Four of the six runic finds so far made (October 1984) during the renewed excavations in the Illerup river valley1 have been presented in two articles in Kuml (Ilkjrer and Ljiinstrup 1981; Moltke and Stoklund 1981). The two most recent discoveries have been summarily discussed by Niels Åge Nielsen (1983; 1984). The present article has two purposes.
First, to give my own brief presentation of this new runic material and to review the interpretations of it so far offered; second, to discuss the implications of the finds for our understanding of certain aspects of the early runic alphabet. In the latter case I shall concern myself particularly with the twenty-second rune, sometimes called the "Ing-rune", which has recently been the subject of renewed controversy (Westergaard 1981; Odenstedt forthcoming).
The first inscription of the six, which came to light in 1977, is runologically and linguistically unremarkable. It is found on a bronze mounting for a shield handle and consists of one word: swarta (see tig.
1). Moltke and Stoklund are ofthe opinion that this isa name, carved by the owner of the shield, to indicate who it belonged to (it would appar-ently have been visible to the bearer). Pursuing one ofMoltke's favourite notions, they draw attention to the position of the last rune, below and at right angles to the others, and conclude that the carver understood what he was writing-unlike the average illiterate metalworker who might well stop in the middle of a word or an inscription if he came to the end of a line or there was no more space. They interpret the word, reason-ably enough, as nom. sing. of a weak masc. noun meaning "den svarte"
(we might say 'Blacky' in English), and assume that this refers in some way to the appearance of the shield owner-possibly he had very dark hair or a swarthy complexion. The inscription, in common with the other five from Illerup, is dated on archaeological grounds not later than A.D.
c. 200. According to the two principal excavators, Jjlirgen Ilkjrer and Jjlirn
This article is a revised version of a paper first given to a meeting of Kungl. Gustav Adolfs Akademien on 11th April 1984.
1 Fora preliminary report on these excavations see Ilkjrer and Lpnstrup (1983).
-,.
Fig. 1. The Illerup bronze shield handle mounting.-Figs. 1--6 are drawings made on the basis of existing illustrations and (in the case of 3-4, 6) sketches of the originals made by Jpm Lpnstrup and/or myself. They are not to scale.
L(l)nstrup, the large number of artefacts discovered at Illerup, including approximately 125 Roman denarii, make fora particularly reliable dat-ing.
The next two discoveries, which are more interesting than the first, were made in 1980 (see fig. 2). Two lance heads were each found to bear the same minute inscription (c. 11-13 mm long, 4.5-7.5 mm high), but while the runes on one were carved in the usual way, those on the other were in relief and had clearly been stamped. A number of uncertainties surround this pair of inscriptions: how are the runes to be read, what message do they carry, and why is one carved and the other stamped?
Moltke and Stoklund take it for granted that a stamped inscription represents the name of a smith or a workshop, since one would only go to the trouble of making a stamp if the same word or words were to be used repeatedly. If this assumption is correct, we have here a different type of inscription from raunijaz on the lance from 0vre Stabu, Norway, which, it is generally agreed, contains the name of the weapon ('tester', or such-like). Inscriptions on other lance heads have also mostly been interpreted as weapon names. Why one of the Illerup heads is stamped and the other carved is a matter about which scholars have not been slow to speculate. According to Moltke and Stoklund, the smith lost his stamp, and was therefore compelled to start carving his name as a guarantee of good workmanship. Nielsen (1983), on the other hand, blames it all on some (envious?) warrior who copied a (famous?) trade-mark.
Attempts so far made to decipher the inscriptions hardly help to solve this problem, but have greater intrinsic interest. Moltke and Stoklund were puzzled by the form f , which resembles the Viking-Age maor-rune, but is clearly something else. They tried first to read it as an
*Ingwaz with rounded side-staves (as opposed to the more usual t), which resulted in the word (i)ngagnijo when read from right to left, and ojingang when read in the opposite direction. The former is quite rightly
Fig. 2. The Illerup lance heads (the upper is stamped, the lower incised).
Fig. 3. The Illerup plane.
described as "en temmelig umulig sprogform", while the latter is merely
"en uforklarlig form", the ending of which nevertheless resembles that of the Slemminge inscription's
~ltR?
