the words ceol music, seol sail (which are already so written in Old Irish), predominates so that the writing ceól, seól cannot be disputed.

25. The long accent is also met with over syllables in which two vowels which originally belonged to separate syllables coalesce to one syllable (cf. § 81). Originally in most such cases the first vowel may have predominated in the pronunciation. For Old Irish at any rate ina díaid after him (dead finis, Cymr. diwedd), téora f. three, bíu, béo living (§ 31), fríu towards them (fri for frith, § 174), líu, léo with them, appear to be more accurate than ina diáid, teóra, biú, beó, friú, liú, leó. The same is true as to drúi Druid, gen. drúad, dat. drúid and druí, druád, druid. The Modern Irish draoi Druid is pronounced as with a short u and a long i.

25. In the 1 and 3 pl. of the secondary tenses -mais, -tais, also appear for -mís, -tís. Here the written form is not correctly déntáis they would do, but déntais, for the a is in such cases only introduced in consequence of an existing or once existing broad vowel in the preceding syllable. Cf. § 254".

25°. In Old Irish a long accent often appears over short vowels before a double consonant, especially before grouped or doubled r, l, n: márb dead, lóndas indignatio (Ml. 18, 10 ed. Ascoli), and here (Ml. 26°, 2), óll amplus (Ml. 20a, 3). However this inclination to pronounce the vowel long in such words has been neither permanent nor consistently carried out, although it may have been repeated in different periods and dialects.

25. In Middle Irish the sign of length is sometimes found in places where it has not been proved in Old Irish, e. g. over the heavy deponent terminations in -ar: 3 sg. ro charastár amavit, 1 pl. do-deochammár venimus, 3 pl. asbertatár dixerunt, 3 pl. bátár fuerunt. This lengthening seems to have come in under the influence of a secondary tone which requires consideration in Irish for words of several syllables.

25%. The chief accent was early thrown back from the termination of the word, as may be concluded from the mutilation of the syllables of flexion. But certain phonetic appearances demonstrate positively many cases in which it did not stand on the radical syllable. Cf. §§ 25, 42, 46, 60, 61, 62, 77, 81, 83, 108, 247, 275, 286, 295, 300, 325.

25h. In verse a short terminal vowel not unfrequently rhymes with a long syllable of a stem. For example in a poem of the Codex S. Pauli cele (socius) rhymes with ré (time), and messe (I) with glé (splendidus). Also Sc. 37, 15 airgdidu (dat. of airgdide silver) with clú (fame). From these and similar appearances it would be unsafe without further evidence to consider as long numerous terminal syllables which are never found in Old Irish prose with a long mark upon them.

207. The possessive pronouns of the third person are in Old Irish (e.g. in the Milan Codex) very often found with a long mark: á ainm his name, á n-íc their salvation.

254. The 1 and 2 pl. active of the absolute flexion are in Old Irish but slightly supported by MSS., and the same is true in later Irish of the past tenses, in which the absolute flexion comes gradually to be altogether disused. To judge from the existing material the oldest forms of the terminations are -me and -mit in the 1st and -te in the 2 pl.; present bermme, bermmit, S-past carsimme (carste), and so on; also in § 275 it ought to be (cechnimme), "bérmme." But, especially in Middle Irish and Modern Irish there are also forms in -mi, -mai, ti, tai, which Stokes has taken into his paradigms: carstai amavistis, téstai ibitis, bérmai feremus; the a in -mai, -tai is only introduced in consequence of a suppressed broad vowel before the termination especially when the preceding syllable contains no slender vowel, so bérmai for bérammi. In Modern Irish the i of this termination (probably under the influence of a secondary tone) is pronounced long, as is shown by O'Donovan, Gr., p. 219, beirimid ferimus, beirthi, beirthidh (dh at the end is silent, cf. § 3) fertis. If the radical

syllable contains a broad vowel aoi (i.e. a long i preceded by a slightly articulated dull vowel u) has established itself in the termination, molamaoid we praise, moltaoi ye praise, molfamaoid we shall praise. Cf. p. 126, § 25a.

254°. From the ordinary formation of the relative of the 3 sg. in -es, -as there deviate: file qui est § 388, teite qui it, the relative form of téit, i.e. do-éit § 264°; to which may be added the perfect form boie qui erat (Goid.3 p. 87, Book of Armagh).

But there also occurs a preterite teite, teiti he went, besides téit, without any relative signification. In the same way luide he went, with luid § 302. Stokes (Beitr. zur Vergl. Sprachforsch. vii. p. 40, 42) is inclined in such cases to separate a pronoun -e, -i sometimes as nominative, sometimes as dative or accusative leigth-i duillen he threw a javelin; geibth-i Loeg cloich Loeg takes it, a stone; is Cuchulainn cobarthe, it is Cuculainn who would have helped him. Cf. § 205 et seq., but on the other hand there are the preterites ending in -ta, -tha § 309, to which perhaps some of these forms belong e.g. budigthe he thanked T. E. 14.

