BMCR 2001.11.17

Prudentiana I, Critica

, Prudentiana. Munich and Leipzig: Saur, 2000-2003. 3 volumes, plates : illustrations, facsimiles ; 25 cm. ISBN 3598774206. DM 238.

Gnilka’s interest in Prudentius can be traced back to his PhD on the Psychomachia (Diss. Bonn 1962, published in 1963). The many articles and books that followed showed a continuing concern for, but no restriction to, the Spanish poet. Most notable have been his two volumes within the series Chresis : Die Methode der Kirchenväter im Umgang mit der antiken Kultur.1

His Prudentiana is a monumental enterprise: it summarises his long-lasting efforts to illuminate the text and the significance of Prudentius. The first volume (subtitled Critica), which I am discussing here, aims at establishing the original text by removing hundreds of interpolations. In 690 pages, he strings together meticulous discussions of textual criticism: this is hard-core philology, but it is classical scholarship at its height! Most of it is clearly written (given the specialist subject this is a merit worth mentioning), full of coherent arguments and in many respects highly instructive, showing the author’s impressively wide-ranging familiarity with all disciplines of classical research.

The book comprises twelve previously published studies arranged in chronological order (p. 1-290), and seven new ones (p. 291-647); they are followed by 43 pages of miscellaneous addenda (p. 648-90), an index of abbreviations used in the latter parts (p. 691-96), a three-part index (p. 697-756), and twenty photographs of Roman statues and early medieval manuscript pages (tables ι with references on p. 757-62). The short preface announces a second volume: Exegetica, which is forthcoming this year (according to the online advertisement of Saur).

One of the principal problems posed by the text of Prudentius is the amount of apparent or alleged interpolation: an oeuvre which is characterised by its prolixitas and occasionally by its obscuritas, may have induced many enthusiastic readers to add a verse or an entire stanza, as becomes evident when the manuscripts are compared. The effect of this phenomenon was passionate interpolation-hunting by former editors, who valued their own taste over the sometimes free flowing style of the poet himself and disregarded the state of transmission, particularly the authority of the two 6th-century codices (cod. Putaneus and cod. β Ambrosianus).

This, in any case, is the view of the more conservative 20th-century editors (Bergman, CSEL 61, 1926; Lavarenne, Budé, 1st ed. 1943-51; Cunningham, CCSL 126, 1966), who generally object to athetesis without strong support by the manuscripts. Gnilka, however, is a declared adherent of Jachmann’s theory that most changes to the text date to (late-) antique redactions (both are fond of the term ‘Diaskeuase’ and ‘Diaskeuast’ respectively), thus challenging the sacrosanct authority of codd. A and B.2 It was probably his suspicion of Psych. 456ex.-458in. already put forward in an appendix of his PhD dissertation (1962/63, p. 129-33, not included in Prudentiana), that was the foundation of his current theses.

Hence I approached the Prudentiana with a great deal of scepticism. In the following I will discuss the individual parts; it is clear that I cannot comment on all of them to the same extent.

I. Zwei Textprobleme bei Prudentius (first published in 1965, p. 1-15): in a very short preface (p.1f.) Gnilka distances himself from a ‘unbesonnene Interpolationsjagd’ as well as from the theory of authorial variants, a subject he frequently resumes in this volume (cf. index, p. 746 and my discussion at the end of part II.).

The first verses under discussion form part of an excursus on the constitutional history of Rome (C. Symm. 2.416ff.): the monarchy had been replaced by the rule of aristocrats, who soon afterwards shared the power with the plebs; v. 423-27 deal with the decemviratus and the restitution of the consulate; the passage is closed by a single verse mentioning the triumvir and alluding to the civil wars (v. 428). Gnilka rightly points out that the v. 423-27 not only reveal historical ignorance but apparently disturb the structure of the entire excursus. Furthermore, the dubious passage is not backed by cod. A (which entirely lacks C. Symm.), and is missing or located at different positions in other early manuscripts. Next it is explained how the interpolator, not sensitive to Prudentius’ train of thought, was induced by Liv. 3.33,36 to add the decemviratus as a third step in the development of the Roman constitution. The only question that remains after reading this chapter is why no one before Gnilka ridded the C. Symm. of this manifest and harmful falsification.

In the next passage (p. 8-15) an interpolation in Ham. (v. 887-91) is uncovered. The case is more difficult, as it is not marked by differing manuscript readings, but the obvious disturbance of the original concept had already been recognised by others (e.g. Lavarenne). An aspect of divine justice is explained: the souls in heaven and in hell can see one another: this necessitates explaining the sight of the souls (v. 863ff.): they outdo the corporeal eyes already in life, but are completely unrestricted after their separation from the bodies. 887-91 interrupt the argument with an excursus (within the excursus) on the invariability of the shape and colour of the souls. The quidquid id est (v. 891) is somehow opaque, and its only function seems to be to fill the last verse. Contrary to most other examples treated by Gnilka, apart from the latter verse, he cannot draw on stylistic or metrical shortcomings. His conclusions remain attractive however, though not compelling.

II. Beobachtungen zum Claudiantext (first published in 1975, 16-67): in a general preface Gnilka gives an introduction to his editorial practices on the text of Claudian, condemning unsound atheteses as well as blind confidence in the manuscript.