(Krause and Jankuhn 1966, 89;see also below). A left-to-right reading also poses the problem of why the sequence /ng/ is written in two different ways: ,}
X
and f (only later do the authors consider the possibility that ,}X
could denote [n]+
[g] and f [ug]-see below). The obvious solution is that the armourer made a mistake-perhaps in trying to copy at
orr
from a model in which the runes had rounded side-staves. The possibilities opened up by this assumption are then explored. The direction of the side-staves on ~suggests a reading from right to left, which gives tagnijo (zagnijo is impossible since /z/ does not occur in word-initial position). This is rejected partly on the grounds that it does not resemble any known Germanic name, and partly because if it were a name the word would be offeminine gender, and that is hardly suitable fora smith. Stressing that the direction of the side-staves on ~ are not crucial in deciding the direction in which the inscription should be read,2 Moltke and Stoklund finally settle for ojingaz, which at least has the advantage of containing the well-known nom. masc. a-stem ending /-az/, although, as they admit, no such name is known from any other source and it is hard to see what the sequence oj- or possibly ojin- might represent. Nevertheless, the authors stress the similarity of ojingaz to the Norwegian Reistad stone's iu(>ingaz-both use the combination
i-X
rather than *Ingwaz-and sug-gest that such spelling indicates a different pronunciation from that denoted by the twenty-second rune. I assume they have in mind [n]+
[y], or possibly [n]
+
[g], which we might expect at a morphemeboundary, as for example in
fl txH1Xi/\
on the Norwegian Nordhuglo stone, wheret /\
is taken as the negative prefix.The last inscription to be dealt with by Moltke and Stoklund-also discovered in 1980--occurs on a plane. The runes are not easy to read, but proceeding from left to right, the two authors discern the word afilaiki (see fig. 3). It is hard to square this with the drawing that accompanies Illrjrer and Lj21nstrup's 1981 article, and the close-up
photo-2 Runes facing in the opposite direction from that of the inscription are relatively common (cf. Diiwel 1968, 10-11), and are known as reversed runes (Swedish vändrunor, Danish venderuner). There is, however, much confusion in the use of this term. It is often employed simply to denote runes which face left, apparently on the (clearly false) assump-tion that the normal direcassump-tion of early runic writing is left to right. Moltke and Stoklund (1981) are not immune from this muddled thinking. On p. 70 they describe ~ on the lance-head inscriptions as a venderune on the assumption both that the inscription runs from left to right and right to left.
graph of the inscription that appears in both is too murky to resolve matters of detail, but Moltke and Stoklund assure us that runes 6 and 8, which they read as
I,
can be distinguished from secondary marks and scratches when examined under a microscope, and conclude that of the various marks between the twoI
symbols, ,.._ gives the most plausible reading. However, afilaiki is not easily recognisable as a word or words.Could the last part be roade to read -laikaz, we should recognise a well-known name element (cf., eg, Hadulaikaz on the Norwegian Kjll)levig stone) and perhaps be able to assume that the inscription gave us the name of the plane's owner, although afi- has no obvious interpretation.
Moltke and Stoklund suggest that an
r
may have been lost when the plane was broken, but a good chunk of wood remains after the finalI
and I could see no trace of any further rune when I examined the plane early in 1984. They then equate the supposed -laikiz with -laikaz, offering as comparative evidence the Swedish Ellestad stone, which has sigimArAz for the expected * sigimariz (and was undoubtedly carved long after the Primitive Norse period). "Vore forfredre har ikke holdt sig så nll)je til navnegrupperne som moderne filologer kunne ll)nske'', they conclude, a statement which begs a number of questions. Formally afilaiki could contain either an i-stem variant of laikaz or some other i-stem word in the acc., dat. or just possibly the voc. sing.3 We cannot, however, simply ignore the absence of corroborative evidence fora word *laikiz, however we interpret it. An isolated acc. seems to lack convincing parallels too, which is doubtless what Moltke and Stoklund mean when they say: "en sådan form giver ingen förnuftig mening", but their statement is directed equally at the thought that afilaiki could be a dat.An elliptical dat. with the sense 'for/to Afilaikiz' does not seem entirely absurd, however, even if convincing early parallels are bard to find.4 A voc. form is unlikely for phonetic as well as semantic reasons; we would not expect Indo-European final -i to survive into the second century A.D. in this position. If it were not so certain-as the archaeologists assure us it is-that the Illerup site 2 finds originated in south-eastern Norway or south-western Sweden (Bohuslän, Västergötland, Halland), one might even be tempted to see in afilaiki a West Germanic nom.
