Was the Commentary on Vergil by Aelius Donatus Extant in the Ninth Century? A Reappraisal

1 Introduction

The student interested in the history of Latin scholia is regularly told that the Vergilian commentary by Aelius Donatus – one of the most influential late-antique commentaries that have not survived^[1] – was still extant in the Carolingian period and is referred, for the relevant evidence, to three articles published by Savage in the first decades of the last century:^[2] “The scholia in the Virgil of Tours, Bernensis 165” (1925), “More on Donatus’ Commentary on Virgil” (1929), and “Was the Commentary on Virgil by Aelius Donatus Extant in the Ninth Century?” (1931). In these three articles, Savage claimed that a few marginal notes in one of the ninth-century primary witnesses to the DS scholia, the so-called ‘Vergil of Tours’ (Bern, Burgerbibliothek, Ms. 165 = T), were drawn almost directly from Donatus’ commentary and that a marginal note in a roughly coeval Servian witness (Bern, Burgerbibliothek, Ms. 363 = B) provided information about a place where a copy of the commentary could be found. This paper aims to re-examine such evidence in the light of the current knowledge on the two manuscripts and the Carolingian scholarship.^[3]

2 Bern, Burgerbibliothek, Ms. 165 (T)

Copied in the scriptorium of Tours at the beginning of the ninth century, T is one of the best-known examples of Carolingian ‘glossed edition’ of Vergil.^[4] It transmits the three Vergilian works along with a disordered mass of interlinear and marginal notes added at different stages by a few almost contemporary scribes.

As was noticed by Savage, one of T’s scribes, who might have gone under the name of Liudramnus,^[5] drew his notes on the Aeneid from an organic Carolingian commentary which was in turn based on Servius as well as on other sources.^[6] Remarking that (a) eight of these notes could be positively suspected of depending on Donatus and (b) three other notes explicitly mention the name of Donatus, Savage concluded that one of those sources should be identified with Donatus’ lost commentary itself.^[7]

In particular, five notes of the former group were suspected by Savage of depending on Donatus because they could be paralleled in texts that he regarded as witnesses to the Donatian exegesis:^[8]

As far as the notes on Aen. 1.140 and 2.5 are concerned, the parallels adduced by Savage seem to be devoid of any probative value: if they are not completely irrelevant, as happens for the scholium on the Eunuchus and for the DS scholium ad Aen. 2.5, they simply agree with Liudramnus’ notes in offering the same obvious interpretations of the Vergilian lines at issue. The other three notes do show some verbal similarity with these ‘Donatian’ parallels,^[10] but the conclusion that Savage aimed to infer from such similarity – namely that the notes also derived from Donatus’ commentary – does not appear legitimate for two reasons.

First, even if one maintained, for the sake of the argument, that such similarity should be regarded as evidence of a relationship between Liudramnus’ notes and the other texts, the former could also be explained as depending directly on the latter or on similar early-medieval texts. In the case of the note on Aen. 1.85, one might theoretically suppose that the notion of the identity of Eurus and Subsolanus was drawn either from the liber glossarum itself or from a related work since, as Savage himself remarked, Liudramnus’ notes often agree with coeval glossaries and some of these agreements concern notions which are unlikely to have been inherited from Donatus.^[11] The note on Aen. 2.65 includes no element that could not have been deduced from Servius, whom Savage proved to be the main source of the Carolingian commentary excerpted by Liudramnus. As for the note on Aen. 2.235, the interpretation of accingunt as accinguntur, which, incidentally, cannot even be attributed with certainty to Donatus,^[12] is admittedly not mentioned in the Servian corresponding scholium; nevertheless, this interpretation could have been found in the liber glossarum^[13] or in a related work or even have been simply inspired by the few similar Servian scholia where an active verb is interpreted alternatively as a middle or as a reflexive.^[14]

Second, the similarity between the three notes and the parallels given by Savage is still so loose that it may well be considered as accidental: nothing proves that the compiler responsible for the text of these notes could not have drawn them from other sources^[15] or even have simply jotted them out of his own knowledge and understanding of Vergil.^[16]

The remaining three notes were suspected by Savage of depending on Donatus on the basis of a comparison with Servius.^[17] Savage remarked that these notes – albeit very similar to the Servian corresponding scholia – preserved some information which could not be found there and concluded that they were excerpted directly from the main source of those scholia, namely Donatus’ commentary:

On the one hand, the etymological explanation of conticinium, which is the only element of Liudramnus’ first note not to be found in Servius, is so common that it does not require, like that of Lenaeus discussed supra,^[19] to be regarded as descending from a specific textual source.^[20] On the other, nothing suggests that the notes on Aen. 3.209 and 6.844 should be judged as emanations from the same commentary to which Servius resorted rather than as conflations of Servian and non-Servian materials.^[21] None of the non-Servian elements at issue exhibits any peculiarly Donatian quality. While the notion that Phineus was a king of Thrace, which Savage found in the DS scholium ad Aen. 3.209,^[22] is also present in a few other extant classical sources,^[23] Phineus’ request to Apollo to be released from the Harpies and the description of Fabricius’ house cannot be paralleled in any such source and might even be early-medieval developments, respectively, of the myth according to which Phineus was punished because he spread the divine knowledge that he had received from Apollo^[24] and of the anecdotes on Fabricius’ possessing only a salinum and a patera.^[25]

