been written, and perhaps with dialectal or even
sociolectal differences, as u writings are not spread equally across texts.
Because alternation between a and is extremely common in Old
Icelandic, it is true that a speaker of the language with a word Gefjn in their
vocabulary would be expected to produce a genitive *Gefjanar. However the
distribution of genitive forms is very inconveniently associated with u
spellings and does not entirely preclude its existence.

93

6. Reconstructing Gefjun
6.1.The first component: gefj6.1.1. The verb gefa to give and derivatives
Though *gean- v. to give was ruled out as the immediate source of the
derivation Gefjun, the roots are strongly suggestive of a relationship, especially
considering the preterite singular stem of the verb, *ga-. The word *geanwill have to be investigated in order to determine what the relationship, if any,
actually is.
6.1.1.1. Reflexes in Germanic
PGmc. *gean-, a strong verb of the fifth class meaning to give, and
with the principle parts pres. *ge-, pret.sg. *ga-, pret.pl. *g-, pret. part.
*geana- can be reconstructed securely from Germanic evidence, including
Goth. giban v. to give, OI gefa v. id., OE gifan/giefan/gyfan v. id., OS
gean/gevan v. id., OHG geban v. id. and others. A clear deverbal derivative
is *ge f. gift, cf. Goth. giba f. id., OI gjf f. id., OE gifu/gyfu/giefu f. id.,
OS. geva/gea f. id., OHG gebe f. id.. From either the verb or the substantive
comes an adjective, OI gjfull munificent, OE gifol generous, probably
PGmc. *geulaz unless it was coined in one of the daughters and spread
afterward. There is another derivative which sgeir Blndal Magnsson (1989,
244 [gift]) reconstructs *gefti-, but which would have become *gifti- f. following
the raising of *e before a high front vowel (Ringe 2006, 2205 [4.2.2 (i)]), cf.
Goth. fragifts f. gift, grant, betrothal, OE gift f. a giving; bride-price; wedding
94

(esp. in the plural), OI gift/gipt f. luck; gift with compounds relating to

marriage, cf. giftar-jr f. dowry farm, giftar-ml n. marriage; German Gift f.
poison. It appears to have formed from *gean- using the feminine action
noun suffix *-i- (Ringe 2006, 113 [3.2.4 (iv)]), so its meaning at the time of
its formation seems to have been act of giving, arriving at the meanings in the
daughter languages over time.
In West Germanic can also be found an -verb, perhaps a derivation
from *ge f. gift, OE gifian v. to bestow gifts, OS gevon/geon/geoian v. to
give, make donations, bestow, OHG gebn v. to bestow gifts. It should be
noted that the OE gifian does not reflect a PGmc. semivowel *j following a *,
as this would cause gemination before being lost, yielding an OE root ending
with -bb- (Hogg 2011, 712 [4.11-4.14]; see also 2.3.2 regarding OE
geofon). Old English and Old Frisian restructured their -verb paradigm,
replacing *-- with *-ja- in most inflectional cases (Cowgill 1959, 1-15), so
that the OE cognate to OI laa, Got. lan, from Proto-Germanic *lan-, is
laian. The 3.sg.pret. gifode supports this development. For lexical reasons
along with the root vowel i, OE gifig adj., which Bosworth (1972, 64 [gifig])
translates as possessing as the result of a gift, is probably denominative from
the OE substantive gifu f. gift, rather than a cognate to Got. gabeigs adj. rich.
Beyond these, it becomes more difficult to determine how other words
may be related to *gean-. Words beginning *g-, with a long *, are typically
considered to descend from the same root as well. A substantive *gijn- f.
seems reconstructable at least for North Germanic, cf. OI gfa f. good luck,
95

Far. gva f. id., No. gjve f. id.. Kroonen (2013, 173 [gbn-]) reconstructs
another related substantive *gn- for OI gfa f. (spiritual) gift, talent, Far.
gva f. id., MHG gbe f. id. but sgeir lndal Magnsson (1989, 223, 292
[gfa; gfa]) suggests instead that MHG gbe f. (spiritual) gift, talent also
descends from *gijn-, and was then spread as a loanword from a West
Germanic language into the North Germanic languages resulting in OI gfa,
etc. Alongside the substantive is an adjective *giz, cf. OI gfr adj. meek;
pleasant, Faroese gvur adj. doughty, excellent, honest, OF jve, gve adj.
usual(?), MLG geve adj. pleasant, MHG gebe adj. acceptable, enjoyable,
convenient. sgeir Blndal Magnsson (1989, 292 [gfa]) adds Ice. gf f.
tranquility, gftir f.pl. weather at sea; care for animals cf. NNo. gjft f.
benevolence; a feeding (of animals). According to Kroonen (2013, 173
[gbi-]), *giz adj. is a

-derivation from *gean- with the original

meaning having been given.

There is also a series of words in *ga-, resembling the preterite singular
stem of the verb *gean-, treated in detail in 6.1.2 below. The reconstructable
PGmc. and a selection of their reflexes are presented in Table 6.1 below.

96

Table 6.1: Proto-Germanic words with roots *ge- and *gReconstructable Proto-Germanic words with a root *ge-,
*gean-, *ga-, *gb-,
*geanaz s.v. to give
*ge f.

Goth. giban; OI gefa; OE gifan/giefan/gyfan; OS gean/gevan;

OHG geban
Goth. giba f. gift; OI gjf f. gift; OE gifu/gyfu/giefu f. gift
*gifig (from *geigaz?); OS geva/gea f. gift; OHG gebe f. gift

*geulaz adj. generous

OI gjfull munificent; OE gifol generous

*getiz > *giftiz f. gift27

Goth. fragifts f. gift, grant, betrothal, ; OE gift f. a giving; brideprice; wedding; OI gift/gipt f. luck; gift (giftar-ml n.
marriage); German Gift f. poison.

*gen- -verb
to give gifts

OE gifian v. to bestow gifts; OS gevon/geon/geoian v. to give,

make donations, bestow; OHG gebn v. to bestow gifts

*gean- n. giver

OI -gjafi m. giver (in compounds); OE gifa/gyfa/giefa/

geofa m. giver; OHG kebo m. giver, donor

Reconstructable Proto-Germanic words with a root *g-,

*gijn-

OI gfa f. luck; Far. gva f. luck; NNo. gjva f. luck

*gn-?28

OI gfa f. (spiritual) gift, talent; Sw. gva f. gift;

*giz adj.

OI gfr adj. meek; pleasant; Faroese gvur adj. good,

advantage; talent; MLG gve f. gift; talent

excellent, honest; MLG geve adj. pleasant; MHG gebe
adj. acceptable, enjoyable, convenient.

Also Ice. gftir f.pl. weather at sea; care for animals, NNo. gjft benevolence; a
feeding (of animals); OI gf f. meekness, Ice. gf f. tranquility? The latter
seems to be formed with a *-i- suffix at a relatively late time because *-iderivations continued to form with PGmc. *-ti- when following an obstruent.
sgeir Blndal Magnsson (1989, 292 [gf, g f]) reconstructs PN *giu but its
declension is that of an i-stem so that is probably wrong..

The PIE suffix *-ti- usually becomes *-i- by Grimms law but this often is not shown
immediately following another obstruent, cf. *khpts which became PGmc. *haftaz m.
captive; all labial stops become *f before *t (Ringe 2006, 93102 [3.2.4 (i)]; 112116; [3.2.4
27

(iv)])
28

Unnecessary if MLG gve reflects *gijn- and was loaned into North Germanic.

97

6.1.1.2. Indo-European Etymology

Establishing an Indo-European etymology for *gean- has proven
difficult. Seebold 1970, 2179 [geb-a-]) explains that it is usually connected to
words such as OIr. gaibid v. take, Gaulish gabi v.2nd.imp. take!, Latin habe
v. I have Polish gaba to grab, and Lithuanian gabnti v. to carry, transport
(cf. also sgeir Blndal Magnsson 1989, 263 [gefa]. The meaning is therefore
roughly opposite its proposed cognates but this is not a major obstacle, as they
all deal with transfer or stasis of location or ownership, cf. Lat. d v. I give and
cognate Hitt. d-/d- v. to take, wed, decide (de Vaan 2008, 1745 [d]).
What remains are phonological problems; the Italic and Celtic forms have an
*a, the origin of which is difficult to determine. There is disagreement as to
whether or not a PIE phoneme *a is reconstructable at all; see for example
Beekes (1995, 1389 [11.7.2.]) who argues that it was limited to expressive
words like Goth. atta father and in loanwords, though it is accepted for
example by Ringe (2006, 101 [2.2.3.]) who nonetheless describes it as
uncommon. A sound a in daughter languages usually came about by an IE *e
in contact with a laryngeal *h, or from any laryngeal in syllabic position, for
which see Lehmann (1980, 2235). In some languages such as PGmc., an *a
can also result from a PIE short *o (Ringe 2006, 1456, [3.2.7 (i)]).
A PIE root *ghabh- was reconstructed by Pokorny (1959, 2:4079
[gab-]) without incorporation of the laryngeal theory, whence the Italic,
Celtic, Baltic, and Slavic terms mentioned above, as well as Germanic words
related to Goth. gabei f. riches (from PGmc. *gan-), OI gfr pleasant, mild,
98

