Among the varied lines of inquiry discernible in Donatus' commentary on Terence are indications of a comparative method that sought to elucidate the Latin text by reference to its Greek model(s). This testimony eventually generated a line of scholarship that made Terence's relationship to them the center of attention. The resulting preoccupation with lost originals no longer dominates scholarship on Roman comedy, but that does not reduce the value of Donatus' evidence. His testimony is worth revisiting both to understand his own interest in making these comparisons and to see what (else) his observations can reveal about Terence's dramaturgy.

Roman writers of palliata comedies never forgot their debt to Greek predecessors and did not let their audiences forget it either, not even when—or perhaps particularly when—they did significant violence to their models. In those situations they seem especially keen not so much to prove, in Eric Handley's words, "that the goods were genuine," but to revel in the fact of their difference. So the prologue-speaker of Casina gleefully announces that this Latin version of Diphilus' Klerumenoi has eliminated the young man whose eventual appearance triggered the original play's climax.

in urbem non redibit: Plautus noluit;pontem interrupit, qui erat ei in itinere.

He will not return to town. Plautus said no.He broke the bridge that was on his route.


That departure from the Greek model must have been of enduring interest, for when Casina was revived a generation or so later, the prologue as expanded for the new occasion retained the original declaration.1 The [End Page 83] acknowledgment of Greek predecessors is no less prominent in the plays of Terence, though in his case the emphasis is a little different. His famously polemic prologues deal less with the mere fact of Greek models than with the process by which they were made Roman: thus their emphasis on creating logical action, on the preference for good writing over farce, and most notoriously on the borrowing from secondary models that we persist in calling contaminatio. The first thing an audience learns upon encountering a Terentian play is what he did with and to its Greek model.2

Small wonder then that later Romans remained mindful of the palliata's Greek roots, from Varro, who is said to have preferred the opening of Terence's Adelphoe to that of Menander, to Cicero, who thought a preference for the original Greek plays highly pretentious, to Gellius two centuries later, who (somewhat to his own surprise) did prefer Menander to Caecilius.3 From there, the next major step in the history of reception brings us to the grammarian Aelius Donatus, who taught the plays of Terence to students in the mid-4th century. In addition to observations, many quite astute, on matters of diction, rhetorical figures, characterization and the like, the commentary that survives under Donatus' name preserves remnants of a comparative method that sought to explain a reference in Terence's text or to explicate his intentions in a scene, a character, or sometimes just a single line by explicit reference to what his model by Menander or Apollodorus had done before him.4 The commentary is [End Page 84] thus itself an important source of information about a range of Greek plays, and its testimony came to figure prominently in a line of Terentian scholarship—long the dominant line of Terentian scholarship—that made his models and what he did to them the center of attention.

It was not always so. For an early critic like Anne Dacier (1625–1720), whose editions of Plautus (1683) and Terence (1688) would set the terms of scholarly discourse for the better part of two centuries, the comparison of interest was between Plautus and Terence. A change from Terence's Greek model, such as the addition of the second lover Charinus and his slave Byrria to Andria, mattered to her not for the structural consequences that preoccupy modern scholarship but for the motivation behind the change.5 Only in the course of the 19th century did a growing admiration for things Greek lead critics to regard the Roman plays less as aesthetic objects to be enjoyed in their own right and judged by their own standard than as imperfect copies of lost Greek masterpieces. The structural repetitions, inconsistencies, and interpolations introduced by the Roman dramatists in the process of adaptation then came to matter more than the comic exuberance and dramatic richness admired by earlier generations, and attention shifted from what Plautus and Terence achieved in creating new Latin comedies to what they had failed to reproduce in adapting the earlier Greek ones.6

Plautus' reputation was a well-known casualty of that aesthetic shift ("a blacksmith mending a watch," in Gilbert Norwood's oft-quoted phrase), while praise of the distinctly more subtle Terence could sound very like faint damns. He was only tinkering when seen to depart from his model and was little more than a translator when taken to be faithful to it.7 These are no longer the dominant attitudes. Over the last forty or so years, as a new generation of Latinists approaches Roman comedy [End Page 85] with new critical tools and, it must be said, a more appreciative sense of humor, and as considerably more Menander emerges from the Egyptian sands to claim the attention of Hellenists, it has become possible to treat Greek New Comedy and the Roman comoedia palliata as distinct, if closely related, art forms.8 The one need no longer be judged by the putative standards of the other. Given this change in perspective, it is worth revisiting Donatus' testimony to see what it is good for besides the reconstruction of lost Greek plays. The aim of the exercise is to set his occasional comments on or about Terence's models within the larger context of his interpretive enterprise and, by doing so, to see how his comparisons of Latin play with Greek original suited his own style of exegesis and, ultimately, what his comparative method reveals about Terence's dramaturgy.


Admittedly, much of Donatus' textual explication is quite basic. Though the problematic nature of the surviving commentary may on occasion leave the point of a note unclear, the practical concerns of the schoolroom seem never far away. The commentary often reflects the struggles of Donatus' students to understand the subtleties of a colloquialism not quite their own: When Terence writes faciuntne intellegendo, is -ne the interrogative particle, i.e., the equivalent of nonne, or an emphatic Greek νή (ad An. 17)? What is the difference between vereor ut and vereor ne (ad Hec. 101) or between the fish-sellers called cetarii and those called piscatores (ad Eun. 257)? So, too, with the emphasis on figures of speech. A rhetorically based curriculum will always be keen to discuss their use, but the penchant of Terentian characters for ellipsis and prolepsis in their conversation requires frequent, often seemingly elementary explication by expanding—as commentators and teachers continue to do—and re-arranging what Terence wrote to make a pattern of thought easier to see and a sentence easier to construe (e.g., ad An. 401, Eun. 557).9