The first example (p. 18-27) is taken from the praise of Stilico’s endeavors to protect the empire (Stil. 1.302-5): efficiens patranda manu, dictare paratus / quae scriptis peragenda forent. non (var.: quae) bracchia centum / [quis Briareus aliis numero crescente lacertis] / tot simul obiectis possent (var.: posset) confligere rebus. According to Gnilka, v. 304 is missing only in one ms., but extant in the five so-called ‘Haupthandschriften’. The edition of J.B. Hall (Leipzig 1985), which does not even take into account Gnilka’s suggestion, notes further mss., among others the Bruxellensis 9974-6 saec. xiii.3 Gnilka convincingly points out v. 304 as an interpolation, taking his cue from the mention of the ‘hundred arms’.

The interpretation is buttressed by the incorrectness of the mythologem, which is a bizarre mixture of the Hecatoncheires and the hundred-headed Hydra, in fact unfeasible for a poet like Claudian, but characteristic of the superficial knowledge of a poetaster (a term frequently used by Gnilka). He may have had Verg. Aen. 6.287 in mind: et centumgeminus Briareus ac belua Lernae. Gnilka further supposes that the interpolator did not want to add a verse, but aimed at replacing v. 303. With both verses being transmitted, later redactors tried to balance out the syntactical unevenness by the changes to v. 303/305 indicated above. Gnilka rejects the alternative interpretation that these adaptations are due to the same redactor, which is not compatible with the theory of a supplementary interpolation: non bracchia centum, / non Briareus… would have been much more plausible in this case. To my surprise he gives up the latter interpretation in the addenda (p.649f.), assigning all the changes to one hand. He thus appears more consistent with his general theory (for which see below), but he overlooks a contradiction to p. 20 n. 7.4

In the subsequent chapter (p. 27-30) Gnilka sets out to eliminate Paneg. Prob. et Olybr. 201-4, which illustrate the preeminence of the father (Petronius Probus) and the mother (Proba) of the consuls of A.D. 395: ceu sibi certantes, sexus quid possit uterque, / hunc legere torum. taceat Nereida nuptam / Pelion. o duplici fecundam consule matrem / felicemque uterum, qui nomina parturit annis. A strong argument against the authenticity of the verses is the fact that they are not attested before the ed. Isengriniana (Basel 1534). Gnilka criticises them as tasteless and inappropriate, a verdict I do not accept. On the contrary, I find it difficult to assign such remarkable verses to any poetaster. He also fails to give any reason for such an addition, and I wonder whether the unidentified old ms. which the aforesaid edition is based on could not preserve the original reading, deleted by Christian redactors in the mainstream of transmission.

In the next chapter (p. 30-35), Get. 128 is convincingly identified as a γνομικὸς στίχος of low quality, a clumsy pastiche of the context. I miss an explicit statement on the textual transmission, which does not, however, detract from the conclusion. The same applies to chapter four which deals with ‘Binneninterpolationen’, first in general (p. 35-37), then with Get. 477ex. / 478in. in particular, of which he masterfully demonstrates the manifold insufficiencies (train of thought, syntax, style). Reference to v. 78-86 makes clear the motivation of the addition.

Chapter five (p. 43-60) deals with the Deprecatio in Alethium quaestorem (carm. min. 23): Claudian admits he criticised the poetry of Alethius (v. 9 versiculos, fateor, non cauta voce notavi), which he ironically calls a crime (v. 10). The verses of Orpheus, Virgil, and Homer, however, are freely commented on (v. 11-14); in order to be pardoned Claudian applauds ostentatiously (v. 17-20). The last two sections are separated by a distich which is assessed as highly inept and therefore excluded by Gnilka: sed non Vergilius, sed non accusat Homerus: / neuter enim quaestor, pauper uterque fuit.

In condemning this passage, Gnilka (p. 48-50) notes the weakness of neuter : a disyllabic reading was considered a barbarism in Claudian’s day and thus undermined the point (cf. 5.389 Keil). The writer obviously lacked sensitivity in this regard since he could have avoided the offence simply by exchanging the word order (quaestor enim neuter). Also the contrast of quaestor and dives is inept (p. 51-55), especially since Virgil was by no means poor. The critic also successfully explains the error with reference to Ad Olybrium 23. But there still remain many problems, which Gnilka does not address: To identify Alethius quaestor with the grammaticus of v. 6 (v. 5f. sic non Tartareo Furiarum verbere pulsus / irati relegam carmina grammatici) renders the entire poem unbalanced, be it with or without v. 15f. Thus Gnilka’s contribution is important, but not the last word (in more detail, I deal with the problem in Prosopon 12 (

What struck me most in this chapter, however, is Gnilka’s harsh criticism of E. Baehrens (Bursians Jahresberichte 18, 1879, 147 on Jeep’s ed. of Claudian, 1876/79), ‘der die Lauge seiner aetzenden Kritik ueber den Jeepschen Claudian ergoss’ (p. 48); this might mirror Baehrens’ style (which I have not checked), but it is not the only example of the little respect Gnilka has for the ‘ignorant’ among his colleagues. However, usually his polemics are harmless and can sometimes even be enjoyable.5

The general conclusion (p. 60-63) is obvious: significant changes to Claudian’s text go back to late antiquity. This parallels the case of Prudentius and justifies the ‘Beilage: Das Interpolationsproblem bei Prudentius’ (p. 63-67), an energetic and detailed argument against the ‘Urvariantenlehre’ (= theory of authorial variants) as well as against Cunningham’s alternative theory that all doublings are to be explained as former ‘Randglossen’ and ‘loci similes’. Although Gnilka’s criticism is highly polemical, I cannot but favour his view; which is repeated throughout the book (cf. the index), but here it finds one of its sharpest pronouncements. Therefore I take this opportunity to comment on this important issue.

