Thesaurus Palaeohibernicus/Description of the MSS (Volume 2)

From Wikisource
Jump to navigation Jump to search
Thesaurus Palaeohibernicus, Volume II  (1903) 
Whitley Stokes and John Strachan, eds.
Preface to Volume 2
[ ix ]


1. Codex Augustini Carolsruhani.[1]

This manuscript formerly belonged to the monastery of Reichenau[2] and is now in the Hof- und Landesbibliothek at Carlsruhe, where it is numbered Codex Augiensis cxcv. It consists of 47 leaves, of which 7, 8, 19, 20, 21, 32, 33, 34, 35, 36, 37, 41, 42 are palimpsest. Most of the pages contain two columns; ff. 26, 28, 30, 31, 33, 37, 45, 46, 47 contain only one. Fo. 1 and fo. 47, which were once attached to the inner sides of the wooden cover, but have now been separated from it and are included in the pagination, do not belong to the codex. Fo. 1 is very faded; in col. 1 twenty-four fresh lines have been written, according to Windisch, probably in the same hand as the bulk of the codex. The greater part of the codex (fo. 2–fo. 39 col. 1) is occupied by the Soliloquies of St Augustine; in these folios and in the twenty-four lines added on fo. 1 the Irish notes and glosses are found.

Windisch considers that the Irish glosses are probably in the same hand as the Latin text, some of them having been written at the same time as the Latin, others having been added afterwards, but that some of the corrections may be in another hand.

According to Mone, Holder, and Windisch the codex belongs to the ninth century. This date is supported by certain linguistic peculiarities in the MS., particularly by some changes which have taken place in final vowels: rei = rée 2d1, etargna = etargne 6a2, taudbartha 7a2, reta corptha 8b1, besgna 13d1, insarta 27b1. But the glosses may in whole or in part be older. Such errors as sochtmacht 5d4, adromarsu 7a1, asrubartmart 12c1, aim 14a2, caisin 28r1, point to transcription[3].

[ x ]

2. Codices Bedae.

Irish notes and glosses have been discovered in two manuscripts of Bede:

(a) Codex Carolsruhanus (Augiensis) clxvii.
(b) Codex Bedae Vindobonensis n. 15298 (or Suppl. 2698).

These two codices shew the same recension of the Latin text. In part the Irish glosses are identical in both. These common glosses must have come from a common source; they have not been copied from one codex into the other.

(a) Codex Augiensis clxvii[4], nunc Carolsruhanus.

This manuscript once belonged to the monastery of Reichenau, and is now in the Hof- und Landesbibliothek at Carlsruhe. It now consists of 49 leaves, of which, however, 5–12 belong to a distinct codex of Beda. The manuscript contains a selection of his works. It is interspersed with notes and glosses in various hands, from one of which come the Irish notes and glosses.

Various chronological notes are added on the margins, most of which are printed below, p. 283. Others are:

fo. 4r [marg. sup.] óengusso†.
fo. 15b [marg. sup. to dcccxvii] aed rex hiberniæ moritur[5].
fo. 17a [Mai.] H xiiii K a u kl. depositio sancti germani episcopi.
fo. 17b [Iul.] Kii n m g uiii id Nataƚ sancti Chiliani cum sociis suis.
fo. 17b [tr 1 β e b d b u 7 Klb. in marg. d] bás muirchatho maicc maile dúin hicluain maccunois á imda chiarain .x. anno.
fo. 17c [marg. inf.] IN gallia sancti Quintini cuius corpus post annos .lv. ab angelo reuelatum est uiii Kl. iuli...7 (to Aug. icu ice ii Kl. IN .h. xiiii đ. h. x).

From a series of marginal entries in another hand, the last of which is dcccxlviii vi·m xlviii ab initio mundi[6] Zimmer concludes that the Latin text was written before the year 848. The date of the addition of the Irish glosses he seeks to determine from the marginal note on fo. 17b in the hand of the scribe who wrote the Irish glosses. The Muirchath mentioned there is identified by him, with great probability, with the Muirchath whose deposition is mentioned in the Annals of the Four Masters a.d. 821. If immediately after his deposition he retired to Clonmacnois, his death, if the identification be correct, would have taken place about a.d. 831, i.e. in the tenth year of his retirement. Zimmer[7] supposes that the scribe was on terms of friendship with Muirchath before he left Ireland, and regards it as probable that the glosses were written about 850 a.d.

[ xi ]With this date the language of the glosses would harmonize: note in particular the treatment of final vowels in aicneta 18b12, tricha 31c9, aesca 32b1, fotha 33b4, oldata 33b8. Attention may be directed further to aine = óine 31c4, and to dunnai 18b10, saidai 18c3, to the single consonant in mais 18a1, deis 19c2, imatrebdidiu 36a2, rucad 40a2, oca turcbail 18c2, ina riaglaib 33b13, and to leissem 32a5, 32b6, lingidsem 31c8, and to fail 18c4.

That the Irish glosses have been copied, in part at least, from an older manuscript is evident from their coincidence in part with the glosses in the Vienna Beda.

(b) Codex Bedae Vindobonensis[8].

In the Royal Library of Vienna there is a fragment, which probably dates from the ninth century, consisting of four leaves of Beda’s De Temporum Ratione, in double columns. It is numbered n. 15928, or suppl. 2698, and at the bottom of col. 1, p. 1 it bears a stamp ‘E cod. P.V. 2269 [Rec. 429]. The leaves have suffered much injury; in some places the margins have been torn away, parts are very hard to decipher, parts are altogether illegible. Between the lines and on the margins are notes and glosses, Irish and Latin, in various hands.

3. Codices Canonum Hibernicorum[9].

(a) Corpus Christi College Cambridge, Parker, 279.

This manuscript is written in a continental hand, and has been assigned to the ninth or tenth century[10]. Among other texts[11] it contains canons excerpted from the books of Exodus, Leviticus, Numbers, Deuteronomy. In these excerpts are found Irish glosses, transcribed from the Irish original by the same hand as the Latin text. The last entry, prescribing the penalty for shedding a bishop's blood, corresponds with the Ancient Laws of Ireland iv. 363, ll. 26–27[12]. The gloss on colirio (leg. collyrio), anre, is British, and is the equivalent of the Irish innrach ‘a tent or plug used to keep wounds open.’

(b) Codex Sangermanensis 121 (now MS. Lat. Paris. 12021)[13].

This manuscript is preserved in the Bibliothèque Nationale in Paris. It has been assigned by some to the eighth century, by others more correctly [ xii ]to a somewhat later date[14]. It contains in ff. 33–127 a collection of Irish canons, with a few Irish glosses. Both text and glosses have been copied from an older codex.

4. Codices Libri de Computo.

(a) Codex Vaticanus n. 5755[15].

This codex consists for the most part of a copy of St Augustine’s work De Trinitate: to this, however, fo. 2, 3 and fo. 63–73 do not belong, but contain fragments of a computus. Fo. 2, which begins with nihil remanserit and ends with sí quando mense martio xiiii · luna pascalis incurrit xxxiii · regulares in primís teneas: ex quibus æpactas cuius uollueris anni deducas, contains the Paschal Arguments of Dionysius viii, ix, x, and xiv. On the margins of this folio there are copious notes in Latin, with the exception of one which is partly in Irish; between the lines are found notes and glosses in Irish and Latin. The codex has been assigned to the eleventh century[16]; on fo. 2, however, the Irish seems to represent the language of the eighth century, and there is no clear evidence that the Irish glosses have been copied.

(b) Codex Nanciacensis[17].

This is a fragment consisting of a single leaf, written in an Irish hand of the ninth century, attached to the inner cover of Cod. 59 of the Library of Nancy. It contains copious Latin notes and glosses on the margins and between the lines, and also a number of Irish glosses. The Latin text contains the Dionysiac Paschal Arguments xi, xiii.

5. Codices Eutychii.

(a) Codex Vindobonensis n. 16[18].

This manuscript, which formerly belonged to the Columban monastery of Bobbio, is now in the Royal Library of Vienna. Ff. 57–68 contain a text of Eutychii de discernendis coniugationibus Libri II. written in a hand of the eighth or ninth century, with Irish glosses. That these glosses have been copied is shewn by the fact that glaidim has become attached to rudo instead of to erado.

[ xiii ]

(b) Codex Parisiacus, MS. Lat. 10,400[19].

A manuscript in the Bibliothèque Nationale, consisting of fragments of MSS. found for the most part in bindings etc. The fragment ff. 109, 110 is in an Irish hand probably of the ninth century; it is a bookbinding and is very hard to decipher. In one instance at least beicim, the Irish gloss, is attached to the wrong Latin word.

(c) Codex Parisiacus, MS. Lat. 11411[20].

This is another fragment, also a bookbinding, in the same library, probably of the ninth century. According to Dr Friedel the glosses are in a different ink and thinner. He thinks that the leaf belonged to the same body as 10,400. Some of the Irish glosses are attached to the wrong words.

6. Codex Latinus Monacensis.[21]

In fo. 222a–226b of this codex, which has been assigned to the ninth century[22], is an alphabetical Latin glossary, in three columns, with glosses added in various hands. Among these glosses there are a few Irish ones, written in the same hand as the text.

7. Codex Iuuenci.

This codex is in the University Library of Cambridge, where it is numbered Ff. 4. 42. It is thus described by Hardwick and Luard: “A quarto, on parchment, 108 leaves, about 28 lines in each page; handwriting as early as the ninth century. The date 1233 is twice written in the margin, but if meant to indicate the time at which the MS. was executed, it is far too modern. ‘Quatuor Euangelia a Iuuenco Presbytero pene ad verbum translata,’ so reads the colophon….” The text contains a large number of British, and a few Irish, glosses[23].

8. Liber Ardmachanus[24].

The Book of Armagh is a small vellum quarto, containing 221 leaves, in the library of Trinity College, Dublin. The writing is generally in double columns (very rarely in three), and all seems the work of the same scribe, Ferdomnach, whose name occurs (fo. 214a) in the following entry: pro [ xiv ]ferdomnacho orés. There were two famous scribes of this name connected with Armagh, one of whom died a.d. 727, the other a.d. 845. That the scribe of the Book of Armagh was the latter has been ingeniously proved by Bishop Graves[25] from the following half-erased entry in a semi-Greek character, which occurs in fo. 52 b:

.….ακhνκ λ..
.. μ…ε δικταντε
…βακh · hηρηδη πατ
ρικιι · ϲκριπϲιτ—

As the only comarbe of Patrick whose name ended in ‑bach was Torbach Bishop Graves restored the entry thus:

f domnach · hvnc · lib
e rvm · : : : e dictante
r torbach · herede · pat
   ricii · scripsit

As Torbach held the primacy for only one year and died in 808, the codex, or at least this part of it, must have been written either in 807 or 808. That the entry refers to the former year is proved by the following entry on fo. 36a:

εxπλικιτ · αευανγοε
λιωn · κατα mat
thvm · ϲκρiπτvm
ατκnive φitvm ·
in φηρια · matthi

= explicit euangelion κατὰ Mattheum scriptum atque finitum in feria Matthei. For Torbach died on the 16th of July and this entry was made on the 26th of September, the Feast of S. Matthew, The marginal entries have in part been mutilated by the cutting of the margins.

The Book of Armagh contains a transcript of older documents. A full description of the contents will be found in the edition by W. S. of the Tripartite Life of Patrick, xci sq. The following portions have been published in the present work:

I. [26]The Irish names in Muirchu Maccu Machthéni’s memoirs of S. Patrick (fol. 2a1–fo. 9a2). This Muirchu professes to write in obedience to the command of Bishop Aed of Sletty, † 698. The importance of these notes for the language of the time has been pointed out by Thurneysen[27]. The following points may be noted:

Long ē has not yet become ía: Fēcc = Fíacc, Cērrigi = Cíarraigi. Of ō there is an example in the Latinised Coolennorum. Unaccented ĕ and ŏ between non-palatal consonants are preserved: Clocher = clochar, Findubrec, Lucetmail, Ulod, Lothroch[28]. Oi [ xv ]does not yet appear as oe: Loigaire, but by ai is found ae: Lucetmail and Lucetmael. After a non-palatal consonant final ‑i is expressed by ‑i, not by ‑ai: Calpdi, Ferchertni, ferti, Machi. But in the same position we find ‑e: Mache, Slane, ‑æ: Arddæ, Esrachtæ, Machæ, and ‑ae: Greccae, Machae. The gen. sg. is ‑o, not ‑a: Dego.