Drawing examples from Terence's Greek predecessors for purposes of comparison and contrast introduces a higher level of observation, but the origin and purpose of those comparisons are problematic. Terence was read primarily as a stylistic model in antiquity, and while his Greek antecedents [End Page 86] might be consulted, he does not seem to have attracted anything like the close scrutiny of sources that produced the early collections of Vergil's debts.10 Donatus himself almost certainly knew Terence's models only at secondhand, and who, if anyone, among his predecessors would or could have troubled to make the kind of systematic comparisons of Latin and Greek texts that modern analytic scholars sometimes envision is unknown. All we can really say is that Donatus' own work was, like his great work on Vergil, a variorum commentary, that he quoted and responded to comments on Terence's Greek models drawn from among those varia, and that remnants of some fuller knowledge of the Greek plays thus informs his style of exegesis.11 Though Apollodorus of Carystos, who provided the models for Hecyra and Phormio, left little impression on the later record, something of his work was evidently available for earlier commentators to consult, and direct knowledge of Menander certainly endured for centuries.12

Among Greeks, Menander enjoyed a status not far below Homer. He was a school author from early on, and familiarity with his work, whether gained through collections of maxims and speeches or from the experience of whole plays, often reflects the book knowledge of a reading public. Yet his plays also continued to be performed in the public theaters of the Roman empire and provided private entertainment (staged or recited) in the houses of the rich.13 Among Romans, the [End Page 87] different strands of knowledge, direct or indirect, derived from books or theater, are more difficult to disentangle. When mentioning Greek performers, for example, Roman sources may not clearly identify the language of performance or the genre of the work performed. Thus Quintilian in the late 1st century describes the relative strengths of the comic actors Demetrius and Stratocles in performance, but what kinds of comedy they performed and whether in Greek or Latin is not clear from his comments.14 The longstanding practice of hiring comic actors to coach aspiring orators in techniques of delivery further obscures the distinction between theatrical performance and rhetorical demonstration.15 So Quintilian refers specifically to the prologues of Menander's Hydria and Georgos as performed by comoedi, but whether his readers would think first of the theater or of the classroom, where such speeches were regularly extracted for exercises in role-playing and the depiction of emotion, remains unclear. When praising Menander as the embodiment of all his principles, Quintilian refers explicitly to reading plays, not to seeing them in performance.16

It is the same problem we face in assessing the survival of palliata comedy in the late Republic and empire, when not the fact of its continued performance becomes uncertain but the role of performance in preserving the comic legacy. While Roscius was certainly famous for his characterization of Ballio, and the palliata's influence on subsequent literature is increasingly recognized by modern critics, the growing emphasis on texts in the record makes it difficult to gauge the relative importance of reading and watching in shaping the Roman literary imagination.17 Quintilian at [End Page 88] one point offers a long list of palliata characters in a context that strongly implies performance (11.3.74), but when he quotes the famous opening of Terence's Eunuchus (quid igitur faciam? 46), three of his four references are in the context of poetry read, while the last (on the delivery of Phaedria's opening monologue) contrasts actor and orator in a context that may recall stage performance but is equally consistent with the kind of knowledge to be derived from a comoedus' classroom demonstration.18

Two centuries later, Donatus was still drawing on comic texts to illustrate techniques of ethopoiesis, but where Quintilian's recommended exercises are so often based on extracts from Menander, Donatus draws on the Latin of Terence:

ad Eun. 986.1 Hem quid amat. haec singula pronuntianda sunt, ut stuporem nimiae indignationis ostendat.

Well what he's in love. These words should be articulated separately [i.e., Well? What? He's in love?] to show the bewilderment of excessive indignation.

ad Ad. 134.2 Profundat perdat . haec sic pronuntianda sunt, ut ostendatur gestu nolle quod loquitur.

Let him squander, waste. These words should be articulated in such a way that he is shown by his bearing not to want what he says.

Observations like these on the use of voice and carriage, frequently couched in the traditional terms of rhetorical actio, are not designed primarily to make students more appreciative spectators or better readers of Terence. Given the larger aims of Donatus' curriculum, they are best understood as instructions to guide students as they practice reciting [End Page 89] these passages as a means to hone their own oral skills.19 So too with his numerous references to facial expression (vultus), since the face may provide clues for the tone and bearing appropriate to the speaker and the situation.20

We are looking, then, at a pedagogical method—in principle not so different from our own—designed to help students understand the literal meaning of the text, appreciate its artistry, and employ that text to advance their larger educational goals. The fact that the Roman plays had Greek models is a given, but to assess the dramatist's "originality" by comparing his work with those models was a task that Donatus was neither equipped nor inclined to undertake. At times, as we will see, he even seems to resist such a tendency in his predecessors. The extant commentary nevertheless preserves details about plays by Menander and Apollodorus that provided much grist for what became the later analysts' comparative mill. What was that knowledge, and how did Donatus himself use it, if not explicitly to preserve knowledge of those models?


Sometimes Donatus casually offers up a fact about the Greek original as if in answer to a student's idle question: Why in Eunuchus is Chremes, the adulescens with an old man's name, so brusque and suspicious? Answer: In Menander he was a farmer (ad Eun. 507). Why is the boys' father so hostile to the meretrix Thais? Answer: In Menander he had long resented her corruption of Phaedria (ad Eun. 1001). Other observations run deeper, as when he notes the several occasions where Terence turned monologue into dialogue or dialogue into monologue, or the moment in Andria when he heightened the emotional pitch of the confrontation between Pamphilus and his father Simo, or how he (but not Menander) has Micio protest the escalating demands on him at the end of Adelphoe.21 Changes [End Page 90] like these usher in significant alterations in characterization and in the quality and pace of Terence's stage action.

Other useful details about Greek plays may emerge from the rhetorical analysis of Terence's text. So, for example, the information that Menander's Perinthia and Andria opened with similar scenes, and that Terence adapted the expository dialogue of Perinthia for his Latin Andria and changed only the senex Simo's companion from wife to freedman, comes not as an observation about Terence's own opening scene, but earlier in explicating the argument in his prologue, where the poet confesses to and defends the appropriation. The note there is phrased in the question and answer format honed by critics of Homer and carried over into the schoolrooms of the empire: a "problem" is posed for which the teacher then supplies the answer: sed quare ergo se onerat Terentius . . . sic solvitur . . . (ad An. 14).22 Donatus' later, dramaturgically astute comments on the action and characterizations of the expository scene as Terence wrote it make no mention of its debt to the Perinthia.23