It is certainly too extreme to deny the possibility of unintentional intrusion of ‘Randglossen’ into the text. In my opinion, they have to be considered when no plausible reason for either replacing or supplementing a passage of the original can be offered. Jachmann’s and Gnilka’s occasional dismissive remarks in this regard (cf. p. 646 quoted below) are counterproductive, for they sometimes provoke a general disagreement even if the rest of the argument is strong.6 Gnilka’s opponents may even be led into failing to notice that explaining the interpolator’s intention is one of his peculiar virtues. One obvious example of ‘Randglossen’ is the incorporation of some parts of Symm. Rel. 3 into C. Symm. 2 (after v. 487, 648, 780, and 909): the reason for the addition—originally no doubt at the margin—must have been to give background information, not to change the poem into a satura Menippea. I would also prefer to explain the inept passage Apoth. 1069f. in this way, rightly assigned to a weaker talent than Prudentius by Gnilka (p. 645-47); it may have been a spontaneous comment by someone reading the surrounding verses but was hardly meant to become part of the poem, let alone to replace any original verse.7

P. 681f. Gnilka resumes the subject again and refers to the observations of M. Deufert: Lucrez und Marullus—Ein kurzer Blick in die Werkstatt eines humanistischen Interpolators, RhM 142, 1999, 210-23, who discusses an addition at the margin of a ms. of Lucretius. This addition is particularly interesting because of the signs indicating where exactly to insert the spurious verses. Gnilka is right to reject Deufert’s model of a twofold process of ‘Textdiaskeuase’ in this case, but is unconvincing when he claims that interpolation was universally a single deliberate deed. What about the authors of marginalia who did not attach such signs: were they commentators, poetastri, or forgers = interpolators? Most of the passages discussed by Gnilka (or mentioned in this review) are certainly of the latter type. Notwithstanding, there are numerous cases in which the order of the verses is not uniform in the mss.; partly this may be owing to contamination or confusion, but I suppose that the author of the marginal text more often simply had not indicated what to do with the adnotation; in such cases, the incorporation required a second deliberate step (e. g., C. Symm. 1.367,a; 2.143,a,b).

20th-century scholarship has been too much in favour of the theory of authorial variants and too sceptical about the possibility of significant text redactions in antiquity, even though there is explicit evidence in the ancient sources.8 Fortunately, the opposite position held by Jachmann and Gnilka is regaining ground, and especially in the case of Prudentius all instances of alleged authorial variants are explained in a different way. However, one should not entirely exclude the possibility of authorial variants either.9

III. Kritische Bemerkungen zu Prudentius’ Hamartigenie (first published in 1984, p. 68-90): although it might seem bold to claim that Ham. 745, 747f., 765-68, 776f. are interpolated, there is nothing wrong with Gnilka’s compelling arguments. He concedes that Prudentius—’as all Latin poets since Virgil’ (why these two restrictions?)—is fond of pleonastic accounts, but denies that any of the authentic repetitions are as ‘stumpfsinnig’ as those of the less skilled redactors. As Prudentius shows brilliance throughout his works, the harsh stylistic passages appearing here and there really deserve critical attention, particularly whenever they are accompanied by syntactical, metrical, thematical, or dialectical insufficienies. Reluctantly, I have to admit that the evidence of the mss. at least in such clear cases is secondary.

In most of the subsequent contributions Gnilka cogently demonstrates that even the oldest codd. (B) contain several 5th-century interpolations. In detecting them, the scholar also illuminates Prudentius’ literary skills, anthropology and theology.

IV. Eine interpolatorische Ehrenrettung Davids (first published in 1984, p. 90-101): Ham. 574-80 is shown to be a doubling intended to supersede v. 569-73.

V. Theologie und Textgeschichte. Zwei Doppelfassungen bei Prudentius, psychom. praef. 38ff. (first published in 1984, 102-25): v. 43f. and 60a had been designed to substitute v. 41f. and 60 respectively.

VI. Zwei Binneninterpolamente und ihre Bedeutung für die Geschichte des Prudentiustextes (first published in 1986, p. 126-137): the discussions concern Ham. 488ex.-488a in. and Psych. 727-29.

VII. Zur Praefatio des Prudentius (first published in 1988, p. 138-58): section I is a thorough philological commentary on v. 1-3, section IV on v. 43-45; as those verses form the first and the last stanza of the programmatic preface, Gnilka’s observations are particularly important. In sections II and III (v. 25-27 and 31-33), two stanzas are claimed to be spurious; Gnilka is followed by Bastiaensen (as note 7) 107f. There is no doubt that all the arguments put forward deserve consideration, as they rightly point out many weaknesses hitherto neglected. However, at least in the second case, I am not sure whether the rejection of the complete stanza is appropriate: Dicendum mihi: ‘quisquis es, / mundum, quem coluit, mens tua perdidit; / non sunt illa Dei, quae studuit, cuius habeberis’. Perdere mundum and haberi with genitive are certainly opaque but still acceptable; I find it petty to criticise the future tense, which must not be played off against Ham. 933ff. and Act. 17.25,27 or the like: writing habeberis, Prudentius does not deny belonging to God already during his lifetime, but death (which he feels approaching now) will entail a difference in quality.

