Cover Page

The handle http://hdl.handle.net/1887/20979 holds various files of this Leiden University dissertation.

Author: Griffiths, Alan Title: A family of names : Rune-names and Ogam-names and their relation to alphabet letter-names Issue Date: 2013-06-18 PART 1

SOURCES OF NAME-FORMS AND MEANINGS

Part 1: Sources of Name-forms and Meanings

PART 1: SOURCES OF NAME-FORMS AND MEANINGS

Names and interpretations of the names of letters, ogam characters and runes are found in homilies and epistles intended to expound on biblical texts, in grammatical tracts and in acrostic poems. I list the principal medieval manuscript sources of the interpretations of Hebrew, Greek and Latin names in Tables 1 (Hebrew) and 2 (Greek and Latin), and give the main sources of name-forms and glosses recorded for the ogam and runic systems in Tables 3 (ogam), 4 and 5 (runes). In the case of Hebrew, Greek and Latin names, I have taken into account only those sources which include interpretations beside the name-forms, since it would be impracticable to give a list of all manuscripts which reproduce letter-forms and names without interpretations. In the case of ogam- and rune-names, however, lists giving only names, i.e. without any further gloss, have been included as well as glossed lists, since ogam and rune name-forms are more debatable than their Hebrew, Greek and Latin counterparts. I present the interpretations given in these manuscripts without prejudice as to the plausibility of letter-names, ogam-names or rune-names having had any influence on each other. All comments at this stage have been restricted to describing the nature of the records, their contexts and, in the case of Hebrew letter-names, their introduction and treatment in early Christian manuscripts and their dissemination by later commentators. I reserve detailed discussion of possible relationships between the names for Part 2. 1.1 Hebrew letter-names

1.1.1 General features of the glosses by Eusebius, Ambrose and Jerome Knowledge of the Hebrew language was certainly not widespread in Europe during the early Christian period and the breakup of the Roman empire. However, in certain circles, i.e. among Greek- and Latin-speaking Christians, there was at least some knowledge of the Hebrew alphabet, if only in connexion with the study of Holy Scripture. In a Christian context, for instance, Eusebius of Caesarea (c. 263–339),1 in his Praeparatio evangelica (Preparation for the Gospel), book X, chapter V, provided Greek interpretations of the Hebrew letter-names, while two prominent disseminators of the names and interpretations of Hebrew letters to a Latin- speaking public were Ambrose (c. 334 or 339–97) and Jerome (c. 342 or 347–419/20). Another commentator in Greek whose glosses on Hebrew letters in Ps. CXLIV (CXLV) were not so well known as those of Eusebius, but will be referred to later, was Hesychius of Jerusalem (d. 450).2 In his Praeparatio, which he saw as an introduction to Christianity for pagans, Eusebius not only gives the Hebrew names in Greek transcription, together with a translation of their meaning into Greek, but also suggests that they can be grouped in sequences of two, three or five, connected syntactically, so that each sequence appears to form a sensible statement. In

1 Eusebius became Bishop of Caesaria c. 314 and is known in particular for his Ecclesiastical History. 2 See §1.1.6.

31 A Family of Names the case of the first five letters – a aleph (a1lf = ma&qhsij [act of learning, instruction]), b beth (bhq = oi1kou [of the house], gen. of oi1koj), g gimel (gi&mel = plh&rwsij [a filling up, becoming full]), d daleth3 (de&lq = de<wn [of writing-tablets, books], gen. plur. of de<oj) and h heh (h = au3th [it], fem. sing. of au0to&j) – he says: o(mou= de\ ta\ pa/nta a)parti/zei toiau/thn tina\ dia/noian [If you join all these together, you will form a sentence of some sort]. The sentence he then produces is: ma/qhsij oi)/kou, plh/rwsij de/ltwn au(/th [the teaching of the house,4 it itself (is) an abundance of books]. The whole passage is given in Table 6. Ambrose, who almost certainly knew no Hebrew, produced what appears to be a series of homilies in which he developed allegorical interpretations of the Hebrew letter-names that are reproduced at the head of each of the twenty-two octaves of verses comprising Ps. CXVIII.5 By contrast, Jerome went to great lengths to learn Hebrew and produced two sets of interpretations of Hebrew letter-names: one in a letter (Epistula XXX) written to Paula, who had asked him to explain the form of Ps. CXVIII;6 the other as glosses on the psalms (De psalterio) in his Liber interpretationis hebraicorum nominum,7 one of two onomasticons that he produced on Hebrew and Greek names occurring in the Bible. In his letter to Paula, Jerome begins by pointing to various biblical passages based on the Hebrew alphabet and concludes with a summary of interpretations in a form very similar to that of Eusebius, which he called connexiones. In his De psalterio glosses, on the other hand, the letter-names are not treated as names in an alphabetical series, but are integrated in the list of Hebrew names that he has gathered from the psalms. Also, unlike his solution in his letter to Paula, where he gives a one-to-one correspondence between letter-name and interpretation, his De psalterio glosses provide several alternatives for some of the names, expanding on the connexiones interpretations and in some cases including an interpretation that is found in Ambrose’s homilies. Consequently, it is important to emphasize that none of the interpretations of Eusebius, Ambrose or Jerome was initially produced in the form of straightforward lists of linguistic equivalents. For Ambrose, the names were a starting-point for his homilies, while Jerome either followed Eusebius in reading a connected meaning into the Hebrew alphabet as a whole or else treated the letter-names as part of his general etymological interpretation of Hebrew names. For this reason, it is worth presenting the

3 The Hebrew letter-name was most likely daleth, but Eusebius, and following him Jerome, treated it as though it was deleth [door, leaf (of a door or a codex)]. This latter form seems to have been preferred because, especially in Eusebius’s transcription, de&lq, it compared more closely to Greek de<oj, and the Greek letter- name delta (see below §1.1.3, s.v. Daleth). 4 ‘The house’ in this context most probably refers to the Church; Jerome translates the first two Hebrew names as doctrina domus, which he interprets as doctrina Ecclesiae, quae domus Dei est (see Table 6). 5 PL XV: 1283–598. Kellner (1893: 31–166) provides a summary of Ambrose’s hermeneutics (based on sensus naturalis, sensus mysticus or rationabils and sensus moralis or ethicus) and a discussion of his homilies on Psalm CXVIII , with an interpretation of some of the letter-names. However, Kellner attempts to explain some of Ambrose’s interpretations via Syriac and Armenian, which seems an unlikely source for someone lacking a knowledge of Hebrew. 6 PL XXIII: 441–5. 7 PL XXIII: 871.

32 Part 1: Sources of Name-forms and Meanings solutions of all three commentators in their contexts. Ambrose’s homilies are too long to be quoted in extenso, but relevant passages are given in Table 7, even though a digest of this nature risks distorting the author’s argument. Jerome’s connexiones are reproduced alongside the text of Eusebius in Table 6, and his glosses on letters-names in his onomasticon are given in Table 8 (a) in De Lagarde’s transcription,8 which is closer to Jerome’s own orthography than Migne’s normalized spelling (see below, §1.1.3). For convenience, these texts have also been excerpted to provide the lists given in Table 9 (a) (columns A and B),9 together with transliterations and transliterations of the Hebrew words on which the glosses may have been based (columns C and D); a list of these words in square Hebrew script (without diacritics) is added in Table 9 (b). The precise sources used by Ambrose and Jerome are difficult to determine. It is clear from Tables 6 and 9 (a) that Jerome’s connexiones arrangement in his letter to Paula (Jerome 1) are based on the Greek interpretations by Eusebius. However, in his De psalterio glosses (Jerome 2), Jerome gives several alternative interpretations to those in his connexiones, the exact source of which cannot be traced. In the preface to his onomasticon, Liber interpretationis hebraicorum nominum, from which the De psalterio glosses (Jerome 2) are taken, he explains that he has been encouraged by brothers Lupulus and Valerianus to improve on the work of the Greek authors Philo of Alexandria (c. 20 BC– AD 50) and Origen (AD 184–254).10 He says he is able to do this because he has recently acquired other works on Hebrew ‘matters’ that he thinks will be new to Greek and Latin scholars. In the preface to his second onomasticon, De situ et nominibus locorum hebraicorum liber, he explains that he is elaborating on (as well as translating) the Greek work of Eusebius. There are indications that Eusebius based his interpretations on a source attributable to Origen. The Codex Marchalianus of the Septuagint, which used not only the Hexapla but also works by Origen, contains a series of marginal glosses with translations of biblical names, including letter-names relating to the first chapter of Lamentations. Klosterman was of the opinion that these glosses were ‘so gut wie sicher’ attributable to Origen.11 A list of the letter-names and glosses are included in Table 9 (a) (the two extreme right-hand columns under O). Glosses for h heh, z zayin, j cheth and  sin are missing from the Codex Marchalianus, but all these letters except h heh are found in another list attributable to Origen in Codex Colbertinus 4124, which Martianay (in the second edition of his works of Jerome, 1699) characterized as Origeniani aliud exemplar.12 From a comparison of the

8 De Lagarde 1909: 48–9. 9 The list labelled Jerome 3? in Table 1, i.e. the anonymous set of interpretations appended to the second of Jerome’s onomasticons, his De situ et nominibus locorum hebraicorum liber, is not included in Table 9 (a), since it is virtually identical to the list of column 4 (Jerome 2), with the notable exception of the gloss on coph, i.e. simius, ‘ape’, which appears to be unique to this list. Thiel (1973: 87) suggests that, since Jerome was probably the only person in the Latin-speaking world capable of interpreting Hebrew, the possibility that Jerome was also responsible for this list cannot be excluded. 10 Philo, also known as Philo Judaeus or Philo the Jew, put the emphasis on allegorical interpretations of the Bible; Origen wrote numerous exegetical treatises but is known in particulr for the Hexapla, a comparative study of various translations of the Old Testament. 11 Klosterman 1903: 136. 12 See Wutz 1914: 2, 210, 216.

33 A Family of Names

Marchalianus and Colbertinus glosses with those of Eusebius in Table 9 (a), it will be seen that Eusebius’s list is particularly close to the Marchalianus version, a notable difference being Eusebius’s omission of the second element in each of the glosses on a1lf, bh&q, gi&mel, de&lq, a1in, kw&f and qau=. Jerome’s dependence on Eusebius, particularly in his connexiones, is beyond doubt. His comments in the preface to his Liber interpretationis also make it clear that he is acquainted with Origen, while the De psalterio glosses (Jerome 2) in that onamasticon demonstrate his acquaintance with the interpretations in Ambrose’s homilies. Ambrose’s sources, however, are less clear. In many cases his interpretations deviate from Eusebius’s and Jerome’s to such an extent that it has been suggested that he must have used sources other than those available to either of these two. For example, Wutz13 points to Ambrose’s use of repeated phrases like ut alibi invenimus and ut alia interpretatio habet as evidence that he made use of a variety of Greek and Latin sources. However, there is no indication, according to Wutz, of any correspondece between these sources and the Marchalianus or Colbertinus lists. Rather, he suggests, several of Ambrose’s interpretations can be explained via misreadings of Latin and Greek sources, e.g. zayin = fornicatio (cf. Jerome 2, Table 9 (a), col. 4) = porneuein, read by Ambrose (Table 9 (a), col. 2) as poreu&ein = duc, with te being taken from a particular context; nun = go&noj (cf. Jerome 2: foetus), read by Ambrose as mo&noj = unicus. An alternative suggestion, put forward by Müller,14 is that Ambrose may have had access to rabbinical writings, in particular two separate recensions of a treatise on the alphabet attributed to Aqiba, who was a rabbi at the academy in Yavne around the time of the second Jewish uprising (c. AD 134). Some of the work attributed to Aqiba appears to have inspired later Kabbalistic thought, but the recensions of The Alphabet of Rabbi Aqiba, as it has come to be known, contain not only mystic passages, but also a body of non-mystic, pedagogic material that does not relate directly to the Kabbalistic tradition.15 There is no mention of these recensions before the tenth century, and it is thought that they originated sometime between the seventh and ninth centuries.16 Consequently, Wutz doubts whether they will have had any influence on either Ambrose or Jerome. In his view it may well have been the Church Fathers who in fact influenced rabbinical interpretations, rather than the other way round.17 However, even if there were an element of truth in Wutz’s suggestion, it would still be necessary to explain the origin of the distinctively Hebrew characteristics of the alphabet interpretations that are found in Ambrose and are not encountered in Eusebius or Jerome. A more likely scenario, in my opinion, is one in which rabbinical ‘Haggadot’ (i.e. ‘tales’) concerning the alphabet influenced both Ambrose and, to a lesser extent, Jerome, and were taken account of in the Babylonian Talmud before being absorbed into The Alphabet of Rabbi Aqiba (which I shall refer to simply as The Alphabet).18 It is therefore worth 13 Wutz 1914: 224–5. 14 Müller 1911: 11. 15 Ginzberg (1909, vol. I, note 10) describes the non-mystic material as belonging to a group of what he calls ‘“pedagogic Haggadot” [i.e. ‘tales’], whose object it is to render the elementary instruction to the young interesting and attractive.’ 16 Strack and Stemberger 1982: 311. 17 Wutz 1914: 229, footnote 3. 18 A brief summary of the lore on the alphabet found in The Alphabet of Rabbi Aqiba is given in the Gemara commentary on the Mishna, Shabbath XII.3, of the

34 Part 1: Sources of Name-forms and Meanings

considering alphabetical passages in The Alphabet alongside those in Ambrose and Jerome in an attempt to appreciate the type of material on which they may have based their interpretations. 1.1.2 The Alphabet of Rabbi Aqiba The two recensions of the alphabet treatise attributed to Rabbi Aqiba were published in German translation in 1909 by Wünsche in his Aus Israels Lehrhallen, IV, under the titles Erste (ältere) Rezension des Alphabet- Midrasch des R. ’Akiba. (Nach dem Krakauer und Amsterdamer Druck. Jellinek Bet ha Midrasch III, pp. 50-64) and Zweite (jüngere) Rezension des Alphabet-Midrasch des R. ’Akiba (Nach dem Konstantinopolitaner und Venediger Drucke. Jellinek Bet ha Midrasch III, pp. 12–49).19 The two versions are quite distinctive in the way they treat the Hebrew alphabet. The earlier one, which is much shorter and generally less discursive, begins with each of the twenty-two letters of the alphabet lining up before the Lord to argue its claim to be the one letter with which God created the world. The procedure is the same in each case, beginning with the last letter, t tav, and working backwards through the alphabet to a aleph. Each letter puts forward a reason to support its claim, citing a passage in Holy Scripture containing a word with a positive connotation in which it is the initial letter. Tav, for instance, cites the word hrwt torah itself, as in Deut. XXXIII.4: ‘Moses commanded us a law (torah).’ God in turn counters with a quotation that puts the letter in a negative light, as for instance for wt tav, Ezek. IX.4: ‘And the Lord said unto him, go through the midst of Jerusalem, and set a mark (wt tav) on the foreheads of men that sigh and that cry for all the abominations that be done in the midst thereof.’ When eventually it is beth’s turn, the letter is able to claim the word ûwrb barukh, ‘blessed’, as in Ps. LXXXVIII.52: ‘Blessed be the Lord for ever more,’ and in many other passages. And the Lord immediately accepts beth as the letter with which to create the world, for it says in Gen. I.1: ‘In the beginning (tyarb be-reshith) God created heaven and earth.’ In other words, Creation began with the first word of the first verse of Genesis, b be, [in]. Aleph, a, is put out at this presumption of primacy by beth, and bewails the fact that, as the smallest number (one), he cannot stand up to the greater numerical values of all the other letters. But the Lord comforts him by pointing out that: ‘You are One (dja achad) and I am One (dja achad) and

Babylonian Talmud (fol. 104a); see the translation by Goldschmidt 1930: 746–8. 19 Wünsche uses the word ‘Midrasch’ to describe The Alphabet, and in Dan et al. (1986: 156) it is also referred to as a ‘Midrash’, although it is debatable this description is strictly correct. The treatise as a whole is not a midrash in the sense that it belongs to the corpus of literature consisting of purely scriptural exegesis. It is in fact an anthology which combines non-mystic, pedagogic elements (see note 2 above) with merkabah mysticism. For the most part, however, the passages explaining the alphabet names can be said to be ‘of the form Midrash’ as described by Goldberg (1987: 39): ‘the single Midrash, the single Scriptural exegesis [as distinct from Midrash as a designation for a corpus of literature], shall be called here a sentence of the form Midrash. ... The constituents of a sentence of the form Midrash are a “lemma”, that is to say, a certain part of Revealed Scripture, as well as an exegetical operation which can or cannot be realized linguistically, and a dictum – a certain proposition. Sentences of the form Midrash ... say, this word or this part of a sentence in Scripture means (= intends) such and such, which is then expressed in the dictum.’

35 A Family of Names the Torah is One (tja achath)’20 and ‘ykna anoki [i.e. the personal pronoun ‘I’] is the beginning of all words (the Ten Commandments [cf. Ex. XX.2: ‘I am the Lord thy God’]) and a aleph is the beginning of all letters.’ 21 As a result, alaph was put in the initial position. In the next section of The Alphabet, each letter is discussed in terms that mostly characterize its shape and position in the alphabet. A neat example concerns the series dgb beth, gimel, daleth:

Beth. Why is his face turned towards gimel and why is gimel turned towards beth? Because ‘the beth’ (tybh ha-bayith) resembles a house, all the doors of which are open.

Gimel resembles a man (rbg gibbor) who, seeing a poor man at the door, enters his house to fetch food for the poor man. Why is gimel’s hip leaning on daleth? Because all tokens of charity are for the weak (ld dal, [poor, needy] ).22 Here the author involves not only the shapes and juxtapositions of the letters in his explanation, but also develops an intricate play on the respective meanings of equivalent words:

bayith tyb ‘house’ gamal lmg ‘bestow on, do (good) to’ (and also the initial of rbg gibbor [man]) deleth tld ‘door’ dal ld ‘door; needy, poor’.23 The third section of the Alphabet is to a certain extent comparable to the connexiones of Eusebius and Jerome, where series of letters are interpreted as forming semantically connected groups. For aleph, beth, gimel, daleth, for example, the commentator proposes:

Another explanation. What is meant by tld lmyg tyb ¹la (aleph beth gimel daleth)? µyld lwmg hnyb ¹la (alaph binah gemul dalim): learn insight (reason), show [mercy] to the poor. And why is gimel’s foot stretched out in front of him? Because those who show mercy run after the poor. Why are daleth’s handles behind him? Because a poor man looks behind him and says: ‘Who is it who is coming after me? Perhaps he will show goodness towards me.’24

20 ‘Du bist Einer (dja achad) und ich bin Einer (dja achad) und die Thora ist Eine (tja achath)’ (Wünsche 1909: 180). All the quotations from The Alphabet are from Wünsche’s German translation (1909: 168–268). The English translations are my own. 21 ‘ykna (anoki [Ich]) ist der Anfang von allen Worten (zehn Geboten) und Aleph ist der Anfang von allen Buchstaben.’ (Wünsche 1909: 181). 22 ‘tyb beth. Warum ist sein Angesicht nach dem Gimel gerichtet und das Gimel nach dem Beth? Weil das Beth (tybh ha-bayith) einem Hause gleicht, dessen Türen allen geöffnet sind. ‘Das lmyg gimel gleicht einem Mann (rbg gibbor), der einen Armen an der Tür sieht und in sein Haus hinein geht, um aus ihm Speise für den Armen herauszuholen. Warum ist die Hüfte des Gimel an das Daleth angelehnt? Weil alle Liebesbeweise für den Schwachen (ld dal, Armen) sind.’ (Wünsche 1909: 181). 23 The word ld dal in the sense ‘door’ is a hapax legomenon, found only at Ps. CXL.3, where it is used metaphorically: Pone, Domine, custodiam ori meo et ostium circumstantiae labiis meis [Set a watch, O Lord, before my mouth; keep the door of my lips]. It is clear from the text of The Alphabet, however, that ld, dal [poor] and ld, dal [door] are intended as pairs. 24 ‘Eine andere Erklärung. Was bedeutet: tld lmyg tyb ¹la ? µyld lwmg hnyb ¹la [Shabbath 104a], lehre Einsicht (Vernunft), erweise den Armen (Gutes). Und warum ist der Fuss des Gimel vor sich hin gestreckt? Weil die Liebeserweise Uebenden

36 Part 1: Sources of Name-forms and Meanings

Then follow several explanations of series of letters based on the principles of:

(i) at bash ger daq, etc., qd rg b ta i.e. the first letter (a aleph) paired with the last (t tav), the second (b beth) with the penultimate ( sin), etc.25 (ii) al bam gan das, etc., sd ÷g µb la i.e. the first letter paired with the twelfth, the second with the thirteenth, etc. (iii) ahs btho gyp,etc., ¹yg [fb sja i.e. first, eighth, fifteenth letters, second, ninth sixteenth letters etc. With the first series, the letters of each pair are initially interpreted separately, i.e. aleph separately from beth, etc., but subsequently the pairs are interpreted as forming actual words, e.g. wb bosh [be ashamed, be confused]. With the second and third series, it becomes necessary to re- interpret the pairs and triplets as words. Thus the reader or listener is instructed: ‘Do not read sja ahs but swjia ai-chus [spare, have pity]’,26 and similarly µb la al bam as µbl labam, which the text connects with Is. V.24 and the word for ‘flame’ hbhl lehabah.27 In the second, later recension of the The Alphabet, the formula ‘Do not read x but y’ is also required quite frequently, but the main feature of this text is the use of a form of notarikon28 to explain the meanings behind the letter-names. Thus aleph is expanded to give a sentence with words beginning first with each of the three letters a l p a l p, and then p l a:

Another explanation. a jtpa ÷wl hp hp ÷wl jtpa, aleph: eftach lashon peh peh lashon eftach, ‘I open my tongue, my mouth, and my mouth, my tongue I open.’ 29 hinter den Armen herlaufen. Warum sind die Henkel des Daleth hinter ihm? Weil ein armer Mensch hinter sich sieht und spricht: Wer ist der, der hinter mir kommt, vielleicht erweist er mir eine Wohltat.’ (Wünsche 1909: 186). 25 The at bash arrangement of the alphabet was one of the principal means of encryption, known as temara. It would appear that the method was known to the author of the book of Jeremiah, where in chapter XXV.26 we read: ‘...and the king of Sheshach shall drink after them.’ In Jer. LI.41 it becomes clear that Sheshach is none other than Babel, which has been encoded on the basis of substituting letters according to the at bash arrangement of letters, with b=sh, and l=ch (i.e. final k): B (a) b (e) l= Sh (e) sh (a) ch. 26 Wünsche 1909: 192. 27 Wünsche 1909: 195. Is. V.24 reads: ...sicut devorat stipulam lingua ignis, et calor flammae exurit, ... ( ... as the [tongue of] fire devoureth the stubble, and the flame consumeth the chaff, ...). 28 Notarikon is a method used extensively by Kabbalists whereby the answer to a question can be divined by taking the initial letters of the words comprising the question, and then taking the final letters. For example, Moses’s question in Deut. XXX.12: ‘Who shall go up for us to heaven?’ produces from the initial letters the answer mylah, ‘circumcision’, while the final letters give the name of the Lord, IHVH. The complete answer is thus: ‘The circumcized will reach the Lord.’ Early Christians justified their use of the symbol of the fish on the same basis: the word for fish in Greek was i0xqu=j, the initial letters of which could be expanded to Ihsou&j Xristo&j Qeou= Ui9o_j Swth&r [Jesus Christ, the Son of God, the Saviour]. 29 ‘Eine andere Erklärung. a jtpa ÷wl hp hp ÷wl jtpa (ich öffne die Zunge, den Mund, und den Mund, die Zunge öffne ich).’ (Wünsche 1909: 200). The verbal root of jtpa is jtp ptch. The allusion here is to Ib XXXIII.2: Ecce aperui os meum, loquatur lingua mea in faucibus meis (Behold, now I have opened my mouth, my tongue hath spoken in my palate).

37 A Family of Names

This process is repeated ten different ways for aleph and is similarly applied to all the other letters, though not as extensively as with aleph. Some examples that will be of relevance in the discussion of ogam- and rune- names are:

Beth: The names of God’s deeds beginning with beth, e.g. the creation. Gimel: God’s beneficence to man. Zayin: The key to God: Zerubbabel. Qoph: The story of Moses. Thus, in both recensions of the treatise, there is not only an elaboration of a series of moral points based of interpretations of Hebrew letter-names, dealt with in succession, as in Ambrose’s homilies on Ps. CXVIII, but also a treatment of the names in alphabetical order as if they composed meaningful sentences, as in the connexiones of Eusebius and Jerome. Some passages in The Alphabet refer specifically to the merkabah, i.e. the divine throne or chariot of Ezekiel’s vision, and clearly belong to that mystical tradition. Also, the relation between letters of the alphabet and the creation of the world, which is developed in the earlier of the two recensions, is the subject of much later writings in the tradition of the Kabbalah arising from the Maaseh Bereshith [History of Creation]. On the other hand, the general approach in The Alphabet of Rabbi Aqiba is quite different from, say, that found in the mystical Sefer Yetsirah [Book of Creation], which was probably written during the fourth century AD.30 In the relevant section of the Sefer Yetsirah, for instance, the twenty-two letters of the alphabet are divided into the three ‘mothers’, seven double letters and twelve simple letters. The three mothers are aleph, mem and shin, which represent the three primordial substances, air (the aspirate aleph), water (mayim [water]) and fire (ish [fire]), which were thought to constitute the whole of creation. No part of this mystical tradition is discernible in the glosses of Eusebius, Ambrose or Jerome, but some of their interpretations do show parallels with the non-mystical explanations found in The Alphabet which probably derive from pedagogic ‘Haggadoth’ tradition (see note 15 above).

1.1.3 Transcription of letter-names by Jerome At this point, it is probably useful to consider two features of the treatment of Hebrew letter-names by Eusebius, Ambrose and Jerome in more detail, namely the degree of latitude in transcription and the didactic motivation that led to a certain diversity of interpretation as well as a treatment of the alphabetical sequence of letter-names as a meaningful series of concepts rather than a simple list. The question of transcription is best illustrated by the problems Jerome seemed to have in transcribing Hebrew in general. In his onomasticons, in particular, Jerome has attempted to develop a consistent system of transcribing Hebrew into Latin. The difficulties he has in this respect are reflected in the comments he has inserted in his lists. Among the comments of relevance to his transcription of Hebrew letter- names in the section De psalterio in his Liber interpretationis hebraicorum nominum are the following.31

30 Dan et al. 1986: 7. 31 The passages quoted here are from De Lagarde 1870, vol. I, the page and line numbers from this edition being cited as, for example, Lag. 4, 4–14, i.e. p. 4, lines 4 to 14, or Lag. 48, 7–49,10, i.e. p. 48, line 7 to p. 49, line 10.

38 Part 1: Sources of Name-forms and Meanings

Aleph, ayin, he, cheth: On the problems (and delights) of transliterating these letters Jerome inserts the following passage between the preface to his Liber interpretationis and his first lemmata:

It should not immediately be assumed that, where names are listed with the letter A, which in Hebrew is called aleph, this is the only [letter] which is listed (i.e. that only words beginning with aleph are listed), for sometimes they begin with ain, often with he, [and] not infrequently with heth, which letters interchange aspirates and their vowels; it should therefore be noted that, in Genesis as in the other books, where a name begins with a vowel, it may begin in Hebrew with [any of] various letters, as we have said above, but since in our language there is not such a diversity of vowels, we simple folk are delighted when it turns out that those names which are not written in the same way in their language are seen by us to be altered in interpretation.32 An illustration of the confusion that can be caused and is still apparent in the English Authorized Version of the Bible can be gauged from a comparison between a random sample of names which Jerome has transcribed with initial Latin O and the same names in AV transcription:

Jerome’s transcription33 AV transcription Initial Hebrew letter Orech, Ochozath Erech, Ahuz-zath aleph Olibama, Ofir siue Afir34 Aholibamah, Ophir aleph Omar Omar aleph Odollamites Adullamite ayin Ofim Huppim cheth

One of the consequences of these considerations is that, in his De psalterio list with letter-names. Jerome has alphabetized ayin under Latin A,35 although he sometimes transcribes initial ayin as G, as in hz[ Gaza and hrm[ Gomorra, adding the comment: ‘... it should be noted that this does not have the letter G in Hebrew but is written with the vowel ain.’36 It seems that he was in fact following traditional name-forms in Latin, as can be seen in another example, this time in his De situ et nomibus locorum hebraicorum liber: ‘Gader ..., which is written without the letter G in Hebrew, Ader.’37 (The AV transcription of this name, rd[, is Edar or Eder.) Fortunately,

32 Non statim, ubicumque ex A littera, quae apud Hebraeos dicitur aleph, ponuntur nomina, aestimandum est ipsam esse solam quae ponitur, nam interdum ex ain, saepe ex he, non numquam ex heth litteris, quae adspirationes suas uocesque conmutant, habent exordium, sciendum igitur quod tam in Genesi quam ceteris in libris, ubi a vocali littera nomen incipit, apud Hebraeos a diuersis (ut supra diximus) incohetur elementis, sed quia apud nos non est uocum tanta diuersitas, simplici sumus elatione contenti, unde accidit ut eadem uocabula, quae apud illos non similiter scripta sunt, nobis uideantur in interpretatione uariari (Lag. 2, 4–14). (In my translations of Jerome, I have tried to keep as close as possible to his vocabulary and syntax without compromising too much on intelligibility.) 33 Lag. 9, 11–17. 34 This Ofir is clearly different from his Ofir (AV: Hepher), Lag. 19, 22, of which he says: ... apud Hebraeos per heth litteram scribitur [written by the Hebrews with heth]. 35 Lag. 48, 7–49, 10. 36 ...sciendum quod G litteram in hebraico non habet, sed scribitur per uocalem ain (Lag. 6,28–29). The original phonemes */’/ (’ayn) and */g´/ (ghayn) were assimilated and subsequently both written with the ’ayin-grapheme. The distinction was maintained in the Septuagint transcription of the name Go&morra, which was probably Jerome’s source (Gzela 2011: 80; Krebernik 2007: 125–6 and 133, f. n. 81). 37 Gader ... , quae absque G littera in hebraeo Ader scribitur (Lag. 124, 32–3).

39 A Family of Names tradition appears to prevail in his transcription of ÷d[ Eden, which also has an intial ayin, although by analogy with his Gader it might conceivably have been reproduced as *Gaden. On the letter j cheth Jerome remarks: ‘Chettaeus ... , but this name does not begin with a consonantal letter, but in fact with heth, ... .’38 Thus, although he does not regard cheth as a consonant, he feels obliged to transcribe it initially as ‘Ch’, and consequently classifies words beginning with cheth alongside those beginning with k kaph, which he also transcribes with ‘Ch’ (see below). He sometimes feels the need to note that the word has an initial cheth (e.g. µj Cham; tj Chettaeus; ÷rj or ÷rh Charran; ÷wrbj Chebron; rbj Chaber39), but not always (e.g. tj Chet).40 With a name like ûwnj Enoch, on the other hand, Jerome ignores the initial aspiration (as with µypj Ofim above) and lists it next to wna Enos, with initial aleph.41 Similarly, in listing the letter-names in his De psalterio, he places Eth after E42 and not as Chet next to, say, Chaf (i.e. kaph). Peh and vav: In dealing with both initial and final Hebrew p, Jerome consistently transcribes it in his onomasticons as Latin F, e.g. initially, ÷wyp Fison,43 and finally ¹la alef, ¹sa asaf, ¹wq cof44 (although aleph occurs in his opening comments on the vowels, see above). This usage corresponds to Eusebius’s use of f for Hebrew p, and is spelled out in a comment Jerome makes when he is confronted with putting forward etymologies for New Testament names like Petrus, Pontius and Pilatus, where he interprets the pi of Pilatus as Hebrew peh (fe) [mouth]:

Petrus ‘recognizing’. Pontius ‘turning away from decision’. Pilatus ‘mouth of a hammerer’, but it should be noted that the letter P is not known among the Hebrews and there are not any names that are pronounced with this letter.45 The consequence for Jerome’s alphabetization is that he transliterates the initial letter p peh as fe and lists his glosses on it and on all Hebrew names in the Old Testament beginning with peh under Latin F, after E, whereas he places w uau (vav) in the position of Latin U. Samech, tsadeh, shin/sin and zayin: Jerome’s solution for the various Hebrew sibilants is to ignore the differences between s samech, x tsadeh and  shin/sin and to transcribe them all with Latin S, while z zayin, the alphabetical equivalent of Greek zeta, is placed at the end of the Latin alphabet as Z. Thus zaith – his De psalterio version of zayin (transcribed in his connexiones as zai, following Eusebius) – occurs at the end of the alphabetical list.46 On the name of Nazareth he says: ‘It is not, however,

38 Chettaeus ... , sed hoc nomen in hebraico non incipitur consonanti littera, uerum ab heth, ... (Lag. 4, 18–20). 39 Lag. 4, 11–13; 4, 18–20; 4, 21–2; 4, 22–3; 5, 2–3. 40 Lag. 4, 28. 41 Lag. 5, 17. 42 Lag. 48, 18. 43 Lag. 6, 11. 44 Lag. 48, 8–12. 45 Petrus agnoscens. Pontius declinans consilium. Pilatus os malleatoris sed sciendum est quod apud Hebraeos P littera non habetur nec ullum nomen est quod hoc elementum sonet (Lag. 65, 18–21). 46 Zaith oliva uel fornicatio siue haec [Zaith ‘olive’ or ‘fornication’ or ‘this’ (Lag. 49, 9f–10), where oliva is equivalent to Hebrew zayith, fornicatio to zanah, and

40 Part 1: Sources of Name-forms and Meanings written with the letter Z but with the Hebrew sade, which is pronounced as neither S nor Z.’47 On samech, tsadeh and shin/sin Jerome remarks:

What we have said in principle to be observed for the vowels, namely that in our language there sometimes will be only one letter while various vowels will be pronounced in Hebrew, the same should now also be noted for the letter S, since in Hebrew there are three S letters: one which is called samech and is defined simply almost like our letter S; another [called] sin, for which there is no equivalent hissing sound that invades our speech; a third [called] sade, which our ears utterly shy away from. So wherever it happens that such things define names differing one from the other, that is the reason that they are written with various letters.48 Tav and teth: A comment that illustrates Jerome’s perception of tav and teth is found in his list of place-names:

Thus far we have been reading [words] with a simple letter T; but since those [words] which follow are written in Hebrew, not with teth, but with thau, that is Greek theta, we should read [them] with aspiration.49 Again, Jerome is not entirely consistent in this distinction (cf. jbf Tabech, with initial teth; but jt Tachos with initial tav50), but the consequence for his list of letter-names in De psalterio is that teth is treated as plain Latin T and tav is transcribed with aspiration:

Teth, ‘good’. The four names which follow should be presented with added aspiration. Thabor, ‘coming light’ or ‘light will come’. Thau, ‘sign’ or ‘below’. Thalasar, ‘he appends the prince’. Thobel, ‘to offer’.51 Kaph and qoph: Much the most problematic Hebrew letters for Jerome to transcribe appear to have been k kaph, q qoph and j cheth. Latin writers preferred to use C rather than K, and although Latin, unlike Greek, used Q as a letter, Jerome uses C for q qoph as well as some instances of k kaph. For initial j cheth he mostly uses Ch, equating it with Greek χ chi, but there are many instances where he also transcribes initial k kaph as Ch, as in his version of kaph in his De psalterio text: chaf alongside cof for qoph.52 Typical of Jerome’s approach to K, C and chi, following a list of names that he has transcribed with initial C, is: haec to zoth (fem. of zeh)]. 47 Scribitur autem non per Z litteram sed per hebraeum sade, quod nec S nec Z litteram sonat (Lag. 62, 25–27). 48 Quod in principio dixeramus in uocalibus litteris observandum eo, quod apud nos una sit interdum littera et apud Hebraeos uariis uocibus proferatur, hoc nunc quoque in S littera sciendum est, siquidem apud Hebraeos tres S sunt litterae: una, quae dicitur samech et simpliciter quasi per S nostram litteram describatur: alia sin, in qua stridor quidam non nostri sermonis interstrepit: tertis sade, quam aures nostrae penitus reformidant. Sicubi ergo euenerit ut eadem nomina aliter atque aliter interpretentur, illud in causa est quod diversis scripta sunt litteris (Lag. 10, 1– 11). 49 hucusque [Trachonitis] per T simplicem litteram legerimus, uerum quoniam quae secuntur apud Hebraeos non ex teth, sed ex thau, id est ex theta graeco, scribantur, cum adspiratione legere debemus (Lag. 155, 28–30). 50 Lag. 11, 14. 51 Teth bonum. quattuor nomina quae secuntur adspiratione addita proferenda sunt. Thabor ueniens lumen uel ueniat lux. Thau signum uel subter. Thalasar adpendit principem. Thobel offerre (Lag. 49, 3–7). 52 Lag. 48, 12–15. The source of Jerome’s chaf appears to the Septuagint’s xaf.

41 A Family of Names

Thus far we should read [the names] with the letter C, that is Greek kappa. From here on [any name] should be both written and read with the letter chi, which contains in it aspiration and is rarely found in Latin.53 The immediately preceding names in fact all begin in Hebrew with q qoph, whereas those coming after the remark begin in Hebrew with kaph, cheth, ayin or aleph (see also the comments on cheth, ayin and aleph above). However, Jerome transcribes the name of the µytk Chittim or Kittim (with an initial kaph) with a C, as Citiis,54 although earlier he spells it Chetim.55 In his spelling of ÷[nk Canaan as Chanaan,56 he is following the Greek Xanaan, but can still give Cananaeus for Canaanite.57 On the other hand, the Ch of Chaleb, Chamos, Chozbi, Chenneroth, Chaslon and Cherubin represents in every case an initial kaph.58 This is a very small and random sample but it would appear that Jerome mostly perceives kaph as equivalent to Greek chi and qoph as an unaspirated [k], to be represented in his Latin orthography by C. In fact, he says of Q:

The letter Q, which is not heard in either Greek or Hebrew, is not found in any other language except Latin, which is why we make no mention of it.59 His use of C for qoph would appear to be justified by the Greek use of K for Latin Q. However, this use of C for Greek K, representing Latin Q, leads him to some unusual etymologies. For example, he recognizes that the biblical Greek Kurh&nioj represents the Latin:

Cyrinus, ‘guardian’, which is pronounced better in our language [i.e. Latin] and more precisely with the letter Q, that is to say, Quirinus.60 The gloss haeres [guardian] in this interpretation, however, plays on a Greek word beginning with K, since it clearly refers here, not to the meaning ‘heir’, but to ‘lord, master, guardian’, which is the sense of Greek ku&rioj and Latin curator. A second example is Jerome’s transcription and glossing of the biblical Greek kodra&nthj [farthing, one fourth of an as]: ‘Codrantes, “mist” or “darkness”, which we pronounce with the letter Q: quadrans.’61 His reference to the Roman quadrans is certainly correct, but where his caligo uel tenebrae comes from is less clear. The Hebrew for ‘mist’ is da ed, with initial aleph, but a Hebrew word closer to the spelling of codrantes would be the verb rdq qadar [to be black or dark]. Finally, in his glossing of cof with auis, [bird],62 i.e. Hebrew ¹w[ oph,

53 huc usque per C litteram, id est graecum kappa, legere debemus: exin per elementum chi, quod adspirationem in se continet et a Latinis minime habetur, scribendum pariter ac legendum (Lag. 111, 24–27). 54 Lag. 17, 4. 55 Lag. 4, 14 56 Lag. 17, 11. 57 Lag. 61, 3. 58 Lag. 17, 11–17. 59 Q litteram, quam neque Graeci resonant nec Hebraei, exceptis Latinis nulla alia lingua habet. unde et a nobis penitus praetermissa est (Lag. 29, 23–25). 60 Cyrinus haeres, qui apud nos melius effertur et verius per Q litteram, ut dicatur Quirinus (Lag. 64, 9–11). 61 Codrantes caligo uel tenebrae. quem nos per Q litteram quadrantem dicimus (Lag. 61, 1–2). 62 Lag. 48, 12.

42 Part 1: Sources of Name-forms and Meanings

Jerome has apparently taken the initial ayin as c-, whence his cof rather than *chof. 1.1.4 Didacticism and diversity of interpretation As already indicated, the interpretations by both Ambrose and Jerome need to be seen as integral to their didactic purposes rather than as purely linguistic exercises. Ambrose’s homilies on Ps. CXVIII were certainly didactic; Jerome’s etymologies of names, although partly motivated by linguistic curiosity, were mainly aimed at delving the deeper meaning of names in the Hebrew texts of the Bible, while his connexiones were intended to draw out the deeper meaning of the Hebrew alphabet. This didactic motivation was apparently aided by the degree of latitude in transcription that allowed the possibility of ‘adjusting’ the form of names to produce words suited to the message Ambrose and Jerome wanted to communicate. That is not to say, however, that they deviously manipulated their interpretations; rather, they appear to have been following a Hebrew tradition exemplified by The Alphabet of Rabbi Aqiba. Ambrose, for example, could associate daleth with nativitas in the same way that the letter daleth in the first recension of The Alphabet pleads its case by citing verse 4 of the acrostic Ps. CXLIV: ‘One generation shall praise thy word to another’, where the Hebrew for ‘generation’ is rwd dor, i.e. nativitas.63 God, on the other hand, counters with Deut. XVII.8: ‘If there arise a matter too hard (wonderful) for thee in judgment, between blood and blood, between plea and plea ...’, where the Hebrew for ‘plea’ is ÷yd din and for ‘blood’ µd dam. It was a matter of building an argument by using letters of the alphabet in series as a starting-point and playing with that letter as an alliterative device in order to choose a word suitable to make the point. To a certain extent it was similar to the use of the initials of letters of the alphabet as a mnemonic or intellectual exercise in an acrostic poem, as is found in the acrostic passages in the Bible, including Ps. CXVIII. Daleth was not eqivalent to rwd dor, ÷yd din or µd dam, but these words all began with the initial d daleth. This type of exercise was what also allowed, for example, the play on ld dal [poor, needy] (beside tld deleth, ld dal [door]) that was quoted above from the The Alphabet in connexion with the interpretation of the series beth, gimel, daleth, although in this case the involvement of two homonyms (ld dal [door], and ld dal [poor]) was more ingenious than relying on simple alliteration. It is therefore too constricting to expect Ambrose’s and Jerome’s interpretations of the Hebrew letter-names to represent one-to-one correspondences between name and meaningful Hebrew word. In Tables 9 (a) and (b) an attempt has been made to summarize the diverse Hebrew words that the two authors were probably influenced by when choosing their interpretations. In the following commentary on the individual letters in this table, ‘Eusebius’ refers to Eusebius’s glosses from his Praeparatio evangelica X. 5 (columns 1, A and B), and ‘Ambrose’ to Ambrose’s glosses from his homilies on Psalm CXVIII (columns 2, A and B). Jerome’s interpretations in the connexiones in his letter to Paula (columns 3, A and B) are referred to as Jerome 1, and his glosses in the De psalterio section of his Liber interpretationis hebraicorum nominum (columns 4, A and B) are

63 Wünsche 1909: 179.

43 A Family of Names referred to as Jerome 2.64 Aqiba 1 refers to the earlier of the two recensions of the Alphabet, and Aqiba 2 to the later one. Aleph: The gloss doctrina in Ambrose and Jerome 1 and 2 is straightforward: Hebrew ¹la alaph [learn, teach]. Aqiba 1 cites a text found in the Gemara commentary on Mishna, Shabbath XII.3, 104a: ‘Teach understanding, do good to the poor’.65 Jerome 2’s mille relates to Hebrew ¹la [thousand], with initial a aleph but commonly transliterated as eleph, Aram. alaph [thousand], cf. Aqiba 1, citing Deut. XXXII.30: ‘How should one (dja echad or achad) chase a [i.e. one] thousand (¹la dja achad eleph)?’66 Beth: Domus in Jerome 1 and 2 is straightforward: Hebrew tyb bayith [house]. Confusio in Ambrose is unclear. Müller points to the juxtaposition of Babylon uel Babel confusio and Bethel domus dei in Jerome’s etymologies of the cities’ names in Gen. XI.9 and Gen. XII.8, respectively.67 Babel in the well-known passage in Gen. XI.9 is related to Hebrew llb balal [confound]: ‘Therefore is the name of it called Babel; because the Lord did there confound the language of all the earth.’ Beth and Babylon are also neighbours in Jerome 2: Beth domus. Babylon confusio (cited from Ps. CXVIII.33 and 57, respectively). It is not at all certain, however, that Ambrose had access to Jerome, and in any case why should he have confused the two words? A more likely explanation is that Ambrose was aware of an interpretation of the b ta at-bash sequence, which in Aqiba 1 is explained as follows: attw ûyl[ wb ta at bosh alekh watata [(You,) shame upon you and below].68 The combination of the two letters b b and  sh here are clearly interpreted as wb bosh [be ashamed, be confounded], beside the noun tb bosheth [confusion, shame]. The association of beth with ‘confusion’ is confirmed in Aqiba 2, where it is explained that the letter is instrumental not only in building the house (tyb bayith) containing understanding (hnyb binah), but also in the creation of the world, as well as in the confusion of languages:

The Lord, praise be to him! said: ‘With beth (i.e. tybB be-beth) I created the world, ... with beth I confounded the language in the world. ...’ And how (do we know) that he created the world with beth? Because it is said (Gen. 1.1): ‘tyarb be-reshith [in the beginning], with b b God created in the beginning.’ ... He confounded the language with beth, how (do we know that)? Because it is said (Gen. 11.9): ‘because the Lord did there confound (llb balal) the language of all the earth.’69

64 As explained in note 9, the unnumbered source in Table 1, referred to there as Jerome 3?, Interpretatio alphabeti Hebraeorum, is a list appended to Jerome’s De situ et nominibus locorum hebraicorum liber, and is virtually identical to Jerome 2; so the same comments apply to it as to Jerome 2, except for the gloss simius against coph. 65 ‘[L]ehre Einsicht (Vernuft), erweise den Armen (Gutes).’ (Wünsche 1909: 186). 66 ‘Weil es heisst (Deut. 32, 30): ‘Wie kan einer (achad) verfolgen Tausend (achad elaph).’ (Wünsche 1909: 181). 67 Müller 1911: 13; cf. Lag. 3, 18. 68 Wünsche 1909: 192. Kellner (1893: 45, n. 2) suggests a derivation from Syriac beheth, Aramaic tjb, meaning perdefiri, confundi. 69 ‘Der Heilige, geb. s. er! sprach: Mit Bet (be-beth) schuf ich die Welt, ... mit Bet verwirrte ich die Sprache (Zunge) in der Welt. ... Und woher (entnemen wir) dass er mit Bet die Welt schuf? Weil es heisst (Gen. 1, 1): “Be-reshith, mit be schuf Gott anfangs.” ... Er verwirrte die Sprache (Zunge) mit Bet, woher (entnehmen wir das)?