Hardly less significant are the verbatim quotations that confront the Latin text directly with its Greek original. Gunther Jachmann performed a significant service by gathering these quotations along with other evidence to facilitate comparisons, and this body of material has proved useful not just for the reconstructive projects of scholars like Otto Rieth and Walter Ludwig that tended to downplay Terence's originality but also for the more literary investigations of, among others, Heinz Haffter and Orazio Bianco that built it back up.24 Yet extracting fragments of Menander from Donatus' commentary is an imprecise task, and inconsistencies necessarily abound. The gloss on Iuno Lucina at Andria 473, illam [obstetriciam potestatem] Menander Dianam appellet, leads editors of Menander to include Ἄρτεμι as a fragment of the Greek Andria (fr. 35 K-T), but Donatus' note that the altar at Andria 726 belongs to Apollo, quem Ἀγυιαῖον Menander vocat, fails to secure the [End Page 91] corresponding phrase Ἀπόλλων Ἀγυιεύς for Menander's play. His claim that Andria 959–61 derives from Menander's Eunouchos is reproduced verbatim as fr. 165 K-T, even though Donatus' observation seems to be either faulty or incomplete: the corresponding passage at Eunuchus 551 is not in fact very close in sentiment, and no corresponding note appears in his commentary on Eunuchus.25 Citing any of this material out of its original context can also obscure Donatus' aims in making comparisons, aims which may differ significantly from those of his modern successors. It is therefore instructive to set some of his observations back in the broader context of his critical method and to consider how his knowledge of the Greek antecedents, however spotty, served the exegetical tasks he appears to have set for himself.

The point at issue in the commentary may be purely technical, as with the double accusative construction of nil me fallis (An. 204), which he identifies as a Grecism (figura ἑλληνισμός) and illustrates with the phrase from Menander's original, νῦν δ' οὐ λεληθάς με.26 A textual variant he found, e.g., audiam or audio at Andria 592?, may be explained by an appeal to Menander, τί ποτ' ἀκούσομαι; (fr. 38 K-T: modern editors retain the present of the MSS.). A garbled note on Andria 406–7 (venit meditatus alicunde ex solo loco: / orationem sperat invenisse se . . .) includes a gloss and a quotation:

ex solo loco "solo" deserto, ubi meditari facilius possit.—Menander from a lonely place "lonely" deserted, where it would be easier to think something through.—Menander

εὑρετικὸν εἶναί φασι τὴν ἐρημίανοἱ τὰς ὀφρῦς αἴροντες

Highbrows say that a deserted spotfosters creativity

We can see in this case how the quotation supports the gloss meditari facilius and that a literal version of Menander's joking reference to philosophers as those with raised eyebrows (cf. Tithe, fr. 395 K-T) might only puzzle Roman audiences, but it is nevertheless striking how little these [End Page 92] lines correspond to anything in Terence's text.27 A far better parallel to his meaning would have been the situation in Samia, when Moschion exits to plan an ethically charged appeal to his father:

τί δ' ούκ ἀ]πελθὼν εἰς ἐρημίαν τινὰγυμν]άζομαι;

Why don't I go off to some deserted spotto practice?

(Sa. 94–5)

The exegetical tradition must have seriously constricted the corpus of Menander available to commentators, which (curiously enough) does not appear to have included Samia.28

Other notes dig much deeper. Among the most interesting are those refuting rival interpretations. Though the eruditi whose preference for Menander's plays over those of Caecilus and Terence so annoyed Cicero remain largely anonymous, their values apparently endured and exerted no small influence on the later reception of Roman comedy. Gellius was evidently not the only reader of later times to confront at least a bit of a Latin play with its Greek model and find it wanting. Discernible behind many of Donatus' notes is a centuries-long tradition of carping and fault-finding. For those critics, as for their modern successors, Terence left many faults to find: he misunderstood a Greek word, failed to recognize an Attic deme-name, ruined a well-turned sententia, did not know that the men of Mykonos were proverbially bald. Yet what is most striking, and surely deserves more attention, is not that Donatus records such complaints together with the Greek texts on which they are based, but that he often responds to these objections with a combination of good sense and good humor. So, for example, on the bald men of Mykonos (ad Hec. 440):

imperite Terentium de Myconio crispum dixisse aiunt, cum Apollodorus calvum dixerit, quod proprium Myconiis est . . . unde etiam proverbium Graecum "μία Μύκονος." sed ego Terentium puto scientem facetius Myconium crispum dixisse. [End Page 93]

They say that Terence ignorantly called a man from Mykonos "curly-haired," although Apollodorus said "bald," which suits men from Mykonos . . . from which sentiment comes the Greek proverb "Mykonos is all one" [i.e., all bald]. I think, though, that Terence knowingly said "curly-haired Mykonian" as a joke.

It is a good joke, too. The detail is part of the description required for one of those impromptu and absurd fool's errands that are invented to remove a meddling slave like Parmeno of Hecyra from the scene. The absurdity of Pamphilus' hasty improvisation combines with Parmeno's inability to spot the mistake to enhance the comedy of the moment.29

A note on Eunuchus 689 is more convoluted, but no less interesting for its insight into Donatus' method. The old eunuch for whom Chaerea will be substituted is described as colore mustelino ("with the complexion of a weasel") because, said the detractores, Terence confused Menander's phrase γαλεώτης γέρων ("a gecko of an old man") with γαλῆ ("weasel"). So Donatus:

erravit Terentius non intellegens Menandricum illud "αὐτός ἐστιν γαλεώτης γέρων." ait autem stellionem, quod animal lacertae non dissimile est maculoso corio; namque ad id genus facies exprimitur eunuchorum corporis, quia plerique lentiginosi sunt. hic ergo erravit ideo, quia γαλῆ mustella dicitur, γαλεώτης stellio . . . [2] at ego †edesionum sequor, qui recte intellexit Terentium scientem mustellino colore dixisse eunuchum velut sublivido, quia vere eunuchi aut ex candidissimis lentiginosi fiunt, ut ex Gallis et eiusmodi occidentalibus, aut ex fuscis sublividi, ut ex Armeniis et aliis orientalibus.

Terence has erred, not understanding Menander's expression "a gecko of an old man." He [Menander] says "gecko" because this is an animal not unlike a lizard with a spotted skin: the body of eunuchs is represented as looking this way because many of them are freckled. He [Terence] has therefore erred since γαλῆ is a weasel, γαλεώτης a gecko . . . [2] But I agree with [?], who rightly understood that Terence knowingly said the eunuch had a "weasel-like complexion," i.e., somewhat livid, since in fact eunuchs drawn from very fair people like Gauls and westerners of that kind become [End Page 94] freckled, while those from darker-skinned peoples like Armenians and others from the east become rather livid.

