The commentary on the plays of Terence written by the grammarian Aelius Donatus in the middle of the fourth century was a work of massive scholarship. Observations about Terence's dramatic technique and characterization, his models, and his rhetoric abound, some of them very acute and many of them quite learned. What is more, his use of Greek terminology, primarily concerning rhetoric, is extensive and has relatively few parallels in other surviving Roman commentaries, with the exception of Porphyrio's commentary on Horace. And unlike most Roman commentaries, it is not clogged with potted mythology and erratic Roman history.
Alas, the commentary of Donatus as it left his pen is long gone. The great grammarian's work was sliced, diced, and squeezed into the margins of copies of Terence in the fifth or sixth century, and in the eighth century someone tried to reconstitute a single Donatus using two of these sets of marginalia. That version is neither complete nor coherent: the many double notes saying the same thing about the same words reveal its double marginal origins, as does the section of the commentary on Phormio which simply has two uncombined sets of notes. Before even reaching the end of Act 3 of Phormio, the compiler stopped compiling: the rest of the play has only a single set of notes, and there is no commentary on Hautontimoroumenos. Furthermore, not everything in our Donatus is the genuine article, even abridged: there are some clearly intrusive notes, some of which are inane or trivial in themselves, and some of which interrupt and make gibberish out of reasonably coherent sentences of the commentary. "The so-called commentary of Donatus" (Aeli Donati quod fertur commentum) in the title of Carmela Cioffi's new Teubner text of the commentary on Terence's Andria displays a fully justified skepticism about what this text actually is.1
The same hesitant title was used by the last scholar to edit Donatus. Paul Wessner was the greatest expert on Latin scholia of his—or perhaps any—generation; he published his edition of the commentary on Andria and Eunuchus in 1902, the remainder in 1905. In the past fifty years, however, important work has appeared about both the commentary and its transmission, notably studies by Otto Zwierlein, Michael Reeve, and Rainer Jakobi; Cioffi herself has discussed the manuscripts of the text in several recent articles.2 A new edition is fully justified, and even if Cioffi's text is relatively close to Wessner's, her apparatus criticus is based on a much fuller knowledge of the manuscripts and transmission.
Few of the differences between Cioffi's text and Wessner's make a significant difference. Some changes are clearly for the better; in particular, she often rightly follows emendations proposed by Jakobi.3 On the other hand, she frequently (but not consistently) treats readings of the fifteenth-century manuscripts (many of which she rejects and Wessner accepts) as conjectural; but in some cases the readings of the later tradition, conjectures or not, are better than her own choices, and some of Cioffi's own conjectures are less than convincing.4 Cioffi is more rigid than Wessner in following the indications of her stemma; in some cases, that is probably right, but the complete absence of lines of contamination from her stemma gives one pause. She promises a volume of Prolegomena which I have not seen; it is only fair to reserve judgment about some of her choices until her arguments for them can be read.
In terms of layout, Cioffi is more sparing than Wessner in the use of italics to indicate intrusive Carolingian material (although at times her choices are puzzling), and she rightly does not follow Wessner in italicizing every et that links one comment to the one before it. On the other hand, where Wessner used quotation marks to indicate words cited from the text and Sperrdruck for other words being discussed or explained, Cioffi uses quotation marks for both; her text is confused (and confusing) in a number of passages. Thus at 800.1, Cioffi's text reads: "SED QVOS P. V. 'interrogatio' est, cui necessario respondetur 'aut etiam aut non', 'percontatio', cui nihil horum. . . ." The point is to distinguish between two types of question, those expecting a yes or no answer (interrogatio) and those to which a broader response is necessary (percontatio. The proper punctuation (eliminating excess commas and without using Sperrdruck) is "SED QVOS P. V. interrogatio est cui necessario respondetur aut 'etiam' aut 'non'; percontatio cui nihil horum . . . 5
Most of the difficulties with Cioffi's edition, however, have to do with the difference between the fourth-century Donatus and the eighth-century compiler. At 5.1, there is a quotation from Cicero, at the end of which the commentator adds hoc in Topicis habetur." Cioffi, following Jakobi, deletes that sentence, and she also inserts into the quotation two words of Cicero that the manuscripts omit. Both changes are wrong: the reference to the Topica is certainly medieval, but may well be by the compiler rather than later; more significantly, we do not know if even the real Donatus quoted Cicero accurately. It is always wrong to emend quotations in a late antique text on the basis of what we think the wording of the quoted text actually was: these quotations may provide evidence, which should not be hidden in the apparatus, of the history of the earlier text. An even worse emendation of this kind appears at both 939.2 and 946.3: the text twice quotes the same short phrase, which it attributes to Sallust. Because it is in fact probably from Cicero, Cioffi emends Sallustius to Cicero twice. Moritz Haupt once said that if the sense required it, he would emend the monosyllable "O" to "Constantinopolitanus." Fair enough; but that does not apply to a text like this.6
Cioffi knows that the only text that can actually be edited is the Carolingian version, and that is the rule she, like Wessner before her, generally follows. But while Wessner preserved the occasional mis-ordering of notes to show what the Carolingian commentary is like, Cioffi consistently rearranges them to follow the order of Terence's text. Easier for the reader, perhaps; but a fair representation of "Donatus"? At times, to be sure, it is hard to know what to do. The real Donatus used a lot of Greek, but some of that is given in our manuscripts in Latin letters—but it is not consistent, and even when Greek is used, one needs to be suspicious of medieval scribes showing off, as they do, their knowledge, often faulty, of the Greek alphabet. Which alphabet should we use? A famous problem in this regard is the Latin spelling of words derived from λαμβάνω. The proper Greek word is σύλληψις and only late and bad Greek texts have σύλλημψις, while the correct Latin spelling is syllempsis or even syllensis. Wessner consistently and wrongly printed σύλλημψις, while Cioffi varies between the two Greek spellings, although the real choice is between σύλληψις and syllempsis. 7 Scholiastic texts also sometimes give Greek words Latin endings; thus at 45.1 Wessner followed A and printed ἀξι ώματe as a macaronic ablative, while Cioffi prints an authentic Greek dative. Wessner is correct for the eighth century, Cioffi for the fourth. Which text does one edit?
