Did the world need a new critical edition of Aulus Gellius’ Noctes Atticae just over half a century after P. K. Marshall published his Oxford Classical Text? Those who thought not are proven wrong by this magnificent edition by Leofranc Holford-Strevens (ed., vols. 1–2), published together with the companion volume Gelliana, in which Holford-Strevens comments on the text and explains his editorial decisions. He has produced not only the best edition of Gellius to date, but also a masterpiece of Latin textual scholarship.
Editing the Noctes Atticae is an unusually complex task for two reasons. First of all, Gellius quotes from, or refers to, scores of earlier texts, and these quotations and references raise editorial challenges of their own. Most of them are very different in language and subject-matter from the Noctes Atticae, some are also known through other sources, and many have a rich scholarly tradition of their own that must be reckoned with. In sum, an editor of Gellius must edit all these other texts as well. Holford-Strevens’ philological skills and outstanding erudition let him rise to this challenge admirably.
Second, the manuscript tradition of the Noctes Atticae is not only rich but also complex, since different groups of manuscripts contain different parts of the text, and medieval florilegia and epitomes represent further lines of transmission. A closer look will show the nature of this material and highlight the advances made in the use of the sources.
Most surviving manuscripts contain either the Preface and Books 1–7 of the text with the lemmata or chapter summaries of Books 1–8 (class Ξ; I use Holford-Strevens’ sigla), or Books 9–20 with the lemmata of all these Books except 19 (class Ψ). The text of Book 8 has been lost, as have the lemmata of Book 19. Some later manuscripts combine what is transmitted by Ξ and Ψ so as to create a complete text of the Noctes Atticae.
Two manuscripts do not fit into this pattern. The third-century palimpsest A (Vatican City, Biblioteca Apostolica Vaticana, Palatinus lat. 24) contains parts of Books 1–4 and of the lemmata of Books 17–18. The mysterious Buslidianus or BUcontained all the books that survive today; we know its readings from several scholars who consulted it in sixteenth-century Leuven, before it disappeared from view. Holford-Strevens suggests that BU may have been a fifteenth-century manuscript with a contaminated text that combined inferior readings with others that are clearly ancient (ed. vol. 1, pp. xxxiii–xxxiv). Notably, BU conserves several short passages that are missing in Ξ and Ψ. Out of these, 1.2.15 and 1.3.1–7 also appear in A, which was not yet known when the Buslidianus emerged in the sixteenth century.
Marshall’s Oxford Classical Text of 1968 may serve as a useful point of reference. Marshall used three manuscripts of class Ξ and eight of class Ψ, as well as A and BU. Holford-Strevens uses five manuscripts of class Ξ, twelve of class Ψand two composite manuscripts that combine the text of the two classes, alongside A and BU. Instead of Marshall’s thirteen manuscripts, Holford-Strevens uses twenty-one. He has studied them both through reproductions and in person and locates them in a more detailed stemma codicum than that of Marshall. It must be said that many of the manuscripts that were added by Holford-Strevens had been discovered after 1968 by other scholars, including Marshall. Still, adding so many codices to the stemma is a major step forward, and not the only one.
The popularity of Gellius during the Middle Ages is shown by the survival of florilegia and epitomes that contain parts of the Noctes Atticae. These are in effect alternative strands of the manuscript tradition. Marshall quotes one florilegium, using two manuscript sources. Holford-Strevens quotes four florilegia, using over a dozen manuscripts and other sources in all. He also takes into account a medieval epitome of Gellius, of which he quotes four manuscripts.
Most critical editions quote a few more manuscripts than their predecessors, or they quote the same sources but simply provide a more accurate report of the manuscript readings. Holford-Strevens has introduced not only new manuscripts and florilegia, but also a new class of textual evidence with the epitome. It would be hard to find another recent editor of a classical text who has expanded our knowledge of the sources so radically.
Let us now take a look at the structure of the edition. The first volume opens with the “Testimonia Antiqua” (pp. vii–x), thirteen passages from antiquity and the early Middle Ages that refer to Aulus Gellius or as he came to be known, Agellius. There follows a long introduction written in Latin (“Praemonenda,” pp. xi–l), a list of abbreviated bibliographical references (“Notae,” pp. li–liii), an overview of earlier editors and editions (“Conspectus editorum Auli Gelli,” pp. lv–lvii) and a list of sigla (“Signa codicum,” pp. lix–lxiv). At the end of the second volume, which has continuous page numbering, there stands further material including a list of the editions of texts other than Gellius that have been used (“Conspectus editionum praeter Gellianas,” pp. 685–88), a bibliography (“Conspectus studiorum,” pp. 689–720) and a useful set of indices, which embrace proper names, laws and edicts, libraries, and miscellaneous words other than proper names that are discussed by Gellius (pp. 721–64).
