Marcus Deufert (henceforward D.) has given us with his Habilitationsschrift a fully-fledged and informative study of the history of the text of Plautus in antiquity, closely sifting the evidence for the reception and transmission of these plays throughout their various stages until the establishment of the direct tradition. D.’s study engages in a contest with two landmarks of modern Plautus criticism, namely W.M. Lindsay’s The ancient editions of Plautus (Oxford, 1904) and F. Leo’s Plautinische Forschungen (Berlin, 1895, 1912 2, with the important review of O. Seyffert, BPhW 16, 1896, 234-41, 251-5, 264-70), against which the progress of D.’s book can be assessed. Other important discussions of this topic were offered by G. Pasquali, Storia della tradizione e critica del testo (Florence, 1952), 331-54, and by R.J. Tarrant, in Texts and Transmission (Oxford, 1983), 302-7.
Chapter I deals with the first phase of transmission of the texts, from the first performances to the establishment of the first complete edition in the last decades of the II century BCE. D. characterizes this period as a phase of ‘Mouvance’, a time of fluid flowing of variants into the texts. Plautus did not write books but scripts for actors. The comedies were sold to the impresarios, who then offered the plays to the aediles (for a recent discussion of the evidence see now P.G.McC. Brown, Actors and actor-managers at Rome in the time of Plautus and Terence, in P. Easterling-E. Hall, Greek and Roman actors, Cambridge, 2002, esp. 229-32). Copies of the scripts, however, remained in the hands of the companies and were altered in the course of later performances. D. discusses the possible appearance of these scripts on the basis of the evidence provided by Greek dramatic papyri of comparable date, their putative colometric lay-out, and orthography.
D. further wonders how much subsequent playwrights such as Terence knew of Plautus’ scripts and when actual revivals took place. D. sides with Ritschl ( Parerga zu Plautus und Terenz, Leipzig, 1845, 198) against Leo ( Geschichte der römischen Literatur, Berlin, 1913, 212-6) in arguing that ‘old plays’ of Plautus started being brought on stage only after the death of Terence. D.’s main argument for supporting the post-Terentian dating for the Casina prologue is that the production of plays during Terence’s life-time would be irreconcilable with this playwright’s determination to defend himself against charges of plagiarism in his prologues (most notably in Eun.). Terence repeatedly asserts that he has written only ‘new plays’, that is plays previously untranslated into Latin (esp. Heaut. 4). Why press this point so much if somebody else was reviving ‘old’ ones quite without embarrassment? There is a difference, however, between presenting as one’s own a patchwork of old scenes written by some predecessor (the accusation against which Terence reacts) and the revival of an old play by a director with no claim on authorship.
Deufert accepts Zwierlein’s conclusions (in the series Zur Kritik und Exegese des Plautus
The cumulative evidence that scholars since Ritschl have brought to bear on the question of interpolations is very strong, but I am worried by D.’s acceptance of previous scholars’ excisions, and in particular of Zwierlein’s highly controversial ‘single-reviser’ thesis, as a platform for further deductions, as in the case of the relationship between Terence and Plautus or his interpolators. When D. does engage in textual analyses bearing on the authenticity of specific passages, his conclusions are not always persuasive. For example, D.’s linguistic objection against the authenticity of Trin. 16-7 sed de argumento ne exspectetis fabulae:/ senes qui huc uenient, ei rem uobis aperient (traditionally thought to be imitated by Ter. Ad. 22-3 dehinc ne exspectetis argumentum fabulae:/ senes qui primi uenient, ei partem aperient) (28, n. 60), is inadequate. The occurrence of exspecto with de and the ablative, as opposed to the accusative, is no argument to support the interpretation of the intrusion of a later linguistic usage. De plus ablative is not a later construction replacing the accusative after exspecto. The accusative is simply left unexpressed. The phrase means ‘do not expect anything more of the plot’, or ‘concerning the plot’.
Chapter II (44-62) is devoted to the lay out and organization principles of the (in D.’s words) first complete edition of Plautus. Important features were the inclusiveness of all variant readings found in the actors’ scripts (see also infra), and the lay out of the cantica according to the colometric principles established by Alexandrian scholars of drama.