and even the Matronae names related to Gabiae, but the PGmc. strong verb
*gean- is not connected to the Indo-European root, but rather said to be an
invention modeled on *neman- v. to take. Rix, et al. (LIV 2001, 193, 195)
give two different PIE roots, one *geb-, pres. *gb-e-, perf. *ge-gb/gb-,
from which he derives PGmc. *gean- and Lith. geb ti v. to be able, noting
some complications in the Gmc. pret.pl. *g- from *gegb-, and the possibiliy
of comparing Lith. gabenti v. to carry, transport, Pol. gaba v. to grab, and
Ved. gbhasti- m. hand. On the other root he is more tentative, giving
?*g()heHb-, aorist *g()heHb-/*g()hHb-, pres. *g()hHb-i-, essive *g()hHb-hi-,
and derives Gaulish gabi v.2nd.imp. take! and OIr. gaibid v. takes from the
present and Latin habe v. I have from the essive. He says that the second of
these roots is only found in Italic and Celtic languages (LIV 2001, 195
[?*g()heHb-]).
Kortlandt (1992, 104-105) tried to explain *gean- as the reanalysis of
the prefix *ga- attached to a verbal root cognate to Hitt. epzi v. siezes and Lat.
apscor v. I reach and cop v. I began. This was accepted by Kroonen (2013,
172-173 [*geban-]), but this is part of a larger argument of his about the fifth
class of strong verbs that suffers from a lack of evidence and examples, as
pointed out by Mailhammer (2007, 8081) who then offered a much more
satisfactory explanation rendering Kortlandts (1992) explanation unnecessary.
See Mailhammer (2007, 6786 [3.1.8]).
Bjorvand and Lindeman (2007, 3513 [gi]) propose that the root a of
the Italic and Celtic reflexes is analogical, while deriving the Lithuanian stems
99

with root vowel a from the o-grade *gob-. While this seems possible, it would
only partially explain the relationship, as the j-present form of the Latin and
Celtic forms is lacking from the Germanic (cf. Seebold 1970, 219 [geb-a-]).
Neither de Vaan (2008, 278 [habe]) nor Matasovi (2009, 149 [*gab-yo-])
accept *gean- as cognate, respectively to Lat. habe and PC *gab-yo- to take,
hold; de Vaan (2008, 278 [habe]) reconstructs pres. *ghb(h)-(e)i-, aor.
*ghb(h)-eh- for the Italic and Celtic forms (2008, 278 [habe]), which does not
conflict very strongly with LIVs pres. *g()hHb-i-, aor. *g()heHb-/*g()hHb- (LIV
2001, 195 [?*g()heHb-]) other than the root-final unaspirated *b. This was
probably chosen because of Umbrian habe hat ergriffen, habetu soll halten;
PIE *b usually becomes f in Umbrian and Oscan (Buck 1904, 79 [124]).
However, Schrijver (1991, 92[f]) explains the root-final consonants in Umbrian
and Oscan reflexes result either from an athematic conjugation or analogically
from capi v. cature, take. De Vaan (2008, 278 [habe]) maintains the
possibility of a *b but regards it as unlikely; Matasovi (2009, 149 [*gab-yo-])
does not even mention the possibility.
It is with the Italic and Celtic derivations that Gmc. *gi- adj. (OI gfr
adj. mild; pleasant) and *gijn- (OI gfa f. good luck) should be considered
to be associated, and if MHG gbe f. (spiritual) gift, talent does not derive
from *gijon- and a PGmc *gbn- must be proposed, the same applies to it as
well. If Schrijver (1991, 92[f]) is correct that Lith. g bti v. to be able is also
cognate, derived from the full grade *gehb-, this better explains the meaning
of the Germanic substantives meaning talent and even good luck from
100

PGmc. *g-, which would therefore have a long root vowel as a result of
compensatory lengthening following the loss of the laryngeal (Ringe 2006, 70
75 [3.2.1(ii)]). On the adjective OI gfr adj. meek; pleasant, Faroese gvur
adj. doughty, excellent, honest, MLG geve adj. pleasant, while clearly not a
parallel formation, are semantically similar to Lat. habilis adj. easy to handle,
derived from habe; the meanings related to mildness and goodness may stem
from a sense of familiarity. The i-stem of the adjective and ijn- in the
substantive are then also explained as simple reflexes of a PIE *i in the
reconstructions proposed by Schrijver (1991, 92[f]), de Vaan (2008, 278
[habe]).
Although the actual reconstructions of PGmc. *gean- on one hand, and
Lat. habe, OIr. gaibid on the other, are clearly still debateable, it is clear that a
connection between them is not easily supportable. While there are derivations
from *gean- within Germanic, it is better not to count Germanic words
beginning *gb- among them. The question then becomes whether or not
PGmc. words beginning *ga-, such as Goth. gabei f. riches and Goth. gabeigs
adj. rich, OI gfugr adj. noble, should be considered derived from *gean- at
all, or whether a completely different origin related to Lat. habe v. have
should be saught instead, as Pokorny (1959:II, 4079 [gab-]) said.

6.1.2. Old Icelandic *gefja?

In 3.1.3, the possibility that Gefjun is a deverbal of the *-ni- type is
discussed, and it is concluded that if this is the case, it is derived from a weak -

101

verb *gefja, as is proposed by Much 1891, 317; North 1998, 223; and
mentioned without necessarily accepting by Sturtevant 1952, 166; the latter two
define it to give without explanation. No reflex of such a verb is found in any
daughter, but Much (1891, 317) draws attention to a kenning for the god
Njrr listed in Skldskaparml in GKS 3267 4to, [22r25], reading gefianda
gu. It is difficult to know how this should be interpreted.
In the position of gefianda, the other principle manuscripts of the
Prose Edda, AM 242 fol Codex Wormianus [48:3], DG 11 4to Codex Upsaliensis
(ed. Heimir Plsson 2012, 144), and Utrecht University Library MS No. 1374
Codex Trajectinus [22v12] all have fgjafa. Faulkes (ed. 1998, 18) keeps the
GKS 3267 text, but emends it to gefanda.
If the reading is correct, (*)gefjanda is a substantive derived from the
present participle of a verb *gefja, here in the genitive (god of *gefjandi(s) or
god of people who *gefja). The present participle does not indicate the class of
the verb and so it cannot confirm *gefja as an -verb. Since it appears in a list
without any context or explanation for why it is a suitable kenning for Njrr, a
meaning giving is only suggested by comparison to variants and not
confirmed. Faulkes emendation to gefanda gu is most likly an accurate
representation of the scribes intention; this is a well-attested word which makes
sense as an alternative to fgjafa gu without change to the meaning. A scribal
error is the simpler and more likely explanation than a hapax legomenon derived
from an unattested verb, and so the reading should be rejected. Furthermore, in

102

5.1 it is shown that representation of Gefjun is emphatically different from

representation of *-ni- deverbals.

6.1.3. Proto-Germanic *ga6.1.3.1. Germanic words and loanwords from Germanic

Words which descend from PGmc. forms beginning *ga- are well
attested, including *gan- (Gothic gabei f. riches) and an adjective
(Gothic gabeigs/gabigs rich, OI gfugr noble). The East- and West-Germanic
universally refer to material wealth, Gothic gabei translating Greek

riches, wealth (Lehmann 1986, 134 [gabei]) and gabeigs translating lo

rich (Lehmann 1986, 134 [gabei]). Old High German possesses a noun listed in
Althochdeutsches Wrterbuch (1986, 174 [gebgi]) as gebgi f.sg.dat. Reichtum,
which appears to be derived from the adjective.

*gan-

Goth. gabei f. riches gabigaba adv. richly, gabigjan v.

to enrich, gabignan v. to be rich

*gagaz adj.

Goth. gabeigs/gabigs adj. rich; OI gfugr adj. noble,

worshipful; ( OHG gebgi f.dat. Reichtum)

*gagn- v.

OI gfga v. to worship

Table 6.2: PGmc. words with roots *ga-, also usually derived from *gean- v. to give

103

According to Krahe and Meid (1969, III:18893 [145]), Gothic gabigs

and its cognates and derivatives are the only examples of such a formation with
the *-ga- suffix with a secure short vowel *i in the second syllable. The form
gabeigs points to a long *, though derivatives gabigaba adv. richly, gabigjan v.
to enrich, and gabignan v. to be rich all suggest a short vowel and the long
variant could be secondary, formed due to the rarity of the short vowel. In
*-ga- formations, the compositional vowel can sometimes be used to determine
the stem vowel of the noun from which the word derives, although many such
adjectives in *-ga- in fact derive from a-, -, ja-, and j-stems (cf. OHG muotig
courageous, OS mdig id., from a-stem noun *mdaz, OHG muot, OS md;
Krahe and Meid (1969, III:18893 [145]). Old Icelandic gfugr does not
necessarily contradict this, as there was considerable variation in these suffixes
in

Old

Icelandic

(cf.

bligr/blugr

adj.

bloody,

nauigr/nauugr

forced/unwilling).
Loanwords probably reflecting the root *ga(j)- appear in non-Germanic
languages as well. Slavic *gobino n. (Old Church Slavonic gobin abundance,
Old Russian gobino abundance, Serbo-Croatian gobino spelt), and derived
adjective *gobn abundant, productive (RuCS, ORu. gobz) are taken to be a
loan of *gain- (Derksen 2007, 171 [*gobino]). A Finnish substantive kapiot pl.
is of special interest. It is only used in the plural, marked by the Finnish caseending -t, and means dowry or trousseau (Kulonen et al. 1992, 305 [kapiot]).
Suomen Sanojen Alkuper Etymologinen Sanakirja (1992, 305 [kapiot]) derives it
from Germanic *gbia- < *gbia-, which should be read as a Proto-Germanic
104

