litteris tres sunt, quae sive breves sive longae ejusdemmodi sunt, A, I,
U: similiter habent sive longae sive breves.

But two (E, O) change their quality:

[Id. ib.] O vero et E non sonant breves. E aliter longa aliter brevis
sonat. Dicit ita Terentianus (hoc dixit) 'Quotienscumque E longam
volumus proferri, vicina sit ad I (i with macron to show length)
litteram.' Ipse sonus sic debet sonare, quomodo sonat I (i without
macron to show short) littera. Quando dicis _evitat_, vicina debet esse,
sic pressa, sic angusta, ut vicina sit ad I litteram. Quando vis dicere
brevem e simpliciter sonat. O longa sit an brevis. Si longa est, debet
sonus ipse intra palatum sonare, ut si dices _orator_, quasi intra
sonat, intra palatum. Si brevis est debet primis labris sonare, quasi
extremis labris, ut puta sic dices _obit_. Habes istam regulam expressam
in Terentiano. Quando vis exprimere quia brevis est, primis labris
sonat; quando exprimis longam, intra palatum sonat.

[Ars Gram. Mar. Vict. de Orthog. et de Metr. Rat., I. vi. 9.] O qui
correptum enuntiat, nec magno hiatu labra reserabit, et retrorsum actam
linguam tenebit.

It would thus seem that the long E of the Latin in its prolongation
draws into the I sound, somewhat as if I were subjoined, as in the
English _vein_ or Italian _fedele._

The grammarians speak of the obscure sound of I and U, short and
unaccented in the middle of a word; so that in a number of words I and U
were written indifferently, even by classic writers, as _optimus_ or
_optumus, maximus_ or _maxumus_. This is but a simple and natural thing.