least occasionally have represented Latin c by σ,—a thing which none of them has ever done. It is probable that the modification of c which is noticed in the modern languages was a characteristic of the Umbrian and Oscan dialects and so prevailed to some extent in the provinces, but there is absolutely not the slightest evidence to show that it formed a part of the pronunciation of cultivated men at Rome.]

4. D: had regularly the sound of English d; but at the end of words nearly that of t.

(a) The position of the vocal organs in uttering this letter is described by Terentianus Maurus (p. 331 Keil); Marius Victorinus (p. 33); and Martianus Capella (III. 261).

(b) That final d was sounded like t is clear from the positive statements of Quintilian and from the fact that in inscriptions, as well as in the best manuscripts of Plautus and Vergil, we find almost indifferently ad and at, apud and aput, haud and haut, quid and quit, as well as adque and atque and many others.

[At about the fourth century A.D., di before a vowel began to be pronounced somewhat like the French j, just as in Aeolic Greek we find ζά for διά. Hence in the modern languages g and j arise out of Latin di. Compare Latin diurnus with the Italian giorno and the French jour.]

5. E: ē had the sound of English e in "they" or of the French ê; ĕ had the sound of English e in "net".

(a) The position of the vocal organs in pronouncing e is described by Terentianus Maurus (p. 329 Keil); Marius Victorinus (p. 32); and Martianus Capella (III. 261). It is regularly represented in Greek transliterations by ε when short, and by η when long.

(b) The sound of the letter e seems to have varied more than was the case with other vowels. The later grammarians give to ē a sound approximating to the sound of i. (Cf. Donatus in Servius p. 421, Keil ^[1]). And confusion of ĕ and ĭ in words like timidus, navibos (written timedus, navebos) is to be seen in early Latin. But too much importance has been given to this. The fact is that one short unaccented vowel is very likely to be mistaken, for another, especially by the uneducated and by careless speakers. The hearer cannot detect the difference, and in fact there is none, practically. The extremely accurate and discriminating elocution of which we hear was in all probability confined to the highly cultivated classes.

6. F: had practically the sound of English f.

Latin f is not like the Greek φ, which was a double sound rather than a single one, namely p + h with each element distinctly audible, as in English top-heavy, uphill. Quintilian says: "The Greeks are accustomed to aspirate; whence Cicero in his oration for Fundanius ridicules a witness who could not sound the first letter of that name."^[2] The descriptions given by Priscian and Terentianus Maurus of the position of the lips and teeth in pronouncing f show that it was formed precisely as our f, i.e. with the lower lip against the upper teeth.

7. G: g always had the hard sound of English g in "get".

(a) "When g comes before an s it produces x, thus showing that it is a guttural: e.g. lex = leg + s; and rex = reg + s.

(b) No Roman grammarian mentions more than one sound as belonging to g, although they treat of the letters minutely.

(c) All the vowels readily interchange after g in the same root, which would hardly be the case if g had had more than one sound. Thus we have maligenus and malignus; lego, legis, legit; gigeno and gigno; tegimen and tegmen.

(d) Latin g is invariably represented by Greek γ, and the Greek γ is invariably represented by Latin g. St. Augustine remarks: "When I say lege, a Greek understands one thing and a Roman another in these two syllables." This shows that Latin lege and Greek λέγε had precisely the same sound.

[About the fifth century A.D., g began to have the soft sound before e and i that is now found in the modern languages. The first change from the old hard sound was to a y sound like that given to g by those who speak the Berliner Dialekt in Germany to-day, and said to be found also in Lowland Scotch. Such variations as magestas for maiestas, and in Greek βειέντι for viginti, occur.]

8. H: had the sound of English h.

(a) H is described as a simple breathing by Marius Victorinus, p. 34 (Keil); Terentianus Maurus, p. 331; and Martianus Capella, III. 261. It is represented in Greek by the rough breathing, and in turn it represents that breathing.

(b) There seems to have existed among the uneducated Romans that irregularity in the use of h which marks the language of the English cockney to-day. Nigidius Figulus, the grammarian, said: "Your speech becomes boorish if you aspirate wrongly." Catullus in one of his epigrams ridicules the cockneyism of a person who said chommoda for commoda, and hinsidiae for insidiae.^[3] In later Latin, the varying spelling shows the growing irregularity of usage. H seems to have been omitted or inserted almost at pleasure; thus hauctoritas, hii, and hinventio, stand beside inospita, omini (homini), and abitat (habitat). The reason for this irregularity seems to have been the gradual weakening of the sound until h became a silent letter, as it is in modern Spanish and Italian. ^[4]

9. I consonant (J): had the sound of English y.

(a) That i had a consonant sound as distinct from its vowel sound is clear from the statement of Priscian (I. p. 13, Keil). Before a vowel and not preceded by an accented syllable with final consonant, he says that i "passes over to the force of a consonant." That it differs from i the vowel, is also clear from the fact that in prosody it lengthens the preceding vowel.

(b) That it was not like English j is clear from the fact that it readily passes into i, which proves the two sounds to have been closely akin; and in Greek transliterations it is always represented by ι. Thus Julius = Ιούλιος.

(c) Nigidius Figulus cautioned his readers that the i (j) in such words as iam, iecur, iocus is not a vowel,—a caution that would have been absurdly unnecessary if i had had any such sound as that of English j.

(d) The true sound of the letter is seen in the alternative spelling Eanus for Janus proposed by some of the ancients, who derived the name from eo, ire. About 300 A.D. the letter got the sound of z or gi.

10. I