*gija-, reflected in OI gfr adj. meek, quiet; pleasant. The entry cites Posti
1981 in support of this derivation, as well as Lindstrm 1859 deriving from
*gn-, a possible earlier form of OI gfa (spiritual) gift, talent.
Kylstra et al. (1991, 40-41 [kapiot]) lists no fewer than six suggested
Germanic origins for the Finnish word, and in fact Gefjun is among them
although Karsten, the first to propose this theory, later revised it to *gajn-,
from which he derives Swedish dialectal gvo Gabe. A reconstruction
*Gabjn is given for Gefjun without further comment, and although that is
certainly possible it is not very helpful without explanation. More interesting is
the suggestion that kapiot derives from a Proto-Germanic *gaj, from which
Kylstra et al. (1991, 40-41 [kapiot]) derives dialectal Swedish gef Glck,
Geschick. Unfortunately, in preparation for the present work, it was not
possible to confirm the existence of a Sw. gef nor even determine the dialect
from which it is supposed to have come, in order to evaluate the likelihood of it
descending from a PGmc *gaj. Kylstra et al. (1991, 40-41 [kapiot]) does not
specify the source of the word in the entry. Reference material conferred
include Ordbok ver svenska sprket (1898), Svenskt dialektlexikon: ordbok
fver svenska allmogesprket (Rietz, 18621867), Svensk Etymologisk Ordbok
(ed. Hellquist 1922), and some dictionaries particular to certain dialects such as
rdbok ver folkm len i vre

alarna (Levander and Bjrklund, 1961), Ordbok

ver lulemlet (Nystrm, 1993), Ordbok ver Njurundamlet (Hellbom 1980),

Ordbok ver Multrmlet (Nordlander 1933), and Ordbok ver de stsvenska
dialekterna (Vendell 19041907), but gef was not found.
105

It seems unsafe to accept that this form is not an orthographic variant of

gv c. gift. The Swedish adjective gv excellent; pleasant is probably cognate
to Old Icelandic gfr adj. meek, quiet; pleasant, and although a substantive
cognate to OI gfa would be expected to retain the -a ending in Swedish (cf.
gva c. gift, talent, cognate to OI gfa id.), a hypothetical substantive *gv c.
meaning Glck, Geschick could be a secondary derivation from the adjective,
possibly with influence from a substantive cognate to OI gfa. If it does exist,
and if the reconstruction to *gaj is correct, it would prove invaluable to the
current study of the name Gefjun, but unfortunately the inability to determine
its existence with reasonable certainty in time for this paper means that it must
remain hypothetical until further research can be done.
Nonetheless, of the suggested origins of kapiot pl. dowry, treussaeu, the
strongest case can be made for a Germanic *gaj- to account for the form the
Finnish takes. Though a PGmc. *gaj meaning luck does not have enough
evidence to accept, much of the evidence adduced throughout previous
chapters suggest that a word *gaj- did exist, although its meaning has not
been elucidated.
Like Germanic roots in *g-, it is more fitting to identify *ga(j)- with
Lat. habe v. have, OIr. gaibid v. take, from a present stem *ghb-(e)i-. The
source of the vowel *a would then be the laryngeal rather than an o-grade,
which if Schrijver (1991, 92[f]) is correct in accepting Lith. g bti v. to be able
as a cognate, would be gohb- and result in a Germanic *g-, although this is
not necessary. Though it is not problematic to associate a word meaning
106

possessions with a verb meaning give, it is certainly even less problematic to

associate it with a word meaning have. As the PIE etymology of these words is
still disputed it may well be that *gean- and habe are indeed built on the
same root, even if it seems unlikely, but in that case *gaj- was derived so long
ago that it is not bound to being interpreted as related to give.
6.1.3.2. The meaning of *gan, *gagaz
Having established the distribution of the root *ga(j)- throughout
Germanic languages and in loanwords, a closer lexical analysis of attested
reflexes is worthwhile. The meaning of *gan- seems to have been not merely
possessions, but also to imply a greater amount than is necessary or expected.
In the Gothic Bible it is contrasted with spiritual attainment, for example in
Luke 8:14 (KJV): And that which fell among thorns are they, which, when
they have heard, go forth, and are choked with cares and riches (gabei) and
pleasures of this life, and bring no fruit to perfection. Considering the
meaning of the Slavic loanwords, it is unlikely to have specifically meant
wealth in terms of actual capital at the time of the loans since it could
apparently manifest as grain or animals as well as money. Presumably
ownership of it, or at least in abundance, was exclusive to distinguished or
fortunate people, and acquired connotations of nobility, leading to Old
Icelandic gfugr adj. noble, worshipful. There is possibly an example of Gothic
gabei being used in a positive religious sense, in Romans 11:15, translating

f. reconciliation: For if the casting away of them be the

107

reconciling (gabei) of the world, what shall the receiving of them be, but life
from the dead? Streitberg (1910, 41 [gabei]) explains this as an error for
gafrins f. reconciliation which is likely correct.
Old Icelandic gfga has strong religious connotations from an early time.
Its first attestation comes from Elucidarius, dated 1150-1200 (authors
normalization, from AM 674 a 4to [2r5], ed. Jensen and Stefn Karlsson 1983,
7): Discipulus: Sv es sagt at manngi veit hvat go es en oss snisk makligt at vita
eigi hvat vr gfgum (Disciple: It is said that no man knows what God is, but it
seems to us unfair to know not what we worship). In the Icelandic Homily
Book Holm perg 15 4to, dated to c. 1200, it translates Lat. adrre (ed. van
Weenen 1993 [27v25]) honrificre ([40r2]), honrre ([89r24]), venerre
([89v21]). It can refer to the worship of either the Christian God or to heathen
deities (Wolf ed. 2003). This must be secondary, however. Verbs ending -ga are
related to the adjective suffix -igr/-ugr, and mean to make the object of the
verb into something to which the adjective applies; for example to blga v. to
bloody (make bleed, cover with blood) something makes it blugr adj. bloody
(blood-covered) (Alexander Jhannesson 1927, 30 [-ga]). Alongside the
meaning to honor; to worship, gfga also means to endow, used often in the
preterite participle such as lgfgum land-endowed in Erfikvi um Magns
berftt by Gsl Illugason (ed. Gade 2009, 420 [strophe 6]) or tri gfgar
endowed with glory in Noregs konungatal (ed. Gade 2009, 776 [strophe 23]). It
seems likely that the lexical development included a stage where it meant only
to endow, subsequently developing to to worship by means of offerings, with
108

the offering eventually becoming a nondistinctive part of the word such that it
was seen as appropriate for use in a Christian context (unlike, for example blta,
which is used only for heathen worship).
6.1.3.3. Proto-Germanic *ga ~ *gajThe feminine n-stems and adjective forming suffix *-Vga- were
derivational suffixes in Proto-Germanic forming to substantives and adjectives.
The n-stem class is widely considered to have formed by an *-n extension to
the inherited dev-type ending, PGmc with *- in the nominative singular and
*-(i)j- in the other cases, from PIE *-ih ~ *-(e)ieh- (Krahe/Meid 1969,
III:102 [93]; Johnsen 2005, 1179 [2.11.4]). The ending then became
productive itself, so that new n-stems could be formed directly from roots
rather than exclusively from /j-stems (Krahe/Meid 1969, III:102 [93]).
Johnsen (2005, 117119 [2.11.4]) argued that *gan was extended from an
earlier /j-stem *ga, itself formed in Proto-Indo-European from the o-grade
of the verbal root *geb-, that is, from the -grade *gob- was formed *gobih. Johnsen takes for granted that *gean- is cognate to Latin habe which is
shown above not to be secure, but disregarding the cognate status would not
greatly impact his argument, which requires that *gan- ultimately originate in
a verbal derivation, but it is not necessary that it come from the root of *geanin particular.
It is useful to briefly summarize the /j-stem class and Johnsens (2005)
findings. The class is formed with an ending going back to an ablauting Indo-

109

European ending, with *-ih in the strong cases, alternating with the ending
*-ieh in the weak cases (the dv-class; cf. Meier-Brgger 2003 2857
[W204]). These endings became PGmc. *- (in the nominative singular only)
and *-j (in all other cases) respectively, and because of Sievers Law the latter
becomes *-ij after roots with heavy syllables, for example *band ~ *bandijo- f.
fetter, cf. Goth. bandi, acc. bandja (Johnsen 2005, 3358 [2.2]). Meanwhile,
the PIE ieh-stems with a non-ablauting ending (which Johnsen calls the
vidy-type) also have a Germanic ending *-ij- after a heavy syllable, and
these forms with *-ij- merge with the dv-type in Germanic with an ending
*- in the nominative (Johnsen 2005, 37 [1.1], 1030 [1.11], 33 [2.1],
12233). Because of this, it was long regarded that the ending *- was
exclusive to roots with heavy syllables, but Johnsen has made a convincing
argument that this was not truly the case using demonstrations too numberous
to reiterate here, but including for example *maw f. girl (cf. Goth. mawi f.
girl, OI mr f. maiden) and *iw (Goth. iwi f. (female) servant, OI r f.
id.; Johnsen 2005, 11024 [2.11]).
In the discussion of the Matronae Gabiae in 4.2.2.1 it was explained that
both Gutenbrunner (1936, 434, 90) and Neumann ([1998] 2008, 3656
[Garmangabis]) considered the inscriptions featuring Romanized Gabi- to show
evidence of an underlying Germanic /j-stem, *ga(z), and while Johnsens
(2005, 1179 [2.11.4]) reconstruction is hardly the agentive giver that they
proposed, it is is still valuable in that the same word was proposed by entirely
different means. Sturtevant (1952, 1667) proposed that an earlier form
110

equivalent to a hypothetical OI *Gef, gen. *Gefjar, preceded the name Gefjun

and he may not have been far off, but rather than a j-stem *gaj there is
greater evidence for an /j-stem *ga ~ *gaj-. It is likely, but not certain,
that this word derives from a PIE *ghb(h)-(e)i- meaning something in the range
of to have. It is probably related to OI gfa f. good luck and MHG gbe f.
(spiritual) gift, talent, coming from different ablaut grades of the same stem.
Furthermore, even if the word does not derive from the same PIE origin as
Latin habe, Johnsen (2005, 1179 [2.11.4]) made it perfectly clear that the
word is old enough that its meaning, the meaning of *gean-, or both, could
have changed drastically between when the word was first formed and the
earliest Gabiae inscription.
Though a j-stem *gaj meaning good luck, proposed by Kylstra et al.
(1991, 401 [kapiot]) as the origin of Finn. kapiot pl. dowry, treussaeu has not
been supported here, *ga seems the most likely source of the Finnish word. It
is only used in the plural so it is unsurprising that it does not show signs of a
nominative singular *ga; the Finnish declension cannot show a distinction
between *ga and *gaj. Therefore there may also have been contexts
wherein *ga related specifically to marriage, although it is also possible that
this developed specifically within Finnish after the loan had already taken place.

