BMCR 2024.03.25

Cassiodorus: Institutiones humanarum litterarum. Textus Phi Delta

, Cassiodorus: Institutiones humanarum litterarum. Textus Phi Delta. Corpus Christianorum series Latina, 99. Turnhout: Brepols, 2022. Pp. 512. ISBN 9782503595894.

In its final authorial version, Cassiodorus’s Institutiones consist of two books: the first (Institutiones diuinarum litterarum) deals with the Holy Scriptures; the second (Institutiones humanarum litterarum) offers a short introduction to the human disciplines included in the Triuium and Quadriuium. The work was conceived as a handbook for the monks living at Vivarium, which is where Cassiodorus published it in around 550–70.[1]

As is well known, the whole work in this two-book final form was edited by Mynors (1937), and his critical edition remains an excellent piece of Late Latin scholarship.[2] This book by Ilaria Morresi does not try to replace Mynors’s edition: it does not deal with Cassiodorus’s final version, but with two interpolated versions of its second book, known as Φ and Δ.[3] Both of them were doubtlessly worth editing: first, they derive from another authorial draft (ω), specifically an earlier version of Cassiodorus’s Institutiones humanarum litterarum, prior to his final version (called Ω). By comparing both texts and Ω, it is possible to reconstruct ω as well as to understand the textual genesis of the work from ω to Ω.[4] Second, both Φ and Δ include a large number of interpolations on the liberal arts, taken from Late Antique and Early Medieval sources: Φ and some parts of Δ can be dated to the 6th–7th century, and their composition may have occurred at Vivarium; some parts of Δ are  the work of one or more Carolingian interpolators, prior to Hrabanus Maurus. These interpolations are interesting per se, and some of them, such as the famous Anecdoton Holderi (a crucial source for the biographies of Symmachus, Boethius and Cassiodorus himself), have attracted considerable attention. Some major variants of both corpora were included in a separate apparatus by Mynors. However, there was no comprehensive edition of any of them, and Mynors was only able to collate a reduced number of manuscripts, without establishing any stemma.

Thus, Morresi’s book, which must be complemented with her own commentary (Morressi 2023), does fill a significant gap in Late Latin scholarship. The book starts with an extensive introduction containing three chapters. The first deals with the work and its versions: despite the complexity of the subject, the exposition is clear, informative and up-to-date, and should be read by everyone interested in Cassiodorus. The second chapter deals with the manuscript tradition. All currently-known manuscripts are concisely described.[5] The Φ-tradition consists of four complete and eight partial witnesses; Δ had more offspring (12 complete manuscripts and 11 partial witnesses), perhaps because its text was longer and consequently of more interest to the Carolingian centres in which it was copied. Morresi has collated the whole manuscript tradition. This has significantly improved our understanding of this work, not only with regard to those manuscripts which were previously unknown, but also the copies which Mynors disdained. A very important case is that of Par. Lat. 12963 (siglum: ζ), copied in the third quarter of the ninth century: this codex, which was dismissed by Mynors as “a poor text”, has now been rightly recognised as one of the manuscripts closest to Δ, together with λ and ι.

The stemmatic analysis is methodologically excellent and succeeds in establishing a credible stemma for each tradition. It is worthy of praise that Morresi never hides inconsistencies and contradictory agreements, nor tries to diminish them; on the contrary, she always deals with them transparently, exposing the pros and cons of each possibility, and opting for the most plausible and economic explanation. Many relations are obscured by the constant presence of contamination, due both to the erudite nature of the work (whose readers and scribes would tend to look for better or more complete texts), as well as to its large diffusion, which made it available in many scriptoria. The history of the text which emerges from this set of data is consistent and allows us to discard previous hypotheses, such as Holtz’s proposal of an insular stage in the line of transmission, for which there is no actual direct evidence (see p. 146); the origin of Φ in the south of Italy (perhaps Vivarium) is well supported with manuscript and phonetic evidence (p. 82). The analysis is almost exhaustive and therefore only minor observations can be made; and most of them reinforce Morresi’s conclusions:

  • On p. 68 she shows that m transmits an emended text, due to process of “correzione ope ingenii”. But given that the three passages are not corrected by conjecture, but in light of their respective sources (Terentius and Boethius), it may be better to speak of “contaminatio ex fontibus”, on which see Delmülle (2018).
  • At pp. 118–119 Morresi discusses a set of passages in which ι sometimes agrees with απ and sometimes with απβ δκ; she correctly proposes that this is probably due to corrections made by the ancestor of β δκ. In support of this, it could be relevant to note that ι agrees with απ mostly at the beginning (between 28,265–72,105); hereafter, when ι agrees with απ, it also agrees with β δκ. This could indicate that the ancestor of β δκ started to correct some erroneous passages, but at some point these emendations were interrupted or spared; henceforth β δκ inherited more of these old errors found in ι απ as well.
  • Morresi has skilfully identified very plausible examples of the practice of reproducing layout, i.e., the process of not only transcribing a text, but also reproducing the mise en page (lineation, pagination, etc.) of the exemplar (see pp. 112–113). Besides Giovanni Orlandi’s classic article quoted on p. 114 n. 224 (Orlandi 1994), on this matter it is useful also to consult Reeve (1989, particularly 11–15) and De Angelis (1995).
  • On pp. 119–120 Morresi rightly dismisses the innovations common to ι and λ as false conjunctive errors. However, as for their origin, they seem to be more polygenetic than a consequence of contamination.
  • The textual tradition issuing from Δ has two branches (εθζλ vs. ιηαπβδκ). On p. 132 Morresi rightly states that 157,14 actu (λ ιαπθ2) and arte (εθc.ζηβδ) are transmitted in both branches, perhaps because both readings were found in the hyparchetype itself (see p. 132*). I agree, but one can perhaps restrict this uaria lectio to one branch. The exemplar of εθζλ did transmit both arte (εθζ) and actu (λ). However, the ancestor of ιηαπβδκ perhaps read only actu, and the ancestors of η and βδ(κ) were independently contaminated with manuscripts belonging to the first branch εθζλ (arte); this would be consistent with the behaviour of these witnesses at other points.

The critical edition follows the principles and criteria explained in detail by the editor on pp. 148–164. The first edited text (p. 1–102) reconstructs II, that is to say, the first interpolated version attested by Φ and used as the basis for the second interpolated version (III, transmitted by Δ). This task has been completed in a consistent and very accurate way. Readers of Mynors’s edition will recognise the use of half brackets ⸤ ⸥ ⸢ ⸣ to include right readings of the first version which were changed by Cassiodorus in Ω (⸤ ⸥) and errors found in ω (or at least unanimous in ΦΔ) which are corrected in Ω (⸢ ⸣).[6] Morresi is methodologically right to distinguish between II (the original of the first interpolated version) and Φ (its corrupt archetype), as well as between III (the original of the second interpolated version) and Δ (its corrupt archetype): if ΩΔ agree against Φ, then Φ’s reading should be considered as an error which was committed in the line of transmission between ΙΙ and Φ. If Φ and Δ transmit different texts, both are separately printed in parallel columns. Different apparatuses are employed to record:

i) sources;
ii) differences between II and Ω, and differences between Φ (the hyarchetype of II) and Δ;
iii) readings in Φ manuscripts;
iv) readings in Δ manuscripts.

After ω’s text, Morresi first edits the additions which were inserted by the first interpolator and are transmitted by Φ and Δ (pp. 103–151). Subsequently, she presents the critical edition of the texts included by the third interpolator, which are only transmitted by Δ (pp. 153–220).[7] The text seems to be well preserved and Morresi’s edition is consequently conservative, but she intervenes when there is need, and her proposals are convincing: on p. 10, 39 hoc <est> signatur is a better solution than Stoppacci’s seclusion (one cannot imagine a good reason for adding hoc signatur, whereas the omission of est is easier, particularly if est or hoc est were abbreviated); on p. 38, 31 Δ the addition of <in> is probably right (cf. ibidem Φ’s text l. 29); the crux on p. 90, 190 Boethius iulatur opusculum composuit is still the best solution. No proposal is wholly convincing, although iu– seems to me to have probably issued from u.i. (uir inlustris), as assumed also by van de Vyver (uir inlustris patricius) and Morresi’s suggestion (uir inlustris Latinum).