masc. sing. at the transitional stage between an -iz and a zero ending. 5 All in all, in our present state of knowledge I think we can do little more than mention the various possibilities of interpretation and try to avoid
3 On supposed traces of the voc. in North Germanic see most recently Stiles (1984).
4 One interpretation of the Norwegian Stenstad inscription: igijonhalaz is "Fiir I. (setzte) Hall (den Stein)" (cf. Krause and Jankuhn 1966, 186), anda new reading of the last line of the Norwegian Årstad stone as: ekwinai leads to the translation: "I, for my friend" (cf.
Antonsen 1975, 35). See also Krause (1971), 131.
5 With the strong likelihood that the Torsbjerg chape and its inscription originate from an area between the Elbe and the Rhine (cf. Ilkjrer and Lpnstrup 1981), we now have probable West Germanic nom. masc. sing. forms in -az and -iz.
Q. ~I? / f
Fig. 4. The first Illerup silver shield handle mounting.too dogmatic an approach, particularly since it is by no means certain that the inscription has been correctly read.
I have gone into some detail about the word (or words) on the Illerup plane because of its possible implications for our understanding of one of the two remaining inscriptions. These were discovered as late as 1983 and do not therefore figure in Moltke and Stoklund's article. However, as noted above, they have been briefly treated by Nielsen (1983; 1984).
Both are found on shield handle mountings which, unusually, are of silver (most such mountings found at Illerup are of iron and carry no runes). The first to come to light bears a clear inscription which almost certainly consists of two words (see tig. 4). J!l)rn L!l)nstrup has suggested (in a letter to me) that the inscription was carved by two different people, and he compares it in that respect with the runic horn from Gallehus.
Whatever the truth of this unusual view, it seems clear that we find the past tense indic. of the verb *taujan 'to make'6 on both inscriptions, with the difference that here it appears in the 3rd pers. sing. (as also on the Garb!l)lle box from Sjrelland), while Gallehus has the 1st pers. sing.
form. Why-if L!l)nstrup is right-two different carvers should have been at work on either inscription is unclear. The evidence from Galle-hus and Garb!l)lle makes it a plausible assumption that the first sequence of runes on the Illerup mount contains the name of the individual who made it, and the rather uneven quality of the characters might indicate that he incised them himself, but why, then, would he get hold of someone else (a professional to judge by the contrast in style) to carve the word tawide?
Of considerably greater interest than this (for the present at least) insoluble problem is the form of the third rune in tawide. Faced with this rune in an unfamiliar context, most runologists would unhesitatingly declare it to be *Ingwaz, but it will have emerged from what has already been said that here it can hardly be anything other than *wuryö-with pockets on both sides. Any doubts are speedily resolved by attempts to interpret it as *lngwaz; neither tangide, which could mean 'joined to-gether' nor edingat, which as far as I know does not mean anything, can easily be preferred to a word that is known from two other inscriptions and which makes sense in the context. What, then, of the other six runes? Nielsen (1983) suggests either ningijo (right-to-left) or ojingin (left-to-right). In favour of the former he cites the direction of the word tawide and the "vendt" form of -1-(see note 2), in favour of the latter the
6 Fora different interpretation of the meaning of this word see Rooth (1984).
::! ~4?