6.2. The ending -un on

The information analyzed and presented thus far is finally enough to
attempt a reconstruction of Gefjun. The case for a deverbal from an unattested
-stem verb *gajn- is not especially strong. It rests on the dubious word
111

written gefianda the Codex Regius which is better explained as a mistake than
an agentive noun derived from an unattested verb (6.1.2). A deverbal
derivation parallel to skipan/skipun from skipa leaves not only the verb itself to
be explained, but also the disproportionate prevelence of the suffix vowel
written o in manuscripts, as more examples with a suffix vowel a would be
expected; indeed that would be expected to dominate.
In 5.2 it was concluded that a suffix descending from the thematicized
Hoffmann suffix attached to a stem *gaj- best accounted for the prevelence of
o as the suffix vowel in Gefjun. It was proposed that a spelling gefion
revealed an underlying Gefjn with the suffix vowel not usually appearing in
unstressed syllables, perhaps due to secondary stress, and not differentiated
from o because it occurs in the sequence j, which is most often written io
(Hreinn Benediktsson 2002 [1963], 160). Syrett (1994, 2034 [7.6.5]) argued
for secondary stress as a factor in the lack of umlaut caused by the suffix *-na-.
It should also be noted that the secondary stress is the most speculative part of
the theory currently being presented on the formation of the name Gefjun, and
the rest of the theory does not entirely rely on it; without the secondary stress
the suffix vowel o is no less confusing than with the deverbal derivation, and
still has the benefit of much stronger evidence for the word from which it
probably derives.
Support for a medial vowel *a following /j-stems compounds or
derivations is difficult to demonstrate explicitly, but seems likely. There are no
examples in the early runic corpus, and there are very few /j-stems with short
112

root syllables at all, but the word mr, acc. mey f. girl, maiden (Gothic mawi,
acc. mauja f. girl), from PGmc. *maw ~ *mauj- (Johnsen 2005, 1134
[2.11.2]) is well-attested with many compounds, such as meybarn n. girl and
meydmr m. virginity, maidenhood, which do not appear to be composed any
differently from compounds like kynborinn adj. high-born from the ja-stem
kyn n. kindred; kind, sort. The umlaut in words like meybarn seem to suggest
that the *j was still present in the original compound, pointing to a PN
*mauja-, and although analogy from the paradigm of mr is not impossible, the
most likely conclusion is that PN /j-stems did compound with *-ja-.
The formation of the word which would become Gefjun cannot be dated
precisely, but the proposed development of the name including a very
hypothetical Proto-Germanic form is as follows:
1. *ga ~ *gaj- + *-n Suffix, forming *gajan
ruling over/pertaining to *ga
2. *gajan > *gajanu (word final > u)
3. *gajanu > *g janu (i-umlaut)
4. *g janu > *[gjnu] > *g jn
(u-umlaut and phonemicization by syncopy of word-final u)
5. *g jn > (*)Gefjn (e, > e; > f; exactly where in this chronology either
of these changes happened is not certain, nor consequential)
6. (*)Gefjn Gefjun (analogical removal of unstressed ,
either structurally-conditioned or on the basis of *-ni- deverbals)

113

7. Conclusion
Throughout the present work, attention has been brought to the fact that
the name Gefjun has not been adequately explained by earlier scholarship, and
it has been examined for data to try to create a more evidence-based estimation
of the names significance. The generally accepted definition the giving one
can be safely rejected, and a new etymology has been proposed. In spite of
similarity to a class of words likewise ending -un, deverbals from -verbs, the
representation of the word Gefjun in medieval manuscripts does not accord with
what would be expected from one, and there is only very weak evidence that a
verb *gefja from which it could have come ever existed, and so a deverbal
derivation is rejected.
The thematicized Hoffmann suffix, the PGmc. reflex of which was
probably *-n for feminine names, accords better with the evidence. The base
word from which it derives does not have an Old Icelandic reflex, but seems to
be attested in the form of devotional inscription in Roman-occupied Germania.
The uniqueness of the name Gefjun, comparable only to Nrun and possibly
*Ingun, is explainable as a result of remodeling of these formations, replacing
earlier endings with *-n; the only difference with Gefjun is that it was not
remodeled, and like jann m. sovereign and Herjann m. a name of inn it
continues its original suffix vowel.

114

It is not altogether clear exactly what the name should be considered to

mean any more specifically than pertaining to, or ruling over, *ga, nor its
relationship to the goddess name Gefn.

7.1. Gefjun and Gefn

It is still difficult to determine the relationship between the words Gefjun
and Gefn. If Syrett (1994, 2034 [7.6.5]) is right that a remodeling occurred
that changed, for example, *wanaz to *wnaz, Gefjun and Gefn could
actually derive respectively from the pre- and post-remodeled form of the same
name: *gajan and *gan. It is true that feminines with a second syllable *-do not usually show i-umlaut in OI, such as vrn f. defense from *warni-, but
i-umlaut is shown in East Norse forms, cf. Old Swedish and Old Danish vrn f.
defense (see 3.1.2). It is worth considering whether Gefn from *gan, in
this scenerio the form which was remodelled, was the East Norse term used by
pre-Christian people in Denmark where Gefjuns primary myth places her and
where there are possible theophoric place-names (see 2.3.1), and whether the
form Gefjun is an archaism that was saved from obscurity by an early Norwegian
poet, Bragi Boddason.
While this is possible, other explanations remain. It cannot be ruled out
entirely that the names are completely different, as Gefn actually could be
related to the verb gefa v. to give. As shown in 3.1.1, feminine *-nideverbals from strong verbs usually place the suffix directly on the root, and
umlaut does usually occur in certain cases, with umlauted or non-umlauted root
115

vowels subsequently being generalized throughout the paradigm, often

resulting in paradigmatic split. Gefjun is probably not a deverbal, but Gefn still
could be. Still, the possibility of an etymological relationship between Gefn and
the single Latin inscription to Gabinis (dat.; AE 1981, 678) is strong.
Much of the discussion of the names Gefjun and Gefn has revolved
around whether or not they are the same, with authors occasionally declaring
Gefjun to be identical with, or merely a regional variant of, the goddess Freyja
(Olrik 1910, Nsstrm 1995, 1001). One objection that should be raised is
that even if the two names share a common origin, declaring them to be the
same goddess for that reason is a projection of the authors linguistic bias
onto an entire society over hundreds of years, a society which was not
homogenous in its belief (Brink 2007). A more productive question might be
Did ragi oddason consider Gefjun and Freyja to be the same? The answer
might still be yes, but hopefully for better reasons than a common origin of
two words.
Beck (2009, 66 f.333) dates the inscription dedicated to Gabinis (dat.) to
the late 2nd or early 3rd century. If the name is cognate to Gefn, it means that
the name is of significant age for the common origin of Gefn and Gefjun to have
time to be obscured to pre-Christian Scandinavians before their earliest
attestations in Old Icelandic. If, more likely, Gabinis is not in continuity with
Gefn (see 4.2.2.1), it is a demonstration that a word beginning *gabj- related
to meaningful social or religious concepts had currency as a name and in
formation of derived names, and supports the idea of multiple independent
116

instances of it. To establish that Gefjun and Gefn are the same goddess (or
different, for that matter) requires more evidence than linguistics, and even if it
did not the linguistic evidence is not adequate at this time to provide an answer.

7.2. The Meaning of Gefjun

In section 6.2 it is argued that Gefjun is composed of a suffix *-n-, the
Germanic successor to the thematized Hoffmann suffix *-hnh-o- which
since Meid (1957) has been called the Herrschersuffix in this context,
attached to a Germanic word *ga ~ *gaj-. What exactly a PGmc *gajan or
PN *gaju would be perceived as meaning to speakers of the language is still
debateable. If the name of the Matronis Gabiabus (dat.) represents a Germanic
-stem with a Latin plural ending, and if the Gabiae really are multiple beings
rather than this plurality being a mode of expression that was introduced to
Germanic peoples along with the idea of devotional epigraphy, then one
possibility is that Gefjun is some kind of leader or archetypical member of the
Gabiae, following the logic by which Meid (1957, 1001; 120) described
opulna as the representitive of the populus m. people, nation, Hludana as the
leader of the Holden (some sort of spirit, perhaps related to Scandinavian
huldrer or the seikona Hul in Ynglinga saga chapter 13, (ed. Bjarni
Aalbjarnarson 1941, 29), and jann m. sovereign as a representitive
member of the j f. nation (rather the primary meaning being rulership of
the j which he regards as secondary; Meid 1957, 77).
Alternatively, Meid (1957, 869) also also draws attention to pairs of
names like r : inn and Ullr : Ullinn, and suggests a development within the
117

spiritual system wherein at an early stage an object or concept was worshiped as

the thing in itself, but later the suffix was used to distinguish the god with
power or representation of the concept. Though he discusses this in particular
with regard to formations featuring a PIE suffix *-tu- (Germanic *wuz,
*wuluz; unattested Latin *fortus, *neptus which are suggested by Fort na,
Nept nus), the concept is perhaps still applicable. A concept of *ga may have
existed, with a goddess or goddesses relating to it developing at different times
in different permutations.
Some tentative suggestions seem advisable. Gefjun is seen in the
mythology claiming land, and perhaps even considered to be nearby when the
young lady swears by her in Vlsa ttr. Perhaps she plays a role in the
regulation of the luck of the household, or is representitive of some kind of
household spirit, of even mrnir whatever they are. In Lokasenna inn
warns that it is inadvisable to fall out of her favor. In this might be seen the
common thread uniting Gothic words gabei f. riches and gabeigs adj. rich
with Old Icelandic terms referring to nobility and worship, gfugr adj. noble
and gfga v. to worship, honor. If pre-Christian pagans believed that
maintaining the good graces of certain beings, perhaps including Gefjun, would
cause them to be successful in their endeavors, the Icelandic and Gothic terms
might be described respectively as the cause (maintenence of good relations
with unseen beings) and effect (material success) within the same religious
system. This is extremely tentative, and as was stated early in this paper, the

118

study of Gefjun as a mythological figure is a topic worthy of persuit in its own

right, and not the goal of the current volume.
Too much of the study of Gefjun in mythology has been burdened by a
false etymology linking her to the ocean and to giving, and attempting to
explain her mythological characteristics as relating to the fertility of the land. In
fact, the issue is much more complicated. Though relatable to abundance and
material well-being it does not seem that Gefjun can be shown being the actual
producer herself; when she plows Sjlland she behaves like a person, and the
episode may have more to do with land ownership than farming. Hopefully the
etymology proposed here will help to break down obstacles to a better
understanding of pre-Christian Nordic religion. There is much more to be said
about Gefjun, and hopefully this paper is one contribution to a more
comprehensive reanalysis.