The alleged motivation of the forger does not convince me either: ‘ Vel bona vel mala (29) im Verein mit jenem id est : das war ihm zu wenig. Und so brachte er hier den mundus als Inbegriff des Welttreibens hinein…’ But if this had really been the reason, would he not have chosen a more precise locution? Most importantly, something would be missing if the stanza were abandoned: the moment of the conversion itself, which is the main subject of the entire preface. I therefore prefer to stick to the transmitted text, without claiming that it is the unaltered original.

VIII. Eine Spur altlateinischer Bibelversion bei Prudentius (first published in 1988, p. 158-66): Gnilka soundly demonstrates that in Psych. praef. 30f. the poet neither draws on the Vulgata nor on the Itala but on another translation of the Septuaginta; for a stemma see p. 166.

IX. Palaestra bei Prudentius (first published in 1989, p. 167-87): I found this part slightly boring; Gnilka discusses two verses (C. Symm. 2.143a,b) which have been identified as interpolations already. He gives minute commentaries on enervare and palaestra, then resumes the questions of the ‘Autorenvariante’ and the motivation of forgers. In an excursus, he treats Iuvencus 2.452a,b

X. Das Templum Romae und die Statuengruppe bei Prudentius, C. Symm. 1,215/237 (first published in 1994, 188-218), with tables I-V: this is a knowledgeable and subtle contribution, which should not be missed by archeologists in particular. It discusses the description of the temple of Venus and Roma and the group of statues probably located in front of it. Gnilka exposes v. 223 and 230 as interpolations on various grounds supported by variant readings of the mss. However, I do not share his disapproval of v. 228f.: in spite of his polemic remarks, the notion of nothus qualifies the sons of Iuppiter, who were not born by Juno, but by Leda, ‘a seduced mother’ ( corrupta de matre nothi), cf. ThlL 4,1056,35ff. s.v. corrumpere and L-S-J (1968) 1178 s.v. νόθος; Ledeia is admittedly a hapax legomenon (instead of Ledaeus), but this does not allow us to reject it; the unique, but succinct periphrasis nocturni equites for Castor and Pollux seems quite suitable to me as well.

Nevertheless, Gnilka convincingly infers that the descriptions of Roman antiquities by Prudentius are reliable, though strongly selective; they must not be understood as an expression of Prudentius’ high regard for the classical heritage but rather fulfill a very precise function in his exposition of the sociology of religion, namely to illustrate how the Romans’ belief was conditioned from their childhood on. He ends with some additamenta to the Lexicon Iconographicum Mythologiae Classicae (Castores, Hercules, Ianus, Italus, Picus, Sabinus, Saturnus, Tros).

XI. Ein missgluecktes Interpretament im Prudentiustext (C. Symm. 2,302) (first published in 1996, p. 219-27): I have nothing to add to the self-explanatory title.

XII. Antike Goetter beim echten und beim unechten Prudentius (first published in 1996, 228-90): this voluminous contribution discusses nine passages on pagan Gods, which have been subject to distorting redactions. Only four instances are obvious from the evidence of the textual transmission. The first two examples (C. Symm. 1.367a and Per.10.302a, p. 230-39 and 239-44) are commonly accepted, but serve to establish a typology of the interpolators’ purposes: either to supplement or to replace original verses, to simplify an expression (or comment on it) or to avoid possible or actual contradictions to biblical themes or theological doctrine.

Section 3 (p. 244-50) gives another example of an apocryphal mythologem transmitted in cod. A (Per. 5.99a,100a) and again introduces the Forschungsgeschichte.

section 4 (p. 250-56): The redaction of C. Symm. 2.326ex.-327 in cod. Leidensis (E) is due to Prudentius’ measuring quadrupes dactylic, which was permitted since Aus. Griph. 38. The ‘correction’ is at the cost of distorting the content. Interestingly, quadrupes is also changed in C. Symm. 2.817 in the same cod. E.

section 5 (p. 256-61): Gnilka soundly condemns the obscure and inept identification of the numen Marcionis with the Charon mundi (Ham. 502).

section 6 (p. 262-65): The metrical shortcoming of Hadrianique exposes C. Symm. 1.274f. as an interpolation. Gnilka shows how both verses clearly violate the train of thought without contributing anything to it.

sections 7 and 8 (p. 265-81) discuss C. Symm. 1.379-407. The exclusion of v. 386f. is compelling: the mythologem is obscure, and the entire phrase violates the original concept blatantly. The arguments against v. 395-99 deserve consideration, but the conclusion to reject them does not seem cogent to me.

section 9 (p. 281-90) deals with the exhortation of Arcadius and Honorius to the Roman senate in C. Symm. 2.57ff.: desine, si pudor est, gentilis ineptia, tandem / res incorporeas simulatis fingere membris, desine terga hominis plumis obducere: frustra / fertur avis mulier magnusque eadem dea vultur. Gnilka’s grammatical arguments (in particular against frustra and ferri) are not decisive. To cover Victory’s body with feathers is admittedly wrong from an iconographical point of view; but with respect to the highly polemical character of v. 59f., this is not sufficient reason to challenge its authenticity. More significant is the observation that Prudentius has just characterised the image of the same goddess as highly graceful, though totally fictional (v. 27-38). The shift in aesthetic is explained by Gnilka with regard to the redactor’s misunderstanding of adsimulatis…monstris (v. 39f.), which indeed implies fictitiousness rather than horror, and by foeda ornamenta (v. 64) where, according to Gnilka, the adjective does not qualify the outer appearance but the immorality of idolatry.