Chapter III traces the spread of the hypothetical Erstausgabe in quotations in poets (Lucilius), critics, and above all the works of antiquarians active in the last quarter of the second century BCE or later, and mainly transmitted by later sources such as Varro’s De lingua Latina and Verrius Flaccus’ De uerborum significatu. Of particular interest is D.’s discussion of the criteria used by Republican scholars to ascertain the authenticity of specific plays, for which he is able to point to significant parallels in the doctrines of Krates of Mallos and in the later aesthetic writings of Philodemus (96-103). The didascalic information prefixed to Pseud. and Stich. in A, the Milan palimpsest, is studied in parallel with the similar information transmitted by Terence’s MSS in the context of the Greek tradition of ‘production notices’ or didascaliae (88-96). D. believes that the presence of this information should be ascribed to influence of Varro’s antiquarian essays, in particular De actis scaenicis; its inclusion in an edition of Plautus is therefore later than the first century BCE and should be ascribed to the Hadrianic editor (96: see infra, on chapter VI). D. also argues that all plays had production notices in a form similar to that of Terence’s plays. From comparison with several Greek dramatic hypotheseis and the Terence stage-notices, D. argues that the final line of the Stichus notice in A, blank now that the original red ink has vanished, contained the number indicating the order of composition of the play in Plautus’ total dramatic production. Likewise, D. (89, n. 182) suggests interpreting the fragmentary second line of the didascalia prefixed to Pseudolus, ACC, as [F]
D. also discusses the evidence for glossaries of Plautus and commentaries (115-20). D. attributes to Schröter ( Studien zur varronischen Etymologie. Erster Teil, Köln, 1959, 76-82) the hypothesis that Varro drew on a line-by-line commentary on Plautus’ Pseudolus in his note in L.L. 7.81, supported, in his view, by the more elaborate introduction formula used for the quote, revealing some knowledge of the onstage action (118; cf. L.L. 7.81 Apud Plautum: ut transuersus, non prouersus cedit quasi cancer solet. prouersus dicitur ab eo qui in id quod est ante, est uersus, et ideo qui exit in uestibulum, quod est ante domum, prodire et procedere; quod cum leno non faceret, sed secundum parietem transuersus iret, dixit ‘ut transuersus cedit quasi cancer, non prouersus ut homo’). In a context in which Varro draws consistently on Servius Claudius, it is strange that Varro, if an extended commentary was available, should draw on it for only this example. In my view, this particular passage is still compatible with derivation from the same syngramma, or specialized lexicographical essay, from which the rest of Varro’s Claudian notes seems to derive. Comment on stage action is not enough alone to suggest derivation from a commentary in our sense, and some knowledge of the action is presupposed also by Varro’s later note on Cistellaria.
Chapter IV (139-75) discusses the evidence provided by the Republican indirect tradition as regards the state of the tradition at this stage. D. is indebted here to some important contributions by the late H.D. Jocelyn, namely his ‘Studies in the Indirect Tradition of Plautus’ Pseudolus (I), in Filologia e forme letterarie. Studi offerti a Francesco della Corte (Urbino, 1987), vol. 2, 57-72; (II), in Studi di filologia classica offerti a Giusto Monaco (Palermo, 1984), vol. II, 569-80; a full discussion can also be found in Lindsay, Ancient editions, 2-23. D.’s main aim is to refute Leo’s view (as expressed in Plautinische Forschungen, esp. 54-7, 61-2) that the textual tradition of Plautus not only suffered from interpolation in its very early stages but, more significantly, underwent massive corruption in the intervening period, until Probus put an end to this process at the end of the first century CE. In Leo’s reconstruction, the MSS collected by Probus transmitted a text that was already severely corrupt after two or three generations of total disregard and neglect, but Probus, though aware of the corrupt state of his MSS, refused to emend more. This is the only text, Leo argued, that the modern editor of Plautus could hope to reconstruct by the means of Lachmannian recensio. The hope to recapture the ipsa verba of Plautus could only be the work of an almost desperate emendatio.
