to such an extent that no firm division can be made. The
language of some manuscripts of the 14th century contains forms which
are really Old Irish, and Middle Irish orthography was partly employed
by historians and antiquarians in the middle of the 17th century. Old
Irish, as compared with Brythonic, preserves a wealth of inflectional
forms in declension and conjugation, but many of these tend to
disappear very early. In the modern dialects of Ireland and Scotland
there is a rigid rule of orthography that a palatalized, or, as it is
termed, slender consonant in medial or final position, must be
preceded by a palatal vowel (i), and a non-palatalized consonant by a
non-palatal or broad vowel (a, o, u). This is the famous rule of the
grammarians known as _caol le caol agus leathan le leathan_ ("slender
to slender and broad to broad"), but it is not so strictly adhered to
in the spoken language as is commonly stated. In the older language
the quality of medial and final consonants is only denoted very
imperfectly, thus non-palatalized final consonants are regularly not
denoted as such, e.g. O. and Mid. Ir. _fir_, Mod. Ir. _fior_. In Old
and Mid. Irish the initial mutations are only regularly denoted in the
case of the vocalic mutation of c, p, t, s, f, and the nasal mutation
of b, d, g. The vocalic mutation of c, p, t, s, f was denoted by
writing ch, ph, th, sh, fh, the first three symbols of which were
derived from the Latin alphabet. Another method of denoting the
mutation was to write a dot over the letter, originally the punctum
delens, which was justified in the case of mutated f as the latter
early became silent. But no such devices were ready at hand in the
case of the medial b, d, g, and the mutated forms of these consonants
were consequently not represented at all in the orthography. The same
remark holds good in the case of the nasal mutation (eclipse) of the
tenues. But it is easy to demonstrate that the same condition of
affairs as we find in the modern language must have obtained in Old
Irish. This insufficiency of symbols renders the orthography of the
early stages of the language very complicated. We find that b, d, g
were used initially to denote the voiced stops, but medially and
finally they represent spirants, the voiced stops in this case being
denoted by c, p, t. It is not until much later times that the h in the
mutated forms of the t