44 Part 1: Sources of Name-forms and Meanings

Gimel: Müller70 refers Ambrose’s retributio to Aqiba 2’s interpretation of gimel: ‘What is gimel? This teaches that the Lord, praise be to Him!, said “I above all show mercy to (recompense) the poor; to the poor I above all show mercy.”’71 A more direct source is Ps. CXVIII.17: a0ntapo&doj tw=| dou&lw| sou, Vulgate: retribue servo tuo, for which the AV translation is: ‘Deal bountifully with thy servant...’. The Hebrew verb for a0ntapo&doj (Vulgate retribue) is lmg gamal, which is translated elsewhere as, for example, tribuit (Ps. XII.6), benefecit (CXIV.7), and retribuas (CXLI.7), all of which the AV renders as ‘deal bountifully’, cf. also the noun lwmg gemul, rendered retributio (Isa. XXXV.4, Isa. LIX.18, and LXVI.6), all of which the AV translates as ‘recompence’. Ambrose’s and Jerome 2’s retributio would appear to correspond more closely to the noun gemul [recompense], whereas Jerome 1’s plenitudo [abundance] appears to follow Eusebius’s plh&rwsij, which is more difficult to account for. It is possible that Eusebius linked the verb lmg gamal, Greek a0ntapo&doj, Latin retribue, with the Greek verb plhro&w [pay back in full, requite; fulfill; become full] and thence the Greek noun plh&rwsij [a filling up, becoming full; satisfaction] and Jerome’s plenitudo. How this association may have come about is discussed in connexion with Hesychius’s interpretation of gimel as thorax (see §1.1.5). Daleth: This name has already been cited above in the example given in Aqiba 1 in relation to the series beth, gimel, daleth. It is noteworthy that Ambrose transcribes the letter-name as daleth, whereas Jerome spells it deleth and Eusebius de&lq. It must be remembered that vowels were not indicated in any texts that Ambrose, Jerome or Eusebius may have been familiar with, so that there would have been no problem in spelling the name daleth or deleth and interpreting this as ‘door’. At the same time, as mentioned above, Eusebius may well have been influenced by the Greek de<oj [writing-tablet (in the triangular shape of the Greek letter delta ∆)] in his transcription of the name as de&lq. The meaning ‘writing-tablet’ as well as ‘door’ is supported by biblical usage of tld deleth (e.g. Jer. XXXVI.23: pagellas [leaves (of a scroll)]; Ezek. XLI.24: ostia [leaves (of a door]; and numerous passages with tld deleth translated as door or gate). However, the form with da- is the one presumably interpreted in Aqiba 1 to provide the associations between ld dal [poor], and ld dal [door]. Jerome 2 also gives the meanings pauper and ianua, which both relate to ld dal in spite of Jerome’s transcription, deleth. Jerome 2’s third gloss, tabulae, however, must go back to Jerome 1’s tabularum, which in turn relates to Eusebius’s de<wn. The clue to this derivation is provided by the case endings. Eusebius has put his de<oj in the genitive plural de<wn in order to give sense to his sequence of translations, and in this respect he is followed in Jerome 1’s connexiones with tabularum. In Jerome 2, the genitive is dropped but the gloss is kept in the plural, tabulae, as though

Weil es heisst (das. 11, 9): “Denn da verwirrte (balal) der Ewige die sprache der ganzen Erde”.’ (Wünsche 1909: 210–211). In the translations from Wünsche I follow his Bible references and note the Vulgate and AV numbering where these differ. 70 Müller 1911: 13–14. 71 ‘Was ist gimel? Das lehrt dass der Heilige, geb. sei er! sprach: Ich tue wohl zumal von den Liebeserweisungen den Armen; den Armen von den Liebeserweisungen zumal tue ich wohl.’ (Wünsche 1909: 214). I use ‘mercy’ in the translation in the sense of Latin misericordia.

45 A Family of Names deleth is to be taken as the plural of dal. Ambrose’s nativitas and timor are more problematic. Müller’s suggestion of an association between nativitas and tdl ledet [birth]72 is not convincing. Hebrew twdlwt toledoth [birth, generations], as in Ex. XXVIII.10 (Vulgate: nativitas), would perhaps be marginally closer to daleth. Wutz, pursuing the idea that Ambrose had before him a set of Greek translations of the Hebrew names, which he misinterpreted, suggests that Ambrose confused a form of the Greek verb e0nde&w, future -deh&sw [be in want of, deficient in] (cf. Hebrew lld, dalal [be impoverished, bring low]), with de&oj [fear, alarm], and the future dei&sw of the verb dei&dw [fear].73 A better solution, however, is provided by Aqiba 1,74 where rwd dor [generation] in the daleth verse 4 of the acrostic Ps. CXLIV is employed by the letter daleth to plead its case: ‘One generation shall praise thy works to another.’ What links nativitas and rwd dor with the first verse of the daleth octave of Ps. CXVIII is apparently the dust to which the text refers and from which human life is born; in his text (see Table 7), Ambrose specifically uses the phrase terrena nativitas [earthly birth]. His explanation linking nativitas and timor, however, is less clear: ‘For fear (timor) is born (nascitur) out of the corporeal and material. What is therefore earthly birth (terrena nativitas) if not fear (timor)? Hence in short the verse begins: “My soul cleaveth unto the dust”.’75 A connexion between fear and ld dal [poor] is discernible in Aqiba 2, which relates the poor to the broken hearted:

‘And when a poor man comes, I shall not turn him away empty-handed, for it is said (Ps. 74.21): “O let not the oppressed (ûd dak) return ashamed...” And I draw my Shechina nigh unto him every day, for it is said (Ps. 34.19): “The Lord is nigh unto them that are of a broken heart;”.’76 The description ‘of a broken heart’ is then said to refer to the poor. The Hebrew word for ‘broken’ in this context is rb shabar, and Ps. XXXIII.19 (= AV Ps. 34.18) continues: ‘and saveth such as be of a contrite spirit’, where ‘contrite’ is a translation of akd dikka. The verb with the root akd dka means ‘break to pieces, be bruised, smite, oppress, crush’, etc. A similar sense of ‘broken’ is reflected by the Hebrew ttj chatath [be beaten down, broken into pieces], but this word also covers the senses of ‘be afraid, make afraid, be dismayed’. The concepts that link these words would appear to be a literal breaking

72 Müller 1911: 14; tdl ledet > tld deleth, with metathesis of d and l. 73 Wutz 1914: 227. 74 Wünsche 1909: 179. 75 Kellner (1893: 45, n. 4) suggests that the link between ‘birth’ and the usual interpretation of deleth as ‘door’ lies in the idea that a door can be seen as symbolizing entry into life. 76 ‘Und wenn ein Armer kommt, so weise ich ihn nicht leer von mir zurück, wie es heisst (Ps. 74, 21): “Nicht weiche der Gedrückte (dak) beschämt zurück.” Und ich nähere meine Schechina zu ihm an jedem Tage, wie es heisst (Ps. 34, 19): “Nahe ist der Ewige denen, die gebrochenen Herzens sind.”’ (Wünsche 1909: 218). In the Vulgate numbering, the psalms are LXXIII and XXXIII. In the AV the reference is to Ps. 34.18, because the title of the psalm is not included in the numbering of the verses. Earlier The Alphabet links the letter deleth with ‘poor’ and ‘afflicted’ by quoting Ps. 18.28 (=27): “Dem armen (bedrückten) Volke hilfst du”. The AV translation reads: “For thou wilt save the afflicted people”. The Hebrew word translated as ‘afflicted’ is ani, which is elsewhere translated as ‘poor’.

46 Part 1: Sources of Name-forms and Meanings into pieces and a metaphorical breaking of the spirit. The question is: could Ambrose have been aware of such convoluted reasoning? It is only possible to speculate, but it is conceivable that a hint of these considerations is actually contained in the Latin translation of the beginning of Ps. CXVIII.25: Adhæsit pavimento anima mea. The Latin pavimentum, which the AV renders as ‘dust’, represents Hebrew aphar [dust, ashes, powder, clay]. Etymologically, the Latin word derives from the verb pavio [beat, strike, tread down], and is akin to Greek pai&w. The participle pavitum comes to be used as a noun, meaning ‘hard-beaten floor’, in the same sense as pavimentum. The dative form pavito (= pavimento in the biblical verse) is indistinguishable from the frequentative verb pavito [tremble, quake with fear], derived from paveo [be struck with fear], the etymology of which, according to Lewis and Short (s. v.), is dubious but is perhaps akin to pavio.77 I would suggest that Ambrose was capable of making this connexion between pavito [unto the dust] and pavito [quake with fear], possibly with an awareness of a rabbinic tradition that linked deleth as the plural of ld dal [door] with ld dal [poor], akd dka [break to pieces, be bruised, smite, oppress, crush] and ttj chatath [be beaten down, broken into pieces; be afraid, make afraid, be dismayed]. Heh: Jerome 1’s ista and Jerome 2’s ipsa and ista as well as Eusebius’s au3th are unequivocally explicable from Hebrew ayh hi [she, it, the same, that, this]. The feminine gender was chosen by both Eusebius and Jerome 1 in their connexiones for reasons of congruence with doctrina.78 In Jerome 2 the feminine forms, ipsa uel ista, were maintained. Once again, however, Ambrose is less easy to explain. It would appear that his est is derived from Hebrew hyh hayah [be], while vivo is from hyj chayah [sustain, keep alive], with an initial j cheth instead of h heh. Confusion between the graphic forms of h heh and j cheth would be perfectly understandable, especially also as the concepts of ‘living’ and ‘being’ are closely related. Ambrose justifies his glosses with the quite reasonable comment: qui enim est, vivit; et qui vivit, est. Jerome 2’s suscipiens could similarly be an attempt to reproduce Hebrew hyj chayah, in the sense of nourishing and preserving life. Vav, zayin and cheth: In Jerome 1, vav is straightforwardly interpreted as Hebrew w va, [and], but Jerome 2’s et ipse, Ambrose’s ille est et non alius or et ille, and Eusebius’s e)n au)th=| require explanation and are best considered in conjunction with the two subsequent letters, zai(n) and (h)eth. Eusebius transcribes vav as ou0au= and interprets it as e)n au)th=|, which is inexplicable unless taken in the context of his ‘running interpretation’ (as in Jerome’s connexiones). His first ‘sentence’ reads: ma/qhsij oi)/kou, plh/rwsij de/ltwn au(/thµαθησις [Teaching of the house (of God): an abundance of books (is) this], to which he could have added his second

77 In the at-bash formation of the Hebrew alphabet, daleth pairs with qoph to produce the ‘word’ daq [small, thin]. The verb daqaq means ‘beat small, beat into pieces, make dust, stamp (into powder)’. There are no connotations of ‘beating’ in the Greek equivalent of pavimentum in verse 25, i.e. e1dafoj [bottom, foundation; ground-floor, pavement; ground, soil]. 78 Oi)/kou in Eusebius is masc. gen. and domus in Jerome 1 is fem. gen.. The parsing of Jerome 1 is evident from Jerome’s interpretation of the pair doctrina domus as doctrina Ecclesiae, quae domus Dei est.

47 A Family of Names

‘sentence’ by means of a simple ‘and’, as Jerome does. But Eusebius creates a problem for himself by transcribing the next two letter-names as zai& and hq, and interpreting them as Greek zh= [live], and "o( zw=n [the living one] (Hebrew hyj chayah [to live, keep alive] or yj chai [living]). It is impossible to retrace his train of thought with certainty, but it is reasonable to assume that he was concerned to provide a sensible reading for his sequence of interpretations. Since ‘the living one’ forms the subject of his verb ‘live’, he may have felt the need to coordinate the two sentences more closely, and appears to have adapted the literal ‘and’ to a coordinating ‘where’, perhaps under the influence of the similarity of his transcription, ou)au=, with Greek ou[ [where]. He then appears to have taken the result – ‘an abundance of books is this, where the living one (God) lives’ – and made it more precise by re-interpreting ou[ as e)n au)th=| [in this], in which the feminine singular of the dative au)th=| refers back to ‘the house’ and not ‘books’. Jerome, in his connexiones (Jerome 1), has approached the problem quite differently. He begins his second ‘sentence’ with the previous letter, he, not with vav, and gives the reading:

Sentence 1: aleph beth gimel deleth doctrina domus plenitudo tabularum. Teaching of the house abundance of books. Sentence 2: he vav zai heth ista et haec vita. That and this life. Whereas ista [that] clearly refers back to the house of God, Jerome makes it clear in his comment on this second ‘sentence’ that ‘this life’ is intended by him to refer to Christ himself, ‘the life of the faithful’: ‘For how different life can be without knowledge of the Scriptures, through which indeed Christ himself is revealed, who is the life of believers?’79 Jerome’s haec relates to the interpretation of zai here as Hebrew taz zoth, the feminine of hz zeh, while vita relates tOo µyyj chayyim [life]. Jerome 2’s et ipse cannot be explained by straightforward reference to the Hebrew and may have been influenced by Ambrose’s et ille. Both ipse and ille are masculine nominative singular, and thus have no reference to any preceding noun. Ambrose explains his et ille or his alternative ille est et non alius80 by linking the gloss to the first verse of the vav octave (Ps. CXVIII.42): ‘Let thy mercies come also unto me, O Lord, even thy salvation, according to thy word.’81 The salvation called for here is, according to him, none other than the coming of Christ: ‘What is being called for is the coming of God’s salvation. It is he who is being called for, that is, the Lord Jesus, and none other. ... Indeed, not only the Father is being called for. It is he, that is, the Son too, who is being called for and having been called on comes and grants salvation to the world.’82 In other words, the word et of the Hebrew vav, or perhaps even the verbal subject affix -v,

79 Quae enim alia potest esse vita sine scientia Scriptuarum, per quas etiam ipse Christus agnoscitur, qui est vita credentium? (see Table 6). 80 Variant manuscript readings, as noted by Migne, are given in Table 7. 81 Et veniat super me misericordia tua, Domine, salutare tuum secundum eloquium tuum. 82 Rogatur ergo ut veniat salutare Dei. Ille est qui rogatur, hoc est, Dominus Jesus, et non alius, sicut habet litterae interpretation. ... Et val sic vel sic non solus Pater rogatur. Est [et] ille, hoc est, et Filius qui rogatur et rogatus advenit, et salutare mundo dedit.

48 Part 1: Sources of Name-forms and Meanings indicating ‘they’, has been construed by Ambrose as implying Father and Son and thereby used to make a theological point underlining the oneness of Father and Son in the salvation of the world, thus reflecting the debate in the early Church about whether the Holy Spirit proceeds from God alone or from God and the Son (Filioque). According to Müller, this Christian interpretation may go back to the reference in Aqiba 2 to Ex. III.15 in explaining the letter-name vav:83 ‘This is my name for ever, and this is my memorial unto all generations.’84 This quotation is followed by the comment that the expression ‘This is my name for ever’ refers to the ineffable name(s) of God, while ‘And this (hzw va-zeh) is my memorial unto all generations’ refers to the seals (rkz zeker [memorial], related to the verb rkz zakar [remember, imprint]) of the name. The expression ‘and this’ hzw (va-zeh) comprises the two letters vav and zayin in sequence, and recourse is had to the concept of seals and their opening in the subsequent explanation of the letter zayin.85 As pointed out above, Jerome 2’s interpretations of zayin are straightforward attempts to pin down Hebrew equivalents with sensible meanings. His transcription is zaith, but he gives three quite different interpretations: oliva (tyz zayith), fornicatio (hnz zanah) and haec (taz zoth). The last of these is the same as the syntactic interpretation in his connexiones (Jerome 1), while the first is perhaps related to Aqiba 2’s reference to the opening of seals and in particular to Zerubbabel (the annointed one) in connexion with zayin.86 This theme will be picked up again in the discussion of the rune-name sigel in Part 2. Müller tentatively links Ambrose’s interpretation of zayin, namely huc, to hz zeh, but is unable to explain duc te.87 Ambrose himself linked his interpretation to the sense of ‘Come hither’ or ‘Here!’, i.e. ‘Follow this way’, i.e. the right path. Perhaps he was simply referring to the deictic function of a pronoun like ‘this’ in pointing out and directing attention. Wutz suggests that the Greek list which Ambrose supposedly had before him glossed zayin as porneu&ein, as a translation of zanah [fornicatio], and that Ambrose misread this as poreu&ein [carry, convey].88 Jerome 2’s vita and vivacitas for the name he transcribes as eth are simple reproductions of his connexiones interpretation based on yj or tyj (Jerome 1), and is not too far distant from Eusebius, but once again Ambrose gives a quite different gloss. His pavor [fear], would appear to correspond to Hebrew ttj chatath [be afraid, be dismayed], tj chath [dismayed, broken, fear], htj chittah [terror], and tytj chittith [terror],89 but, as Müller points out, Ambrose refers to Ps. CXV.2 (= AV Ps. 116.11): Ego dixi in pavore meo, where pavor translates Hebrew zpj chaphaz, which the Vulgate translates as Ego dixi in excessu meo [D-R: I said in my excess. AV: I said in my haste].90

83 Müller 1911: 17. 84 Wünsche 1909: 224–5. 85 Wünsche 1909: 228. See Ambrose’s interpretation in Table 7. 86 See Wünsche 1909: 227–8. 87 Müller 1911: 17–18. 88 Wutz 1914: 226. 89 As Kellner points out (1893: 46, n. 3), Philo glossed chet as stupor. 90 Müller 1911: 18. In a footnote Müller, also refers to Philo’s Cheth = stupor [stupidity]. Jerome glosses Chet, the name of the ancestor of the Hittites at Gen. XX.3, as percutiens [smiting]. See also Kellner (1893: 46, fn.3).

49 A Family of Names

Teth and yod: Eusebius, Jerome 1 and Jerome 2 all agree on glossing th/q (Eusebius), tet (Jerome 1) and teth (Jerome 2) as ‘good’, i.e. kalh/ (Eusebius) and bonum (Jerome 1 and 2), where the case endings all agree with the subsequent glosses on iw&q (Eusebius), iod (Jerome 1) and ioth (Jerome 2) as ‘the first cause or principle’, namely a)rxh/ (Eusebius) and principium (Jerome 1 and 2). The Hebrew for ‘good’ would be the bwf tob which occurs throughout the teth octave of Ps. CXVIII, but ‘the first principle’ is less clear. Müller suggests that the interpretation of yod derives from the name of Yaveh.91 He offers no corroboration, but significantly in Aqiba 1 the letter yod pleads its cause by referring to ‘Yah (hy, i.e. Yaveh, Jehovah), the everlasting’, as in Isa. XXVI.4.92 The same passage goes on to say that ‘With me is the beginning of your name as the only one (dyhy yachid)’. In a footnote, Müller suggests that Jerome 2’s dominator and scientia for ioth may have the same source as principium,93 but a more likely derivation is from dy yad and [dy yada, among whose translations are ‘dominion, power, knowledge’ and ‘have knowledge’, respectively. Ambrose’s exclusio for teth is obscure, but his reference to 2 Reg. (= 2 Sam.) XII.13: Et Dominus abstulit peccatum tuum [The Lord also hath put away thy sin] suggests a link with Hebrew lwf thul [cast forth], even though the Hebrew word for ‘put away’ in the equivalent Hebrew passage is rb[ abar. His desolatio for yod might derive from dyjy yachid, which is translated as ‘desolate, solitary’ as well as ‘only (child or son)’, but other candidates would be ay yaash [despair] and ÷wmyy yeshimon [desert, solitary wilderness].94 As Müller suggests, his confessio could be linked to hdy yadah [confess] and hdy yadah [cast out]], where the act of confession is symbolized by ‘casting out’ one’s hand, dy yad. Kaph: Eusebius transcribes kaph as xa&f and interprets it as "o(/mwj [all the same, nevertheless], probably based on Hebrew yk ki [although, but, nevertheless], whereas Jerome stays with the straightforward interpretation of ¹k kaph: Jerome 1 manus; Jerome 2 manus palma uel uola, although the case ending of palma is not clear.95 Ambrose’s curvati sunt reproduces Hebrew ¹pk kaphaph [bow down, curve], as is clear from his comments on humility, which end with: Nec si flectas ut circulum collum tuum, in reference to Isa. LVIII.5, which the Vulgate translation renders as: nunquid contorquere quasi circulum caput suum;96 the Douay-Rheims translation reads: ‘is this it, to wind his head about like a circle...?’, whereas the AV version – ‘is it to bow down his head as a bulrush...?’ – is closer to the agmon אְגDespiteexplanations,therefore,kenningsandglossesonninremainenigmatic.Thenamecertainlyappearstobetermforforkaweaver’sbeam,whichwasraisedintimespeace;itisalsosaidA4,5ash-tree.But,giventreebiasA4,5,difficultbelievethatbeamgotitsfromAndwhatconnexionwith‘Boastbeauty’women’?Waswoman’sroleweaving?Werewomenweavingbeautiful?Why,anycase,suchanesotericobjectchosenogamcharacter?WhymorecommonbeginningNnotchosen?choi agmonas‘bulrush.thewordetoftheHebrewvav,orperhapsevenverbalsubjectaffix-v,indicating‘they’,hasbeenconstruedby䭎低乗ݨ
Ψ ןֹומו ן Hebrew in translating the verb ¹pk kaphaph as ‘bow down’ and as ‘bulrush.

91 Müller 1911: 19. 92 Wünsche 1909: 176. 93 Müller 1911: 19. 94 Kellner (1893: 46) notes that the yod verse of Lam. I describes the laying-waste of Jerusalem. 95 The gloss palpa in column 5 may be a mistake for palpus, ‘soft palm of the hand’, or possibly a corruption due to the influence of palma. 96 Cf. the LXX version; οὐ δ' ἂ ν κ ά μψ ῃ ς ὡ ς κρ ί κον τ ὸ ν τρ ά χηλ ό ν σου [nor even should you bend as a ring your neck.]

50 Part 1: Sources of Name-forms and Meanings

Lamed: Jerome 2’s doctrina uel disciplina is related to Hebrew dml lamad [learn, teach], although the noun forms would correspond more closely in meaning with jql leqach [doctrine, learning]. The verb form is more precisely reproduced in Eusebius’s interpretation of his la&bd as ma&qe [learn], giving the xa&f la&bd pair with the meaning ‘nevertheless learn!’ It is possible that Jerome 2’s noun forms were carried over from Jerome 1’s forms which in connexion with the manus for caph required a genitive: disciplinae uel cordis. The word cordis here, like Ambrosius’s cor, is derived from Hebrew bl leb [heart], and is confirmed in Aqiba 2: ‘Do not read dml lamed but h[d ÷ybm bl leb mebin deah [my heart understands knowledge]’,97 which in turn accords with the Hebrew rendering of ‘the mind, intellect’ by bl leb [heart]. It is in this sense that Jerome 1’s fourth connexio is to be understood: ‘The hand is perceived in terms of action, the heart and understanding are interpreted in terms of the capacity for sensing (i.e. feeling and understanding); we cannot do anything unless we know beforehand what is to be done.’98 Ambrose’s servo is seen by Müller to be directly linked to the sense of ‘heart’ in bl leb,99 but it may be a reproduction of Hebrew db[l l-ebed [belonging to a servant] (cf. the verb db[ abad [to be servants, to serve]). Mem: Eusebius’s interpretation of mh&m as e)c au)tw=n [from them] corresponds with Jerome 1 and 2’s ex ipsis, and would appear to be a rendering of Hebrew µhm me-hem [from them, from which], whence also Jerome 2’s ex quo. Ambrose also cites ex ipsis, but as an interpretation given by ‘others’. His prime interpretation is viscera [inner parts, internal organs], from Hebrew µy[m meim [bowels], which he uses for his own didactic ends in explaining charity as proceeding from our innermost being (see Table 7).100 Jerome 2’s aqua is from Hebrew µym mayim [water]. Nun: Ambrose’s pascua eorum is most likely derived from Hebrew ÷whwn navehon [their fold, sheepcote] (cf. also: han naah [pasture]). His unicus, however, is more controversial and for once appears to be related to the interpretations offered by Eusebius and Jerome. Eusebius’s ai)wni/"a for nou=n and Jerome 1 and 2’s sempiternum for nun possibly derive from the ÷wn nun as used in Ps LXXI.17 in the sense of permanet [shall be continued, shall persist], although this is the only passage in the whole Bible where ÷wn nun is found with this meaning. Another possible source is the name of the father of Joshua, ÷wn Nun (cf. Ex. XXXIII.11), which has been glossed as ‘continuation’;101 unfortunately, however, Jerome does not gloss this name at all. A verse in the nun octave (Ps. CXVIII.111) contains an expression for ‘for ever’ (Vulgate: in

97 Wünsche 1909: 240. 98 Caph Lamed: manus disciplinae siue cordis. Manus intelliguntur in opere, cor et disciplina interpretantur [al. intelliguntur] in sensu; quia nihil facere possimus, nisi prius quae facienda sunt scierimus 99 Müller 1911: 20. 100 This interpretation is in line with Hebrew usage, where the bowels represent compassion, as when the prophet addresses God with the words: ‘the sounding of thy bowels and of thy mercies towards me’, Isa. LXIII.15 The variant readings in the Ambrosian manuscripts, noted by Migne (see Table 7), all have the same intent, including ignis ex ultimis. 101 E.g. Young (no date).

51 A Family of Names aeternum), but the Hebrew word used is not ÷wn nun but µlw[ olam [ever(lasting)]. An alternative word with an initial letter nun which has the sense ‘for ever’ is jxn netsach (cf. jxn al lo netsach [never, not to perpetuity]), but a more complex and perhaps more satisfactory explanation may be gathered from Ambrose’s gloss of unicus and his explanation that this refers to Jesus, the only Son of God:

These letters of the Hebrews bear witness to the Lord Jesus being the only Son of the Father, the Word of God. And thus, in the first verse, David says of the only Son of God: Thy word is a lamp unto my feet, and a light unto my path.102 The Hebrew for ‘lamp’ in this context is rn ner,103 and the theme is pursued by Jerome 1 in the explanation of the fifth connexiones:

The fifth connexio is Mem Nun Samech: from these [there is] everlasting support. There is no need of an explanation, but it is more evident in every light that everlasting support is provided from the scriptures.104 The link between ‘the (only) son’ and the concept of ‘everlasting’ is found in Ps. LXXI (= AV Ps. LXXII), of which Barton and Muddiman say:

[This psalm] was treated as messianic in Jewish and early Christian tradition, the Targum paraphrasing [...] v. 17 as “His name shall be remembered for ever; and before the sun existed his name was prepared; and all the people shall be blessed in his merits”. It is never quoted in the NT, however, though at an early period it was adopted as the special psalm for Epiphany.105 Müller points to the Septuagint’s translation of this verse: e1stw to_ o1noma au0tou= eu0loghme&non ei0j tou0j ai0w=naj, pro_ tou= diamenei= (Vulgate permanet) to_ o1noma au0tou=, from which, he says, Eusebius’s ai)wni/"a and Jerome’s sempiternum are explicable.106 Hence it would appear that both Ambrose and Jerome perceived ÷wn nun as referring to the Word, the only Son of God, the everlasting Light. Jerome 2’s foetus is possibly linked to his glosses in relation to the name of the city of Nineveh throughout his Liber interpretationis hebraicorum

102 Ecce ipse [l. ipsae] Hebraeorum litterae testificantur Dominum Jesum esse unicum Patris Filium, Verbum Dei. Denique primo statim versu David de unico Filio Deo dicit: Lucerna pedibus meis Verbum tuum, et lumen semitis mei. 103 In Aqiba 1 (Wünsche 1909: 174), the letter nun also refers to ner [light], in pleading his case, by citing Prov. XX.27: ‘The spirit of man is the candle (ner) of the Lord’ (cf. the same citation in Aqiba 2, Wünsche 1909: 246). 104 Quinta connexio est Mem Nun Samech: ex ipsis sempiternum adjutorium. Hoc explanatione non indiget, sed omni luce manifestius est, ex Scripturis aeterna subsidia ministrari. 105 Barton and Muddiman 2001: 387. The AV translates Ps. LXXII.17 as: ‘His name shall endure for ever: his name shall be continued (nun) as long as the sun: and men shall be blessed in him: all nations shall call him blessed’ with the alternative translation for the second clause: ‘shall be as a son to continue his father’s name for ever.’ The Vulgate translation reads: Sit nomen ejus benedictum in sæcula; ante solem permanet nomen ejus. Et benedicentur in ipso omnes tribus terræ; omnes gentes magnificabunt eum. [D-R: Let his name be blessed for evermore : his name continueth before the sun. And in him shall all the tribes of the earth be blessed: all nations shall magnify him.] 106 Müller 1911: 22. Müller links the first clause of v. 17 to: ‘janîn oder jinnon “sprossen, Sprossen treiben” ..., womit man wohl auch nin “Sproß, Sohn” zusammenstellen kann’.

52 Part 1: Sources of Name-forms and Meanings

nominum. The relevant lemmata are:

Transcription Gloss Biblical ref. Ref. in Lag. AV form Niniue pulchra uel germen pulchritudinis Gen. X.11 9, 4 Nineveh Naue semen uel germen uel pulchritudo Ex. XXXIII.11 14, 11 Nun Nuath pulchritudo 1 Sam. XIX.18 36, 12 Nai-oth Niniueh foeta siue germen pulchritudinis 2 Ki. XIX.36 46, 25 Nineveh uel speciosa Nun foetus uel piscis siue sempiternum Ps. CXVIII.105 48, 26 (Nun) Niniue foeta uel germen pulchritudinis Isa. XXXVII.37 50, 18 Nineveh aut speciosa Niniue speciosa Nah. I.1 52, 13 Niniveh Naa pulchra Lc. VII.11 65, 16 Nain Niniuitae nativitas pulchritudinis aut speciosi Lc. XI.30 65, 17 Ninevites

Apart from the remarkable transcription Naue by Jerome, where the Vulgate and AV form for the father of Joshua is Nun, this summary shows that Jerome associated the name of the city of Nineveh with two basic concepts: ‘beauty’, i.e. pulchritudo, and ‘offshoot, issue, shoot, scion’, i.e. germen, semen, foetus – the city attracting feminine forms (pulchra, speciosa, foeta), and the nun of the psalm masculine forms (foetus). The coupling of semen and foetus with germen [sprig, offshoot], moreover, indicates that the literal meanings of semen [seed], and foetus [foetus], are not intended. The concept of beauty appears to have been taken by Jerome from homonyms of han naah [pasture] and hwn naveh [sheepcote] – on which Ambrose’s pascua eorum is probably based – namely han naah [be beautiful, be comely] and hwn naveh [comely, seemly, becoming, pleasant]. The concept of ‘offspring, issue, shoot’ could have derived from ÷yn nin, Latin progenies [offspring, progeny], but there are two other possibilities. The first is that Jerome associated the rn ner [light, lamp, candle] of the rabbinical texts and the first verse of the nun octave in Ps. CXVIII with r[n naar [child] (the son of Nun at Ex. XXXIII.11 is referred to as r[n naar [young (growing) man]). The second possibility is that he associated the concept of ÷wn nun [be continued, continuation] with jxn netsach [always, ever], and thence with rxn netser [branch, shoot], with the connotations of Jesus, scion of Abraham and David, the only son of God, the everlasting Word (cf. Isa. XI.1: ‘And there shall come forth a rod out of Jesse, and a Branch (rxn netser) shall grow out of his roots.’).107 Jerome 2’s piscis is more difficult to explain. Speculatively, piscis might be a Christian interpretation which associated Jesus, the only Son of God, with the acronym formed from the Greek word i0xqu=j [fish]: Ihsou&j Xristo&j Qeou= Ui9o_j Swth&r [Jesus Christ, the Son of God, the Saviour]. However, there is also a possible link with Hebrew r[n naar [young man] in its more literal sense of ‘growing’. The usual Hebrew for ‘fish’ is gd dag or hgd dagah, but the verb hgd dagah is used at Gen. XLVIII. 16 in the sense of ‘grow (like fish)’, precisely at the point where Jacob blesses the sons (seed) of Joseph: et crescant in multitudinem super terram [let them grow into a

107 In the acrostic Ps. CXLIV, the nun verse is missing from the Masoretic text and is not included in the English AV translation. It is, however, included in the Qumran scroll and LXX and Syriac versions (Barton and Muddiman 2001: 403–4). In the New English Bible, the verse is translated as the opening lines of verse 14: ‘In all his promises the Lord keeps faith, and he is unchanging in all his works.’ The Hebrew for ‘keep’ here is natsar which, without diacritics, would read precisely the same as netser [branch].

53 A Family of Names multitude in the midst of the earth]. This concept is in turn reminiscent of the Christian use of pisciculi [little fish] as a term for those (newly) baptized in the piscina [font]. At Is. XIX. 10, fish in a pond are equated with living souls as a result of the Vulgate’s translation of Hebrew pn nephesh [living (breathing) creature; soul] as pisces: Et erunt irrigua eius flacentia, omnes qui faciebant lacunas ad capiendos pisces. [D-R translation: And its watery places shall be dry, all they shall mourn that made pools to take fishes. AV translation: And they shall be broken in the purposes thereof, all that make sluices and ponds for fish.]. In fact, The Alphabet text makes a direct connexion between the letter nun and a person’s soul:

Then ... spoke [nun]: Lord Almighty, would that you would create the world with me, for with me you once gave spirit to the people you created, and it was called a light, as is written (Prov. XX. 27): ‘The spirit of man is the candle of the Lord’.108 The Hebrew word for ‘spirit, soul’ here is not pn nephesh but another word with initial nun, hmn neshamah; significantly, however, the soul here is referred to as a rn ner [candle]. Although Jerome is unlikely to have been familiar with this passage as such, but he shares with it the concepts that led to his association of pn nephesh with pisces in the Vulgate translation and his gloss of piscis on nun.

Samech: Eusebius’s boh&qeia [aid, support] for σαµχ and Jerome 1 and 2’s adiutorium [aid, support], for samech are in unison, although Jerome 2’s use of quidam ... putant [some think ...] to bracket his glosses of erectionem uel adiutorium siue fulturam [setting up or support or prop, support] suggests that these interpretations might not be Jerome’s own first choices. Also, adiutorium in this context might be perceived in more material terms than Eusebius’s boh&qeia. On the other hand, in his connexiones (Jerome 1) Jerome clearly uses adiutorium as spiritual and intellectual assistance from the Scriptures. His first choice in Jerome 2 would appear to be the same as the second of the two glosses that Ambrose cites as given by ‘others’, i.e. firmamentum [a strengthening, support, prop; the firmament], which extends the idea of support to the concept of the sky fixed and supported above the earth. The concept of ‘support’ has most likely been derived from Hebrew ûms samak [uphold, sustain, be borne up], while firmamentum would appear to refer to µym shamayim [the heavens] (spelled with an initial shin).109 Ambrose’s involvement of a word with an initial shin rather than samech is confirmed by his gloss audi [listen, hear], which in Hebrew would be [m shama, and, as elsewhere, he goes to considerable lengths to make a connexion between the heavens and the act of listening (see Table 7).110

108 ‘Darauf ... sprach [nun]: Herr der Welt! möchte es dein Wille sein, dass du mit mir deine Welt erschaffst, denn mit mir gibst du dereinst die Seele den Geschöpfen (den Menschen) und sie wird eine Leuchte genannt, wie es heisst (Prov. 20, 27): “Eine Leuchte des Ewigen ist des Menschen Seele (hmn)”.’ [Vulgate: Lucerna Domini spiraculum hominis] (Wünsche 1909: 174). 109 Kellner (1893: 47, n. 4) cites ûms samak [firmamentum, fulcimentum] from Buxtorfii Lexicon Chaldaicum (1875: 751). The lead word in Ps. CXVIII.116 is ûms, samak, which the Vulgate translates as suscipe. The lead word in v. 117 is d[s, saad, which is translated in the Vulgate as adiuva. 110 Müller (1911: 23) points out that the linking of samech and shamayim is

54 Part 1: Sources of Name-forms and Meanings

Ayin: All three authors agree on the glosses of oculus [eye] or fons [fountain] for Hebrew ÷y[ ayin.111 Peh: Eusebius and Jerome 1 and 2 agree on the sense of Hebrew hp peh, namely [mouth]. There are, however, some notable differences. Eusebius and Jerome 1 agree on the transcription: Eusebius fe=, Jerome 1 phe, but Jerome 2 gives the Hebrew letter as fe and treats it as equivalent to Latin F. Jerome has to make clear to his Latin readers that os is ab ore, i.e. from os [mouth], and not ab osse, i.e. from os [bone]. Eusebius has no such problems with the Greek for ‘mouth’, i.e. sto/ma, but his sixth connexio leaves sto/ma in the nominative, phgh\ h)\ kai\ o)fqalmo\j kai\ sto/ma dikaiosu/nhj [source, or eye, and mouth of justice], whereas Jerome’s connexio puts os in the genitive along with justice, fons siue oculus oris justitiae [source, or eye, of the mouth of justice]. Jerome 2’s two other interpretations, laqueus [noose, snare, gin, trap] and decipula [snare, gin, trap], have nothing to do with Hebrew hp peh, but relate to jp pach [gin, snare]. Similarly, Ambrose’s erravi [I have gone astray, wandered from the truth], is not related to hp peh, while his os aperui [I have opened my mouth], would seem to contain more information than simply a connexion between hp peh and os. Müller points to Ps. CXVIII.131 as an explanation for os aperui [I opened my mouth,...], but admits that he is at a loss to explain erravi, which he thinks is verdächtig [suspect], and more at home in a gloss on tav (on which see below).112 However, although he quotes the Hebrew for CXVIII.131 (jtp hp peh pathach[I opened my mouth]), he fails to recognize that the acrostic Ps. CXLIV.16 has jtp pathach as the word with initial peh, in the sense of ‘open’: ‘Thou openest thy hand’, and that a near homonym, htp pathah [deceive, be deceived] is very close indeed to Latin erravi, as much as the h[t taah [err, go astray, be deceived], which is picked up by Ambrose in his glosses on tav (see below). Tsadeh: Jerome 1 and 2’s iustitia [justice] agrees with Eusebius’s dikaiosu/nhj, and in both Eusebius’s interpretations and Jerome 1’s connexiones the gloss is in the genitive. The Hebrew word here is qdx tsedeq [justice, righteousness]. The Hebrew words on which Jerome 2’s other glosses are based are regio: hd sadeh [field, country] (spelled with an initial sin) and venatio: dyx tsayid [hunting, venison], dwx tsud [hunt], while Ambrose’s consolatio [consolation, comfort] would appear to derive from d[s saad [comfort] (spelled with an initial samech).113 The involvement of words with initial sibilants other than tsadeh reflects the difficulty of coping with the different Hebrew sibilants in Latin. Notable in this context, in anticipation of the discussion of ogam-names, is Ambrose’s citation of Ps. CXVIII.140: Ignitum eloquium tuum nimis, et servus tuus dilexit illud (Vulgate: Ignitum eloquium tuum vehementer, et servus tuus dilexit illud [Thy word is very pure (i.e. ¹rx tsaraph [tried or confirmed in Aqiba 2 (Wünsche 1909: 250), but he does not explore the link with shama. 111 The same interpretation is given in Aqiba 2 (Wünsche 1909: 254). 112 Müller 1911: 24. 113 Müller (1911: 24) points to Aqiba 2’s comment: ‘Don’t read tsade but tsedeq’. He is only able to explain Ambrose’s consolatio by construing the first verse in the tsade octave as providing consolation or comforting: ‘Righteous art thou, O Lord, and upright are thy judgments’.

55 A Family of Names refined by fire]), therefore thy servant loveth it]). Qoph: Eusebius transcribes this letter as kw&f, Ambrose as koph, Jerome 1 as coph, and Jerome 2 as cof. None of the glosses provided by Eusebius, Ambrose or Jerome, however, are a direct translation of Hebrew qoph. The only commentator to do so is the compiler of the list labelled in Table 1 Jerome 3?, who gives the ‘correct’ simius, ‘ape’, which is not found in any other list.114 Eusebius’s gloss, klh=sij [a call, invocation, appellation, invitation, legal summons] would appear to have been taken over in both Jerome 1 and 2’s vocatio [a summons, calling]. Müller relates this gloss to Hebrew lwq qol [voice].115 Jerome’s explanation of his connexio does indeed refer to vox [voice], but he does not gloss coph with vox. Instead, it appears that he seems to have preferred Eusebius’s association with Hebrew arwq qara [to call, call upon, be called], with its connotation of being called to Christ, the head of ‘all’, caput omnium. Ambrose fails to give an equivalent of either lwq qol or arwq qara as a gloss, but he does make a point of quoting the first verse of the qoph octave: Exclamavi in toto corde meo; exaudi me, Domine. [I cried (lwq qara) with my whole heart; hear me O Lord]. Jerome 2’s auis [bird] has already been referred to above in the discussion of Jerome’s problems with transcription, the initial ayin in the Hebrew ¹w[ ’oph being reproduced in Latin with a gutteral. Jerome 2’s excussio and Ambrose’s conclusio, as well as Ambrose’s aspice, are more difficult to account for. Müller suggests that Ambrose’s conclusio does not mean ‘conclusion’ in a logical sense, but ‘enclose, encompass’, as indicated in Ambrose’s comment qui concluditur circumspicere se debet [who is compassed about, needs to look about].116 This interpretation, Müller concludes, agrees with the explanation given in Aqiba 2:

Koph. That is Moses, father of the wise, father of all prophets who brought together (¹yqh hiqiph) before Pharaoh all words of wisdom and all words of reason and all words of good sense and insight ...117 Hebrew ¹yqh, hiqiph, can, in Müller’s opinion, be just as suitably translated by the passive ‘encompassed, enclosed’ as by the active ‘encompass, enclose’. In other words:

Not only the ‘encompassed’ but also the ‘one who encompasses’ should be on the watch out, Ambrosius possibly having confused active and passive.118 This commentary is helpful as far as it goes, but there remain at least two serious difficulties. Firstly, why is qoph associated with Moses? And secondly, did Ambrose really confuse active and passive? For Moses’s association with qoph, it is necessary to consult the Gemara

114 See Thiel 1973: 87. See also note 9. 115 Müller 1911: 25. 116 Müller 1911: 25. 117 Koph. Das ist Mose, der Vater der Weisen, der Vater aller Propheten, der vor Pharao alle Worte der Weisheit und alle Worte der Vernunft und alle Worte der Klugheit und der Einsicht ... zusammen reihte (Wünsche 1909: 260). 118 Nicht nur der ‘Eingeschlossene’, sondern auch der ‘Umschließende’ muß Ausschau halten, wobei Ambrosius Aktiv und Passiv verwechselt haben kan (Müller 1911: 25).