Varro’s works, mainly De lingua Latina, transmit some 50 quotations of Plautus’ extant plays. D. does not discuss all of Varro’s Plautine quotes, for which he refers to A. Klotz, ‘Die Plautuscitate Varros’, Philologus (1944), 21, for a number of cases in which Varro corrects trivial errors of the Palatine family. D. lists 11 cases in which Varro exhibits a text clearly at fault, either through errors of memory or corruptions of Varro’s MSS. D. stresses that Varro does not exhibit significant coincidences in error with the direct tradition (the agreement of A and P, the archetype of the Palatine family of mss. which proves that this tradition rests on solid ground. He then discusses 5 cases in which Varro allows editors to restore the correct text (but only P is available for all but one).1
Verrius Flaccus’ lexicographical collection is extant only in the epitome of Sextus Pompeius Festus (II-III century CE, supplemented from the later work of Paulus Diaconus (of the Carolingian age). Verrius-Festus-Paulus (henceforward Festus) preserves some 100 quotes from Plautus. D. discusses 10 cases in which Festus’ text is clearly at fault, 3 cases of trivial, and therefore stemmatically irrelevant, errors common to Festus and P (A is absent),2 and 7 more cases in which Festus transmits a text superior to that of our MSS (170-1 and nn. 196-7). To these he adds discussions of 6 more cases in which Festus preserves what he regards as variants stemming from the grammatical traditions, not clearly inferior but likely to have originated as glossae. For the remaining passages, including cases in which Festus allows to correct the MSS, D. refers to E. Leidolph, De Festi et Pauli locis Plautinis, Commentationes philologae Ienenses 2, 1883, 238 ff.
D. emphasizes the fact that Varro and Verrius-Festus share no significant errors with AP. This he rightly sees as an indication of the stability and reliability of our tradition and as a refutation of Leo’s claim for a massive corruption at pre-archetypal stage (although Leo meant by that the Augustan and post-Augustan period). The fact that Varro and Verrius allow editors to correct the direct tradition at significant points, sometimes where no corruption was apparent, might seem to go some way towards supporting Leo’s view, but it is impossible to establish when these errors occurred. For practical editorial purposes, however, the discussion seems to me of very little consequence.
Chapter V (176-99) looks at the evidence for the knowledge of Plautus in the first and early second century CE, in poets and grammarians, notably in the fragments of L. Annaeus Cornutus, and in the letters of the younger Pliny. The chapter includes a full discussion of Probus’ philological work on the text of Plautus (183-92). In line with recent studies on this subject, and above all indebted to Jocelyn’s important study in CQ 34 (1984), 464-72, D. argues that no evidence supports the claim, decisively championed by Leo, that Probus produced an edition of Plautus. D. downplays the significance of the romantic story narrated in Suet. De gramm. 24, according to which Probus collected MSS of early Roman poetry still available in the provinces, and he rightly refuses to read it as a proof of his editorial activity.
Chapter VI (200-37) studies the establishment of the ‘Selection’ in the age of Hadrian, or perhaps slightly later. This edition is seen as a product of the greater demand for legible Plautus texts in an age which started to reevaluate archaic literature (with Fronto and Apuleius as the standard-bearers of the archaizing movement). There is no direct evidence concerning the date of the edition, the first to include only the twenty-one plays universally accepted as authentic; D. interprets the renewed interest in questions of authenticity revealed by Gellius, N.A. 3.3 as a reaction against the recent Selection (213-4). The Selection was the first edition to include, in his view, indications of speakers and scene-headings (217-224). Scene-headings marked new entrances or changes in the delivery or in the manner of the interaction of the characters simultaneously present on stage. Headings evolved from an original form marking speakers continuously present with EIDEM to a pattern in which the names of the same characters are spelled out in full.3 The editor responsible for these editorial features also put together production notices, prefixed to all plays. The similarity of scene-headings and production-notices to those of the putative contemporary edition of Terence suggests that they may have originated from the same scholarly milieu (224-6). D. then goes on to discuss the evidence of the non-acrostical argumenta, which he sees, following R. Opitz (De argumentorum metricorum latinorum arte et origine, Leipziger Studien zur classischen Philologie 6, 1883, 195-316), as descended from the scene-headings. This is shown by verbal parallels linking the argumenta to the scene-headings, especially euphemistic expressions for rape and sexual violation, expressed in Terentian language. The argumenta are seen as slightly later than the scene-headings, although composed for the same edition.