I have left for last what is in many respects the most important part of an edition of a scholiastic text, the testimonia, which ought to guide the reader either to bibliography or to Donatus' sources or place his notes in the context of late antique and medieval scholarship. Wessner's appendix is austere, but useful. Cioffi's apparatus of testimonia is larger, but omits some of Wessner's useful parallels and adds a great many that are completely irrelevant. At 10.4, for instance, she gives a reference to Donatus' note on Ad. 288, presumably for the use of ad aliquid. It is not closely parallel, but a great many examples of this common grammatical usage (meaning "relative" terms, like brother or slave, which imply a relationship to someone else) are collected by Samantha Schad in her invaluable lexicon, which Cioffi should have cited here. She gives a host of references for glosses on idoneus at 757.1: most of them are completely irrelevant to this passage, and she omits (as she seems to do consistently) any reference to Goetz's Thesaurus Glossarum Emendatarum.8 I have found very few additions to Wessner that are in fact useful, omissions of citations he gives that are relevant (as to the Bembine Scholia at 811.1), and a great many that are completely pointless and do not belong at all. That in at least one case she takes over from Wessner a typographical error in a reference (on 790.1, where RhLM 526.11 should read 521.11) does not inspire confidence.
No edition of a text like this can ever represent it completely and accurately, and it takes courage to undertake an insoluble problem. Cioffi's edition has its problems, and the testimonia in particular could have used more work. But on the whole her text is accurate and careful, even if there are mistakes; one awaits her Prolegomena for more detailed discussion of the text. But Cioffi's edition only supplements Wessner's; it does not replace it.
1. I discussed the early history of the commentary in HSCP 79 (1975) 335-54; fuller discussion and bibliography forthcoming in Critics, Compilers, and Commentators (New York, 2018). I base this review on a sample: I have fully collated Cioffi's edition against Wessner's for the commentary on the first fifty pages (on Andria 1-106) and the last fifty-five pages (Andria 740-981) and all my examples (cited by lemma and sentence numbers) come from those sections.
2. O. Zwierlein, Der Terenzkommentar des Donat im Codex Chigianus H VII 240 (Berlin, 1970); M. Reeve, particularly in Hermes 106 (1978) 608-18 and CP 74 (1979) 310-326; R. Jakobi, Die Kunst der Exegese im Terenzkommentar des Donatus (Berlin, 1996) (and other articles). Cioffi lists seven of her own articles in her bibliography.
3. R. Jakobi, Göttinger Forum für Altertumswissenschaft 20 (2017) 25-33. They include changes at 28.1, 35.1, 73.3, 85.1, 855.7, 856.1. At 820.1 she follows his deletion of officiorum but does not mention his suggested supplement (which is plausible) of iniuriosam. She probably should have followed him in some other places, e.g. Praef. 2.2.
4. Why, for instance, posit a lacuna at 797 rather than accept eligere from the later tradition? or reject D's dicta at 911? They may be emendations, but they are still plausible. Other dubious choices Cioffi has made include se scilicet at 14.1, commodis eorum at 64.1, and the deletion of uel principium at 748. At 69.2 Jakobi's addition of Sinonem is better than Cioffi's deletion of paulo post nominaturus. The two truly bad emendations I have found are ἔμφυτον dici scimus at 795 and Sunium for Rhamnus at 930.1. At 64.1 she omits the transmitted scilicet; at 28.3 she correctly prints remaneat where Wessner has maneat. The only manuscript I have consulted is Paris, Lat. 7920 (=A): (Gallica BNF). The only serious typo I have noted is QVOD for QVO at 804.1.
5. Some other examples of bad punctuation at 4.1-2; 25.4; 763.1.
6. Cited in C. Belger, Moritz Haupt als academischer Lehrer (Berlin, 1879) 126; quoted also by A. E. Housman in his lecture "The Application of Thought to Textual Criticism" in J. Carter, ed., A. E. Housman: Selected Prose (Cambridge, 1961) 142.
7. The classic treatment of this is by Wilhelm Schulze, Orthographica (Rome, 1958; originally published Marburg, 1894).
8. S. Schad, A Lexicon of Latin Grammatical Terminology (Pisa, 2007); G. Goetz, Thesaurus Glossarum Emendatarum (=Corpus Glossariorum Latinorum Vols. 6-7)) (Leipzig, 1899).