However, the scholar does not tell why someone should have been induced to add the verses. Was it simply the desire to be more precise? Or did someone feel the need to counter the all too charming description of the winged goddess? But in the latter case he would probably have done so in the closer context of v. 36-38, or he could even have censured the latter passage itself. Hence the former motivation would be more plausible. In any case, the discrepancies can also be explained with respect to the fictional address by the emperors and their aggressive tone. It is therefore not necessary to abandon the transmitted text.

XIII. Doppelter Gedichtschluss (p. 291-356): this title inaugurates no fewer than 357 pages of original contributions formerly unpublished; I wonder why the additional notes to these (p. 674-90) have not been integrated into the text (or notes). Ham. praef. 36-47 is revealed as one of the most extended interpolations in the text of Prudentius. The forger had intended to exchange the original ending of the preface (which in the ms. tradition follows in v. 48-63); his brusque identification of Cain with Marcion not only destroys the subtle train of thought, but also entails a misreading of the entire poem to follow, as its scope is much wider than the rejection of the one archheretic and deals with dualistic theologies in general. At the same time, many inconsistencies, which emerge within the preface and throughout and cast doubt on Prudentius’ literary skills disappear with the exclusion of the spurious passage.

XIV. Falscher Marcion (p. 357-63): an appendix to part XIII, which exposes Ham. 93f. as interpolation (though not necessarily by the same redactor).

XV. Flickverse (p. 364-72): the term (coined by Jachmann) denotes cento-poetry drawing on other verses either of the original author under redaction or on classical writers. Ham. 69 and C. Symm. 2.1100f. are verified as such.

XVI. Eine gefaelschte Strophe im Romanushymnus (p. 373-84): stanza 82 of the hymn (Perist. 10.406-10) is marred by oblique motives and obscure phrases. The interpolator was tempted to complete Prudentius’ selective account of Roman history. Regardless of the fine composition, he clumsily disrupts it by his useless addition. He had obviously difficulties in filling five complete verses, so that v. 409 is of striking triviality: quidquid novellum surgit, olim non fuit.

XVII. Erweiterte Kataloge (p. 385-433): this study is only partly an original contribution, as it strongly draws on the article “Der Gabenzug der Staedte bei der Ankunft des Herrn—Zu Prudentius”, Peristephanon 4,1-76, in: Iconographia sacra—FS K. Hauck 1994, 25-67. The hymn figures as a general example of the temptation to extend catalogues. The pride of Hispanian Caesaraugusta resides in its 18 martyrs: Carthago (v. 16f.), Corduba (v. 19f.), Tarraco (v. 21-24), Gerunda (v. 29f.), Calagurris (v. 31f.), Barchino (v. 33), Narbo (v. 34), and Areles (v. 35f.) are outdone in number. The 7th stanza interrupts the speedy catalogue, and Gnilka is probably right to hypothesize that a redactor felt the desire to comment on the preceding verses.

The rejection of three entire stanzas that continue the catalogue adding the martyrs of the Lusitani (v. 37-40), Conplutum (v. 41-44), Tingis (v. 45), and the Massyli (v. 46-48) (p. 397-404) is radical. Gnilka’s first case is cogent: the picture of the altar of Christ (v. 39) is not compatible either with the concept of the fictive procession of the cities or with the holy scripture, and the stanza neglects the concentration on the number of the saints, which is the condition of the triumph of Caesaraugusta. The objections against the second stanza are of a different nature: the metaphor sanguinem Iusti, cui Pastor haeret (v. 41) is tasteless; the next verses tell three times that the martyrs in question were two, whereas in the preceding stanzas numerals are only used as an alternative to names. V. 46-48 are not only opaque but also suffer from the same shortcomings as stanza 10. Consequently, Portugal, Alcalá de Henares and Marocco may still be proud that their saints were praised in 5th-century verses, but they no longer have the right to claim the authorship of Prudentius.

The totally inept v. 61-64 (=st. 16) are to be discarded in an apparatus of a critical edition (p. 404-9).

Next v. 177-92 (=str. 45-48) are rejected (p. 409-33). The first two stanzas are also missing in cod. A, and they are marked with brackets by Bergman. But Gnilka adds further arguments: to him, they seem to reveal the same inaccuracy of thought as the two subsequent stanzas, as is certainly right for the inappropriate soteriology implied in st. 48 (cf. Gnilka’s statement against ‘moderner Heilsoptimismus’ on p. 431). He also argues that the periphrasis of the confessors (v. 187f. ambo gustarunt leviter saporem / martyriorum) establishes a difference of rank between the latter and the matyrs, contrary to Perist. 4.109-44, particularly 113, 135, 144 (on Encratis) and Paulin. Nol. carm. 12.9, 14.4 (on Felix). This is subtle, but maybe too subtle, because the reference given does not justify a generalisation. Nevertheless, st. 47 cannot stand without st. 46, as ambo resumes Gaius and Crementius. St. 45-47 were designed to complete the number of the 18 saints inscribed into the heavenly book (st. 44), and st. 48 was intended to fit the spurious text better into the context.

XVIII. Zu Paulinus Nolanus (p. 434-58): the incorporation of this contribution is only owing to the subtitle of the book ( Critica); at the same time, there are many parallels to the rest of Prudentiana I. The tertium comparationis is the fact that many interpolations in these authors antedate the oldest manuscript, notwithstanding its comparatively late date in the case of Paulinus (Vossianus 111 saec. IXin.). Seven passages are discussed: carm. 13.86-88, 25.239-41 (p. 442 ‘interpolatorische Drittfassung’), 6.172, 7.568f., 31.209f., 611f.; epist. 32.17 (prose!).