In the discussion of the editorial features of the selection, D. is greatly indebted to Bader’s important 1970 Tübingen dissertation Szenentitel und Szeneneinteilung bei Plautus. Bader was the first to suggest that a chronological clue to the age of the insertion of scene-headings is yielded by the appearance of the word lorarii, designating the slaves administering punishment (for example in the headings at Merc. 272, Pseud. 133). This is a word not found in Plautus but occurring in Gellius, NA. 10.3.19, a passage in which the attendants of Roman magistrates employed for coercive purposes are compared to the comic characters in scaenicis fabulis qui dicebantur lorarii. As a consequence of this linguistic argument, Bader is credited with the argument that scene-headings go back to the Hadrianic edition (221, 223 n. 121).
Bader was in fact more cautious as regards the textual appearance of the second century edition, and about the date of the migration of scene-headings from the margins, where they originally stood, into the text (cf. 108, 151). Bader thought in fact that the II-century edition contained only a first kernel of marginal and interlinear Sprecherbezeichnungen, on the model of contemporary and later Greek books of drama such as Pap. Bodmer of Menander’s Dyskolos. According to Bader, a fully deployed system of scene-headings set in much later. Questa, for example, has championed the thesis that scene-headings developed at a much later stage, in the IV century codex edition (now in Numeri innumeri, Roma, 1984, 161-92).4 I am not entirely persuaded by the argument that the word lorarius was a pseudo-archaic coinage of the Hadrianic editor (223). In Plautus and in later authors, -arius formations are normal to indicate craftsmen and even slaves overseeing to/discharging specific household functions, but -arius formations are attested throughout the whole history of Latin and in the Romance languages, and they are not conspicuously archaic formations. Finally, the word lorarius exists in epigraphic evidence, even if, as D. is right to say, it seems to denote mainly harness-makers. There are, however, a few borderline cases, and Matteo Della Corte suggested interpreting lorarius in CIL IV 7989 as ‘qui beluas loris incitat’.5
Chapter VII deals with the reception of the Selection in this new form. The establishment of the selection-edition triggered a new wave of scholarly interest in Plautus, as documented by the commentary of the elusive Sisenna (fragments mainly in Rufinus, Comm. in metra Terentiana, GLK VI, 560.28-561.1-10). This must have been the time when plays not included in the selection started disappearing, with the exception of the fragments of the Caecus quoted by Iulius Romanus (apud Char.). Contrary to established opinion, D. attempts to argue, in my view successfully, that Plautus was one of the classics read by advanced students, and he collects the relevant evidence, focusing on Donatus, Hieronymus, and Servius.
An important section of this chapter deals with some late-antique attempts at recreating the early dramatic meters, most notably Ausonius’ senarii of the Ludus septem sapientum, and the acrostical argumenta. D. argues that Ausonius exhibits a considerable metrical expertise, and he makes a good case for the defence of transmitted cases of hiatus at the caesura in the work, usually emended away by normalizing editors (277-83). I am less convinced by D’s discovery of loci Jacobsohniani in Ausonius. D. also analyzes the acrostical arguments, extant only for the Palatine MSS. He argues, following Opitz, that they are later than the non-acrostical, as proved, in his view, by some linguistic borrowings from the ‘Terentian’ language of the non-acrostical argumenta and by their contorted narrative style, which he takes as a further sign of derivation from the non-acrostical summaries. Priority questions are notoriously difficult to decide on the basis of internal evidence, and I think the point is not so certain as D. avers. Seyffert ( JAW 47, 1886, 22) thought that the acrostical argumenta were much older, on the basis of their prosodical and metrical virtuosity, especially apparent from comparison with the non-acrostical summaries, and even with Ausonius’ technique. D.’s analysis of Ausonius’ Ludus goes some way towards suggesting that even in the third or fourth century, on the eve of the establishment of the two late-antique recensions, there were people capable of reproducing the metrical and prosodical features of Plautus’ verse, an element which D. will put to good use in his subsequent chapter.
Chapter VIII (293-339) gets to grips with the question of the archetype of the direct tradition. Following Questa, D. maintains that the last common ancestor of the two families A and P was already in codex form, mainly on the strength of the argument that the two families break long verses of more than 45-50 letters in the same way. He then gives a typology of errors common to AP.6 Study of these common errors shows that the archetype already incorporated explanatory unmetrical glosses and was defaced by numerous other trivial corruptions. This he regards as an element in favour of Pasquali’s suggestion of an ‘edizione commerciale’, rushed into the market by a profit-minded book-dealer. D. also counters Leo’s view that the errors go back to the ‘Hadrianic’ edition rather than to the reconstructed IV century codex archetype.