II[29]. The Irish names in the miscellaneous notes on the life of S. Patrick, which Bishop Tírechán is said to have written ex ore uel libro of his foster-father or tutor (aite) Bishop Ultán, † 656 (fo. 9a2–fo. 16a1). The language shews the same characteristics as I, only not so strictly:

ē: Cēnachtæ, Cēnnani, Cēranus, Cērrigi, Clēbach, Fēccus, Fēchach, Fēchrach, Neel. It has become ea in Druimleas (cf. feadinne in the glosses on Philargyrius), and at the end of a word, Bandea (cf. deadía in the Cambray Homily). It has become ia in fīan.

ō: Bōin, Booin, Boonrigi, Bōidmail, Coonu, Clōno, Crōchān, Gōsacht, Gōsachtus, Irlōchir. But also ūa: Būain, Būas, Chonlūain, Es Rūaid, Mūaide, Latinised Muadam, thūaithe. It will be observed that, except in Būas, ūa appears only before a palatal consonant.

ĕ preserved: Ached, Argetbor, Echredd. By Congleng and Ercleng, however, appear Conlang and Erclang in the list of names on fo. 9b2.

ŏ preserved: Adrochtæ, Cenondas, Hirotæ, Martorthige, Nioth, Teloch. The later Fochlad appears both as Fochloth and Fochluth. Corresponding to the Ogham name Gosuctias[30] we find Gōsacht, where Gōsocht might have been expected.

After a non-palatal consonant i: Argi, Cetni, Congi, Chungi, Elni, Endi, Ferti, Fidarti, Luchti; also in the interior of a word: Amolngid, Caplit, carric, Cerrigi, Irlōchir, Taulich, sertib. But Chungai and Irai, Humail.

After a non-palatal consonant final ‑e is commonly written ‑æ: Adrochtæ, Brigtæ, Cēnachtæ, Comgellæ, Corræ, Ercæ, Herotæ, Machæ, Sinnæ, Succæ, But also Core, Erce (MS. Cerce), and once Machae. Æ appears sometimes after a palatal consonant: Columbcillæ, Dumichæ, Slicichæ[31].

oi regularly: Coimanum, Loiguire, loigles, Oingus[32].

ai happens to occur only before a palatal consonant: Maini, Boidmail[33].

From ‑i- and ‑u- stems the gen. sg. is regularly in ‑o: Ailello, Alo, Arddsratho, Clōno, Drommo, Fergusso, Itho, Nento, Temro. Once a: Airddsratha.

III. Additions to the notes of Tírechán in Latin and Irish (fo. 16a2–fo. 18b2)[34]. The language of these additions shews later characteristics than the notes of Tírechán. It seems on the whole to represent the Irish of the early eighth century, though some later forms may have been introduced by the copyist, e.g. ni fetorsa = ní fetarsa:

ĕ preserved: adcotedæ, atrópert, ōchter, toidached. But adopart, contubart.

ŏ preserved: cathboth, conacolto, edocht, fithot, oitherroch. But aidacht, cathbath[35].

[ xvi ]ē appears only in dfēch, in a legal formula, and in pretonic position re m‑bás. It appears once as ea in Druimm Leas, as īe in Fīechrach, and in pretonic position in iersin, iersuidiu, but commonly as īa: blīadin, Cīarrige, Críathar, Dīarmuit, Fíacc, aníar, Lías, iarsuidiu, iarsin, iartain.

ō: clōin, diróggel, lóg, forlóg, ōchter, óg, ó, ódib, ótha. More rarely ūa: būachaele, clūain, hūaimse, húad.

‑i after non-palatal consonant: cuci, Dumi, Endi, Ferni, léni, lobri, orpi, Achid, Alich, anis, argit, arith, Berich, blīadin, Brēchmig, Carnib, cennadich, Cīarrige, Cremthinn, cumil, dlomis, eclis, Feradig, forrig, manchib, Ōchtir, Patricc, pridchiss, Tamnich. But ungai, abbaith, argait, Broccaid, cétaig, gabais, maccaib, Diarmuit, Gabuir, idpuirt, manchuib, Themuir.

e after non-palatal consonant. e: blīadne, omne, tigerne. æ: daltæ, damnæ, Endæ, Gimmæ, Odræ, mac Rímæ, ríthæ, ‑ructhæ, sommæ, Tamlachtæ, telchæ, ungæ[36]. ae: adcotedae, Machae, ungae.

oi: cōicid, Lōiguire, nóib, nōinomne, ōinsetche, Toicuile. But sóer, ōentuistiu.

ai: Bāitán, Cāichán, Fāilgnad, Fáiláin, Forfāilid. But Āed passim[37].

Pretonic to- appears in the verb as du: dutét, dulluid, dufōid; pretonic as di‑: digéni, diróggel, or du: duchooid. Before nouns the prepositions do and di are still kept apart.

From ‑i- and ‑u- stems the gen. sg. is always in ‑o‑: brátho, Conacolto, Daro, Drommo, Dublocho, Fergosso, Fetho, Forfáilto, Feidilmedo, Fedeilmtheo[38], locho, Santo.

In the verb may be noted: boie, fācib (by fácab), ‑fetor = ‑fetar.

duaberrad for dia berrad is peculiar, but it cannot be put down as an archaism, for already in the Cambray Homily we find dea.

IV. [39]A series of notes or catchwords, written in a very small hand and abounding in contractions, which represent in the main that portion of the Tripartite Life which is not embraced in Muirchu’s Memoir and Tirechán’s notes (fo. 18b2–fo. 19a1). The language shews later peculiarities more than the foregoing pieces. Note, for example, Dīarmit, Fīac, Fīachrach, Būail…, Esrūaid, Mūadan, Lūan, Tūadmumu, Achad, Láthrach, Itha, Aeda.

V. [40]The Irish names in Muirchu’s prologue and in the headings of his chapters (fo. 20a).

VI. [41]The Irish names in the Liber Angeli (fo. 20b1–fo. 21b2).

VII. [42]Irish glosses on fol. 6a–21b.

VIII. [43]The Irish names in the Confessio Patricii (fo. 22b1–fo. 21b2).

IX. [44]The Irish glosses on the New Testament (fo. 31b2–fo. 190).

That these glosses are later than the older Patrician documents is shewn by the diphthongization of ē: iar, iarfichid, (i)armifoistis (but ren‑), and of ō: [ xvii ]hūasal. That they are earlier than the bulk of the glosses in Wb. seems to be indicated by the fact that the gen. sg. of ‑i- and ‑u- stems is always in ‑o: folo, gléso, senso, spirto. After a non-palatal consonant i is regularly expressed by i: gabis, fodil, samil, delbich, grádich, æclis, dālire, dāldi, etc.[45] After a non-palatal consonant final ‑e is expressed by e: derbensde, tarsende, or by æ: rúnæ, sechtæ, etc.[46] The diphthong oe does not yet appear.

9. Codices Philargyrii[47].

Two series of excerpts from Iunius Philargyrius’ scholia on the Bucolics are preserved in three manuscripts:

P. (= N. Hagen) = Codex Parisiacus Lat. 7960, saec. x. In this MS. the second series of excerpts is found fo. 1a–14a, the first series of excerpts on fo. 14'b–41b.

L. = Codex Laurentinus, Plut. xlv. Cod. 14, s. x. According to Hagen the manuscript was brought from France. The colophon to the first series (Rheinisches Museum, N. F. xv. 119) seems to contain a Latin rendering—Fatosus—of the name of the Irish excerptor, which may have been Toicthech: see The Academy for July 28, 1894, Rev. Celt. xvi. 123.

P². (= P. Hagen) = Codex Parisinus 11308, s. ix.

With regard to the relations of the three manuscripts the following may be noted. P and L are very closely related and come from a common source, but P was not copied from L nor L from P. P² is fuller and more correct than P and L, but is not the source from which they have been copied. Errors common to the three MSS. indicate that they go back to a common source, into which many mistakes had already crept through the transcriber’s ignorance of the Irish language and his unfamiliarity with the script. The three codices are all in a continental hand. It is not improbable that the archetype from which all the three MSS. finally came was written by a continental scribe.

The Irish of the glosses is of the same archaic character as that of the Book of Armagh. Its peculiarities have been discussed by Thurneysen, Celt. Zeitschr. iii. 52 sq.[48]

[ xviii ]Codices Anonymi breuis expositionis Vergilii Georgicorum[49].

P. ( = N. Hagen) = Cod. Paris. Lat. 7690.

P². ( = P. Hagen) = Cod. Par. 11308.

G. = Codex Burmannianus, nunc bibl. Leidensis publ. Lat. n. 135, s. xi.

10. Codices Prisciani.

The following manuscripts of Priscian contain Irish glosses and notes:

A. Codex Sangallensis No. 904.

B. Codex Augiensis No. cxxxii, nunc Carolsruhanus.

C. Codex Leidensis. Cod. Lat., G7.

D. Fragmentum Ambrosianum. Cod. A. 138 sup.

It has been shewn by Hertz[50] that the first three codices come from a common source, and that A and C are especially closely related. From a photograph of a page of D it is clear that it belongs to the same family[51], and that it is even more closely related to C[52] than A is, while at the same time it can neither have been the source of C nor can it have been copied from C.

Irish glosses are by far most numerous in A. In part B has similar Irish glosses to A, but it has also many Irish glosses peculiar to itself. Most of the Irish glosses in C are also found in A. Corresponding to the Irish glosses in D are generally found Irish glosses in A, but while they agree in sense, they often differ in form[53]. It is to be observed further that no two of the other collections have an Irish gloss in common which is not shared by A; the only instance, however, is ḟoilenn (Sg. 93a 1) = failen (Pcr. 37a 1) = foilenn (Pld. 59a).

In addition to the Irish glosses the MSS. have a larger or smaller number [ xix ]of Latin glosses. B and C have sometimes a Latin gloss corresponding to an Irish gloss in C[54]; occasionally the Latin gloss is found in both B and C[55].

A. Codex Sangallensis 904[56].

This manuscript is in the Stiftsbibliothek of St Gall. It consists of 240 pages[57], and contains the Latin text of the first sixteen books of Priscian and of part of book seventeen down to the word “naturaliter” Hertz ii. 147 l. 18. Traube has shewn that it was written by some of the friends of Sedulius; he supposes that it was copied in some Irish monastery in the first half of the ninth century, and brought by wandering Irishmen to the continent[58]. The Latin text is in different hands[59]. The margin has been cut in binding, so that some of the notes have been mutilated, particularly on the upper margin.

The manuscript contains between the lines and on the margins both Irish and Latin notes and glosses; the Irish, however, predominate. The glosses are written in different hands from those which wrote the Latin text. At least three hands are to be distinguished[60]. The chief glossator (A) extends to 65b5. 64b6 .i. uás · lestar is from the second glossator (B); 65b7 atriur is again from A. B wrote the glosses from 65b8 to the end of the page, the glosses on 66a and 66b, the glosses on 67a and the glosses on 67b except 67b19, and 22, which are from A, who wrote the bulk of the glosses from 68a to the end. A third hand (C) has added scattered glosses throughout, often short Latin explanations. With regard, however, to the scattered glosses which do not come from A or B, Professor Thurneysen writes: “The question is more difficult how far the glosses of another hand (i.e. than A and B) have the same [ xx ]or different authors, for in the case of these isolated additions it is difficult to decide as to the hand. I have consequently examined them again:

“Certainly C are: 15b 11 buaid lię, 33a 24 genus doloris, 49a 2 lothor, 54a 12 iouis, 57b 5 sabinus, 62b 10 ligo .i. bacc buana fínime, 63a 12 .i. mocoll lín, 67b 14 soror uiri, 69b 10 .i. obedientia, 144b 2 .i. hastas colligo and ꝉ quero ꝉ populo alloquor, 145a 5, 6 .i. nutrio (but the marginal arbiathim may be from the usual hand).

“Probably C: 46b 13 ingen, 50b 21 .i. féle, 63b 17 bestia.

“The following shew lighter ink than C has elsewhere: 46a 2 ꝉ bóc, 46a 3 ꝉ, 53a 15 tened, 92a 4 uestimentum.

“Doubtful if C: 46b 13 rite (may be from the writer of the Latin text), 49a 3 and 7 .i. derg.

“A different ductus probably appears in 52b 9 cís rigda, 53a 13 .i. cliab noiden (in rasura).

“Certainly not C: 105b 1 7 capus sebocc, 106a 3 ꝉ soror, 106b 10 uersio (the writing of these glosses reminds one of that of the writer of the Latin text).

“Likewise not C: 143b 2, 146b 14 ꝉ foalgim.”