119

Works Cited
Adami Gesta Hammaburgensis ecclesiae pontificum ex recensione Lappenbergii.
1876. 2nd ed. Hannoverae: Impensis Bibliopolii Hahniani.
AE = Lanne pigraphique. 1888. Paris.
Alexander Jhannesson. 1927. Die Suffixe im Islndischen. Halle (Saale): M.
Niemeyer.
Antonsen, Elmer. 2002. Runes and Germanic Linguistics. Berlin, New York: de
Gruyter.
sgeir Blndal Magnsson. 1989. slensk orsifjabk. Reykjavk: Orabk
Hskolans.
ammesberger, Alfred. 1999. Wednesday. In Words, lexemes, concepts approaches to the lexicon: Studies in honour of Leonhard Lipka. Edited by
Wolfgang Falkner and Hans-Jrg Schmid. 16. Tbigen: Gunter Narr.
attaglia, Frank. 1991. The Germanic Earth Goddess in
Quarterly, 31 (4): 41546.

eowulf? Mankind

eck, Noemie. 2009. Goddesses in Celtic Religion. PhD diss., Universit

Lumire Lyon 2.
Bek-Pedersen, Karen. 2011. The Norns in Norse Mythology. Edinburgh:
Dunedin.
Bjarni Aalbjarnarson, ed. 1941. Heimskringla: I. slensk fornrit 26. Reykjavk:
Hi slenzka fornritaflag.
Bjorvand, Harald and Fredrik Otto Lindeman. 2007. Vre arveord: Etymologisk
ordbok. Oslo: Instituttet for sammenlignende kulturforskning.

120

Bosworth, Joseph. and T. Northcote Toller. 1898-1921. An Anglo-Saxon

Dictionary. Oxford: Oxford University Press.
Boutkan, Dirk. 1995. The Germanic Auslautgesetze. Amsterdam: Rodopi.
rink, Stefan. 2007. How Uniform was Old Norse Religion. In Learning and
Understanding in the Old Norse World. 10535. Turnhout: Brepols.
Buck, Carl Darling. 1904. A Grammar of Oscan and Umbrian. Boston: Ginn &
Company.
CIL XIII = Corpus Inscriptionum Latinarum Vol. XIII: Inscriptiones trium
Galliarum et Germaniarium latinae. 19041907. 1:22:2: Inscriptiones
Belgicae, Inscriptiones Germaniae superioris, Inscriptiones Germaniae inferioris.
Miliaria Galliarum et Germaniarum. Edited by Domaszewski, Alfred von,
Mommsen, Theodor, Hirschfeld, Otto, Zangemeister, Karl Friedrich
Wilhelm. Berlin: Apvd Georgivm Reimervm.
Cleasby, Richard and Gubrandur Vigfsson. 1874. An Icelandic-English
Dictionary. Oxford: The Clarendon Press.
Clunies Ross, Margaret. 1978. The myth of Gefjon and Gylfi and its function
in Snorra Edda and Heimskringla. Archiv fr Nordisk Filologi. 93:14965.
. 19941998. Prolonged Echoes: Old Norse Myths in Medieval Northern
Society. 2 vols. The Viking collection: Studies in Northern civilization 7 &
10. Odense: Odense University Press.
Cowgill, Warren. 1959. The Inflection of Germanic -Presents. Language 35
(1): 115.
Davidson, H. R. Ellis. 1968. Gods and Myths of Northern Europe. London:
Penguin.
. Gefjon, Goddess of the Northern Seas. In Islanders and WaterDwellers. Edited by Patricia Lysaght, Samas Cathin & Dith hgin.
519. Dublin: University College Dublin.
121

Derksen, Rick. 2007. Etymological Dictionary of the Slavic Inherited Lexicon.

Leiden Indo-European Etymological Dictionary Series. Leiden: Brill.
Dronke, Ursula, ed. and trans. 1997. The Poetic Edda II: Mythological Poems.
Oxford: Clarendon Press.
Faulkes, Anthony, ed. 1998. Skldskaparml: 1. London: The Viking Society
for Northern Research, University College London.
. 2005. Prologue and Gylfaginning. London: The Viking Society for
Northern Research.
Finch, R. G., ed. 1965. Vlsunga saga. London: Nelson.
Finnur Jnsson, ed. 1892. Hauksbk. Copenhagen: Thieles.
, ed. 1896. Vlsngsrmur. In Fernir Fornslenskir Rmnaflokkar.
Copenhagen: S.L. Mllers. 4359.
. 1905. i : io i norsk-islansk. Arkiv fr nordisk filologi 21: 24453.
, ed. 1927. AM 623, 4: Helgensagaer. Samfund til udgivelse af gammel
nordisk litteratur 52. Copenhagen: S.L. Mllers.
, ed. 1930. Flateyjarbk (Codex flateyensis) MS. no. 1005 fol. in the Old
Royal Collection in the Royal Library of Copenhagen. Copenhagen: Levin &
Munksgaard.
Friedrich, Paul. 1970. Proto-Indo-European Trees: The Arboreal System of a
Prehistoric People. Chicago: The University of Chicago Press.
Gade, Kari Ellen, ed. 2009. Poetry from the Kings sagas 2. Skaldic Poetry of the
Scandinavian Middle Ages II. 2 vols. Turnhout: Brepols.
Glare, P.G.W. 1968. Oxford Latin Dictionary. Oxford: Clarendon Press.
Grimm, Jacob. (1835) 1880. Teutonic Mythology: 1. Translated by James Steven
Stallybrass. London: George Bell & Sons.
122

Gubrandur Vigfsson and C. R. Unger, eds. 18601868. Flateyjarbk: en

samling af norske konge-sagaer. 3 vols. Christiania: P. T. Mallings.
Gunnell, Terry. 2007. The Goddess(es) In The Marshes. Paper presented at
Old Norse Myth: International Symposium. University of Aarhus, 22-23
November.
Gutenbrunner, Siegfried. 1936. Die germanischen Gtternamen der antiken
Inschriften. Rheinische beitrge und hlfsbcher zur germanischen
philologie und volkskunde 24. Halle (Saale): M. Niemeyer.
Heaney, Seamus, ed. 2000. Beowulf. New York: W. W. Norton & Company.
Heimir Plsson, ed. 2012. The Uppsala Edda: DG 11 4to. London: The Viking
Society for Northern Research, University College London.
Hellbom, Algot. 1980.
Universitetsbibliotek.

Ordbok

ver

Njurundamlet.

Ume:

Ume

Hellquist, Elof. 1922. Svensk Etymologisk Ordbok. Lund: Gleerup

Hersch, Karen K. 2010. The Roman Wedding: Ritual and Meaning in Antiquity.
Cambridge: Cambridge University Press.
Hopkins, Joseph S. 2012. Goddesses Unknown I: Njrun and the Sister-Wife
of Njrr. RMN Newsletter 5. 3944.
Hogg, Richard M. 2011. A Grammar of Old English I: Phonology. Chichester:
Wiley-Blackwell.
Hreinn enediktsson. (1962) 2002. The Unstressed and the Non-Syllabic
Vowels of Old Icelandic. In Linguistic Studies, Historical and Comparative.
Edited by Gurn rhallsdttir, Hskuldur rinsson, Jn G. Frijnsson,
and Kjartan Ottoson. 7491. Reykjavk: Institute of Linguistics. Originally
published in Arkiv fr nordisk filologi 77:731.

123

. (1963) 2002. Some Aspects of Nordic Umlaut and

Linguistic Studies, Historical and Comparative. Edited

reaking. In
by Gurn

rhallsdttir, Hskuldur rinsson, Jn G. Frijnsson, and Kjartan

Ottoson. 14263. Reykjavk: Institute of Linguistics. Originally published
in Language 39:40931.
. (1986) 2002. OIcel. oxe, uxe: Morphology and Phonology. In
Linguistic Studies, Historical and Comparative.

Edited

by

Gurn

rhallsdttir, Hskuldur rinsson, Jn G. Frijnsson, and Kjartan

Ottoson. 32353. Reykjavk: Institute of Linguistics. 32353. Originally
published in NOWELE 7: 2997.
Jakobsen, Jakob, ed.. 19021903. Droplaugarsona saga. In Austfiringa sgur.
13980. Copenhagen: S. L. Mllers Bogtrykkeri.
Johnsen, Sverre Stausland. 2005. The Germanic (i)j-stem declension: Origin
and development. Mphil diss., University of Oslo.
. 2012. Syncope of long * in Old Norse nouns. In Germansk filologi
og norske ord. Edited by John Ole Askedal, Tom Schmidt, and Rolf Theil,
3351. Oslo: Novus forlag.
Jn Helgason, ed. 1952. Eddadigte 2: Gudedigte. Nordisk filologi 7.
Copenhagen: Munksgaard.
Jn Sigursson, ed. 1843. slendingabk in slendinga sgur vol. 1.
Copenhagen: Kongelige nordiske oldskrift-selskab.
Kern, H. 1870. Noms germaniques dans des inscriptions latines du Rhin
infrieur. Revue Celtique 2. 15377.
Kock, Axel. 1898. Der a-Umlaut und der Wechsel der Endvocale a : i(e) in
den altnordischen Sprachen. Beitrge zur Geschichte der deutschen Sprache
und Literatur 23. 484554.

