Reference edition:
Victorinus, Caius Marius, Liber de definitionibus, ed. Th. Stangl in Tulliana et Mario-Victoriniana, Munich 1888, 17-48.
This work, handed down among the logico-rhetorical works of Boethius and once assigned to the latter author, is in fact a short treatise attributable to Marius Victorinus, the well-known 3rd-4th century rhetorician and philosopher, commentator of Cicero's Topica, converted to Christianity in later life and author of important theological writings. The constant quotations from Cicero, the mention of a translation from Porphyry as his own work, and the fact that Boethius, Cassiodorus (especially in the Expositio psalmorum and in Book II of the Institutiones; but on this aspect see Stoppacci's 2018 reservations) and Isidore of Seville quoted passages from this text in their works, which constitutes an important testimony to the development of logic in late antiquity, are the fundamental evidence to support its authorship.
The text is based on the following manuscripts:
M = Codex Latinus Monacensis 14272 (10th-11th c.), a miscellaneous codex containing mainly Boethian works.
N = Codex Latinus Monacensis 14819 (10th-11th c.), contains logical works by Boethius.
V = Codex Bernensis 300 (11th cent.), contains logical writings by Boethius, Apuleius and Augustine.
They largely depend, though presumably not by direct derivation, on a single archetype. M is the best codex; N is not descriptus of M, despite being later, but derives from a different subarchetype; V is generally marred by more errors, but also preserves a fair number of good lessons not attested in M and N.
It is the only logical work by Victorinus that has come down to us intact, and is a complement to the commentary to the Ciceronian Topica. The work remained virtually unknown during the early Middle Ages, but reappeared between the end of the 10th and the beginning of the 11th century, within miscellaneous codices on logical-rhetorical subjects, in which it is often associated with Boethius' dialectical work, so much so that it has been misattributed. [D. Di Rienzo]