The codex also contains a number of marginalia[61] in Irish and Latin, in various hands; some of them are in the Ogham character.[62]

p. 5a [marg. sup.] bene est hic.
p. 42 [marg. sup.] faue brigita.
p. 50 [marg. inf., Ogham] feria cai hodie.
p. 52 [marg. sup.] daman ṡianach.
p. 70 [marg. sup., Ogham] fel martain[63].
p. 77 [marg. sup.] omnium.
p. 92 [marg. sup.] sancta brigita intercedat pro me.
p. 114 [marg. sup.] bendacht for anmmain ferguso. amen. mar uar dom[64].
p. 118 [marg. sup.] traces of a gloss cut away.
p. 150 [marg. sup.] v……e[65] faue.
p. 156 [marg. sup.] traces of a gloss cut away.
p. 157^ [marg. sup.] hvcvsque caluus patricii[66] depinxit.
p. 157 [marg. sup.] xp̅e faue.
p. 158 [marg. sup.] s(é)n (anói)bing(e)n[67].
p. 159 [marg. sup.] ruadri[68] adest.
pp. 163, 165 traces of letters cut away on the upper margin.

[ xxi ]

p. 165a [marg. inf.] is dorchæ dom[69].
pp. 168, 169 traces of letters cut away on the upper margin,
p. 170 [marg. sup., Ogham] minchasc[70].
p. 171 [marg. sup.] faue brigita.
p. 173a [marg. sup.] faue brigita.
p. 175a [marg. sup.] patricie adiuua.
p. 176a [marg. sup.] sancta brigita.
p. 176b [marg. sup.] uit mo chrob[71].
p. 177a [marg. sup.] sancta trinitas.
p. 178b [marg. sup.] patricie benedic.
p. 181 [marg. sup.] faue patricie.
p. 182 [marg. sup.] faue brigita.
p. 182b [marg. sup.] finguine[72].
p. 184b [marg. sup.] sancta brigita oret pro nobis.
p. 189a {marg. sup.] lathæirt[73].
p. 190a [marg. sup.] patricie faue.
p. 190 [marg. sup.] follega[74].
p. 191a [marg. sup] faue brigita,
p. 192 [marg. sup.] sancta brigita adiuua scriptorem istius artis.
p. 193 [marg. sup., Ogham] cocart[75].
p. 194 [marg. sup.] donngus[76].
p. 194a [marg. inf.] do inis maddoc dv́n .i. meisse ⁊ coirbbre[77].
p. 194b [marg. sup., Ogham] cocart[75].
p. 195 [marg. sup.] sancta brigita.
p. 195 [marg. sup., Ogham] cocart[75].
p. 195a [marg. sup.] is gann in memr˘ ⁊ ascribend[78].
p. 195b [marg. inf.] ní ǽrmall roscríbad inletraimso[79].
p. 196 [marg. sup.] sancta brigita.
p. 196a [marg. sup., Ogham] acocart inso[80].
p. 197a [marg. sup.] sancta brigita.
p. 197a traces of letters cut away on the upper margin.
p. 199a [marg. inf.] ní mmall[81].
p. 203a [marg. sup.] sancta brigita.
p. 202a [marg. inf.] :::::thas patric ⁊ brig˘ ar máel bri(g)tæ namba olcc amenma frimm (arin)scribund roscribad indulso[82].

[ xxii ]

p. 203 [marg. inf.] maraith sercc céin mardda aithne a máellecán[83].
p. 204b [marg. sup., Ogham] latheirt.
p. 206 [marg. sup.] brigita adiuua.
p. 207 [marg. sup.] Dongus.
p. 208 [marg. sup.] auctor adiuua lucis aeternae.
p. 209 [marg. sup.] faue ihu.
p. 209 [marg. inf.] sudet qui legat difficilis est ista pagina.
p. 210 [marg. inf.] tiach didiu mad ferr lat[84] .i. d. o. o.
p. 211a [marg. inf.] uch mochliab anóibingen[85].
p. 212b [marg. inf.] tertia hora.
p. 213a [marg. sup.] grácad[86].
p. 214 [marg. sup.] ac? an de drochdub faigde dim (? or lim?) ⁽īdiu[87].
p. 215 [marg. sup.] sanctus patricius.
p. 217 [marg. inf.] memmbrum naue droch dub ó ní epur na haill[88].
p. 218 [marg. svip.] in nomine Almi Patricii.
p. 219a [marg. sup.] inmaith[89].
p. 219b [marg. sup.] cobthach.
p. 220 [marg. sup.] gracad.
p. 221 [marg. sup.] in nomine sancti diormitii.
p. 222 [marg. sup.] sanctus diormitius oret pro nobis.
p. 223a [marg. sup.] feria diormitii.
p. 223b [marg. sup.] grácad.
p. 226 [marg. sup.] mochoe noiṅdrommo[90].
p. 228 [marg. sup.] is gann membrumm[91].
p. 228 [marg. inf ] tempus est prandii.
p. 231a [marg. sup.] medon lai[92].
p. 231b [marg. sup.] faué xpe.
p. 233 [marg. sup.] satharnn samchasc[93].
p. 233b [marg. sup.] amen.
p. 235a traces of letters cut away on the upper margin.
p. 236 [marg. sup.] saulus qui fuerat ad(emp)to nomine pauhis.
p. 239 traces of letters cut away on the upper margin.
p. 240a [marg. sup.] iob.
p. 241 [marg. sup.] adiuua x̅p̅e̅.
p. 242a [marg. sup.] aarón iulius.
p. 242b [marg. sup.] sancta maria.
p. 243 [marg. sup.] brigita.
p. 246a [marg. inf.] nox adest.
p. 247 [marg. sup.] faue brigita.
p. 247 [marg, sup.] grácad.
p. 248b [marg. sup.] is tana andub[94].
p. 249 traces of letters cut away on the upper margin.

[ xxiii ]p. 249b [marg. inf.] spiritui sancto semper dignissima gloria. For sigla scattered through the manuscript see Nigra, Rel. Celt. 27. Two Irish quatrains and one poem written on the margin are printed below, p. 290. For the Latin poems in the codex, one of which is in praise of Bishop Gunthar of Cologne, see Nigra, Rel. Celt. 6 sq., Traube, Roma Nobilis 51 (347), Poet. Carol. iii. 238 sq.

As we have seen, the codex was probably written about the middle of the ninth century. The date of the Irish glosses has been much disputed; sometimes they have been considered earlier, sometimes later than Ml., and opinions have varied according as attention has been directed to one point or another[95]. The explanation of the fluctuation of opinion is that the collection of glosses is not homogeneous, but comes from various sources and is of a varying antiquity[96].

With regard to the relation of the Irish glosses to the Latin text it is important to note a large number of instances in which the Irish clearly explains the corrupt Latin of the manuscript. Such are libralibus 1a 1, auctori 7b 11, pudicitia Penelopae 29a 8, ciclasias 32b 12, capsa 36a 8, curta 57a 6, aut amatoriae 63b 7, teretes 66a 22, excipiuntur 67a 12, abriza 73a 4, causdico 138a 12, uisionem 149b 5, nomina 156b 6, opheogenistum 181a 4, potest 189b 3, retransit quae 199b 1, passeris 203a 20, pasiua—liquefiunt 209b 19–21. At 155a 1 it would seem as though the glossator had knowledge of a reading αἰτοπάθειαν. At 191a 3 he was apparently acquainted with the true reading.

Instances of misinterpretation of the Latin will be found at 15b 11, 17b 13, 20a 4, 24a 9, 13, 36b 4, 38a 6, 49b 8, 57a 7, 8, 9, 59b 14, 60a 4, 62b 8, 64a 18, 67a 5, 92a 1, 95a 6, 139a 1, 144a 3, 146b 7, 154b 1, 185b 7, 188b 1, 217b 3.

The authority most frequently cited is Isidore, 13b 2, 47b 7, 49b 16, 20, 52a 11, 53a 12, 20, 95a 1, 96a 3, 96b 2, 106b 12, 111b 5, 152a 2, 159a 7. Others are Cicero[97] 7b 15, 73a 4, 92b 1, 102a 2a, 106b 14, 107a 3, Beda 35a 12, 49b 8, 124b 6, Orosius 23b 4, 57a 8, 95a 7 (?), Virgilius 106b 13, 143b 7, 152b 1, Ambrosius 96b 7, Boeotius 57a 7, Cassianus 41a 1, 131b 1, C˘[98] 8b 5, 190b 3, Com˘ 100b 2, Dionysius Thrax 18a 4, Gaudentius 70a 15, Hieronymus 62b 2, Hono˘ 7b 14, Lactantius 22a 2, Maximianus 136a 2, in libro Niciae 65b 16, Papirinus 4a 9, Polibius Medicus 49b 22, Probus 155b 2, …pho˘ 47b 6. Two Irish ‘erratici’ are mentioned in abbreviation Mael˘ and Cua˘ 31b 12, and probably a Mail Gaimrid 183b3[99]. A manuscript called the Liber Romanus is referred to 4a 12[100].

[ xxiv ]

B. Codex Augiensis cxxxii, nunc Carolsruhanus[101].

This manuscript formerly belonged to the monastery of Reichenau: it is now in the Hof- und Landesbibliothek in Carlsruhe. It consists of 107 leaves written in an Irish hand of the ninth century[102]. On the margin and between the lines Latin and Irish notes have been added by different hands.

C. Codex Leidensis Lat. 67[103].

This manuscript is preserved in the University Library of Leyden. It consists of 219 leaves, and was written about the middle of the ninth century[104]. Fo. 9a sqq. contain the text of Priscian’s Latin grammar, with some lacunae, written by more than one hand. On the margin and between the lines are a considerable number of Latin glosses and notes and a few Irish glosses in different hands, printed infra p. 231.

D. Fragmentum Ambrosianum[105].

Cod. A 138 of the Ambrosian Library, Milan, contains Haymonis Comm. in Epistolas S. Pauli ad Hebraeos, ad Corinth. 1 et 2. This is preceded by three leaves, of which the first two contain a fragment of the vulgate text of the prophet Ezechiel, the last a fragment of Priscian, Bk. iv auribus excipitur man (Hertz 1. 139, 9)—Bk. v consonantis quidem antecedente (Hertz i. 150, 14). Haymonis Comm. has come undoubtedly from Bobbio. Consequently it may be inferred that the Priscian came from the same place. On the verso of the second leaf an Irish hand has added the contents of Haymon's Commentary, which shews that the two leaves were attached to the commentary from a very ancient date[106]. The nine Irish glosses contained in this fragment are printed infra p. 232.

11. Codex Ambrosianus, F. 60 sup.[107]

The manuscript bears the title “Sententiae sanctorum Doctorum et Patrum.” The five Irish glosses printed infra p. 234 are on fo. 60.

[ xxv ]

12. Codex Bernensis 363[108].

This codex is preserved in the Stadtbibliothek of Berne, and contains Servii Mauri grammatici Commentarius in Bucolica Georgica et Aeneidem Virgilii, fo. 1–142, Horace, fo. 167a–186d (odae, epodi, carmen saeculare, ars poetica, et sermonum lib. 1 usque ad sat. iii., v. 134), part of Ovid’s Metamorphoses, Bedae Historia Britanniae, and a variety of other works. According to Traube the codex (which is wholly in an Irish hand) is not earlier than the end of the ninth century[109]. It is a copy of one or more older Irish manuscripts, and it is not impossible that all the marginalia have been transcribed from the original[110]. These marginalia carry us into the circle of Sedulius[111] and the middle of the ninth century. The original belonged to North Italy, probably to Milan[112].

13. Codex Bibl. Reg. Monacensis, Cod. Lat. 14846[113].

This is a manuscript in the Hof- und Staatsbibliothek of Munich, assigned to the tenth[114] or eleventh century[115]. It has on the back the title: In Donatum de Grammatica, Saec. ix., and consists for the most part of Erchanherti commentarius in Donatum minorem. Ff. 106–121, however, contain a collection of Latin sortes; on fo. 106r, which is otherwise blank, another hand has written: Sortilegia per literas et sacros libros quorum meminit diuus gregorius turonensis. These sortes are of various kinds. In those printed below (pp. 236, 237) the prefixed letters have reference to the consultation of the Psalter; unless it was otherwise prescribed, the initial letter of the word which first met the eye would seem to have been decisive. The operation is denoted by the phrase librum tenere.

The Latin text is corrupt, and it has had incorporated with it both Irish and British glosses, much distorted in the process of transcription. [ xxvi ]The Celtic glosses are written in the text, but are generally indicated by perpendicular or horizontal strokes. For the most part the Celtic words stand out of construction in the sentence; sometimes they are obviously misplaced. They seem to have been originally notes and glosses on the text, which later copyists incorporated therein[116].