The real point here is not whether Donatus and his unknown informant are right or wrong about geckos and lizards and eunuchs (whether Gallic or Armenian), but that, as in the case of the curly-haired man from Mykonos, he works from the assumption that Terence knew what he was doing and that the commentator's job is to make the dramatist's intention clear.

Donatus often does that quite well. He is a very close reader, alive to nuances of language and to the wider implications of Terence's stylistic choices. His note on the famous opening of Eunuchus is typical. This is the scene, echoed by Horace and Persius and an eventual favorite of rhetoricians, where the lovesick Phaedria, temporarily banished from Thais' presence, poses in despair the famous question quid igitur faciam? (46). Donatus' discussion centers on igitur, which he glosses as deinde. He offers a parallel from Plautus (AmPh. 209–10) and then quotes Menander's original, ἀλλὰ τί ποιήσω; (fr. 161K-T). He understands Menander's ἀλλά as an example of what Denniston calls hypophora, "the proffering and rejecting of successive suggestions . . . by a single speaker, who conducts, as it were, a dialogue with himself" (1975, 10–11). This leads Donatus to a perceptive observation of wider import: "the young man apparently thought over many possibilities silently and has at last burst out with these words."30 What began as a question about language thus grows into an implicit invitation to think about staging and, ultimately, about Terence's handling of traditional Roman material: Do we see Phaedria thinking silently, like Plautus' Palaestrio (Mil. 200–18)? Does he speak immediately upon entry, or does he first pace about and scratch his head, striking poses before he speaks, like the infatuated young lovers do at the beginning of Curculio and Pseudolus? Is he speaking aloud to himself or to the audience, or is he addressing his slave, Parmeno? And (to recall a longstanding crux for editors of this scene) when does Parmeno speak and in so doing turn this monologue into a conversation? Donatus may [End Page 95] well have posed such questions himself: he specifically notes the difference between monologue and dialogue in Menander's Andria and Perinthia (ad An. 140), and many observations elsewhere in the commentary suggest as interest in matters of staging and delivery.31 No such explicit questions survive in the extant text here, but the alignment of Terence's Latin with its Greek model at least helps us phrase the appropriate questions.

The exercise is no less valuable in cases of contrast between the Latin text and its model, though what Donatus made of the contrasts he preserves is often lost, since the composite nature of the surviving text can make his critical approach difficult to recognize. So at Hecyra 286–7, which is a bit of moralizing by the prissy young husband Pamphilus:

nam nos omnes quibus est alicunde aliquis obiectus labos,omne quod est interea tempus prius quam id rescitumst lucrost

All of us for whom some trouble lies in wait from somewhere,all the time that intervenes before it becomes known is sheer profit.

The extant commentary struggles to explain nos omnes, pointing out that Pamphilus should more logically have said either apud nos omnes or nobis omnibus, but Donatus then quotes from Cicero a helpful parallel to this anacoluthon, praetor quoque, ne se pulchrum ac beatum putaret, ei quoque carmen compositum est . . . ("And the praetor, so he shouldn't think himself sitting pretty and care-free, a jingle has been made up for him, too . . . ," Mur. 26). Then, even more interestingly, he cites the Greek original of Apollodorus:

οἱ γὰρ ἀτυχοῦντες τὸν χρόνον κερδαίνομενὅσον ἄν ποτ' ἀγνοῶμεν ἡτυχηκότες

We who are unfortunate count as profit all the timewhen we do not know we are unfortunate.

(PCG, fr. 10)

Donatus' own comment on the Greek does not survive, though the disjointed notes that follow in the surviving text make clear that his discussion continued beyond the quotation. [End Page 96]

What might be gleaned from the juxtaposition of Terence and his original here was pursued by Jachmann, who saw in Terence's omnes an otiose expansion that puts clumsy stress on Apollodorus' gnomic sentiment: the tightly shaped Greek expression is thus greatly extended in the Roman version and pulled out of shape (1934, 614). That is, in itself, quite true. What Jachmann fails to observe—and the point that surely lay behind Donatus' quotation of the original—is that the resulting clumsiness of expression is deliberate. The fault lies not with Terence but with his character. A well-turned phrase of the kind found in Apollodorus is beyond Pamphilus' ability. Terence makes him instead as awkward a moralizer as he is amoral a husband, just as he has the muddle-headed Ctesipho praise his brother at Adelphoe 258–9 with an awkwardness of syntax to match his characteristic incoherence of thought.32 The commentary thus not only preserves the illustration of a certain distance between Terence and his model, but also invites us to consider the logic of Terence's dramatic decision by showing what stood as a viable alternative to it. Donatus is not inclined here to make a value judgment on the decision, but he does indicate the possibility of a different dramatic value.

In other contexts, value judgments could come easily to him. Our one glimpse of his classroom manner suggests a mordant wit,33 and the commentary does not hesitate to note occasions when—to mention just a few examples—Terence has in his opinion done something mire (Eun. 254, 356; Ad. 560), docte (Eun. 981; Ph. 735), magna virtute (Ad. 539), mira arte (An. 965), and even magno ingenio (Ph. 352). Donatus may also note instances where he believes Terence has actually improved upon his model. Sometimes it is a small thing, like the unceremonious abruptness that immediately characterizes Demea on his first appearance in Adelphoe (melius quam Menander, ad Ad. 81), but the alteration can create a more significant difference. Donatus seems particularly wary of long expository monologues and regularly praises Terence for avoiding them, e.g., by adapting the expository pattern of Menander's Perinthia to prevent old Simo's long disquisition in the Andria from becoming tedious, or by recasting Chaerea's account of his outrageous escapade disguised as a eunuch as a dialogue by introducing Antipho to share the scene.34 He [End Page 97] clearly admired the great confrontation scene between Pamphilus and his father Simo in Andria (872–903), teasing out the emotional subtleties and explicitly noting that its elevation in tone and temperature was Terence's innovation.35