XIX. Unechtes in der Apotheosis (p. 459-647): in this gorgeous part, which could have been published as a separate monograph, Gnilka examines nearly the entire Apoth. and rejects 88 out of 1152 verses. Thus he sees the book on Christology as the one that suffered most from 5th-century redactions, which were due to an ongoing theological dispute (p. 459-61). As I do not want to duplicate the index of the book itself, I refrain from listing the passages and confine myself to one passage.

Gnilka’s most daring attempt is to reject the entire Hymnus de trinitate (p. 461-88), which precedes the so-called second praefatio of the Apoth. The introductory Forschungsgeschichte covers three centuries of scholarship (p. 461-66), naming F.A.Th. Obbarius (ed. Tuebingen 1845) as the only editor who denied the authenticity of the hymn. Contradicting contemporary philologists (such as K. Smolak, Diss. Wien 1968, or A. Bastiaensen, Studia Patristica 28, 1993, 3-14), who defend or even praise the obscurities of the twelve verses, Gnilka considers them a cheap pastiche exploiting the works of Prudentius and the Bible; he admits that its author chose a nice and meaningful frame: 3 verses each on the holy trinity, the father, the son and the holy spirit, but in doing so, he imposed conditions on himself which exceeded his abilities.

Gnilka convincingly notes many problems here. Est tria summa deus (v.1) aspires to brilliance, but the interpolator fails because of the paradoxical plural of the superlative. Secondly, trinum specimen instead of trina persona (sc. for tres personae) is inappropriate, since Prudentius relates specimen (as well as species) only to the son (cf. Apoth. 6, 18 sed tamen et patris est specimen and 23-25 for species); note that combination retro semper (v. 5) is obscure and without parallel, probably an unsucessful imitation of Apoth. 271. Fourthly, v. 6, 10-12 are even more opaque. Fifthly, v. 8-12 excessively exploit original Prudentian verses (C. Symm 2.194, Apoth. 875 etc.). And finally, the verses on Christ do not even mention his crucifixion. Although I do not share Gnilka’s reservations about other sections,10 it is indeed difficult not to agree with him: so many shortcomings in only twelve verses are untenable, especially because they pretend to be of outstanding significance.

In addition, Gnilka has given a short, but sound comment on the dispute about where to situate the creed: according to the manuscript tradition, which places the hymn after the title Apotheosis, it had been intended to head only this christological book and must not include Ham. and Psych. nor to C. Symm. (p.462f.).

But Gnilka paid little attention to the question of the overall composition of Prudentius’ edition A.D. 404. Most recent scholarship tends to prefer an analytical view and either challenges the whole idea of a voluminous edition subtly arranged by the author, or speaks of no more than a loose collection of works already published before (hence the dubious theory of the ‘Autorenvariante’ magisterially rebuked by Gnilka in the case of Prudentius). However, most of the strong arguments for a unified design put forward by Steidle (1971) are still valid. After the rejection of the spurious hymn, the Prudentian opus magnum emerges even more brightly:11


Liber Cathemerinon (12 hymns)

[Hymnus de trinitate] Apotheosis ([Hymnus de trinitate] 1st allegorical preface, 1st didactal poem) Hamartigenia (2nd allegorical preface, 2nd didactical poem)

Psychomachia (3rd allegorical preface, 3rd didactical poem: the ‘soul’)

Contra Symmachum liber I (4th allegorical preface, 4th didactical poem) Contra Symmachum liber II (5th allegorical preface, 5th didactical poem)

Peristephanon liber (12 hymns; 8 and 10 are to be excluded from the 14 transmitted)


A vivid and subtle discussion of the only authentic preface (p. 488-96) and the poem itself (496-647) follow. What Gnilka says at the end of this learned examination in order to defend the athetesis of v. 1069f. is quite typical of his character and style (p. 646): ‘Und in der Tat, auch wenn sich kaum ein plausibler Grund dafür wird finden lassen, weshalb “dem Falsar die Kehle juckte” ‘ (quoting Jachmann), ‘als er die Zeilen 1069/70 einschob: muessen wir sie deswegen ertragen? Der Grundsatz superflua non nocent gilt für die Jurisprudenz, nicht für die Kunst. Wie totes Treibholz liegt das Stueck im lebendigen Wellenschlag der Dichterverse, die hier besonders feierlich und kraftvoll dem Gedichtende zustroemen’. Such a powerful prose undoubtedly resembles 19th-century philology, which partly discredited the search for interpolation. What marks the difference is that Gnilka bases his arguments not only on taste, but on linguistic and substantial arguments most of which will still convince scholars of the centuries to come.

What is missing in this book? In place of the tiresome repetition of the theory of textual transmission and the methodology applied, I would have appreciated a systematic introduction to this highly important subject. This lack is even more regrettable in view of Gnilka’s promise to present such an overview of ‘Jachmanns Theorie’ in a third volume of Jachmann’s ‘Schriften’, which, as far as I know, never appeared. The detailed index (particularly part III: Interpolationswesen, p. 745-55) is highly valuable, but no sufficient substitute. It would be useful to incorporate something of this kind in the forthcoming Prudentiana II, if still possible.

