The yew rune in the Pforzen inscription

IlL Das Scheidenbiech von Bergakker

The Yew-Rune in the Pforzen Inscription0

Tineke Looijenga I. In this paper, I propose to investigate the occurrence of the so-called yew rune in runic texts of the oldest period, and its significance in the Pforzen inscription. Whereas I initially took the presence of the yew rune in the Pforzen inscription as a deliberate graph, I now assume that it might have been the product of a scribal error. Below I shall explain how I was lead to this assumption.

II. The question of the original sound value of the yew rune SI is most interesting. The problem has been treated by many scholars without reaching a consensus. According to Odenstedt (1990:75) Below I give seven occurrences from the period between circa 400 and circa 700 AD. In these texts I transliterate SI as Y. It appears that, in at least six cases, the yew rune indicates a vowel or perhaps a semi-vowel; in the seventh case, the value is not clear, In at least two instances, the yew rune occurs in an inscription that exhibits the sequence N al. One inscription is on an object found in England, although it probably originates from Scandinavia; the other inscription is from southern Bavaria. Both inscriptions are dated rather early, to the fifth and sixth centuries.

(1) The first inscription is on an astragalus from Caistor-by-Norwich, East Anglia, dated circa 425-475 [Hines (1990a:442)]; the runes read raïhan 'roedeer', 0E râha (0E a < Gmc. *ai). The astragalus was a roedeer's bone.

This inscription may well be our oldest instance of the yew rune in an inter- pretable text. The yew rune appears to be used here instead of the i-rune; perhaps for the sake of variety. According to Sanness Johnson (1974:40), the two runes ï and i were used as graphic variants (in "historisk runetid").

(2) The second instance of the sequence al is on a buckle, found in 1991 near Pforzen in Bavaria, and dated to the second half of the sixth century. The inscription is transliterated aigil andi alirun 1 tahu gasokun. Clearly two people, Aigil and Aïlrun, quarrelled about (or 'condemned', or 'sought') something, which might be hiding behind the single 1. The names of the two persons, a man and a woman, are well-known because they appear in the ON VQlurzdr saga as Egill and Qlrfln. Egill is the brother of Vqlundr, and Qlruin2 is one of the swanmaidens. She is the one who becomes Egill's partner. Egill is known as an exceptionally skilled archer (Marold 1996).

(3) Another Anglo-Saxon instance of the yew rune, transliterated ï, can be found in the inscription sïbbd on the Loveden Hill urn (Lincolnshire), dated fifth or sixth century [Hines (1990a:443)]. Note that also in this inscription, the yew rune probably has been used as a variety for rendering the sound i.

(4) Another attestation from Germany (Freilaubersheim, Rheinhessen, last third of the sixth century)3 shows the yew rune presumably again as a variant for rendering the sound i in da?lna.

(5) Uncertain, but possible, is an instance of J' on a square fitting with rivets, dated last third of sixth century, from Heilbronn-Böckingen (Baden-Württemberg, Germany). The initial rune has been perforated by the rivets, but I conjecture J' may have been carved, since some remains of the sidetwigs can be seen. I read Ik arwi 'I, Arwi'.

(6) A sixth instance of the yew rune is found on the Charnay fibula (dép. Saône-et-Loire, France), dated to the second third of the sixth century. The brooch is inscribed with a nearly complete fukark containing a yew rune, and furthermore the legend: ufnkai iddan liano ha. The part ha has not been explained. Curiously enough, an inscription from Weingarten (Baden-Württemberg) has the legend aergu: feha:writ ja, and here too it is not clear what ja means [Looijenga (1997:152)].

2. Qlrûn was the daughter of King Kiár of Valland. Kjárr = Caesar, according to De Vries (1962:312). Valland is the land of the Romans or the Merovingians. Vqlundr is the name of a famous smith, a hero's name, according to De Vries. So the story is about a relation between a daughter of a Roman emperor and the brother of a famous Germanic weaponsmith. She came disguised as a swanmaiden, e.g. a Valkyrie, from the South, although the motif of the swanmaidens may have been a later addition to the original Story [see - Nedoma (1995:196)]. The second part, -run may be either of Celto-Germanic or of Celtic origin [Schönfeld (1965:196)]. The name element may mean 'secret'. It is remarkable that the name-element -nin is otherwise not attested in the surviving names of any of the early runic traditions (see Peterson (1994a) for instance).

The datings of the objects from the Continental corpus are based on Roth (1981).