After the critical editions, the book includes the graphic schemata of contents found in the manuscripts, as well as a set of very useful indices of biblical sources, literary sources and loci paralleli, manuscripts, keywords, Greek words, and personal names. The book has been carefully produced: there are almost no typographical errors, and the very few cases I have detected are trivial and do not affect the reading.[8]

To sum up, this is an excellent piece of Late Latin scholarship. The knowledge of manuscripts, editions, stemmatics and previous literature is outstanding. From now on, every reader of Mynors’s edition should always have this book (and the complementary comment) in the other hand.



De Angelis, Violetta, “Fedeltà al antigrafo nell’impostazione della pagina. Il caso dei tesi commentate.” Filologia mediolatina 2 (1995) 57–67

Delmulle, Jérémy, “La ‘contaminatio ex fontibus’ dans la transmission des florilèges. Réflexions à partir du cas d’étude des florilèges augustiniens.” Filologia mediolatina 25 (2018) 1–43.

Morresi, Ilaria, Le Institutiones humanarum litterarum di Cassiodoro: Commento alle redazioni interpolate. Instrumenta Patristica et Mediaeualia, 88. Turnhout: Brepols, 2023.

Mynors, Roger and Aubrey Baskerville, Cassiodori Senatoris Intitutiones. Edited from the Manuscripts by R. A. B. Mynors. Oxford: Clarendon Press, 1937, repr. 1961.

Orlandi, Giovanni, “Apografi e pseudo-apografi nella Nauigatio sancti Brendani e altrove.” Filologia Mediolatina 1 (1994) 1–35.

Reeve, Michael D., “Eliminatio codicum descriptorum: A methodological problem.” In Editing Greek and Latin texts: Papers given at the Twenty-Third Annual Conference on Editorial Problems, University of Toronto, 6–7 November 1987, edited by John N. Grant. New York: AMS Press, 1989, 1–35 [repr.: Michael D. Reeve, Manuscripts and Methods: Essays on Editing and Transmission. Storia e letteratura, 270. Roma: Edizioni di storia e letteratura, 2011, 145–174].



[1] The best extant copy, Bamberg, Staatsbibliothek, Msc.Patr. 61, still preserves a subscription confirming that it ultimately derives from Cassiodorus’s “official exemplar”, which should be used to check and correct other copies (fol. 67v): Codex archetypus ad cuius exemplaria sunt reliqui corrigendi. The manuscript, quoted as B, is online; see

[2] See Mynors 1937.

[3] Mynors called them ‘II’ and ‘III’ (Mynors 1937, xxiv); version ‘I’ was Cassiodorus’s final version. Morresi uses II and III as the original interpolated versions, whereas Φ and Δ denote their corrupt archetypes. This is a valid distinction and has its consequences: see p. 61 and infra.

[4] Even though they ultimately constitute the second book, the Institutiones humanarum litterarum were actually written first. In its final form, Cassiodorus places his definitive version of this book (Ω) after the Institutiones diuinarum litterarum. This division is of consequence in the manuscript transmission: of the ca. 100 manuscripts containing at least some parts of the work, only three transmit the two-book final structure: ca. 60 codices preserve the Institutiones diuinarum litterarum in isolation, and ca. 40 manuscripts transmit the Institutiones humanarum litterarum in one of the three versions (Ω, Δ, Φ). See Morresi pp. 9–11.

[5] Morresi deals also with the earliest indirect tradition, Isidore of Seville and the so-called Paulus abbas (both deriving from an intermediate stage between ω and Ω), as well as Alcuin of York and Hrabanus Maurus (who are witnesses for Φ and Δ, respectively).

[6] In these cases, ω’s paradosis printed between ⸢ ⸣ may truly be an authorial error, or a mistake committed between II and ω.

[7] In all these editions, reconstructing Δ is one of the most problematic issues: an isolated reading of Δ can be either the original text of this version (= III), or an innovation committed in the line of transmission from III to Δ.

[8] P. 115* limita > limitata; p. 137 n. 266 Sànchez > Sánchez.