124

Krahe/Meid = Krahe, Hans and Wolfgang Meid. 1969. Germanische

Sprachwissenschaft. vols. 2: Formenlehre and 3: Wortbildungslehre. Berlin: de
Gruyter.
Kristjn rnason. 2011. The Phonology of Icelandic and Faroese. Oxford: Oxford
University Press.
Kroonen, Guus. 2013. Etymological Dictionary of Proto-Germanic. Leiden IndoEuropean Etymological Dictionary Series. Leiden: Brill.
Kulonen, Ulla-Maija, Erkki Itkonen, Aulis J. Joki, Reino Peltola, Marita
Cronstedt, Eino Koponen, Meri Puromies, Klaas Ph. Ruppel, and Satu
Tanner. 1992. Suomen Sanojen Alkuper Etymologinen Sanakirja: 1. Helsinki:
Suomalaisen Kirjallisuuden Seura Kotimaisten Kielten Tutkimuskeskus.
Kunin, Devra, trans. 2001. A History of Norway and the Passion and Miracles of
the Blessed lfr. Edited by Carl Phelpstead. London: The Viking Society
for Northern Research, University College London.
Kylstra, A. D., Sirkka-Liisa Hahmo, Tette Hofstra, and Osmo Nikkil. 1991.
Lexikon der lteren germanischen Lehnwrter in den ostseefinnischen Sprachen:
II. Amsterdam: Rodopi.
Lalis, Antanas. 1915. A Dictionary of the Lithuanian and English Languages.
Chicago: Lietuva.
Larsson, Ludvig, ed. 1885. Islndska handskriften N 645 4 i den
Arnamagnanska

samlingen

universitetsbiblioteket

Kbenhavn

diplomatariskt aftryck. Lund: Gleerup.

Lassen, Annette. 2011. Odin p kristent pergament: En teksthistorisk studie.
Copenhagen: Museum Tusculanum
Lehman, Winifred P. 1986. A Gothic Etymological Dictionary. Leiden: Brill.
Levander, Lars and Stig

jrklund. 1961. Ordbok ver folkm len i vre

Dalarna. Stockholm : Almqvist & Wiksell

125

Lexicon Poeticum = Sveinbjrn Egilsson and Finnur Jnsson. 1931. Lexicon

poeticum antiqu lingu septentrionalis/Ordbog over det norsk-islandske
skjaldesprog. Copenhagen: S. L. Mllers.
Liberman, Anatoly. (1992) 1994. Snorri and Saxo on tgaraloki, with Notes
on Loki Laufeyjarson's Character, Career, and Name. In Word Heath.
Wortheide. Orheii: Essays on Germanic Literature and Usage 19721992.
176234. Rome: Editrice il Calamo.
. 2012. The Theonym Wodan ~ Wuotan ~ inn. In .
. . II. Edited by .. . Moscow:
.
Lindow, John. 2001. Handbook of Norse Mythology. Santa Barbara: ABC-CLIO.
Lindstrm, Johan Adolf. 1859. Om den Keltisk-Germaniska kulturens inverkan p
Finska folket. Tavastehus.
LIV = Rix, Helmut, Martin Kmmel, Thomas Zehnder, Reiner Lipp, and
Brigitte Schirmer. 2001. Lexikon der Indogermanischen Verben. 2nd ed.
Wiesbaden: Reichert.
Loth, Agnete, ed. 19691970. Reykjahlabk: Islandske helgenlegender. 2
vols. Copenhagen: Munksgaard.
Maher, J. Peter. (1973) 1977. *Haemon: (Stone) Axe and Sky in IndoEuropean/Battle-axe

culture.

Creation and Tradition in Language.

Amsterdam: John Benjamins. 85106. Originally published in The Journal

of Indo-European Studies 1 (4). 44158.
Mailhammer, Robert. 2007. The Germanic Strong Verbs: Foundations and
Development of a New System. Berlin: De Gruyter.
Matasovi, Ranko. 2008. Etymological Dictionary of Proto-Celtic. Leiden IndoEuropean Etymological Dictionary Series. Leiden: Brill.
126

Meier-Brgger, Michael. 2003. Indo-European Linguistics. Translated by

Charles Gertmenian. Berlin, New York: De Gruyter.
Much, Rudolf. 1891. Germanische Matronennamen. Zeitschrift fr deutsches
Altertum und deutsche Literatur 35. 31524.
Mundal, Else. 1990. The Position of the Individual Gods and Goddesses in
Various Types of Sources with Special Reference to the Female
Divinities. In Old Norse and Finnish Religions and Cultic Place-Names.
Edited by Tore Ahlbck. 294315. bo: The Donner Institute for Research
in Religious and Cultural History.
Murphy, Luke John. 2013. Herjans dsir: Valkyrjur, Supernatural Femininities,
and Elite Warrior Culture in the Late Pre-Christian Iron Age. MA thesis,
Hskli slands. http://skemman.is.
Mrur rnason, rni Bvarsson, and Laufey Leifsdttir. 2007. slensk
Orabk. Reykjavk: Edda.
Naktinien, Gertrda, Jonas Paulauskas, Ritut Petrokien, Vytautas Vitkauskas,
Jolanta Zabarskait. 2005. Lietuvi Kalbos odynas. Electronic edition.
Prepared for web by Evaldas Oeraitis and Vytautas Zinkeviius. Vilnius:
Institute of the Lithuanian Language. http://www.lkz.lt/.
Nsstrm, Brit-Mari. 1995. Freyja the Great Goddess of the North. Lund
Sudies in History of Religions 5. Lund: University of Lund.
Neckel, Gustav and Hans Kuhn. 1983. Edda: die Lieder des Codex Regius nebst
verwandten Denkmlern. Heidelberg: Carl Winter. Electronic edition.
Prepared for web by David Stifter and Sigurdur H. Palsson. 1994. TITUS.
http://titus.uni-frankfurt.de/indexe.htm.
Neumann, Gnter. (1983) 2008. Die Sprachverhltnisse in den germanischen
Provinzen des Rmischen Reiches. In Namensudien zum Altgermanischen.
Edited by Heinrich Beck, Dieter Geuenich, and Heiko Steuer. 2652.
Berlin/New York: Water de Gruyter. Originally published 1983 in Aufstieg
127

und Niedergang der rmischen Welt. Geschichte und Kultur Roms im Spiegel
der neueren Forschung. II. 106188. Berlin/New York.
. (1987) 2008. Die germanischen Matronenbeinamen. In Namensudien
zum Altgermanischen. Edited by Heinrich Beck, Dieter Geuenich, and Heiko
Steuer. 253289. Berlin/New York: Water de Gruyter. Originally published
in Matronen und verwandte Gottheiten. Beihefte der Bonner Jahrbcher 44:
10332. Kln: Rheinland Verlag GmbH.
. (1998) 2008. Garmangabis. In Namensudien zum Altgermanischen.
Edited by Heinrich Beck, Dieter Geuenich, and Heiko Steuer. 3656.
Berlin/New York: Water de Gruyter. Originally published in Reallexikon der
Germanischen Altertumskunde. 10. Berlin/New York: Walter de Gruyter.
. (2001) 2008. Matronen. In Namensudien zum Altgermanischen.
Edited by Heinrich Beck, Dieter Geuenich, and Heiko Steuer. 4025.
Berlin/New York: Water de Gruyter. Originally published in Reallexikon der
Germanischen Altertumskunde. 19. Berlin/New York: Walter de Gruyter.
. (2003) 2008. Germanische Gttinnen in lateinischen Texten. In
Namensudien zum Altgermanischen. Edited by Heinrich Beck, Dieter
Geuenich, and Heiko Steuer. 226237. Berlin/New York: Water de
Gruyter. Originally published in Namenkundliche Informationen 83/84: 41
54.
Nordlander, Johan. 1933. Ordbok ver Multrmlet. Stockholm: Fritze.
Noreen, Adolf. 1923. Altnordische Grammatik I: Altislndische und
altnorwegische Grammatik: (Laut- und Flexionslehre) unter Bercksichtigung
des Urnordischen. Halle (Saale): Verlag von Max Niemeyer. Electronic
edition. Prepared for web by Andrea de Leeuw van Weenen. 2010.
http://www.arnastofnun.is/solofile/1016380.
North, Richard. 1998. Heathen Gods in Old English Literature. Cambridge:
Cambridge University Press.

128

Nystrm, Jan-Olov. 1993. Ordbok ver lulemlet. Ume : Dialekt-, ortnamnsoch folkminnesarkivet i Ume.
Olrik, Axel. 1910. Gefion. In Danske Studier. Edited by Marius Kristensen
and Axel Olrik. 1-31. Copenhagen: Gyldendalske Boghandel/Nordisk
Forlag.
Olsen, irgit Anette. 2004. The Complex of Nasal Stems in Indo-European.
In James Clarkson and Birgit Anette Olsen, eds. Indo-European Word
Formation: Proceedings of the Conference held at the University of Copenhagen
October 20th 22nd 2000. 215248. Copenhagen: Museum Tusculanum
Press.
Olsen, Karin. 2001. ragis Ragnarsdrpa, a Monstrous Poem. In Monsters
and the Monstrous in Medieval Northwest Europe. Edited by K.E. Olsen and
L.A.J.R. Houwen. 12339. Leuven: Peeters.
Olsen, Magnus. 1949. "Sproget i de grnlandske runeinnskrifter." In Fra norrn
filologi. Oslo: Aschehoug. 51-71.
ONP = Ordbog over det norrne prosasprog. 1989. Copenhagen: Den
Arnamagnanske Kommission.
Pinault, Georges-Jean. 2000. Vdique dm nas-, Latin dominus et lorigine du
suffixe de Hoffman. Bulletin de la Socit de linguistique de Paris 95. 61
117.
Pokorny, Julius. 1959. Indogermanisches Etymologisches Wrterbuch. vol. 2.
Bern and Mnchen: Franke Verlag.
Price, Neil. 2002. The Viking Way: Religion and War in Late Iron Age
Scandinavia. Uppsala: Uppsala University.
RIB = Collingwood, R.G. and R.P. Wright. 1965. The Roman Inscriptions of
Britain. Volume I: The Inscriptions on Stone. Oxford: Clarendon Press.

