352 Proceedings of the Royal Irish Academy. 



*fteiiiiaviiidagni ? = QmRvendani, Hiibiier, Inscr. Britt., = Chcninddn L. Arm.) 

 with similar influence of v. 



6. i is the junction-vowel for i-stems, but may possibly extend to other 

 stems as in Gaulish names (Holder, ii, 2, 1. 6). The instances of all kinds 

 noted are — Assicona, Battigni, Cassitas, Cunigni, Ditibeas, Dolatibi? Drutiquli, 

 Gamicunas (= G-aimclion, cf, gaimred), Glasiconas, Lodimani, Muibiti, Nisigni, 

 Qeniloci, Q,eniloc[a?]gni,^ Vlatiami. 



7. as junction- vowel seems usually due to influence of v or b (aspirate), 

 Calunovica, Casoni, Denoval, Eracobi, Ivodacca, Lagobbe? Veqoanai, Meddo?geni, 

 Vendogni. 



8. e occurs in Erxeim, . . . eneggni, perhaps from ?o-stems, giving 

 *-iagni, -egni. These are the only instances noted corresponding to the 

 very frequent MS. ending -dn. 



IV. — Compensatory Lengthening. 



1. The Ogham inscriptions prove clearly the important fact that at 

 least two distinct epochs of compensatory lengthening occurred before the 

 MS. period. 



2. The change of nt into d, and of oic into g, has already taken place before 

 the period of the Ogham inscriptions. 



3. The disappearance of g before a liquid, w^ith concomitant lengthening 

 of the preceding vowel, took place within the Ogham period. Early Ogham 

 -agni becomes late Ogham -an, -ann.- 



4. I have been unable to find any instance of d + liquid in oghams, but the 

 lengthening in cafhair from cathedra, and the resultant vowel a from ad- in 

 composition before liquids, seem to show that the change belongs to the same 

 period as the loss of g before a liquid. In the two instances of Dalo, J, 1895, 

 p. 133, the critical syllable has been supplied by Barry. 



5. When g disappears before a liquid, the preceding vowel, whether 

 accented or not, is long in MS. Irish. 



Instances of -agni = -an are abundant. 



Cunamagli = nom. Connidl. 



Netta Sagru = Nazar-us, Nazar-ius., L. Arm. [z = ts), gen. Natsair, 

 Nastair, Nazair in many genealogies. I have not found the nom. except 

 in the latinized form. If the reading of the ogham is certain, the MS. 



1 Gen. Cennlocain BB 122a25, nom. Cenlacan 123)341, doubtless = Cellachdn, Qeniloci 

 = Cellaig, nom. Cellach. 



2 The frequency of -ann = -agni = dw seems to indicate that the doubled consonant has here 

 phonetic value. If so, it can only be a tentative late device. 