272b. Stokes has lately (Three Middle-Irish Homilies, Preface, p. ix) explained as of a particular imperfect form, though not expressly ro chreti, yet similar forms of verbs of the II conjugation, e.g. ro labra he spake, ro scríbai he wrote, and especially (without ro) adcobra he would, Hy. 2, 45.

276. A reduplicated future has also been proved in carim II amo: ni con chechrat act ní bas tol doib they will only love, what is their desire (Gloss on et erunt homines se ipsos amantes Wb. 30°, 2); as a rule it forms the B-future, § 282.

289. Difficult to explain is dudichestar (i. miastair i. huaid fesin), Gloss on ducetur in the clause gravis illi vita tristisque ducetur Ml. 30. The future passive should be du-diastar, as is shown by du-diastae § 321 and du-dí § 287.

PIECES FOR READING.

Pieces V., VI. have not hitherto been printed. The Echtra Condla Chaim was published by O'Beirne Crowe, Journal Arch. and Hist. Association, 1874, p. 118. The learner had better begin with the

Old Irish sentences in I. Of the remaining pieces V. is especially easy, for IV. Hennessy's exact translation may be examined. Italics indicate the expansion of a contraction in the MS.

I.

SENTENCES FROM THE OLD IRISH GLOSSED MSS.

1.

Ní mebul lemm precept soscéli (Wb. 1).

2.

Is uisse lóg a saithir do chách (Wb. 29).

3. Is sí ar n-ires hi sin atá mor dechur etir deacht ocus doinacht (Ml. 26).

4. Is triit dorolgetha ar pecthi duún (Wb. 26°).

5. Ni ru foraithmenair Duid isin t-salm so a n-durigni Abisolón fris (Ml. 24c).

6. Denid attlugud buide do Dia di cach maith dogní frib (Wb. 273).

7. Sech ni coimnactar ar namit son fortanbristisni (Gloss on obprimi nequivimus Ml. 135).

8. Ni bat litre nota aram cia scríbtair hi fers (Sg. 6b).

9. Amal fongníter ídil, síc fogníther donaib ánib (Wb. 27°).

10. Na taibred cách úaib bréic imm alaile (Wb. 27").

11. Gaibid inimib a n-etach macc cóimsa, amal nondad maicc cóima (Induite vos ergo, sicut electi dei sancti et dilecti per viscera misericordiæ, benignitatem... Wb. 27b).

27°).

12. Attlugud boide do Dia di bar n-ícc trit-som (Wb.

I. G.

9

27°).

13.

14.

Adib moga-si dano, atá far cóimdiu innim (Wb. 27).

Is airi am cimbid-se hore no predchim in rúin sin (Wb.

15. Bid di bar n-ág-si ron bia-ni indocbál (Wb. 25a).

16. Is hé in tecttaire maith condaig indocbáil dia thigerni (Wb. 8).

17. Ro bad bethu dom, dian chomalninn (Wb. 3o).

18. Ni riat na dánu diadi ar a n-indeb domunde (Wb. 28€).

19. Ni tairmthecht rechto, mani airgara recht (Wb. 2o). 20. Sech ni thartsat som ní comtachtmar-ni (Wb. 24). Berir do imchomarc uaidib (Wb. 31d).

22. Ar osailether hires tri degním; innarbar (read innarbanar) hires dano trí droch-gnimu (Ml. 14).

23. Ceni tormastar ho méit is trom cenae ho aicniud ut sunt lapides (Ml. 20a).

24. Tairchechuin resíu forchuimsed (Wb. 4a).

25. Nob sóirfa-si Dia dinab fochidib (Wb. 11b).
26. Is glé lim-sa rom bia buáid (Wb. 11a).

27. Nachin rogba uáll (Wb. 15a).

28.

Mani ro chosca som a muntir in tain bíis cen grád, ni uisse toisigecht sochuide do (Wb. 28).

29. Is immaille ro scaich in bolc do blith ocus in t-immun do denam (Lib. Hy. 11a).

30. Dobert goiste imma bragait fadesin conid marb, huare nad n-digni Abisolón a chomairli (Ml. 23).

31. Foillsigthir as n-isel in dóinacht iar n-aicniud, huare as in deacht fodaraithmine ocus no da fortachtaigedar (Ml. 25°).

32. Is hed dí (or dino) alligitime, scarad fri indeb in domuin ocus tol De do dénum (Gloss on non coronatur nisi legitime certaverit, Wb. 30a).

33. Is do thabirt díglae berid in claideb sin (Wb. 6a).
34.

"Cia atagegalldathar" ol Sencha. "Atagegallar-sa" ol Triscoth (L. U. p. 19b).

35. Cid bec cid mar ind inducbál ó dia tar hesi denmo ind libuir, bith má de do buith dait-siu hi coimthecht oco (Sg. 2a).