129

Rietz, Johan Ernst. 18621867. Svenskt dialektlexikon: ordbok fver svenska

allmogesprket. Lund : N.P. Lundbergs
Ringe, Don. 2006. From Proto-Indo-European to Proto-Germanic. A Linguistic
History of English 1. Oxford: Oxford University Press.
Schiller, Karl and August Lbben. 1876. Mittelniederdeutsches Wrterbuch.
Bremen: Verlag von J. Khtmann's Buchhandlung.
Schrijver, Peter. 1991. The Reflexes of the Proto-Indo-European Laryngeals in
Latin. Leiden Studies in Indo-European 2. Amsterdam/Atlanta: Rodopi.
Schck, Henrik. 1940. Ingunar Freyr. Fornvnnen 35. 289296.
Schtzeichel, Rudolf. 2012. Althochdeutsches Wrterbuch. Berlin, Boston: De
Gruyter.
Seebold, Elmar. 1970. Vergleichendes und etymologisches Wrterbuch der
germanischen starken Verben. The Hague: Mouton & Co.
Shaw, Philip A. Pagan Goddesses in the Early Germanic World: Eostre, Hreda and
the Cult of Matrons. Sudes in Early Medieval History. London: Bristol
Classical Press.
Sigurur Nordal and Guni Jnsson, eds. 1938. Gunnlaugs saga Ormstungu in
Borgfiringa sgur. slenzk fornrit 3. Reykjavk: Hi slenzka fornritaflag.
Simek, Rudolf. 1996. Dictionary of Northern Mythology. Translated by Angela
Hall. Cambridge: D. S. Brewer.
Skj = Finnur Jnsson. 19671973. Den norsk-islandske skjaldedigtning. 4 vols.
Copenhagen: Rosenkilde og Bagger. Originally published 19121915.
von Steinmeyer, Elias, Elisabeth Karg-Gasterstdt, Theodor Frings, Rudolf
Groe, Gotthard Lerchner, and Hans Ulrich Schmid. 1986. Althochdeutsches
Wrterbuch. vol. 4. Berlin: Akademie-Verlag.

130

Streitberg, Wilhelm. 1910. Die gotische Bibel. Zweiter Teil: Gotisch-griechischdeutsches Wrterbuch. Heidelberg: Carl Winter.
Sturtevant, Albert Morey. 1952. Regarding the Old Norse name Gefjon.
Scandinavian Studies. 24 (4). 166-7.
Sveinbjrn Egilsson and Finnur Jnsson. 1931. Lexicon Poeticum: Antiqu
Lingu Septentrionalis. Second Edition. Copenhagen: S. L. Mllers
Bogtrykkeri
Svenska Akademien. 1898. Ordbok ver Svenska Sprket. Lund: AB
Gleerupska Universitetsbokhandeln.
Syrett, Martin. 1994. The Unaccented Vowels of Proto-Norse. Odense: Odense
University Press.
Tacitus, Cornelius. 1914. Germania Edited and Translated by Maurice Hutton.
Loeb Classical Library 35. London: Heinemann/New York: MacMillan.
Tiefenback, Heinrich. 2010. Altschsisches Handwrterbuch/A Concise Old
Saxon Dictionary. Berlin, New York: De Gruyter.
Torp, Alf. 1919. Nynorsk Etymologisk Ordbok. Krisiania: H. Aschehoug & Co.
Unger, C. R., ed. 1862. Stjorn: gammelnorsk Bibelhistorie fra Verdens Skabelse til
det babyloniske Fangenskab. Chrisiania: Feilberg & Landmarks.
, ed. 1874. Postola sgur: Legendariske Fortllinger om Apostlernes Liv,
deres Kamp for Kristendommens Udbredelse, samt deres Martyrdd.
Christiania: Trykt Hos B. M. Bentzen.
, ed. 1877. Heilagra Manna Sgur: Fortllinger og Legender om Hellige
Mnd og Kvinder. 2 vols. Christiania: Trykt Hos B. M. Bentzen.
de Vaan, Michiel. 2008. Etymological Dictionary of Latin and the Other Italic
Languages. Leiden Indo-European Etymological Dictionary Series. Leiden:
Brill.
131

Vendell, Herman Albert. 19041907. Ordbok ver de stsvenska dialekterna.

Helsingfors: Tidnings- & Tryckeri-Altiebolagets Tryckeri.
de Vries, Jan. 1931. Contributions to the study of Othin especially in his relation to
agricultural practices and popular lore. Folklore Fellows Communications 94.
Helsinki: Societas scientiarum fennica.
de Vries, Jan. 1963. Altnordisches Etymologisches Wrterbuch. Leiden: Brill.
van Weenen, Andrea de Leeuw. 2004. Lemmatized Index to the Icelandic Homily
Book. Reykjavk: Hsklatgfan.
Wessn, Elias and Sven Birger Fredrik Jansson, eds. 19431946. Upplands
runinskrifter: II. Sveriges runinskrifter 7. Stockholm: Kungl. Vitterhets
Historie och Antikvitets Akademien.
Wolf, Kirsten, ed. 2003. Heilagra meyja sgur. slensk trarrit 1. Reykjavk:
Bkmenntafristofnun Hskla slands
Wright, Joseph and Elizabeth M. Wright (eds.). 1925. Old English Grammar.
London: Oxford University Press.
rhallur Vilmundarson and Bjarni Vilhjlmsson, eds. 2009. Stjrnu-Odda
draumur. In Harar saga. slenzk fornrit 13. Reykjavk: Hi slenzka
fornritaflag.

132

Appendix A: Attestations of Gefjun

In the following table, the attestations of the name Gefjun that were gathered for
analysis in chapter 5 are presented along with the manuscript of their origin, the
location in the manuscript if possible, otherwise the page and line number of
the edition that was used instead. The text in which it appears is listed in
abbreviated form in the interest of formatting. The abbreviations represent:
Agns

Agnesar saga

Kat

Katarine saga

Bret

Breta sgur

Klem

Klements saga

Dpl

Droplaugarsona saga

Loka

Lokasenna

FSK

Fdesar saga, Spesar ok Kartasar

Nik

Nikolaus saga erkibyskups

Gylf

Gylfaginning

Pls

Pls saga postula

Gylf:Bragi

Bragi Boddason (Gylfaginning)

Skm

Skldskaparml

Hkr

Heimskringla

Stjrn

Stjrn

Hkr:Bragi

Bragi Boddason (Heimskringla)

Troj

Trojumanna saga

Hl

Haustlng

Vlsa

Vlsa ttr

Jn

Jns saga postula

ul

ulur verses (or nonpoetic lists of names)

Manuscript

Location

Dating

Text

Attestation

Case

AM 645 4to (part 1)

ed. Larsson
1885, 67:6

1220

Klem

gefion

acc

AM 645 4to (part 2)

ed. Unger 1874,

223:16

1225-1250

Pls

Gefionar

gen

AM 645 4to (part 2)

ed. Unger 1874,

224:2

1225-1250

Pls

Gefion

nom

GKS 2365 4to

15v20-21

1270

Loka

gef|ion

acc

GKS 2365 4to

15v18

1270

Loka

gefion

nom

1250-1300

Pls

Gefionar

gen

1250-1300

Jn

Gefiunar

gen

AM 655 XVI 4to

AM 652 4to

ed. Unger 1874,

223:23 (var.)
ed. Unger 1874,
451:37

133

AM 544 4to
AM 544 4to
AM 544 4to
AM 544 4to
AM 544 4to

ed. Finnur Jns.

1892, 241:8
ed. Finnur Jns.
1892, 241:2
ed. Finnur Jns.
1892, 241:8
ed. Finnur Jns.
1892, 241:15
ed. Finnur Jns.
1892, 199:17

1302-1310

Bret

Gefionar

gen

1302-1310

Bret

Giefivnar

gen

1302-1310

Bret

Gefion

dat

1302-1310

Bret

Gefivn

acc

1302-1310

Troj

Gefion

dat

AM 748 I b 4to

21r15

1300-1325

ul

gefiun

nom

AM 748 I b 4to

18r20

1300-1325

ul

gfivn

nom

AM 45 fol

2ra7

1300-1325

Hkr

gefion

acc

AM 45 fol

2ra14

1300-1325

Hkr:Bragi

Gefion

nom

AM 656 I 4to

ed. Unger 1874,

451:37

1300-1325

Jn

Gefiunar

gen

1300-1325

Jn

Gefiunar

gen

1325

Jn

gefionar

gen

AM 623 4to

ed. Finnur Jns.