56 Part 1: Sources of Name-forms and Meanings commentary on the Shabbath text119 and other passages in The Alphabet. The Gemara commentary on Shabbath XII.3 interprets qoph as ‘holy one’, and contrasts this with its interpretation of resh, namely ‘wicked one’. ‘Holy’ in Hebrew is dq qodesh and ‘wicked’ is [r rasha. It is not enough, however, to link Moses to qoph simply on account of the association with holiness. Aqiba 1 asks of the appearance of the letter q why it is tall and why it has a horn.120 The answer is that the horns of the wicked will be cut off, on account of their pride, but not those of the righteous, as it says in Ps. LXXIV.10: et omnia cornua peccatorum confringam, et exaltabuntur cornua iusti (AV: All the horns of the wicked also will I cut off; but the horns of the righteous shall be exalted). The Hebrew for ‘horn’ is ÷rq qeren. When Moses descends from Mount Sinai it is said, in the Vulgate translation of Ex. XXXIV.29 reads: et ignorabat quod cornuta esset facies sua, which the Douay-Rheims translation renders as: ‘and he knew not that his face was horned’, whereas the AV translation renders the Hebrew more accurately as ‘Moses wist not that the skin of his face shone’. The Hebrew for ‘shone’ here is ÷rq qaran, although it is also used in Ps. LXVIII.32 (AV: 31) to mean ‘horns’. 121 The Vulgate translation for Moses’s shining face as facies cornuta had the consequence that medieval and Renaissance representations of Moses showed him with horns. The Aqiba 1 text gives a further exegesis on qoph which expands on the Shabbath Gemara text, where it is asked why the holy one (q qoph, dq qodesh) turns his head away from the wicked one (r resh, [r rasha). The holy one in the Gemara text says that he is unable to look on the image of the wicked one. Aqiba 1 follows this assertion up with a quotation from 4 Reg. III (2 Chron.).14: si non vultum Iosaphat regis Iudae erubesurem, non attendissem quidem te, nec respexissem [D-R: if I did not reverence the face of Josaphat king of Juda, I would not have hearkened to thee, nor looked on thee. AV: Were it not that I regard the presence of Jehoshaphat the king of Judah, I would not look toward thee, nor see thee].122 The Hebrew for ‘regard’ here is an nasa, for ‘see’ har raah, and for ‘look’ fbn nabat, none of which has an initial q qoph. On the other hand, the Shabbath Gemara and The Alphabet texts establish a link with ‘looking’, whence Ambrose may have drawn his aspice (cf. also bq qashab [regard, give attention to]). As for Ambrose having confused active ‘enclose’ and passive ‘enclosed’, it is more likely that, like Jerome with his excussio, he was not considering ¹yqh hiqiph at all, but the word translated in the Vulgate at Isa. XVII.6 by excussio oleae [the shaking of an olive tree], namely ¹qn noqeph (with an n- preformative in place of an h-), which in turn is related to ¹qn naqaph [enclose, encompass, strike, knock, bruise].123 At the same time, it is

119 See above, footnote 18. 120 Wünsche 1909: 185. 121 This confusion of qeren and qaran might be compared with the possible confusion of netser and natsar suggested above. 122 Wünsche 1909: 185. 123 In Zech. IV.11–14, for example, the olive tree is taken as an image of the chosen people of God, on account of its provision of oil for annointing the chosen. In Isa. XVII.6, but more especially in Isa. XXIV.13–15, the image of the shaking of the olive tree is that of the harvesting of olives, the chosen ones. The description of Moses and Aaron before Pharaoh in The Alphabet, however, would appear to relate more to the ‘shaking’ of Pharaoh and his court: ‘... and there fell upon them [Pharaoh and his court] fear and trembling, fright and shaking and horror, and they

57 A Family of Names notable that Eusebius’s klh=sij (< klei&w, kale&w [call]) has a homonym klh=sij (< klei&w [shut up, shut in, enclose]). It is conceivable that it may have been this homonym that influenced Ambrose when he interpreted his koph as conclusio. Resh: The interpretation of resh as meaning ‘head’ (Hebrew ar rosh, Aramaic resh) is found in Jerome and Eusebius as well as in Ambrose, the transcription of the letter-name being rh&j in Eusebius and res in Ambrose and Jerome. Also, both Eusebius and Jerome 1 give their interpretation in the genitive, to fit their turn of phrase: kefalh=j and capitis. Ambrose’s alternative gloss, primatus, relates to another principal meaning of ar rosh, resh, namely ‘chief, captain’. Shin/sin: Again Eusebius’s and Jerome’s ‘teeth’ as an interpretation of this letter-name are uncontroversial, with again the turn of phrase chosen by Eusebius and Jerome 1 requiring the genitive: o)do/ntwn and dentium. The transcriptions are: Eusebius se&n, Jerome 1 and 2 sen and sin (Hebrew ÷). For his shin, Ambrose gives super vulnus, which is difficult to explain. Müller quotes from Aqiba 2: ‘Shin, that is the teeth of the thoroughly wicked, which the Lord will break thrice over’124 and refers to Ps. III.8: dentes peccatorum contrivisti (AV, v.7: thou hast broken the teeth of the ungodly). He then suggests that the n shen in the expression hrb µy[r yn sheni rashaim shabberah [has broke the teeth of the ungodly] be equated with vulnus and that rb shabber could be construed as a phonetic equivalent of Latin super. Such speculation, however, smacks of desparation. In fact, Ambrose is quite explicit when he interprets the first verse of the shin octave (Ps. CXVII.162; see Table 7) as ‘Princes have persecuted me without a cause’ – vulnus est – and the second verse as ‘I rejoice at thy word, as one that findeth great spoil’ – super vulnus est, quia verbis Dominicis vulneris dura curantur [for the hardships of a wound are cured by the Lord’s words]. The Hebrew word for ‘spoil’ here is ll shalal and for ‘rejoice’ it is y sis. But from Ambrose’s opening words in this homily it would appear that he is thinking in particular of the word for ‘oil, ointment’, i.e. ÷m shemen, since it is oil that is poured into the wound and it is the wine and oil of Jesus Christ, the annointed, that are poured into our wounds:

Medicine therefore is super vulnus: for the Lord Jesus himself is the doctor who heals our wounds, pouring (over them) wine and oil, and in like manner binding the wounds of him who, coming down from Jerusalem, was injured by thieves.125 Tav: Jerome follows Eusebius in transcribing this letter-name with an aspirate: thau and qau=, and Jerome 1 also follows Eusebius in putting the gloss in the plural: signa and shmei="a. Their interpretation, ‘sign(s)’, is a straightforward translation of Hebrew wt tav. Jerome 2’s subter translates threw their pens from their hands and their parchment from their shoulder-bags, and fell on their faces before Moses and Aaron and threw themselves down before them, as is said (Ps. CIV.38): “Egypt was glad when they departed: for the fear of them fell upon them”.’ 124 Müller 1911: 26: ‘Shin, das sind die Zähne der vollendeten Freveler, welche der Heilige, geb. sei er! dereinst dreimal zerbrechen wird (Wünsche 1909: 265).’ 125 Medicamentum igitur super vulnus est; quia Dominus Jesus ipse medicus, qui vulnera nostra curavit, infundens vinum et oleum, et alligans vulnera adae illius, qui descendens a Jerusalem, a latronibus vulneratus est.

58 Part 1: Sources of Name-forms and Meanings

Hebrew tjt tachath [under, beneath]. Ambrose’s consumavit probably relates to µmt tamam [be consumed, be finished, have done], and µymt tamim [perfect, complete] (cf. also tylkt taklith [end, perfection]), in an allusion to the termination of the psalm, but it could also allude to tav as a sign of the tau-cross and Christ’s final cry (Joh. XIX: 30): Consummatum est. His erravit, on the other hand, refers to h[t taah [err, go astray], as in the very last verse of the psalm: ‘I have gone astray like a lost sheep’. At the same time, there is Jerome’s gloss on the name Thou (Hebrew w[t or y[t) as amens uel errans [out of one’s senses or wandering].126 1.1.5 Hesychius A set of glosses on the Hebrew alphabet in Greek which was for a long time attributed to Athanasius, archbishop of Alexandria (from 328 to 373), but is now accepted as being by Hesychius of Jerusalem (d. 450), shows correspondences with those of Eusebius, Jerome and Ambrose, but diverges sufficiently to indicate that a number of them were drawn from quite different sources. They are found in Hesychius’s commentary on the acrostic Ps. CXLIV (CXLV), De titulis Psalmorum (see Table 10 (a)).127 His comments are not always transparent, but one gloss in particular throws light on the patristic interpretation of gimel, and is of relevance to the way the ogam-name (n)getal was derived and interpreted (see §2.3.3). Hesychius interpreted gamel as thorax, which he put forward as referring to Paul’s breastplate metaphor:

interpretatur enim thorax, quod est Dominus noster, quemmodum dixit Apostolus. The association with the Hebrew letter gimel, however, has more to do with the specific sense of thorax as ‘breastwork of a wall; parapet’, which compares with the Latin equivalent lorica in its sense of ‘fence, hedge, enclosure’. In Hebrew, the noun for ‘fence, hedge, wall’ is rdg gader and the verb meaning ‘wall up, hedge, inclose’ is rdg gadar. In the acrostic Lam. III, which devotes three verses to each letter, rdg gadar is the word with initial gimel in verses 7 (translated in the Vulgate as circumaedificavit) and 9 (Vulgate: conclusit).128 Confirmation of the link between thorax and gimel, and with Eusebius’s and Jerome’s glosses of plh&rosij and plenitudo, is provided by the expression in II Reg XXIII: 7: armibitur ferro. The Hebrew for armibitur here is almy yimmale, literally ‘be(come) filled’. The Douay-Rheims

126 Lag. 36,28, where the gloss is referenced to Reg. I,I.1, i.e. I Sam. I.1. In the AV and NEB the name of this king is transcribed as Toho, which is glossed by Young (no date) as ‘humility, depression’, although the noun toho is translated elsewhere as, inter alia, ‘confusion’ (Isa. XXIV.10; XXXIV.11; XLI.29), ‘nought’, i.e. ruin, vacancy (Isa. XLIX.4) and ‘wilderness’ (Ps. CVI.40). At I Chr. VI.34, however, the same king is named as Toah, while the noun toah is translated at Isa. XXXII.6 as ‘error’. Jerome does not appear to gloss the name of the king Toi (= Tou) mentioned at II Sam. VII. 9–10 (= I Chr. XVII.9–10), which cloud also be related to toah, taah. 127 Published as Athanasii, Archiepiscopi Alaexandrae, Interpretatio Psalmorum sive De titulis Psalmorum, ed. Nicolaus Antonellus, Rome, 1746: 375–79; also in PG XXVII, cols 1311–12. 128 Another Hebrew word for ‘hedge about, defend’ is ganan, with its related gan [garden]. The latter is directly associated in Aqiba 1 (Wünsche, 1909: 189) with gimel, presumably on the basis of an al bam gan series.

59 A Family of Names

translation of the expression is ‘must be armed with iron’, but the AV renders it as ‘must be fenced with iron’. The senses ‘be filled with’, ‘be fenced with’ and ‘be armed with’ all converge in the senses of the Greek verb qwrh&ssw, which means not only ‘arm (with a breastplate)’, but also ‘satisfy (with drink)’, the latter sense being reflected in plhro&w [fill full (of food); requite, pay (back) in full], whence Eusebius’s plh&rwsij [repletion (with food); recompense], Jerome’s plenitudo, and Ambrose’s retributio. 1.1.6 Successors to Ambrose and Jerome Ambrose’s and Jerome’s interpretations of Hebrew letter-names had long- lasting and far-reaching influence. The interpretations in Jerome’s Liber interpretationis hebraicorum nominum (Jerome 2, Table 8) were followed by Cassiodorus (c. 485–582) in his Explicationes psalmorum, which were conveniently summarized in the De titulis psalmorum commentary attributed to Bede (Table 10 (b)). Moreover, Jerome’s interpretations from his letter to Paula (Jerome 1, see Table 11, nos. 3 and 5) were reproduced virtually unchanged in a pseudo-Jerome commentary, In Lamentationes Jeremiae (s. vii/ix?), while Hrabanus Maurus quoted the whole letter at the beginning of his explanation of Lamentations in his Expositio super Jeremiam prophetam XVIII.1.129 To give an idea of the wide distribution of these interpretations of Hebrew letter-names, I have assembled in Tables 11 and 12 a number of later lists containing glosses based on either Jerome or Ambrose. In determining which of Jerome’s versions was used, the inflected forms of the connexiones list can prove to be a convenient tool. Some compilers of such lists, however, picked and mixed glosses from both Jerome and Ambrose, and in some cases also integrated interpretations from other sources. I give some examples of this mixing of glosses in Table 13. A particular curiosity of relevance in the context of the English and Scandinavian rune-poems, which I shall be discussing later, is a Carolingian poem based largely on Jerome’s connexiones and probably composed by an Irishman (Table 14 (a)).130 (A similar poem from the same source, but relating to the Greek alphabet, will be dealt with below, together with two other alphabet poems (§1.2.3).) Finally, a Hebrew alphabet with Jerome-based glosses from the transcript of the Auraicept na nÉces in the Yellow Book of Lecan is listed in Table 14 (b), since this particular version of the Hebrew alphabet includes some unusual features that will be of relevance in the discussion of ogam- names. An overview of the interconnexion between these lists is given in Table 15, where the connecting lines are not intended to indicate a direct relation between manuscripts, but rather to show which set of interpretations was probably copied from which, whether directly or via an intermediary. An example of the type of corruption that could occur in the transmission of the lists is the list of Ps. CXVIII tituli derived from Ambrose in the Salaberga Psalter (Table 12, no. 7). Here, glosses like confusio and duc te against beth and zai are clear indications of an Ambrosean source. There are, however, several aberrations. Discit against aleph is not far removed semantically from doctrina, but it is not immediately obvious 129 PL CXI: 1182D–4A. 130 See Traube 1896: 698–9. Thiel gives a list of the interpretations of the Hebrew letter-names, but I prefer to give the full text, since it illuminates other commentators’ interpretations based on the same source.

60 Part 1: Sources of Name-forms and Meanings where angustia against gimel has come from. A possible explanation is that at some stage tributio replaced Ambrose’s retributio – the first word of the Hebraicum version of the gimel octave is tribue, as opposed to retribue in the Romanum and Gallicanum versions – then tributio (cf. also Table 11, no. 11) was re-interpreted as tribulatio, whence a synonym, angustia, was derived. The turbati sunt gloss on caph in the Salaberga Psalter is a clear case of an insular c, in curvati, being misread as t. The uulnus gloss on sen has lost its super, while mem may initially have been glossed with Ambrose’s uiscera, which was taken over in the singular, uiscus, leading first to uisu(s), then to uisu(m). In the case of teth, this has apparently been given a gloss similar to thau, which has then been contaminated by the gloss on eth above it – perhaps reflecting the sense of Hebrew chattath [sin], which Aqiba 1 associates with the letter cheth.131 Finally, with nun, the copyist seems to have muddled the minim strokes of m, i, n and u, so that: nun unicus > * nununicus > n inimicus. A distinct group of glosses with an apparently mixed input from various sources is discernible in a series of English and Anglo-Norman manuscripts (Table 13, nos. 6a, 6b, 15, 18a and 18b). The Anglo-Norman manuscript, Paris 1 (no. 18a),132 is late (s. xiv), but is included here because one of the two sets of glosses it contains appears to be related to glosses in the earlier (s. xii) Cambridge manuscript (no. 15), which in turn show similarities with glosses in the much earlier Vespasian Psalter (no. 6a, London, British Library, Cotton Vespasian A.i, s. viii2/4; sometimes also known as the Canterbury Psalter) and a French double psalter (no. 6b, Vatican City, Biblioteca Apostolica Vaticana, Reginensis, lat. 11, France, s. viii). A feature of the Vespasian and Vatican lists is the assignment of the glosses deus and filius to aleph and beth, respectively. These interpretations would appear to be based on the Christian association of Greek alpha with God at the beginning of all creation, followed by His Son. The Vatican list (6b) extends these two glosses to complete the Trinity by assigning the gloss consolator to gimel – consolator being equivalent to the Paraklete, whereas the Vespasian list (6a) gives consolatio (probably from the Romanum version of the psalter text at Ps. CXVIII.24.133 Elsewhere, I have pointed to the possibility that some of the alphabet glosses in these lists were derived from rabbinical material of the sort which is encountered in The Alphabet of Rabbi Aqiba but was not used directly by either Jerome or Ambrose.134 Moreover, apparent defects in the Vespasian list (two versions of the letter uau, with separate glosses, and the omission of the letters peh and tsadeh135) may in fact not be errors of copying but deliberate revisions. If this suspicion is correct, then whoever drew up the Vespasian Psalter manuscript may have had more than a mere copyist’s knowledge of Hebrew.136

131 Wünsche, 1909: 177, 182. 132 In the manuscript the Paris glosses, which are accompanied by English translations, are jumbled together, but are reproduced here as two separate lists (Paris 1 and 2; nos 18a and b), as restored by Bonnard (1882), Darmester (1882) and Thiel (1973: 102–13). 133 The Carolingean poem on the Greek alphabet by an unknown Irishman (Table 17 (a)) similarly links the first letters to the Trinity (see §1.2.3). 134 Griffiths 2007:213–51. 135 See footnote 34 in Table 13. 136 Other manuscripts containing the same list, with the same apparent ‘defects’, as found in the Vespasian Psalter are: the Mondsee Psalter (also known as the Montpellier Psalter), Montpellier, Bibliothèque Universitaire (Faculté de

61 A Family of Names

The probability that the Cambridge and Paris 1 lists (Table 13, nos. 15 and 18a) were influenced by the Vespasian list can be deduced from the opening trio of glosses on aleph, beth and gimel, and from the gloss of uia salutis on he, which is not encountered in any other set of glosses. The aleph and beth glosses are exactly the same, and although gimel has uox [voice]137 rather than consolatio or consolator, this variant could have been due to an interpretation of the Greek para&klhtoj, i.e. the Paraklete or Holy Spirit, as literally ‘the one who is called alongside’, i.e. the advocate. The uia salutis appears against he in the Vespasian, Vatican and Cambridge lists, but the Paris 1 list differs slightly, in that the gloss is split into two, so that uia is assigned to he and salutis to uau. The uau gloss in the Cambridge lists does not follow the Vespasian version but is derived from Jerome, as are the interpretations in both Paris 1 and Cambridge for teth, yod, samech, phe, coph and resh. On the other hand, both manuscripts gloss daleth and sade according to Ambrose (timor and consolatio), while sermo for lamech may be an erroneous reading of Ambrose’s seruo, and unicus for mem is displaced from nun. As in the Vatican list, there is confusion between the glosses for sade and shin, with shin being assigned Jerome’s gloss of iustitia (iusticia) for sade. The operatio gloss on caph appears to have been based on Jerome’s association of manus with opera (see the connexiones in Table 6), and uiuus for ain probably derived from fons uiuus, as in Cassiodorus (see Table 10 (b)). Sensus for nun could be related to the mens in the Vespasian list, but the source of fortitudo and narratio for tau and heth remains obscure. Paris 2 (Table 13, no. 18b), as restored by Bonnard, Darmester and Thiel,138 is corrupt, particularly between the letters lameth and sade. The sine cordis should read siue cordis, and belongs to lameth, not to mem; unus belongs to nun and not to samech; sine oculis should read siue oculus and belongs to ain, not to phe; and the os iustici0 should be split between phe and sade. Finally, the signa beside uau presumably belongs to tau. In all of these cases, the English translation has been added after the corruption. For the rest, the list is based on Jerome 1, as can be seen from the endings of tabularius (i.e. tabularum), cordis, iustici0, capitis and dentium, as well as, probably, signa.

Médecine), H. 409 (Mondsee, c. 780) CLA VI, 795; St. Gall, Stiftsbibliothek, Cod. 110 (Verona, c. 800), CLA VII, 907; Verona, Chapter Library LII (Burgundy, s. viiiex), CLA IV, 505; Manchester, John Rylands University Library, lat. 116 (Trier, s. ix); Graz, Universitätsbibliothek 86 (?, s. xii); and Vatican City, Biblioteca Apostolica Vaticana, Vat. lat. 6018 (s. ix1/4), fol. 54v. The letter-names and interpretations in the preface of the Montpellier Psalter are the same as in the Vespasian Psalter, but the letter-forms against which the names and interpretations are placed are Greek in the order of the Latin alphabet, whereas the letter-forms in the Vespasian Psalter are Greek in the Greek order. The Vatican Reginensis list is without the Vespasian’s ‘defects’, as are those in Vatican City, Biblioteca Apostolica Vaticana, Chigi D VI.79 (Rome, s. xi2) and in a copy of Cassiodorus’s Expositio psalmorum text on Ps. CXVIII in St Gall, Stiftsbibliothek, 202 (St Gall, s. ixmed). 137 In the Vespasian glosses, uox is the gloss on coph, presumably based on Jerome’s uocatio. In fact, the Cambridge and Paris 1 manuscripts themselves gloss coph as uocatio. 138 See note 132.

62 Part 1: Sources of Name-forms and Meanings

1.2 Greek and Latin letter-names As indicated at the beginning of this discussion of alphabets, it was accepted by Greek and Roman grammarians that the names of letters in Greek and Latin alphabets did not have meanings as such but were either meaningless derivatives from Semitic names (e.g. Greek alpha, beta, gamma, etc., as described by Eusebius in his chapter on the alphabet in Praeparatio evangelia) or purely phonetic (as Latin be, ce, de, etc.). The main sources of knowledge about the way the Latin alphabet was perceived are the grammarians Donatus (fl. 350), Servius (fl. 390) and Priscian (fl. 500), whose views were largely encapsulated by Isidore of Seville († 640) in his encyclopedic Etymologiae. Before turning to the Latin alphabet, however, Isidore gives a potted history of the alphabet in general, along similar lines to Eusebius, but with the addition of details such as the supposed invention of Egyptian hieroglyphs by Isis, who came from Greece (!). 139 He has nothing to say about Hebrew letter-names, other than to point to the similarity of aleph to Greek alpha, beside Latin A, and to note in passing that Hebrew thau means ‘sign’. The examples of Greek and Latin letter-names listed below are relatively late, some as late as the eleventh century. Their relevance in the present context is that they are evidence of the living tradition of inventing names to compete with Hebrew names that, I suggest, is encountered in the invention of names for runes and ogam characters. 1.2.1 The Greek alphabet In discussing the Greek alphabet, Isidore comments on the significance of five letters only:140

• Υ, which he links to Pythagoras’s description of the letter as exemplifying human life, from childhood (the lower stem) to life from adolescence on (the bivium, starting at the fork), with its arduous righthand arm, tending towards happiness, and its more lax lefthand arm, leading to disgrace and ruin; • Θ, which was used on ballots in Greek jury trials as an abbreviation to signify a verdict of θανατος [death]; • Τ, which Isidore relates to the Christian crucifix (T-cross) and to the Hebrew meaning of thau, i.e. sign, as in Ezek. IX.4: Transi per medium Ierusalem, et signa thau in frontes virorum gementium et dolentium [Go through the midst of Jerusalem, and set a mark on the foreheads of the men that sigh and that cry]; • Α and Ω, which he relates to the Christian connotations of beginning and end. Thus, in only one case, thau, is there any question of interpreting the name of a letter as such. In the other four cases that he comments on, significance is not in the name but is a matter of allegory based on shape (Υ), use as an abbreviation (Θ), or symbolism based on position in the alphabet (Α and Ω). Otherwise he simply refers to the use of Greek letters as numbers, which Latin only replicates with its I and X. Typical of medieval listings of Greek names, without any assignment of meaning but with numerical values added, are the first two given in Table 16 (nos. (1) and (2)). The Auraicept Greek alphabet (no. 2) illustrates the confusion that could arise as a result of scribal ignorance of what he was copying. In the more consistent listing of names in Vienna 795 (no. 1), it is important to note the names episinon (i.e. e0pi&shmon), copin and sincope for

139 Etym. I, iii, 1–6. 140 Etym. I, iii, 7–11.

63 A Family of Names

the digamma, coppa and san or sampi, respectively. The perceived meaninglessness of Greek names, however, did not diminish the apparent need to seek or invent meanings. A notable series of glosses in English manuscripts is that listed as nos. 3–6 in Table 16. According to Bischoff, the Erfurt manuscript (no. 3) is the first to give Latin glosses on Greek letter-names.141 Bischoff estimates that the glosses were additions of s. ix–x to a Greek alphabet of s. ixin. In the twelfth century, they were included in a ‘Corpus von Hilfsmitteln für hebraistische und biblisch- antiquarische Studien’, locatable to Normandy and England. In this form, they are to be found in Oxford, Digby 184 (s. xi2/xii2, no. 5), and (with variants) in Cotton Vitellius A. xii (s. xi2/xii2, no. 6). According to Derolez, however, the alphabet on folio 45r of the Vitellius manuscript – which is mixed in with the Hebrew alphabet already given in Table 12 (no. 13) – is identical to that on folio 65r of Exeter, Cathedral Library, 3507 (no. 4), which he dates to the tenth century.142 Bischoff points to the apparent pairing of the interpretations in the Erfurt and Digby lists, as in agricola vetus, virga correctionis, etc. – a feature that is disturbed in the other two manuscripts – and suggests that the compiler was making a conscious attempt to imitate Jerome’s connexiones. One of the pairs – virga correctionis – could, he implies, be traced to Prov. XXIX.15, but it is unclear what might have been the inspiration for the rest of the list. Some of the interpretations appear to be acrophonic, as in agricola vetus, where the v of vetus could reflect a spirant pronunciation of the b in bet; the beatitudo gloss on beata in the Vitellius manuscript might also be intended to be acrophonic. The agricola gloss on alpha could be related to the association of alpha with deus, God the Creator, as in the Vespasian Psalter (Table 13, no. 6a), with a reference to Io. XV.1, where Christ calls God ‘husbandman’ (Vulgate: agricola).143 Some interpretations appear to have been derived from Jerome or Ambrose, but they were attached to different letters (see table below). The motive for the difference is unclear, but the intention may have been to create the pairs. In any case, it is clear that whoever drew up these lists was aware of the tradition relating to the Hebrew letters and was creative enough to develop it for his own purposes.

Gloss Attached to Greek letter in Gloss Attached to Hebrew letter Erfurt/Digby Vitellius/Exeter by Jerome by Ambrose (et) hec e brevis haec zayin scientia zeta scientia yod doctrinae csi doctrina aleph/lamed aleph signorum simma signa tav bonitas delta bonum teth retributio noi retributio gimel gimel oculos chi oculus ayin ayin

1.2.2 The Latin alphabet Isidore treats the Latin alphabet grammatically, along similar lines to earlier grammarians such as Donatus, Servius and Priscian.144 First, vowels and consonants are distinguished, and then the consonants are divided into semivowels and mutae. This categorization has been cited as a possible

141 Bischoff 1966–81: II, 253. 142 Derolez 1954: 219–27. Bischoff does not mention the Exeter manuscript. 143 Leiden, Universiteitsbibliotheek, Vossianus lat. F.24, fol. 88r (glosses on Eucherius of Lyon, d. 455), contains the gloss agricola d(eu)s. 144 For a detailed discussion of the Latin alphabet letter-names, see Gordon 1973.

64 Part 1: Sources of Name-forms and Meanings rationale behind the structure of the ogam scheme (see §2.1.1), and is the probable basis for the naming of Latin letters from before the time of Varro (ob. 27 BC). Donatus draws close parallels between the three categories: vowels, semivowels and mutae (IV, 367, 11; IV, 367, 21; IV, 368, 5):

Vowels are what are pronounced on their own (per se proferuntur) and form a syllable on their own. Semivowels are what are also pronounced on their own, but do not form a syllable on their own. Mutae are what neither are pronounced on their own nor form a syllable on their own.145 How the semivowels were to be pronounced according to this definition (per se ... proferuntur) is unclear. As Schulze points out,146 Cledonius was probably referring to this problem when he stated (V, 27, 27):

What also are pronounced on their own: that is wrong. The sound is in fact pronounced with a vowel and finished naturally, i.e. in its own sound.147 In commenting on Donatus, Servius rephrases the definition (IV, 421, 6; 12):

They are called vowels since they form a sound on their own and form a syllable on their own. [Semivowels] also form a sound on their own, but do not form a syllable on their own. [Mutae] neither form a sound on their own nor form a syllable on their own.148 and he goes on to specify the pronunciation of semivowels (IV, 422, 15):

There are seven semivowels, which are pronounced in such a way that they begin with the letter E and end with their natural sound, as EF EL EM EN ER ES IX, although X begins with I and is used as two consonants.149 and mutae (IV, 422, 32):

There are nine mutae, which should begin with the natural sound and end in the vowel E, as B G and the rest, of which three form an exception, since they do not end in E.150

Priscian refers directly to Servius’s commentary and quotes his definitions almost word for word, but he also explains why X, unlike the other semivowels, is pronounced with an I rather than an E (II, 8, 6): ‘except

145 Vocales sunt quae per se proferuntur et per se syllabam faciunt. Semivocales sunt quae per se quidem proferuntur, sed per se syllabam non faciunt. Mutae sunt quae nec per se proferuntur nec per se syllabam non faciunt. 146 Schulze 19662: 449. 147 Quae per se qvidem profervntnr: falsum est. profertur autem sonus a vocabilis et finitur in naturalem, id est in propium sonum. 148 Vocales dicuntur quia per se sonant et per se syllabam faciunt. [Semivocales] per se quidem sonant, sed per se syllabam non faciunt. [Mutae] nec per se sonant nec per se syllabam non faciunt. 149 Semivocales sunt septem quae ita proferuntur ut incohent ab e littera et desinant in naturalem sonum, ut EF EL EM EN ER ES IX, sed X ab i incohat et duarum consonantium fungitur. 150 Mutae sunt novem quae debent incohare a naturali sono et in vocalem e desinere, ut B G et reliquae ex quibus tres, qoniam non desinunt in e, contumeliam patiuntur. The three exceptions he is presumably referring to here are H, K and Q.

65 A Family of Names for X, which begins with I through reversal [metathesis] of the Greek name ci’’151 Isidore follows the rules expressed by Servius and Priscian, except that he makes no mention of the deviant pronunciation of IX. He does, however, hint at the trifold categorization of the earlier grammarians in his remark: ‘vowels, semivowels and mutes, in fact called by the ancients sonae, semisonae and insonae.’152 What exercises him more than any of these distinctions are:153

• the consonantal as well as vocalic values of I and V, and the liquid qualities of L and R (and of M and N in Greek); • the aspirational value of H, making it a nota rather than a littera; the relationship between K, C and Q – Latin being the only language to have a Q (which is absent in Hebrew as well as Greek!); • the duplex nature of X, i.e. being a phonetic combination of the sounds of C and S; • the use of SS for Z and I for Y before Z and Y were introduced into Latin at the time of Augustus. In addressing the question of alphabetical order, Isidore makes the only comment that might be construed as relating to an interpretation of a letter’s sound-value in conceptual terms:

A is in fact the first of the letters with all races from the idea that it is the first sound to be made by new-born babies.154

This idea is one that is found in particular in the glosses on ogam-names. The development of the names of Latin letters, based on the conclusions of Gordon,155 might be summarized as follows:

Before Varro (all possibilities uncertain): ?bē cē dē fē gē hā kā lē mē nē pē kū rē sē tē xē? ?bə cə də fə gə hə cə lə mə nə pə cə rə sə tə xə? ?bē cē dē f ̣ gē hā kā l ̣̣̣̣ m ̣̣̣̣ n ̣̣̣̣ pē kū r ̣̣̣̣ s ̣̣̣̣ tē x?̣ ?bē cē dē əf gē hā kā əl əm ən pē kū ər əs tē ix? Varro (ob. 27 BC), who mentions only mutes ending in -ē: bē cē dē ef gē ? ? el em en pē ? er es tē ex Terentianus Maurus (flor. AD 290): bē cē dē f ̣̣̣ gē hā kā l ̣ m ̣̣̣ n ̣̣̣̣ pē kū r ̣̣̣̣ s ̣̣̣̣ tē x ̣̣̣̣ Ausonius (poem AD 390):156 bē cē dē ? gē hā kā ? ? ? pē kū ? ? tē ? From ca. AD 400 on: bē cē dē ef gē hā kā el em en pē kū er es tē ex The latter-day inventiveness with the Latin alphabet is illustrated by the list of letter-names in Paris, Bibliothèque nationale, lat. 4841 (see Table 16, no. 7, cited by Bischoff157) . Here the letters are given the names of parts of the body, based on the acrophonic principle. The purpose was probably to

151 absque X quae ab i incipit per anastrophen graeci nominis ci. 152 Vocales autem et semivocales et mutas a veteribus sonas et semisonas et insonas dictas (Etym. I, iv, 4). 153 Etym. I, iv, 1–15. 154 A autem in omnibus gentibus ideo prior est litterarum, pro eo quod ipsa prior nascentibus vocem aperiat (Etym. I, iv, 16). 155 Gordon 1973: 63. 156 It is not possible to tell from Ausonius’s poem what the (monosyllabic) names of the semivowels are (see below, §1.2.3). 157 Bischoff 1966–81: III, 142, n.13

66 Part 1: Sources of Name-forms and Meanings

provide a sign language that involved pointing to the relevant body-part for each letter.158 1.2.3 Three alphabet poems As Halsall points out, there was a long tradition of alphabetical poems or hymns, in which each successive letter of the alphabet formed the initial letter of a stanza or line in much the same way that the Hebrew letters are used in the acrostic psalms.159 These poems, however, did not take the letters of the alphabet as such as their subject but simply used them as an ordering device. Two poems that do take the Latin and Greek alphabets as their subject are the fourth-century De litteris monosyllabis graecis ac latinis, by Ausonius (Table 17 (b)),160 and an anonymous Irishman’s Versus de nominibus litterarum (alternatively recorded as Versus cuiusdam Scoti de alphabeto; Table 17 (c)). Halsall calls the Irishman’s poem Carolingian, but it is dated by Wright and Halliwell to the middle of the seventh century. 161 Halsall does not think that either of these poems could have had any influence on, say, the English rune-poem. In the first place, she correctly points to the fact that Ausonius’s poem is not acrostic, and although Ausonius’s work was probably known in England, his alphabetical poem, in her view, ‘could have supplied no more than the idea that the alphabet might be described in verse.’162 Of the Irishman’s poem, from which she quotes only the first four verses, she says:

It would be foolhardy to suggest that this particular Latin poem was familiar to the author of the Old English rune-poem, especially since there is at least one other example of the composition of Latin verses on alphabet letters, emanating from Northumbria as early as the eighth century.163 Nevertheless, she emphasizes parallels between the Irishman’s poem and the rune-poem, namely both poets’ devotion of one stanza to each letter, the use of riddles, the length of the stanzas,164 and the ordering of the stanzas according to the traditional sequence of the letters described.165

158 Bischoff (see previous note) follows Süß (1923: 148) in comparing this sign language to suggestions of Ovid in Ars Amatoria. I, 137, but Ovid merely refers to the use of fingers to attract a paramour’s attention. One could just as well point to the passage in Prov. VI.13: ‘He winketh with his eyes, he speaketh with his feet, he teacheth with his fingers,’ which Isidore (Etym. I.xxvi.2) refers to at the end of his comments De notis digitorum – Et Salomon: ‘Annuit oculo, terit pede, digito loquitur.’ 159 Halsall 1981: 42–3 160 Ausonius’s dates are: 310–395. 161 The earliest manuscripts in which the poem occurs date from the tenth century (for details, see Table 2.B, footnote 7). Wright and Halliwell’s dating to the seventh century is partly based on references in the poem to Gregory the Great (line 20) and to Carthage as an African capital (line 30), i.e. before the Arab invasion. The poem has even been linked to Columbanus and his circle (Wright and Halliwell, 1841–3: I, 164–5). 162 Halsall 1981: 43. 163 Halsall 1981:79, note 52; she is referring to Hwætbert’s alphabetical riddles. 164 The Irishman’s stanzas are all three lines in length, but the Anglo-Saxon poem’s stanzas vary quite considerably. 165 Ausonius’s poem does not follow the alphabetical order, but its structure is far from arbitrary. What Ausonius has done is to start with the vowels A, E, O (both Latin and Greek); then I and V, Y (Latin and Greek); then the Greek semivowels in the middle of the alphabet, M, N, C; then the labial and dental mutae, voiced and

67 A Family of Names

The parallels that I would like to draw, however, are different. They are not based on the structure of the poems but on the descriptions of the individual letters, not only in the Irishman’s poem but also in Ausonius’s. This is because I am not so much concerned with the general phenomenon of acrostic poems concerning the alphabet; there is little that is unusual in that. The more important features of Ausonius’s and the Irishman’s poems are the language that is used to describe the letters and the type of allegory, metaphor and general lore that is used in the descriptions, both in general and in particular. It is these features that will, I suggest, illustrate the process of name-giving, i.e. the conversion of gloss, metaphor or riddle into an actual name. In this regard, irrespective of their poetical form, the lines and stanzas of Ausonius and the anonymous Irish poet can in fact be seen as nothing more nor less than extended versions of the interpretations and glosses that have been listed up to now. For that reason it is important, as in the case of Eusebius, Ambrose and Jerome, to present the complete texts of the poems, together with annotations, as in Tables 17 (b) and (c). Ausonius treats the alphabets quite differently to the Church Fathers. In the first place, since the Greek and Latin letter-names could not be interpreted as representing meaningful concepts, other devices had to be found to provide glosses. In the second place, unlike the Church Fathers, Ausonius had no ulterior religious motive. The devices he used to gloss the letters can be categorized as follows: (a) the position of a letter in the alphabet; (b) the relation between Greek and Latin letters; (c) the monosyllabic pronunciation of a letter as a word or part of a word in Greek or Latin;166 (d) the shape of a letter; (e) orthography or pronunciation; (f) a letter’s numerical value or its use as an abbreviation. The Irishman’s poem uses the same devices, but at the same time contains obvious Christian references. From his comment on the letter F, it would appear that that the poet was aware of the Hebrew alphabet. His comment on B (principium libri) could possibly refer to the be-reshith [in the beginning] at the opening of the Hebrew Genesis, but it could equally well be limited to the first word of the Psalms, beatus. On the other hand, his subsequent comment (mutis caput alter et ordo) is reminiscent of the older recension of The Alphabet (Aqiba 1) referred to earlier, in which the Lord has to resolve the contention for first place between aleph and bet. The categories of devices, (a) to (f), are used in the notes to Tables 17 (b) and (c). The notes on the Irishman’s poem in Table 17 (c) are based on a commentary found in one of the manuscripts containing the poem, Chartres, Carnot 55, fol.1, published by Omont.167 voiceless, B, D, P, T (together with Greek rho because of its similarity in shape to Latin P), then the aspirate and gutterals, H, K, C, G; and finally the terminal consonants of the Greek alphabet, F, X (= Latin X), Y, with the letters used in Greek only as numerals, the e0pi&shma V and J, digamma and coppa. The interpretation of the last five lines is disputed (see Di Giovine 1996: 215–23), but the assignments suggested here would fit a rational pattern. Why Ausonius did not treat the semivowels L, S and Latin F, R, is unclear. 166 The letters which Ausonius places at the end of a line all have monosyllabic names; any Greek names which are not monosyllabic are included within the line. 167 Omont 1938, 42: 436–40. It has been suggested that the commentary may even be

68 Part 1: Sources of Name-forms and Meanings

Besides these two poems cited by Halsall, I also give the text of a Carolingean poem on the Greek alphabet (Table 17 (a)) by the same ‘unknown Irishman’ as wrote the poem on the Hebrew alphabet already given in Table 14 (a).168 The poem on the Greek letter-names is mainly an elaboration of the numerical values attached to Greek letters, but its interpretations of the first three Greek letters correlate with those given for the interpretations of the first three Hebrew letter-names in the Vespasian Psalter (no. 6 (a) in Table 13). 1.2.4 Virgilius Maro Grammaticus’s Metrofia Finally, in connexion with the Latin alphabet, mention must be made of the unusual names and Latin glosses which Virgilius Maro Grammaticus added to the Latin alphabet as part of what he called his fifth ‘Latin’ of Metrofia in his Epitomae (Table 18).169 A detailed investigation into the personalia of Virgilius170 – his identity and dates – would extend beyond the scope of the present thesis. However, his glosses on the Latin alphabet and his (idiosyncratic) list of letter-names, which probably date from about the middle of the seventh century, 171 are relevant to the discussion of ogam- and rune-names for two reasons: 1. As Calder recognized, the names and Latin glosses on the Latin alphabet presented in the longer recension of the Auraicept na nÉces inbetween (a) the Hebrew and Greek alphabets and (b) the ogam list are virtually the same as the names and glosses given by Virgilius in his Epitomae; an advantage of the Auraicept version is that it includes Irish translations of the glosses given by Virgilius.172 See the lists in Tables 18 (a) and (b). 2. It can be demonstrated that a large number of Virgilius’s names for the Latin letters, which at first sight may seem to be pure nonsense, in fact prove to be Irish words – with an ‘Irish’ twist which I would compare with the enigmatic approach found in the kennings and glosses on the ogam-names. Attempts to unravel Virgilius’s names have included pre-eminently those by Michael Herren and Vivian Law. It was Herren who recognized that some of the names appeared to be Irish words, if the alphabetic letter to which each refers is removed (e.g. p in gabpal > gabal):173

by the poet himself (see Mantius 1911, I: 191). 168 See Traube 1896: 699–700. 169 Cf. Tardi’s 1928 edition of the Epitomae, and also Stangl’s collation of the manuscripts (1891). 170 For ease of reference, the name of Virgilius Maro Grammaticus will be referred to simply as Virgilius (not to be confused with the Roman poet Virgil, i.e. Publius Vergilius Maro). 171 In this I follow Herren 1979: 27–71 and 1980: 243–50. 172 Calder 1917: xliv. The alphabet lists are to be found at Aur. 4211–22 but not at Aur. 1131, where they might have followed the Hebrew and Greek alphabet lists.Herren (1979: 65, n. 192) notes that: ‘Even the close parallel between Virgil’s metrofia and the Aipgitir Laitianda (Aur. 4211 ff.) reveals a common source rather than borrowing in either direction. Dr. Alqvist, who has made a study of the quotations of Latin grammarians in Auraicept (both text and commentary), assures me that there are no direct citations of Virgil in Auraicept.’ 173 Herren 1979: 54–5.

69 A Family of Names

Two words from Virgil’s fifth Latinity … appear to be Irish. They are gno, glossed utilitas, and gabpal, glossed obsequium. Gnó in Old Irish means ‘business’, ‘concern’ or ‘venture’ (see the Contributions [= DIL], s.v.). In he form gabpal, the p is extraneous, according to the principle of the alphabetical game involved in metrofia.174 The intended word is therefore gabal. Gabal is a kind of tribute tax (sic)175 in Old Irish, hence obsequium (see the Contributions [DIL], s.v. gabal, 2). Law did not accept that the names might be comprehensible words but at the same time tried to determine a key – or rather a series of keys – that might explain the link between the names and the glosses:176

Running throught these words [Virgilius's letter-names] is a complete alphabet — a in dicantabat, b in bora, c in gcno, and so on.177 But even if these letters are eliminated, the words that remain are no more comprehensible. Applying letter substitution and other techniques works on a few: bora is presumably inspired by robur, pl. robora, ‘strength’;178 sade, as has often been remarked, is the name of a Hebrew letter which has the meaning ‘justice’; rihph, meaning ‘hilarity’, is a lightly disguised version of risus, and similarly rfoph ‘veneration’ conceals ritus ‘rite, service’. Clitps ‘nobility’ is based on inclitus ‘noble, renowned’. Might gal ‘kingdom’ hide rex ‘king’, fkal ‘religion’ prex ‘prayer’, and spadx ‘longevity’ senex ‘old person’? Even after these puzzling forms are unscrambled (and to do them justice several keys, not just one, are needed) the ultimate enigma remains: ‘the whole world is ruled by these things and prospers in them’. Neither of these approaches is entirely satisfactory. In some instances, Herren resorted to the idea of ‘misplaced glosses’:

Another word in metrofia is gatrb hoc est pax. By removing the extraneous t we have garb. One of the meanings of garb in Old Irish is “torrent” or “rough weather” (Contributions [DIL], s.v. 1 and 2). It is possible that pluuia glossing merc in the line above was meant to go with gatrb = garb.179 He even suggested that the glosses might not be glosses at all:

Another curiosity is ‘blaqth hoc est lux solis’. If we “de-metrofy” blaqth, we have blath: bláth in Old Irish means ‘flower’. Perhaps the glosses are not glosses at all, but rather associative words and phrases.180 A drawback to Law’s suggestions is that they have to rely on a ‘fluctuating code’, and for that reason I find Herren’s approach to be more productive of a cogent solution. I would even suggest that, in a large number of cases, the

174 Herren 1979, n. 150: ‘In metrofia redundant letters are inserted in alphabetical order into the list of words, viz. fkal, clitps, mymos, fann, ulio, gabpal, blaqth, etc.’. This comment would seem to imply that metrofia was based on a clearly defined principle, but no such definition is known. 175 DIL: ? gabal tribute, tax (i.e. tribute or tax). 176 Law 1995: 89–90. 177 Law notes (1995: 140, n. 8): ‘In order to restore alphabetical order I have transposed sade and bora. The spelling of these strange words, where the context offered no guidance to the struggling scribe, was particularly subject to corruption, making all conjectures as to their original form and interpretation fraught with uncertainty.’ 178 With this and her subsequent suggestions, Law is attempting to find a consistent connexion between Virgilius’s names and his Latin glosses, which Herren did only incidentally. 179 Herren 1979: 54–5. Also, 1979, n. 151: ‘Another possible “misplaced gloss” – could it be deliberate? – is “fkal hoc est religio”. By removing the extraneous k we get fal. The previous word gal is glossed as ‘regnum’. But fál in Old Irish means ‘king’ Contributions [= DIL], s.v. 2)’. 180 Herren 1979, n. 151.

70 Part 1: Sources of Name-forms and Meanings principle of deducting the alphabet letter from the name not only provides a comprehensible Irish word but also a word that directly relates to Virgilius’s gloss as well as to interpretations interpretations found in alphabet tradition in general. In appreciating these connexions, moreover, the Irish glosses to Virgilius’s names in the Auraicept version of his alphabet provide valuable evidence. In some other cases, however, it is not a matter of deducting the alphabet letter, but of adding a syllable representing the letter, as, for example, in Virgilius’s name for E, namely ter, where addition of e provides an Irish word, eter, which, as I demonstrate below, fits both Virgilius’s Latin gloss, dualitas coniugalis [conjugal duality] and the Irish gloss eathnadh (i.e. étrad [lust, concupiscence, fornication]) given in Auraicept list. As well as the example of E, I give below three other examples illustrating Virgilius’s approach, namely the letters P, Q and D. My reason for choosing the name and gloss on P is that it indicates that Virgilius may have been familiar with the tradition called on by Ausonius; his name and gloss for Q suggests that he may have had some knowledge of the Jewish tradition reproduced in The Alphabet of Rabbi Aqiba, while the example of D, although not entirely transparent, provides clues that may bridge the Irish interpretations of ogam D and the description of the d-rune in the Anglo- Saxon Rune Poem. Finally I deal with the letter Z, which is glossed in the Auraicept version of the metrofia as preciositas (see Table 18 (b)). This gloss is of significance in the present context because it would appear to link with the interpretation of the name of the s-rune, which I discuss in §2.4.3.