(7) The legend of bracteate Nebenstedt (1)-B (Niedersachsen, dated to the end of the fifth, beginnning of the sixth century) is well-known; it reads: glïaugiz ulu rnz 1. It contains two instances of the yew rune, both times transliterated I. The reading glïaugiz wî(h)ju riirthz 1 is interpreted a.s: 'Glïaugiz. I consecrate (the) runes l[aukaz]'. The sound value represented by ï is uncertain. In Glïaugiz it may represent something like -ij-. ulu may reflect wi(h)ju, 3 singular present indicative of the Gmc. infinitive * wz7j which may or may not have been pronounced with a velar fricative in the middle. I presume the pronunciation of I in ulu may have been -ij-.

1 -1 1. Occurrences of the yew rune in later inscriptions Apart from denoting a vowel, the yew rune could also denote a consonant and as far as it is known, it was used as such exclusively by Anglo-Saxon runewriters.

(8) The oldest inscription that shows the yew rune denoting a consonant, (transliterated as 3 to avoid confusion with ï and h) is alme3ttig 'almighty ' (Ruthwell Cross). The inscription is dated to 700-750 AD. (9) Other instances are toro3treda in Great Urswick, North Lancashire, dated 750-850, and (10) eate3nne 'Eategn' in Thornhill (II). It is curious that the yew runes in alme3ttig and eate3nue should be rendered by 3, since the pronunciation probably was that of the semi-vowel j, corresponding to a palatalised g (Page (1968/1995:137); he states that the yew rune in alme3ttig and toro3treda represents a spirant).

(11) The Dover (Kent) rune stone (ninth/tenth century) bears a name jïslheard, +TLirMTK N, in which the yew rune clearly denotes a vowel.

(12) The Brandon (East Anglia) pin from the ninth century exhibits g, h, j, and 1 () in a fuJorcquotation: fuorcgwhnijïpxs. The g is rendered by the Sternrune +, the j has its so-called 'epigraphical form' 4 (known from manuscripts only) and the s has the so-called 'bookhand' r form. This would point to ecclesiastical influences [see Parsons (1994)]. The J' is in its usual place in the fuborc. Its sound value cannot be deduced from this inscription.

IV.

In three of the earliest inscriptions, Caistor-by-Norwich raïhan, Nebenstedt glïaugiz uïu, and Pforzen aïlrun, ..I' was probably used to render a sound such as long palatal jj: ij or ji(i), perhaps caused by the fact that it denotes the transitional stage between two syllables. In the other three inscrip-tions Freilaubersheim (da?ïna), Loveden Hill (sïbbd), and Heilbronn-Böckingen (Ik arwi) the yew rune denotes an i sound.

The value [ç] may reflect a secondary development, used only at a rather late date and in an ecclesiastical context in Anglo-Saxon England. In Old English, the name of the rune is variously given as eoh and th; one may conclude that J represented two sounds, one of a vowel and one of a consonant: [e] or [i] and [ç].4 On the other hand, in the Norse rune-poems the name was lr < Gmc. *thwaz, *eihwaz; the initial sound (confirms the acrophonic principle of the rune names) is that of a front vowel which is followed by the voiceless fricative [ç]. On the whole, it appears that e and i could and did interchange.5

In my opinion the problem is connected with both the linguistic value and the graphic representation of the yew rune. Analysing the sparsely recorded runic texts containing the yew rune, I am inclined to assume that it may have been devel9ped graphically as a bindrune, consisting of i and j: I and <> J' 6 The pronunciation might have been something like -ijor -ji-. If so, the yew rune may be a later graphic development that was not yet present in the initial runic alphabet (note that .J' occurs not before the fifth century).

It may be that the yew rune was designed to represent a combination of a vowel, i, and a semi-vowel j. The sequence -iiis known from the oldest runic evidence, e.g. Vimose talijo and Illerup wagnijo (circa 200 AD). One may have wanted at a later stage to merge i and j into one rune, because it sounded more like a monophthong than a diphthong, assuming that runographers wrote according to their pronunciation.' Thus, J' rendered a glide, ij, or j1 Both graphically and phonologically, it appears that J' combined the sounds j and i(i). A little later perhaps, J' could have been and was used to denote just [i] and [i:].