Fig. 5. The Fr~yhov statue.similarity with ojingaz on the two lances. In fäet there is nothing to indicate in which direction the runes should be read: the form of *nauoiz could vary freely between ,} and
i-
in the older runic alphabet (see, eg, Diiwel 1968, 8, v. Friesen et al. 1933, 40, 42-3), and there are several examples of changes in direction in one and the same inscription (eg, the Norwegian Fonnås brooch). Commenting on ojingin, Nielsen finds it unsatisfactory that both this and the other alleged name, ojingaz, are otherwise unknown and that neither can be explained on the basis of recognisable North Germanic elements. ningijo be leaves to the mercy of bis readers, and attempts instead to broaden the perspective. Noting the probable Swedish or Norwegian provenance of the Illerup finds (see above), be ponders the likelihood that these areas gave birth to variants of the normal runic symbols and sees evidence for such a view in the Fq'lyhov inscription (see fig. 5). This is carved on the small bronze statue of a man (7 .5 cm high), found in Romerike in south-eastern Norway, and dated to the third century. Not everyone has recognised this inscription as runic, but remembering that ton the Illerup shield handle mounting denotes /w/, Nielsen suggests Fr~yhov now be read wanda; C8J is taken as a variant of *dagaz, and said to denote /nd/-a common practice in runic writing. The word that emerges is, we are told, the forerunner of Old Icel. vandi 'difficulty, obligation', and on the basis of the single occur-rence of the compound vaNpablot, explained as "genstand, der ofres til en gud" (cf. Fritzner 1883-1972, IV s.v. vandabl6t), Nielsen concludes that wanda bad "kultisk betydning". Fortified by this little excursus, be returns to the inscription on the two lances and decides that it can be read wagnijo, a possible variant of "wagnaR 'vagn, den k~rende eller farende' ''. He stresses how suitable a word denoting movement would be as the name of a lance head and points out that the majority of early inscriptions of this type appear to contain the name of the lance or the head. Nielsen thus believes, as I do, that Moltke and Stoklund's in-sistence that5l SI ./'X1f
gives us the name of the weaponmaker is ill-considered. I also think bis interpretation of ~ is likely to be correct, but not necessarily for the same reasons as he gives in bis 1983 article. The key is supplied by the latest runic find from Illerup-the second silver shield handle mounting (see fig. 6).If one is not prepared to believe that this inscription says lagungenga, or possibly angengugal-and I am not-it must read laguI>ewa. And it is not really so surprising to find * purisaz with pockets on both sides when
/ i n
'</>I\ 'X 17
Fig. 6. The second IIJerup silver shield handle mounting.we already have examples of *wunjö with exactly the same characteris-tic. That is not to say that lagu(>ewa can be interpreted without difficulty.
For one thing, it is not entirely clear whether the inscription is to be read as one word or two. Strictly speaking, lagu could be 1) nom. or acc. pi.
of the neut. noun *laga, Old Icel. l9g 'law', 2) voc. or acc. sing. of the masc. noun *laguz, Old Icel. l9gr 'water', 3) the stem form of *laguz acting as the fint element of a compound. As for (>ewa, it strongly suggests the word pewaz which we know from two early inscriptions:
Valsfjord (Sjl)r-Trjl)ndelag, A.D. c. 400?) and the Torsbjerg chape (Schleswig, A.D. c. 200, see note 5). It could be the acc. sing. form of this word, or possibly a weak variant-an an-stem-in the nom. An acc.
sing. would be bard to explain since there is no obvious syntactic reason for it (lagu is hardly a verb), but it would make a good companion for afilaiki on the Illerup plane. I cannot help feeling there is some connec-tion between these two forms, both of which one would expect to be nom., but both ofwhich appear to be something else. Ifwe accept this as a working hypothesis, the possibilities become somewhat limited: (>ewa cannot be dat., nor, unless it is an analogical form, is it likely to be voc.;
-laiki can hardly be a North Germanic nom. Taken together, then, the two forms can only be acc. (the object of an unspecified verb?), or nominatives without the usual /-z/ ending. There is, however, no reason to expect such nominatives, ifthese inscriptions emanate from the North Germanic area. If there were any suspicion that they were of West Germanic origin, it would, as indicated above, be tempting to claim them as transitional forms, coming between Torsbjerg's -(>ewaz, for example, and later W est Germanic peow, deo, but present indications are that this is unlikely.
To return to the inscription as a whole, I am inclined to read it as a single, compound word, although I admit that an interpretation 'water-servant' does not carry immediate conviction. Nielsen (1984) suggests
"sjl)kriger" on the grounds that pewaz might be used of a retainer as well as the more humble kind of servant suggested by later Germanic reflexes and related words such as Old Icel.
py
and pja. He also thinks this a fitting designation for a man who sailed to Jutland from N orway or Sweden in quest of booty. That there is no gainsaying, though whether a third-century warrior would enter battle with the equivalent of 'soldier' or 'sailor' scratched on his shield handle mounting is another matter.Apart from stressing that ''runen
p
er rendret efter samme princip som w-runen" (i.e., has two pockets), Nielsen has little more to say aboutApart from stressing that ''runen