• E : ter hoc est dualitas coniugalis (.i. eathnadh) This is perhaps the most easily understood of the names that require the addition of a syllable as a prefix to yield a comprehensible word, although in this instance the syllable is a simple alphabetic letter, e, giving Irish eter [between (two)]. The clues to understanding Virgilius’s line of thought here are the Latin gloss, dualitas coniugalis (Aur. 4213 has duailitas without coniugalis), and the Irish gloss in Auraicept: eathnadh, which Calder suggests correcting to eathradh (with the r being a mistake for n), i.e. étrad.181 Irish eter is in essence the preposition meaning ‘between (two)’, while étrad means ‘lust, concupiscence, fornication’ and eadradh ‘lust, lechery, whoredom’. The gloss in O’Mulconry’s Glossary 457, cited in DIL, s.v. étrad, is particularly illuminating: ‘etred o eit, quia illia concupiunt et parturiunt [since they covet and give birth], no etred .i. riad n-etere [< e9tai&rh].’182 For eit read ét [emulation, jealousy] while riad n-etere, glossed εταιρη, refers to the ‘double life’ involving an etaira [female companion],183 but more specifically a companion as opposed to a wife ‘with various shades of meaning, from concubine (who might be a wife in all but the legal qualification of citizenship) down to a courtesan, but distinguished from a po&rnh [prostitute]’,184 hence dualitas coniugalis. An Irish form ét could have been suggested by the Hebrew letter- name (ch)eth, but another connexion between eter and letter-names for E

181 Calder 1917: 231. 182 Stokes and Meyer, vol. 1, 1898. 183 Michael Herren (personal communication) points out that ‘Gr. e9tai&ra (female companion) might have been conflated in meaning with e3tera (the other one, female), as the pronunciation of the old diphthong ai changed to e fairly early.’ 184 Liddell and Scott, s.v. e9tai=roj.

71 A Family of Names may be seen in the Greek name e1yilon. The adjective yi&loj means literally ‘bare, stripped of hair’ and refers grammatically to the spiritus lenis as opposed to the spiritus asper, which in Greek is described as da&soj [rough, thick with hair], i.e. with rough breathing. At Aur. 769, the Irish term for the spiritus asper, i.e. tinfedh [breathing], is analysed as tiniudh feadh [melting of signs],185 which is in neat contrast to Isidore’s description of yilh&, the spiritus lenis: ‘yilh&, which is interpreted as ‘dryness’, or pure.’186 The Irish for ‘dryness’ is terad, and ‘dry’ tir, tur, tair. On that basis, Virgilius’s name of e-ter may have been based on an interpretation of Greek e1-yilon as e-tair > e-ter, the ‘dry e’ (as opposed to h0ta), which was then playfully associated with riad n-etere, i.e. e9tai&rh, the ‘double life’, with an e9tai&ra, i.e. dualitas coniugalis.

• P : gabpal hoc est obsequium (.i. umla)187 Of all the names in Virgilius’s list, the name for P is the easiest to understand, although the solution does not seem to be the one suggested by Herren, i.e. gabal [tribute, tax].188 As Herren noted, removal of the alphabetic letter P from Virgilius’s name gabpal does give a sensible Irish word gabal. However, although the concept of a tribute or tax may seem to be re-inforced by the Latin gloss obsequium [compliance, yielding, obedience, allegiance] and by the Irish gloss of umla [humility, submission, obedience] at Aur. 4219, these meanings do not appear to have any connexion with the letter P. It may be argued that they were not meant to have: ‘[p]erhaps the glosses are not glosses at all, but rather associative words and phrases.’189 But loosening the ties between name and letter is an easy way out and is not necessary in this instance. Irish gabal, as well as referring to a tribute or tax, was also an alternative spelling of gabul [fork (of a gallows), gibbet], and this does indeed link to the letter P, or rather to the shape of the Greek P. Corroboration of this observation is provided by a line in Ausonius’s poem De litteris monosyllabis graecis ac latinis: ‘The shape of the “hostile yoke” will be given you by P’190 The Roman iugum consisted of two spears fixed upright in the ground with a third lashed horizontally to join their tops. A conquered army was required to pass under a iugum to demonstrate their obsequium [compliance, obedience, allegiance] to the Romans. Originally, most gallows were in the form of a Roman yoke and even came to be referred to in English as ‘a pair of gallows’ on account of the two upright supports for the crossbeam on which the criminal was hung.191 Admittedly, gabal does not refer to the Roman yoke as such but the fact that Ausonius links the shape of a iugum to

185 Cf. the the Anglo-Saxon Rune Poem’s description of the rune-name hægl [hail] for h: wealcaþ hit windes scúra, worþeþ hit tó wætere [gusts of wind toss it about; then it turns to water] (Table 31-6 and 7). 186 Etym. I, xix, 10: yilh&, quod interpretatur siccitas, sive purum. 187 The text between brackets in the headings are the Irish glosses given in the Auraicept (see Table 18 (b)). 188 Herren 1979: 54. 189 Herren 1979: 55, n. 151. 190 hostilis quae forma iugi est, hanc efficiet ? (seeTable 17 (b)). 191 See OED, s.v. ‘gallows’: ‘So far as our material goes, Caxton is the first writer to speak of “a gallows”, though he also uses the older expression “a pair of gallows”.’

72 Part 1: Sources of Name-forms and Meanings that of a Greek P and that Virgilius links his P to a gabal cannot be fortuitous. In the case of gabal, there is no immediately obvious connexion with the Hebrew name of the letter, peh, or with Jerome’s or the Auraicept’s glosses on the name (Tables 9 (a) and 14 (b)). However, if Virgilius had access to a copy of Jerome’s glosses listed in his De psalterio (Table 8), he will have found that fe is listed as an entry in an alphabetical list (not specifically a list of letter-names) between eth and gibal, i.e. beginning with an f. The word following fe in Jerome’s list, gibal, is glossed definiens uel disterminans [limiting or dividing by a boundary] and relates to the Hebrew gabal [be a border, set bounds]. The Hebrew word has nothing to do with Irish gabul/gabal, but the similarity in the form of the words in combination with the preceeding gloss on fe as laqueus uel decipula [noose or snare] might have helped in confirming the image of P as a gallows (with a noose) that will have been suggested in the imagery already familiar to Ausonius. Such speculation on the minutiae of Virgilius’s possible process of ‘creative thinking’, however, is secondary compared to the general indication provided by the gabpal example, namely that the rest of Virgilius’s list is likely to contain ‘manipulated’ Irish words and that it may also relate to letters other than those in the Latin alphabet, in this instance Greek P.

• Q : blaqth hoc est lux solis (.i. griansolustur) Like gabpal, the name blaqth is an Irish word with the alphabetic letter inserted: bla-q-th. The situation here is again more complex than Herren supposes, because it is not immediately obvious what bláth [flower, blossom] has to to with lux solis [sunlight], or indeed with the gloss on the Hebrew name cop(h): uocacio [= vocatio] .i. gairm [a summons, cry], which is found at Aur. 4175. However, bláth is also used in an extended sense to refer to ‘bloom, bright colour, flourishing appearance’, for which a synonym is solustar [brightness, light], cf. solustacht [brightness, radiance], solusta [bright, shining] and solus (< so + lés [bright, clear, light-giving]). It is thus conceivable that in particular solus called to mind Latin sol and lux solis. But that still does not explain the connexion with the letter Q. The Latin gloss of uocacio on the letter-name cop in the Hebrew alphabet in the Auraicept (see Table 14 (b)) is the same as Jerome’s gloss of uocatio,192 which in turn is apparently based on a reading of Hebrew qol [voice, cry, noise, sound, proclamation] or qara [cry, call]. In the ‘alphabet’ psalm, Ps. CXVIII, the subject of several of the verses in the qoph section is ‘crying’ or ‘voice’. In vv. 145, 145 and 149, for instance, the head-words are arq qara, arq qara and lwq qol, respectively, compared with µdq qadam [prevent] in vv. 147 and 148.193 Virgilius, on the other hand, seems to have followed a completely different line of thought, which is found in rabbinical writings on the alphabet. There is no evidence that he was acquainted with Jewish commentaries, but his choice of blath as a name for Q indicates that he may well have been aware of the Jewish tradition, described above in §1.1.4, whose roots are found in the Babylonian Talmud and are worked out

192 See Thiel 1973: 92. 193 Jerome probably chose vocatio rather than vox in imitation of Eusebius’s κλησις (see Table 6). Both commentators appear to have interpreted qoph in terms of the vocation, of God.

73 A Family of Names in detail in the two recensions of The Alphabet of Rabbi Aqiba. As described in §1.1.4, s.v. qoph, the earlier of the two Aqiba texts explains that qoph means ‘holy one’ (Hebrew vdq qodesh), by contrast with the subsequent letter, resh, which is said to mean ‘wicked one’ (Hebrew [xvr rasha).194 It is then asked why the letter qoph has a horn (Hebrew ÷rq qeren), to which the answer is given by way of reference to Psalm LXXIV.11: et omnia cornua peccatorum confringam, et exaltabuntur cornua iusti [AV translation: ‘All the horns of the wicked also will I cut off; but the horns of the righteous shall be exalted.’]. More significantly, however, the later text of The Alphabet of Rabbi Aqiba says that qoph is Moses, but without explaining why.195 An explanation can be found in Ex. XXXIV, 29–5, where Moses’s return from Mount Sinai is described. In modern translations of the passage it is said that ‘the skin of his face shone’.196 The word ‘shone’ here is a translation of the Hebrew verb ÷rq qaran, but Jerome mistakenly translated it as cornuta, reading it (without vocalization) as ÷rq qeren [horn]. In relation to Virgilius’s gloss on the letter Q, it is quite remarkable that his choice of the word bláth, in the sense of ‘bloom, bright colour, flourishing appearance’, should coincide with Jewish interpretations of the Hebrew letter-name qoph in terms of Moses and his shining skin. If this similarity is not purely coincidental, then it must be concluded that Virgilius was aware of Jewish traditions like those expressed in the Babylonian Talmud or The Alphabet of Rabbi Aqiba (and perhaps even Jerome’s confusion of ÷rq qeren and ÷rq qaran — although this need not have been the case).197

• D : bora hoc est fortitudo (.i. nert) Law’s solution for bora = (ro)bora, i.e. Latin robur, pl. robora [strength],198 is attractive since it would also link to the ogam-name for D, daur, dair [oak], the Latin robur also being a name for the oak-tree. Law does not include the latter possiblity because she is intent on demonstrating bora to be Virgilius’s name for B, with inclusion of the alphabetic letter, which means she has to transpose the name for B, sade, with that for D, bora. Instead of resorting to a transposition of names, I would suggest that the Latin gloss at least can be explained in terms of the Latin letter-name, de. As I argue in relation to the ogam-name dair [oak] for D (§2.3.3), Irish dé is the

194 See Goldschmidt 1930: 747. 195 See the German translation from The Alphabet of Rabbi Aqiba (Wünsche 1911: 260): ‘Qoph: Das ist Mose ... Als sie Mose und Aharon sahen, ... deren Glanz des Antlitzes dem Glanze der Sonne glichen.’ 196 Matth. XVII: 2 connects Jesus’s transfiguration to the Exodus passage by using the same image to describe his shining countenance. 197 It is important to emphasise that these traditions predate Kabalistic writings, which did not surface until the twelfth century in southern France, with the Sefer ha- Bahir [Book of Brightness]. The earliest writers (Isaac the Blind, Azriël and Ezra of Gerona, Nachmanides, and Josef ibn Gikarilla) all date from the twelfth to fourteenth centuries. In responding to Bischoff’s article on Virgilius’s Jewishness (1988: 11–16), Herren (1995: 71) is probably correct in pointing up Virgilius’s possible contact with Jews, but not correct in suggesting contact with Kabalistic writings: 'Wherever he [Virgilius] went [on the Continent], it seems that he came into contact with Jews. He learned of the existence of cabalistic practices and acquired a number of Hebrew words, some of which were not attested in Latin patristic sources.' 198 Law 1995: 89–90.

74 Part 1: Sources of Name-forms and Meanings

vocative singular (used in exclamations) of día [god] = biblical elah [god]; synonym elah [oak] = Latin robur [oak; strength] = fortitudo; alternatively Irish dé = voc. sg. of día [god] = biblical El/Saddai = fortis/robustus > fortitudo. Some corroboration of this line of reasoning is provided by Jerome’s citation (in his Epistula 25 ad Marcellam) of Aquila’s interpretation of El as i0sxuro&j id est fortis and of Saddai (another name for God) as i9kano&j, which Jerome interprets as robustus/sufficiens.199 Virgilius may indeed have derived his name bora from Latin robur, pl. robora, but if he did, it would be an example of the use of a Latin, rather than an Irish, word, and it is difficult to see the omission of the syllable ro- as anything other than an arbitrary choice, there being no manifest connexion between ro- and D.

• Z : presiositas .i. loghmairecht The Auraicept list ends (Aur. 4222-3) with the letter Z, which was introduced into Latin, along with Y, from Greek alphabet. Neither Y nor Z figures in Virgilius’s metrofia, which terminates with the letter V (see Table 18 (a)). In the present context, however, the gloss on Z in the Auraicept list is of relevance to my interpretation of the name for the s-rune as discussed in §2.4.3. The Latin letter Z was used to represent Greek zeta, which in turn was based on Hebrew zain, zai, zaith. In his list in De psalterio (Table 6), Jerome in fact places zaith after thau and vau, thereby equating it with Latin Z. In the Auraicept’s list the letter is glossed as presiositas [precious object], which is the same as that found in the alphabets in Exeter, Cathedral Library, 3507, 65r, and London, British Library, Cotton Vitellius A.xii, 45v (Table 16, nos. 4 and 6 ). None of these glosses has anything to do with any of Ambrose’s or Jerome’s glosses, but the Anglo-Saxon rune-name for s has the same connotation: sigel [gem, jewel], cf OIr. lómar [precious], lia lógmar [precious stone, jewel]. In §2.1.2–4, I argue that both runic s and ogam S appear to reflect Greek zeta, Hebrew zayin, rather than sigma/shin or even tsade. In this light, it is possible that both runic sigel and the Auraicept’s presiositas are based on Jerome’s spelling of zaith for the Greek/Hebrew names zeta/zayin, which is given an Irish interpretation, namely saith [wealth] or sét [object of value, a treasure].200 This limited selection of examples from the list of letter-names attributable to Virgilius in the Auraicept is illustrative of Virgilius’s approach to the letters of his metrofia in general. It is an approach which is characteristic of that found in the creation of ogam-names and of the kennings and glosses explaining them, as I discuss in Part 2, so that it is not inappropriate that Virgilius’s metrofia should have immediately preceded treatment of ogam-names in the version of the Auraicept recorded in the Yellow Book of Lecan. 1.3 Hebrew, Greek and Latin alphabets: a résumé

The inventions of Ausonius, the anonymous Irish poet and Virgilius are all at some remove from the exegetical tradition of Eusebius, Ambrose and Jerome. Ambrose’s approach seems to have been the closest to the rabbinical commentaries found in the Talmud and expanded on in The Alphabet of

199 Thiel 1973: 69, 74. 200 See also Griffiths 2011: 123–43.

75 A Family of Names

Rabbi Aqiba. Like Ambrose, Jerome also had a didactic purpose in his glossing of the Hebrew alphabet. In particular, his and Eusebius’s interpretation of the letters of the alphabet as connected, meaningful sequences would appear to be an attempt to ‘christianize’ a Jewish tradition. Jerome’s more linguistically oriented interest in Hebrew is evident in his lists of multiple interpretations. Isidore appears to have shown no interest in the Hebrew alphabet or interpretations of its letter-names, and it is only towards the end of the eighth century that any such interest was revived. It is possible to trace the particular influences of Ambrose and Jerome in the treatment of the Hebrew alphabet, and also in the imaginative treatments of the Greek alphabet, where commentators appear to have felt the need to emulate the meaningful interpretations of the Hebrew letters. Imaginative treatment of the Latin alphabet was much less common. The poems by Ausonius and the later Irish poet concentrate more on the sounds and shapes of the letters than on possible meanings of their names and in so doing they illustrate a discernible riddling or puzzle-setting approach to the alphabet. But these riddles and ingenious descriptions appear to be designed to show off the eruditeness of the poet rather than to provide material with which a teacher might test his pupils. A visualization of the interplay of the various inputs into this process is presented in Table 19. Apart from these re-interpretations of the three sacred alphabets were ingenious systems like the alphabet based on the names of parts of the body, while Virgilius’s invention of names for the Latin letters would appear to be a new departure in that he evidently created totally new names – mostly vernacular, Irish names – on the basis of earlier glosses on Hebrew, Greek and Latin alphabets. The question is: do the ogam- and rune-names, and their interpretations in kennings and poems, belong to one or other of these alphabet traditions, or are they quite distinct from them? The answer may be that both schemes were created on the same basis as Virgilius’s names, with the exception that no attempt was made to emulate Virgilius’s orthographical distortions. 1.4 Ogam-names No names relating to ogam characters are found in epigraphical inscriptions. Sources for the names are all relatively late manuscripts (fifteenth century at the earliest, see Table 3) in which the focus of passages dealing with ogam- names is on what the names mean and on largely mythical accounts of why or how they came to be applied to the characters. In spite of the lateness of the texts and their debatable historicity, however, arguments have been put forward to support an early origin of the names, perhaps even coeval with the origin of the script itself. These arguments will be rehearsed in §1.4.7. First, the textual sources and their contents will be presented. 1.4.1 The sources The main source for ogam-names is In Lebor Ogaim, generally known in English The Book of Ogams, but sometimes also referred to as The Ogam Tract.201 The text of this book, an edition of which Calder appended to his

201 The Irish title of the tract is not found as a title in the manuscript; it is taken from a passage in the Auraicept, where at lines 2813–4 a quotation from the tract is introduced with the phrase amal isber in leapar ogaim.

76 Part 1: Sources of Name-forms and Meanings

edition of the Auraicept na nÉces, presents two of the three known traditions that list descriptions of each of the ogam characters. Each description takes the form of what may be called a two-word kenning, known in the Irish text as bríatharogam [word-ogam].202 None of the kennings includes the actual name of a character but each is formulated so as to provide a clue to the name. The clues, however, are enigmatic enough to require glossing with more or less explicit interpretations which do usually include the name. The two lists of kennings in The Book of Ogams are quite distinct. One of them is said to have been devised by Morann mac Moín, the other by Mac ind Óc. On linguistic grounds, McManus dates them to the Old Irish period, before 800.203 Graves pointed out that The Book of Ogams cannot be earlier than the beginning of the eighth century, since it refers to Adamnan (ob. 704), and cannot be later than the beginning of the tenth century, since it is referred to by Cormac mac Cuillenáinn (ob. 909).204 The full text of The Book of Ogams is found in at least three manuscripts: Dublin, Royal Irish Academy, 23P12/536, fols. 308b44–314 (s. xiv/xv; known as The Book of Ballymote, abbreviated here to BB); Dublin, Trinity College, H.3.18, fols. 26a1–35 (s. xvi); and Dublin, National Library of Ireland, G53, fols. 1–22 (s. xvii).205 In addition, the two latter manuscripts provide briefer columnar listings of Morann’s and Mac ind Óc’s kennings at the end of their texts, together with a third set of kennings and glosses, also in columnar form, attributed to Cú Chulainn. Using the sigla adopted by McManus,206 the distribution of passages listing ogam-names can be summarized as follows (their locations in the manuscripts and published sources are detailed in Table 3):

Kennings attributed to: Morann mac Moín (series A), Mac ind Óc (series B), Cú Chulainn (series C): text of The Book of Ogams: text of The Book of Ogams: BB A1 BB B1 G53 A3 G53 B3 H.3.18 A7 H.3.18 B5 columnar listings: columnar listings: columnar listings only: H.3.18 A2 H.3.18 B2 H.3.18 C1 G53 A6 G53 B4 G53 C2.

One further source for Morann’s kennings, with different glosses from those in The Book of Ogams, is found in the commentary on the text of the Auraicept na nÉces and reproduced in slightly differing forms in each of the two manuscripts carrying the text, namely Dublin, Royal Irish Academy, 23P12/536 (Book of Ballymote), fol. 325a20–325b2, and Dublin, Trinity College, H2.16/1318 (known as The Yellow Book of Lecan, abbreviated to YBL), fol. 536.207 These two sets will be referred to as A4 and A5, respectively. The texts from these manuscripts are given in Table 20 in versions as presented by McManus, but in a different arrangement to his and with a few additions and emendations. The different arrangement is intended to

202 The relevant passages are at Calder 1917: 276–89. 203 McManus 1988: 131. 204 Graves 1876: 449. 205 To this list Derolez (1951: 9) adds London, British Library, Add. 4783. 206 McManus 1988: 133. 207 Foliation as given by McManus 1988: 133. Equivalent texts are also found in Edinburgh, National Library of Scotland, Gaelic I/72.2.1, fol. 26a, and London, British Library, Egerton 88, fol. 71v–72r, respectively.

77 A Family of Names

highlight the parallels between the three columnar versions (A2,6; B2,4; and C1,2) as well as between the texts in The Book of Ogams (A1,3,7; and B1,3,5).208 At the end of the table (Table 20-12), I add a separate list from The Book of Ogams, which McManus did not include in his survey, but which Meroney did include (without full glosses) under the siglum IIb.209 This list – given by Calder at Aur. 5511–5527, just before the Morann mac Moín kennings A1 – is attributable to none of the three traditions treated by McManus, but is nevertheless significant in that it illustrates the ultimate distillation, from these traditions, of what has been termed the tree or ‘vegetal’ alphabet – all the names here except one (muin) being glossed as trees or plants.210 To appreciate these texts the following general comments may be useful. 1.4.2 The form of the kennings The basic text in each of the three traditions is a list of kennings. The glosses which expand on these kennings form separate commentaries, added at different times by different hands, and should be read as such. That is why McManus gave the kennings in list form and in his extended discussion added the glosses separately.211 For the purposes of the discussion in Part 2, however, it is important to be able to refer back and forth between a particular kenning and the different ways commentators glossed it. Consequently, in Table 20, I have grouped kennings and glosses together. At the same time, it is essential to see the list of kennings within each tradition as an entity, as is evident from the form of the kennings and their relation to one another. In all instances, the form of the kennings is the same: two words providing a characterization of the object to which the relevant ogam-name refers. McManus initially compared the kennings to crossword-puzzle clues of the sort that teachers might put to their students.212 Later, however, he described the opposition between kenning and letter-name as ‘that of a semantically marked synonym to an unmarked norm, or of poetic to everyday language.’213 Technically, the two elements forming a kenning are either a noun combined with a second noun, occasionally a noun with an adjective, and, more often, an adjective combined with a noun. In the adjective-with-noun construction, the adjective is usually in the superlative form, followed by the genitive form of the noun (e.g. glaisem cnis [greyest of skin], ardam dosae [highest of trees]), but the comparative plus a dative is also found (e.g. caíniu fedaib [fairest of trees], milsiu féraib [sweetest of grasses]). I have followed McManus in making no distinction between these two constructions in translation, choosing the superlative construction in English – thus ardam dosae214 ‘highest of trees’ and the variant ardu dosaib, lit. ‘higher than trees’, i.e. ‘highest’ or ‘most exalted of trees’. Normally, the noun would be genitive plural but in certain instances some manuscripts appear to give a genitive singular (e.g. caínem éco [fairest of fish]). Within each of the three traditions (A, B and C), there is little variation in the kennings for individual names. I have signalled any exceptions in Table 20 in the notes, although not all orthographic variants have been listed. The main deviations from this uniformity may be summarized as follows:

208 References such as A1,3,7 signify the three versions A1, A3 and A7, etc. 209 Meroney 1949: 21–3. 210 See the Introduction, footnote 12. 211 McManus 1988: 127–68 (extended discussion) and 1992: 42–3 212 McManus 1988: 130. 213 McManus 1991: 42. 214 McManus restores dosae from the manuscript do(s)a. The latter could be genitive singular or plural of a u-stem noun, but McManus takes it to be plural.

78 Part 1: Sources of Name-forms and Meanings

• In the A tradition, the kennings for Muin, Gort, Straif and Uilleann in A1,3,7 differ from those in A2,4,5,6; in the table, the A1,3,7 forms are given in the main text, the A2,4,5,6 forms in the notes. The kenning for Ruis in A1,2,3,6,7 differs from that in A4,5; in this case, the A1,2,3,6,7 forms are given in the main text, the A4,5 forms in the notes. Nin and Onn also have two sets of kennings – Nin: A1,2,3,4,6,7 as against A5; Onn: A1,3,7 as against A2,6 and A5. • There are no kennings in A4 for the vowels or the last five supplementary characters. • There are no kennings in either of the manuscripts of the C tradition for the last four supplementary characters. A notable feature in the B tradition, which highlights not only the poetic nature of the kennings but also emphasizes the intentional linking of each item in a particular list, is the connecting alliteration between the second element of one kenning and the first element of the next (including alliteration of g/c and d/t). There is, however, a break in this pattern of alliteration at the transition from the first twenty characters to the last five (i.e. between Idad and Ébad). Other exceptions breaking the pattern occur in B2,4, where the kenning for Luis is taken from the C tradition and the kennings for the last four suplementary characters are taken from the A tradition. 1.4.3 The form of the glosses There is a fairly regular pattern to the glosses within each of the two textual traditions A1,3,7, and B1,3,5, as well as within the Auraicept commentaries, A4,5. There are also common features in the entries across all three of the traditions as presented in columnar form, i.e. A2,6; B2,4; and C1,2. By its very nature, the columnar presentation is the simplest. Here the pattern is:

(a) phonetic equivalent of the character in the Latin alphabet; (b) kenning; (c) gloss(es). For example, for gétal the lists give:

A2,6 GG luth leighe .i. gilcach ar imat a icce [sustenance of a leech, i.e. broom/reed on account of the abundance of its healing (?powers)]. B2,4 GG eitiu[d] midach .i. gilccach [raiment of physicians, i.e. broom/reed]. C1,2 GG tosach n-ec[h]to .i. icce [beginning of slaying, i.e. healing]. The presentation of the kennings and glosses in the texts is more complicated and less consistent. The A series in The Book of Ogams (A1,3,7) follows the following general pattern, though not always in precisely the same order or with all the elements:

(a) kenning; (b) i.e. name; (c) reason for name; (d) kenning repeated; (e) ‘That is [Morann’s] kenning for xyz’; (f) ‘[The kenning] was transferred to the Ogam letter, which took its name from it’; (g) reason: either the name and the name of xyz are identical, or the character forms the initial of xyz. The xyz in (e) and (g) may be a tree or plant but it may also be some other object, such as a weaver’s beam (in the case of nin), or action, such as blushing (in the case of ruis). For straif, for example, the A1,3,7 kenning and glosses read:

79 A Family of Names

(a) Tressam rúamnai [Strongest reddening (dye)] (e) .i. straif leis-sium sin in ogaim. [i.e. that is his [kenning for] Ogam Straif.] (c) Straif iar ræt, ar is i in straif is tresiu ruamna ic dathadh na ræt, ar is i do-gni in airget ngeal conad gorm ic denum airgit decht de. Is i berbthar tresin fual isin or mban co ndene (or) derg de. [Straif in substance, for straif is the

strongest of reddening agents for dyeing things, for it is it which causes white silver to become blue while making pure silver of it. It is it which is boiled in urine into white gold and makes red gold of it.] (d) Tresim ruamna in straif iar ræt. [‘Strongest reddening’ [applies to] straif in its concrete meaning.] (f) Tugad uaid-side isin fid dianad ainm straif ar æntaid anma aturu; [The kenning] was transferred thence to the letter called Straif on account of their identity in name;] (g) .i. straif ainm cechtar de. [i.e. straif is the name of each of them.] The usual pattern for the B series in The Book of Ogams (B1,3,5) is:

(a) kenning; (b) ‘i.e.that is the Ogam (name called) after xyz’; (c) reason: either a function, or the name is idential with the name of xyz; (d) ‘transferred thence to the Ogam name (or to a cognate or cognominal name)’. (McManus’s translation, ‘cognominal’ = ‘synonymous’.) For sail, for example, the B1,3,5 kenning and glosses read:

(a) Lúth bech [Sustenance of bees] (b) .i. sail sin [i.e. that is the willow-tree] (c) ar a blath 7 ar a canach. [on account of its flower and its down.] (d) Tucad uad-side ara fid coibhnesa in ogaim. [[The kenning] was transferred thence to its cognate Ogam letter.] Although not every element in the pattern occurs in every entry, the principle followed by the glossators in the texts is the same in each case: they start from the kenning and are at pains not only to name xyz, with which the ogam letter is to be identified, but also to give a reason for the identification. This naming and listing of reasons is in contrast to the glosses in the simple columnar presentations (A2,6; B2,4; and C1,2), where not even the name of the ogam letter is given. In Table 20, the kenning which appears at the head of each of the entries in the manuscript has been given below the ogam-name (which is not given in this place in the manuscripts) and is repeated only if it occurs in the course of the commentary given in the gloss. The name of xyz is not translated unless a translation is essential for an understanding of the context. An interpretation of each name will be given in the course of the discussion of each ogam-name in Part 2. In the commentaries on the ogam-names in the Auraicept na nÉces (A4,5 in Table 20), the pattern of the glosses is quite different from that in The Book of Ogams. In these texts, the glossator starts, not from the kenning itself, but from the ogam-name and works his way towards the kenning, which he usually introduces with the phrase ut dicitur, and in several instances is quite different from the kenning given in The Book of Ogams.215 The general pattern for the consonants is:

(a) ‘name then was called after a tree’; (b) ‘i.e. (after) xyz’; (c) ut dicitur: kenning is name, i.e. xyz; (d) (in some case also:) a reason. 215 In his transcription, McManus (1988) does not always give the full text for the A4,5 glosses, whereas I have followed the text in the Auraicept.

80 Part 1: Sources of Name-forms and Meanings

For muin, for example, A4,5 gives:

(a) Muin dono (A5 adds: is o chrand ro hainmnigther) [Muin, then, (A5: was called after a tree,)] (b) .i. finemain, [i.e. the vine,] (c) ut dicitur airden maisi muin .i. finemain [ut dicitur, ‘Highest of beauty’ is muin, i.e. vine] (d) .i. iarsani fasas a n-airdi .i. finemain. [i.e. because it grows upwards i.e. the vine.] For the vowels, on the other hand, a shorter version is found:

(a) ‘name, i.e. xyz’; (b) (to which A5 adds:) kenning, i.e. xyz (or a gloss). For idad, for example, the A4,5 gloss reads:

(a) Idho .i. ibhar [Idad, i.e. yew]216 (b) to which A5 adds: ut dicitur sioneim fedha iobair; iobhar .i. eo barr simper. [ ut dicitur, ‘Oldest tree’, yew-tree; iobhar, i.e. eo barr ‘salmon/yew-tree end’ simper.] Again, the A4,5 entries are not entirely consistent, but this general description highlights the main differences with the glosses in The Book of Ogams. 1.4.4 The ‘tree alphabet’ or vegetal alphabet An ogam-name is not always interpreted in the glosses in terms of a tree or plant. The fact that the term ‘tree alphabet’ or alphabet végétal, as sometimes applied to the complete series of ogam-names,217 is a misnomer can be verified by reference to Tables 21 (a) and (b), which present all the tree or plant names that are either used as ogam-names as such or are mentioned in the glosses on the names. Six of the lists, i.e. those labelled with McManus’s sigla (A1,3,7; A2,6; A4,5; B1,3,5; B2,4; C1,2), have been extracted from Table 20-2 to 20-11, while the list with Meroney’s siglum IIb is the one cited in Table 20-12.218 The two other lists given by Meroney, Ia and Ib (given in Table 20-13), are the only non-cryptic lists (nos. 27 and 26)219 from among the ninety or so ciphers called certogam [correct ogam] at the end of The Book of Ogams (fols. 312–14). List Ia (no. 27 in the series)220 simply re-iterates the ogam-names from Meroney’s IIb list. List Ib (no. 26 in the series), on the other hand, appears to have replaced most of the non- vegetal ogam-names by the names of trees or plants. (The names altered in this way are indicated in Tables 21 (a) and (b) in bold print.) On the basis of these lists it is possible to identify four types of glosses:

(a) glosses containing no vegetal names or reference; (b) glosses containing at least one vegetal name that begins with the same initial as the sound-value represented by the ogam character and is used as the ogam-name (i.e. an acrophonic vegetal name); (c) glosses containing acrophonic vegetal names, none of which is used as the ogam- name; (d) glosses containing only non-acrophonic vegetal names.

216 The A5 gloss spells the name idedh. 217 See the Introduction, footnote 12. 218 Meroney 1949: 21–3. 219 See Calder 1917: 302; 304 (Aur. 5925–30). 220 Aur. 5928–30.

81 A Family of Names

The distribution of these types of glosses across the various interpretative traditions (A, B and C) and their locations (i.e. whether in the texts, e.g. A1,3,7, or in end-lists, e.g. C1,2) is presented graphically in Figure I (Volume 2). This analysis does not pretend to be rigorously statistical; I shall return to it in more detail below, but a preliminary conclusion would be that some glossators found it more important to interpret the kennings in terms of trees and plants than others, who were content with non-vegetal interpretations. It might be speculated that the introduction of an increasing number of, first, non-acrophonic (type d), and subsequently acrophonic (type c) vegetal names into the glosses was symptomatic of a process of rationalization, whereby an attempt was being made to demonstrate that all the ogam-names were vegetal, or should be interpreted as such. In Tables 21 (a) and (b), the most advanced stage of this process would appear to be represented by list Ib, where the acrophonic vegetal names from the glosses of list IIb have been promoted to the status of actual ogam-names (names in bold print). 1.4.5 The vegetal alphabet and the tree-lists Another way of approaching an apparent shift of emphasis towards a vegetal interpretation is to consider the lists of trees as classified into four groups in the Auraicept and the Bretha Comaithchesa [Laws of Neighbours]. The Bretha Comaithchesa (referred to below as BC) is an eighth-century tract, but the list itself could be earlier, since thirteen of the tree-names in it are cited in a legal poem which has been dated to the seventh century. 221 It is reproduced in a number of fifteenth-century manuscripts, all of which are published in the Corpus Iuris Hibernici (CIH),222 as follows:

Sigla Manuscript Reference in CIH BC1a Rawlinson B.487, p. 67a vol. i, pages 78–9 BC1b E.3.5, p. 5b vol. i, page 203 BC1c H.3.17, col. 309 vol. v, page 1857 BC2 H.3.18, p. 14b vol. ii, page 582. Essentially, the BC1 lists are in agreement with each other and less so with the BC2 list. Similar lists are included in the Auraicept commentaries at Aur. 1150– 1157 and Aur. 4247–4253, just before the summary of A4,5 kennings and glosses on the ogam-names at Aur. 1157–1198 (A4) and Aur. 4253–4308 (A5). The Aur. 1150–1157 tree-list contains three fewer trees than Aur. 4247–4253, but otherwise the two Auraicept versions are in closer agreement with each other than they are with either of the Bretha Comaithchesa versions. In each tree-list the trees are divided into four groups: A Aire fedha [Chieftain trees], B Aithig fedha [Peasant trees], C Fodla fedha [Shrub trees] and D Lossa feudha [Herb trees]. For comparison, the Auraicept and Bretha Comaithchesa lists are given in Table 22 (a), in their four groups (A to D), with English translations as cited by Calder (Auraicept) and Kelly (Bretha Comaithchesa).223 In group A, the Auraicept and Bretha Comaithchesa versions tally with respect to the types of tree, except for gíus [fir] in the Auraicept versus ochtach [Scots pine?] in the Bretha Comaithchesa (see the italicized names in Table 22 (a)). In the Bretha Comaithchesa, however, ochtach is glossed .i. crand gíus [fir], 221 Ma be ri rofesser (Celtica 9: 152–68), see Kelly 1976: 107. 222 Binchy 1978. 223 Kelly 1976:108–123.

82 Part 1: Sources of Name-forms and Meanings so that ochtach and gius may be considered equivalent.224 Consequently, the only real difference between the Auraicept and the Bretha Comaithchesa in this group is in the placing of aball, ibar and gíus/ochtach. In groups B and C, on the other hand, there is a set of more fundamental differences between the lists in relation to the trees scé [whitethorn], critach [aspen], cærthand [rowan], fidath [bird-cherry] and idath/hida [wild cherry], as can be seen in the table below, where these names have been extracted from Table 22 (a) but kept in the same sequence of occurrence in the lists.

Group Aur. 1150–1157 Aur. 4247–4252 BC1 BC2 B scé sceu sceith crithach critach caorthand caerthann hida cærthand crithouch idath caerthann C fidath fidhout critach sci

From this extract, it will be seen that:

•Aur.1150–1157 has cærthand, crithach in reverse order to the equivalent trees in Aur. 4247–4252. •In group B, the BC1 lists have sceith versus critach in BC2, whereas in group C, the BC1 lists have critach versus sci in BC2. •In group B, Aur. 1150–1157/4247–4252 have critach/crithouch versus idath/hida in BC1/2, whereas in group C, Aur. 1150–1157/4247–4252 have fidhat/fidhout ( idath with prosthetic f-) versus critach/sci in BC1/2. These (mis)matches would appear to indicate that, compared with the Araicept lists, the BC1 lists interchange crithach and (f)idath between groups B and C, whereas the BC2 list interchanges sci and hida (= idath). One other difference between the Auraicept and BC lists in group C is caitne [arbutus] in BC1 versus feithlenn or eidlenn [honeysuckle] in the other three lists. In group D, both the content and order of the Auraicept and BC lists are slightly different. When the tree-lists are compared with the ogam-name glosses in the Auraicept texts, it is found that the sequence of tree-names in the tree-lists shows a degree of correspondence with the order in which vegetal references occur in the ogam texts. As an example, Table 22 (b) compares the Auraicept tree-list at Aur. 4247–52 with the various ogam lists225 and indicates the ogam characters which are given names from the tree-list or whose names are glossed with a reference to trees from the list. A character is given in bold print if it uses the tree-name as its ogam-name, while a character not in bold indicates that the gloss on that character contains a reference to the tree from the list. In particular, the group of aithig fedha [peasant trees] and the beginning of the group of aire fedha [chieftain trees] contain the names of trees that are found as names or in the glosses on the names relating mainly to the ogam characters F S B L H D C Q, as illustrated by the highlighted portions of the following extract from Table 22 (b), where Group B, the aithig fedha [peasant trees], is listed before Group A, the aire fedha [chieftain trees] :

224 According to Kelly (1976: 111), the Scots pine may have survived in Ireland until at least AD 900. 225 The sigla of the lists, e.g. A4,5 etc., are those of McManus and Meroney, as listed in Table 20.

83 A Family of Names

B2, C1, A1,3, List from Aur. Calder's A4,5 A2,6 4 2 7 B1,3,5 4247–4252 translation Ib IIb IV III     Aithig fedha Peasant trees fern alder F F F F F F F F sail willow S S S S S S S S bethe birch B B B B B B B B lem elm L L - L L L - L sceu whitethorn H H H H H - H - caorthand rowan - Q L - R - L - crithouch aspen - Q E, Ea Ea - - E - Aire fedha Chieftain trees dair oak D D D D D D D D coull hazel C C C C C C C C cuillend holly Q Q, T T T - - T -

The sequence F S B L H D C Q is too close to the ogam order of B L F S (N) H D (T) C Q to be purely coincidental, although by no means close enough to reach a conclusion on the precise relationship between ogam and the tree-lists. However, it is almost certain that the order of characters in the ogam scheme was a development from the order of letters in the alphabet on which the scheme is based.226 Consequently, any similarity between the order of ogam characters and the order of tree-names in the tree-lists would suggest that: (a) an association of the ogam characters B L F S H D C Q with tree-names is likely to have predated the compilation of the tree-lists, i.e. in the seventh century,227 and (b) the order of these eight characters may have had an influence on the ordering of the tree-names in Groups A and B of the tree-lists, rather than vice versa. Moreover, a further indication of the primacy of the ogam order over the order of trees in the tree-lists is provided by a fifth tree-list, found in The Book of Ogams, at Aur. 5491–96, which shows an even closer relationship with ogam. This list, given in Table 23 (a), is defective and differs so much from both the Auraicept and Bretha Comaithchesa versions that it is difficult to compare with them. But it is particularly striking that the order of names in the list appears to have been adjusted to follow the order of names in ogam. This is illustrated in Table 23 (b), where it will be seen that there is a symmetry in the selection of tree- names for group B and the beginning of group A, when these names are compared with the order of the ten ogam-names in the first two aicmi:

aicme 1 aicme 2 Group B: bethi luis sail nin huath tinne quert Group A: fernn dur coll

Other indications of cross-fertilization between the vegetal references in ogam-names and glosses and the Auraicept tree-lists (but not the Bretha Comaithchesa lists) can be deduced from a comparison of Tables 21 and 22:

• In group D (lossa fedha) of the Auraicept tree-ists, gilcach is followed by raid (see Table 22 (a)); both these plants occur together in the ogam glosses on GG/gétal in A4,5 and A1,3,7 (see Table 21 (a)). • In the same group D of the Auraicept tree-lists, aitean is followed by fræch (Table 22 (a)); in the ogam lists A4,5 and A2,6, O/onn is glossed with aiten, followed by U/úr glossed with fræch (Table 21 (b)).

226 See below § 2.1.3. 227 See Kelly 1976: 107 (cited above in footnote 221).

84 Part 1: Sources of Name-forms and Meanings

• In group C (fodha fedha) of the Auraicept tree-lists, crand fir is followed by fedlend (Table 22 (a)); in the A2,6 and B2,4 glosses on Ui/uilleann the names, elann and eilenn, i.e. forms of edlenn (= fedlend), occur together with crann fir (Table 21 (b)). • In the same group C of the Auraicept tree-list, feorus is followed by crand fir (Table 22 (a)); in the ogam lists Oi/oir is followed by Ui/uilleann; in A2,6 the Oi/oir and Ui/uilleann glosses contain feorus and crann fir, respectively (Table 21 (b)). Despite these correspondences between the selection of vegetal names in the glosses and the Auraicept tree-lists, it is notable how arbitrary the connexion between ogam-names and vegetal interpretations seems to be, as illustrated in Tables 24 (a) and (b), which show:

• The frequency of several different trees in a single gloss (Table 24 (a)). • The shift of a particular tree from gloss to gloss across the various traditions (Table 24 (b)). Other details brought out in these two tables are:

• The occurrence of both edlenn and edhend in the glosses on G in A2,6 could indicate some confusion by the glossators between these two plant names (Table 24 (a)); a similar type of confusion may have existed between edlenn (edleand) and elenn, eilend in the glosses on Ui as well as between edind in the A4 glosses and edlend (eithlend) in the A5 glosses on Oi (Table 24 (b)). • The occurrence of trom in the gloss on tinne in list IIb and as an actual ogam-name in place of tinne in the Ib list of ogam-names, as opposed to its occurrence in the glosses on ruis in A2,6, A4,5 and B2,4, could be due to the phrase te(i)ne truim [fire of elder wood] in the A2,6 and A4,5 ruis glosses (see Table 24 (b)), which may have led to an association of teine [fire], and hence trom [elder], with tinne. The main conclusion I draw from these observations is that the glosses on the ogam-names clearly demonstrate that not all the names were tree- names from the inception of the ogam system. Rather, there was a certain cross-fertilization between the glosses and the tree-lists. On the one hand, sequences of characters in ogam are echoed by the order of some of the entries in the tree-lists, particularly the tree-lists in the Auraicept and the deffective tree-list given in The Book of Ogams. On the other, interpretations of ogam-names included vegetal references in varying degrees across the different traditions (see Figure I), perhaps collected from the tree-lists. In the C tradition, hardly any vegetal references are found. In the B tradition, there are many more, with a larger number of acrophonic forms in the glosses in the text of The Book of Ogams (B1) than in the end-list (B2). The A tradition has still more vegetal references, with more actual vegetal ogam-names in The Book of Ogams text (A1) than in the end-list (A2), while nearly all the Auraicept glosses (A4,5) contain vegetal references, as do the glosses in the IIb list in The Book of Ogams (which includes none of the traditional kennings). The major difference between A4,5 and IIb is the larger proportion of acrophonic forms, which are eventually adopted as actual ogam-names in the Ib list, the final stage in the evolution of the alphabet végétal. 1.4.6 Nature of the interpretation of ogam-names in the glosses The above review of the forms of ogam-names and their kennings as recoverable from manuscript sources leaves numerous questions unanswered. As with the alphabet letter-names, the recorded forms of the ogam-names are fairly consistent. The problem arises with the interpretation of the meanings of those names – if indeed the names had meaning.

