LATIN INSCRIPTION AT CIRTA. 399 



silent myself, I speak in my verses. If mea vita is right, it proves, as does clara in 

 V. 2, luxuria in v. 6, and sanctam for sancta in v. 8, that the final m, as Mr Blakesley 

 observes, was now mute : and this is confirmed by thousands of late inscriptions. In the 

 best classical times it had, as is well known, a dull half-suppressed sound which was 

 often represented in writing by a half letter. In the oldest inscriptions it is frequently 

 omitted, before the date when the poet Attius, a great grammatical reformer, fixed its 

 place in the orthography of the language; thus rendering at least one essential service to 

 the grammar of his language, on the declensions and conjugations of which the loss of 

 the final m would have had the most disastrous consequences, as is well proved by the 

 transition of Latin into the Romance tongues. The early loss of this final letter con- 

 tributed much to the rapid decay of the Umbrian, as we know from many existing 

 monuments. In demonsfro the n was almost or quite mute, but the o was proportionately 

 lengthened. We know on the authority of Cicero and others that this was generally 

 the case when n preceded * or f. Hence thensaurus is the genuine Latin form of the 

 Greek Qtiaavpos- The unaccentuated de was thereby rendered probably shorter than it 

 would otherwise have been to Praecilius. Commodian opens his Instructiones with this line 

 Praefatio nostra viam erranti demonstrat. The later poets, even those who profess to 

 observe quantity, perpetually shorten this de in composition. Even so good a grammarian 

 as Servius, who lived of course when quantity had to be acquired by artificial rules, tells 

 us that in a word like amicus you know by its accent the second syllable to be long, 

 but must learn the quantity of the first arte, that is by your gradus. Now as Praecilius 

 had no gradus, he took the liberty of making the first syllable of demonstro short at the 

 conclusion of his verse, and the first syllable of honeste long in the first part of verse 9. 

 He knew no difference in quantity between demonstro and recondo, honeste and venisse, 

 the accent in all cases determining only the length of the penultimate. In the same way 

 versibv^s sounded to him the same as profugus, titulos in v. 11 the same as litora. His 

 first line had for him precisely the same cadence as Hie ego qui taceo, numeris mea fata 

 recondo would have had. 



In V. 2 the last two syllables of Praecilius coalescing probably rendered the accentuated 

 i peculiarly long, and the prae proportionately short, though even to writers who profess 

 quantity in the fourth and following centuries this prae was essentially a short syllable. 

 Even so early as the first century ae came to be used to denote the short open e in words 

 like praemo etc. 



V. 3 was no less rhythmical to Praecilius than Emollit mores didicisse Jideliter artes. 

 The first syllable of tare with its clear liquid sound was perhaps more distinctly long 

 to him as to a modern Italian, than the a in arma or fato. Compare the Italian pro- 

 nunciation of mare, mano, rosa etc.; and fydes in v. 4 and the Italian fede. In the ac- 

 centual church hymns, attributed to St Ambrose by Bede and others, quantity is observed 

 with more or less care; yet we find such a verse as Qui eras ante saecula. The e of 

 eras was as long as the a in lare. The frequency with which Commodian uses words 

 like duce, quoque, neque, homo and such like for spondees or trochees is very striking. 

 The last syllable was of course quite as indifferent to him and Praecilius as to a modern 

 Vol. X. Part II. 51 