After a full discussion of Nonius’ stemmatic position in the tradition of Plautus (D. argues, against Pasquali, that Nonius’ MSS of Plautus were all descendants of the Hadrianic edition, with contamination via grammatical sources accounting for some better readings), D. goes on to discuss the thorny question of the presence of duplicate passages in the two late antique editions that gave rise, in turn, to A and P.
Lindsay believed that the great variety of readings found in the two families of Plautine MSS went back to two different Republican editions, one more faithful to Plautus’ original text, the other a ‘Revival’ text reflecting later performances. Lindsay accordingly downplayed the extent of common errors in A and P, denying in essence the existence of an archetype ( Ancient editions, 104-18). Very few subsequent scholars have followed Lindsay’s denial of an archetype, but the problem of double versions present in both branches, or omitted by one or the other of them (though usually P is more inclusive than
D. adheres to the now-prevalent interpretation of the problem, namely that early interpolations were included ‘conditionally’ in the second-century BCE Erstausgabe, with critical marks highlighting what the editor suspected were spurious additions. When the two families took their separate ways, the editors responsible for A and P acted differently with regard to these annotations, with P mainly including, and A omitting. The classic examples of long passages present in only one recension include Most. 940-5, Bacc. 540-51, and Capt. 1016-22.
It is difficult to object to this picture in the face of the evidence for a late-antique archetype. One wonders, however, how these critical annotations survived so long down the centuries (D. addresses the question at 58-9, where he suggests that commentaries helped to ensure the survival of these notes and clarify their meaning) and why, if they did reach the age of A and P, they vanished altogether thereafter. It is odd that the late-antique editors, especially the editor of A, felt it incumbent upon them to delete so much that must have appeared, even to them, if not authentic, at least of considerable antiquity, unless this is seen as the final part of an ongoing process of pruning that had started as early as the first edition or the so-called ‘Hadrianic’ edition.
D. attempts to explain away the rest of the not clearly erroneous divergent readings of the two families with the hypothesis of late-antique interpolation, for which he is able to draw on much comparative evidence from research on the traditions of other classical authors, most notably Zwierlein’s on Seneca’s tragedies, his own earlier book on Lucretius, Zwierlein’s more recent studies on the text of Vergil, and other studies devoted to Martial and Juvenal (329-39). D. himself anticipates the objection that the selfsame editors responsible for these virtuoso rewritings of Plautine passages did not bother to emend the trivial nonsense they found in their MSS, reminding us that the competence and consistency of nineteenth-century professional editors is not to be expected of these Late Antique scholars. D. makes a good case for Truc. 246 ff. and Epid. 164-5 as the product of Late Antique editorial tampering with a difficult text, but other examples he discusses are less straightforward (so e.g. Stich. 631 and Pers. 597). My impression is that it is difficult to draw a firm line between Late Antique readjustments of difficult or puzzling passages and ‘genuine’ interpolations going back to the Revival period, and I find that D.’s assertiveness is not always appropriate.
The last chapter (IX, 340-81) deals with hiatus in the corpus of Plautus. D. provides an interesting survey of nineteenth-century scholars’ varying approach to hiatus in the corpus and particularly of Ritschl’s changing attitude in his successive editions (358-9). Ritschl started from an extreme reluctance to accept hiatus in Plautus, but his attitude became more hesitant with the inspection he made of the Ambrosian Palimpsest in 1836/37, which he saw as confirming several cases of suspected hiatuses of P. In the sixties, Ritschl came to the conclusion that several instances of hiatus could be done away by restoring features of archaic orthography which, current in Plautus’ time, had become obsolete at the time of the Plautine revivals, such as final -d in the ablative of first and second declension nouns and final -s in the nominative plural of first and second declension nouns (our MSS preserve them only in pronoun forms such as, respectively, ted and his). To this Deufert rightly objects that the selfsame revisers who could replace obsolete forms were well able to do away with hiatus, and if they were put off by obsolete forms, why could they tolerate hiatus?