Yet the problematic nature of Donatus' composite text sometimes renders his testimony less than straightforward. He tells us clearly enough, for example, that Charinus and his slave Byrria were Terence's addition to Andria and were not found in Menander, but whether that means they were entirely Terence's invention, i.e., not found in either Menander's Andria or Perinthia, or that they were inspired by something Terence did find (like the opening expository dialogue) in Menander's Perinthia and then recast for the Latin Andria, is unclear from Donatus' testimony.36 The note praising Antipho's introduction at Eunuchus 539, ne unus diu loquatur ut apud Menandrum, is unclear in a different way: "as in Menander" could conceivably mean that Menander also avoided a lengthy monologue, though the phrase is not generally understood that way.37 The famous comment about the changed ending of Adelphoe, apud Menandrum senex de nuptiis non gravatur: ergo Terentius εὑρετικῶς, is less problematic but likely to be incomplete. Though εὑρετικῶς is widely understood to mean simply "by his own invention," the adverb may conceal a value judgment: εὑρετικός in ancient criticism often has, like English "inventive," a distinctly positive connotation. Terence, on that reading, has not just changed the ending but improved it.38 [End Page 98]


It is of course true that the vast store of knowledge represented by Donatus' commentary on Terence is in its present form incomplete (commentary on Heauton timorumenos is entirely missing), often incoherent, inconsistent, or obscure, and occasionally corrupt. The interpretive point he wants to make is not always evident, a fact too often forgotten as commentators through the millennia ransack his work to support their own interpretive arguments. The diversity of his critical range supported a significant variety of objectives, whether the question at hand was purely grammatical or ran to matters of style or dramaturgy. In reading his commentary both as the teaching text it originally was and as the document in the history of reception that it became, one thing stands out clearly. The breadth of knowledge required to open Terence to his 4th-century students suggests that he expected either to pose to them or to have posed to him a range of questions almost certainly wider than commonly encountered in modern classrooms. That difference is significant. We no longer read Terence, either on our own or with students, primarily to improve our Latin style or our ability to speak in character or our grasp of rhetorical figures and their power. Our interest in Roman comedy is less pressing, less immediate. That is why Donatus seems to give us more than we really need (and sometimes more than we want), but even as we cut his commentary down to a more manageable size and mine it for our own purposes, we need to remember that a work on this scale, however it may seem to mirror our present needs, once had an integrity all its own, with its own standards and its own methods in pursuit of what over the millennia remains our goal, viz. a fuller appreciation of Terence's dramatic art.39 [End Page 99]

Sander M. Goldberg
University of California, Los Angeles
e-mail: sander@humnet.ucla.edu