85 A Family of Names

As with the Hebrew alphabet, there are differing interpretative traditions relating to ogam-names, and within these traditions it is questionable whether even the coiners of the kennings really understood the meanings of the names. Tree-names like dair, coll and beithe would seem to be uncomplicated. But what of luis, muin, and nin,228 not to speak of gétal, straif and cert/quert? McManus has even questioned the antiquity of these last three names in the forms in which they occur in the manuscripts as well as the sound-values attributed to the characters (see §1.4.7). Moreover, if the coiners of the kennings may not have been completely sure of the meanings of the names, the glossators who interpreted the kennings were struggling: on the one hand to understand the names as such, and on the other to understand and justify the application of the kennings. Their intention was presumably to make it easier for those who followed them – which cannot always be said of the coiners of the kennings. In some cases a kenning provides a quite literal characterization, e.g. Mac ind Óc’s kenning for fern: ‘Milk container’, which the glossator has no trouble in explaining – ‘fern, i.e. alder (wood) is used to make milk pails’. However, some kennings are thoroughly enigmatic and the glossator gives no help either, as in the case of Mac ind Óc’s and Cu Chulain’s kennings for nin in the B and C series: ‘Boast of women’ and ‘Boast of beauty’, which are not fully explained by the glosses: ‘i.e. (fork of) a weaver’s beam’. The nin kennings and glosses in fact provide a useful illustration of the problem of understanding the interpretations of the ogam-names in general. Although the relation of Mac ind Óc’s and Cu Chulainn’s kennings (B and C series) to the explanation of nin as being ‘a fork of a weaver’s beam’ is unclear, this gloss is corroborated by Morann’s glosses in the A series. But here the kenning is quite different, Costud side [Establishing of peace]. In this case the A2,4 glosses do make a connexion between the kenning and the interpretation of nin as a weaver’s beam by pointing out that a nin is raised in times of peace. On the other hand, the A1,3,7 glosses do not explicitly refer to the raising of the nin but say only that the nin is ‘a sign of peace’. Apparently the glossator here found it superfluous to explain the link between nin as a pure object and its signification as a sign of peace. In the Auraicept texts, the glosses were apparently influenced by the idea that most, if not all, the ogam-names were based on the names of trees and plants. Thus, nin is interpreted as ‘ash-tree’, which is put first in the explanation, while the meaning ‘fork of a weaver’s beam’ is derived from the fact that a weaver’s beam was made of ash. But so, too, were spearshafts, so that spearshafts were made the prime reason for explaining the ‘Establishing of peace’ kenning. Unfortunately, however, spears were more often responsible for destroying rather than establishing peace, so the glossator in A5 found it necessary to present two different kennings:

• He changes Morann’s costud side [establishing of peace] to coscrad sida [destruction of peace] to fit the gloss that refers to spearshafts. • He keeps Morann’s costud side in the form costad sidha as the kenning to which the gloss of ‘fork of a weaver’s beam’ refers. Not only has the kenning been changed to fit an alternative gloss, but the line of reasoning found in the A2,4 and A1,3,7 glosses has been distorted,

228 McManus points out that the kennings on these names have different possible starting-points: luise from the root *leuk- [to shine] versus lus < *leudh- [to grow] (1988: 150); muin [upper part of neck] versus muin [love, esteem] (1988: 156); nin [forked branch] versus ninach [lofty] (1988: 153).

86 Part 1: Sources of Name-forms and Meanings

apparently in order to comply with the concept of a vegetal alphabet. Despite the explanations, therefore, the kennings and glosses on nin remain enigmatic. The name certainly appears to be the term for the fork of a weaver’s beam, which was raised in times of peace; it is also said in A4,5 to be the name of the ash-tree. But, given the tree bias of A4,5, it is difficult to believe that the weaver’s beam got its name from the ash-tree. And what was the connexion with ‘Boast of beauty’ or ‘Boast of women’? Was it the woman’s role in weaving? Were the women or was the weaving beautiful? Why, in any case, was such an esoteric object as a weaver’s beam chosen as the name for the ogam character? Why was the name of a more common object beginning with N not chosen? Was the choice arbitrary? The predominantly vegetal alphabets (Ib and IIb) in fact drop all reference to the ash-tree, preferring ‘nettles’ instead, although the IIb gloss maintains the connexion with a weaver’s beam (see Table 20-12):

Nin in ogaim .i. ginol garmna no nenaid isna feadaib. Nin of the Ogam, i.e. fork of a weaver’s beam or nettles in the woods. The sort of questions posed here could perhaps be dismissed as unanswerable, or as not worth answering. They are, however, the sort of questions that will be addressed in Part 2 of this thesis. 1.4.7 Plausibility of the manuscript record and the question of chronology In the discussion so far, doubt has been cast on the primacy of the ogam tree alphabet, but otherwise the assignment of names and sound-values in the manuscript interpretations of the A, B and C traditions has not been called into question. However, it is important to bear in mind that the manuscript record includes comments on characters that were added to the ogam scheme after it had largely ceased to be used epigraphically, in the seventh century, and its forms were being adapted to manuscript usage. Also, as McManus has rightly emphasized,229 the texts in manuscripts comprise descriptions of how ogam was perceived at the time and are not a historical linguistic account. Where the composers of the kennings and their successive interpreters had problems fitting the material they were describing to their knowledge of Irish and other languages, their solutions were given in contemporary terms and they did not pretend to undertake historical analysis. Thus, where certain Primitive Irish sound-values had merged by the Old Irish period (e.g. Primitive Irish /kw/ > /k/, merging with existing /k/), commentators developed their own explanations as to why ogam had, say, two signs for one sound-value. In McManus’s view, it is misleading to accept without question the Old and Middle Irish descriptions and interpretations of, for example, gort and (n)gétal230 or cert and coll as reflecting the situation at the time of ogam’s inception. The way the manuscript tradition handles these two aspects of (a) innovation (i.e. the addition of extra characters) and (b) revision (i.e. the adaptation of existing characters to contemporary sound changes) can be defined in terms of the characters mainly involved: (a) the forfeda [supplementary characters], and (b) the foilchesta [composite characters] and

229 McManus 1986, 1988 and 1991. 230 The spellings (n)gétal, (h)úath and s(t)raif are intended to indicate that the letter in brackets is sometimes added by scribes in order to match the name to the perceived sound-value. In general, the spelling of names in this section has been normalized, unless it represents a citation from a manuscript.

87 A Family of Names the character known as (h)úath. The forfeda are the last five characters described in the kennings in Table 20-10 and 20-11, four of which were added in the Old Irish period. The foilchesta were s(t)raif, (n)gétal and cert, while (h)úath is problematic in that it was transcribed in the manuscripts as H although, in McManus’s view, there had been no /h/ in Primitive Irish. A summary of the passages on all of these characters in the Auraicept na nÉces and the tract De Dúilib Feda na Fored la Filedaib is given in Tables 25 (a) and (b). According to McManus’s view of the later development of ogam, (h)úath, the foilchesta and the forfeda were first involved in a revision by which, as Sims-Williams describes it:231

(a) ‘foreign, cosmetic’ H, Q, agma [i.e. NG], and Z values were assigned to four of the original letters, (b) an early forfid _X_ (variously denoting short /e/ and a velar consonant) was equated with Greek eta, that is,long /e:/,232 and (c) the other four forfeda were added to correspond to Greek omega (that is, /o:/) and Latin Y, P, X. In McManus’s own words:233

This first revision took place after Ogam had ceased to be used in a practical capacity and had nothing to do with writing Irish. It was inspired by a comparison of Ogam with the Latin and Greek alphabets and was designed to generate an alphabet which would be capable of catering for words borrowed from these languages. The first revision was followed by a relatively late second revision by which:234

the forfeda was turned into a series of diphthongs and vowel digraphs in order to enhance the ogam alphabet as ‘an inventory of sounds in the Irish language’. Sims-Williams would go even further and suggest that in the first revision, too, the long é sometimes assigned to –X– was independent of Greek eta and ‘similarly that the other four forfeda were invented primarily to represent Irish phonemes ... not to imitate Latin and Greek’.235 Running through all these observations is the question of chronology, i.e. a chronology of:

i) the development of Irish sound-values, ii) the development of the ogam signary, iii) the assignment of ogam-names, and iv) the manuscripts’ treatment of ogam. A detailed discussion of Irish phonology cannot be embarked on here, but a synopsis is given in Table 26. Within this very broad framework, the manuscript treatment of (h)úath, the foilchesta and the forfeda (including the accommodation of phonemes like /p/, originally foreign to Irish) can be taken as a touchstone for the plausibility of the manuscript account of ogam in general and may provide some indication of the chronology of the signary’s inception and development, particularly with regard to ogam-

231 Sims-Williams 1992: 36. 232 Sims-Williams (1992: 36, n. 17) points out that, although McManus normalizes the name of _X_ as Ébad, the twelfth-century Book of Leinster marks length in Ór and Ifín but not on the e of its Eba. 233 McManus 1991: 145. 234 Sims-Williams 1992: 36, again citing McManus. 235 Sims-Williams 1992: 36.

88 Part 1: Sources of Name-forms and Meanings

names. (a) (h)Úath and the foilchesta236 Lists of ogam-names and their kennings are recorded in only one specific part of the Auraicept na nÉces, namely the section that is commonly referred to as the Book of Fenius, or Book IV,237 attributed to Fénius, Íar mac Nema and Gaídel mac Ethéoir, at Aur. 1102–03 (Book of Ballymote) and 4136–38 (Yellow Book of Lecan). Passages relating to the particular uses of (h)úath and the foilchesta, on the other hand, are distributed throughout the Auraicept (see Table 25 (a)). Authorship of the sections preceding the Book of Fénius was attributed to: Cenn Fáelad (comprising Aur. 63–734/2616– 3492; Book I), Ferchertne (comprising Aur. 735–1027/ 3492–3984; Book II), and Amergen (comprising Aur. 1028–1101/3984–4101; Book III).238 Cenn Fáelad is said by Calder to have died in 679. However, Thurneysen could find no evidence of the text being older than the end of the eighth century.239 The Book of Ballymote version of Book I, according to him, was probably completed towards the end of the eleventh century.240 Book II is dated by him to the second half of the tenth century at the earliest,241 and Book IV to around the same period. Book III is considered by Calder to be ‘of earlier date and language than the general run of the tract’. It will be seen from Table 25 (a) that the use of (h)úath is dealt with in Books I, II, and IV. In Book I, (h)úath is considered solely in terms of the aspiration of b in order to provide an equivalent of p and is classed together with two other fuilti [augments, accents], namely forsail and arnin.242 In Book II, it is considered in terms of aspiration in general (no examples being given), and in Book IV, it is defined in terms of both aspiration and lenition. All of these uses of H are, in McManus’s view, purely scholastic.243 All three foilchesta are given special attention in Book I, but there is no mention of them in Book II, while Books III and IV mention only ngétal but do not elaborate on it. The name (h)úath is sometimes written hÚath to emphasize the connexion between the reference in the kennings to úath [fear] and the fact that the character bearing this name was transcribed as an H and might be compared with the initial h in, say, Latin hora, which was not pronounced. In other words, an initial h was needed in the name to preserve the acrostic principle. In some Auraicept commentaries, the character is described as a sign of aspiration and at one point it is assigned the function of indicating

236 Besides McManus, Sims-Williams also examined problems surrounding (h)úath and the foilchesta. In a 1993 article (i.e. subsequent in date to the article cited above), he largely agrees with McManus in relation to straif and gétal but not húath and cert. 237 See Calder 1917: xxvi; Thurneysen 1928: 285. 238 Calder 1917: xxvi. 239 Thurneysen 1928: 281. 240 Thurneysen 1928: 283. 241 Thurneysen 1928: 285. 242 According to Meroney (1949: 40–41) forsail originally meant ‘over S’ and indicated the punctum delens over a lenited s, but it later was interpreted as meaning ‘S over (a vowel)’and was seen as an abbreviation of síniud [lengthening], the punctum being supplanted by the Latin acutus as a mark of vowel length: Airnin [for N] was seen as a titulus or suspension for n and originally used to indicate the doubling of a final n)]. 243 McManus 1991: 129.

89 A Family of Names lenition (see Table 25 (a)), neither of which uses occur in epigraphic ogam. In fact, ogam H is not reliably attested in any ogam inscription. Likewise, s(t)raif is not reliably attested in inscriptions, so that it is not possible to determine the precise sound-value attributed to it;244 assumption of a value /st/, /ts/ or /z/ appears to derive from the name itself. As for (n)gétal, there is only one inscription in which it is possible to make out a reliable case for its use, where it appears to have represented the sound-value gw245 – a value that had merged with /g/ by the time the Auraicept commentators came to describe the characteristics of individual ogam signs. The assignments of Z to straif and NG or GG to (n)gétal appears to have been introductions from outside Irish: Z Greek-cum-Latin, and NG possibly from the Latin grammarians’ agma, or even from runic ng,246 perhaps in an attempt to establish a distinction from the sound-values represented by the ogam characters sail and gort. The third foilchesta [composite character], named cert, does occur quite frequently in inscriptions, particularly in the sequence transcribed as MAQ(Q)I (gen. sg. of Primitive Irish *maqqas [son], Old Irish macc, gen. sg. maicc), in view of which it clearly represents the /kw/ that merged with /k/ before the manuscript records came to be written. According to McManus, the manuscripts’ insistence that cert should be used to represent C before U, as Q in Latin (see Table 25 (a)), was again due to Latin influence, the result of a process of choosing between the two Latin supervacuae, K and Q, and not the result of historical linguistic knowledge.247 In general McManus considers that the manuscript presentation of the names and sound-values for H and the foilchesta should be treated with suspicion and not relied on as representing the original Primitive Irish situation. His view is that the ogam series was conceived specifically to reproduce the sound-values of Primitive Irish and that it is difficult to reconcile such a conception with the inclusion in the series of characters representing sound-values that did not exist in that language. In disputing this view, I consider the situation with first (h)úath and straif, then cert and (n)gétal.

• (h)Úath and straif Specifically, in relation to (h)úath and straif, McManus argues that both the names and the values given in the manuscripts for these characters are cosmetic re-interpretations of earlier forms, introduced in order to suit manuscript conventions. His arguments, however, are not entirely satisfactory. He notes Thurneysen’s doubts as to the veracity of the value assigned to straif and accepts his suggestion that whoever was to establish the etymology of the name would be close to ascertaining the original sound- value of the character.248 As a consequence, McManus speculates that,

244 McManus 1986: 25. 245 McManus (1986: 20–25) cites six inscriptions in which the character named ngétal might occur: CIIC 10, 189, 224, 256A, 299 and 439. Of these, however, only CIIC 299 contains a sign that can indisputably be read as a ngétal. The inscription has been read as MOGwEDIAS, and interpreted as a name meaning ‘my prayer’, i.e. Primitive Irish *Mo Gwediā [my prayer] > Old Irish *Mo Guide (see McManus 1986: 20 and Sims-Williams 1993: 146–7). 246 See, for example, McManus 1991: 29–32. 247 McManus, 1991: 33. 248 Thurneysen 1937: 207.

90 Part 1: Sources of Name-forms and Meanings since /s/ was represented by the character sail, straif must have represented a sound distinct from /s/, such as /ts/ or /sw/.249 Similarly for (h)úath, he posits that the assignment of H to this name was cosmetic250 and suggests an original value /y/ (sic, he means the phoneme usually transcribed as /j/) as a consonantal parallel to vocalic /i/ on a par with the parallel between ogam /u/ (i.e. úr) and /w/ (i.e. fern). He admits, however, that he is unable to provide a convincing etymology251 or to explain the absence of a representation of /j/ in inscriptions.252 It is undoubtedly correct to say that the manuscript record reflected contemporary views and did not set out to present an authentic version of the Primitive Irish forms of ogam-names and sound-values. In this connexion, McManus cites the example of fern:253 It is well established that there was no /f/ in Primitive Irish and that the character named fern in the manuscripts represented the phoneme /w/ in inscriptions. Consequently, if the character for /w/ already had a Primitive Irish name, it must have been *werna (cf. Welsh gwern(en) [alder tree(s)]) and changed to fern after /w/ > /f/.254 In other words, if the ogam-name had existed in Primitive Irish, the sound- value of its initial would have changed in concert with the change in the sound-value of /w/ in general; the sound-value of the character did not undergo a radical change from /w/ to, say /k/. If McManus is correct, however, the situation with (h)úath will have been quite different. If this name had originally been assigned to a sound-value other than /h/ – say /j/, which, McManus suggests, was lost initially – then its re-assignment to the character H in the manuscripts would have been due to a desire to find a name for H rather than to the development in the name leading to a change in the value of the character. In other words, it is not possible to invoke here the supposed principle according to which ‘if the initial sound in a name underwent change, a corresponding change in the value of the symbol took place’.255 In anticipation of the arguments I put forward in Part 2, I briefly summarize here my reasons for not sharing McManus’s distrust about the plausibility of the manuscript record in relation to (h)úath and straif. In the first place, in response to McManus, Sims-Williams concludes that ‘the traditional equation of h1 [i.e. H, the first character of the H-aicme] with Latin H is likely to be correct, and possible native reasons can certainly be found for the inclusion (or retention) of such a letter in the ogam alphabet.’256 Moreover, Schrijver agrees with Pedersen that word-initial h- and word-final -h were not merely orthographic in Old Irish, but represented a phoneme of

249 McManus 1986: 25; 1989: 160. Sims-Williams suggests (1993: 161) that, besides /sw/, another possibility is ‘an Irish equivalent of the Gaulish ‘’ phoneme < *st’. He also offers an etymology for straif in the kenning’s sense of ‘sulphur’; the matter will be pursued in Part 2, under the letter-name. 250 McManus 1986: 16–17. 251 McManus 1991: 36; the issue is pursued in §2.3.2, s.v. (H)úath. 252 For example, no distinction is made in ogam inscriptions between yo-stem and o- stem gen. sg., both being written with -I (McManus 1991: 115). 253 McManus 1991: 33. 254 McManus 1991: 36. 255 See McManus 1991: 35–6 and explanations summarized in §1.5.2 (d) for the change of runic a to o. 256 Sims-Williams 1993: 171.

91 A Family of Names

‘composite origin’.257 Secondly, it seems unlikely that a character would have been especially created for a specific sound-value and then not have been used, or at least was not in sufficiently frequent use to be a Primitive Irish word attested in inscriptions. On the other hand, it is legitimate to ask why characters representing H and Z should have been (‘cosmetically’) included in the list if they were not liable to be used, especially when other characters, like P, were not included. This type of question arises from the concept of ‘perfect fit’, according to which each character of a script represents one, and only one, phoneme of the dialect where the script originated, and each phoneme of the dialect is represented by one distinct character.258 In the history of writing systems, however, the adoption of a script used for one dialect for use in another has often tended to be less than perfect. For instance, in adopting a Greek alphabet, the Etruscans never used the letters for /b/, /d/ or /o/.259 In the case of ogam, it is possible that the characters designated by the scribes in the manuscripts as úath, straif and (n)gétal could have been a (to them) redundant part of the original list adopted by the Irish in the same way that the Etruscans took over B, D and O in their model alphabet, as found on the ivory tablet found at Marsiliana d’Albegna.260 A third reason for not completely distrusting the manuscript assignments of H and Z is that the ogam signs named (h)úath and straif occur in the signary at precisely the places where they would be expected to occur when compared with an alphabet containing the equivalents of H and Z. Admittedly, there is a grave danger of circularity in this argument, since my derivation of the ogam scheme from the alphabet (see §2.1) presumes a one- to-one correspondence between ogam and a foreign alphabet (a correspondence contested by McManus) which contains equivalents of H and Z at the appropriate places. There is, however, a fourth reason for accepting the manuscript tradition on H and Z, and that is the correspondence I demonstrate in Part 2 between the scribal interpretations of the ogam-names úath and straif and the patristic interpretations of the equivalent letter-names. In particular, it is noteworthy that ogam úath has the same meaning as Ambrose’s Latin gloss of pavor [fear] on the Hebrew letter-name heth (see Table 7). It might still be objected that the Primitive Irish sound-values for the characters named (h)úath and straif were nonetheless not /h/ and /z/ (or /st/ or /ts/). Given the current lack of evidence in inscriptions, however, there is no way of knowing. What I will seek to show in Part 2 of this thesis is that the manuscript tradition was justified in assigning to these signs the values /h/ and /z/ or /ts/ and in recording names that corresponded with equivalent letter-names.

• Cert and (n)gétal These two foilchesta are taken together here because McManus considered

257 Schrijver 1997: 205–27; Pedersen 1909: 410–11. O’Neill (2009: 13) notes that ‘... there is evidence in Early Old Irish of experimentation with : to mark internal hiatus in a word (for example rehe); and to represent /x/, as in the gloss menmnihi which occurs in the earliest layer of the Würzburg glosses, dating from c. 700 (Thurneysen 1946: §25).’ O’Neill also refers to Schrijver’s article. 258 This concept has been applied to the fuþark (see Derolez 1952: 31–33; also 1990: 406; Musset 1965: 96; Antonsen 1975: 1). 259 See, for example, the tables in Caffarello 1975: 71 and Bonfante 1990: 16. 260 Dirringer 1968: 388.

92 Part 1: Sources of Name-forms and Meanings

them both to represent labiovelars in Primitive Irish – /k w/ and /gw/, respectively – which by the Old Irish period had merged with the velars /k/ and /g/ – ogam coll and gort, respectively. As already mentioned, the ogam character named in the manuscripts gétal or ngétal is reliably attested in only one ogam inscription, whereas cert is common. The Auraicept commentaries prescribe for the two characters that cert be used in words in which it is followed by u and that gétal be used in words in which the spelling ng is required (see Table 25 (a) under foilchesta). McManus maintained that the initial n in the spelling ngétal could be disregarded as a cosmetic addition to tally with the artificial assignment of NG, as distinct from the G of gort, there having been no known radical initial NG (= /þ/) in Primitive Irish.261 Taking the name gétal, without initial n, he suggested that this was a verbal noun (admittedly unattested) of gonid [wounds, slays] (cf. Welsh gwanu [pierce, stab] from the root *gwhen-), which would tie in with the interpretation of the name in the kennings as meaning ‘slaughter’.262 On this basis, he posited an original value of /gw/, which then merged with /g/, and had to be distinguished from it by the creation of a cosmetic transcription NG. The example of ogam MOGwEDIAS (CIIC 299)263 certainly provides strong evidence that the character named gétal did indeed represent a Primitive Irish phoneme /gw/. But I would suggest that McManus’s particular interpretations of the kennings on the ogam-name gétal are too tenuous to support his suggestion that the name was coined in a period before /gw/ was delabialized and merged with /g/. He admits that there are problems with his etymology of the name he postulates. More fundamentally, however, it is possible to demonstrate that interpretaions of the kennings and glosses on gétal, as well as on (h)úath and straif, point to a period for the coinages of these names that is much later than McManus assumes (see my discussion of these names in Part2). As regards cert, it would superficially appear that the manuscript tradition preserved a connexion with an original labio-velar /kw/, i.e. Q, by insisting on the use of cert before u, in emulation of Latin Q. But, as mentioned above, McManus prefers to see this practice as another example of re-invention under Latin influence.264 However, whether the association of cert with Q reflects an older tradition or, if McManus is correct, it is the fortuitous result of a later re-invention, the fact remains that the epigraphic sources (e.g. MAQQI) confirm that association.265 (b) The forfeda In addition to the names and kennings on the basic fifteen consonants and five vowels, the manuscript commentaries also deal with five extra characters, known as the forfeda [supplementary characters],266 which have

261 McManus 1986: 18; 1988: 156; 1991: 31. 262 McManus 1988: 157–9; 1991: 37–8. 263 See footnote 245. 264 McManus 1991: 33. 265 Sims-Williams (1993: 147) is similarly sceptical of McManus’s standpoint on cert: ‘The /kw/ symbol remained useful for representing Latin Q, a value presumably preserved in literary tradition, as supposed by Cowgill [1980: 60].’ 266 The forfeda are discussed in detail by Meroney (1949: 19–43), McMannus (1991: 141–6), and Sims-Williams (1992: 29–75). The present discussion concerns itself mainly with the views of McMannus and Sims-Williams. Meroney (1949: 42–

93 A Family of Names

quite different forms to the simple strokes of the other characters. Various versions of the forfeda, as found in different sources, are summarized in chronological order in Volume 2, Figure 2, and are dealt with separately below, but not chronologically. An example giving the shapes, sound-values and names of the series is found at Aur. 1138/4229 (see Figure 1.A; sound- values are in italics, names in roman):267 Figure 1.A ô ea oi ui io æ ebhadh oir uilleand iphin emancoll eba ipin Of these five characters, the last four were added to the scheme in the Old Irish period, but evidence for them on monuments is doubtful.268 The first of them, however, was used in the inscriptions, but the sound-value assigned to it changed over the period of its use. Originally, it probably had the value /k/ or /x/.269 It occured as the initial of the word KOI, which appears to have been equivalent to Latin HIC in the funerary expression hic iacit [here lies].270 When related to its supposed velar sound-value, the character’s shape would suggest a connexion with Greek chi rather than Latin X. Post seventh century, in the period when ogam began to be written in manuscripts along horizontal stemlines rather than cut along the edges of stones, it was known by the name of ébad and came to represent [e:]. About the same time, the second supplementary character was added, in the form of an O straddling the stemline, representing [o:]. This development at least is what is implied by the beginning of the passage in the Auraicept na nÉces which is said to be from the Auraicept Muman [The Munster Auraicept], as recorded by Fenius:

Na cuig fedha cétumus: The five fedha first271 ag fregra dona cuig guthaigib tuc na secht in response to the five vowels he gave the seven fedha fedha, fregra dona defograiph tuc .i. ea oi: as a response to the diphthongs he gave, that is, ea oi; na x feudha iphin ar defougur ata the ten fedha: iphin stands for a diphthong, emhancoull ar x emancoll stands for x, no ar emnad ata or for a doubling, pin ar p, pin for p, conid x samlaidh. There thus being ten.

43) had some interesting ideas about the derivation of the forfeda signs and names, in particular about the relation between ogam-names and the names of the supplementary runes in the OE fuþorc. These suggestions will be considered in Part 2, under the relevant names. His comments on the development of the forfeda as a whole will be referred to in footnotes to the present discussion. 267 Similar series are given at Aur. 1057 with the values aea, oi, ui, iiii, aaee, and at Aur. 1143 with the values ee(aa), oo, uu, ii, aa(ee) 268 McMannus (1991: 79) cites seven inscriptions listed by Macalister as possibly having supplementary characters: CIIC 4, 7, 102, 193, 204, 235 and 240. ‘... of these all but 235 … are doubtful.’ CIIC 235 contains a name with O (or A) and E in supplementary characters. Two British inscriptions, 327 and 409, apparently attempt to reproduce an equivalent of Latin P, on which see below. 269 Sims-Williams (1992: 46–9) makes a strong case in favour of the value /x/, pointing out that, apart from its use in KOI or OI, it is always found in inscriptions in intervocalic (lenition) position. 270 McManus 1991: 51. 79. 119. 271 I leave fedha untranslated, because it appears to be used ambivalently for both ‘vowel’ and ‘letter’ in general.

94 Part 1: Sources of Name-forms and Meanings

This version (given here with my own translation) is that found in the Yellow Book of Lecan at Aur. 4503–06. A similar description, but with the phrases in garbled order, is also found at Aur. 1361–66. In Table 25 (b), I give both versions and attempt to demonstrate the relation between them. According to the first three lines, the introduction of the two new vowels meant an increase in the number of ogam vowels from five to seven, implying a parallel with the seven vowels of the Greek alphabet and inviting comparison of ébad (= [e:]) and ór (= [o:]) with Greek eta and omega.272 The same interpretation would seem to be what is implied by Cenn Faelad at Aur. 255–60/2565–70; 2870.273 An ogam sequence in the manuscript Berne, Burgerbibliothek 207, folio 257r (s. viiiex or ix), from Fleury-sur-Loire, terminates in a forfeda series comprising the first two signs transcribed as vowels ae and oo, followed by two more transcribed as ach and rr or ss (see Figure 1.B):274

Figure 1.B ae oo ach rr

The values given here are those noted by Losch and Sims-Williams. 275 Derolez thought the second vowel could possibly be oe but concluded ‘oo appears to be the most plausible reading’.276 Notably the signs usually given in the third and fourth places, ui and io (see the forfeda given at the beginning of this section, and the discussion below), are absent; also, the ach sign hangs below the line instead of sitting on it, as is mostly the case. The rr or ss sign was not counted amongst the forfeda in later schemes, but according to the ogam syllabary which follows the single line of ogam on fol. 257r of the Berne manuscript, a similar sign occurs in a position that would equate it with Latin Z; so the transcription may be intended to represent two long-stemmed Ss, i.e. Z.277 The stage in the development, after the addition of ea and oi, is less clear but the passage from the Auraicept Muman cited above continues by stating

272 At Aur. 2872, they are explicitly said to be seichimh nGreigde [in imitation of the Greek (i.e. alphabet)]. The idea that ébad and ór were added first and the other three forfeda came later is also expressed in In Lebor Ogaim (ll. 5504–10). 273 Meroney (1949: 38) thought that the first two forfeda may have been added to bring the number of ogam signs up to the number of letters in the Hebrew alphabet, i.e. 22. According to him, the first forfed represented CH; the second was a Greek  , without the cross-stroke, representing P, possibly in emulation of the , i.e. chi rho, monogram, with the well-attested confusion of rho with Latin P. This stage was then followed by a stage where CH and P were altered to EE and OO to give seven vowels as in Greek. 274 A photograph is given by McManus 1991: 134. 275 Losch 1885: 300, Sims-Williams 1993: 140–1; Losch’s reproduction (1885: following p. 288) gives the value oo, but in his footnote on p. 300 he says ‘Im Codex ist  zu sehen; da aber i sonst unten keinen Bogen hat, so vermuthe ich oe. 276 Derolez 1951: 6–7. 277 Losch’s drawing of the sign, showing two sets of four diagonal lines athwart the stemline, is comparable not only to the Z symbols in the Bern syllabary but also to the double straif in ogam series no. 64, Book of Ballymote, folio 313 (Calder 1917: 307). In McManus’s photograph of the Berne manuscript, it is not clear exactly how many strokes make up the sign labelled rr or ss. Closer examination of the manuscript is required.

95 A Family of Names that the remaining three forfeda which were incorporated in the scheme, after ébad and ór, were:

• iphín, with the value of a diphthong or vowel digraph (the term used here is défogur, which can mean both diphthong and digraph),278 • pín, with the value [p], and • emancholl, with the value [x]. This summary does not include the character which elsewhere is called uilen, but in a separate passage – at Aur. 1367–74 in the Book of Ballymote and at Aur. 4420–30 in the Yellow Book of Lecan (i.e. immediately following the Auraicept Muman quote) – uilen is mentioned and is said to have the value y but u medially, while iphín is said to have the value p but i medially. On this basis, it would appear that iphín and pín were both names for p, in which case the full forfeda series would read: ébad = é; ór = ó; uilen = y, or u medially; iphín or pín = p, or i medially; emancholl = x.279 Except for medial u and i, which are not explained more fully,280 these sound-values all appear to have been borrowed from Greek or Latin. At Aur. 1370–74/4427–30, it is said explicitly of emancholl that it was created for foimtin na focul nGrecde no Laitinda do thabairt isin nGaidelg [in readiness for the adoption of Greek and Latin words into Irish]. Possibly, the scheme with ébad = é, ór = ó, uilen = y or u, iphin = p or i, and emancholl = x is what is represented in the twelfth-century Book of Leinster (Figure 1.C):281 Figure 1.C ifin

eba uillend emancholl

Here, it is notable that the shape of uillend is more like an open hook rather than the spiral of later versions, and the emancholl hangs from the stemline, as in Berne 207, rather than standing on it. The ‘hanging’ emancholl may therefore be the earlier form. Unfortunately, only the names are given, without sound-values, but it may be possible to recover the intended values from the tract which immediately precedes the forfeda scheme in the Book 278 It became the convention in Irish orthography to use a vowel digraph to indicate the palatal or neutral quality of a following consonant, back vowels (a and o) being appended to front vowels (e and i) before neutral consonants (giving the combinations ea and io) and front vowels being appended to back vowels before palatal consonants (giving the combinations oi and ui). Diphthongs, on the other hand, were indicated by a length mark placed on the lead vowel. The length mark, however, was not always written, which led to confusion between diphthongs and digraphs. ‘This was not surprising since these two Old Irish categories had become confused in Middle Irish speech: for example, ea was pronounced [ea] verging on [ea], éo was pronounced [eo], and áe had been monphthongized’ (Sims-Williams 1992: 33, citing Jackson 1951: 82–4, and Greene 1976: 39–44). 279 Meroney (1949: 39–40) suggests that this series came about in two stages: first EE and OO replaced CH and P (see footnote 273), which moved to third and fourth places; subsequently, the CH was perceived as ‘double-CH’ or ‘double-C’, i.e. [x], and moved to the end, its place in third place being taken by Y, to give EE, OO, Y, P, X (= CC and CH). 280 Do they perhaps foreshadow the use of these characters as vowels? 281 Reproduced in Sims-Williams 1992: 32.

96 Part 1: Sources of Name-forms and Meanings of Leinster and is entitled De Dúilib Feda na Fored la Filedaib [Concerning the letter(-form)s of the forfeda according to the poets] (Aur. 5416–63). In this tract, the forfeda symbols are not arranged along a single stemline but integrated in the text, which illustrates the values by means of minimal pairs, as, for example:

ebad emancoll ebad iphin uillenn oir beoil baile feo fia cual coel beoir baire feagh fiag buar boir nuar noer However, the order in which the characters would occur if arranged along a stemline can be surmised from the sequence in which they are first mentioned (Aur. 5421–22; Figure 1.D): Figure 1.D ô éo óe úa ía áe éa ói, óa ái,aí ebad oir fin emancoll euad ifin

Four of the five forfeda are given names, but, more significantly, the distinguishing feature compared with the other schemes is that, apart from monophthongized áe, the values are all diphthongs.282 Also, the third and fourth symbols are in reverse order to the actual scheme given in the Book of Leinster, but the name ifin remains in fourth place, and the value úa (not named but presumably equivalent to uillend) is in third place, assigned to the double cross instead of the hook/spiral. This scheme of mainly diphthongs – éa, ói, úa, ía, áe – is somewhat more sophisticated and coherent than the mixed scheme given in the Book of Ballymote, namely – é, ó, y/u, p/i, x (Aur. 1367–74). An intermediate stage between the two may have been that made up of vowels as described at Aur. 1139–46: ee, oo, uu, ii, ee, where the doubled vowels indicate length.283 The scheme which Fenius sees as apparently the definitive version, however, is the one he gives just before this version with doubled vowels (at Aur. 1138). This is the scheme with which I began this section, and with which Fenius begins his discussion of ogam, before explaining the development of the forfeda and the assignment of ogam-names. All the symbols in this scheme, except again emancoll, are assigned values represented by vowel digraphs: ea, oi, ui, io, æ.284 This is also the version given as ogam series nos. 28 and

282 There are one or two exceptions, e.g. trad [flock], where ebad is used for an e- digraph, and bairend [large stone] and baile [town], where emancholl is used for the digraph ai [ai] (Sims-Williams 1992: 34). 283 Calder emends this series to ee(aa) oo uu ii (aa)ee, but Meroney (1949 :40, n. 39) comments: ‘Calder’s emendations are improper, since the signs are not doubled diphthongs; the scribe is likewise in error, it seems, at Aur. 1057, 1144.’ At Aur. 1055–57, Amergen gives a modified version apparently comprising long vowels and diphthongs or digraphs, aea oi ui iiii aaee, where, probably, aea = æa = éa, iiii = í, and aaee = áe. 284 Meroney (1949: 42) appears to consider the diphthongal series of scheme C to be

97 A Family of Names

30 (Book of Ballymote, fol. 312285) at the end of The Book of Ogams (see Figure 1.E):

Figure 1.E

oii ea ui io æ ebad oir uilleann pin emancoll The only differences compared with Fenius’s scheme are that oir has the rectilinear shape of the Bern 207 version, and the order is the same as in the De Dúlib Feda.286 An idiosyncratic, but at the same time informative forfeda sequence is given as series no. 24 in The Book of Ogams (see Figure 1.F):287

Figure 1.F

cai ł æei .c. ł oñ p ł ui ł ł ł ach ui oi ai au .. .. .i. ol ō Here the glosses should probably be read and interpreted as given in the notes to the figure. The cai gloss over signs 1 and 2 might be a scribal error for eai, but it could be a name, a hapax legomenon, for the velar sound-value of the ebad sign found in inscriptions where it stands for /k/ or /x/. As an Irish word (if it is a word), cai means ‘lament’ or ‘cuckoo’, or is an alternative form of cae [way, road, or the custom of coshering]. Sign 8 is reminiscent of the rr or ss sign in the Bern 207 scheme, but it has two sets of only three diagonal strokes, rather than four, and is comparable to the gort sign in series no. 64. The gloss, ach, above sign 8 is the same as that relating to the emancholl in the Bern manuscript and may be intended to refer to the sign 9 next to it. Sign 11, for oi, is the same rectilinear sign as sign 3 but with a stroke inside it. The gloss below the stemline between signs 3 and 4, .i. ol Ð, may relate to sign 3 and mean something like ‘i.e. like o(nn)’. Sims-Williams suggests that sign 5 could represent a BB monogram, indicating a beithe bog(tha) [soft (i.e. voiceless) b].288 He also compares it to the St Andrew’s cross to the right of the stemline in the British dual-script inscription at Crickhowell (Breconshire, Wales; CIIC 327), which the final stage of development, arrived at in two steps: first EO (influenced by the name éo [salmon] ), OI, UI, IO, AE; then, with a change of EO to EA, a ‘neatly reversible’ series of EA, OI, Y, IO, AE. He suggests (1949: 40–2) that the change from a forfeda consisting of long vowels to one composed of diphthongs was connected with the establishment of the accent forsail ([over-S], originally indicating a punctum delens) as an accent indicating vowel length. As a result it was no longer necessary to have separate signs for long vowels, so the forfeda co-opted as diphthongs. 285 Calder 1917: 302. 286 The text of In Lebor Ogaim (Aur. 5508–10) gives a sequence of seven ogam vowels, terminating with the first two forfeda, followed separately by the remaining three forfeda, using the rounded shapes. However, no values are attached to the characters. 287 Book of Ballymote, fol. 312. 288 Sims-Williams 1992: 41–2.

98 Part 1: Sources of Name-forms and Meanings

corresponds to the P of the Roman equivalent of the ogam script TURPILLI on the same stone.289 The cross, however, could possibly be an unsuccessful attempt at an ifín sign, as in the Book of Leinster, or even a specially marked B using the cross form of ogam series no. 64 (snait(h)i snimach [interwoven thread])290 or no. 23, which is glossed bact (beside lact, fect, sect, nect, etc.) and would appear to belong to the preceding sequences described as og(am) ea(m)nac(h) .i. da fid i(n)un(d)a ar in fid .i. da b(ethi) ar B [twin ogam, i.e two identical letters for the letter, i.e. two bethi for B].291 The inverted V- shape of sign 5 could be just that, an inverted Latin V, and the three prongs of sign 4 could be a V plus a vertical stroke (as in sign 11) or a V plus an I, i.e. ui (cf. runic yr, which appears to be a runic ur with an added stroke or I). One other British inscription to display an ogam equivalent of Latin P is CIIC 409, where the two Ps of the name PVMPEIVS correspond with an ogam character which could possibly be in the form of a ‘hanging’ St Andrew’s cross with a vertical stroke through it, but this is guesswork on the part of Macalister.292 On British stones like this, therefore, there was clearly a need for the /p/ phoneme to be represented, in particular where an equivalent of Latin P was required.293 This is not surprising in view of the contact with Latin in Britain and the presence of the /p/ phoneme in British. In Ireland, on the other hand, introduction of a character for /p/ was part of the creation of the forfeda, in a revision of ogam which, according to McManus, ‘took place after Ogam had ceased to be used in a practical capacity.294 In this instance, it is definitely the influence of manuscript conventions and the need to adapt to Greek and Latin orthography and vocabulary that is evident in the development of the forfeda and their names. The same may also be said of the other ogam characters as treated in the manuscripts. As McManus emphasizes at the end of his account of the forfeda developments:

the manuscript tradition presents us with a contemporary modified form of the Ogam alphabet, and not the original preserved in cold storage. This is in fact precisely what one might expect, and the recognition of this clearly undermines any attempt to identify the prototype of Ogam using this material.295 I would in fact go further and suggest that none of the ogam-names presented in the manuscripts had been taken out of ‘cold storage’. All the names were new creations within the process of modification that the ogam scheme underwent at the hands of the manuscript commentators. (c) The foilchesta names, lack of /p/, and the question of chronology In assessing the date of ogam’s origins, McManus cites the absence of a sign for /p/ as one of a number of pointers to a terminus ante quem for the creation of ogam:296

The absence of /p/ is significant as this sound must have been becoming increasingly

289 Sims-Williams 1992: 40. 290 Calder 1917: 307; 309 (Aur. 6085), referred to by Sims-Williams 1992: 42; 1993: 152. 291 Calder 1917: 302; 304 (Aur. 5907–10; 5918), not mentioned by Sims-Williams. 292 Sims-Williams 1992: 42. 293 For Irish accommodation of loanwords with P in inscriptions, see the next section. 294 McManus 1991: 144–5 295 McManus 1991: 146. 296 McManus 1991: 40.

99 A Family of Names

more common in Latin loanwords by the late fifth century (McManus, 1983, 48). The fact that it never had the honour of being assigned a letter name of its own (McManus, 1988, 167, ...),297 therefore, is testimony to the date of the coining of the letter names in general, and it is reasonable to suppose that this was contemporaneous with the creation of the alphabet. Similarly, the labio-velars /kw/ and /gw/ (Cert and Gétal) are consistent with a date prior to the sixth century, when these were being delabialized to /k/ and /g/ respectively. This assessment would probably be correct if it related to the creation of the ogam scheme as such rather than linking the scheme’s creation to the coining of names. The fact that ogam initially had no need of a specific sign to represent /p/ would indeed appear to point to a date of origin before those who used ogam were confronted with substantial numbers of Latin or Greek loanwords containing P.298 In the syllabary on folio 257r of Bern 207 an ogam character for P is included in the form of a Y athwart the stemline,299 which recalls the pín/ifín symbols in the ‘idiosyncratic’ forfeda sequence given as series no. 24 in The Book of Ogams (see above). I would suggest that this manuscript representation of ogam P is an indication of an older association of alphabetical P/P with the letter U/U, which I postulate in my comparison of the ogam list with the alphabet (§2.1). As a consequence of McManus’s insistence on tying the names of ogam characters to their sound-values, the lack of a name for /p/ becomes testimony for the general coining of ogam-names before a name for /p/ was needed. The argument is difficult to sustain if the creation of ogam and the coining of names are decoupled. It is not inconceivable that the ogam scheme could initially have existed without the meaningful names recorded in the manuscripts; after all, the Latin alphabet was not graced with a system of meaningful names, and Greek letter-names were certainly not meaningful. It is theoretically quite possible for the ogam characters to have been known initially by a simple phonetic system similar to that employed for the Latin alphabet and to have been assigned meaningful names later when contact through Christian missionaries with the Hebrew alphabet suggested that meaningful names might be useful. According to this scenario, ogam-names could well have been invented to match the Hebrew system, and only then will they have become a required part of students’ learning programmes. In that case, the names gétal, straif and úath, rather than being vestiges of the names of sound-values that had become redundant in Old Irish, will have been the Irish equivalents of the Hebrew names for characters which had formed part of the original alphabet list as taken into the ogam scheme but not made use of in the inscriptions. Their presence in the scheme, however, led commentators in the manuscripts

297 The omitted reference here is to a subsequent footnote (McManus 1991: 184, n. 49): ‘The consequence of this was that the sound /p/, which was now well established in Irish [i.e. by the time pín or iphín in the forfeda was being used for io or ío rather than P], was left without a distinctive letter name of its own and had to make do with surrogate names such as Beithe bog ‘soft B’, Peithe and Peithbog (see McManus, 1988, 167).’ 298 Ogam QRIMITIR illustrates the early use of Q for P in a Latin loanword. The Old Irish nom. sg. of this word was cruimther. The Irish form would appear to based on a loan from Latin via *praebiter or praemiter (McManus 1983: 46, n. 60). The ogam nom. sg. was probably *qrimiter (< *qremiteras), with vowel affection in the first syllable; the form qrimitir is gen. sg., with palatization following apocope of *qremitirí (see McManus 1991: 60–1). 299 See the photograph in McManus 1991: 134.

100 Part 1: Sources of Name-forms and Meanings

to pay special attention to them and define functions for them in contemporary Irish. Similarly, after the merging of /kw/ with /k/ in Old Irish, the existence of two characters for /k/ required the creation of two names and a redefinition of uses, such as cert before U in imitation of Latin Q. This view is at variance with McManus’s view on húath and the foilchesta and with Sims-Williams’s view on ngétal and straif. On the other hand, it is in agreement with Sims-Williams’s view on húath and cert. On the forfeda, it largely follows McManus’s acceptance of the traditional development of the forfeda, which is disputed by Sims-Williams.300 Chronologically, the developments are summarized in Table 26. The creation of ogam as a signary appears to belong to the Primitive Irish period, before any pressing need to cater for a /p/ phoneme. The crystallization of names – whether this involved a re-assessment of the roles of (h)úath, the foilchesta, and the first of the forfeda (ébad) or (as I would suggest) the actual creation of the whole set of meaningful names – belongs to the first contact with manuscript tradition in the Old Irish period. A further development of supplementary characters to cater for loanwords from Latin and Greek (pace Sims-Williams) was initiated at a later stage in the Old Irish period (with its first record in the Bern manuscript), while a final exploitation of the forfeda was initiated sometime before the record in the Book of Leinster (twelfth century). 1.5 Rune-names In dealing with the names of runes, it is useful to take account of two different types of sources: those in which the names or interpretations of them are recorded, and those from which the names can only be deduced or reconstructed. In the first category are the lists of rune-names found in manuscripts from the eighth century on, as well as the rune-poems, for which the manuscript sources are somewhat later. The second category comprises not so much sources of actual rune-names as sources from which it has been deduced that rune-names existed before they are recorded in manuscripts. I find it convenient to keep these two categories separate. 1.5.1 Sources recording rune-names or their interpretation Sources recording rune-names or their interpretation include, on the one hand, manuscript lists which give the graphical forms of the runes together with their names but do not give their meanings, and, on the other, the

300 Sims-Williams has not linked the foilchesta and forfeda to the creation of ogam- names in general, but he has suggested (1992: 35) that: ‘The medieval reinterpretation of the forfeda as diphthongs and/or vowel digraphs is therefore comparable with the creative reinterpretation by which the redundant thirteenth and fourteenth symbols of the basic alphabet, which originally denoted the obsolete sounds /gw/ and something like /ts/ or /sw/, were reinterpreted so as to represent the phoneme /þ/, which arose in the Old Irish period, and the letter Z. ... Although it has been suggested there was a deliberate attempt to create equivalents for the Latin grammarians’ H, Q, and agma (NG), the case is far from conclusive; H and Q (/kw/) may be the original Irish values of the sixth and tenth letters, passively preserved, while the re-use of the redundant thirteenth letter as a symbol for /þ/ may have been inspired by the rise of that phoneme in Old Irish itself in the eighth or ninth centuries.’ Discussion of Sims-Williams assignment of specifically Irish phonemes to the forfeda (such as /eä/, /æ(:)/, /:/, //, /ll/, /rr/, as well as /x/ and /p/) would digress beyond the confines of the present thesis.