14. Codex Canonum Hibernicorum Camaracensis[117].

This is a manuscript of the eighth century[118], preserved in the public Library of Cambray, nr. 619. It consists of 72 leaves, and contains the text of the Irish Canons down to Lib. xxxviii. 18 med. It has been copied by a continental hand from a manuscript in the Irish character[119]. In the archetype there had been inserted by chance a leaf containing a fragment of an Irish homily. This was copied by the continental scribe along with the rest of the codex; the words are often wrongly divided, and there are many clerical errors resulting from the scribe’s unfamiliarity with the Irish script.

The Irish is very archaic, and dates from the second half of the seventh or the beginning of the eighth century[120]. The following points may be noted. Internal ē has not yet become īa, e.g. fēda = later fíada; at the end of a word, however, it appears as ea, dea = later día. Similarly ō has not yet become ūa, e.g. ood = later úad. Unaccented ĕ is preserved in autrubert, le, nundem, and in the possessive pronoun, ine[121] chuis, ine lāim, ine mēraib, inae lobri, inae seth, inae dommetu, faire chomnessam; but a appears where no preposition precedes: a bees, a deserce, a fuil, and after fri: fria thola, fria tola; similarly are n‑indarbe, are n‑airema, aire sechethar, aure coicsa, but ara tinōla. Unaccented ŏ is preserved in fēdot, tuthēgot, tuesmot. Pretonic to- is preserved before verbal forms: tu-thēgot, tuesmot, tondecomnacuir, but before nouns we find du, do. In the article pretonic nd is preserved in dundaib, but has become nn in inna. A long vowel is often expressed by doubling, e.g. isee, bees, duun. Amail, intain appear, not amal, intan.

[ xxvii ]

15. Incantationes Sangallenses[122].

Codex Sangallensis nr. 1395, which is made up of a collection of fragments of ancient manuscripts made by von Arx when librarian, contains (pp. 418, 19) a single leaf of an ancient Irish manuscript of the eighth or ninth century. The verso of the leaf contains the Irish spells written by three hands. The first hand wrote the first three spells down to fortchiunn[123]. The second hand is much coarser; from it comes the spell Tessurc–forsate, A third hand added focertar–aleth, words which indicate the application of some form of incantation which is not described.

16. The Stowe Missals[124].

The Stowe Missal is a small manuscript of 67 leaves, 5⅝ inches long by 4½ broad, now in the library of the Royal Irish Academy, Dublin. The principal divisions of the volume are: (1) Extracts from St John’s Gospel (ff. 1–11); (2) the Missal (ff. 12–65); (3) the Irish treatise on the Eucharist (ff. 65 b–67 a); (4) the Irish Spells (fo. 67 b). Section (1) forms a separate quire by itself, so that there is no evidence when it was attached to the rest of the book.

[125]In the liturgical portion the discrimination of the original hands is far from easy. In the first part of it, the Ordinary and Canon of the Mass (ff. 12–38), two hands seem to be found, one (A¹) extending from fo. 12 to fo. 28, the other (A²) beginning at fo. 29. Both are bad and decadent hands, and the difficulty is to know whether this is due, as often, to lateness of date or to the inexperience of the scribes. A² probably continues to fo. 46, where the Missal proper ends. The Ordo Baptismi (ff. 47–65a) begins in a hand (B) akin to A¹, but probably not the same. Apparently several hands occur in it, but it is impossible to determine exactly the [ xxviii ]points of change, or how far the differences are due to progressive deteri- oration on the part of a single scribe. The latter part of the treatise, however, appears to be certainly by a different hand from the beginning of it. There is also to be distinguished the hand of a corrector, Moelcaich[126], whose signature appears on fo. 37, where his activity ends. As to the rubrics, up to fo. 23 inclusive only the words lethdirech sund on fo. 18 appear to be in the hand of Moelcaich. From fo. 24 Canon dominicus papae Gilasi onwards they appear to be all in his hand, except the Irish notes inserted in a small hand on fo. 34. After Moelcaich disappears there are at least two hands apparent in the titles, one on ff. 38 and 47, the other on ff. 42 and 44b. The title and prayer on fo. 46b, before the Ordo Baptismi, are in a hand resembling that of Moelcaich, and may be his. The Irish treatise on the Eucharist and the Spells are written in different rough hands.

With regard to the date of the script, Dr Kenyon would assign that of Moelcaich to the tenth century. If that be so, he would assign the original hands (A¹, A², B) to the beginning of the tenth century or possibly the end of the ninth, but not earlier. The Irish treatise and spells are written in rough hands which are difficult to date. According to Dr Kenyon they can hardly be earlier than the eleventh century, and they might well be later[127].

If the codex is to be put so late, there is evidence from the language that the texts have been transcribed from a much older original, Noteworthy is cache Mass § 18 by cacha Wb. 13b 28, Sg. 26b 9, 198a 14, cecha Ml. 56b 22, 96b 7, 134a 3. Further in the tract on the Mass the preposition to before verbs remains to‑: toresset, tanaurnat, tocing, totét, cf. tofasci in the Spells[128], while before nouns it has become do. to ‘thine’ appears in the Mass § 19, and in the Spells; what weight is to be laid on these isolated cases is not clear. The preposition di before a noun has not yet become do: diobli, deobli, Mass § 16. In § 19 amail still appears by amal. On the other hand there are instances of later phenomena, which may be put down to the chances of transcription; such as dana = dánae, Mass § 16, by anmæ, oblæ, menme, menmæ, nd for nn in brond § 3, colind § 11, the expression of aspirated f and s by a dot over the letter. As peculiarities of orthography may be noted: forsen Mass § 5, insen § 10, hoṡen[129] § 18; cælech § 4, rosaegeth § 19, cf. saele Spells[130]; fuel Spells[131]; coer[132] Mass § 19 ; tuib[133] Mass § 15.

[ xxix ]

17. Cooperculum Codicis Bedae Caroliruhensis[134].

(Cod. Aug. clxvii.)

These fragments were discovered by Dr Holder on the verso of one of the leaves of vellum in which the Reichenau manuscript of Beda was formerly bound. The writing is of the eighth or ninth century.

18. Liber Dimmai.

The Book of Dimma is an ancient copy of the Gospels, now in the Library of Trinity College, Dublin. It also contains an office for the visitation of the sick, O’Curry, Lectures, p. 651, where ‘nunc’ should be .N. The four Irish notes printed infra p. 257 come at the end of the Gospels of S. Matthew, S. Mark, S. Luke and S. John respectively; the Irish quatrain is at the end of the codex. The only form linguistically noteworthy is Dimma; the change of final ‑ae to ‑a seems to have begun about 800 a.d.

19. Liber Dairmagensis.

The Book of Durrow is likewise a copy of the Gospels, now in the Library of Trinity College, Dublin, and sometimes assigned to the sixth century. The Irish note printed infra p. 257 is on fo. 173r, and is in quite a different hand from the text.

20. Liber Deirensis.

The Book of Deir is a small octavo codex of 86 folios now in the University Library of Cambridge, numbered i. i. b. 32. Its principal contents are [ xxx ]the Gospel of S. John (Hieronymian version), portions of the other Gospels and a fragment of an office for the visitation of the sick. These and the colophon printed infra p. 257 are in one handwriting certainly as old as the ninth century. In fo. 28^ occurs the rubric Hisund duh&rx sacorfaicc ddu, 'here the Host is given to him.' The only remarkable form is rodscribai; such preterites do not yet appear in the Old-Irish glosses, but they are found in the Felire of Oengus.

The legend of the foundation of the Abbey of Deir in Buchan, and the grants and charter (interesting as the sole existing specimens of the Gaelic spoken in Scotland in the twelfth century), have been published and trans- lated by W. S. (Goidelica, pp. 106—111) and by Stuart (The Book of Deir, edited for the Spalding Club, Edinburgh, 1869).

21. Vita Fintani^

The Irish sentences are found in three MSS.^

A = Codex C. 23 in the Stiftsbibliothek library of St Gall, assigned to the eleventh century.

B = Codex Augiensis Ixxxiv. (ff. 20-24) in the Hof- und Landesbibliothek of Carlsruhe, assigned to the eleventh century.

C = A codex in the library of the monastery of Engelberg, assigned to the twelfth century.

Zimmer has shewn from the erroneous translation licet tibi a deo post alios remeare which in A follows isket duit odia, anatheset indabdane, and in B takes the place of the Irish, that A and B go back to a common archetype. He further holds that C is copied from A. This is less clear. In a few cases C has the correct Latin text where it is corrupt in A, and it is not easy to see why the scribe should have changed doit to doitiis^.

[135] [136] [137] [ xxxi ]It has been calculated that S. Fintan died in 878 A.D. His life then may have been written towards the end of the ninth century.

22. Adamnani Vita Columbae

The manuscript (= Codex A, Reeves'), from which the Irish names in Adamnan's Life of Coluraba are printed below, belonged formerly to Reichenau and is now in the Stadtbibliothek of Schafibausen, where it is numbered 32. As Reeves has shewn^ it was written by Dorbbdne, who was elected to the chair of S. Columba in lona in 713 A.D. and died in the same year, nine years after Adamnan himself In the time between the composition of the Life and its transcription by Dorbbene the Irish language had under- gone certain changes, and occasionally, as a comparison with other MSS. shews*, Dorbbene introduced the forms of his own time. The language is of the same general character as that of the oldest portions of the Book of Armagh, e' and o' are still preserved, e.g. Fechnus, Neth, Modam, Gloithe, Tomme. Unaccented short vowels preserve their quality, e.g. Ached, Lathreg, Nemaidon: at is still universal; Aido etc. oi appears in Broichdnus by oe in Mess Loen^. The variation between Columm and Golumh{<o^ 1) is remarkable. The gen. of -i- and -u- stems is in o : Aido etc.

23. Antiphonarium Benchorense.

This liturgical manuscript commonly, but inaccurately, called an Anti- phonary, was written in the monastery of Bangor (Ir. Bennchor), on the southern shore of Belfast Lough, during the abbacy of Cronan, i.e. between the years 680 and 691. It contains six canticles; twelve metrical hymns; [138] [139] [140] [141] [142] [143] [144] [145] [ xxxii ]sixty-nine collects for use at the canonical hours ; seventeen collects on behalf of special persons or for use on special occasions ; seventy anthems and versicles; the Creed; and the Pater noster. The evidence as to the birth- place and date of the MS. is contained in three hymns : a hymn (f. 15^) to the first abbot S. Comgell, entitled Ymnum sancti Comgilli, a hymn (f. 30) entitled Versiculi familiae Benchuir, — and containing the line Munther^ Benchuir beata, and a hymn on f. 36^, which we have printed infra (p. 282) as preserving the names of the first fifteen abbots and as proving that the MS. was written during the lifetime of Cronan. There is also in f. S¥ an Irish rubric — Common oroit diin^ — over a prayer beginning ' Oustodi nos Domine ut pupillam occuli.' The MS. formerly belonged to the Irish monastery at Bobbio in the Apennines, and is now in the Ambrosian library at Milan, where its press-mark is C. 5. inf.^

24. Old-Irish Inscriptions.

The inscriptions in this collection, most of which were copied by the distinguished antiquary Dr Geo. Petrie, and redrawn by Miss Margaret Stokes, are of very different dates. The most interesting linguistically are lie Luguaedon macci Menueh (p. 288 1. 35) and in loc so tanimmairni Ternohc mac Geran hie er cul Peter (p. 289 1. 18)*. The eclipsis of the c of Ciarain in the comparatively late Orait ar Gilla Giarain may also be noted.

25. Codex Sancti Pauli".

This codex is preserved in the monastery of St Paul in Carinthia, where it is numbered sec. xxv. d. 86. It consists of four leaves, before which has been fastened a smaller leaf, written on one side, of a manuscript of the [146] [147] [148] [149] [150] [ xxxiii ]Bible. The couteuts are of a miscellaneous character ; on ff. 1** and 4!^ stand the Irish poems printed below, all written in the same hand. The codex has been assigned to the eighth century by Windisch, to the ninth by Zimmer The latter date is in accordance with the indications of the language*. But some at least of the poems are of a considerably earlier origin.

The contents of the poems are as follows :

I. Some sort of charm or incantation, in part unintelligible,

II. A poem treating of the doings of the bookish writer and his favourite cat Pangur bdn, edited by Windisch, Jr. Texte, i. 316 ; and with a French translation, in Rev. Celt. v. 128. The following peculiarities of the language may be noted, some of which would seem to point to the ninth century. Such are :

(a) Final -a = -ae: menma.
(b) Hth monosyllabic = 6ftrf (the regxilar form in the Old- Irish glosses).
(c) ndr by ndthar.
(d) -sem=-som. This is frequent in Sg.^, and is established for about 800 a. d. by the rhyme in Fel. Oeng., Ep. 524. The poem has also -sarn for -som ; once it is established by the rhyme with gal. Such a rhyme does not yet appear in Fel. Oeng.
(e) The aspiration of the object: cheist*.
(f) Neut. iiach for na: nach r4^.
{g) dufuit=odQT dotuit.