Arnott, W. G. 2003. "Diphilus KΛHROΥMENOI and Plautus' Casina." In Lecturae Plautinae Sarsinates VI: Casina, ed. R. Raffaelli and A. Tontini, 23–44. Urbino: Quattro Venti.
Augoustakis, A., and A. Traill, eds. 2013. A Companion to Terence. Blackwell Companions to the Ancient World, 103. Chichester: Wiley-Blackwell.
Barsby, John. 2000. "Donatus on Terence: The Eunuchus commentary." In Dramatische Wäldchen. Festschrift für Eckard Lefèvre zum 65. Geburtstag, ed. E. Stärk and G. Vogt-Spira, 491–513. Hildesheim: Georg Olms.
Bianco, Orazio. 1962. Terenzio. Problemi e aspetti dell'originalitä. Rome: Edizioni dell'Ateneo.
Bonner, Stanley F. 1977. Education in Ancient Rome. Berkeley: University of California Press.
Brown, Peter. 2013. "Terence and Greek New Comedy." In Augoustakis and Traill 2013, 17–32.
Büchner, Karl. 1974. Das Theater des Terenz. Heidelberg: C. Winter.
Dacier, Anne. 1688. Les Comédies de Terence, avec la Traduction et les Remarques, de Madame Dacier. Paris: Thierry and Carbin.
Demetriou, Chrysanthi. 2014. "Aelius Donatus and his Commentary on Terence's Comedies." In Fontaine and Scafuro 2014, 782–99.
———. 2015. "Donatus' Commentary: The Reception of Terence's Performance." In Terence between Late Antiquity and the Age of Printing, ed. A. J. Turner and G. Torello-Hill, 181–99. Leiden: Brill.
Denniston, J. D. 1975. The Greek Particles. 2nd ed. Oxford: Clarendon.
Fantham, Elaine. 1984. "Roman Experience of Menander in the Late Republic and Early Empire." TAPA 114:299–309.
———. 2002. "Orator and/et Actor." In Greek and Roman Actors. Aspects of an Ancient Profession, ed. P. Easterling and E. Hall, 362–76. Cambridge: Cambridge University Press.
Feeney, Denis. 2016. Beyond Greek. The Beginnings of Latin Literature. Cambridge, Mass.: Harvard University Press.
Fontaine, Michael. 2014. "The Reception of Greek Comedy in Rome." In The Cambridge Companion to Greek Comedy, ed. M. Revermann, 404–23. Cambridge: Cambridge University Press.
Fontaine, Michael, and Adele C. Scafuro, eds. 2014. The Oxford Handbook of Greek and Roman Comedy. Oxford: Oxford University Press.
Fraenkel, Eduard. 1968. "Zur römischen Komödie." MH 25:231–42.
———. 2007 (1922). Plautine Elements in Plautus. Trans. T. Drevikovsky and F. Muecke. Oxford: Oxford University Press.
Gammacurta, Tatiana. 2006. Papyrologica Scaenica. Alexandria: Edizioni dell' Orso.
Germany, Robert. 2016. Mimetic Contagion: Art and Artifice in Terence's Eunuch. Oxford: Oxford University Press.
Goldberg, Sander M. 1986. Understanding Terence. Princeton: Princeton University Press.
———. 1987. "Quintilian on Comedy." Traditio 43:359–67.
———. 2005. Constructing Literature in the Roman Republic. Cambridge: Cambridge University Press.
———. 2011. "Roman Comedy Gets Back to Basics." JRS 101:206–21.
---, ed. 2013. Terence: Hecyra. Cambridge: Cambridge University Press.
Gowers, Emily. 2004. "The Plot Thickens: Hidden Outlines in Terence's Prologues." Ramus 33:150–66.
Haffter, Heinz. 1953. "Terenz und seine künstlerische Eigenart." MH 10:1–20, 73–102.
Handley, E. W. 1975. "Plautus and His Public: Some Thoughts on New Comedy in Latin." Dioniso 46:117–32.
Hanses, Mathias. 2015. "Juvenal and the Revival of Greek New Comedy at Rome." In Marshall and Hawkins 2015, 25–41.
Holford-Strevens, Leofranc. 2003. Aulus Gellius. An Antonine Scholar and His Achievement. Oxford: Oxford University Press.
Ihne, W. 1843. Quaestiones Terentianae. Bonn: Karl Georg.
Jachmann, Gunther. 1934. "P. Terentius Afer." RE 5:598–650.
Jakobi, Rainer. 1996. Die Kunst der Exegese im Terenzkommentar des Donat. Berlin: De Gruyter.
Jocelyn, H. D. 1964. "Ancient Scholarship and Virgil's Use of Republican Latin Poetry, I." CQ 14:280–95.
Jones, C. P. 1993. "Greek Drama in the Roman Empire." In Theater and Society in the Classical World, ed. Ruth Scodel, 39–52. Ann Arbor: University of Michigan Press.
Kaster, Robert A. 1988. Guardians of Language. The Grammarian and Society in Late Antiquity. Berkeley: University of California Press.
Kragelund, Patrick. 2012. "Evidence for Performances of Republican Comedy in Fourth-Century Rome." CQ 62:415–22.
Lefèvre, Eckard. 1999. Terenz' und Apollodors Hecyra. Zetemata 101. Munich: Beck.
———. 2008. Terenz' und Menanders Andria. Zetemata 132. Munich: Beck.
Leigh, Matthew. 2004. "The Pro Caelio and Comedy." CP 99:300–35.
Ludwig, Walther. 1964. Review of Bianco 1962. Gnomon 36:152–60.
———. 1968. "The Originality of Terence and His Greek Models." GRBS 9:169–82.
Manuwald, Gesine. 2011. Roman Republican Theatre. Cambridge: Cambridge University Press.
Marshall, C. W. 2006. The Stagecraft and Performance of Roman Comedy. Cambridge: Cambridge University Press.
Marshall, C. W., and T. Hawkins, eds. 2015. Athenian Comedy in the Roman Empire. London.
Maynes, Craig. 2016. "Comic Callimacheanism in Catullus 67." TAPA 146:281–323.
McGill, Scott. 2012. Plagiarism in Latin Literature. Cambridge: Cambridge University Press.
Müller, Roman. 2013. "Terence in Latin Literature from the Second Century BCE to the Second Century CE." In Augoustakis and Traill 2013, 363–79.
Nervegna, Sebastiana. 2013. Menander in Antiquity. The Context of Reception. Cambridge: Cambridge University Press.
———. 2014. "Graphic Comedy: Menandrian Mosaics and Terentian Miniatures." In Fontaine and Scafuro 2014, 717–34.
Nocchi, Francesca Romana. 2013. Techniche teatrali e formazione dell'oratore in Quintiliano. Berlin: De Gruyter.
Pfeiffer, Rudolf. 1968. History of Classical Scholarship. From the Beginnings to the End of the Hellenistic Age. Oxford: Clarendon.
Rieth, Otto. 1964. Die Kunst Menanders in den Adelphen des Terenz. Hildesheim: Georg Olms.
Ritschl, Friedrich. 1845. Parerga zu Plautus und Terenz, I. Berlin.
Sharrock, Alison. 2009. Reading Roman Comedy. Poetics and Playfulness in Plautus and Terence. Cambridge: Cambridge University Press.
Shipp, G. P. 1960. P. Terenti Afri Andria, with Introduction and Commentary. Oxford: Oxford University Press.
Uden, J. 2006. "Embracing the Young Man in Love. Catullus 75 and the Comic Adulescens." Antichthon 40:19–34.
Victor, Benjamin. 2013. "History of the Text and Scholia." In Augoustakis and Traill 2013, 343–62.
Wessner, P. 1905. Aemilius Asper. Ein Beitrag zur römischen Literaturgeschichte. Halle: Hathi.
Wessner, P., ed. 1902. Donatus. Commentum Terenti. 3 vols. Leipzig: Teubner.
Zetzel, J. E. G. 2018. Critics, Compilers, and Commentators: An Introduction to Roman Philology, 200 BCE–800 CE. Oxford: Oxford University Press.


1. Only lines 5–22 of the extant prologue are generally taken to be additions for the new performance. For its composite nature, see Abel 1955, 55–61, and for revival performances in the Republican period, Manuwald 2011, 108–19. The novae comoediae denounced at Cas. 9–10 may actually be those of Terence. Plautus' warning has encouraged caution among modern scholars assessing the relationship of a Roman play to its Greek original. So for Casina, the wariness of Arnott 2003, 24–6. Handley is quoted from 1975, 117.

2. The argumentative quality of the Terentian prologues is approached in various ways by, inter alios, Goldberg 1986, 31–60, Gowers 2004, Sharrock 2009, 63–95, McGill 2012, 115–45. For the special problem of contaminatio, see most recently Germany 2016, 161–7. Particularly striking in the present context is Terence's willingness to use these speeches to advance aesthetic arguments in a style less reminiscent of Plautus than prefiguring Lucilius.

3. Varro ap. Suet. Vita Ter. 5.11, Cic. Fin. 1.4, Gell. 2.23. The change in attitude is largely attributable to Gellius' view of Roman comedy less as drama than as an exercise in translation. See Holford-Strevens 2003, 198–201. Cicero was also thinking of books, not performance, but his emphasis on the merits of the sermo patrius naturally leads to a different conclusion. The shift in taste is discernible in Quint. 10.1.100, regretting Latin's inability to reproduce the stylistic polish of Attic Greek (concessam Atticis venerem).

4. Donatus' original commentary, itself the product of a centuries-old exegetical tradition, was apparently reduced in antiquity to marginalia and the present text reassembled from its disiecta membra no earlier than the 7th century. Good introductions to its strengths and peculiarities by Barsby 2000, Victor 2013, Demetriou 2014. For Donatus' place in the record of Terentian commentary, see Zetzel 2018, 253–7, and for his career more generally, Kaster 1988, 275–9. The text is quoted throughout from the edition of Wessner 1902; an annotated text with French translation is available online at http://hyperdonat.tge-adonis.fr.