There is no obvious order to the contents of the Noctes Atticae; the books must have stood originally on individual papyrus rolls, and nothing guarantees that we have the order in which they were arranged by their author. Modern editors follow the order found in the manuscript tradition except for BU, which was radically different (see ed. vol. 1, pp. xxxiv–xxxvi). In any case, Gellius’ preface and the lemmata originally stood at the start of the work, before Book 1. Marshall printed the lemmata there and repeated each lemma before the chapter it summarised, where he added a critical apparatus. Holford-Strevens makes two reasonable changes to this arrangement: he adds a critical apparatus to the edition of the lemmata before Book 1, and he puts the lemma not before the chapter summarized but after it, which reflects not only the relative importance of text and lemma but also their order of composition.
The edition uses complex layout to convey a large amount of information in a clear and intuitive format. Chapters are introduced by alphabetic Greek numerals, with Arabic numbers added in the margin for the faint-hearted. The text is accompanied below by an apparatus of the sources quoted by Gellius; an apparatus of the indirect transmission, listing the florilegia, epitomes and other early texts that quote each passage and thus supplement the manuscript tradition; and by a critical apparatus. On occasion a fourth apparatus is added for miscellaneous notes (cf. ed. vol. 1, p. 98). This may sound complicated, but the function of each apparatus is revealed by its contents at a glance. This format allows for a large amount of evidence to be presented clearly in a limited amount of space. It is used with skill; I have found few misprints.
Gelliana too is admirably compact; it offers just 165 pages of commentary on 684 pages of Latin text. Most textual problems are discussed in under a dozen lines. A consistent focus on the key issues at stake yields transparency. The commentary is preceded by brief “Prolegomena” (pp. ix–xv) on some recurring editorial problems such as errors by Gellius; a “Note on Christianisms,” which constitute a distinctive class of textual corruption (pp. xvi f.); and a precise note on “Conventions” (p. xviii). An “Appendix” (pp. 167–80) offers extensive corrections to a book and a book chapter by Holford-Strevens on Gellius.
The textual commentary in Gelliana is not limited to new proposals made by Holford-Strevens, but embraces a broad range of problems and points of interest. In effect, the commentary serves to explain the constitutio textus, which is no less important in a critical edition than the study of the sources.
An outstanding linguistic and stylistic sensibility enables Holford-Strevens to detect hidden problems in the text and to reopen well-known ones if the solution does not strike him as satisfactory. More often than Marshall, he is ready to abandon the majority of the authoritative early manuscripts in favor of a reading found in a minority of them, or in the codices recentiores, or in a distinguished outlier such as the Buslidianus, or in the indirect tradition of ancient quotations and florilegia. He is also more prone to reject the transmitted text altogether and to resort to conjecture. This skepticism towards the sources is balanced by prudence in choosing his remedies. There is a minimalistic quality to many of the textual interventions that he adopts, including his own conjectures: a letter or two of the text might be altered, a word may be replaced by a similar one, or a lacuna may be posited and a short word added in supplement. I have not come across any conjecture admitted in his edition that lacks a plausible iter corruptionis.
This careful and balanced editorial approach yields results of a very high quality. Here I shall discuss a small selection of editorial decisions.
At the end of book 1, chapter 12, Gellius quotes the jurist M. Antistius Labeo on the inheritances of the Vestal Virgins. In the lemma (1.12 cap.) this is summarized as ut Labeo dicit nec intestato cuiquam nec eius intestatae quisquam iure heres est. Here the genitive eius intestatae is awkward, especially after the dative intestato. Holford-Strevens considers two options: writing ei, the reading of manuscript C, which was also conjectured by Hosius; or deleting eius altogether, which is an idea of Holford-Strevens’ and is close to the text of the quotation from Labeo, Virgo Vestalis neque heres est cuiquam intestato, neque intestatae quisquam (Gellius 1.12.18 = Labeo, Iurisprudentiae anteiustinianae reliquiae ed. 6, fr. 24). The deletion of eius does not make it into Holford-Strevens’ text, but it is convincing.