In the case of such peculiarities as are not established by the metre, it is uncertain whether they are to be imputed to the writer or to the scribe. For fein, fessin etc. the poem has cein, and cesin by fesin. Such forms likewise appear in the St Gall glosses, in the Cambray Homily, and in the Imram Brain. Whether they are archaic or dialectical is not clear.

III. A riddling poem ascribed to Suibne Geilt, a king who is said to have lost his reason in the battle of Moira. The form durigni is found in Ml. and Sg. but not in Wb., which, however, has sg. i. dorignius. The word for ' star ' is still Htglu not rMa, as it became later.

IV. Verses extracted from a poem ascribed to St Moling f697. The poem in its present form is of a much later date, but there is nothing in the rhymes to prevent its ascription to an early period. If J. S. be right in his conjecture that nem is for ndin = niam, the poem cannot be later than about 700 ; if so, the verses may actually have been written by that saint. The whole poem is found in five MSS., viz. the Book of Leinster p. 284, the Book of Ballymote, p. 256', the Book of Lismore, f. 45*, and the Bodleian codices [151] [152] [153] [154] [155] [ xxxiv ]Rawl. B. 512, f. 141^ and Laud 610, and is printed in Goidelica, p. 180. In the Ballymote copy the poem is said to be taken from the Book of Glendalough, now lost.

V. A poem in praise of some Leinster princeling called Aed^. If this Aed could be identified, the approximate date of the poem would be established, for it is evidently the work of some contemporary bard who sought to please his patron. In its present form the poem shews the language of the original, but none of the ninth century peculiarities are established by the rhyme ^ so that the poem may be safely ascribed to an earlier date^.

26. Codex Epjstularum Pauli Boernerianus*.

This Codex is now in the Royal Library in Dresden, and consists of 111 leaves. Fo. 2'*-99^ contain the Greek text of the thirteen epistles of St Paul with a Latin interlinear version ; on fo. 1 begins an interpretation of the Gospel according to St Matthew, which is continued on fo. 109-111"; on the upper portion of fo. Ill" there is a fragment of Marcus Monachus de lege spiritali. Traube regards it as almost certain that the codex was written by Sedulius. The marginalia are such as appear in other manuscripts belonging to his circle: dongus fo. 5% 16% BS"" (do. 18*); dubthach.^ fo. 8"; fergus 82" 94* ; comgan fo. 68** ; ayavov' (aya., ayavo, ag.) fo. 22% 26% 28% 36% 43^ 54% 58% 59% 64% 65% 65% 70% 74% 81% 90% 93% 96% 96^ 98^ ; lii er du er musci monachi 36"^; Angelherti fo. 52% yoUaKaXKo^" fo. 22^ 87% 88% 90% 93^; yvaw, yiao) fo. 34"; yovddp^ fo. 71"; Aar^garius" / hildmnxxs^" fo. 69*; >ap. fo. 30" 32% Map." fo. 39% 43^ 44% 45% 48% 48", 64% 66% 77% 89*; scotti 95".

[156] [157] [158] [159] [160] [161] [162] [163] [164] [165] [166] [ xxxv ]27. Liber Hymnorum^

The Liber Hymnorum is contained in the following MSS.^

T. = E. 42, Trinity College, Dublin^ This manuscript consists of 34 vellum folios, about 10^ inches long by 7 broad, with three scraps of vellum bound at the end. After fol. 31 the writing deteriorates, and this later portion seems to be younger than the rest of the codex. In a number of cases the marginal glosses have been mutilated by the cutting of the margin.

F*. A manuscript formerly in the library of S. Isidore's, Rome, from which in 1872 it was brought to the Franciscan Convent, Merchants' Quay, Dublin, where it now is. It consists of twenty-three leaves in small folio, and is in a pasteboard cover, endorsed ' 9 vel 10 saecul.' In this codex the Faeth Fiada and Mael Isu's hymn are wanting. On the lower margin of fo. 2 a is written in a hand of the seventeenth century ' Ex libris conventus de Dunnagall,' and Sir James Ware, in the year 1639, quotes it as 'Lib. net. hymn, conuent. Dunnagalliae.' It appears from a remark of Ussher's that the manuscript was once in his hands, and there is a paper MS. in Ussher's collec- tion in the Library of Trinity College, Dublin (E. 3. 28), containing a copy of some of the Irish Hymns from F. In 1630 the MS. was still in Donegal, for it was used by Michael O'Clery in his Martyrology of Donegal, written in that year. Afterwards the MS. went to Louvain, where it was used by Colgan for his Trias Thaumaturga, 1647.

T and F represent two independent recensions of the text, both of which are indispensable for its reconstruction. Such a reconstruction is full of difficulty. For the restoration of the language of the Hymns depends upon

[167] [168] [169] [170] [ xxxvi ]their date, and the chief, in some cases the only, evidence for the date of composition is furnished by the language. In our reconstruction we have been guided by the evidence of date supplied by the language, and particularly by the evidence of the rhymes, which often give valuable indications how far certain sounds had fallen together at the time of the composition of the poems. We are well aware of the uncertainty of the process, and that in this matter opinions will be sure to differ. But even an imperfect reconstruction seemed preferable to a mere recording of the readings of the manuscripts.

Both T and F contain Irish prefaces which, except in Colmdn's Hymn, are practically identical, and a number of glosses and scholia. In part these are common to both manuscripts, so that they must have been copied from a common source ; in part they are peculiar to one manuscript or the other. In particular the margins of F are filled with long Irish notes, which are unfortunately in part very hard to decipher and in part altogether illegible. For the sake of completeness these notes have been printed below, though they are of little real value for the interpretation of the text. The notes on the language are for the most part either superfluous or erroneous, and the various stories narrated to illustrate the text, particularly in Broccan's Hymn, may represent forms of the legends long subsequent to the composition of the poems.

With regard to the date of the manuscripts, that of T has been disputed. We have seen that T and F contain common scholia, which must have been derived from a common older source. Now the language of these common scholia shews that they cannot be earlier, in part at least, than the eleventh century. Hence T cannot be dated earlier than the end of the eleventh century or the beginning of the twelfth. As to F, for the same reasons it can hardly be put earlier than the end of the eleventh century or the beginning of the twelfth, and it may be somewhat later than T.

I. Golmdn's Hymn^.

This hymn is traditionally assigned to Colman, lector of Cork, and it is said to have been composed by way of defence against the plague which devastated Ireland in the middle of the seventh century I The indications in the hymn itself point to the early part of the ninth century. A superior limit may be found in the mention of Joseph, the husband of Mary, who seems not to be commemorated in Irish liturgies before about 800 A.D. An inferior limit may be found in the treatment of final vowels. As the rhyme shews, there is yet e.g. no confusion between final -ae and final -a, a confusion well established for the middle of the ninth century by the St Gall glosses [171] [172] [ xxxvii ]on Priscian. The language of the hymn strongly resembles that of the Fdlire 6engusso, which belongs to about 800 A.D. The most striking de- parture from Old-Irish usage is the use of anacht (1. 22) for the relative anachte. As the repetition of the initial words after 1. 45 shews, the hymn falls into two parts. The conclusion, certainly a later addition, contains an invocation of Irish saints, one of whom, Adamndn, died in 704. According to the glossator of F, D^rmait ua Tigernan, bishop of Armagh in 848, added 11. 47 — 54, and Mugron, abbot of lona from 964 to 980, added 11. 51, 52.

As to the substance of the poem, M. Henri Gaidoz (Revue celtique V. 94 — 103) points out that it falls into three divisions : 1. the original work (11. 1—37), 2. a first addition (11. 38—48), and 3. a second addition (11. 49— 56); and shews that the prototype of the first (11. 1 — 37) is in the Roman breviary, where it is entitled: Commendatio animae quando infirmus est in extremis.

II. Fiacc's Hymn^.

This hymn is traditionally ascribed to Fiacc bishop of Sletty, whose consecration by S. Patrick is recorded in the Book of Armagh (infra p. 241). But it must have been composed after the desolation of Tara (a.d. 561), which is mentioned in vv. 20 and 44. For the real date of its composition the language of the poem furnishes the surest criterion. A superior limit is given by the fact that mm and mh have fallen together {imini : timmi 11. 31, 32) ; as to nc? and nn there is unfortunately no evidence, but the change of mh to mm and vd to nn belong to the same period of the language. Roughly these changes may be assigned to the close of the eighth century ; for the Felire 6engusso they are established by the rhyme. An inferior limit is given by the treatment of final vowels : -a, -ae and -ai are still kept apart ^, also -e and -i This points to a date at the latest not much later than 800. The Middle-Irishisms which have crept in in the course of transmission can be removed without detriment to the metre.

Zimmer and Atkinson find extensive interpolation in the hymn. But, as so shai'psighted an observer as Thurneysen has remarked, the language of the verses supposed to be interpolated differs in no way from that of the vei-ses admitted to be genuine. Nor, with one exception to be spoken of below, is

[173] [174] [175] [176] [ xxxviii ]there anything in the subject-matter to suggest interpolation. The stories are of the same kind as those related in the Patrician documents in the Book of Armagh ; a little variation is found vv, 45 sq., and an exaggeration of the older story in v. 56. But on the whole the narratives agree closely. The exception is v. 34. The documents in the Book of Armagh (fo. 13^2) record only one occasion on which Patrick raised the dead to life, and then it was a giant whom he raised from the grave to shew his unbelieving followers that there had been actually a man of such a stature as his tomb indicated. As Thurneysen has pointed out^, the verse seems to be an imitation of vv. 19 — 24 of the hymn to Christ ascribed to S. Hilariusl It is very improbable that the author of the poem should have written two successive verses with the rhymes bethu : lethu.

III. Ninines Prayer.

This prayer is traditionally ascribed to the poet Ninine (notes to Fel. Oeng. July 6, Dec. 11) or to Fiacc, bishop of Sletty. It is a highly alliterative piece, without rhyme, and with no well-defined metrical system. As to the number of syllables, the first two lines are based on the division 7 + 5. Windisch, omitting primapstal in line 6, would divide the poem thus : 7 + 5. 7 + 5 ; 10. 5. .9 ; 10. 5. 9 ; 10. 10. The absence of rhyme deprives us of the most important criterion for fixing the date of the poem ; the language shews no signs of lateness, except what may be fairly imputed to the tran- scribers.

IV. Ultaris Hymn.

This hymn, traditionally ascribed to Columcille or to Ultan of Ard Breccain (-f* 656), to three of Brigit's community, or to Brenainn, is the only one of the Irish hymns which shews high poetic art. There is nothing in the language to shew that the poem cannot go back to the seventh century a.d. The text has been restored below on the assumption that the poem is of so early a date.

V. Broccdn's Hymn^.

In the hymns previously considered nothing has been discovered that would point to a later date than the early part of the ninth century. In the present hymn, taken as a whole, there are to be observed changes in the [177] [178] [179] [ xxxix ]language that would seem to point to a later origin. The following points may be noted :

(a) the rhymes dara : wimada 11. 95, 96 ; cuire (or cuiri) : A?w7e (or hiiili) 11. 99, 100.
(b) Olinn da Loch for Olinn da Locha 1. 20.
(c) ro-das-gdid for ro-da-gdid 1. 35, ro-daa-doad for ro-da-cload 1. 35 ; ni-s-digaib 1. 36 (linn is not fern.) ; no-d-guidiu 1. 17 (where (f has probably a relative function)^.
(d) the verbal forms ernais (for asi'ir) 1. 7 ; smastar, rodglmnestar, millestar 11. 45, 46 (all in one verse); -aimecht 1. 86 (O.Ir. arrlclu); cech thucai 1. 85 ^,
(e) anUacht 1. 10 may possibly be used in the sense of coniUackt ; on doddecha 1. 81 see the note.

On the other hand this hymn, which is ascribed to Broccan Clden, a disciple of Ultan of Ard Breccain, contains many interesting Old-Irish forms e.g. gdde 1. 49, dith (for did, the perf. sg. 3 of dinim) 1. 76, both 1. 70, conacna 1. 100, senta 1. 38, goita 1. 66. And in a large proportion of the verses there are no deviations from the Old-Irish standard except such as may fairly be put to the account of the scribes.