The latter will also include a detailed index of ‘Namen und Sachen’ as announced in the preface of the first volume; I hope that ‘Namen’ is not confined to ancient authors, as so many of Gnilka’s valuable contributions to Forschungsgeschichte would be lost in hidden notes. Moreover, it would be desirable to include most of the items of the ‘Register III’ once again, because the entries are arranged from the point of view of the ‘Interpolationswesen’, which makes it awkward to use for a different purpose, e.g., entries concerning metrics or the manuscript tradition are annoyingly scattered over those eleven pages, which would have to be read entirely in order not to miss one. It would also be worthwhile to provide the second volume with a bibliography of Prudentius, as complete as possible.12

From a reading of Prudentiana I, two further desiderata emerge: a critical praefatio to the textual transmission of Prudentius, which I had expected to be included in this volume; and a re-assessment of Prudentius’ style, because I have learnt that such dubious pleonasms as parente natus alto et ineffabili / parente inenarrabili atque uno satus (Psych. praef. 60,60a) are in fact not typical of the Christianorum Maro, as commonly assumed.13 A short, but well-knit essay could still find its place in the second volume. The readers’ eyes will appreciate not only the virtual lack of printing errors (I did not find one on more than 700 pages!) but also the standardised and overall generous print, which gives the original page number as well as a continuous pagination. Unfortunately, the type setting of the quotations, especially of the alleged interpolations, is inconsistent: once they are in normal letters, occasionally in brackets; italics and boldface sometimes have the same function, but elsewhere are used to underline particular words. Abbreviations also vary, but never at the cost of precision. To conclude: in more than 600 pages (related to Prudentius), Gnilka tries to demonstrate that out of a total of ca. 10839 verses, ca. 247 are interpolations; while a few of these were considered spurious before, a new core has been put forward by Gnilka.14 Even those not prepared to follow his basic assumption that most of the changes are owing to 5th-century redaction, cannot deny that his arguments are at least serious. Of course, I myself go much further: I regard Gnilka’s premise as overwhelmingly supported; interpolations of cod. A can no longer be treated as ‘Sache des persoenlichen Credo’ (p. 245); the question is no longer whether, but to what extent the text of the Putaneus had already undergone redactions. In roughly two out of three concrete cases (within or without cod. A) put forward, it seems rather difficult not to agree with Gnilka; but Prudentian scholarship needs to take all the cases into account.

What follows from all this? Gnilka (p. 88) quotes a letter in which E.R. Curtius replies to Jachmann’s observations (4.2.1936): ‘Wenn man das Facit aus Ihren Abhandlungen zieht, hat man das Gefuehl, ein Erdbeben miterlebt zu haben. Denn Sie fordern ja nichts Geringeres als die kritische Revision sämtlicher Texte. Wir haben lauter interpolierte Texte gelesen! Ein unheimliches Gefuehl…’ This comes close to what I felt reading Gnilka: the figures mentioned above—less than 2.5 per cent of the whole—may seem small but their acceptance or their rejection affects the interpretation of far more verses and demands a reconsideration of the entire work of Prudentius, as is clear from my discussions of Ham. praef. (part XIII) and most clearly of the Hymnus de trinitate (part XIX). At the end of the 20th century, Prudentian scholarship could be regarded as divided into two periods: the one that antedated Bergman’s critical edition (1926) and the one that followed. Without questioning the historical merit of the Swedish classicist who undertook the Herculean labour of collating more than 300 manuscripts of the most read and most copied Christian poet, I venture the suggestion that the generations to come will consider Gnilka’s Prudentiana as the inauguration of a third period in the Forschungsgeschichte of Prudentius.


1. Vol. 1: Der Begriff des ‘rechten Gebrauchs’, 1984; vol. 2: Kultur und Conversion, 1993, both reviewed by A.A. Bastiaensen, Gnomon 70, 1998, 33-37.

2. Cf. G. Jachmann: Ausgewaehlte Schriften; Textgeschichtliche Studien, ed. Chr. Gnilka, Koenigstein im Taunus 1981/82. Most significant for the history of textual redactions in late antiquity is the contribution in I, 470-527: Das Problem der Urvariantenlehre in der Antike und die Grundlagen der Ausoniuskritik (1941); particularly instructive is also Gnilka’s preface to the latter volume; p. IXf. he announces a ‘Registerband’ with an introduction to ‘Jachmann’s Theorie’, which, as far as I see, unfortunately was never published. Throughout Prudentiana I, Gnilka refers to Jachmann’s categories, most frequently in part XII, p. 228ff.; cf. also the entries ‘Autorenvariante’, ‘Interpolatoren’, ‘Interpolatorische…’, ‘Urvariantenlehre’ in the index.

3. The Bruxellensis 9974-6 saec. xiii ( γ 1) is to be distinguished from the Bruxellensis 5381 of A.D. ca. 1050, which does not have this poem, cf. Hall XVIII and 201 (apparatus). Hall further mentions cod. Cracoviensis 71 saec. xii/xiv (g), Cantabrigiensis Coll. Trin. O. 3.22 saec. xii (ξ Montepessulanus 330 saec. xii/xiii ( μ); the missing v. 304 has been added on the margins of codd. γ 1, C, and μ. Thus they appear related to the source of the editio princeps (1482), the excerpta Florentina (after 1482) and exc. Gyraldina (not before 1523). —With the majority of the codd., Hall reads quae (v. 303), posset (v. 305), and maintains v. 304, explaining the omission in some of the mss.: ‘propter homoeoarcton, ut mihi videtur’.

4. Gnilka 20 note 7 assumes that the ‘Bruxellensis’ (which he considers one of the ‘Haupthandschriften’) has quae (v. 303) and v. 304, but reads posset (v. 305), but see previous note.