Even more compelling probatory value was assigned by Savage to three notes where the name of Donatus is explicitly mentioned:

[f. 57^r, ad Aen. 1.179] Donatus dicit histeroproteroproteron esse, id est ordinem permutatum: primum enim frangitur frumentum mola, deinde torretur, id est quoquitur. sed Seruius dicit non esse ita: nam primum siccauerunt frumentum, deinde moluerunt.[f. 63^v, ad Aen. 1.559] hic non est pleonasmos, sicut Donatus dicit, quia possunt et armis (ex aliae res) fremere.[f. 95^v, ad Aen. 4.207] Leneus dicitur non a leniendo, sicut Donatus dicit, sed apotis lenu, id est a lacu in quo premitur.

The structure of these notes is almost identical: each proposes a Donatian and a non-Donatian explanation and seems to favour – implicitly or explicitly – the latter over the former. The three non-Donatian explanations at issue can all be found in the corresponding Servian scholia:

ad Aen. 1.179: et torrere parant flammis et frangere saxo multi hysteroproteron putant ... quod falsum est; nam et hodieque siccari ante fruges et sic frangi uidemus. et quia apud maiores nostros molarum usus non erat, frumenta torrebant et ea in pilas missa pinsebant; et hoc erat genus molendi.ad Aen. 1.559:ore fremebant ... bene ore, quia et armis possumus fremere.ad Aen. 4.207 lenaevm Bacchicum; nam Liber Lenaeus dicitur quia lacubus praeest, qui et Graece ληνοί dicuntur; nam, cum sit Graecum, a mentis delenimento non potest accipi.

The first two scholia should be regarded as, in all likelihood, the direct sources of the two non-Donatian explanations: the former is explicitly attributed to Servius, and the latter corresponds almost uerbatim to a Servian clause. Whilst it is true that the two alternative explanations could not be attributed to Donatus on the basis of these scholia since one of them is simply attributed to multi and the other is not even explicitly enunciated, both could be directly or indirectly deduced by a Carolingian scribe from Donatus’ most widely circulating work, namely the ars maior:^[26]

p. 401.6–8 K.: hysterologia uel hysteroproteron est sententiae cum uerbis ordo mutatus, ut torrere ... saxo [Aen. 1.179].p. 395.3–4 K.: pleonasmos est adiectio uerbi superuacui ad plenam significationem, ut sic ore locuta est [Aen. 1.614], pro ‘sic locuta est’.

The possibility that the two explanations were indeed drawn from the ars, contemplated but dismissed without any argument by Savage,^[27] seems to be corroborated by a broader examination of Liudramnus’ notes. The Donatian definitions of hysterologia and pleonasmos just quoted belong to the chapters de ceteris uitiis and de tropis, which refer to a few other Vergilian lines as examples of specific rhetorical phenomena. Not only are eleven of these lines explained by Liudramnus and by Donatus in the same way, but at least two of the notes copied by Liudramnus exhibit close verbal parallels with the ars and may well depend on it:^[28]

Although the two explanations proposed in the third note can be found in the corresponding Servian scholium, it is not possible to establish whether the two should be conceived as genealogically related. In any case, the fact that Liudramnus’ note mentions the paternity of the Latin etymology of Lenaeus and the Servian scholium does not cannot be regarded as a compelling proof of the dependance of the former on Donatus. As was pointed out by Thilo and Wessner,^[29] such information could be easily found in another Servian scholium, as well as in the corresponding scholium preserved by one of the traditions to the ‘Philargyrian’ exegesis to which T’s scribes seem to have had access, namely the breuis expositio in Georgica:^[30]

Seru. ad Georg. 2.4: hvc pater o lenaee ... Lenaeus autem ἀπὸ τῆς ληνοῦ dicitur, id est a lacu: nam quod Donatus dicit ab eo, quod mentem deleniat, non procedit: nec enim potest Graecum nomen Latinam etymologiam recipere.breu. expos. ad Georg. 2.4: o lenaee Lenaeus pater Liber ἀπὸ τοῦ ληνοῦ, id est a lacu, in quo uuae premuntur ... nam quod Donatus ait ab eo, quod mentem deleniat, non procedit: non enim potest Graecum nomen Latinam etymologiam recipere.

In conclusion, Savage’s discussion of these eleven notes is not persuasive,^[31] and the thesis that Liudramnus had an almost direct access to Donatus’ lost commentary remains unproven.