The connexion of the verses is of the loosest character. Miracle upon miracle is recorded with a brevity which is often obscure, and no connexion is apparent between one miracle and the other. In such a disconnected poem interpolation is easy. To later versifiers it would have been at once a pious and a simple task to add a few more marvels to the greater glory of S. Brigit. And this would be the easiest explanation of the mixture of old and new discernible in the hymn.

In the miracles narrated and in the order of their narration there is a close agreement between the hymn and the Vita Brigidae by Cogitosus^, which forms the best commentary on the hymn. In all probability the narrative of Cogitosus is based upon the hymn.

VI. Sanctdn's Hymn*.

In this hymn may be noted the rhyme finda : thenga 11. 17, 18, the relative d in no-d-guasim 1. 3, and possibly fitir for rofitir 1. 7. The date of composition is probably the ninth century. The hymn is in two parts, of which the first, ascribed to Sanctain, ends at line 20, and the second is addressed to that saint, to the Virgin Mary, and to Christ. Sanctain is said to have been a Briton, brother of the pilgrim Matoc, and grandson of Muredach Muinderg, king of Ulaid, who died A.D. 479.

[180] [181] [182] [183] [ xl ]VII. Patrick's Hymn^.

This hymn, or rather incantation, said to have rendered S. Patrick and his monks invisible as such, is not in metre, but in a sort of rhythmical prose. It bears upon it marks of antiquity, such as the prayer to be delivered from the spells of women, smiths^ and druids or wizards. The date of its composition cannot be determined. An inferior limit is fixed by the mention of the work in Lib. Ardm. fo. 16" 1, canticum eius (sc. Patricii) scotticum semper canere; and the Milan glossator may possibly refer to it when he writes cluasa Dm diar n-eitsecht (Ml. 24* 18). The title, fdeth Jiada, is a mis-spelling of foid^ (Cymr. gwaedd) fiada, and this is still further corrupted in the feth jia of the Book of Ballymote, 345" 26, where wizards are said to make feth jia (' magical invisibility') or prophecy (druid .i. doniat in feth fia no in aisdinecht). The verbal forms of the hymn are interesting : atomriug from ad-dom-riug ' me extoUo, assurgo,' as Ascoli (Gloss, pal. hib. cxcv.) for the first time rightly rendered this word : mi-duthrastar the deponential 8-conj. of mid'dthraccur : arachuiliu, where the final u has not been explained. So in the declension : niurt the instrumental sg. of the neuter o-stem nert : cretim the same case of the fem. a-stem cretem; and foisitin the same case of a stem in n. The hymn has been edited by Geo. Petrie (Antiquities of Tara Hill), by W. S. (Goidelica, p. 150), by Crowe (Journal of the Kilkenny Archaeological Association), and, lastly, by Bernard and Atkinson (the Irish Liber Hymnorum i. 133 — 135).

VIII. Mael Isu's Hymn.

This hymn is found only in the later portion of T. The author may have been Mael Isu, the coarb of S. Patrick, who, according to the Annals of Tigernach, died in 1086, and whose day is Jan. 16. The metre is rinnard.

28. Codex Taurinensis, F. iv. 1^

This manuscript contains six leaves of an Hiberno-latin liturgy. An Old-Irish gloss is found in fo. 3*. According to W. Meyer the codex is more probably prior than posterior to 700 A.D."

[184] [185] [186] [187] [188]