5. Mme. Dacier thus seizes on Donatus' idea (ad An. 301) that Terence was anxious to provide a husband for the otherwise jilted Philumena: "Cette remarque me paroît importante pour le Théatre, et merite qu' on y fasse reflexion" (1688, 72).

6. The course of this older scholarship is traced by Fraenkel 2007 (1922), 1–4 and, from a slightly different perspective, Goldberg 1986, 61–6, 69–72.

7. So Jachmann 1934, 625, "Auch Übersetzen ist eine Kunstleistung," and Ludwig 1968, 182, claiming a "fundamental difference between Terence's achievement and that of a creative poet." This attitude was not exclusively German. Thus Shipp 1960, 23, "Terence is essentially a translator, or, if one prefers, an adapter of Greek plays, not a creator." Translation is in this sense much narrower a concept than what Feeney 2016, 45–64 means by "the Roman translation project." The barb at Norwood 1923, 1 was in fact taken without attribution from Bernard Shaw's description of the actor Charles Fulton performing in a revival of Ibsen's A Doll's House, The Saturday Review (London) for 15 May 1897, 539–41 (often reprinted).

8. The shift in modern approaches is traced by Goldberg 2011, 206–10, Fontaine 2014.

9. Jakobi 1996, 114–20 observes the stylistic effect of such figures but is less attuned to the practical pedagogy reflected in the surviving annotation of them.

10. The collections discussed by Jocelyn 1964, 286–9 are catalogued at VSD 44–6. In contrast, Gell. 23.4–7 makes the confrontation of Caecilius with his Menandrean model sound like something of a novelty.

11. For the variorum nature of Donatus' commentary on Vergil, a useful comparandum, see Zetzel 2018, 131–6. The belief that Donatus' knowledge of Terence's Greek models was secondhand goes back to Ihne 1843, 13, promptly endorsed by Ritschl 1845, 326. Wessner 1905, 33 identified Donatus' source of information as the 2nd-century commentary on Terence by Aemilius Asper, of which next to nothing is known (Zetzel 2018, 282).

12. An extreme but telling example of the difference: in 5th-century Gaul, Sidonius Apollinaris could draw upon a copy of Menander's Epitrepontes when discussing Terence's Hecyra with his son, but evidently not Terence's actual model, the Hekyra of Apollodorus (Ep. 4.12.1–2). Apollodorus is known today only through book fragments (32 in PCG, vol. 2, pp. 485–501) and one excerpt on papyrus, the remnant of an anthology (P. Berol. 9772 = Austin 10).

13. Full discussion by Nervegna 2013. See also Fantham 1984, Jones 1993, Marshall and Hawkins 2015, 14–17. The material record provides some indications of ancient priorities: the 2nd-century P. Oxy. 409 + 2655 containing excerpts from Menander's Kolax, was annotated to facilitate delivery by students (Nocchi 2013, 190–3); the Bodmer codex preserving Samia, Dyskolos, and Aspis was a somewhat rough-and-ready copy likely commissioned for private reading. It is dated to the 3rd/4th century c.e., i.e., less than a century earlier than Donatus. Evidence for performance scripts is sparse. Only twenty-three papyri, representing all dramatic genres, are identified in Gammacurta 2006, who notes the popularity of mime and of anthologies over entire plays (265–73). The relation of visual evidence (mosaics and the like) to performance is highly problematic. On that, see Nervegna 2014.

14. The different acting styles of Demetrius and Stratocles are vividly evoked at 11.3.178. Nervegna 2013, 102–4 assumes their performances were Greek, as the reference to them at Juv. 3.99 certainly implies; Fantham 1984, 308 was less confident.

15. Quint. 1.11.1–14 expressly recommends the employment of a comoedus to teach gesture and diction, as does D. Chr. 18.6, noting the practical difficulties of reading a dramatic text. See Nervegna 2013, 230–8, Nocchi 2013, 42–54. Even Cicero owed a professional debt to a comoedus, Roscius. Thus the tradition recorded by Macrob. Sat. 3.14.12, for which see Fantham 2002, 364–7, Nocchi 2013, 21–3.

16. Quint. 10.1.69: Menander . . . diligenter lectus (cf. 10.1.72: alii quoque comici . . . si cum venia legantur). The prologues of Hydria and Georgos are cited at 11.3.91. Cf. 10.1.70–2 on the practice of excerpting speeches, and more generally Goldberg 1987, 360–3.

17. For Roscius, Cic. Q. Rosc. 20. Cf. Varr. R. 2.11.11 for old farmers as a fixture of the stage, clear evidence of continuous performance (if less certainly of Terence's Heauton timorumenos, as often claimed). The shift in emphasis from stage to study in the course of the 1st century b.c.e. traced by Goldberg 2005, 54–62 affected the formation of a Roman literary canon, but how individual authors experienced comedy is often less significant than the simple fact of its influence on their work. Thus the distinction between knowledge derived from books and that derived from performance has little bearing on the analysis of Catullus 75 by Uden 2006 and Catullus 67 by Maynes 2016, or of Cicero's Pro Caelio, rich in allusions to palliata comedy but referring to performance only with reference to mime, by Leigh 2004. Maynes 2016, 284–9 notes the difficulty of disentangling motifs common to palliata and mime.

18. Quint. 9.2.11, 9.3.16, 9.4.140, where lines of Terence are quoted in company with lines of Vergil and Catullus; only 11.3.182 explicitly evokes an actor's delivery. Phaedria's question was a longstanding rhetorical topos. This evidence does not support the claim at Hanses 2015, 29 that Eunuchus was a favorite of Flavian audiences. For Donatus' treatment of Eun. 46, see n. 30 below.

19. These and other examples in Jakobi 1996, 8–14. For Terence's status as "mandatory reading for the budding orator" in earlier rhetorical education, see Müller 2013, 370–4, while for Kragelund 2012, 416, even Donatus' frequent remarks on action and delivery reflect "a well-established reading strategy." Then again, the distinction between reading and performance may in antiquity have been less rigid than in modern practice.