3 Bern, Burgerbibliothek, Ms. 363 (B)

B is a miscellaneous manuscript copied in the late-ninth century, perhaps in St. Gallen or in Milan, by an Insular scribe. Among the other texts, it preserves the Servian commentary on the Eclogues, on the Georgics, and on the Aeneid, up to 7.16, with several brief marginal notes added by the same scribe.^[32]

Savage deserves credit for drawing the attention of the Servian scholars to a note written in the left margin of f. 41^v, opposite the words Leneus autem aπο τηc ληνου dicitur, id est a lacuna; quod Donatus dicit ex eo quod mentem deleneat. nec enim potest Graecum nomen Latinam ethimologiam recipere (Seru. ad Georg. 2.4): Donatus alter qui in totum Virgilium exposuit in Leotica.

Taking the indefinite adjective as meant to distinguish between the two commentators sharing the name of Donatus – namely Aelius Donatus and Tiberius Claudius Donatus –, Savage translated the note as “the Donatus [there are two of them] who wrote a commentary on all [the works] of Virgil [is] in the city of Liège”. Since Tiberius Claudius Donatus’ commentary confines itself to the Aeneid alone, he concluded (a) that the scribe must have been referring to Aelius Donatus’ commentary and (b) that the note attested its survival in the ninth century.^[33]

These conclusions, which have been since unanimously accepted,^[34] deserve to be discussed afresh. For the sake of simplicity, the author referred to in the note and the other homonymous author from whom he was meant to be distinguished, identified by Savage with Aelius and Tiberius Claudius Donatus, respectively, will be henceforth labelled as Donatus¹ and Donatus².

Conclusion (b) was presented by Savage as a direct corollary of conclusion (a) but is, strictly speaking, built on conclusion (a) as well as on the premise that the scribe knew that a copy of Donatus¹’ work was still extant. This premise was understandably taken for granted by Savage since he interpreted the whole note as a reference to a place where the scribe knew that a copy of the work at issue could be found. However, this interpretation, intrinsically problematic from a linguistic perspective,^[35] has been proven to be no longer tenable: as was pointed out by Vocino, the words in Leotica are written with a different ink and do not belong to the original note, which must have been meant simply to clarify the identity of the author cited in the Servian scholium.^[36] For this reason, it is not possible to determine whether the scribe had any direct knowledge of Donatus¹ when he wrote the original note^[37] and even the possibility that, while adding the words in Leotica, he might have not paid the due attention to the distinction that he had drawn between two homonymous writers cannot be completely ruled out.^[38]

More significantly, conclusion (a) is not only implausible but also based on a gratuitous assumption. On the one hand, it is per se hard to believe that the scribe regarded Aelius Donatus’ name as requiring any clarification and even harder to believe that he feared that his name could be erroneously taken for that of Tiberius Claudius Donatus.^[39] To say nothing of the circulation of the artes in the Carolingian period, the other notes added by the same scribe in B provide clear evidence of his familiarity with their author. The name of Donatus occurs, either abbreviated as dō or written in full but with no further clarification, in at least seven notes alongside Servian scholia;^[40] in all these cases the reference concerns scholia dealing with grammatical matters, and in all but one the grammatical matter dealt with in the Servian scholia is indeed discussed also in the artes.^[41]

On the other hand, as already mentioned, conclusion (a) was considered by Savage as a direct consequence of the premise that the words qui in totum Virgilium exposuit could only have been meant to distinguish Aelius Donatus from Tiberius Claudius Donatus since the former wrote a commentary on the three works of Vergil and the latter wrote a commentary on the Aeneid alone. In fact, this premise, which was not explicitly discussed by Savage, rests on the assumption that the scribe must have known that Aelius Donatus wrote a line-by-line commentary on the three Vergilian works. This assumption is, if not unlikely, certainly unproven.^[42] In fact, the scribe might well have added those words because he wanted to clarify that Donatus¹ should not be identified with the Donatus referred to elsewhere, that is Aelius Donatus, whose artes comment in Virgilium but not in totum Virgilium,^[43] but rather with another homonymous author who was responsible for a commentary in totum Virgilium.

If this were the case, one might even speculate that the scribe could have meant to identify Donatus¹ with Tiberius Claudius Donatus. The phrase in totum Virgilium certainly sounds imprecise as a reference to this author since his work – albeit far more complete as a commentary than Aelius Donatus’ artes – concerns the Aeneid alone; nevertheless, the imprecision could perhaps be excused if one considers not only that it goes under the generic title of interpretationes Vergilianae but also that the scribe could simply have known it as a continuative commentary on Vergil^[44] and have had no particular direct knowledge of it.^[45]

In any case, whether or not the scribe meant to identify Donatus¹ with Tiberius Claudius Donatus, the fact that Savage’s conclusion (a) seems to be both implausible and based on an unproven assumption prevents one from regarding his conclusion (b) as validly demonstrated.

4 Conclusion

None of the arguments adduced by Savage to prove that Donatus’ commentary was extant in the ninth century stands scrutiny.^[46] Although this does not imply that the commentary could not have been extant during that period, it seems more prudent, on the basis of the available evidence, to antedate the terminus post quem for its loss by at least one century.^[47]