  1. Edd. Windisch, Irische Texte ii. 146–163, W. S., The Old-Irish glosses at Würzburg and Carlsruhe, 143–163. For a full description of the codex see Windisch, op. cit. 143–146. A specimen of the writing will be found in Silvestre-Madden, Palaeography p. 609.
  2. At the bottom of the first page of the codex proper is written “Liber Augie Maioris.”
  3. A still more decisive proof of this is to be found in 12d, if the conjecture be right that after innahí some words like adciat indhí have been omitted per incuriam.
  4. Edd. Zimmer, Glossae Hibernicae, 229–233, W. S., The Old-Irish Glosses at Würzburg and Carlsruhe, 210–237; cf. Zimmer Gloss. Hib. xxiv. sq.
  5. On the margins of 14c–15b are the Annales Augienses breuissimi; cf. Moae, Anzeiger für Kunde der deutschen Vorzeit iv. 14; Pertz, Monumenta Germ. iii. 136 sq.; Zimmer, Glossae Hibernicae, xxv. sq.; F. Kurze, Neues Archiv xxiv. 444.
  6. = the Annales Augienses breuissimi, see preceding note.
  7. Gloss. Hib. xxv. sq.
  8. Ed. Stokes, Goidelica, 51 sq.; Zimmer, Glossae Hibernicae, 253 sq., Sapplementum, p. 13 ; cf. Strachan, The Vienna Fragments of Bede, Rev. Celt. xxiii. 40 sq. The text is here re-edited from photographs of the codex.
  9. The Irish glosses have been edited by W. S., Remarks on the Celtic additions to Curtius’ Greek Etymology, p. 73, and by Zimmer, Glossae Hibernicae, 218.
  10. Cf. Haddan and Stubbs, Councils and ecclesiastical documents relating to Great Britain and Ireland, i. 108.
  11. Cf. Zimmer, Glossae Hibernicae, xx.; Wasserschleben, Die Irische Kanonensammlung² xxiii.
  12. See Seebohm, Tribal Custom in Anglo-Saxon law, pp. 102–103.
  13. The text of the canons has been published by Wasserschleben, Die Bussordnungen der Abendlandischen Kirche, Halle, 1851, pp. 136 sqq., the Irish glosses by Zimmer, Glossae Hibernicae, 284.
  14. Cf. Wasserschleben, Die Irische Canonensammlung, xxx. sq.
  15. Ed. Dziobek, Bezz. Beitr. v. 63 sq. (see Güterbock ibid. vii. 342); Zimmer, Glossae Hibernicae, 259 sq. The glosses are here edited from photographs.
  16. Reifferscheid, Bibliotheca Patrum Latinorum Italica, tom. i. 469.
  17. Edd. d’Arbois de Jubainville, Bibliothèque de l’École des Chartes, vi. série, tom. deuxième, 1866, p. 509, 1867, p. 471; Gaidoz, Proceedings of the Royal Irish Academy, x. 70 sq.; W. S., Goidelica 54; Zimmer, Glossae Hibernicae, 262. The glosses are here edited from a photograph.
  18. Ed. Nigra, Rev. Celt. i. 58 sq.; W. S., Goidelica 51; Zimmer, Glossae Hibernicae 228, Suppl. 12; W. S., KZ. xxxv. 587 sq.; cf. Nigra, Rev. Celt. xxiv.
  19. Ed. Loth, Rev. Celt. v. 470, W. S., The Academy, Sep. 25, 1886, p. 209, KZ. xxxv. 588.
  20. Ed. Loth, Rev. Celt. v. 161, W.S., The Academy, Sep. 25, 1886, p. 209, KZ. xxxv. 588.
  21. The Irish glosses have been edited by Zimmer, KZ. xxxiii. 274, who also gives a description of the contents of the MS.
  22. Graff, Althochd. Sprachschatz I. xli.
  23. The British and Irish glosses have been published by W.S., in Kuhn and Schleicher’s Beiträge iv. 385 sq.; cf. Thurneysen, Rev. Celt. xi. 915 sq.
  24. The whole of the Book of Armagh is about to be published by Dr Gwynn.
  25. Proceedings of the Royal Irish Academy, iii. 316–324.
  26. infra pp. 259–262.
  27. Celt. Zeitschr. i. 347 sq.
  28. There is a variation between a and o in Dubthach and Dubthoch; cf. Echach by Echoch, Ann. Ul. 817. In Latinised form Ulathorum appears by Ulothorum; according to Thurneysen, this is perhaps due to the scribe. Unaccented a appears as o in moccu Echach.
  29. v. infra 262–269, and see C.Z. i. 348, iii. 276.
  30. Brash, p. 190, 198.
  31. Cf. Celt. Zeitschr. iv. 477; the orthography is probably due to the confusion of ae and e in Latin.
  32. From Froech the gen. is Fruich, cf. ruig Philargyrius, tuib Stowe Missal.
  33. By Sele there is also found Saele; so Campum Caeri corresponds to Mag Ceræ, Trip. Life, 110.
  34. The Irish in the Latin notes is printed infra pp. 269–271, the Irish notes infra pp. 238–243.
  35. The preposition oc is written ucc or uc: ucc Ráith Bilich, ucc Domnuch, uc Scí Pátric. So in the Annals of Ulster uc Cuinciu 710, uc Biliu 713, uc Etarlinddu 735.
  36. But ‑æ also expresses ‑e after a palatal consonant: bicæ, Muinæ.
  37. In būachaele, ae expresses the ‑i umlaut of ‑a, cf. saele in the Stowe Spells and infra p. xxviii.
  38. There is a peculiar gen. in ‑eo in inseo (leg. inse) and Bōindeo = inse and Bóinde, cf. inseo Ann. Ul. 737, 740, 836, 870. Apparently final unaccented eo and e had become confused.
  39. infra Appendix II.
  40. infra p. 271.
  41. infra p. 271.
  42. infra p. 45.
  43. infra p. 271.
  44. Vol. i. supra pp. 494–498.
  45. Hence we should write in 171a2 ruminiged, in 176b frisintomaltid, in 182a2 siltid.
  46. Hence in 170b1 adamra should be corrected to adamre. In 177a2, as Mr Edward Gwynn informs us, the last half of the last letter of etalacda has been lost by the cutting of the margin; etalacdæ should be restored.
  47. A few of the glosses were published by Thilo, Rhein. Mus., Neue Folge, xiv. 132, and were reprinted by Zimmer, Gloss. Hib. Supplementum 5, with the addition of two from the same codices contributed by Hagen. The glosses in P were printed by W. S. in the Academy for Jan. 17, 1891, and again in the Rev. Celt. xiv. 226 sq. The glosses from L were published by W. S., KZ. xxxiii. 62 sq. The glosses from P² did not come to our knowledge till after the sheet of the Thesaurus containing the other glosses had been printed off. They are published in Appendix I. from a transcript kindly made for us by the late M. L. Duvau. Both the Latin text and the Irish glosses have now been edited by Hagen, Serv. Gram. vol. iii. Fasc. ii. Appendix Serviana; to this edition we are indebted for some additions and corrections. Cf. also Servius ed. Thilo iii. i. v.
  48. As to coennich, P² shews that the archetype had coinnich.
  49. Through inadvertence the few Irish glosses in this text have been published under the name of Philargyrius, and so they are given from P on p. 48 of the present volume. The readings of P2 and G will be found infra p. 418. The full Latin text has now been published by Hagen, Serv. vol. iii. Fasc. ii., Appendix Serviana.
  50. Gramm. Lat. II. xvi.
  51. Thus in Hertz I. 149 l. 13 after aqua D has the same addition as the three other MSS.: lar (quando signi)ficat κατονκαναιον φων [leg. κατοικίδιον θεόν] laris facit genitiuum • sín imperatorem……(l)artis • quern mactauit cosus . et testis ouidius in epigrammatis (lart)e ferox cesso cosus opima tulit: liuius • iniiiiab urbe tolumnio rege ueient(um).
  52. Some examples of agreement with C are: protulit hoc idem in prima epistolarum = Hertz I. 144, 17; apud Latinos = Hertz I. 145, 5; uel per duas terminationes uel per tres = Hertz I. 145, 6; unum in ro = Hertz I. 146, 3; et amatoriae =Hertz I. 148, 15.
  53. Instances of peculiar readings in D are: democritus etheu = Hertz I. 144, 21; statius followed directly by soluerat = Hertz I. 145, 26; argo mango (argo apparently cancelled and mango added over the line) = mango Hertz I. 146, 4 (C has margo with ꝉ ango on the margin); in al neutra sunt latina omnia ut tribunal = Hertz I. 147, 1; in im quoque inuenitur (neutrum?) (pro)prium cim = Hertz I. 148, 6; uirgilius in uii. solforea = Hertz I. 149, 11. Corresponding to Hertz I. 150, 11 D has, not in the text, but between the columns, ħ plaŭ Collũs (i.e. collus uel collum) i columbari • haut multũ (?) (the rest is lost by the mutilation of the page).
  54. Instances from B will be found in Celt. Zeitschr. iv. 472. Instances from C are á deletionibus fricando enim deletur littera = Sg. 3b4; custos mulorum = Sg. 33b3; pelex = Sg. 38b7; lapis triumphi = Sg. 69a19; celer = Sg. 69a21; hortulanus = Sg. 92b1; ab eo quod est tronitr˘ = Sg. 94a4; sanguineus tumor = Sg. 96b1.
  55. Thus .i. anchora B, .i. anchora vel onus quodlibet quo naues stabiliuntur C = Sg. 22a5; quia dicitur Euripides (Aeripides C) qui in illo die natus est quo Athenienses cum Persis in Eurupo (Aeripo C) bellum commisserunt = Sg. 31a6.
  56. Ed. Ascoli, 1879 (Archivio Glottologico Italiano vi.); cf. Zeuss, Grammatica Celtica² xi. sq.; Nigra, Reliquie Celtiche; Hertz, Grammatici Latini II. xv. sq.; W. S., Notes on the St Gallen Glosses, Celt. Zeitschr. ii. 473 sq.; Strachan, Some Notes on the Irish Glosses of Würzburg and St Gall, Celt. Zeitschr. iii. 55 sq.. On the Language of the St Gall Glosses, Celt. Zeitschr. iv. 470 sq.
  57. According to the pagination of the codex itself, it should contain 249 pages. This pagination is correct down to p. 78. But the following page is numbered not 79 but 88, so that every page after p. 78 is numbered 9 too much; cf. Nigra, op. cit. 4. In the present edition the pagination of the manuscript has been followed.
  58. O Roma Nobilis, pp. 50 (373) sq. Güterbock, KZ. xxxiii. 92 note, has sought to determine the date more accurately from some notes on the margin of the codex. If his data are correct, the manuscript was written either in the year 845 or in the year 856.
  59. Gramm. Celt.² xi. note. Nigra, op. cit. 27 sq.; at the end of his book Nigra gives specimens of different hands.
  60. For the information here given we are indebted to the kindness of Prof. Thurneysen.
  61. With regard to the proper names in this and other manuscripts from the circle of Sedulius see Traube, O Roma Nobilis, 54 (350) sq.
  62. Cf. Nigra, Rel. Celt. 18 sq.
  63. = feria Martini.
  64. ‘A blessing on the soul of Fergus. Amen. I am very cold.’
  65. Die Buchstaben halb weggeschnitten, ausserdem der Rand geglättet. Ganz sicher wohl nur v dann am Ende e und faue, Thurneysen. Nigra conjectures vinniane = St Finnen of Mag bile.
  66. = Máil Patricc from the writer of the Latin text down to p. 157 a.
  67. ‘Bless, Holy Virgin.’ The margin has been cut; restored by Nigra.
  68. King of Wales, 844–878. Cf. Reeves, Adamnan, 390 sq., Nigra, Rel. Celt. 12, Traube, O Roma Nobilis, 66 (352).
  69. ‘It is dark to me.’
  70. According to the probable restoration of Nigra; the characters have been cut away in part. Minchasc means ‘Little Easter,’ Dominica in Albis, ‘Low Sunday.’
  71. ‘Alas! my hand.’
  72. According to Nigra, Rel. Celt. 28, one of the scribes.
  73. See the ogmic latheirt infra at p. 204b. Is it = lathirt ‘crapula’? J. S.
  74. ‘Probabilmente è questo un nome proprio irlandese,’ Nigra. But it rather seems a verb.
  75. 75.0 75.1 75.2 ‘A correction’ or ‘correct.’
  76. According to Nigra, Rel. Celt. 28, probably one of the writers of the Latin text; the name, however, occurs in other manuscripts belonging to the circle of Sedulius, Traube, O Roma Nobilis, 54 (350).
  77. ‘We are from Inis Maddoc, to wit, Coirbbre and I.’
  78. ‘The parchment is rough (? ‘difficilis’ Nigra) and the writing.’
  79. ‘This page has not been written very slowly.’
  80. ‘This is a correction’ (acocart from ad-cocart? W. S.). Nigra would read cocart inso.
  81. ‘Not slow.’
  82. ‘……of Patrick and Brigit on Mael Brigte, that he may not be angry with me for the writing that has been written this time.’ W. S.
  83. ‘Love remains as long as property (lit. deposit, ‘opes,’ Nigra) remains, Maellecan.’
  84. ‘I will go then, if you prefer it.’
  85. ‘O my breast, Holy Virgin.’
  86. This obscure word reoccurs in pp. 220, 223, 247.
  87. Die obere Hälfte aller Buchstaben ist weggeschnitten. Das letzte Wort wohl indiu, das dritteletzte etwa faigde, das zweite und dritte ziemlich sicher de drochdub. Das erste Wort aus vier Buchstaben kann ich nicht sicher ausmachen (vielleicht daan oder baan), Thurneysen.
  88. ‘New parchment, bad ink. O I say nothing more.’
  89. ‘Is it good’ or possibly ‘well.’
  90. ‘Mochoe of Oendruim.’
  91. ‘The parchment is scanty (‘difficilis,’ Nigra).’
  92. ‘mid-day.’
  93. = ‘Sabbatum aestiui paschi,’ cf. Güterbock, KZ. xxxiii. 93 n.
  94. ‘The ink is thin.’
  95. Cf. Thurneysen Rev. Celt. vi. 318; Pedersen, KZ. xxxv. 316; Strachan, Trans. Phil. Soc. 1899–1901, pp. 47, 57, Rev. Celt. xx. 304 sq.; Zimmer, KZ. xxxvi. 471; Thurneysen, KZ. xxxvii. 55.
  96. For a detailed proof of this from linguistic evidence see Celt. Zeitschr. iv. 470 sq.
  97. Probably the obscure grammarian cited frequently by Vergilius Maro: possibly (as Prof. Goetz of Jena suggests) the author of the so-called Synonyma Ciceronis, ed. Mahne, Leiden, 1850.
  98. We have not been able to verify the references, so that the name is doubtful.
  99. Cf. vol. i. p. xviii.
  100. Cf. Hertz I. xv.
  101. Ed. Zimmer, Glossae Hibernicae, 219 sq.; W. S., Old-Irish Glosses at Würzburg and Carlsruhe.
  102. According to the Rev. H. M. Bannister it was transcribed in 848 a.d., see Journal of Theological Studies, 1903, p. 51. For a specimen see Silvestre-Madden, Palaeography, p. 609.
  103. Ed. Pott, Intelligenzblatt zur allgemeinen Litteraturzeitung, 1846, pp. 28, 89; W. S., Goidelica, p. 56; Zimmer, Glossae Hibernicae, 226 sq. Through the kindness of Dr de Vries the MS. was deposited for some weeks in the Rylands Library, Manchester; a few additions and corrections will be found at the end of this volume.
  104. From a chronological entry on fo. 7b the MS. may be more precisely assigned to the year 838. Cf. Hertz I. xiii.; Zimmer, Glossae Hibernicae xxi. sq. For Dubthach the writer see Traube, O Roma Nobilis, 56 (352).
  105. Ed. Zimmer, Supplementum, 3.
  106. For the information here given we are indebted to the kindness of Professor Ascoli.
  107. Ed. Zimmer, Supplementum, 4.
  108. The Irish glosses have been edited by W. S., Goidelica 54; Nigra, Rev. Celt. ii. 446; Zimmer, Gloss. Hib. 263 ; Hagen, Codex Bernensis 363, phototypice editus, Lugduni Batavorum, 1897, pp. xli. sq. (where the Irish is often misread); the whole codex may now be studied in the aforementioned facsimile. Cf. Zimmer, Gloss. Hib. xxxi. sq., Supplementum, 14; Gottlieb, Wiener Studien, ix. 151; Hagen, Verhandlungen der 39 Vers. deutscher Phil. u. Schulm., Leipzig, 1888, pp. 247 sq.; Reuter, Hermes, xxiv. 161 sq.; Traube, Roma Nobilis, 52 (348) sq.; Stern, Celt. Zeitschr. iv. 178 sq.
  109. Op. cit. 54 (350).
  110. Traube, op. cit. 53 (349).
  111. Traube, op. cit. 53 (349). The Irish names are printed below, p. 235; for the others see Zimmer, Gloss. Hib. xxxi. sq.; Traube, op. cit. 54 (350) sq.; Hagen, Cod. Bern. 363, xliii. sq.
  112. Traube, op. cit. 53 (349); Stern, Celt. Zeitschr. iv. 178.
  113. Ed. Thurneysen, Sitzungsberichte der Münchener Akademie, philol.-histor. Classe, 1885, pp. 90 sq. Corrections in Rev. Celt. xi. 90 sq. The Latin text has been published by Winnefeld, Sortes Sangallenses, Bonn, 1887.
  114. Catalogus codicum manuscriptorum bibliothecae regiae Monacensis, iv. 2, p. 241 sq.
  115. Keil, De grammaticis quibusdam Latinis infimae aetatis commentatio. Erlangae, 1868, p. 23.
  116. Thurneysen, op. cit. 95.
  117. Ed. Zeuss, Gramm. Celt.² 1004, O’Curry, Bibliothèque de l’École des Chartes, iii. série, tome iii. 197 sq.; Zimmer, Gloss. Hib. 213 sq. A facsimile is given by C. P. Cooper in his Report on the Foedera, appendix A. The text is here published from photographs; the lines correspond to the lines of the original.
  118. The date is fixed by the closing words: Explicit liber canonum quem dominus Albericus episcopus urbis Camaracensium et Atrebacensium fieri rogauit. Deo gratias Amen. Albericus was bishop of Cambray and Arras from 763 till 790 a.d.; cf. Wasserschleben, Die Irische Kanonensammlung² xxx.
  119. Zimmer, Gloss. Hib. xix.
  120. Cf. Thurneysen, Celt. Zeitschr. i. 348 sq., iii. 53 sq.
  121. Noteworthy is n corresponding to nn in Wb.; cf. Pedersen, Aspirationen, 119.
  122. Ed. Keller, Mittheilungen der antiquarischen Gesellschaft in Zürich, Bd vii. tab. vii. p. 75; Zeuss, Gramm. Celt.² 949; Zimmer, Gloss. Hib. 270 sq.; Windisch, Berichte der Königl. Sächs. Gesellschaft der Wissenschaften, 1890, ss. 92, 93. And see Verzeichniss der Handschriften der Stiftsbibliothek von St. Gallen, Halle, 1875, ss. 462–463. The text is here edited from a photograph.
  123. The words Prechnytφcan–κnaatyonibas are written in peculiar half Greek characters.
  124. The Irish portions have been edited by W. S., KZ. xxvi. 497 sq., and by MacCarthy, Transactions of the Royal Irish Academy, xxvii. 135 sq. Cf. also O’Conor, Stowe Catalogue; Todd, Transactions of the Royal Irish Academy, xxvii.; Warren, The Academy, Feb. 8, 1879, Jan. 1, 1881, and Oct. 20, 1894; Liturgy and Ritual of the Celtic Church, 198 sqq.; Plummer, KZ. xxvii. 441 sqq.; Zimmer, KZ. xxviii. 376 sqq. Specimens of the script will be found in the Transactions of the Royal Irish Academy, xxvii. plate vi.
  125. For these palaeographical notes we are indebted to Dr Kenyon, who through the liberality of the Council of the Royal Irish Academy was able to examine the manuscript at the British Museum.
  126. Cf. Warren, Liturgy and Ritual of the Celtic Church, 201, note 1.
  127. Dr Kenyon gives these results with diffidence, and thinks there is far more hope of arriving at an approximate date on liturgical or linguistic grounds than on palaeographical considerations alone.
  128. But in the Rubrics doberar fo. 50a.
  129. Cf. vol. i. p. 4.
  130. Cf. Dun Cuaer Ann. Ul. 803, 804, 817, Aedhaein 806, Iellaen 825, Aerdd 835, Cluaen, 844, Tommaen 870.
  131. Cf. Maileruen in the list of saints fo. 32a, Cluen, Ann. Ul. 817.
  132. Cf. moer = móir Ann. Ul. 745, 755, 759, 778, 780, 782, 813, 827, 832, 834, 839, 841, 844, 850, 855, 872, Roes = Roiss 746.
  133. Cf. fruich Philargyrius.
  134. Ed. W. S., KZ. xxxi. 246 sq. Cf. Rev. H. M. Bannister, Journal of Theological Studies, 1903, pp. 49 sq. The first fragment is written on the margin and has been mutilated by the cutting of the leaf. How much has been oat away may be conjectured from a mutilated piece of Latin on the margin of the other side of the leaf, which, as the Rev. H. M. Bannister saw, agrees closely in its first part with the Stowe Missal, fo. 24 a:
    pro īcolumitate
    lorũ ac re:
    tís adstant
    tirũ ∴ pro re
    nostror̅ •:
    et pro requie d
    iteneris • scī •:
    astico or:
    et ōnib; regib
    = Stowe Missal: pro incolumitate regum et pace populorum ac reditu captiuorum, pro uotis adstantium, pro memoria martirum, pro remisione pecatorum nostrorum, et actuum emendatione eorum, ac requie defunctorum, et prosperitate iteneris nostri, pro domino papa episcopo et omnibus episcopis, et prespeteris, et omni aeclesiastico ordine, pro imperio romano, et omnibus regibus christianís etc.
  135. The life of S. Fintan was edited from A by Goldast, E«rum Alamannicarum scriptores aliquot uetusti, i. 203 sq. (Frankfurt, 1730), by Mabillon, Acta Sanctorum, iv. 1, 378 sq., and again by Mone, Quellensammlung der badischen Landesgeschichte, i. 54 sq. (Carlsruhe, 1848), with readings from other MSS. The text has been re-edited by Holder-Egger, Monumenta Germaniae Historica Scriptorum, Tomi , xv. Pars i. 502 sq. (Hanover, 1887). The passages containing the Irish notes have also been published by Zimmer, Glossae Hibernicae, 272 sq.
  136. In an edition of the Vita Findani prepared by him, and preserved in the Hof- und Landesbibliothek of Carlsruhe (Codex Sanblasianus 33 fol. 1 — 115), Van der Meer gives the readings of a Saint Gall codex communicated to him by the Abb6 Stocklin of Disentis. The text resembles a very corrupt copy of C. Feket diu todia anatheset in dahdane.