5. I refrain from further quotations; examples of strong, harsh or surprising expressions can be found on p. 144, 152f., 181f., 246, 262 (‘verhunzt’), 547 (‘unverständliches Kauderwelsch’), 647, 745 (‘Allerweltswoerter’ in the index!).

6. Cf. J. Delz: Textkritik und Editionstechnik, in: Einleitung in die lateinische Philologie, ed. by F. Graf, Stuttgart 1997, 51-73, 69: linguistic or material insufficiencies alone do not justify the claim of interpolation; whoever rejects the transmission is obliged to suggest a reasonable motivation. However, K. Dover: Textkritik, in: Einleitung in die griechische Philologie, ed. by H.G. Nesselrath, Stuttgart 1997, 45-58, 52 states that the motive cannot always be explained. These two points of view do not necessarily contradict each other.

7. For a short survey of the general problem cf. Delz (as note 6) 59: to explain differing manuscript readings by authorial variants is now regarded as inappropriate, adding (speaking of Ovid’s Metamorphoses): ‘Interlineare Erklaerungen und Randbemerkungen koennen in den Text eindringen’, followed by cogent examples; p. 68-70: short typology of interpolations; p. 70: with reservations against Jachmann, without mentioning Gnilka. Disappointingly, J.K. Eberlein: Interpolationsforschung, DNP 14, 2000, 617-19 deals only with juridical sources. For discussion of the text of Prudentius cf. the instructive account of A.A.R. Bastiaensen: Prudentius in Recent Literary Criticism, in: Early Christian Poetry, ed by J. den Boeft and A. Hilhorst, Leiden 1993, 101-34, in particular p. 101-8, with a favourable review of Gnilka’s contributions.

8. Cf., e.g., A.E. Hanson: Galen—Author and Critic, in: Editing Texts, ed. by G.W. Most, Goettingen 1998, 22-53, in particular 27 and 31f. —Gnilka frequently quotes O. Zwierlein: Die Ovid- und Vergilrezeption in tiberischer Zeit, vol. 1, Berlin 1999. For further references cf. notes 6 and 7.

9. Telling is the case of Ausonius: while I object to the assumption of authorial variants in the Mosella (cf. my forthcoming article in REA) and in epist. 23/24 to Paulinus (of Nola) (pace R. Green: The Works of Ausonius, Oxford 1991, 655, cf. my forthcoming book ( Die gens Ausoniana an der Macht, Prosopographica et Genealogica 8, Oxford ca. 2001/2), the so-called Fasti prove that fragments of different editions of ancient works could indeed survive (cf. also P & G 8).

10. Gnilka further disqualifies the following words: vigor unus (v. 1); filius ipse est (v. 2): Gnilka comments (p. 474) after the first part of the verse ‘liess der muede Pegasus erstmals die Fluegel sinken’: sanctus ab aeterno subsistit spiritus ore (v. 3): the verse does not tell explicitly, from whose mouth the spirit rose, but I consider it hypercritical to invoke Ham. 931-37 and Cath. 1.1-8 against it; in any case, the dispute about the filioque (which was added to the catholic credo under Charlemange) is of no relevance in this context.

11. Cf. W. Steidle: Die dichterische Konzeption des Prudentius und das Gedicht Contra Symmachum, in: Vigiliae Christianae 25, 1971, 241-81. For more recent views opposing the unitarian concept cf. Bastiaensen (as note 2) 108-15; p. 110 and 111 (note 48), with reference to Gnilka’s reservation (Interpretationen fruehchristlicher Literatur, dargestellt am Beispiel des Prudentius, in: Impulse für die lateinische Lektuere, ed. H. Krefeld, Frankfurt 1979, 138-80, 179 note 93) see most recently L. Rivero García: La poesía de Prudencio, Huelva 1996, 21-28. I will return to the subject elsewhere.

12. In spite of some shortcomings, the most useful bibliography on Prudentius is still to be found in Rivero García (see note 11 above) 242-82.

13. Cf., e.g., Lavarenne vol. I, p. XII-XIV: ‘Prudence ne saurait prétendre au titre de grand poète. Il lui a manqué ces deux dons essentiels: la finesse et la mesure.’

14. The figures are based on the ed. of Bergman; where I simply took the number of the last verse of each poem. As for Gnilka, I counted all the verses printed bold in his index; the alleged number of interpolated verses are added in brackets. The summaries of the account are the following: Praef. 45 [6]; — Cath. I 100 [-]; II 112 [-]; III 205 [1]; IV 102 [-]; V 164 [-]; VI 152 [-]; VII 220 [-]; VIII 80 [-]; IX 114 [-]; X 172 [9]; XI 116 [-]; XII 208 [-]; —Apoth. Hymn. 12 [12]; praef. 56 [-]; Apoth. 1084 [88]; —Ham. praef. 63 [12]; Ham. 966 [43]; —Psych. praef. 68 [3]; Psych. 915 [4.5]*; —C. Symm. I praef. 88 [-]; I 657 [13]; II praef. 66 [-]; II 1132 [19.5]; —Perist. I 120 [-]; II 584 [-]; III 215 [-]; IV 200 [36]; V 576 [2]; VI 162 [-]; VII 90 [-]; VIII 18 [-]; IX 106 [-]; X 1140 [6]; XI 246 [-]; XII 66 [-]; XIII 106 [-]; XIV 133 [-]; -Ditt. 196 [4]; Epil. 34** [-]. * I did not add Psych. 456ex.-458in., for which see his PhD Dissertation (1962/63) p. 129-33. ** I do not count v. 35, which is an obvious interpolation; it is not discussed by Gnilka though.