101 A Family of Names

Anglo-Saxon and Scandinavian rune-poems, which do not always include the name-forms, or have had them inserted later, while the meanings of the names are indicated in sometimes enigmatic stanzas or kenning-like descriptions. To compare rune-names with ogam-names and alphabet letter- names, I have needed to collate both types of information, i.e. both name- forms and the meanings. However, I must emphasize that it has not been my intention to provide a comprehensive collection of manuscript lists or new editions of the poems. The store of runes and rune-names in manuscripts is being added to year by year, building on Derolez’s pioneering work published in his Runica manuscripta: the English tradition (1954). I have taken account of some of the new discoveries, as well as some of the Scandinavian lists, which Derolez did not include in his initial work. As for the poems, good editions have been published in recent years by, for example, Ray Page, Maureen Halsall and Alessia Bauer,301 and I make grateful use of these, as I have made use of Damian McManus’s work on ogam-names in §1.4 above. My purpose is not to compete with any of this work, but to provide adequate material in a suitable format, as a basis for my arguments in Part 2. (a) Sources of manuscript lists For the lists of rune-names in early manuscripts, it is convenient to follow the categorization adopted by Derolez in his Runica manuscripta. He divided the material he connected with the English tradition into (1) lists of names ordered according to the English fuþorc, and (2) rune-names in alphabetical order. Among the fuþorcs, he further identified a group associated with a tract on the cryptographic use of runes, which is generally known as the isruna tract, while he associated a large number of the runic alphabets with various versions of a text describing ‘the invention of letters’: De inventione litterarum.302 To this material I have added a selection of Scandinavian lists. Details of the manuscripts containing lists of rune-names are listed in Table 4, while the lists themselves are given in the tables listed below. As Derolez pointed out, the names in the runic alphabets appear to have been largely derived from English fuþorc names, with some German influence. What he did not do, however, was to take the names of the alphabet runes connected with the De inventione tract and to relate their different spellings to the variant versions of the text. This I have done in Table 30 (a), which as a result highlights individual scribal preferences.

301 Page 1998: 1–36 (Icelandic poem); Page 2003b: 553–66 (Norwegian poem), Halsall 1981 and Bauer 2003a (Anglo-Saxon poem and versions of the Norwegian and Icelandic poems). 302 The title De inventione litterarum is that used by Derolez (1954: 279 ff.). Goldast (1606) entitled it De inventione linguarum ab Hebraea usque ad Theodiscam, & notis antiquis. This edition appeared on pages 91–3 of Volume II of Goldast’s Alamannicarum rerum scriptores aliquot vetusti. Derolez (1954: 304) notes that ‘Goldast’s original does not seem to have come down to us’, and that ‘no title occurs in any versions [of the tract] that have come down to us’. The various versions of the text (given by Derolez, 1954: 349–59) clearly refer to the invention of letters in the different languages. For convenience, I abbreviate the title to De inventione.

102 Part 1: Sources of Name-forms and Meanings

1. Names of runes ordered according to the English fuþorc (a) not associated with the isruna tract (including those added to the Anglo-Saxon Rune Poem) Table 27 (a) (b) associated with the isruna tract Table 27 (b) 2. Names of runes in the order of Scandinavian fuþąrks (including those added to some of the rune-poems) Tables 28 (a) and (b) 3. Names of runes listed in alphabetical order (a) not associated with the De inventione tract Table 29 (a) (b) associated with the De inventione tract Table 29 (b) 4. Distribution of variant spellings of the rune-names over the different versions of the De inventione tract Tables 30 (a) and (b)

(b) Sources of the rune-poems, their form and dates Complete texts of the Anglo-Saxon, Norwegian and Icelandic rune-poems are given in Table 31. The presentation is in no way intended to imply that the texts represent either original or definitive versions. They are the result of comparing and selecting from various editions with the purpose of providing a basis for discussion of the perceived meanings of the rune- names. Variant readings and translations are given in the notes, but the choice of one particular variant for the main text does not necessarily mean that it is to be preferred over the one given in the notes. The choice has been made in order to present a running text. In some cases, however, the variants provide important information, which will be taken up in the course of the discussion of individual names. At the start of Table 31, I summarize the main sources from which the texts have been assembled.303 In view of the complex history of this material, however, I give more details below for each poem in turn, together with a brief analysis of its form and general content, as well as possible dates for its composition. In her work on the rune-poems, Bauer also treats the Abecedarium Nordmannicum as a poem and includes the Swedish rune-poem.304 I have not included the Swedish rune-poem as one of my sources for rune-names, because the text is incomplete and the names are largely derived from the Norwegian and Icelandic poems.305 I shall refer to it where it appears to add a particular insight into a name. As for the Abecedarium Nordmannicum, although I include this text among the list of poems in Table 5, I agree with Musset that its ‘poetic’ nature is debatable306 and will discuss it in the present context in connexion with the evolution of the Scandinavian fuþąrks.

• The Anglo-Saxon Rune Poem For the Anglo-Saxon poem in Table 31, the basic text is that copied from London, British Library, Cotton Otho B. x and reproduced by Hickes on page 135 of his Linguarum vett. septentrionalium Thesaurus grammatico- criticus et archælogicus (see Table 32 (a)). The original Cotton manuscript was damaged in a fire in 1731, in which the folio containing the poem, 165v, was among the pages completely destroyed, leaving Hickes’s reproduction as the oldest source for the poem. However, there is some doubt as to 303 I list the same sources in Table 5 as part of a general summary of sources for this thesis. For the numerous secondary sources for the Norwegian and Icelandic poems in particular, see Bauer 2003a: 125–31 and 179–83. 304 Bauer 2003a: 58–77 and 209–33. For the Swedish poem, see also Quak 1987: 81– 92 305 See the discussion in Bauer referred to in the previous footnote. 306 Musset 1965: 117–18. However, von der Leyen (1962: 153–60), Arntz (1935: 104) and Düwel (2008: 191–2) all treat the text as a poem.

103 A Family of Names whether the folio originally belonged to the codex.307 The material down the lefthand side – consisting of the sound-values of the runes, the rune-signs in fuþorc sequence with several alternative forms, and the rune-names, apparently inserted in a smaller hand – was printed from a copper plate, while a second copper plate was used to print the material at the bottom of the page below the text of the poem. Although it is not certain whether originally a rune-name was included at the head of each stanza of the Anglo-Saxon poem when written, it is clear from the metric scheme that a name formed part of the line when spoken. It is possible that the names inserted beside the rune-signs on folio 165v of Hickes’s version of the Anglo-Saxon poem, as well as some of the rune- forms, were taken from another manuscript. There are a number of similarities with the fuþorc in British Library, Cotton Domitian A. ix, fol. 10v, which Hickes also published (see Table 32 (b)), but there are too many differences to infer that one was copied from the other. The layout of the Anglo-Saxon poem in Table 31 follows that of most modern editions, with stanzas consisting of two to five lines and each line comprising two alliterating half-lines. Hickes’s presentation is not in stanza format but does divide the text into separate sections for each rune and places a rune-form, with alternative forms and sound-values, in the margin next to the relevant section. In the original manuscript, the text was probably in scripta continua, without rune-by-rune division, but the occurrence of a rune-form in the text would have indicated the start of a new ‘section’. Why the number of lines per stanza varies is unclear. A division into stanzas, while not unknown, is rare in Old English poetry, but in the present instance it is understandable in that each stanza of the poem relates to one rune. In her analysis of the poem,308 Halsall comes to the conclusion that each stanza is ‘essentially discrete’, with ‘few overt connections between one stanza and the next’ and ‘no thematic significance for most aspects of the stanza order.’ In pointing to the apparent irregularity in the number of lines assigned to each of the runes, she notes that the first eight stanzas all consist of three lines each (lines 1–24 in Table 31-2 to31-5). This latter observation leads her to suggest that the stanzaic structure might be linked to the traditional division of the older 24-rune fuþark into its three ættir or ‘families’, of eight runes each, to which the five extra vowels were added at the end of the Anglo-Saxon version (see Table 33 (a)). Unfortunately, her explanation for the subsequent stanzas following the first eight is tortuous, to say the least. According to her, the opening of the second œtt ‘is marked unmistakably as a new beginning by the crowded hypermetric lines of the first two stanzas, which consist of two lines each. The next four stanzas consist of three lines each, followed by a four-line stanza, which is supposed to ‘thrust into prominence’ the next two runes, i.e. the final rune of the second ætt (the s) and the first rune of the third ætt (the t). There are then five four-line stanzas, followed by two three-line stanzas at the end of the third ætt, which she says have the effect of emphasizing ‘by contrast the two final runes of the original Germanic fuþark’. Finally, the beginning of the last set of five stanzas relating to the additional Anglo-Saxon runes is

307 Bauer 2003a: 80. For the text of the poem, Hickes used a font specially designed by Franciscus Junius for the publication of his 1655 edition of Oxford, Bodleian Library, Junius 11, commonly known as the Cædmon manuscript (see Lucas 1992: 177–97). 308 Halsall 1981: 49–51.

104 Part 1: Sources of Name-forms and Meanings signalled by a four-line stanza for the ac rune and rounded off by ‘five full and complexly constructed lines.’ The ad hoc pleading required by Halsall to justify this analysis is far from satisfactory. An alternative suggestion, which not only takes account of the number of lines per stanza, irrespective of the division into ættir, but also considers the thematic content, would be to divide the poem into four unequal sections (see Table 33 (b)). The first section consists of four three- line stanzas, followed by a section of eight three-line stanzas divided in the middle by the pair consisting of the only two-line hypermetric stanzas in the composition. This second section starts and ends with stanzas on subjects, rād and peorð, whose meanings are not quite clear as such, but which both refer to activity ‘in the hall’: on recyde, ‘in the hall’, and on beorsele, ‘in the beer-hall’, respectively. The rād stanza is followed by the cēn stanza, while the peorð stanza is preceded by the eōh stanza. The text specifically connects both cēn and eōh with fire: cēn, the torch, and eōh, the guardian of fire, hyrde fyres. The gyfu stanza, following cēn, and the gēr stanza, preceding eōh, also echo each other’s subjects in that both are about generosity: the first, gyfu, the generosity of men; the second, gēr, the generosity of God. If there is a relationship between the wen and īs stanzas, it is not clear; but the two central two-line stanzas in this section both refer to the theme of transformation: one physical (hœgl, hail turning to water) and the other spiritual (nyd, oppressive need transformed into a source of help and salvation). The third section of the poem shows a similar symmetry to the second. In §2.4.3. I will argue that the rune-names eolhx secg and sigel reflect patristic interpretations of the Hebrew letter-names sade and zain, i.e. iustitia and oliva, respectively; the secg [sedge] or [sword], represents the sword of justice, while the sigel [seal] represents the seal of the Holy Ghost, symbolized by the olive branch brought by the dove as a sign of salvation to Noah in his ark, as appears to be described in the poem’s verse. Justice was a cardinal virtue, so that it is appropriate for the justice represented by secg in the first stanza of the third section of the poem to be balanced by a second cardinal virtue, fortitude, as symbolized by the āc [oak] (cf. Latin robur [oak, strength, vigor]), in the last stanza of the section, while the light and hope (of salvation) implicit in the sigel is echoed by the light and hope of the dæg. In the middle of the third section stands ‘man’ in his frailty, flanked by stanzas referring to journeying, either on land (on horseback) or on the sea (on a courser of the deep). Parallels between the other stanzas are also discernible but discussion of these would be too much of an excursus at this point. The final section of the poem reflects the first section of four three-line stanzas, in that it consists of three three-line stanzas with a final ‘heavy’ stanza of five (i.e. three plus two) lines on the burial of the corpse. The subject of the ēar stanza may have influenced the switch in the positions of the ēar and iar runes found in other rune-rows so that the grave was given a suitable finality, but there is also a parallel discernible between the second and second-to-last stanzas of the whole poem: ūr, a wild creature, inhabiting the moorland or heath, and iar, a wild creature inhabiting water but feeding on land (see further §2.4.6). The implication of this analysis is that the poem is not a disparate collection of unconnected stanzas purely dependent on the sequence of the runes in the fuþorc. It is a complexly structured composition with a thematic symmetry in which the varying stanza lengths point up the stucture and

105 A Family of Names themes. Moreover, the structure not only ignores the elder fuþark’s system of ættir — it fully integrates the final five runes characteristic of the Old English fuþorc, which would appear to confirm that the poem was composed in its present form after the creation of the five supplementary runes, although it cannot be ruled out that the main part of the poem could have been based on an earlier version.309 Various linguistic features point to a date in the tenth century for the poem to have been written down but they do not necessarily pinpoint its composition.310 More significant for dating the composition is the alliteration of gyfu and gumena in line 19 of the poem, which from the tenth century on would have been phonemically distinct, with respectively non-alliterating palatal and velar /g/.311 Dobbie sees the alliterating structure of the poem, as opposed to the use of end-rhyme, as indicating a date for composition in the eighth to ninth century.312

• The Norwegian rune-poem The text of the Norwegian poem in Table 31 is largely based on the versions given by Page which he derives from the transcriptions in Copenhagen, Kongelige Bibliotek, E Donatione Variorum I. Barth. D, p. 818 (in the hand of Árni Magnussón; c. 1686–9) and Stockholm, Kungliga Biblioteket, Papp. fol.64, part IV, p. 74 (in the hand of Jón Eggertsson; c. 1686–7). 313 These two transcriptions are almost identical and were almost certainly made from the same manuscript, namely one consisting of several loose folios of parchment which were lost in a fire in Copenhagen in 1728. Both of these versions give a rune-form at the start of each stanza, but no names. Another important transcription, also given by Page, is that made earlier by Ole Worm for his RunéY seu Danica Literatura Antiquissima, Amsterdam, 1636, pp. 105–7 (2nd ed. Copenhagen, 1651, pp. 95–7). This version differs from the other two in many respects and is considered less reliable, the differences being attributed to Worm’s alledged inadequate command of Norwegian.314 Each stanza of the poem in Worm’s edition begins with a rune-name, but the forms of these names are questionable (cf. the names listed in Table 28 (a); the stanzas on m and l are reversed by Worm). Based on Worm’s edition, but with emendations, was a dissertation published by Rúnolfur Jónsson, Linguæ septentrionalis elementa tribus assortibus adstructa, Copenhagen, 1651, signatures B3v–4r (where not only m and l are reversed, but so are r and o). Several other transcripts of the Norwegian poem which appear to have been based on Worm’s or Jónsson’s edition (because of the accessibility in print315) are listed in a footnote to Table 5.

309 Bauer 2003a: 92: ‘Es is jedoch nicht auszuschliesen, dass der Text ursprünglich nur die 24 Runen der älteren Fuþark behandelte, und dass alle weiteren Strophen erst zu einem späteren Zeitpunkt als Ergänzung dazu gedichtet wurden.’ [It cannot be ruled out, however, that the text originally treated only the twenty-four rune of the elder fuþark, and that the extra stanzas were composed only at some later date as an addition.] 310 For details, see Bauer 2003a: 92–3. As Bauer points out, a large proportion of Anglo-Saxon poetry was not written down until the tenth century, although composed earlier. This circumstance would undermine Halsall’s argument for a date for the composition of the poem at the end of the tenth century (Halsall 1981: 32). 311 Luick 1914–40, II: 922–3. 312 Dobbie 1942: xlix. 313 Page 2003b: 553–66. 314 Bauer 2003a: 117, 125. 315 Bauer 2003a: 117, 125–31.

106 Part 1: Sources of Name-forms and Meanings

Structurally, the Norwegian rune-poem consists of one stanza for each of the sixteen runes of the Scandinavian fuþąrk, with each stanza comprising two half-lines connected by alliteration and, with two exceptions, end-rhyme – the two exceptions being the stanzas on the k- and l-runes. The use of end- rhyme is in contrast to the Anglo-Saxon poem. In all but two cases, the half- lines are complete in themselves – the two exceptions again being the stanzas on the k- and l-runes, where the first half-line runs over into the second. Although the rune-names are not explicitly recorded in either Magnússon’s or Eggertsson’s transcripts, they form the subject of each of the first half-line of every stanza except that on the i-rune, where the name is the object of kollum [we call]. The second half-lines, however, would appear to have no meaningful connexion with the first,316 although Clunies Ross has argued, on the basis of the stanza on the f-rune, that each second half-line has a metaphorical association with the first.317 I shall discuss this idea in my consideration of the individual rune-names in Part 2. Besides the employment of end-rhyme, another feature that distinguishes the Norwegian poem from its Anglo-Saxon counterpart is the number of references to pagan mythology alongside clearly Christian motifs, which has led Bauer to use the term ‘Mischdichtung’.318 She also notes that the poem appears to fall into three parts, following the fuþąrk’s division into three ættir (cf. Halsall’s analysis of the Anglo-Saxon poem described above). According to her, the first six stanzas, corresponding to the first ætt, deal with matters related to heathen mythological themes, including in particular the Niebelungen saga, with predominantly negative connotations associated with evil in the world. The next five stanzas, corresponding to the second ætt, reflect a positive Christian outlook, beginning with the reference to Christ in the h-stanza and including the peaceful reign of the Danish king Fróði, while the last five stanzas return to heathen themes involving the wolf Fenrir and Loki and ending with a sign of the end of the world implicit in the burning of the world-tree. While I find this analysis useful in highlighting the poem’s syncretic treatment of Christian and heathen themes, which may be linked to the revival of interest in pre-Christian mythology in Scandinavia from the thirteenth century on (as opposed to a revival of actual pagan religion),319 I discern a number of other elements that may have provided a thematic structure to the poem. In particular, it is the theme of each second half-line that is relevant here. These themes for the first ten stanzas can be represented schematically as follows:

316 On page 141 of his Runologia (AM 413 fol., written 1732, revised 1752), Jón Ólafsson describes the ‘couplets’ of the Norwegian rune-poem Tvý-deilur and observes that the first half-line is simple to interpret (semper analogia est nuda) but the second half-line is much more obscure (obscurior, atque longe petita) – see Bauer 2003a: 131. 317 Clunies Ross (1990: 23–39) has even taken the exceptional k and l stanzas, in which the first half-line runs over into the second, as possible examples of the technique which Snorri (in Háttatal 13, Finnur Jónsson, 1931: 221–2) describes as hjástælt [abutted], where the first word of the last line belongs to the previous sentence both in sense and metrically (i.e. alliterativelly) but the rest of the line is a proverbial statement, separate from what has gone before and providing a wider vision. 318 Bauer 2003a: 135–6. 319 See Bauer 2003a: 135, footnote 252, especially the reference to Lange 1958: 163– 5.

107 A Family of Names

Stanza Theme Stanza Theme Stanza Theme

1 Animal (wolf) 3. Misfortune (to women) 7. Christ 2. Animal (reindeer) 4. Swords 8. Vulnerableperson (naked) 5. Swords 9. Vulnerable person (blind) 6. Misfortune (to children) 10. Fróði

The structural elements in the last six stanzas are more opaque, but some parallels can be discerned:

Stanza Theme

11. ? Divine judgment 12. Smith – Fenrir wolf 13. (First half-line) tree 14. ? Hawk’s claw 15. Gold – Fafnir dragon 16. (First half-line) tree

The connexion between stanzas 11 and 14 is dubious, while that between stanzas 12 and 15 requires particular knowledge of the associations between Tir and Fenrir, on the one hand, and gold and Fafnir, on the other. Also, the linking of the first half-lines of stanzas 13 and 16, rather than the second half-lines, breaks the pattern. The employment of end-rhyme alongside an alliterative scheme has been used by Lindroth to date the poem.320 End-rhyme entered Nordic poetry under the influence of ecclesiastical Latin poetry, which leads Lindroth to suggest a date between 1200 and 1337, the terminus ante quem put forward by Kålund for the original transcription on the basis of texts prefacing the poem in Árni Magnússon’s manuscript.321

• The Icelandic rune-poem and the question of a common source The text of the Icelandic poem in Table 31 is, with one or two exceptions, the edition presented by Page, which in turn is based on versions in three manuscripts in Reykjavik, Stofnun Árna Magnússonar, namely AM 687d 4to (c. 1500), AM 461 12mo (1539–58) and AM 749 4to (s. xvii).322 The version in AM 687d 4to has two sets of Latin glosses, on folios 1v and 2r (see Table 34); that in AM 461 12mo is written in scripta continua with a rune-name with initial capital letter indicating the start of each stanza; on the other hand, the version in AM 749 4to arranges the rune-names and stanzas in alphabetical order rather than the usual fuþąrk order. Runólfur Jónsson included a version of the Icelandic poem in his 1651 dissertation, with not only Latin glosses on the names but also a Latin translation of the stanzas. A version that does not actually reproduce the poem as such but gives a 320 Lindroth 1913: 288. 321 Kålund 1884–91, I: 3. 322 Page 1998: 1–36. Bauer (2003a: 165–7) bases her edition on AM 687d 4to. However, the sequence of the last five stanzas in this manuscript (see Bauer 2003a: 266, Fig. 4) follows the rune sequence lbmt= (lbmty) rather than the usual tbml= (tbmly). It would appear that in the manuscript the runes t and l (t and l) have been transposed. Bauer follows the manuscript sequence in her German translation, but transposes l and b (l and b) in her transcription of the Icelandic. In my own transcription and translation in Table 31, I have followed the usual sequence, bmlty, which also corresponds with the Norwegian poem as transcribed by Jón Egertsson and Árni Magnússon.

108 Part 1: Sources of Name-forms and Meanings selection of the periphrastic descriptions for each name, together with Latin translations of the names and descriptions, is that of Olaus Verelius in his Manductio Compendiosa ad Runographiam Scandicam Antiquam recte intelligendam. En kort underwysing om then Gambla Swea-Gotha Runa ristning. Codex Argent Izvis atgiban ist kunnan Runa þiudangardjos guþs, Uppsala (1675: 24–33). I have not included a verbatim copy of either Jónsson’s or Verelius’s versions,323 but I give the Latin glosses on the names in Table 34 and quote from the poems in Part 2 where they add to Page’s edition. Structurally, the Icelandic poem is composed of sixteen stanzas, each consisting of a þrídeilur, that is a set of three two-term periphrastic descriptions of the rune-name. Like the bríatharogam [word-ogam] used to describe an ogam-name (see §1.4.1), the type of description in the poem is not a kenning in the strictest sense as defined by Meissner,324 but in the looser sense allowed by Snorri.325 The first two kennings of each stanza alliterate with each other, while the third kenning alliterates internally, and there is no end-rhyme. This structure is quite different from that of the Norwegian poem and, superficially, gives the impression that the Icelandic poem is the more archaic of the two, although it could have been the result of a considered attempt to create an archaic effect. Despite the structural differences between the two poems, however, the descriptions in ten of the Icelandic stanzas include virtually identical wording to those in their Norwegian counterparts (see Table 35 (a)). On the other hand, the text in six of the remaining stanzas is quite different – reið, ís, úr, óss, sól, ýr – and in three of those instances – úr, óss, ýr – the texts are not only different but the names are interpreted as having different meanings. In this light, I agree with Bauer that the Norwegian and Icelandic poems were structurally independent compositions but drew on a common source of descriptive phrases for ten of the rune-names. When the Anglo-Saxon poem is included in an analysis of this nature, the comparison becomes more complex. In Table 35 (a), it will be seen that two of the Anglo-Saxon stanzas – hægl and gēr – contain almost identical wording to the Nordic stanzas – hagall and ár – while the wording in another two stanzas – beorc and man – is similar to that in their Nordic counterparts. Bauer compares the description of ‘hail’ in the hagall/hægl stanzas as ‘cold(est) grain’ to lines 32b–33a in the Old English poem of The Seafarer: hægl feol on eorþan | corna caldast [Hail fell on the earth, the 323 Verelius’s text and translations of the Icelandic poem are given by Bauer 2003a: 174–7. Integrated with the Icelandic texts are also names and texts which more or less correspond with eight of the stanzas of the Swedish rune-poem which Bauer gives separately (Bauer 2003a: 219–20). 324 Meissner (1921: 12): ‘Wesentlich für die Kenning ist, daß sie als Ersatz empfunden wird und als solcher etwas allgemeingültiges, typisches, variables hat.’ [The essence of a kenning is that it is felt to be a substitute and as such has something universally valid, typical, variable about it.] 325 Faulkes (1998: xxxiv): ‘Snorri uses the term “kenning” to refer to a structural device, whereby a person of object is indicated by a periphrastic description containing two or more terms (which can be a noun with one or more dependent genitives or a compound noun or a combination of these two structures).’ See also Meissner (1921: 15): ‘Die nord. Dichtersprache bedient sich neben der eigentlichen Kenning zweigliedriger Ausdrücke, wie sie allgemein in der germ. Dichtung zur Erhöhung der Rede verwendet werden.’ [Besides actual kennings, Nordic poetic language employs two-part expressions such as are used generally in Germanic poetry to heighten speech.]

109 A Family of Names coldest of grains].326 The references to earth and dust in the man/maðr stanzas, moreover, recall biblical connexions between man and the earth.327 Biblical or Christian references are not identifiable elsewhere in the Icelandic poem, unlike the overt Christian allusions in the Norwegian and Anglo-Saxon versions. In other words, the Icelandic poet appears to distance himself from a Christian tradition. Even in stanzas where the descriptions are quite different, similar themes, and hence meanings, are attached to the runes – feoh [wealth], nyd [need], lagu [water], rād [journey (on horseback)], īs [ice]; in one instance – Old Norse ýr, Old English eōh – an interpretation of yew-tree is attached to quite different rune-names but includes the same association with fire in both the Norwegian and Anglo-Saxon poems but not in the Icelandic version (Table 35 (b)).328 At the same time, four rune-names – tīr/týr, ūr/úr, ōs/óss, yr/ýr – have similar forms but relate to quite different themes in the Nordic and Anglo-Saxon poems, and in three cases both name-form and theme are different – þurs/þorn, kaun/cēn, sól/sigel – although attempts have been made to link the themes of these names, as I discuss in Part 2. According to Bauer, the structural differences between the poems are too great to justify the idea of a common German rune-poem as such, from which the existing versions in the various poems will have derived, but the similarities and differences between the poems could be accounted for by a common oral source for the descriptions of the rune-names. 329 The assumption of a common oral source would allow not only for poetic structures being adapted to the conventions of the environments in which written versions of the poems were composed, but also for atrophy in remembrance of the meaning of rune-names like ūr/úr, kaun/cēn and yr/ýr, and for the suppression or transformation of certain pagan meanings of names such as ós, týr, þorn and, perhaps, sól in a Christian context. However, the assumption of a common oral source for the descriptions of the rune-names dating from before the composition of the Anglo-Saxon poem, and therefore presumably pagan in origin,330 is not the only possible explanation for the observed similarities and differences in the poems. For instance, Bauer acknowledges that the Icelandic poem’s pagan imagery cannot be taken to imply that it had pagan roots; it is more likely that it was following the Icelandic skaldic tradition of exploiting pagan mythology for poetic effect.331 The same would also apply to the pagan references in the Norwegian poem. In this light, I would suggest that, rather than all three poems being independent compositions drawing on a common oral source,

326 Bauer 2003a: 197, citing Krapp and Dobbie 1936: 144. 327 Gen. II.7: Formavit igitur Dominus Deus hominem de limo terrae. [D-R: And the Lord God formed man of the slime of the earth.] Gen. III.19: In sudore vultus tui vesceris pane, donec revertaris in terram, de qua sumptus es; quia pulvis es, et in pulverem reverteris. [D-R: In the sweat of thy face shalt thou eat bread till thou return to the earth, out of which thou wast taken: for dust thou art, and into dust thou shalt return.] 328 Bauer (2003a: 103) compares the Anglo-Saxon poem’s reference to fire in this stanza with riddle 30a from the Exeter Book (Krapp and Dobbie 1936: 195–6). This riddle, however, is not specifically concerned with a yew-tree but with a tree in general. 329 Bauer 2003a: 237–9. 330 Bauer (2003a: 94) implies that the Anglo-Saxon poem belonged to the same pagan tradition as the Nordic poems and was reformulated in a Christian context. 331 See Bauer 2003a: 135, footnote 252.

110 Part 1: Sources of Name-forms and Meanings

the Anglo-Saxon poem was the oldest rune-poem, based on rune-names derived from ogam-names and alphabet letter-names, composed in a Christian context, and that the rune-names and the idea of composing a rune- poem was transmitted by the Anglo-Saxon missions to Scandinavia,332 where the poem was given characteristic (and partly archaic) Nordic forms, with names to match. (c) Choice of rune-name interpretations as a function of the rune-poems’ purposes It is commonly held that the purpose of the rune-poems was to provide a mnemonic aid to learning the names of the runes in the order of the fuþark. Bauer’s conclusion exemplifies this view:

Die Runengedichte dienten dem Erlernen der Runennamen in ihrer Reihenfolge. Allen Runengedichten liegt ein auf dem Stabreim basierendes mnemotechnisches Verfahren zugrunde, das das Einprägen der Runennamen erleichtern soll. [The purpose of rune-poems was to learn the rune-names in their set order. All rune-poems rely on a mnemonic technique based on alliteration, aimed at making it easier to memorize the rune-names.]333 However, an alliterative rhyming structure was typical of Anglo-Saxon and Nordic poetry, and not specifically a mnemonic technique. Also, most early transcriptions of the poems – those usually considered the most authentic334 – contained only a rune at the head of each stanza and not an actual name- form, which would not have made it easier to learn the names as such. There are also no obvious connexions between the stanzas to help remember their sequence, as there is in the case of Jerome’s connexiones (Table 6), where the letter-names are connected sequentially to form a memorable (Christian) message. Even if the stanzas of the rune-poems are interpreted in terms of general concepts that might represent a Germanic world-view, as has been suggested by Elliott and Polomé,335 it is difficult to discern in the order of the stanzas any rational structuring of these concepts which may have helped in remembering the order of the runes in the rune-row. A learner may have fixed the order in his (or her) mind by memorizing the whole poem, but in that case, the interpretation of each stanza would have served only to aid recall of the relevant name that will have to have been learnt previously by some other means. A poetic structure, as opposed to a straightforward list, would certainly have been an aid to memory. Also, it is not inconceivable that the gnomic, riddling nature of the stanzas may have been intended as instrumental in a teaching process. In a discussion of a pedagogical dialogue between Alcuin and Charlemagne’s son Pippin – Disputatio Pippini regalis et nobilissimi iuvenis cum Albino scholastico336 – Reuschel pointed to the kenning-like nature of the disputation,337 which comprises 101 questions and answers, 332 For an account of the Christianization of Norway from AD 900 onwards, see Sanmark 2004. 333 Bauer 2003a: 236. Cf. Page (2003a: 63): ‘Each [stanza] begins with a rune whose name the rest of the stanza expounds, so that the whole comprises an elementary guide to the fuþorc, presumably designed to help the memory. The Anglo-Saxon Rune Poem is not unique, for there are Norwegian and Icelandic parallels, suggesting that this was a common mnemonic verse type.’ 334 That is, those discussed in the previous section, with the exception of Worm’s version of the Norwegian poem and versions based on it. 335 Elliott 1959/80: 45–61; Polomé 1991: 421–38. 336 PL CI: 975–80. 337 Reuschel 1938: 143–55

111 A Family of Names along such lines as:

Quid est homo? Mancipium mortis, transiens viator, loci hospes. Quid est sol? Splendor orbis, cœli pulchritudo, naturæ gratia, honor diei, horarum distributor. Quid est gelu? Herbarum persecutio, vinculum terræ, fons agrorum. [What is a man? Slave to death, transient wanderer, guest of the earth. What is the sun? Ornament of the (world’s) orb, beauty of heaven, esteem of nature, honour of daytime, distributor of the hours. What is ice? Persecutor of plants, fetter of the earth, source for fields.] Bauer suggests that the dialogue shows parallels with the Icelandic poem, which would appear to be composed of similar answers in two-element periphrastic form, with the implicit question in each case: what is the rune- name that fits this description? Other themes dealt with in the Disputatio include ‘rain’, ‘summer’ and ‘water’, as well as some reflecting the subjects of rune-names from the Anglo-Saxon Rune Poem, such as ‘day’ and ‘hope’, which might indicate a Latin source of this nature as inspiration for the rune- poems in general. The themes, however, are so wide-ranging that it would be surprising if there was no overlap between the themes of the Disputatio and those of the rune-poems. The periphrastic style is certainly common to both the Latin and Icelandic texts but is not sufficiently unique to adduce a direct connexion between the two. However, memorizing the names and order of characters was not the only purpose of texts dealing with the alphabet. As well as helping a potential learner to memorize the names, supposed meanings and alphabetical order of the Hebrew letters, Jerome’s device of connecting the names in meaningful sequences also had a didactic intention, similar to that of Eusebius, which implied that the letter-names contained divinely inspired messages. Ambrose, too, had a decidedly didactic approach when he used the Hebrew names in the tituli of the psalms to provide a structure for his homilies in his Expositio in Psalmum Davidis CXVIII (Table 7). On the other hand, Jerome’s De psalterio glosses (Table 8) had a completely different approach: the listing of the names together with different meanings was in the form of a dictionary, with the lemmata arranged not according to the order of the Hebrew alphabet but according to the alphabetical order of the transcriptions into Latin. These differences in approach – didactic or encyclopedic – reflected the authors’ different intentions, which in turn affected their choice of interpretations. Because of the need to transcribe the names into Latin, even the name-forms varied in Jerome’s De psalterio listing. Other possible intentions of the authors of alphabet texts comparable to the rune-poems might have been to display the authors’ knowledge, their imagination or their artistry. The Carolingian poem on the Hebrew letter- names given in Table 14 (a) is an example of an author displaying his knowledge, with the addition of a Christian message at the end, and a similar poem on the Greek letter-names given in Table 17 (a) likewise displays the author’s knowledge and his connexions with Christianity. Ausonius’s poem, in Table 17 (b), is probably slightly more imaginative, while the Versus cuiusdam Scoti de alphabeto, in Table 17 (c)-1, overflows with erudition, as can be judged from the annotations given in Table 17 (c)-2. The Anglo-Saxon Rune Poem is not particularly erudite, but its riddling, gnomic style, without direct reference to specific names, would appear to invite the listener or reader to recall what is being referred to in a particular stanza. At the same time, most stanzas appear to have a didactic intention

112 Part 1: Sources of Name-forms and Meanings

other than teaching the names and order of runes in the fuþark, since most carry a decidedly Christian message. It is this second didactic intention that appears not only to have shaped the interpretations given to the rune-names in the poem but also to have determined the names themselves. In other words, the rune-names were not anchored in ancient pagan beliefs but were invented to match the tradition of Hebrew letter-names and were interpreted in the same way that Jerome, Ambrose and all those that followed interpreted letter-names, partly to support their Christian message and partly to display the interpreter’s erudition. In the case of the Anglo-Saxon Rune Poem, the intention was to present a decidedly Christian message in a gnomic style, whereas the intention of the Norwegian and Icelandic poets seems to have been more to present the runic material in an archaic erudite style, incorporating themes from pagan mythology with, in the Norwegian poem, a syncretic view of Christianity and the Nordic gods of the past in line with the fashion of ‘Mischdichtung’ of the time.338 1.5.2 Sources from which the early existence of rune-names has been deduced My hypothesis that there was no established system of rune-names before the arrival of Irish Christian missionaries in southern Scotland and Northumbria at the end of the sixth century is at odds with a widely prevailing assumption that rune-names existed at a much earlier date on continental Europe,339 long before they first occurred in the manuscripts referred to above. Evidence for such an assumption is drawn from three sources. The first is the Gothic alphabet, created by Wulfila as a script for his translation of the Bible. The names of the letters of this alphabet are recorded in a manuscript now known as Vienna, Österreichische Nationalbibliothek, Codex 795 (s. viiiex/ixin). They are thought to show evidence of having been influenced by runic practice, which has led even someone as cautious as Page to consider that ‘If the Gothic material in Vienna, National Library, MS 795 is to be believed – and there is some question about this – letter-names of the Gothic alphabet descend from a related system [i.e. related to Old English and Nordic names], and it then looks as if the bulk of the Anglo-Saxon rune-names go back to common Germanic times.’340 I am even more cautious than Page and would question the grounds for reconstructing early forms of rune-names from this Gothic material even more strongly than he does. A second source for deducing the early existence of an established tradition of meaningful rune-names is a number of early epigraphic inscriptions in Scandinavia and England in which individual runes appear to have been used as ideographs (Begriffsrunen), logographs or tachygraphs. The application of the term Begriffsrune to certain single runes in epigraphic inscriptions was developed in particular by Krause.341 In runic contexts, the German term is usually translated into English as ‘ideograph’, although strictly speaking, ‘logograph’ might be more precise. In other contexts, an ideograph applies to a graph that stands for a concept (Begriff), such as the

338 See Bauer 2003a: 135–6, especially footnote 252. 339 See, for example, Düwel 2008: 7–8; Spurkland 2005: 10–11; Nedoma 2003: 556; Polomé 1991: 421–38; Krause 1970: 26–31; Elliott 1989: 45–61. 340 Page 1999: 63. 341 Krause and Jankuhn 1966, passim; detailed references are given in the discussion below in §1.5.5 (b).

113 A Family of Names

Egyptian hieroglyph for ‘house’: . To understand this hieroglyph, it is not necessary to know the Egyptian word het [house]; the shape of the graph resembles a house. When, on the other hand, the rune F is interpreted as being used as a Begriffsrune, it is assumed to stand for the word feoh or fé, and as such is a logograph, although this logograph may have been intended to stand for the concept of ‘wealth’. When the rune m is used in the name ‘SALOm’ [Salomon] (in Solomon and Saturn I in Cambridge, Corpus Christi College, MS 41), it is an abbreviation for the word mon and does not refer to a man but is a form of spelling shorthand or ‘tachygraph’. As the ‘SALOm’ example illustrates, there can be no doubt that single runes were used as tachygraphs in manuscripts. It is questionable, however, whether they were used as ideographs in inscriptions predating their use as such in manuscripts. If they were, then this could imply that there was an established tradition of rune-names well before the advent of and well outside the ambit of any Irish missionaries. A third source for evidence from which to deduce the early existence of rune-names lies in the development of the Anglo-Saxon and Scandinavian rune systems as a whole. In this development – which was due to shifts in the sounds, particularly the vowel sounds, of the various dialects in Scandinavia, England and Frisia between AD 400 and 800 – rune-names are assumed to have played a role. That is to say, runes had already been assigned names before the sound changes took place, and, as a consequence of some of them being affected by the ensuing changes, new or different sound-values were attached to the relevant runes. For example, the original Proto-Norse name of the j-rune is assumed to have been *jara [year]. By 600, with the loss of initial j- and syncope of the final -a, this word had become ár. As a consequence, the j-rune came to be used to represent oral [a], while the original a-rune came to be used mainly for nasal [ã]. In the following sections, I examine (a) the evidence of runic influence in the names of the Gothic letters, (b) the development and possible use of runes as ideographs in early inscriptions, and (c) the possible role of rune- names in the re-assignment of sound values to runes as well as the creation of new runes. In dealing with ideographs, I focus mainly on the Scandinavian evidence, since there are few early examples of possible ideographs from Anglo-Saxon England. In dealing with the re-assignment of sound-values, however, I consider the Scandinavian and Anglo-Saxon evidence separately, since the re-assignments in the two regions were different. (a) The Gothic alphabet The names attached to the characters of the Gothic script used by Wulfila in his translation of the Bible are recorded on folio 20v of the manuscript known at one time as the Codex Salisburgensis, but now registered as Vienna, Österreichische Nationalbibliothek, Codex 795, and here referred to, for convenience, as Vienna 795. A summary of the history, make-up and contents of this manuscript is given by Derolez.342 The first section of the manuscript begins with two letters from Alcuin to a pupil of his, Arn, who was the former abbot of St Amand and later archbishop of Salzburg. Following the two letters there is a treatise on orthography attributed to Alcuin (fols. 5v–18v), then a Greek alphabet and syllabary and Roman

342 Derolez 1954: 52–6.

114 Part 1: Sources of Name-forms and Meanings numerals (fol. 19rv). On fol. 20r there is an Anglo-Saxon fuþorc, an incomplete Gothic alphabet and a cryptogram, and on the verso two Gothic alphabets: one parallelling the Greek order of the letters, with numerical values, the other closer to the Latin order of letters, with the ‘Gothic’ names of the letters. Fols. 5–20 appear originally to have been independent of the remainder of the manuscript. But it is fol. 20 in particular that is of interest in the present context. Von Grienberger343 and Rotsaert344 detected Visigothic rather than Ostrogothic traits in the material on the verso of this folio and suggested the involvement of a Visigoth from southern France, aided, in Von Grienberger’s opinion, by a German speaker from the circle of Alcuin around 800. Both Baesecke and Bischoff, on the other hand, thought the material may have been collected in northern Italy.345 Baesecke suggested that it may have then been brought to Germany by Arn around 799, on his way back to St Amand in northern France after receiving the pallium in Rome, whereas Bischoff saw in the material the hand of the Salzburg Cathedral librarian, Magister Baldo, who Bischoff thought might have copied or commissioned its copying while in the company of Arn in Italy, or while on an independent mission to Ravenna and Rome a little later.346 Either way, the manuscript appears to date from the end of the eighth or beginning of the ninth century. Treatment of the names of the Gothic characters on folio 20v of Vienna 795 has tended to encourage the impression that these names were unequivocally the Gothic names of runes. In Leiden University’s IEED project,347 for example, names from the Codex Salisburgensis, i.e. Vienna 795, are explicitly interpreted as ‘name of n-rune’, ‘name of b-rune’, etc. Düwel, in his treatment, is admittedly careful to emphasise that the names in the manuscript were Gothic Buch-stabennamen,348 but the impression created by his tabulation of rune-names (see Table 36)349 is that the Gothic letter- names are identifiable with rune-names. The list of rune-names presented by the TITUS project (Table 37)350 gives the same impression, although numerals beside the Gothic names are intended to indicate that the letters were arranged in the order of the Greek alphabet, rather than that of the fuþark. Even this annotation, however, is misleading, because there are two lists of Gothic letters on folio 20v of Vienna 795: one with the letters in the order of their Greek equivalents, the other in an order closer to the Latin sequence of letters (Table 38, columns 1 and 2). The letters in the Greek order are listed with numerical values, whereas it is the letters in the order 343 Von Grienberger 1896: 185–224. Derolez (1954: 57, note 1) agrees with Von Grienberger in assigning the text on folios 19 and 20 to the tenth century: ‘fols. 5– 20r, including the runes, must be dated in the tenth century; the Gothic material would be a still later addition.’ But he does not agree with Von Grienberger that the runic and Gothic material was attributable to Alcuin, on account of the ‘Northumbrian character’ of the rune-names. 344 Rotsaert 1983: 140. 345 Baesecke 1930: 156–7; Bischoff 1980: 76–83; 115–19. 346 Bischoff 1984b: 256–8. 347 Accessible at http://www.indo-european.nl/index2.html.ws The Gothic database in the Indo-European Etymological Dictionary (IEED) project is based on Köbler 1989. For the names in the Codex Salisburgensis, Köbler follows Krause 1970. For example, enguz is defined by Köbler as ‘a god, spear (N.)?, man (M.)?, name of n- rune [presumably, ŋ-rune].’ 348 Düwel 2008: 197. 349 Düwel 2008: 198–9. 350 Accessible at http://titus.fkidg1.uni-frankfurt.de/didact/idg/germ/runennam.htm.