20. Don. ad An. 552, 753; Ph. 184; Eun. 224, 274, 499, 523, 948; Hec. 74, 468; Ad. 852, 907. Theatrical masks can be highly expressive, a dimension of ancient performance rightly stressed by Marshall 2006, 127–9, but whether a comoedus maintained or set aside his mask when coaching students is unknown.

21. Brown 2013, 20–5, 30–1 provides a useful summary of these observations.

22. Cf. the pattern διὰ τί . . . ἔστι δὲ λύσις in the resolution of Homeric difficulties (ζητήματα) discussed by Pfeiffer 1968, 69–71 and the corresponding treatment of Roman quaestiones surveyed by Bonner 1977, 239–40.

23. Don. ad An. 28, appreciating how the exposition is woven into the dramatic action (1) and suggesting that Sosia is gently taken aside to receive his patron's confidence (3). When Sosia is identified as a persona protatica (6), the parallels cited are drawn from other Terentian plays, not from Menander.

24. Jachmann 1934, 613–17; Reith 1964 and Ludwig 1968; Haffter 1953 and Bianco 1962. Even Büchner 1974 cannot entirely free himself from the need to distinguish Terence from his models.

25. Nor does Donatus mention the equally analogous Heau. 679–80, though it may have been noted in the missing commentary on that play. A Greek quotation may of course have dropped out of Donatus' text.

26. Donatus' text is incoherent, and the Greek printed by Wessner is impossible. I quote the line as it appears as fr. 33 K-T, with a helpful note by Thierfelder.

27. The fragment, 34 K-T, is presumed to derive from Menander's Andria, but the attribution could be questioned. And if the line simply means "solitude is creative," the thought itself is rather different. So too ad An. 611, where the line Donatus cites from Menander, ἂν θεὸς θέλῃ/οὐκ ἂν ἀπολοίμην (fr. 39 K-T), bears no direct relation to the Terentian line.

28. Samia is preserved on a range of papyri from approximately the 2nd to 5th centuries c.e. and was among the plays that found their way into the design books used by the illustrators of later antiquity, as at Mytilene, but only one (dubious) fragment is preserved in the grammatical tradition (Phryn. Ecl. p. 187 Lobeck).

29. Cf. Ad. 577–80, where Demea spots an analogous error in Syrus' instructions. Donatus' sed ego puto in the Hecyra note suggests the observation is his own, but it may of course derive from his source (Aemilius Asper?). So Lefèvre 1999, 50–1.

30. Ad Eun. 46.1.2: et apparet multa tacitum cogitasse adulescentem et tandem in haec verba erupisse. (Donatus' composite text illogically puts the conclusion before the reasoning that produced it.) In glossing en at Aen. 4.534 (and reading agam), Servius makes a similar observation (per quam intellegimus eam [Didonem] multa cogitasse et sic prorupisse) and cites Eun. 46 as a parallel (nam haec coniunctio multa eum cogitasse significat). Donatus, his source, has probably allowed one gloss to contaminate the other: modern editors of Vergil prefer the reading ago ("rather πράσσω than δρῶ," Conington ad loc.).

31. The attention to delivery may often reflect an interest in rhetorical rather than specifically stage performance. So Nervegna 2013, 245–6, Demetriou 2015, 184–93. Kragelund 2012, 417–20 understands the more explicitly theatrical notes on Ad. 511.2 (stage structure) and An. 716.1 (use of female actors) as references to 4th-century productions of Terence, but there, too, other interpretations are possible.

32. Noted by Donatus ad Ad. 259, disentangling the syntax of the rusticus adulescens . . . titubans ac paene balbutiens.

33. His pupil Jerome (In Eccles. 1) recalls what was probably a viva voce comment on Eun. 41 (nullumst iam dictum quod non dictum sit prius), pereant qui ante nos nostra dixerunt.

34. Thus ad An. 28.1: ut sine fastidio longus sermo sit ac senilis oratio; ad Eun. 539.3: bene inventa persona est [i.e., Antipho], cui narret Chaerea, ne unus diu loquatur, ut apud Menandrum. Extended monologues describing off-stage action are well attested in Menander, e.g., Dysk. 522–45, 666–90; Sa. 206–82, 908–32; also apparently Epitr. 583–602, 908–32. They are much less common in Terence, though Hec. 361–414, 816–40 are noteworthy examples. (Don. ad Hec. 825.2 says that the monologue there is Terence's innovation, nam in Graeca haec aguntur, non narrantur, but the full ramifications of this change are much debated. References at Goldberg 2013, 193, to which add Ludwig 1964, 156–8.)

35. So ad An. 891: mira gravitate sensus elatus est; nec de Menandro, sed proprium Terentii. He found the play's final scene, where the integration of Charinus was clearly Terence's creation, to be executed mira arte (ad An. 965).

36. The note ad An. 301 is deceptively explicit: has personas Terentius addidit fabulae,-nam non sunt apud Menandrum. So too ad 977 (on the function of a second lover): et id extra praescriptum Menandri. Critics agree on the awkwardness of this addition, but its source is a longstanding crux of Terentian scholarship. See Lefèvre 2008, 58–63, a thorough (and cautious) discussion.

37. The suggestion goes back to Fraenkel 1968, 235–42.

38. Greek terms as they appear in Donatus' text are often problematic. This one is a conjecture derived from Lindenborg's 17th-century edition, not from any extant MS., and the comment may well be missing at least a verb to complete the sense. Cf. the structure of the note on Eun. 588 (on what may be a Terentian elaboration), φιλοσοφικῶς Terentius demonstravit. The ancient treatise De comoedia (no. 1 Koster) praises Epicharmos for being quotable, inventive, and skillful (γνωμικὸς καὶ εὑρετικός καὶ φιλότεχνος, 16) and notes the high regard Pherekrates enjoyed for his introduction of new themes and inventive plots (πράγματα δὲ εἰσηγούμενος καινὰ ηὐδοκίμει γενόμενος εὑρετικός μύθων, 31).

39. A preliminary version of this paper was presented at the annual meeting of the Classical Association in 2015. I am grateful to Ruth Caston for organizing the panel on the reception of Roman comedy and to all who joined in the discussion on that occasion. Its subsequent revision owes much to the referees and editors of AJP.

Additional Information

Print ISSN
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.