In Marshall’s edition, the lemma to 4.14 starts Narratur historia de Hostilio Mancino aedilium. The last word is a conjecture of Cramer’s; aP read (a)edulium, which has been turned into aedili in the codices recentiores (or part of them at any rate). Holford-Strevens observes that ‘Hostilio Mancino aedilium is no more Latin than Cicero consulum or Cato censorum’ (Gelliana, p. 3) and conjectures aedili curuli, which is the form used by our accurate author in the text at 4.14.2 Aulus Hostilius Mancinus aedilis curulis fuit. At Gelliana p. 3, Holford-Strevens proposes a mechanism of corruption: ‘-um came in from 4.2. cap.’, which reads in edicto aedilium.
At 2.8.8, the manuscripts of class Ξ read conclusionem synlogismi, while A has coniunctionem synlogismi. Holford-Strevens rightly treats the two readings as equally well attested and chooses the latter on the basis of the principle lectio difficilior potior. He makes a plausible case for this reading in Gelliana (p. 35); but students of philosophy under the Roman Empire may have more to say about this matter.
At 2.14.1, Gellius reports a passage found in libro uetere M. Catonis, qui inscribitur Contra Tiberium exulem (Cato fr. orat. 202 Malcovati = fr. 184 Cugusi–Sblendorio Cugusi). Holford-Strevens observes that the “only Tiberius who is likely either to be called by praenomen or to be designated an exile is Augustus’ unhappy successor”; the “title therefore seems to be corrupt, though its appearance both in the lemma and here suggests that the corruption is pre-Gellian” (Gelliana, p. 36). The notion that there is something wrong with these words is not new, but it is helpful to have it spelt out so clearly, especially as it has bearing on the title of a lost speech of Cato the Elder’s. Holford-Strevens prints the words Tiberium exulem between cruces and quotes the view of Antonio Cima from 1903 that exulem (written as exule in part of the manuscripts) conceals a cognomen. If the corruption in the title indeed predated Gellius, then it may have been better not to obelize what he wrote; but in substance, Holford-Strevens’ position is convincing. Marshall on the other hand prints the transmitted text without indicating that it might be corrupt.
At 2.18.8, the form incelebris (αP), with a rare nominative plural ending, is well defended by Holford-Strevens (Gelliana, p. 38). In Marshall’s edition it does not even make it into the apparatus criticus.
At 14.7.13, Gellius reports from the lost Coniectanea of the Augustan jurist C. Ateius Capito the view of Q. Aelius Tubero that nullum senatusconsultum fieri posse non discessione facta, quia in omnibus senatusconsultis, etiam in iis quae per relationem fierent, discessio esset necessaria, idque ipse Capito uerum esse adfirmat; the readings of the manuscripts are all likely to stem from per relationem. Holford-Strevens notes that this phrase is an “error, certainly not Capito’s,” since the correct term for the procedure was per relationem discessionemque, as the relatio, the formal introduction of a motion, was followed by a discessio, a vote by division (Gelliana, p. 135). He explains how Gellius could have fallen into this error and why the conjectures proposed by Mommsen and Vassis are unlikely.
At 17.2.14 Gellius quotes one of his favourite authors, the Republican annalist Q. Claudius Quadrigarius (FRHist 24 F3), as writing et ueteres conpluscule ita dixerunt. conpluscule is the reading of the manuscripts; this hapax legomenon, not attested anywhere else, is translated by the Oxford Latin Dictionary as ‘fairly often, not infrequently’. Holford-Strevens emends without hesitation to conplusculei, which is an excellent conjecture: the adjective is well attested and easy to make sense of in this context, unlike the adverb, and the Republican ending -ei could well have been corrupted to -e.
In sum, this edition makes major steps forward in the study of the manuscripts and the reconstruction of the text of the Noctes Atticae. In part, this is due to the general progress of scholarship and technology. Latin manuscripts are easier to track down today than they were a few decades ago due to the publication of manuscript catalogues, the availability of digital research tools, and the work of scholars at the Institut de Recherche et d’Histoire des Textes and elsewhere. Manuscripts are easier to study from afar thanks to the availability of microfilms and (more recently) digital photographs, and research trips have become cheaper and more convenient than before (despite the momentary lapse caused by the Covid-19 pandemic). But the availability of these resources does not guarantee that someone will actually use them, or that this will be done with skill.
The late Martin L. West once pointed out “a common fallacy concerning the qualifications required” to prepare a critical edition of a classical text: “it is not a sufficient qualification to have a long-standing interest in it, to have written books or articles about it, in short, to be firmly associated with it in the public’s mind … A better policy for publishers, when they want a good edition of something, would be to look for someone who has done a good edition of something else, even if he has not hitherto concerned himself with what they want.” West’s maxim is proven wrong by Holford-Strevens, who has spent a lifetime working on the Noctes Atticae and is almost certainly the greatest authority on Gellius in the world today. His passion for the author, his deep knowledge of the text and his skill as a textual critic have helped to create a truly outstanding edition.