115 A Family of Names closer to the Latin order that are provided with names. This observation is important in several respects:

1. The Gothic letter q, named quertra, appears in the ‘Latin’ list in the position of Latin Q, following P, Gothic p, pertra. In the ‘Greek’ list, on the other hand, it appears in the position of the Greek stigma/digamma, with the numerical value of 6. 2. The letter j, named gaar, appears in the ‘Latin’ list in the position of Latin G, whereas the same letter in the ‘Greek’ list is placed in the position equivalent to C, xi. 3. The name ezec/ezet is given to the Gothic letter z in the position of ‘Latin’ Z, whereas this letter in the ‘Greek’ list takes the position of Greek Z, zeta. 4. The name enguz is ascribed to Gothic c, which is placed in the position of Latin X, [ks], whereas this letter in the ‘Greek’ list is equated with Greek X, [x], chi. 5. The Gothic semivowel w, [w], named uuinne, is placed after t, which would agree with the placing of Greek U (upsilon), whereas one might expect it to be placed between c and z if the order was intended to be ‘Latin’. The Gothic w is followed by Gothic o, [u], so that there are two Gothic letter-names with initial ‹u›: uraz for u, in the ‘Latin’ position of O, with a shape and name related to runic U; and utal for o in the position of ‘Latin’ U, with a shape and name related to runic o, although according to the ‘Greek’ list, Gothic o is related to Greek W. 6. The placing of Gothic x, #/hw, uuaer, and v, þ, thyth, at the end of the sequence follows the common practice of adding ‘foreign’ letters at the end of an alphabet. But the fact that thyth appears in this position, and that the names fe and quertra are in the ‘Latin’ positions of F and Q, underlines the scribe’s apparent insistence on correlating the Gothic names with the Latin alphabet rather than the Greek. The only inconsistencies in this correlation are the placing of Gothic w after t and of Gothic g, geuua, in the Greek G, gamma, position, presumably under the influence of the graphic form. The nature of the connexion between the Gothic alphabet and runes, in relation to both the forms of the characters as well as the names, has been the subject of considerable debate, whence Page’s caution when broaching the subject (see §1.5.2 above). A detailed discussion of the letter-forms would be beyond the remit of the present thesis, but briefly it may be said that the characters are, as Grimm already pointed out,351 mostly based on Greek letters (see Table 38, column 3), with only three borrowings from Latin (R, S and F), only one possible borrowing from a runic list (the u), and two 352 characters of debatable origin (v and x). These last two characters are precisely the two characters that are placed at the end of the list of names in 351 Grimm 1821: 38–47. 352 Grimm points to four possible derivations of letter-forms from runes: Gothic o, u, w and v; he had no suggestion as to the origin of the x, #. In my opinion, the

116 Part 1: Sources of Name-forms and Meanings

Vienna 795, in the place traditionally reserved for ‘foreign’ additions.353 When siting the two letters g and w in the ‘Latin’ list, the scribe seems to have been influenced by the similarity of the letter-shapes to Greek forms rather than by the phonetc values associated with the letter-names. On the other hand, in three cases – f, j and z – the similarity to Latin letter-shapes seems to have taken precedence. In other words, the scribe seems to have vacillated between Greek and Latin alphabets as models. As far as the names are concerned, it is evident that many of the Gothic letter-names bear resemblances to rune-names, but it is debatable whether, for that reason, they bear witness to early Germanic forms of rune-names, or whether they were created at a later stage on the basis of Anglo-Saxon and Nordic models. The arguments in this regard divide broadly into (a) arguments by those who maintain that the Gothic names derive from early Germanic rune-names which are presumed to have been current in the fourth or fifth century, and (b) arguments by those who are sceptical of such claims. Representative of the sceptical standpoint is Blomfield’s conclusion that the so-called Gothic rune-names:

have no claim to represent runic nomenclature among the Goths in IV [i.e. the fourth century]. They do represent alphabet lore current among some Germanic peoples in VIII [the eighth century]; and their connection with a dialect which was then, as far as we know, obsolete makes it probable that they are riddled with antiquarian confusions. As a source of runic nomenclature they are worthless. Yet there are obvious points of contact. If it were possible to analyse the overlay of sophistication, some relation might be established between these names and rune-names.354 Derolez was of the same opinion as Blomfield, as were Marchand, Scardigli, Ebbinghaus and Hachmann.355 However, a robust defence of the idea that the Gothic letter-names were not merely adaptations of Anglo-Saxon and Nordic rune-names but show evidence of being specifically Gothic rune-names with early Germanic antecedents has been put forward by both Wagner and Mees.356 Their suggested derivations are summarized in Table 39. Mees agrees that nearly all the Gothic names ‘seem to match Anglo-Saxon names recorded here [in Vienna 795] and elsewhere, or where they vary, the Nordic names’, but suggests that ‘some of the Salisburgensis names retain the forms that seem to have been lost to the Anglo-Saxon tradition, but are retained among the Nordic names.’357 More significantly, however, he also concludes that a large number of the names ‘show phonological features that we might expect of a oand w are sufficiently close to Greek W and U graphically to justify a Greek derivation, whereas u, v and x are not obviously derivable from Greek or Latin with the phonetic values that are assigned to them in Gothic. Ebbinghaus (1979: 26– 8) considered that all Wulfila’s letters were derived from the Greek alphabet, agreeing with Boüüaert (1950: 423–37) that Greek cursive was used as a model. 353 Below the last character v there is a dot, and next to it, below the name, thyth, is written t · & h, indicating either that the character is to be written th or that it is equivalent to teth. 354 Blomfield 1937–45: 211–12. 355 Derolez 1954: lii; Marchand 1959: 277–91; Scardigli 1973: 238–9; Ebbinghaus 1990: 207–14; Hachmann 1993: 356–7. 356 Wagner 1994: 262–83; Mees 2002–3: 55–77. Similar arguments have also been put forward by Seebold 2001: 71–109. I refer to his article only where it differs significantly from or adds to Wagner’s and Mees’s views. 357 Mees 2002–3: 57.

117 A Family of Names late Gothic tradition’,358 and draws on a series of ‘consistent phonological developments’ that point to a derivation separate from the Anglo-Saxon names.359 These developments, according to Mees, are:

• voicing of *s > z (seen by Wagner as a sign of northern Gallo-Roman influence in the spelling360); • names which ‘quite patently’ retain the nominative endings that were rhotacised, syncopated or lost in all other Germanic dialects (e.g. uraz, eyz, enguz and daaz); • loss of postvocalic *g and *h (with resulting long -a-, as in daaz and laaz, as well as haal361); • unstressed u > a (e.g. uraz < *uruz and *laaz < *lagaz < *laguz); • īw, ew, iu spelled ey initially, y medially (e.g. eyz < *ewz < *ehwaz, thyth < þiuþ, tyz < [*tīws?] < *tīwaz); • lowering of i > e (e.g. geuua and enguz);362 • *ai > e (e.g. *bairkan * > bercna, *pairþra > pertra, *quairþra > quertra and *raida > reda). It is important to consider carefully the implications of the argument that Mees is putting forward, which is that (originally nameless) Gothic letters were provided with names from an external source, i.e. not names relating to the alphabet as such, but names derived from an established set of Proto- Germanic rune-names. This list of names and letters was then arranged alphabetically, not according to what is assumed to be the original order of the Gothic alphabet, i.e. the Greek order, but according to the Latin order. When seen in this light, this process of providing names for Gothic letters closely parallels that applied to Anglo-Saxon rune-names in the runic alphabets from the eighth century on (see Tables 29 and 30). There is no denying that the Gothic names are ‘runic’ in appearance. The question, however, is whether the phonological developments outlined by Mees are sufficient to support the view that these names can be traced to forms much older than the Anglo-Saxon or even Nordic forms. My critique would be as follows (see Table 39 for details of derivations suggested by Mees and Wagner and for references to their work). In the first place, there are quite a few forms for which exceptions have to be made to the ‘rules’ proposed by Mees (unless otherwise indicated, reconstructed forms are Proto-Germanic): a) If the name aza is to be derived from ahsa [axle, axis], then it does not belong to the list of names usually assumed for the runes. On the other hand, if it is to be derived from *ansuz > *azaz, as suggested by Wagner, then the loss of the final -z would be exceptional. b) As Wagner points out, in the derivation of the name fe < faihu, loss of -h- and unstressed u > a should have given *fea, but the scribe appears to have found the combination *-ea unacceptable, which again is exceptional. c) Derivation of gaar < jer requires the Gothic name to have come from a learned Anglo-Saxon spelling of the [j]-allophone of /g/ in a hypothetical form gǽr of the rune-name gēr. The vocalization of gaar is problematic if it was derived from jer. A possible solution to this problem is offered below.

358 Mees 2002–3: 60. 359 Mees 2002–3: 60–2. 360 Wagner 1994: 274–5. 361 Wagner 1994: 275. 362 According to Mees (2002–3 [publ. 2004]: 62), lowering of the i in uuinne < *winja was prevented by the following j.

118 Part 1: Sources of Name-forms and Meanings d) To compare the name sugil with the attested sauïl, Mees relies on

Lehman’s proposed derivation of Old English sigil < IE *sh2uél, *séh2uol to explain the ‘hardening of medial -w- to -g-’, whereas elsewhere the loss of postvocalic g is taken as the rule (cf. daaz, laaz, haal). e) The name chozma (< *kusma [boil?]) has to be explained as a translation of Nordic kaun. The Gothic equivalent of kaun would be *kauns, which, as the IEED database notes, would have given *chonz or *choz as the form of the Gothic name. f) To explain the form noicz, Wagner proposes that oi represents a long o < Germ. *au, spelled oi according to Old Frankish, cz is an orthographic variant of /ts/, which could mean that the name was derived from Gothic nauþs. g) Mees’s tentative suggestion of ezec < *aizik [coin, bronze bit]? or *aiziþa [mercy, honour]?363 involves words outside the list of names usually assumed for runes. h) Mees’s derivation of uuaer (= Gothic #aír [?caldron]) as ‘a deliberate acrophonic adaptation of an earlier *ei#s’ would involve rhotacization of a final s or z, which he elsewhere excludes as a phonological development, as in tyz. i) Gothic thyth was not derived from any word in the usual list of rune- names, but was probably a reflex of the attested Gothic noun þiuþ [(something) good], on which see below. j) The nominal endings of manna, geuua and reda are justified by the (attested) Gothic forms manna, giba and (reconstructed) *raida; but whereas loss of final *z is assumed for manna < *mannaz, and also for haal < *haglaz, no comparable loss is assumed for uraz, daaz or laaz. In assembling this critique, I find I am in complete agreement with Blomfield and Scardigli’s comments cited above, namely that the list of names in Vienna 795 is the work of a scribe with antiquarian interests, copying from an eclectic collection of rune-names but applying the spelling habits of his own dialect, in particular the use of for . On this basis, geuua, gaar, manna, utal, sugil appear to be derived from Anglo-Saxon forms, and perhaps also iiz and reda, although these last two could be either Anglo-Saxon or Nordic. Apparently Nordic in form are aza, bercna, fe and tyz, this last name possibly being taken from a reading of a final runic M or 7 as z. The final a in both aza and bercna (< *bairkan, with syncopation of -a- before -n) could have been influenced by a perceived parallel between Greek alpha, beta, gamma and aza, bercna, geuua. The names pertra and qertra almost certainly depend on their pairing in the ‘Latin’ ordering of the Gothic names, which would point to a rationalizing influence of the compiler of the list. Their forms could be based on Anglo-Saxon peorð and cweorþ, and the ending -ra might possibly derive from the ending of the Gothic noun *qairþra [lure, bait, decoy].364 The name haal was most likely based on

363 Following Von Grienberger (1896: 112) and Marstrander (1928: 157), respectively. 364 Seebold (2010: 93–101) hypothesizes that the names of the p- and q-runes and their Gothic counterparts are ascribable to Celtic influence at a time when Germanic and Celtic peoples were in contact on the continent of Europe, i.e. before Anglo- Saxons came into contact with insular Celtic languages. He emphasizes that his suggestions are hypothetical but in my opinion they exemplify the difficulties of

119 A Family of Names

Anglo-Saxon hægl, while daaz could have come from Gothic dags, based on Anglo-Saxon dæg. The name laaz may have been formed by analogy with daaz. Of the various proposals for the origin of eyz (see Table 39, note 14), the most attractive is Bugge’s suggestion that it derived from Go. *ehws.365 Arntz suggested that eyz could have been a scribal error for Gothic *aí#s [horse].366 However, I would suggest that the scribe translated Old English eh with a late-Gothic form eyz < *ews < *ehws (with loss of post-vocalic -h) < *aí#s. As for the name gaar, it is difficult to reconcile this form with either Gothic jer or Old English gēr. In the Anglo-Saxon fuþorc on the recto of the folio on which the Gothic alphabets appear on the verso, the rune-name is given as gaer. Derolez suggested that the ae was a mistake for ea.367 Whether he is right or not, the Gothic name gaar for /j/ does not appear to comply with the spelling rule set out in the text that accompanies the Gothic alphabets. This text gives an Old High German transcription of a passage from Wulfila’s Gothic Bible (which confirms that the scribe was familiar with Gothic) and also gives the rule:

ubi dicitur genuit j ponitur ubi gabriel g ponunt & alia his similia [Where genuit is said, j is put, where gabriel, they put , and similarly elsewhere.] Accordingly, the name gaar would presumably have been the name of a letter with a sound-value equivalent to the hard /g/ of Latin. Possibly, the scribe associated the Gothic letter j with Latin G, due to the similar shapes of the two letters. As a result, he placed the j in the position of G in the Latin sequence of letters and , perhaps taking his cue from the name gaer in the Anglo-Saxon fuþorc, treated j as representing a hard /g/, with appropriately a long [a:], i. e. aa, rather than a long [e:], in its name. Of the remaining eight names, chozma and noicz have been dealt with above: chozma as a translation of Nordic kaun, and noicz as a late spelling of Gothic nauþs, which could have been a translation of Anglo-Saxon nyd or Nordic naudz/nauþR. The final six names are literally the final names in the list: uuinne, uraz, enguz, ezec, uuaer and thyth. The last three would all appear to be Gothic forms that have nothing to do with names in either the Anglo-Saxon or Nordic lists: ezec < *aizik [coin, bronze bit]? or ezet < 368 *aiziþa [mercy, honour]; uuaer < *#aír [caldron]; and thyth < þiuþ [(something) good]. In this light, it is conceivable that the previous three are also Gothic forms but, in this case, reflecting the Anglo-Saxon names: uuinne < winja [pasture], reflecting OE wen, wynn; uraz < *ūrus, reflecting OE ūr; and enguz < *iggws [man], reflecting OE ing. This interpretation would not necessitate the involvement of early Germanic names to explain why these last three retained the grammatical endings that had undergone syncopation in Old English. demonstrating the antiquity of Gothic letter-names. See further my discussion of the rune-name peorð in §2.4.8. 365 Bugge 1905–13: 61–2. 366 Arntz 1944b: 221. 367 Derolez 1954: 61. 368 If the name is ezet, reading -t for -c, then it could possibly be related to Greek zeta.

120 Part 1: Sources of Name-forms and Meanings

The final position of thyth in the ‘Latin’ sequence of the Gothic names may be particularly noteworthy, since it would appear to link up with a patristic interpretation of the Hebrew letter-name, tet. As far as I know, no one has yet pointed out that the sense of the Gothic noun from which the Gothic letter-name is assumed to derive, þiuþ [(something) good], is precisely the sense that Eusebius and Jerome attributed to the Hebrew letter- name tet(h), i.e. bonum (Tables 6 and 8), and was repeated in numerous later English lists of Hebrew names (Table 13-1), as well as in the Carolingian poem by an unknown Irishman (Table 14 (a)) and the Auraicept (Table 14 (b)). The Hebrew word that Eusebius and Jerome were probably translating was bwf, tob (Table 9 (b)).369 Was the compiler of the list of Gothic letter- names aware of Eusebius’s and Jerome’s interpretation, or at least aware of the tradition that was based on Jerome’s interpretations? If he was, then it is conceivable that he was aware of his linking the list to a wider tradition and not just recording an odd list of names. Moreover, he may even have been going further than that and have been linking his list to a Judaeo-Christian tradition. It may be a coincidence that thyth was placed at the very end of the list, but there are a number of letter-names that could have been placed in this position. Was it intentional that a letter-name meaning ‘good’ was placed at the end of this compilation of ‘elements’, i.e. letters? Might it have been intended to echo the words of Genesis when, on the sixth day, God completed His creation of the elements: ‘And God saw all the things that he had made, and, they were very good’ (i.e. tob)?370 My conclusion, in response to the arguments like those put forward by Wagner and Mees, is similar to the conclusions of Blomfield and Scardigli given above: the list of Gothic letter-names in Vienna 795 can be explained as an adaptation of rune-names arranged in the Latin alphabetical order, comparable to the runic alphabets compiled from the eighth century on, but in this instance based on an eclectic mixture of Anglo-Saxon and Nordic names, with the addition of some names based on late forms of Gothic nouns.371 In other words, although the material in Vienna 795 is related to Anglo-Saxon and Nordic forms of the rune-names, it does not provide sufficient evidence to support the idea, echoed by Page, that these names ‘go back to common Germanic times.’ (b) Ideographs and abbreviations A second body of evidence that has been put forward to support the assumption that rune-names go back to a common Germanic origin comprises the Begriffsrunen, i.e. ideographs,372 that are supposed to occur in early epigraphic inscriptions. A prerequisite for the use of runes as ideographs is the existence of an established set of rune-names that will have been recognizable to anyone expected to understand the runes. As I noted in the introduction to §1.5.2, evidence for ideographs, and hence for an established set of rune-names, is indisputable in manuscripts containing Old

369 Seebold (2010: 92) suggests that the -y- represents -i- and the final -th should be read as a simple -t, giving þit as a representation of Greek theta, which was pronounced [qita]. 370 D-R, Gen. I.31 (Vulgate: viditque Deus cuncta quae fecit et erant valde bona). 371 Nedoma (2003: 557) uses the term krypto-gothisch, but at the same time suggests the forms can be related to early Germanic names. 372 For convenience, I use here the now traditional English translation of Begriffsrune as ‘ideograph’; but see my comments in the introduction to §1.5.2.

121 A Family of Names

English and Icelandic texts. Page cites various texts in which o, D and m ‘occur as convenient abbreviations or representations of eþel, man and dæg.’373 The question is whether such evidence also exists in epigraphic inscriptions predating the manuscript occurrences and thus lending support to the idea that rune-names go back to a time earlier than I would predicate for my thesis, i.e. the late seventh century. In discussing possible ideographs in inscriptions in England, Page suggests that the legend WBERHT on a coin, ‘recording the name Wynberht for a moneyer of Alfred and Edward the Elder’, might be an English example of the epigraphic use of W, ‘but here the inspiration was probably from manuscript texts since epigraphical runes hardly exist in Wessex’.374 Another apparent rune-form, %, found on a sceat of Wigræd, in fact turns out to be part of the ornamentation.375 On the other hand, Page concedes that the single runes d and a scratched, respectively, on a brooch and inside the base of a bronze bowl could be owners’ marks, representing names containing the elements Dæg... (or ...dæg) and Æsc.376 Also, a |t| pattern on either side of a sword from Faversham in Kent could be an ideograph referring to the war- god Tiw from whom the rune t derives its name in the Norwegian rune- poem, but it could equally well be purely decorative.377 In relation to the possible use of ideographs in Scandinavia, Düwel is cautious:

Die zuerst von Lindqvist gefundene und von Krause ausgebaute Methode, vor allem einzeln stehenden Runen mit ihrem Begriffswert aufzulösen, kan nur mit großer Behutsamkeit angewandt werden [...], zumal in manchen Fällen eindeutig eine Abkürzung vorliegt [...].378 This is in stark contrast to Elliott’s standpoint:

In this connection [i.e. the ritual import of rune-names] might be noted that magic words which occasionally occur in runic inscriptions, like alu, auja, laukaz, generally appear written in full or intelligibly shortened, whereas the words denoted by rune- names are represented by the rune alone. Conversely, where single runes occur, they should always be interpreted as standing for their name.379 According to Krause380 and Elliott, it would seem that: 1. Assumption of the early existence of rune-names (before their record in manuscripts) rests on the assumption that some single runes in early inscriptions were used to stand for their names, i.e. they had a Begriffswert. 2. This second assumption is only valid, however, if the single runes cannot otherwise be understood.

373 Page 1999: 77–8. The examples he gives are o from ll. 520–1 of Beowulf (British Library, Cotton Vitellius A. xx, fol. 141v), m on l. 23 of The Ruin in the Exeter Book, fol. 124r, d and m in the glosses to the Durham Ritual and the Lindesfarne Gospels. 374 Page 1999: 78–9. Wynberht was active at the end of the ninth century. 375 Page 1999: 90. Probably c. 740–50. 376 Page 1999: 91. The brooch is datable to the sixth century, the bowl to the fifth or sixth century; 377 Page 1999: 92. The sword is dated to the late sixth or early seventh century. For more examples of ideographs in English manuscripts, see Silva 2006: 395. 378 Düwel 2008: 8. 379 Elliott 1989: 60, fn.1. 380 For example, Krause 1970: 33–4.

122 Part 1: Sources of Name-forms and Meanings

In other words, any assumption of an early existence of rune-names rests on our inability to understand a few sporadic single runes, which in fact need not be Begriffsrunen at all but could simply be abbreviations, or perhaps magically endowed notae, or even runes cut at the idiosyncratic whim of a carver. I discussed this subject briefly in the appendix to my article ‘Rune- names: the Irish connexion’.381 My conclusion there was that there was no substantial evidence of ideographs in early inscriptions, as put forward by Krause. Düwel is of the same opinion, with the notable exception that the ú rune on the Stentoften stone (KJ96, DR357, dated c. 650)382 is ‘[d]as einzige sichere Beispiel für die Auflösung einer Rune mit ihrem Begriffswert’. 383 Because the whole issue of Begriffsrunen is still contentious, however, and because my previous treatment of the material was very brief, a fuller assessment of the instances of ideographs in early inscriptions suggested by Krause is given below. In their 1966 catalogue, Krause and Jankuhn listed twenty-one inscriptions containing single runes that he could not interpret without expansion. Of these twenty, three are almost certainly abbreviations:

• Femø bracteate (KJ132, Ant61, DR Br.45; Sønderby; dated 400– 650): ekfakaRf (right-left) Krause reads the final f-rune as an abbreviation for fahi [I write] and compares it with the inscription on the Åsum bracteate (KJ131, DR Br. 64): ẹheikfakaRfahi< , which Düwel reads as ẹhe< ek FākaR fāhi [Ich Fak schreibe], FākaR being a poetic name for ‘horse’.384 • Sievern bracteate (KJ134, Ant70; Land Bremen; dated 400–650): rwrita (right-left) Krause reads the r-rune as an abbreviation for rūnōR. Düwel reads rwrila, with l for t.385 • Wapno bracteate (KJ138, Ant68; Poznan, Poland; dated to 400– 650): sabar (right-left) Antonsen sees the final r as an abbreviation for runoR [Saba wrote the runes]. Of the remaining eighteen, Krause’s interpretations of single runes are mostly dubious, if not fanciful. There is a group of seven inscriptions which Krause suggests may contain ideographs with references to a god:

• the Allesø bracteate (KJ113, DR Br. 40; Fyn, Denmark; dated 450– 550): lauRoþaRlut:eaþl or Rlut:eaþllauRoþa Krause reads lauR oþa Rlut:eaþl and sees t:e as possible ideographs for *t(īwaR) – e(hwaR), i.e. ‘Gott-Pferd’; • the Lekkende bracteate (KJ126, DR Br. 46; Sjælland, Denmark, dated 400–650): ea (right-left) Krause suggests a possible interpretation of *e(hwaz) – a(nsuz) for the two runes ea and compares the occurrence of these two runes 381 Griffiths 2006: 105–7. 382 The reference numbers to inscriptions throughout this section are from the standard catalogues: KJ = Krause and Jankuhn 1966; Ant = Antonsen 1975; DR = Jakobsen and Moltke 1941–2; Br. = Bracteate number. 383 Düwel 2008: 8. 384 Düwel 2008: 50. On ẹhe< and the role of the horse, see ibid.: 54–5 and Krause and Jankuhn 1966: 244. 385 Düwel 2008: 47.

123 A Family of Names

with two other inscriptions, one on an amulet from Utgård (KJ51) and the other on a fibula from Dischingen (KJ155); • the Austad bracteate (KJ127, note; Romania; date ? ): gt Krause suggests ‘Gabe (an?) Týr’; • the Sjælland bracteate 2 (KJ127, Ant71, DR Br. 61; Sjælland, Denmark; dated 400–650):    Krause sees the column of three t’s as either a magic tree-symbol, or a repetition of t as the name of the god Týr; and • the Kragehul spearshaft (KJ27, Ant15, DR196; Fyn, Denmark; dated 350–550): ekerila< zasugisalasm< uh< ah< aiteg< ag< ag< a< ginuga...<

This last inscription is the most complicated of the series, where according to Krause the ga< ligatures could be abbreviations for the expression gibu auja [ich gebe Glück] which occurs on the Sjælland bracteate 2 (KJ127), or might be ideographs for ‘Gabe-Ase’, to indicate that the spearshaft had been dedicated to the gods (see Figure 1.G). The Sjælland bracteate 2 (KJ127) and the Kragehul spearshaft (KJ27) have received particular attention. For the spearshaft, Antonsen prefers Krause’s first suggestion and says that the ga< ligature ‘must be read as ag = Kragehul spearshaft (KJ27) auja gebū “I give protection”, as can be seen from alliteration (Uha Figure 1.G haite/auja gebū) and the sequence ginu ag, which must be ginnu - auja gebū; cf. [Ant]71 Sjælland bracteate 2 [KJ127].’386 Unfortunately, Antonsen’s ‘alliteration’ ignores the fact that the u of uha forms a ligature with the preceding m (see arrow in Figure 1.G). Also, it is impossible to ignore a comparable inscription on the Undley bracteate, which has been dated to the second half of the fifth century (see Figure 1.H). This find, which was made on Undley Common, near Lakenheath, Suffolk in 1981, has been the subject of considerable controversy as a result of publications by Hines and Odenstedt, according to whom the bracteate was manufactured in Schleswig-Holstein, or at least in southern Scandinavia, in the middle of the second half of the fifth century.387 This ascription is based on the typology of the bracteate, with its representation of a she-wolf suckling twins below a head in profile, which was copied from VRBS Romana coins and medalions of Constantine the Great.388 The controversy arises because the inscription on the bracteate The Undley Bracteate contains three ligatures reading right to left: British Museum, no. 1984,11-1,1, © Trustees of the British Museum

Figure 1.H

386 The spearshaft is in Copenhagen Nationalmuseet, Inv. C 3158. Pieper (1999: 316–8) reads the three ga< ligatures as g[ebu] a[nsu] [gift to the god (Odin)]. 387 Hines 1984: 204–9; Hines and Odenstedt 1987: 73–94; see also Parsons 1999: 62–7. 388 Hines 1984: 204–9.

124 Part 1: Sources of Name-forms and Meanings which echo the three ligatures reading left to right on the Kragehul spearshaft:

except that the central ligature on the Undley bracteate consists of a g plus an ‘Anglo-Frisian’ O-rune. If Odenstedt and Hines’s dating and provenance are correct, then not only is this the earliest known instance of an ‘Anglo- Frisian’ O-rune, but its occurrence on an artefact of Scandinavian origin is quite remarkable. In the present context, however, it is less important to follow up Odenstedt and Hines’s conclusions, or the specific interpretations that have been offered for the inscription,389 than to consider the ligatures in connexion with their possible interpretation as ideographs. If the a, O and g in the Undley ligatures stood for their respective rune-names, then in Old English the a will have represented æsc, the O os and the g gyfu. If the name os referred to ‘god’ — and that is by no means certain (see §2.4.4, s.v. Ōs, óss) — the middle ligature, presumably go< , could possibly have referred to a ‘Gabe-Ase’[gift-god], as Krause posited for the three Kragehul ga< ligatures. However, an interpretation of the other two Undley gæ< ligatures as ‘gift’ and ‘ash’ would be difficult to place in the context of a bracteate, which in turn calls into question the whole idea of the ligatures being ideographs based on rune-names. Moreover, the Undley and Kragehul ligatures are so similar that it must also be questionable whether the Kragehul ga< ga< ga< ligatures can be ideographs. In both instances, the ligatures appear to be part of a formula, 390 in which case the middle Kragehul ga< is likely to be equivalent to the middle Undley go< , with both of them being distinct from the two ligatures on either side. On the Sjælland bracteate 2 (KJ127), the images are those of a man’s head in profile above a horse, which have been associated with Odin as the divine healer of Balder’s foal, as described in the second Merseburg Charm.391 The last sign in the inscription appears to be composed of three superimposed t-runes, i.e. Týr-runes, which Hauck and Heizmann interpret as a monogramme representing Odin’s name Fimbultýr, ‘the mighty god’ referred to in V@luspá, verse 60.392 Whether such an interpretation is accepted or not, the inscription as such would appear to underline the amulet nature of the bracteate: hariuhahaitika … farauisa : gibuauja „ (followed by the three superimposed t-runes) [Hariuha is my name, the dangerous knowing one (or: expert traveller); (I) give protection (or: happiness)]. In this light, the final t-runes could be considered an appropriate conclusion, whether they represent a triple invocation of ‘god’ or are simply a final flourish. (The fuþark on the Kylver stone slab, dated c. 350–475 (KJ1, Ant30), described in detail in section (c) below, concludes with a similar

389 There have been various attempts at interpreting the Undley inscription; see, for example, Odenstedt 1983: 5; Hines 1984: 206–7; Hines and Odenstedt 1987: 91; Eichner 1990: 315–18; Bammesberger 1991: 398–400; Grønvik 1996: 44–5 390 See Moltke 1985: 104–5; Derolez 1990: 417; Page 1996: 133. 391 Hauck 1998a: 320–1; 1998b: 43; 1998c: 502; Düwel 2008: 45–6. 392 Heizmann 1998: 530–5; Hauck 1998c: 489–544; see also Nowak 2003: 166–7 and Düwel 2008: 49–50.

125 A Family of Names

sign resembling several superimposed t-runes, although this version has an unequal number of side-branches to left and right.) Another inscription which, like the Kragehul and Undley inscriptions, is tantalizingly cryptographic, consisting of a series of single runes with possible magic intent, is that found on the Lindholm amulet (KJ29, DR261): A. ekerilaRsawilagaRhateka… B. aaaaaaaannnRRR-bmuttt:alu… The first part of the inscription (A) is intelligible and has been translated as either ‘I “the eril” here am called the crafty one’393 or ‘I “the eril” am called the one who knows magic’.394 The second part (B), however, has been demonstrated by Flowers to be a magical formula, although I would suggest that it is reminiscent of an example of cryptography offered by Virgilius Mario Grammaticus, i.e. ‘Cicero’s’ RRR SS PP MM NT EE OO A V I for: spes romanorum perit.395 The remaining examples cited by Krause (KJ) as Begriffsrunen are:

Two inscriptions with l-runes: KJ38 Gjersvik (scraper, date?). KJ133 Nebenstede 1 (bracteate, c. 500) Two inscriptions with h-runes: KJ21, Ant3 Torsbjerg (shield boss, c. 200); KJ52, Ant93 Kinneve (Bautastein, c. 600). Two inscriptions with o runes: KJ20 Torsbjerg (sword-chape, c.200); KJ41, Ant97 Pietroassa (gold ring, c.380). Two inscriptions with f-runes: KJ37, Ant19 Fløksand (scraper, date?); KJ95, Ant116 Gummarp (Bautastein, c.650). Two inscriptions with j-runes: KJ96, Ant119 Stentoften (Bautastein, c.650); KJ105, Ant103 Skodborg (bracteate, c. 500).

The l-rune on the Gjersvik scraper (KJ38) is cited by Krause396 in connexion with Begriffsrunen because he interpreted it as standing for laukaR in the sense of ‘Gedeihen’ and because, as such, it could be used to support his supposition that the name of the l-rune was at one time *laukaz. There is, however, no definitive evidence to support the idea that the name of the l-rune was *laukaz. On the other hand, Düwel may be right in claiming that the frequent occurrence of laukaR on bracteates may indicate that the word constituted a ‘Formelwort’.397 It may be in this function, as an abbreviation of a ‘Formelwort’ laukaR, that the single rune l appears on the Nebenstede 1 bracteate. Düwel transliterates the inscription as: glïaugiRuïurnRl GlīaugiR wīu r[ū]n[ō]R l[aukaR] with the translation: ‘Ich, der Glanzäugige, weihe die R[u]n[e]n, L[auch]’ [I, the gleaming-eyed one, dedicate the runes, leek], the reference being to Odin, who is dedicating the runes. For the retrograde inscription on the Torsbjerg shield boss: aisgzh Krause (KJ21) suggests that the final h stands for h(agala) [hail] = ‘jähes Verderben’ [sudden ruin] and compares the reference to hail with the

393 Krause and Jankuhn 1966: 70. 394 DR 1942: 317. 395 Tardi 1928, Epitomae XIII. De scinderatione fonorum. 396 Krause 1970: 34. 397 Düwel 2008: 201.

126 Part 1: Sources of Name-forms and Meanings inclusion of hagala at the very end of the inscription on the Kragehul spearshaft. Antonsen (Ant3) agrees that the h may refer to hail and combines the word with the preceeding runes, to give the interpretation aisk-z h(agalas) ‘Challenger of the hail [i.e. of spears and arrows]’. As an alternative, Düwel offers:

ais(i)g(a)R »der Wütende, der Rasende«, wozu entweder abgekürztes h(aitē) »heiße ich« oder wiederum die Begriffsrune »Hagel« tritt, wieder mit dem Ergebnis eines Waffennamens.398 He concludes, however, ‘Alle diese Versuche operieren mit Voraussetzungen, die sich nicht zweifelfrei nachweisen lassen.’ A similar conclusion of ‘not proven’, moreover, applies to Krause’s interpretation of the inscription on the Kinneve stone (KJ52): ...sizaluh (...) where he again links the h to hail in the sense of ‘jähes Verderben’:

Die schließende h-Rune ist wohl am ehesten als Begriffsrune für h(agala) “Hagel” – “jähes Verderben” aufzufassen. Es handelt sich also um Grabmagie. Das Steinchen lag ursprünglich wohl im Innern eines Grabes. The interpretation of the character supposed to be an o-rune on the Pietroassa ring (KJ41) as an ideograph is also uncertain. The inscription reads: gUtani ‹> Wi hailag where the character between the first I and the W is illegible because the ring has been cut through at this point so that the character has been divided in two. The two halves, ‹ >, have been variously interpreted as j (j), ø (s) or 5 (ŋ).399 Grimm and Stephens,400 who saw the inscription before it was cut in two, reproduced it as: Figure 1.I Figure 1.J

Grimm’s drawing Stephens’s drawing A common interpretation takes the severed character to be an o-rune representing an ideograph for ōþal [possession] so that the whole inscription reads, in Gothic: gutani o wi hailag Gutanī ō(þal) wī(h) hailag ‘possession of the Goths, dedicated [and] sacrosanct’. Nedoma401 has recently rediscovered a photograph, taken in 1869, which he sees as confirming the disputed rune as an o and hence justifying its interpretation as an ideograph.402 However, the apex of the o is missing and

398 Düwel 2008: 17. 399 Reichert (1991–3: 237–8) j; Looijenga (2003: 106) j or s; Pieper (2003: 632–3) ŋ. 400 Grimm 1856: 602; Stephens 1867–8: 567. 401 Nedoma 2010: 27, 29–30. See also Dúwel 2008: 31.

127 A Family of Names

the two halves, ‹ >, are not symmetrical. On this basis, I still find the interpretation of the o-rune as an ideograph to be unproven. As for the Torsbjerg sword-chape (KJ20), the part of the inscription with the o-rune is on side A of the object and reads: owlþuþewaR, of which Düwel says:

Die re. Inschrift ... is verschieden gedeutet worden: [Seite] A als »Erbbesitz (Begriffsrune). – W(u)lthuthew«, »W.« (mit einem die Inschrift eröffenden Zeichen) oder nur »W.« (mit Vertauschung von o und w).403 There is thus no agreement on the status of the o-rune as ideograph in either the Pietroassa or Torsbjerg inscription. The f-rune on the Fløksand scraper (KJ37) is inverted with respect to the original inscription (dated c. 350) and has been added later, there being absolutely no means of interpreting its meaning. On the other hand, claims for the ideographic status of the f-runes in the inscription on the Gummarp stone (KJ95) are more substantial. The stone was removed from Blekinge and taken to Copenhagen, where it was destroyed in that city’s great fire of 1728. The inscription, however, is known from a drawing made of it by Jon Skonvig, the draughtsman employed by Ole Worm (see Figure 1.K404). It belongs to the transitional period for Scandinavian runes, when the g, W and ú runes of the older 24-rune fuþark were still in use alongside, for instance, the J of the younger 16-rune fuþąrk. The J was a new rune used to represent an oral [a] while the older a came to be reserved for a nasal [ã]. The Gummarp inscription reads on sides A to D: A. (h)AþuwolAfA B. sAte C. stAbaþria D. fff The Gummarp stone (KJ95) from a drawing by Jon Skonvig in which A transcribes the J-rune. In translation, it reads ‘Haduwolf placed three staves fff”. The stone is linked via its original location in Blekinge, as Figure 1.K well as the name Haduwolf cited on it, to three other stones, in particular to the Stentoften stone (KJ96), shown in Figure 1.L.405 The relevant part of the inscription on Stentoften reads: niuhAborumR | niuhagestumR | hAþuwolAfRgAfj

402 A reading of gutanio as Gutanjō [of the godesses] (Mees 2002–3 [publ. 2004]: 72–3; Macloed and Mees 2006: 58–9) is rejected by Nedoma (2010: 53, note 76) on morphological grounds. 403 Düwel 2008: 26. Moltke (1985: 99) prefers a simple name, but with o for w and w for o: ‘The first line [A] contains a man’s name, “WulþuþewaR” (o stands for w – compare perhaps gaois on the lance-head from Mos, Gotland? – and w stands for u), which has a meaning of its own, namely “servant of (the god called) Ull”. The sense of the second line [B: niwajemariR< ] is disputed: either “the not ill-famed”(mariR = famed), an adjective qualifying WulþuþewaR, or “do not spare (waje), MariR” (the last word then being the sword’s name).’ The “do not spare, MariR” interpretation is Marstrander’s (1952: 13). Düwel points out that the god Ull is only known in Scandinavia and points to another possibility for the first element of the name: got. wulþus [Glanz]. 404 Accessed at http://runer.ku.dk//userfiles/image/IggCache/354/rune02b_501.jpg, June 2010. 405 Accessed at http://www.arild-hauge.com/arild-hauge/se-rune-stentoften.jpg, June 2010.

128 Part 1: Sources of Name-forms and Meanings

which has been translated as ‘Haduwolf gave to the new farmers, the newcomers, j’. Both the three f’s on the Gummarp stone and the j on Stentoften have been interpreted as ideographs in the context that Haduwolf was a member of a Wulfinge clan that was intent on demonstrating their generosity to their underlings — the fff meaning ‘wealth’ (fé = cattle as a sign of wealth) and j meaning ‘(good) year, harvest’. But, if one wants to be pedantic, why should three f’s be ‘placed’ on Gummarp whereas a j is ‘given’ on Stentoften? It could be that the ‘placing’ of fff is of a different order to the ‘giving’ of j. The three f’s were apparently perceived as ‘staves’, but signifying what? The Stentoften j, on the other hand, might have related to a concept such as a good harvest, but it could equally well have referred to a person or an object. The only reason for interpreting the j-rune ú as meaning ‘(a good) year (or harvest)’ was that the rune ú was assumed to be the forerunner of the rune J and that the name given to the later form of the J-rune in manuscript texts, i.e. N, was ár [year].406 The sound-value of ú was undeniably [j] and that of J was undeniably [a]. The place of ú in the older fuþark was after the i-rune, as was the place of J in the younger Stentoften stone Scandinavian fuþąrk. The name ár [year] is derived from *jara, with loss of (DR 357) the initial j- and syncope of the final -a. A name *jara would be an with h as oral [a] and an elder-fuþark j- appropriate name for the j-rune according to the acrophonic principle. These rune, ú (arrowed) observations have accordingly led to the assumption that the older ú-rune was replaced by the J-rune in the younger Scandinavian fuþąrk and that the Figure 1.L sound-value of the rune changed from [j] to [a] as a result of a change in the name from *jara to ár. An example of this reasoning, based on an assumption of the early existence of rune-names, is provided by Spurkland:407

The great language changes that took place at that time [in the seventh and eighth centuries] meant that the words which functioned as rune-names also changed. In the word *jāra, initial /j/ disappeared, and as a result of syncope the final /a/ was also dropped, resulting in ON ár. In this way the j-rune changed its name from * jāra to ár, and at the same time also changed its sound value from /j/ to /a/. But that was not all – it also changed its appearance from j to J. It is on this type of speculative reconstruction that the Stentoften j-rune has been interpreted as an ideograph meaning ‘a good year/harvest’. However, it is remarkable that the form of the supposed ideographic rune on the Stentoften stone is that of the older j-rune ú, i.e. the older fuþark form, rather than the newer form J, which is used throughout the rest of the

406 For example, the Abecedarium nordmannicum (St Gall, Stiftbibliothek, 878) and Leiden, Universiteitsbibliotheek, Vossianus lat. Q 83 (see Table 37). 407 Spurkland 2005: 81. The Stentoften inscription has been interpreted as referring to a ritual sacrifice (Santesson 1989: 221–9; 1993: 241–52), with which ‘Haduwolf gave a (good) year’, which might explain the special use of the j-rune as an ideograph in the sense of ‘good year’. But even with this interpretation it is difficult to draw the general conclusion that ú was regularly used as an ideograph rather than incidentally enlisted as an abbreviation. According to Adam of Bremen’s (unique) description of a similar sacrifice, c. 1075, the purpose of the ritual was to placate the gods, without reference to a ‘good year’: Ex omni animante, quod masculinum est, novem capita offeruntur, quorum sanguine deos placari mos est. [From every living thing that is male, they offer nine heads, with the blood of which it is customary to placate the gods.] (Gesta Hammaburgensis ecclesiae pontificum, Descriptio insularum aquilonis, ch. 27, accessed at http://hbar.phys.msu.ru/gorm/ chrons/bremen. htm#descrip, January 2013.)

129 A Family of Names inscription. Does the choice of the older rune-form for the supposed ideograph indicate that the older ú and the newer J were considered to be distinct not only in form and sound-value, but also in significance? If the ú- rune did signify something that it was not possible to express with the J- rune, then it may well be anachronistic to project back on to the ú-rune a name that is not recorded until it occurs in the ninth century (in Leiden, Universiteitsbibliotkeek, Vossianus lat. Q 83) as the name of a later form of J, i.e. N. In fact, even in this manuscript, the name given as ar in Latin script was reproduced in runes as aR, with an initial a rather than the N-rune to which the name was assigned. (These developments are dealt with in more detail in the next section.) The final inscription in the above list cited by Krause is the Skodburg bracteate (KJ105). It reads: aujaalawinaujaalawinaujaalawinjalawid The first thirty runes repeat the phrase auja alawin three times, with auja meaning ‘protection, happiness’, and alawin as a proper name. Düwel quotes Müller’s interpretation of the name as being that of a god:408

Die Anrede des Gottes als allumfassenden, allmächtigen Freund [ala-winiR] […] erscheint gerade für ein Brakteatenamulett als sinvoll. Daß dabei der […] Götterfürst, der »Brakteatenhauptgott« gemeint ist, dessen Gestalt auch de Skodburg-Brakteat wiedergibt, liegt nahe. [Addressing the god as an almighty friend … is precisely meaningful for a bracteate amulet. That it is the … lord of the gods, the ‘principal god of bracteates’, whose figure is also reflected by the Skodburg bracteate, would seem very likely.] The last seven runes, jalawid, begin with a j-rune which is reversed with respect to all the other runes in the series. Düwel sees Alawid as a name comparable to Alawin – he dismisses a form *Jalawid as being less plausible – and suggests that the j-rune is an ideograph representing the wish for a ‘(good) year’. Krause himself is less certain:

Die j-rune vor dem letzten Namen kan entweder (nach Salberger) als Begriffsrune j(āra) (vgl. Stentoften) oder (mit Bæksted) als Verschreibung für nochmaliges auja aufgefaßt werden. [The j-rune can be interpreted either (according to Salberger) as an ideograph j(ara) (cf. Stentoften) or (following Bæksted) as a mistake for another auja.] However, the reversal of the j-rune may be significant and could possibly have been intended to separate the final name from the rest of the invocation. To sum up, Krause’s idea that certain single runes in early inscriptions could be interpreted as representing concepts expressed by the names of the runes, i.e. Begriffsrunen or ideographs, presupposed that the runes concerned had established names from an early date. His suggestions were in essence intended as possible solutions to interpreting inscriptions that were otherwise intractable. In relation to the Scandinavian bracteates, the idea of ideographs was helpful in highlighting two broader concepts, namely that: 1. the restricted space offered by bracteates required those composing inscriptions on them to make use of single runes as abbreviations similar to those on Roman and other coins; and 2. the amulet function of most bracteates encouraged the use of runes, and in particular single runes, to represent cryptic formulae with invocatory or possibly magic intent.

408 Düwel 2008: 51; Müller 1988: 134.

130 Part 1: Sources of Name-forms and Meanings

Most of the single runes used on bracteates or amulet-like objects that Krause suggested might be ideographs can be interpreted in terms of these two broader concepts but not as ideographs in Krause’s narrower sense. They do not appear to have represented rune-names from an established set of names, but to have been pressed into service according to the requirement of each individual inscription. I therefore conclude that in pre-eighth-century inscriptions on artefacts other than amulet-like objects, the evidence for ideographs is inconclusive. Most of the single runes in these cases are uninterpretable or can be explained as abbreviations or perhaps mistakes. It is on this basis difficult to use the concept of Begriffsrunen to adduce the existence of rune-names before they first occur in manuscripts. (c) The possible role of rune-names in the fuþark and fuþąrk in Scandinavia The third source of evidence for the early existence of rune-names is the supposed role that some names played in the changes in sound-values represented by certain runes. The passage from Spurkland cited above provides an example of the role that the supposed name of the ú-rune, i.e. *jara, is thought to have played in the change in sound-value of the rune that apparently replaced the ú-rune. The name *jara lost its initial j and final a to become ár, as a result of which, it is assumed, the rune came to represent [a] instead of [j].409 A difficulty with this particular assumption is that, as Spurkland also indicates, the shape of the rune apparently changed at the same time as the sound-value changed, the ú being replaced by J. These changes, however, were more complex than this description implies and need to be considered in the context of two general changes that were taking place in conjunction during the period between about 600 and 800, both in Nordic dialects and in the system of runes used to represent them. On the one hand, there were linguistic changes, involving in particular the vowel system; on the other hand, the system of runes was subject to radical reform, due partly to the linguistic changes, and partly to other independent influences which are not entirely clear. The most obvious change to the runic sytem was a gradual reduction in the number of runes from twenty-four to sixteen. Up to about 600, the number and forms of the runes in the elder fuþark show remarkable consistency. The oldest known fuþark is the one found on the Kylver stone on Gotland, dated c. 350–475 (KJ1, Ant30), where the j-rune has the form ÷, followed by p and ï (the last two runes being d and o):

f u þ a r k g w h n i j p ï z s t b e m l ŋ d o

409 It is perhaps easy to see how interpretation of ‘the acrophonic principle’ in relation to runes becomes extended in its application. For example, Stiles (1995: 185) says quite correctly that: ‘It is a principle of the runic alphabet that the name of a rune gives its value (the acrophonic principle).’ But later, he extends this by saying (1995: 187): ‘When a rune-name underwent a sound-change, its value was liable to change too (this is part of the acrophonic principle).’ I would suggest, however, that this latter process, demonstrated by the development in the name *jara > ár, could better be described as the acrophonic principle working in reverse: instead of a character’s sound-value being reflected in the sound-value of the initial letter of the name chosen for it, a change in the sound-value of the initial letter of an established name is said to determine the sound-value of the character.