1325
1927, 20:23

Jn

Gefiunar

gen

GKS 2367 4to

1v14

1300-1350

Gylf

gefi

nom

GKS 2367 4to

8v10

1300-1350

Gylf

gefiun

nom

GKS 2367 4to

1v20

1300-1350

Gylf:Bragi

Gefiun

nom

GKS 2367 4to

42v 3

1300-1350

ul

gefion

nom

GKS 2367 4to

18r10

1300-1350

Skm

gefiun

nom

GKS 2367 4to

27r29-30

1300-1350

Skm

gef|iun

nom

GKS 2367 4to

24r12

1300-1350

Skm:Hl

lge

nom

AM 242 fol

22:12

1350

Gylf

gefion

nom

AM 242 fol

8:23

1350

Gylf

gefion

nom

AM 242 fol

8:26

1350

Gylf

gefion

nom

AM 242 fol

8:29

1350

Gylf:Bragi

Gefion

nom

AM 242 fol

40:8

1350

Skm

gefion

nom

AM 242 fol

59:1

1350

Skm

Gefon

nom

AM 242 fol

51:24

1350

Skm:Hl

lgefion

nom

AM 132 fol

147v27

1330-1370

Dpl

gefion

acc

1350-1375

Kat

Gefion

dat

AM 656 I 4to
AM 623 4to

AM 233 a fol

ed. Unger 1874,

451:35
ed. Finnur Jns.
1927, 20:21

ed. Unger 1877,

417: 15

134

AM 226 fol
GKS 1005 fol

ed. Unger 1862,

90:18
ed. Finnur Jns.
1930, 122ra:44

1360-1370

Stjrn

Gefion

acc

1387-1395

Vlsa

gefion

acc

AM 235 fol

38r

1400

FSK

gefion

acc

AM 748 II 4to

11v14

1390-1410

ul

gefion

nom

AM 757 a 4to

8v 6

1390-1410

ul

gefon

nom

AM 235 fol

37v

1400

FSK

gefion

acc

AM 235 fol

37r

1400

FSK

gofion

acc

1425-1445

FSK

Gefion

acc

1425-1445

FSK

Gefion

acc

Holm perg 2 fol

Holm perg 2 fol

ed. Unger 1877,

373:18
ed. Unger 1877,
370:20

Holm perg 2 fol

ed. Unger 1877,

372:1

1425-1445

FSK

Gefion

dat

Holm perg 2 fol

ed. Unger 1877,

16:34

1425-1445

Agns

Gefion

acc

Holm perg 2 fol

ed. Unger 1877,

17:17

1425-1445

Agns

Gefion

acc

Holm perg 2 fol

ed. Unger 1877,

30:11

1425-1445

Nik

Gefionar

gen

Holm perg 2 fol

ed. Unger 1877,

30:28

1425-1445

Nik

Gefion

nom

T: MS No. 1374

2r21

1595

Gylf

Gie

nom

T: MS No. 1374

2r24

1595

Gylf

Ge

nom

T: MS No. 1374

2r27

1595

Gylf:Bragi

Ge

nom

T: MS No. 1374

44r23

1595

ul

gefion

nom

T: MS No. 1374

24v18

1595

Skm:Hl

aulgefion

nom

AM 758 4to

1r16

1609

Gylf

Gefjon

nom

AM 758 4to

10r16

1609

Gylf

Gefjon

nom

AM 758 4to

1r20

1609

Gylf

Gefjun

nom

AM 742 4to

2v11

1611-1650

ul

Gefion

nom

AM 292 4to

55r16

1600-1699

Vlsa

Gefion

acc

AM 751 4to

27v

1611-1700

ul

gefjn

nom

AM 751 4to

5r22

1611-1700

Gylf

Gefion

nom

AM 751 4to

5r26

1611-1700

Gylf

Gefion

nom

AM 751 4to

14r29

1611-1700

Gylf

Gefion

nom

AM 751 4to

5v6

1611-1700

Gylf:Bragi

Gefion

nom

AM 741 4to

38v12

1639-1672

ul

Gefion

nom

135

AM 741 4to

4v17

1639-1672

Gylf

Giefon

nom

AM 741 4to

4v21

1639-1672

Gylf

Giefon

nom

AM 741 4to

14v11

1639-1672

Gylf

Giefon

nom

AM 741 4to

4v24

1639-1672

Gylf:Bragi

Giefon

nom

AM 750 4to

38vc4-5

1650-1699

ul

gi|efin

nom

AM 750 4to

5r1-2

1650-1699

Gylf

AM 750 4to

5r6

1650-1699

Gylf

giefion

nom

AM 750 4to

14v25

1650-1699

Gylf

Gefion

nom

Holm papp 18 fol

2v (margin)

1650-1700

Hkr

Gefian

dat

Holm papp 18 fol

3r3

1650-1700

Hkr

Gefion

acc

Holm papp 18 fol

3r9

1650-1700

Hkr:Bragi

Gefion

nom

AM 35 fol

9v5

1675-1700

Hkr

Gefian

dat

AM 35 fol

9v5

1675-1700

Hkr

Gefion
dat
(correction)

AM 35 fol

10r4

1675-1700

Hkr

Gefion

acc

AM 35 fol

10r19

1675-1700

Hkr:Bragi

Gefion

nom

Thott 1768 4to

236r16

1675-1700

Vlsa

Gefion

acc

AM 164 k fol

20v17

1690-1697

Dpl

Gefion

nom

AM 761 a 4to

50r2

1690-1700

Bragi

Gefion

nom

AM 744 4to

62v7

1700-1725

ul

gefon

nom

AM 761 b 4to

524v4

1700-1725

Vlsa

gefion

acc

LBS 1442 4to

270v24

1720-1740

Dpl

Gefion

nom

AM 38 fol

4v18

1675-1800

Hkr

Gefiun

nom

AM 38 fol

4v29

1675-1800

Hkr:Bragi

Gefion

nom

AM 746 4to

107v

1725-1750

ul

gefiun

nom

Lbs 636 4to

86r12

1750-1760

Loka

Gefion

acc

Lbs 636 4to

86r10

1750-1760

Loka

Gefion

nom

LBS 139 4to

68r8

1760

Dpl

Gefion

nom

NKS 1867 4to

148r10

1760

ul

Giefion

nom

NKS 1867 4to

117v11

1760

Gylf

Gefin

nom

NKS 1867 4to

117v15

1760

Gylf

Gefion

nom

NKS 1867 4to

127v3

1760

Gylf

Gefion

nom

NKS 1867 4to

117v:19

1760

Gylf:Bragi

Gefion

nom

NKS 1867 4to

26v11

1760

Loka

Gefn

acc

NKS 1867 4to

26v9

1760

Loka

Gefion

nom

NKS 1867 4to

141r2

1760

Skm

Gefin

nom

LBS 1321 8vo

100v2

1750-1799

Dpl

Gefion

nom

Lbs 756

43rc24

1777

ul

Geffion

nom

136

Gie on/
gief in

nom

Lbs 756

8r2

1777

Gylf

Gefion

nom

Lbs 756

18r8

1777

Gylf

Geffion

nom

JS 160 fol

163r23

1772-1799

Dpl

Gefion

nom

B 184 4to

51v20

1775-1799

Dpl

Gefion

nom

LBS 1339 4to

161v20

1790

Dpl

Gefion

nom

BR 25 8vo

78r12

1792

ul

Gefjn

nom

BR 25 8vo

24r15-16

1792

Gylf

Gief|jn

nom

BR 25 8vo

24r22

1792

Gylf

Giefjon

nom

BR 25 8vo

37v22

1792

Gylf

Giefon

nom

BR 25 8vo

24v3

1792

Gylf:Bragi

Gefjon

nom

LBS 1001 4to

6260r10

1800

Dpl

Gevion

nom

LBS 327 fol

41r4-5

1800

Dpl

G[..]|on

nom

LBS 2462 4to

47v9

1775-1825

Dpl

Gefion

nom

LBS 3713 4to

107r6

1776-1825

Dpl

Gefion

nom

LBS 1846 4to

134v8

1798-1806

Dpl

Gefion

nom

JS 450 4to

34r12

1807

Dpl

Gievion

nom

JS 630 4to

78v4

1808

Dpl

Gevion

nom

LBS 718 4to

49v29

1810

Dpl

Gefion

nom

BR 8 4to

108v10

1801-1820

Vlsa

Gefion

acc

LBS 1489 4to

132v18

1810-1814

Dpl

Gefion

nom

LBS 356 4to

92r5

1810-1815

Dpl

Gefion

nom

LBS 1634 4to

47v9

1815

Dpl

Gefion

nom

BR 6 4to

31r16

1820

Dpl

Gefion

nom

LBS 997 4to

67v19

1820

Dpl

Gefin

nom

LBS 997 4to

67v19

1820

Dpl

Gefin

nom

AM 932 4to

177v19

1821

Dpl

Gefion

nom

LBS 221 fol

174r34

1819-1832

Dpl

giefion

nom

LBS 2116 4to

19r15

1825-1827

Dpl

Geveon

nom

JS 435 4to

135v21

1805-1850

Dpl

Gefion

nom

B 418 4to

77v6

1825-1830

Dpl

Gefjon

nom

BR 38 8vo

81r2

1828-1831

Gylf

giefion

nom

BR 38 8vo

104v9

1828-1831

Hkr

Giefion

acc

BR 38 8vo

95r13

1828-1831

Skm

giefion

nom

JS 19 fol

31v13

1840

Bret

Ge/fionar

gen

JS 19 fol

31v8

1840

Bret

Giefivnar

gen

JS 19 fol

31v12

1840

Bret

Gefon

dat

JS 19 fol

31v19

1840

Bret

Gefivn

acc

JS 19 fol

11v6

1840

Troj

Gefion

dat

LBS 1573 4to

16v13

1810-1877

Dpl

Gefion

nom

137

LBS 63 4to

111r10

1843-1848

Dpl

Gefion

nom

Lbs 756

7v17

1847

Gylf

e n

nom

LBS 992 8vo

109r3

1870

Dpl

Gefjon

nom

LBS 747 fol

238v26

1871-1875

Dpl

Gefion

nom

Lbs 5157 4to

15v18

1850-1899

Dpl

Gefjun

nom

B 474 4to

69r18

1870-1880

Dpl

Gefion

nom

LBS 3936 4to

248r24-25

1880-1883

Dpl

Gif|in

nom

LBS 1511 4to

111v11

1888

Dpl

Gefjun

nom

138