Page (1968/1995:137) suggests that the names eoh, Th do "not derive from the primary rune-name, but from a renaming brought about by the fact that [either] the letter's primary function had been superseded so that its form could be re-used with a new

value, ( ... ). V. The yew rune in the Pforzen inscription, The Pforzen legend alirun presents a baffling situation. There are potential etymological problems in connecting Alirun to the later OHG name Alrtin, or to ModHG Airaun 'mandrake', although Airaun is likely to be the (linguistic) counterpart of the ON Qlrtmn, the partner of Egill (aigil in the 'alerune'.8 Pforzen inscription). Her name literally means One should expect *Alurtjn(a) as the forerunner of both Alraun and Qlrtmn, but this is simply not what was carved in the Pforzen inscription. I suppose we are confronted here with either a scribal error, or that al represents a development that cannot yet be reconstructed.

But if the yew-rune rendered a sound in between [e] and [i:], which could not be

In both aigil and alirun the first part is written with a diphthong. In OHG, ai> ei, and in OS ai> in ON ai> in 0E ai> ã> (through fronting), see Jgili on Franks Casket. The historical Egill's forerunner may therefore *Aigil have been spelled as whereas the spelling in aïlrun is confusing. I assume that, on the analogy of aigil, the carver wanted to carve ailrun, probably instead of *alrun. Remarkably, both spellings, ai and al, occur in the first syllables of the names, and both syllables are stressed, according to alliterative verse. The other a's in the text are in andi and gasokun, and these a's occur in an unstressed position. The runographer may have wanted to express this controversy graphically. But why is there a difference in orthography: ai versus al with a yew rune? I wondered whether this may have resulted from a scribal error. By looking at how the runes are carved, we realize that the a in alirun has very long sidetwigs and the lower twig even crosses the bottom line. It looks as if this twig has been lengthened, i.e. carved in two strokes. I considered the possibility that this may have been the result of a graphical mistake. Instead of carving an i, the carver made an 1 too many or too soon and therefore changed it into a yew rune. The mistake may have occurred because an 1 was to follow in ailrun. Since a yew rune could be used instead of an i-rune, as we have seen in the ralhan inscription, the carver tried to repair his mistake by changing r into .1, by carving an extra sidetwig. This sidetwig coincides with the end of the lower twig of the preceding a-rune (giving the impression as if the lower twig was lengthened). Therefore, I think the sequence al is a scribal error. The question arises whether scribal errors occur more often in rune-inscriptions,

_isjUI I fli1II-KO II [s) fl•

(1) Freilaubersheim reads boso:wraetruna kda?ina:go1ida. The second row of the inscription starts with '< 'lj[i]k'. At a close look, the initial rune, which is quite abraded, has its pocket nearly at the bottom of its headstaif. Vaguely another pocket can be perceived higher up on the headstaif. Probably by mistake, a b-rune was at first intended, perhaps due to confusion, caused by b of boso right above it.

(2) A well-known instance of a mistake that evidently could not be repaired is Krogsta, with the inscription 114 I 1R, reading from right to left sïai naz, which generally is considered to be a mistake for stainaz.

(3) A third example is the Sievern bracteate with the legend All 5j1, which should be read from right to left as rwrilu. It was probably meant to be writu 'r[unes] I write'.

(4) The inscription on the brooch Neudingen-Baar (Baden-Württemberg) s?u?d?? midu klefiija is executed in three rows of runes beneath one another. The top row is very difficult to read since the runes have nearly vanished, because the tinned surface of the back has suffered badly from corrosion. One may read udim [with Düwel (1997a:492)], but especially the last rune is uncertain. The second row has clearly midu. One gets the impression that the carver was not satisfied with his first row and repeated the word in the second row, now from left to right.'

As far as the early English corpus is concerned, one wonders whether the carver made a mistake in the inscription on the scabbard mouthpiece of (5) Chessel Down (Wight), AF :K I, transliterated ko:?ori. The rune r cannot be k since that would disagree with A in aeko. It might be s, since the rune has the form of the so-called bookhand-s, such as the ones in later English and Frisian inscriptions. But what is son? A third possibility might be to transliterate it as 1, analogous to similar runeforms in bracteate legends. But again, what is ion? A solution may be to regard the rune as a scribal error,10 although it is unclear what was meant.

(6) The West Heslerton (North Yorkshire) brooch bears an inscription M I reading from right to left mien. The meaning might be 'jewellery', 0E mene, OS merii, pointing to the object, a piece of jewellery. If this interpretation is correct, the runecarver made an orthographical error in misplacing the letters (metathesis), perhaps as the result of some degree of illiteracy. Both Lagman (1989:33) andMeijer (1995:30ff.) have given several instances of metathesis (omkastning) in Viking Age inscriptions (see below).