131 A Family of Names

Another fuþark in which the j-rune has the form ÷ is the Charnay fibula, dated c. 550–600 (KJ6), where it is followed by ï and p. In the retrograde fuþark on the Vadstena bracteate from Östergötland, dated c. 500–550 (KJ2, Ant90), the j-rune has the form 2 and is follows by ï and b, with b for p (the last two runes being o and probably d, although the final rune is unclear):

(d) o ŋ l m e b t: s z b ï j i n h : w g k r a þ u f In the period between 600 and 800, when the major changes were taking place, there are no known records of fuþark sequences. Reconstructions based on the runes used in inscriptions of the period demonstrate the gradual reduction in the number of runes. The inscriptions on the Blekinge group of stones (KJ95–8), dated c. 600–675, and on the Eggja stone (KJ101), dated c. 700–750, make use of 21 runes. The p, ï and ŋ-runes are unused, the a-rune represents a nasal [ã], transcribed as ą, while the J-rune represents an oral [a], transcribed as A:

F u q a R ß g w f u þ ą r k g w h n i J . . 7 S h n i A . . R s t b e m l . d o t b e m l . d o In the inscription on the Ribe cranium, dated between 700–765,410 there are only fifteen distinct runes, although the number in the fuþąrk in use is probably sixteen since the a-rune was not used in the inscription but can be assumed to have belonged to the sequence:

F U Q ? R ß . . f u þ (ą) r k . . h n I J . . 7 S h n I A . . R s t B . m l . . . t b . m l . . . From 800 onward, two main types of fuþąrk – or ‘younger fuþąrk’ – are found, both consisting of sixteen runes, as exemplified by the ‘long-twig runes’ in the Norwegian-Danish (or Swedish-Danish) fuþąrk on the Gørlev stone (ninth century, DR239), and the ‘short-twig runes’of the Norwegian- Swedish fuþąrk on the Hedeby stick 1 (c. 800–1000): Figure 1.M Figure 1.N ‘Long-twig’ runes ‘Short-twig’ runes

f u þ ą r k h n i A s t b m l R f u þ ą r k h n i a st bm l R As a result of the reduction to sixteen runes, certain runes came to be used for multiple sounds. The B-rune, for example, came to be used to reproduce [p] as well as [b] and [β]. The most significant changes, however, were those affecting the vowels where the reduction in the number of runes was contemporaneous with a process of mutation which actually increased the number of vowels. As a consequence, each of the four vocalic runes remaining from the ninth century on – a, J, i, u – came to represent a range of vowels, as summarized below.411

410 Düwel 2008: 89.

132 Part 1: Sources of Name-forms and Meanings

Sound-values Pre-600; c. 600–75; c. 700–50; Ninth century; Ninth century; Sound-values in the elder the elder Blekinge Eggja stone, Gørlev stone, Hedeby stick 1,415 in the younger fuþark fuþark412 group413 Ribe cranium414 ‘long-twig’ ‘short-twig’ fuþąrk [a] a a a a Ä (») [ã], [õ]416 h, c h æ æ (A) [a], [æ], [e], [@] [e] e e (e) - - - [i] i i i i i [i], [], [e], [j]417 [j] & (÷ h) ú - - - - [o] o o (o) - - - [u] u u u u u [u], [o], [y], [w] [w] W W (W) - - - In this evolution, the runes where it has been suggested that rune-names played a major role are the vowels represented by the J, c and a-runes. In inscriptions up to c. 600, the a-rune represents [a], whereas the c-rune appears in the Kylver and Charnay fuþarks in the position occupied in other fuþarks by the j-rune, & or ú. The J-rune is not recorded in any known fuþark; its first known occurrence is in the inscription on the Noleby stone (KJ67, end of the sixth century418), with the value [j] as the third rune in the word toJ ße a tojeka< , which is interpreted as ‘I prepare’.419 In the next hundred years or so, however, the J-rune appears on the Blekinge group of stones mentioned above, with the value of an oral [a], as opposed to the nasal [ã] represented in the same inscriptions by the older a-rune. At about the same time, in an inscription on one of the stones in this group – Istaby (KJ98) – oral [a] was represented by c rather than J in comparable rune sequences. For example, a comparison of the names on the Stentoften and Istaby stones shows:420

Stentoften: h J q U w o l J F 7 h J R i w o l J F 7 h A þ u w o l A f R h A r i w o l A f R Istaby: h S q U w U l a F 7 h S R i w U l a F a h A þ u w u l ã f R h A r i w u l ã f ã

411 Only broad bands of dates are indicated since there is much debate about precise dating (see, for example, Isakson 2000 and Düwel 2008). 412 Elder fuþark: The form ÷ appears in the Kylver fuþark (KJ1, 350–475) and the Charnay fuþark (KJ6, second half of the sixth century); the form h appears as a j- rune in an inscription on the Noleby stone (KJ67, end of the sixth century). 413 The c-rune form occurs only on the Istaby stone. In the Istaby inscription, the a- rune is also used parasitically; in the Gummarp and Stentoften inscriptions, the h- rune is also used parasitically. 414 The forms in round brackets occur only on the Eggja stone. 415 ‘Hedeby’ is also spelled ‘Haddeby’ or ‘Haideby’; on rune-stones haiþabu. The forms in brackets appear on Hedeby stick 2. 416 Also a nasalized [@]. 417 The i rune in the younger fuþąrk also represented incidentally [ai] and [ei]. 418 Düwel 2008: 35. 419 Antonsen (2002: 183) notes that ‘... the dating of the inscription to the end of the 6th century, based on the forms of the various runes, cannot be correct. The inscription displays no “monophthongizations”. The long vowel of tōj- is not the result of East Nordic monothphthongization of */au/. It stands in the same ablaut relationship to /aw/ in tawido as Go. /ō/ in (fulla-)tojis ‘complete’ to Go. /aw/ in tawida. … The Noleby inscription cannot be younger than ca. AD 450.’ 420 The o-rune on Stentoften almost invariably alternates with the U-rune on Istaby.

133 A Family of Names

It will also be seen that an epenthetic J on Stentoften, in the sequence l J F, appears as a on Istaby. The similarity in form of the Istaby c to the Kylver and Charnay ÷, as well as the use of J as [j] on the Noleby stone before its appearance at Blekinge with the value [a], would appear to confirm that both ÷ and J initially had the value [j] and that by the seventh century they both came to be used for oral [a]. These developments can be considered beside the fact that the name later recorded for the A-rune was ár [year], as illustrated by the ninth- century ABECEDARIUM NORD(mannicum), found on page 321 of St Gall, Stiftbibliotheek, 878, where æ is a later form of J (see Table 40 (a)).421 In the light of the Blekinge inscriptions and the Abecedarium, it is certainly not unreasonable to explain the apparent change in the sound-value of ÷ and J from [j] to [a] by assuming that these runes originally had the name * jara and that the change in sound-value was a consequence of the change in the name from *jara to *ár. The sound-value of the æ-rune in the Abecedarium, which developed from J, would appear to be confirmed by the interlinear gloss on ‘æ ar’, namely A, which represented [a] in the extended Anglo- Saxon fuþorc. There may be some doubt as to the date of the gloss, however, because the Nordic 7 in the Abecedarium is glossed with an Anglo-Saxon . This latter rune had the Old English name yr, with the sound-value [y:]. The Nordic 7 also had the name yr, but in the ninth century, it still represented a palatal r < Germ. /z/ (which is usually transcribed as R); it was not used for [y] until the eleventh century. Bauer suggests that the Anglo- Saxon gloss reveals the scribe’s inaccurate knowledge of rune-names or their sound-values,422 but it is also possible that the glosses (if not the Abecedarium itself) are to be dated to the eleventh century.423 When, however, the Abecedarium is compared with a tenth-century list of Nordic rune-names on folio 24v of Leiden, Vossianus lat. Q 83,424 a number of anomalous features are found besides that relating to the 7-rune. The most obvious is that the name of the æ-rune is spelled aR, i.e. not with an initial æ, but with the supposedly nasal [ã], a. Also, the name aR is reproduced in the Latin script as ae, which may be intended to indicate the sound-value of the rune, although no specific sound-value is given for any of the other runes in the Latin transcription of the text. The transcription of the runes and Latin interlinear text given in Table 40 (b) is my own, based on

421 This copy of the Abecedarium is from Grimm 1883: 112. The manuscript has been dated to the second half of the ninth century (see note 423 below). The Abecedarium has been considered a poem on account of the alliteration in the first line in particular and in the other lines more sparingly. The runes are Nordic with interlinear Anglo-Saxon additions. The rune-names (and the text) show a mixture of ON, OE, OS and OHG forms: man (OE/OS/OHG); hagal, ar, sol (ON); lagu, os (OE); ur, is (common Germ.); t- (?); yr (OE/ON?); rat (-a- OE, -t OHG); naut (-au- ON, -t OHG); thuris (OS?); brita (b(i)rica? OS/OHG); cha- (OHG); feu (fe(h)u OS?); see Bauer 2003a: 58–77. 422 Bauer 2003a: 65. 423 Bischoff (1981: 221) saw the Abecedarium as the work of Walahfrid Strabo around the middle of the ninth century (see also Derolez 1954: 75). Others (e.g. Scherrer, 1875: 307, and Steinmeyer and Sievers, 1898: 445) have dated the manuscript to the eleventh century. See also note 421 above. 424 The manuscript itself is dated to the ninth century (De Meyier 1975: 193), but it is thought that the runic list was added in the tenth, although the names are thought to be based on ninth-century forms (Isakson 2000: 3–43).

134 Part 1: Sources of Name-forms and Meanings one made by H. Kern,425 since the original is in no state to be reproduced in facsimile. Where the æ-rune is used in the spelling of the rune-names, it is transcribed in the Latin text as e – biercan and reidu – and although the nasal [ã], a, occurs before n in names like biercan and manr, as is to be expected, it is also used in non-nasal positions such as hagal, JaKal. In other words, there was an overlap in the use of æ and a, and the latter was not reserved for [ã] or other nasal vowel, as might be inferred from the assumption that the name of the a-rune developed from *ansuR to *ãsuR and then to áss, with an ensuing re-alignment of the rune’s sound-value from [ān] to [ã:] to [a:]. The anomalies in Leiden, Vossianus lat. Q 83 could have been due to scribal error or ignorance,426 but they suggest to me that the differentiation between oral [a] æ and nasal [ã] a was a later rationalization. Likewise, the differentiation between Noleby’s J for [j] and Blekinge’s J for oral [a] may not be so clear-cut as is suggested by the explanation that *jara > ár. In particular, Antonsen has pointed to the occurrence on the Blekinge stones of both J and c in digraphs such as Je and ce, in particular in the Björketorp stone’s hJeRJmJlJUø7 (hAerAmAlAusR) and the Istaby stone’s hSeRUwUlaFi7 (hAeruwulafiR). He finds this spelling “suspicious” and suggests:427

If one proceeds directly from the runic inscription itself instead of from the transliteration, one must seriously question the value of the c-rune before e in the word hceRUwUlaFi7. While it is true that this rune (from older j, ú, J, see Barnes 2001: 105) can be used to express the vowels /a, ā/, as a consequence of the phonological development of its name from */jæran/ > */jara/ > */ara/, it is not possible to conclude that it could no longer be used to designate /j/ where the latter still occurred (i.e. non-initially), as on the Noleby stone’s J = /j/ in toJ Ke a tojeka< The patronymic on the Istaby stone must be transcribed simply as HjøruwulafīR, and the form is the most natural that could be expected and shows that breaking was already completed in this period. The same is true for the Björketorp stone’s hJeRJmJlJUø7 hAerAmAlAusR, in which Ae expresses /jæ/ in hjøramaløsR ... This explanation further complicates an already complex situation. If Antonsen is right, then the runes J and c on the Blekinge stones had the value [a] both initially and medially except when combined non-initially with e, in which case they retained the value [j] to indicate breaking. But this resolution of Ae into jø is difficult to reconcile with the interpretation of iJ as ja or jø in a word like hiJlB, i.e. hjalb, representing either the verb hjalp or the noun hjølp [help], on the Ribe cranium, which is dated to c. 720. The jara-rune, however, was not the only rune where it has been suggested that changes in the initial sound-value of the name played a role in shaping the younger Nordic fuþąrk. Changes in the supposed rune-names have also been cited as contributing to the loss of runes like W, w, and e, e (see the table above). It has been argued, for instance, that the original rune W had the name *wunju and that loss of the initial w- led to the name *unju, with the result that there were two runes representing /u/, i. e. U and W. ‘But the overlap was avoided by dropping W, w, from the fuþark.’428 Similarly, the

425 On an inserted leaf in the codex, dated October 1872. The interlinear r is short- stem rather than long-stem r. 426 Note also the Anglo-Saxon C (k) in the name Caun (kaun), and the roman C in biæRCin. 427 Antonsen 2002: 70. 428 Spurkland 2005: 82.

135 A Family of Names supposed name for e, i. e. *ehwaR, developed into ON jór, which as a result could no longer stand for /e/. In theory, e, e, with the name jór, could have taken the place of the old jara-rune to represent /j/, when jara > ár. according to Spurkland, the reason why this did not happen is that:

after the transition from *jara >*ara, a good deal still had to take place before *ehwaR became jór, and by that time the system had already adapted to representing /j/ using i. By the time *ehwaR became jór, there was no longer any need for a separate rune for /j/. And this is where the element of graphic simplification might have come in: a rune like e has very little to recommend it once the system has gone in the direction of single-stave runes.429 The process of simplification into single-stave runes has also been called upon to explain why the o-rune was dropped. According to this explanation, it is assumed that the name for o, ōþila, developed into þila. But instead of the rune being kept to represent //, as in the Anglo-Saxon fuþorc, it was dropped, again due, perhaps, to the principle of choosing a simpler graph.430 The result of these simplifications, however, was fewer runes to represent a far more complex system of vowels, with only four runes – ą, A, i, u – to represent ten vowels – /ã/, /a/, /æ/, /e/, /i/, /o/, //, /@/, /u/, /y/ – in contrast to the expansion in the number of runes in the Anglo-Saxon system to represent the increase in the number of vowels in Old English. In view of the complications created by the contraction of the Scandinavian system of runes, I am inclined to take the view that two more important factors than changes in the sound-values of supposed rune-names were responsible for the contraction. One factor was the development towards simpler graphic forms. The other was a drive to reduce the overall number of runes, as appears to have been the motivation behind the reduction in the number of consonantal runes, where B came to represent both /b/ and /p/, t represented /t/ and /d/, and ß represented /k/ and /g/, as well as i and U representing not only vowels but also the semivowels /j/ and /w/ as in Latin. This reduction in the number of graphs representing consonants and semivowels was reflected also in the reduction, rather than increase, in the number of graphs representing vowels, despite the growing complexity in the vowel system as such due to mutation. As Spurkland points out: ‘... it was not so easy to take care of the need for graphic simplification and simultaneously increase the inventory of symbols significantly. One need or the other had to give way; and it ended up being the inventory of symbols and the one-to-one relationship between phoneme and grapheme.’431 Far from being ad hoc, this process appears to have taken account of grammatical structure. As outlined by Spurkland, the runes i, u, A and ą were retained, not only because they had simple graphic forms, but also because in Old Norse they stood for the only vowels – /i/, /u/, /a/, /ã/ – that appeared in inflectional endings, which contained the important grammatical information of whether the word was a verb (in the present or past tense), or an adjective, noun or pronoun (singular or plural, which gender and which case). ‘The other vowels, /e/, /o/, /æ/, /y/, /i/, //, /@/,

429 Spurkland 2005: 82. 430 Spurkland 2005: 82. 431 Spurkland 2005: 83, for this and subsequent citations.

136 Part 1: Sources of Name-forms and Meanings

appeared mainly within the word, where where distinguishing one from the other was not crucial.’ To sum up, it is not possible to infer that rune-names existed in a Nordic context before 800 on the assumption that it was changes in rune-names that were partly responsible for changes in the sound-values represented by the runes. In the first place, the only runes employed in the younger versions of the fuþąrk where this idea might have applied are the runes ÷, J and a, as used to represent oral [a] and nasal [ã]. From epigraphic evidence, it is clear that before the seventh century, the runes ÷ and J were both used to represent [j] but were subsequently employed for oral [a]. The form of the ÷- rune echoed the form of the j-rune in the Kylver and Charnay fuþarks, but J was a new form. The Stentoften inscription’s use of an older form of the j- rune, ú, apparently intended as an abbreviation, alongside the J-rune representing [a] suggests that the two runes were regarded as distinct, and that the J-rune could not be used for the same abbreviation as the ú-rune. There is nothing to confirm that either rune had an established name. The a-rune at this time represented nasal [ã] but was also used incidentally as equivalent to J – for instance, as a parasitic vowel. Between the seventh and ninth centuries, the form of J was simplified to æ, and by the ninth century, in the Abecedarium of St Gall 878, the name ár is recorded for this rune. However, the runic spelling of ár in Leiden Vossianus lat. Q 83, a tenth-century list, was aR, not æR. The name haKal hagal was also spelled with a, which suggests that the scribe, at least, did not always differentiate between æ and a. The name assigned to a in the same list was aUS aus, which could be a reflex of an earlier *ansuR, but could equally well be a Nordic form of the Old English os, which is how the name is transcribed in the Abecedarium. There is no unequivocal evidence to confirm an earlier rune- name of *ansuR for a. The other runes where it has been supposed that an older rune-name changed due to a change in the initial sound-value were all vowels or semivowels which disappeared from the younger fuþąrk. In each case, however, a different explanation is given for the rune’s disappearance. In the case of the W-rune, representing /w/, the disappearance of the initial w- of its supposed name *wunjo meant that it would have come to represent /u/, which was already represented by U, thus making the W rendundant. In the case of e, the supposed name *ehwaR > jór, so that the rune would have been expected to represent /j/, but this phoneme came to be represented by a much simpler graph, namely i. As for the supposed name for the o-rune, *oþila, would have become þila, but rather than representing //, the rune was dropped from the rune-row and // was represented by the simpler i. An explanation for these developments which would appear to be far stronger than that involving rune-names is one based on the dual principles of choosing simpler graphic forms and, in tandem with the reduction of the number of consonantal runes, dealing with the increasingly complex vowel system by reducing the number of runes representing vowels to the minimum required to maintain a comprehensible grammatical structure. (d) The possible role of rune-names in the fuþark and fuþorc in England and Frisia In Anglo-Saxon England and in Frisia, the fuþark evolved quite differently from the Nordic fuþark. However, the assumption that a change in the initial

137 A Family of Names

letter of certain rune-names played a role in the reassignment of sound- values to runes in the Nordic fuþark has similarly been applied to English and Frisian developments.432 In Anglo-Saxon England, changes in the sound-values of certain vowels led to a new rune being placed in the fourth position in the fuþark series. This new rune eventually came to be used to represent [o] rather than [a].433 Also, a later convention in manuscript usage led to [k] being represented in Latin script by the letter . As a consequence, it is usual to refer to the Anglo-Saxon form of the fuþark as a fuþorc. In contrast to Scandinavia, epigraphical versions of the fuþark are not known in either England or Frisia, and only one complete epigraphical fuþorc is known from England, namely that adorning the blade of the sword known as the Thames scramasax, which dates from the tenth century, i. e. after fuþorcs were being listed in manuscripts (see Tables 27 and 29). Here, whereas the fourth rune is transcribed as o, the twenty-fourth rune is transcribed as œ, the i-mutated form of o. Other notable features are the four vowels added after œ – a æ y ea< 434 – and the order of the four runes preceding œ – ng d l m – which deviates from other known versions, where the usual sequence is m l ng d. The forms of the runes for c h j s ng and œ also differ from Nordic older fuþark forms, while the s appears to have been inserted as an afterthought (Figure 1.O).435 Figure 1.O The fuþorc on the Thames scramasax:

f u þ o r c g w h n i j ï p x s t b e ng d l m œ a æ y ea< A fuller, manuscript version of the fuþorc, complete with serifs and several alternative rune-forms, is that given in the copy of the Anglo-Saxon Rune Poem in Hickes’s Thesaurus as taken over from British Library, Cotton Otho B. x (Figure 1.P).436 The runes in this manuscript are written beside each relevant stanza of the poem, but are reproduced below as a sequence, together with a number of alternative rune-forms which are also included in the manuscript. Figure 1.P The fuþorc as copied beside the Anglo-Saxon Runic Poem from British Library, Cotton Otto B. x

f u þ o r c g w w h h h n n i j ï ï p x s t b e m ng œ d a æ y ia ea q k st ğ

432 Derolez 1990: 411: ‘As Ludwig Wimmer suggested just over a century ago (Wimmer 1887:33, 194f., 201 fn.1) it was phonological changes affecting the names of some runes that sparked off the reform that was to produce the Anglo-Saxon fuþorc.’ 433 Since the runes make no distinction between long and short vowels, I indicate length in phonetic transcriptions only where it is significant, as in [ã] > [o:]. 434 The type of transcription I use here for ea< is usually reserved to indicate a ligature, but in this particular instance I follow Page (2003a: 43–4) in using it to indicate that the rune represents a diphthong, possibly formed as a ligature of two runes, e and A (see also Page 1961: 65–79; 1995: 85). 435 Both this and the following transcription are from http://www.arild- hauge.com/eanglor.htm, accessed June 2010. 436 It is possible that some of the forms of these runes were copied from London, British Library, Cotton Domitian A. ix, fol. 10 (see, for example, Bauer 2003a: 91).

138 Part 1: Sources of Name-forms and Meanings

The j-rune L here is a form mostly found in manuscripts, whereas J, used in this series for ia, is the usual form of the j-rune found in epigrahical inscriptions. The last four runes listed in the sequence do not belong to the poem but are given in Hickes’s transcription at the bottom of his page. If these last four runes and the various duplications (for w h ï, etc.) are ignored, an fuþorc can be drawn up, the order of which agrees with that in most manuscripts, except for the runes œ and d, which in a majority of manuscripts are in reverse order:

F u Q O R c g W H n i L 4 p y ÷ t B e m l % o d A a v J q f u þ o r c g w h n i j ï p x s t b e m l ng œ d a æ y ia ea< 1 2 3 4 5 6 7 8 9 10 111213 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

As in Scandinavia, the fuþark in England evolved quite considerably, but there are major differences as well as similarities in the evolution in these regions. The main difference is that, in England, various sound- changes were reflected in the creation of new runes, whereas Nordic sound- changes were largely not accounted for by any increase in the number of runes. In fact, as dicussed in the previous section, the total number of runes in the Nordic fuþark was reduced from 24 to 16, whereas the Anglo-Saxon fuþorc expanded to 29 runes, to accommodate new vowels, and eventually to a total of 32 runes or more, to take account of extra consonants. An important similarity between the changes to the fuþarks used in Scandinavia and England, however, concerns the representation of [a] and related vowels, since it is here that rune-names are assumed to have played a role in the changes. Admittedly, there is no question of any loss of initial [j] in Old English giving rise to a new a-rune, but most scholars have assumed that it was a development in the name for the fourth rune in fuþarks throughout Scandinavia and England – i.e. *ansuR/ansuz > áss/ōs – that led to this rune representing, first, [ã], then [o:]. Unfortunately, this assumption raises more questions than it answers, as is highlighted by my own roundabout description of the rune involved as being ‘the fourth rune in the fuþark series’. The reason for this circumlocution is that the original fourth rune, which is generally supposed to have had the Germanic name *ansuz, was a in the older fuþark, whereas the fourth rune in the Anglo-Saxon fuþorc was a new rune O, with the name ōs and the value [ã] > [o:].437 The original a came to represent [æ], reflecting the fronting of [a] to [æ],438 and was moved to the end of the rune series at number 26, with the name æsc, while (nonfronted) [a] was represented by another new rune A, placed at number 25 in the series. In Frisia, the same developments with regard to a, A and O are found, but the runes for the extra vowels y, ia and ea< are not recorded. Because the runes A and O are found only in English and Frisian inscriptions, and because they appear to coincide with sound-changes that are common to Old English and Old Frisian, the developments in England and Frisia have sometimes been described as ‘Anglo-Frisian’. But as Page emphasizes, there is uncertainty about whether there ever was a ‘period of Anglo-Frisian unity’.439 In his view, the close similarities of the two languages ‘are the effect of convergence, possibly the effect of cultural interchanges between

437 Campbell 1939: §9; Campbell 1959: §121; Luick 1964: §113. 438 Campbell 1959: §131; Luick 1964: §187. 439 Page 2003a: 44.

139 A Family of Names

England and Frisia after the Anglo-Saxon settlement of this country [i. e. England].’440 The issue is of relevance in connexion with the Old English rune-name āc [oak]. The Frisian word for ‘oak’, ēk, could not have been the name of the A-rune. Both āc and ēk derive etymologically from Germanic *aik, Germanic [ai] becoming [a:] in Old English, but mostly [e:] in Old Frisian, although in some contexts it did become [a:]. Another important difference between Old English and Old Frisian is the development of Germanic [au], which became [ea] in Old English, but [a:] in Old Frisian.441 The creation of the new runes A and O and the relation to sound-changes in Old English and Old Frisian have been discussed extensively in recent decades.442 These discussions have tended to focus on two questions: (a) Where did the new runes originate, England or Frisia? And (b) In what sequence were they invented – O before A, or vice versa, or simultaneously? No consensus has been reached. However, I would agree with Parsons on the question of an English or Frisian origin for the two runes; he finds it difficult ‘to challenge Campbell’s conclusion, that “since ai is so differently treated in OE and OFris., it must have been monophthongized after the separation of the two languages” (1959 § 132). Consequently, I conclude that A, and perhaps O as well, was probably an Anglo-Saxon, not an Anglo-Frisian creation.’443 In relation to the sequence in which A and O were created, I also find Parson’s ‘deduction’ in favour of the simultaneous creation of A and O particulary attractive:444

... by adding to the conventional view the observation that the relative order of A and a in the fuþorc is suggestive, it could be deduced that O was not created before A.445 Accepting this deduction, I have further suggested that one particular explanation of the reform, involving the creation of A and O in a single reorganisation occasioned by first fronting, might be favoured (though others remain possible).

440 Page 2003a: 44; Bremmer (1981: 79–83) comes to a similar conclusion. Stiles (1995: 177–220) is dubious about the idea of Anglo-Frisian unity, whereas Fulk (1998: 139–54) argues in favour. 441 Campbell 1939: §9; Campbell 1959: §§132, 252, 255. Seebold (1991b: 508) suggests that, since [a:] developed from [ai] in only restricted cases in Frisian, the A- rune must have emerged in England. Stiles (1995: 185–6) proposes that the Frisian value of [a:] for Awas borrowed from Old English, whereas Fulk (1998: 144–5) sees the development of [ai] as common to Old English and Old Frisian. 442 See, for example: Nielsen 1995: 19–34; Odenstedt 1983:14; Odenstedt 1990: 139–40; Odenstedt 1991: 56; Hines and Odenstedt 1987: 92–3; Hines 1991: 82; Page 1985: 35; Page 1996: 131–50; Bammesberger 1991: 393–7, 405–6; 1996: 15– 23; Quak 1990: 358; Quak 1991: 290; Looienga 1996: 102–6; Parsons 1996: 151– 70; Parsons 1999: 34–9. 443 Parsons 1996: 167–8. Earlier in the same article (1996: 152), Parsons acknowledges that both Quak (1991: 291) and Seebold (1991: 508) had indicated a linguistic objection to the ‘Anglo-Frisian’ origin of A. As pointed out in footnote 441 above, Stiles (1995: 185–6) also thought that the Frisian value of [a:] for Awas borrowed from Old English. 444 Parsons 1996: 167. 445 In a later publication, Parsons (1999: 37) says that he ‘cannot accept that our meagre early archeological evidence, which records two instances of O before the first A, is statistically significant.’ He is referring to the O on the Undley bracteate, dated by Hines (1984: 204–9) to the middle of the second half of the fifth century, and the Chessel Down scabbard-mount, found in a sixth-century grave (Evison 1967: 76), whereas the earliest A occurs on the skanomodu solidus, currently tentatively dated to the late sixth century (Blackburn 1991: 141–2).

140 Part 1: Sources of Name-forms and Meanings

However, Parsons does not feel able to stand by his own suggestion:446

...the balance of probability does not seem to me so much in favour of the second deduction, that A could not have been created before O. The possibility that the order of the displaced and additional runes was not immediately established seems to me too strong to stand by the deduction. The theory is neat and remains attractive, but it is not a firm enough foundation on which to build conclusions. Into this minefield of uncertainty and subtle balancing of probabiltiy and possibility, I would insert three propositions relating to (1) the forms of the a, A and O runes, (2) their names, and (3) their role in the reorganization of the fuþark. (1) Rune-forms The first proposition concerns ideas regarding the creation of A and O. Hempl (and after him Von Grienberger) suggested that the form of A could be the result of a combination of a and i, while O involved a combination of A and n (or, according to Page, a and n).447 Odenstedt objected to these derivations:448 ‘This theory seems to presuppose that the inventor of A knew that OE ā derived from Prim. Gmc. ai, a fact he could not possibly have been aware of. Even more far-fetched seems the notion that O is a combination of a and n = O.’ To be fair to Hempl, he did not imply that the inventor of A was aware that OE ā derived from Prim. Gmc. ai; he simply suggested that the two runic symbols ai, presumably in the spelling of a word like *aik (*aic), gradually blended into a ligature with the i attached to the upper arm of the a. However, I do agree with Odenstedt’s own derivation of A and O:449

The inventor had no doubt noticed that the a and æ, a and o frequently alternated in the same word (…) Since a, æ, o were somehow related in these words, the inventor thought it fitting to allocate symbols to them, based on the old a rune. He achieved this in the simplest possible way, by adding one short angular stroke to the upper arm in A and two angular strokes in O. These strokes are not reflexes of other runes; they are merely arbitrary graphic conventions.’

In effect, while maintaining the link with a, the reduction of the derivation of A and O to ‘mere graphic convention’ divorces the origin of these runes from specific sound-changes in either Old English or Old Frisian: the A represented [a:] whether derived from Gmc [ai], as in Old English, or from [au], as in Old Frisian; its distinguishing feature was its opposition to a, representing [æ], and to O, representing nasalized [ã, õ] and eventually [o]. It is these distinctions that are of relevance to rune-names and to the reorganization of the fuþark, not the chronological or geographical distribution of the first occurrence of the runes themselves. (2) Rune-names Whereas the first proposition decouples the forms of the runes A and O from specific sound-changes, my second proposition decouples the names of these runes from the chronology of the sound-changes and the reorganization of

446 Parsons 1996: 168. 447 Hempl 1896: 175; Von Grienberger 1899; Page 1973: 45; also Bammesberger 1996: 20: ‘It is totally thinkable that O arose out of a combination of a + n.’ 448 Odenstedt 1990: 140. 449 Odenstedt 1990: 140.

141 A Family of Names

the fuþark, in which O displaced a in fourth position and a moved to twenty- sixth position. The orthodox view is that, in assuming the original name of a, *ansuz, the new O-rune also assumed the a-rune’s fourth position – the position of *ansuz being too important to be changed. Parsons expresses this orthodox view as follows:450

The well-preserved twenty-four-rune order and, especially, the evident importance attached to maintaining the name ōs in fourth place, suggest that the order was of great importance. I imagine that it gained this importance because reciting the row in fuþorc order (as, in Anglian Old English, feh, ūr, þorn, ōs, rād, cēn, etc.) was as natural a step in becoming rune-literate as reciting the alphabet is to become roman- literate. Other scholars might invest the order with more mystical importance. I share Parsons’s scepticism of attempts to invest the order of the runes in the fuþark with any mystical importance. At the same time, it is important to stress that the idea of rune-names as a mnemonic tool is not an established reason for the names’ existence, but purely a surmise to explain why runes would have names at all. In fact, I am sceptical of the orthodox view in general and would propose that no complete set of names was assigned to any of the runes, whether in Old English, Old Frisian or any Nordic language, until some time after the new Anglo-Saxon runes A and O had become established, i.e. until after the first fronting of [a] in Old English, at which point the a-rune assumed the value [æ], thereby triggering the creation of both A, with the value [a], and O, with the value [ã]. Parsons, like many before him, sees the Old English names of these runes as ‘suggestive of the early developments [in their sound-values]’,451 and as a guide to their creation.452 But the names are known only from later manuscripts, which document the end of the process of development, not the intermediate steps in that development. (3) Reorganization of the fuþark Nevertheless, it does appear that the the traditional order of the runes in the Anglo-Saxon version of the fuþark was rigorously maintained until it was disturbed by the introduction of the new runes. The fuþark, however, did not need meaningful names for it to maintain the order of the runes any more than the Latin alphabet needed meaningful names for it to maintain the order of its letters, or indeed for the alphabet to be learned. Yet there must have been a reason for the new runes to have been allowed to disturb the fuþark order by just so much and no more. The reason was, I suggest, part of a general reform that led not only to a modest reorganization of the fuþark order, but also included the assignment of names and the graphic reforms described by Parsons, affecting the runes c j s ng and o (as exhibited in the Thames scramasax) and the addition of runes such as k ğ st and perhaps q (as found at the bottom of Hickes’s reproduction of the Anglo-Saxon Rune Poem).453 My third proposition, therefore, is that rune-names did not play a

450 Parsons 1996: 162–3. 451 Parsons 1996: 154. The list of scholars who have thought similarly is impressive; they include: Wimmer 1887: 194–5, 201 n. 1; Hempl 1896: 174–6; Von Grienberger 1899: 19–20; Von Friesen 1918–9: 25; Keller 1938b: 24–5; Arntz 1944b: 230; Musset 1965: 185–6; Page 1973: 44–5; Elliott 1989: 44; Derolez 1990: 411–2; Bammesberger 1991: 393–7. 452 Parsons 1996: 166. 453 Parsons 1999: 85–130. The double-barred h, H, was an earlier innovation, found also elsewhere on the Continent.

142 Part 1: Sources of Name-forms and Meanings role in the reorganization of the fuþark but were assigned either contemporaneously with or soon after the reorganization and graphic reforms. The reorganization of the fuþark was primarily a matter of reordering the vowels, the basis for which was not a purely internal runic matter, but was a consequence of a comparison with the Latin alphabet. A comparison of the order of vowels in the Latin alphabet with the order in the fuþark and Anglo-Saxon fuþorc can be made as follows: If the fuþark is divided into its three ættir (ignoring minor variations in forms and position, such as the order of d and o):

ætt 1 F u Q a R k g W f u þ a r k g w ætt 2 h n i j p 4 M S h n i j p ï z s ætt 3 t B e m l 5 d o t b e m l ng d o it will be seen that the distribution of the six vowels over the three ættir is:

ætt 1 u a u a ætt 2 i 4 i ï ætt 3 e o e o On the basis of the Anglo-Saxon Rune Poem, the sequence in the fuþorc was:

u O u o i 4 i ï e o e œ A a a æ v J q y ia ea<

If the eight vowels in the columns u to A (u to a) and O to a (o to æ) are considered in the first instance, the sound-values of those in the column u to A (u to a) correspond to the vowels in the Latin alphabet but in reverse order and without O: Latin U I E A Runes u i e a while the vowels in the column O to a (o to æ) comprise phonetic parallels with the first column: o ï œ æ. These considerations, together with the assumption that new characters were usually added to the end of an alphabet, lead me to postulate a three- stage development of the fuþark, with an intermediate stage between the older fuþark order and the revised Anglo-Saxon fuþorc:

143 A Family of Names

ætt Stage 1 Stage 2 Stage 3 (fuþark) (hypothetical) (fuþorc) 1 u a u a u æ u a u o u O 2 i ï i 4 i ï i 4 i ï i 4 3 e o e o e o e o e œ e o ‘extra’ a ã,õ A O a æ A a

Stage 2 represents the situation where the Old English fronting of [a] > [æ] and nasalization of [an] > [ã,õ] were recognized and a need was felt for new signs to represent on the one hand [a:] < [ai] and nonfronted [a], and on the other hand [ã,õ] < [an]. The order of A and O could be arbitrary, but I agree with Parsons’s view that the shapes of these runes are ‘suggestive’ and that A, with a single extra stroke, was probably created before O, with two extra strokes.454 In stage 2, original [o] had yet to undergo i-mutation and continued to be represented by o alongside the new rune O for [ã,õ] < [an].455 There are no records of fuþarks to confirm how the runes at this stage might have been arranged in sequence, but it is conceivable that the new runes were added to the end of the sequence. When original [o] did eventually undergo i-mutation to [œ] in certain environments, then:

(a) the original rune o came to be used for mutated [œ], in the same way that the original rune a came to be used for fronted [æ]; (b) the new rune O came to be used for unmutated [o], in the same way that the new rune A was used for nonfronted [a].

Finally, in stage 3, a changed places with O, so that, phonetically, [æ] paired with [a] and [o] paired with [u]. An important implication of this reconstruction, if it is accepted, is that the fourth rune in the fuþark did not acquire a new sound-value because the supposed name of this rune evolved from *ansuz to ōs [(pagan) god] and was subsequently ‘depaganized’ to ōs [mouth]. Rather, as part of a structural reform to account for the shifts in the Old English vowel system, two new runes, A and O, were invented, based on but differentiated from a with regard to both shape and sound-value, and given names to reflect the new system. Similarly, the positioning of Aand O and the repositioning of a were not ad hoc, but were a consequence of the rationalization that reflected this new system. A further implication is that, if the name ōs did not derive from *ansuz but was consciously assigned to O in order to correspond with its value [o], as distinct from o with its value [œ], then a probable terminus post quem for the assignment of this name, and possibly of all the other names, would be the i-mutation of [o] > [œ]. In other words, as with O, so with o – the original name of o was not *oþila, and the sound-value of this rune did not change to [œ] when the name changed to œþil, as is generally presumed;

454 Parsons 1996: 167. The order and attribution of names to the a and A-runes in the fuþorcs accompanied by the isruna tract differ from those of other fuþorcs (see comment in Table 27 (b), and Derolez 1954: 129). 455 Parsons 1996: 159. If O had represented [o] at this stage, it would also have undergone i-mutation, which means that I disagree with Oldenstedt (1990: 139): ‘It would therefore seem that from at least 450 to the period of i-umlaut (when o was reserved for œ) o and A were allographs for o.’

144 Part 1: Sources of Name-forms and Meanings rather, o acquired the name œþil only after the i-mutation of [o]. This chronology is at odds with the view of those who interpret the óss stanza on the Icelandic rune-poem (Table 31-2 and 31-3) as alluding to Odin and as a consequence consider the establishment of Odin as hea of the nordic pantheon of Æsir in the early part of the Migration Period (AD 300–500) to constitute a terminus ante quem for the creation of the rune-names.456 My arguments against this view will be given in my discussion of the individual rune-names in §2.4.4, s.v. Ōs, óss. The three runes that are the subject of the last three stanzas of the Anglo-Saxon Rune Poem – v y, J ia, 6 ea< – were probably added to the fuþorc sequence after the A- and a-runes had been established in positions 25 and 26. They have so far not been found in a Frisian context. Rune 27 – v, y or X – was almost certainly created on the basis of the u-rune u, in order to represent a mutated u. The addition of | , ┬ or x as an apparent diacritic is comparable to the idea that the strokes added to a to create A and O were also diacritics.457 Its name yr appears to confirm the connexion with the u-rune’s name ur. The J-rune is not included in the Thames scramasax fuþorc, in which 6 follows immediately after X, and it occurs only rarely in manuscript fuþorcs.458 The names of these runes will be dealt with in Part 2, but as far as their order in fuþorc is concerned, I suspect that their inclusion at the end of the Anglo-Saxon Rune Poem may have been due to Irish influence. Daunt points out that Northumbrian and Mercian scribes probably learnt to write Latin letters from Irish missionaries.459 Indeed, I wonder whether it is a coincidence that the Rune Poem ends with a set of five supplementary vowels, similar to the set of vowels that were eventually assigned to the five supplementary characters (forfeda) at the end of the ogam series – see §1.4.7 (b). Not only does the similarity lie in the number of supplementary characters and the fact that they are all vowels, but the quality and order of the last three vowels in the forfeda show a degree of correspondence with the last three at the end of the Rune Poem:

456 See Bauer 2003a: 34, following the ideas of Arntz 1944: 191. 457 Daunt 1939: 120: ‘The same thing was done when it was necessary to adapt the Late Latin alphabet to the Old English sounds. In Latin V an i wasVi inserted to represent [y], and this is found in the coin of Eanred of Northumbria, 807–41.’ See Arntz 1935a: 148. See also O’Neill 2009: 13–14, who, in pointing to the possibility that runic v may have been modelled on the Old English digraph ‹ui› (for i- mutated /u/, later written ‹y›), wonders why there were not similar runic creations for the digraphs ‹oi› and ‹ei› (for i-mutated /o/ and /a/, i.e. ‹œ› and ‹æ›). There was, however, a form of o with the addition of i, namely ô, which does not appear to have been universally accepted, while the specific system of diacritics created early on to account for the mutation of /a/ (i.e. a A O) may have precluded any further ‘tampering’ with these vowels. 458 Some manuscripts listed the sound-value of the J-rune as io, with the name ior, as, for example, in the list in the now lost British Library, Galba A. ii, in Table 27 (a). See also footnote 461 below. 459 Daunt 1939: 111. Critical and more favourable views of Daunt’s ideas are reviewed by O’Neill (2009: 5–7), who himself puts the case for Irish influence on Old English orthography.

145 A Family of Names

forfeda (Aur. 1138/4229)460 ea oi ui io æ Rune Poem a æ y (i.e. ui) io or ia eo or ea461

Admittedly, the sound-value of the Anglo-Saxon io/ia-rune, [jo, ja], is not the same as the forfeda io, and Old Irish æ does not correspond with the Anglo-Saxon ea. Also, manuscript listings of fuþorcs include an assortment of supplementary runes at the end of their lists. On the other hand, the first additions after the original twenty-four runes are invariably the supplementary vowels treated in the Rune Poem, even if the actual sequence of a æ y ia ea is found only in this poem.462 In fact, the similarities are especially notable if account is taken of the creation of the first two ‘supplementary’ vowels. In the Rune Poem, these vowels had the values [a:] and [æ]; in the Irish forfeda, they were [e:] and [o:]. According to my suggested stage 2 in the evolution of the Anglo-Saxon fuþorc, the Old English vowels were [a:] and [ã, õ] > [o:]. According to the Auraicept, the long vowels [e:] and [o:] were added to the ogam series ‘in readiness for the adoption of Greek and Latin words into Irish’.463 I would suggest that the addition of the first two supplementary vowels at the end of the Anglo- Saxon fuþark-cum-fuþorc – initially [a:] and [ã, õ] > [o:], subsequently [a] and [æ] – is to be seen as separate from the other three supplementary vowels, in the same way that the addition of the ogam supplementary vowels [e:] and [o:] was a separate, first step in augmenting the ogam series, as described in §1.4.7 (b). I do not intend to imply that learning to write in Latin letters from Irish missionaries was responsible for the creation of the new runes A and O. The dates and first occurrences of these runes preclude any such idea. What learning to write in Latin letters probably did bring about, however, was a reassessment of the runic vowel system and a rearrangement of the sequence of vowels in the fuþark. It was in this context that a reform of the runic graphs as proposed by Parsons took place. It was in this context, too, that a set of names was created for the runes of the fuþorc, to emulate the names invented for the characters, which in their turn had been invented to emulate the names of Hebrew letters.

460 A similar series is found at Aur. 5421–2. At Aur. 1057, the values are given as aea, oi, ui, iiii, aaee and at Aur. 1143 ee(aa), oo, uu, ii, aa(ee). See also ogam series nos. 28 and 30 (Book of Ballymote, fol. 312). 461 Daunt (1939: 116–17) wonders whether the confusion between eo, ea writings in Northumbrian and other Anglian texts may be due to the Irish practice of interchanging a in diphthongs such as aí, aé with o, so that the same word can be spelled aís, oís or oés [people]. The same might be said of the alternation of Old English iar and ior. 462 According to Page (2006: 216–32), five of the eight manuscript fuþorcs listed by Derolez (1954), apart from the Rune Poem, include the supplementary J-rune or its name. Two list the name ior among the names of other supplementary runes; two give J as a supplementary vowel, with the value œ, between the runes d and a; while one gives J as a supplementary vowel after the sequence a æ y ea. Manuscripts that list the supplementary vowels a æ y ea give them in this order, except Vienna 795, which gives a æ ea y. The situation is more complex than this summary suggests, but cannot be explored in detail here. 463 Aur. 1370–4/4427–30.

146