    Ata oblec (ohleecl) ichi Xpm christ ochus Patri grat machie forna fell tarn nakisel thcil chur (or crur7) tart doittis teilco ilfar kisel.

    Cucendo chach chuchen det faden maicf de hachf.

    Quine ilaocus innadichi in loge et longe celederemut voferfas torithius.

  137. Zimmer lays weight on the fact that in A and G are added over the Irish words isket, etc. and ataich, etc., the Latin versions licet tibi a deo ire in ahbatiam (quoted by Holder-Egger from A only), and obsecra christum et patricium nomen ciuitatis. If these additions are in A in a different hand from the text, and in the text-hand in C, that would prove that C was copied from A, otherwise not.
  138. Ed. Reeves. The Life of St Columba... written by Adamnan, ninth abbot of that monastery, Dublin, 1857.
  139. The MS. is described by Reeves, op. cit. xiii. sq., who gives specimens of the script.
  140. Op. cit. XIV.
  141. The other MSS. are described by Reeves, op. cit. xxiv. sq. The most important linguistically is Reeves' Codex B, a vellum MS. of the middle of the fifteenth century, preserved in the British Museum, Bibl. Reg. 8 D. ix., and which represents a text independent of A ; cf. Zimmer, KZ. XXXII. 199. The part of this MS. containing the names of S. Columba's disciples and relations is printed infra, p. 281.
  142. In some cases Dorbbene has introduced a later orthography: Ceannachte 56» { = Cenacte B), Ceate 58» { = Cete B), Feachnaus 32* {=Fechtiaus B, C, F, S), Deathrib 52" {=zDethrib B), Leathain 118» (=Lethani B), Clied 55'> { = Cleeth B); ea appears in final position in Lea 28» ( = Lea B), cf. dea in the Cambray Homily. The later ia appears in niath 25^ { = math B); this is doubtless due to the transcribers, not to Adamnan ; as to Miathorum 18» it may be remarked that this is a foreign name, which Reeves, p. 33, identities with the Motarat.
  143. At the end of a word we find MoLua 76».
  144. The gen. Colgion ZS*' by Colgen is remarkable. Attention may be directed to the middle vowel of Fechureg 23 (by Feclireg 121») and Ainmurech 49*", Ainmureg 108».
  145. In 59» Boend (cf. Boend Lib. Ard. 11») comes from Bqfind, but the reading Bofind in B shews that the form Boend is not to be imputed to Adamnan.
  146. Cf. violthu infra p. 353.
  147. 'an orate common for us,' d'Arbois de JubainVille, Itev. Celt. xv. 137.
  148. The MS. was first printed with many omissions and inaccuracies by Muratori in his Anecdota Ambrosiana, Padua, 1713, torn. iv. pp. 119 — 159. Reprinted without alteration in his Opera Omnia, Arezzo, 1770, tom. xi. pars iii. pp. 217 — 225, in Migne's Patrologia Curs. Lat. tom. Ixxii. coll. 579 — 608, and somewhat more fully in O'Laverty's Historical Account of the Diocese of Down and Connor, Dublin, 1884, vol. ii. Appendix, pp. ix. — xlv. A photographic edition of the whole MS. was excellently edited in 1892 by the Rev. F. E. Warren for the Henry Bradshaw Society; and from his introduction we have drawn the greater part of this description. The hymn beginning with Precamur Patrem (no. 3 in the Antiphonary) is called Imimmd na n-Apstal in Adamnan's Second Vision, Rev. Celt. xii. 432. The hymn printed infra, p. 782, has been edited with more or less inaccuracy by Peyron (Ciceronis Oratt. Fragmenta inedita, Stuttgard, 1824, pp. 225—6) and by Zeuss-Ebel (Grammatica Celtica, p. 944). Dr MacCarthy (Transactions of the Royal Irish Academy, vol. xxvii. p. 239) undertook to correct Muratori's and Peyron's misreadings of this hymn, and not only misprinted Congillum, Fintendnum, Boe- taenus, Noster and Cumineus for Comgillum, Fintenanum, Berachus, Notus uir and Cumenemts, but read a hole in the parchment (f. 30 v.) as corde, ' the only instance of picture-writing that I have found in Irish MSS.'
  149. Tern6c died 716 a.d.
  150. Ed. W. S., Goidelica, 175 sq. (in part); Windisch, Irische Texte, i. 312 sqq.; Zimmer, Glossae Hibernicae, 267 sqq., cf. Supplementum, 14 sq.
  151. Supplementum, 16.
  152. Note in particular the confusion of nn and nd, and the treatment of final vowels in the fourth poem.
  153. Cf. Celt. Zeitschr. iv. 490. In Ml. and Wb. -sem is rare, Celt. Zeitschr. iv. 66.
  154. Cf. KZ. XXXV. 325 sq.
  155. Cf. nach quod Ml. 101» 1.
  156. An Aed mac Derviato is mentioned in the Annals of Ulster, 713 a.d., but it is not stated to what part of Ireland he belonged.
  157. This final -e, -i could be restored throughout, also mb, nd for mm, nn of the MS.
  158. Perhaps the form aue may help to fix an inferior limit. In the Annals of Ulster the forms aue, auib, auu cease about 760 a.d. When we take into account the fondness of these Annals for archaistic forms, e.g. Uloth=Ulad 809, 894, 897; Mumen 778, 792, Irmumen 834; Dermait 822, 834, 847, 850, 851, 869; er cath-iar cath 865; Clona 759, 764; Nodan 808, Nodot 817, Tomce, Tomae 739, 748, 750, 751, 767, 780, 781, 793, 812, 850, it is probable that forms like aue were disappearing from literary use about the middle of the eighth century.
  159. The Irish verses have been edited by Zimmer, Gloss. Hib. 264, by W. S., Goidelica-, 1872, p. 18, and by Windisch, Berichte der Konigl. Sachs. Gesellschaft der Wissenschaften, 1890, 8. 84; the text of the codex has been edited by Ch. F. Matthaei, xiii. epistolarum Pauli codex Graecus cum versione Latina veteri, Misenae, 1791 ; cf. Zimmer, Gloss. Hib. xxxiii. sq. ; Traube, O Eoma Nobilis, 52 (348).
  160. According to Traube, probably identical with the scribe whose name appears in the Leyden Priscian (see above p. xxiv. note 1).
  161. According to Traube, possibly Aganon Bishop of Bergamo (837—867).
  162. The heretic Gottschalk of Orbais; cf. the words cited by Zimmer, Gloss. Hib. xxxvi.
  163. Gunthar, Bishop of Cologne.
  164. Hartgar, Bishop of Luttech.
  165. Hildewin, predecessor of Gunthar as Bishop of Cologne.
  166. Marcus? or Mariauus Capella?
  167. Of the Liber Hymnorum ff. 1 — 15* were edited from T by Todd, Leabhar Tmmuin (sic). The Book of Hymns of the ancient Church of Ireland, 2 vols. Dublin 1855, 1869. The whole of the Liber Hymnorum has been published in 1898 by Bernard and Atkinson for the Henry Bradshaw Society. The Irish hymns have been published from T by W. S., Goidelica 121 sq., and by Windisch, Irische Texte i. 1 sq., with variants from F, p. 321 sq.
  168. For individual hymns the following MSS. have been used :

    E = Egerton 93, British Museum. According to O'Curry, Cat. of Irish MSS. in the British Museum, the first 19 folios of this vellum MS. were written in 1477. This codex contains (fo. 19, col. 1) a copy of Patrick's Hymn. The text approximates to that of R. The two MSS. represent a different recension from T.

    L = The Book of Lismore, a fifteenth century manuscript in the possession of the Duke of Devonshire. It contains a copy of Ultan's Hymn, published by W. S., Lives of Saints from the Book of Lismore, pp. 51 sq. It belongs to the same recension as F.

    R=Rawl. B. 512, Bodleian Library, written in the fourteenth and fifteenth centuries. It contains the text of Patrick's Hymn, printed by W. S., Trip. Life, pp. 48 sq.

    X = Laud Misc. 615, Bodleian Library. It contains Ultan's Hymn; the variants have been published by Bernard and Atkinson.

  169. Cf. Todd, op. cit,, W. S., Goidelica^ 61, Windisch, Irische Texte i. 3 sq., Bernard and Atkinson, Liber Hymnorum I, x. sq,
  170. Cf. Zimmer, Keltische Studien i. 9 sq., W. S., Trip. Life, p. cii., Bernard and Atkinson, I. XIII. sq.
  171. Cf. Bernard and Atkinson, The Irish Liber Hymnorum II. xxxv. sq.
  172. Cf. Ann. Ul. 664, 667, and note to Fdl. Oeng. Sep. 4 (p. cxlii.).
  173. Cf. Windisch, Irische Teste i. 10 sq. ; Zimmer, Eeltische Stadien ii. 162 aq. ; Thurneysen, Rev. Celt. vi. 326 sq. ; Bernard and Atkinson II. xl. sq., 175 sq.
  174. Confusion of ae and a must be assumed if bebae 1. 6 is to rhyme with feba 1. 7, but the rhyme here is not obligatory. In 1. 45 adcobra seems to rhyme with lobrai, but adcobrai may be restored, cf. adrannai, F4i. Oeng. Sep. 6.
  175. Zimmer, Keltische Kirche 217 ( = p. 40 of Miss Meyer's translation), assigns the poem to the tenth century. But the language is decisive against this.
  176. Thus dos/uc 1. 36 may be replaced by domic, doxfiufced 1. 34 by doxnhuced (if the verse be not au interpolation), ismalle v. 66 by immalle. For dod/etis in place of dafetis, wliich the metre would equally allow, analogies may be found in later Old-Irish, cf. Celt. Zeitschr. iv. 67.
  177. Eev. Celt. vi. 334 sq.
  178. Qui refertur post Erodem nutriendus Nazareth | multa paruus multa adultus signa feoit celitus I quae latent et quae leguntur coram raultis testibus | praedicans celeste regnum dicta factis approbat. | DeMlesfacituigere,cecosluceilluminat, uerbis purgat leprae morbum,mortuos resmcitat.
  179. Cf. Windisch, Irische Texte i. 25 sq., Bernard and Atkinson ii. 1 sq., 189.
  180. In I. 73 ar-do-utacht is for ar-da-utacht, but this may be an error of transcription.
  181. Argairt for argart 1. 33 may be an error of transcription.
  182. Reprinted by Windisch, Ir. Texte i. The close agreement may easily be perceived from the concordance given by Bernard and Atkinson, ii. 1 sq.
  183. Cf. Bernard and Atkinson TT. lvi. sq.
  184. Cf. Bernard and Atkinson II. Lvn. sq., 208 sq.
  185. See J. M. Eodwell's Koran, p. 179, Sir R. Burton's First Footsteps in East Africa, p. 33, and A. Maury, Journal des Savants, Juin 1873, p. 745. With the whole incantation cf. the twelfth Assembly of Al-Hariri, translated by T. Chenery.
  186. As dith infra p. 346, of did, perf. sg. 3 of dinim.
  187. The Irish gloss has been printed by W. S., Goidelica, p. 2, the whole fragment by W. Meyer, Nachrichten von der Konigl. Gesellschaft der Wissenschaften zu Gottingen, Philologisch- historische Klasse, 1903, pp. 163 sq.
  188. Op. cit. 168 sq.