 The reviewer gratefully acknowledges postdoctoral research grant no. 116524 from the Hungarian National Bureau for Research, Development and Innovation and Ramón y Cajal grant no. RYC2018-024411-I of the Spanish Ministry for Science, Innovation and Universities, supported by the European Social Fund.
P. K. Marshall (ed.), A. Gellii Noctes Atticae, 2 vols., Oxford 1968. A case was made for its replacement by J. Velaza, “Hacia una nueva edición comentada de Aulo Gelio: la Praefatio de las Noctes Atticae,” Anuari de Filologia. Antiqua et Mediaevalia 2 (2012), 11–48.
 Here Holford-Strevens is for once too succinct. The Flemish scholar Ludovicus Carrio, who studied the Buslidianus in the sixteenth century, wrote that it was four hundred years old. Holford-Strevens suggests that Carrio could have misdated a fifteenth-century manuscript to the twelfth century. Indeed, fifteenth-century humanistic minuscule script from Italy can resemble certain twelfth-century hands, the fifteenth century produced countless contaminated manuscripts, and BUappears to have been strongly contaminated. But the twelfth century also produced many contaminated manuscripts; and if BU had been copied in twelfth-century Flanders, for example, then Carrio’s dating would be correct. That it is not is made more plausible by the readings shared by BU with the codices recentiores, which are mentioned briefly by Holford-Strevens (ed. vol. 1, p. xxxiv). It would have been helpful to have more detail about these shared readings and the manuscripts that contain them. If they only appear in fifteenth-century Italian manuscripts, then Holford-Strevens’ suggestion about the date of the Buslidianus is indeed plausible.
Incidentally, it is surprising to read that contamination should have occurred in one direction only: that many inferior readings have entered the Buslidianus from the recentiores, but none of the valuable readings of the Buslidianus have travelled in the opposite direction. Have the recentiores been checked in depth for possible influence of BU?
 The name A. Gellius must have been misinterpreted as Agellius at an early date. Augustine already refers to our author as Agellius, and Priscian includes among a list of substantives with the derivative suffix –ius the disconcerting sequence agellus agelli Agellius (Prisc. Inst. 18.4.21 = Grammatici Latini 2.135.14 Keil; cf. ed. vol. 1 p. ix).
 In the critical edition of the lemma to 1.18, Holford-Strevens writes M. Varro in quarto decimo diuinarum, accepting Bernardi Perini’s conjecture, where the manuscripts read humanarum. He explains his reasons in Gelliana, p. 2. But when the lemma is repeated after the chapter, we read in quarto decimo humanarum. Is this inconsistency a sign of a pentimento, a change of mind by the editor?
Gelliana, p. 10 has the amusing error lector brevior potior for lectio b. p.
Some further corrections: Ed. vol. 1, p. vii testim. 3 line 4: in Cecropiae noctes read Cecropias (cf. p. xiv). P. ix testim. 5(b), below, bis should surely be ter. P. 88 apparatus 1 line 2 “82§11” should be divided by an ample space: “82 §11.” P. 97, line 11 sol uitur should read soluitur. Ed. vol. 2, p. 513 line 2 de bere should be debere.
Gelliana p. xii n. 10 ‘Jan Van der Wouwer’: after a first name, Dutch usage requires lower-case ‘van’. P. 34, on 2.6.17, line 7: enim should not be followed by a quotation mark.
 L. Holford-Strevens, Aulus Gellius: An Antonine Scholar and his Achievements, Oxford 2003, reprinted with additions 2005; id., ‘Recht as een Palmen-Bohm and Other Facets of Gellius’ Medieval and Humanistic Reception’, in L. Holford-Strevens, A. Vardi (eds.), The Worlds of Aulus Gellius, Oxford 2004, 249–81.
 Holford-Strevens writes (at Gelliana, p. 11) that he ‘cannot explain the corruptions’ when two nouns appear to change gender at 1.2.7 atroque uerborum et argutiarum fuligine ob oculos audientium iacto (thus most manuscripts; atraque BU, iacta Beroaldus) and 18.4.11 origines … quos requisitos … et repertos (-as -as -as Rutgers). He may be too skeptical here; scribes commit many kinds of grammatical mischief.
 M. L. West, Textual Criticism and Editorial Technique applicable to Greek and Latin texts, Stuttgart 1973, 62.