D.’s principal polemical idol is Leo’s view of hiatus as the result of extensive corruption and also of the growing dislike for synaloephe in the Augustan period. Readers and editors aiming at removing the odd cases of synaloephe in Plautus introduced or tolerated hiatus (362-77). D.’s main contention is that hiatus is a feature of Plautus, that can be traced to the origins of the Latin native meters, most notably the trochaic septenarius. D. convincingly shows that hiatus is a feature well established not only in Plautus but also in the fragments of early drama, especially comedy, up to Terence. Of particular interest is D.’s observation of Ribbeck’s waverings in his treatment and acceptance of hiatus in the course of his successive three editions of the fragments of early drama ( Scaenicorum Romanorum poesis fragmenta, 1852-5, 1871-3, 1897-8), which went in parallel with the changing approach of Ritschl and his pupils in the same years (396).
This is an informed, accurate, and intelligent discussion of a question that has occupied some of the most brilliant Latinists of modern times. D.’s ability to organize and subject to critical analysis the enormous evidence bearing on the problem, primary and secondary, commands vast admiration, and his textual discussions, some involving sophisticated metrical expertise, are always technically adequate and stimulating, even when one ultimately disagrees with them. It is in the nature of all such wide-reaching projects to depend largely on previous scholars’ results, and it would have been unwise to neglect the earlier literature on the subject. However, D.’s book successfully complements and completes all previous general accounts of the Plautine question, notably Leo’s and Lindsay’s, updating them in several important respects.
Notes
1. D. makes good cases for Varro’s text vis-à-vis a corrupt or inferior reading of the direct tradition at 145-6, especially on Aul. 446, suggesting we write mihi uasa iubes, hic pipulo te differam ante aedis in place of P’s mihi uasa iubes, pipulo hic differam ante aedis, solving a difficult metrical problem; Most. 245; Cist. 8-9 (building up on Schoell), and Mil. 24. However, the use of adverbial insanum modifying bene in this last passage seems unobjectionable to me.
2. D. argues that Verrius-Festus never side in significant error with either of the two branches of the MS tradition, but I disagree with D.’s discussion of Cas. 523 (169-70), transmitted by Festus and P in the unmetrical form sed facito dum merula per uersus quod cantat colas (the most recent editor, Questa, inserts Lindsay’s tu before colas). D. proposes his own conjecture pueris uersus in place of per uersus, an emendation which allows him to argue that the corruption occurred independently in P and in Festus. The argument depends on the persuasiveness of the emendation, and I am not convinced by it. The merula is not imparting lessons particularly suited for pueri, and D.’s own parallel from Apul. Flor. 17 is not decisive. The error is probably very ancient (possibly already in the Hadrianic edition). Festus puts up with another unmetrical reading at Mil. 94).
3. It is not true, however, that scene-headings in Seneca’s MSS can be seen to have evolved along lines similar to Plautus’, nor does Zwierlein say that (217, n. 97).
4. OLD s.v. scaena does not comprise a separate subheading for our technical meaning of ‘scene’, or minimal narrative unit of a dramatic text. This meaning seems to occur for the first time in grammatical writers, such as Rufinus, GLK VI, 561.9 haec scaena anapaestico metro est; cf. also Don. Ad Ter. Hec. Praef. 3.6 (= fr. 307 Funaioli) docet Varro in hac fabula neque in aliis esse mirandum quod actus impares scaenarum paginarumque sint numero. One wonders if the extended meaning of the word came about as a result of the, specifically Roman, editorial appearance of books of drama, in which different units of dialogue were separated by scene-headings within the text. The only earlier possible instance of this meaning is the metaphorical use in Petr. Sat. 33.5 conuertit ad hanc scaenam Trimalchio uultus, although the sense could be more generally ‘charade’, ‘theatricals’, ‘comic act’.
5. It is also not quite right to say that the absence of lorarii in A is caused by the disappearance of role nouns in the second line, originally in red ink, of the scene-headings. As Bader (15) showed, there are different typologies of scene-heading in the Palimpsest, and it is conceivable that lorarii could have appeared on the first line, although it remains to be seen whether any of the headings where the word could have stood is still legible.
6. I have checked D.’s list of common AP-errors (for Pseudolus alone) using the apparatuses of Leo and Lindsay. I have observed the following omissions on Leo (the first is the correct text): 138-9 (false verse division), 185 (factust Lambinus : factum est
7. Particularly the evidence collected for Senecan tragedy by O. Zwierlein, Prolegomena zu einer kritischen Ausgabe der Tragödien Senecas, Wiesbaden, 1983, 25, 38, and by W. Schmid for Martial ( Ausgewählte philologische Schriften, Berlin-New York, 1984, 400-44).