9. Meijer (1995a:79) lists in his article the category: "Omission discovered: the mistake remedied by starting anew". He finds three occurrences. Another category that Meijer mentions is "The mistakes 'corrected' by adding the correct form". Here he finds five instances, for example Svista, U 193: litu ritu rita. Meijer: "On finding that the spelling ritu was wrong, the runographer added the correct form: rita".

io. Eichner (1990:329; note 45) assumes that the rune might be taken for a somewhat levelled w-rune. Thus one may read wor[hta]/ 'made'.

In two cases a b may have been carved incompletely with only one pocket instead of two.

(7) The Watchfield fittings bear an inscription HK I AI. PH transliterated as hriboki wusre. If the rune F> actually should be R, one would get busa, bu(r)sx 'purse'. This interpretation, 'H's purse', would refer to the object, a leather purse" (decayed, only the copper-alloy fittings were left).

(8) A second instance of an incomplete b is on the Bezenye brooch (Komitat Mosony, Hungary) arsiboda segun, of which Krause (1966:309) writes: "... nur den unteren Buckel, scheint einen oberen auch niemals gehabt zu haben".

(9) Finally, the Grdlösa inscription has an incomplete z-rune in M<FN r, transliterated ek unwodz.

Svante Lagman (1989) investigated cases of misspellings (feiristnirigar) in Viking Age inscriptions in Sweden. He found all sorts of deviating spellings, often due to the writer's orthographic uncertainties, differences in pronunciation, or a slow pronunciation, which leads to the writing of svarabhaktivowels (extra runor). Other categories he listed are uncertainties about the two r-sounds (R and r), and the difficulty in distinguishing between t, b and in certain positions. Besides, he has counted left-out runes (uteldmnad eiier utegiömd runor), metathesis and grafisk förväxiing, which occurs when runes are much alike, such as is the case with the younger futhark runes a and n-1 and.

Jan Meijer (1995) collected a number of scribal mistakes of a certain kind in Viking Age inscriptions. These mistakes are cases of (pseudo-)metathesis, i.e. "the inversion of the order of twogenerally neighbouringphonemes". He distinguishes various categories: "due to mistakes"; "sound-analysts"; "genuine metathesis" and "metathesis?". (1995a), Meijer discusses instances of corrections. He describes different types: "the correction was made across the original character" (56 times); "the correction results in a genuine bind-rune" (2 times); "the mistake discovered when a shallow 'sketch' had been made" (13 times); "the mistake 'wiped out"' (0 times); "omitted rune inserted" (54 times); "insertion of the omitted rune results in a bind-rune" (14 times); "omitted rune(s) added outside the text-band" (31 instances in 27 inscriptions); "omission discovered: the mistake remedied by starting anew" (3 times).

In a second investigation, Corrections in Viking Age rune-stone inscriptions

Neither Lagman nor Meijer mentioned a 'repair' or correction such as might have been made in the Pforzen inscription. However, they did not focus their research on this category of mistakes (i.e. a correction that results in a rune that was not intended, but which does render an acceptable text). As 11. Odenstedt (1991:62) suggests that the word should be read as pusa 'bag'. The whole inscription would be taken to mean 'the army's book's bag'. regards cases of graphic change (grafisk förvarling) repairs indeed occurred, although with a wrong, i.e. not wished-for result, e.g. h * resulting from + + a/n corrected into n/a by adding an extra side-stroke. It should be mentioned that there is one case in which the supposed repair of the rune did not result in a not-intended h +. On the contrary, this rune should be taken at its face value, which is z in a form that resembles z in the Charnay futhark: X . I refer to the Skâäng inscription, which in my opinion reads harijaz leugaz [Looijenga (1997:52f.)]. The peculiar 'mirrored' variety of z occurs also in the inscription on the Balingen brooch [Krause -Jankuhn (1966:302)] a n/r zdnloamiluk.

As a conclusion I suggest the possibility that the Pforzen inscription contains a scribal mistake in the sequence aïlrun. In this respect, the yew rune was not intended but the result of an attempt to repair an error. I suspect that the runographer wanted to write ailrun, but when the carver discovered that he had omitted the i-rune and wrote one 1 too soon or one too many, he made up for his mistake and emended this 1 rune with one extra side-twig, thus creating a yew rune.

Rijksuniversiteit Groningen NL-Groningen