Challenges of Editing Latin Patristic Texts: A report from inside the Corpus Scriptorum Ecclesiasticorum Latinorum

Google ‘Ars edendi’ on a computer in the Salzburg area, and one of the first sites to pop up is ‘Ars edendi – the art of eating’, by Monika Speier, a nutritionist from near Munich. Though eating and editing in Latin do not share the quantity of the first vowel, they do have some things in common.1 For example, an editor is used to making collations, and so does everyone in Italy when having breakfast, the prima collazione. To ruminate is to both chew food and to think over a problem again and again – something that everyone producing an edition knows very well from experience, and finally the delight that results from a nice meal can be similar to the delight a well-produced edition gives to its user. The enjoyment of well-produced editions is not the least outcome of the Ars edendi project at Stockholm University. This large and excellent project is now reaching its end, and has made editors better aware of the possibilities, aims, risks and limits which they encounter in the editing of medieval texts.

and patterns. 2 The period in question began around the year 200 and, depending on regional developments in the different regions of the former Roman Empire, ended between the late fifth and the early seventh century. 3 Classical philologists in the German speaking countries agree nowadays that the responsibility for editing and commenting on the Latin texts of that period, whether they be Christian or pagan, belongs to the field of classical philology, whereas previously only theologians studied the Christian literature of Late Antiquity. Moreover, that period was judged to be a time of deterioration and degeneration; researching it was regarded as investing time and energy in a culture that produced no art but only objects devoid of artistic value. 4 Defining historical periods may often create more problems than it resolves, it is true. We may even suspect that establishing periods means no more than projecting the opinions of contemporary art criticism 2 See Alexander Demandt, 'Die Spätantike als Epoche', in Spätantike. Mit  For a classicist specialising in editing, Late Antiquity has some interesting new features: due to historical and cultural circumstances that favoured textual transmission, the extant literature of Late Antiquity represents far more varied artistic levels than in the case of classical literature. We have highly ambitious literature, for example, panegyrics to emperors or the pretentious and sophisticated letters of Ennodius, as well as primitive literary products such as, as early as from the end of the fourth century, the so-called Peregrinatio Egeriae or Aetheriae, 6 a description by a nun named Egeria or Aetheria of her pilgrimage to the Holy Land, or monastic rules (Regula Magistri, Regula Benedicti, Regula Donati, etc.) from the sixth century onwards. In other and new aspects of everyday life a specific terminology developed, for example, the legal terminology necessary for the administration of the Roman Empire and the subsequent Germanic kingdoms, 7 or the homiletic terminology. Each of these new developments generated specific customs and language patterns. Furthermore, borders between literary genres, formerly neatly separated, began to be blurred. Remember, for example, the Satura Menippea as used by Boethius in his Consolatio Philosophiae by which he refines the De nuptiis Mercurii et Philologiae by Martianus Capella. 8 New genres appeared, as can be seen in Augustine of Hippo's Confessiones, which combine biographical elements and Bible exegesis. For some of these shifts, the emergence of Christianity seems to have been decisive. Thus, simple style and language were made respectable by the sermo piscatorius of the Bible. The genres of speech before a court or before a political gathering transformed into homilies, and the philosophical dispute eventually became a dispute against heretics.
It is evident, however, that in many respects Christian literature underwent the same changes as the non-Christian literature of the time. In fact, the distinction between them is nowadays judged as inappropriate because both reflect their position in regard to classical antiquity and recur to it in an almost identical way. This similarity results from the uniform institutional schools which were spread all over the Roman Empire. Since the contents of the curricula remained mostly as they had been and had not been Christianised, there was only one canon of literature and educational subjects that could be taken as default from Spain to Illyria and from North-Africa to Gaul. 9 The contacts with Greek literature of that time, however, gradually became looser, as the knowledge of Greek declined.

The Foundation of CSEL in 1864
The Corpus Scriptorum Ecclesiasticorum Latinorum (CSEL), therefore, has not limited itself to editing Christian Latin texts of Late Antiquity on the assumption that there is any essential difference in editing these texts and editing classical texts. On the contrary, the limitation to Christian text of Late Antiquity is due to the state of classical scholarship around the middle of the nineteenth century, when the long-term project of the CSEL was founded. 10 13 The catalogue, it is true, has been a useful tool for editors, although we are now far more cautious in attributing the highest value to the oldest manuscripts. I will come back to this. The editions of Sulpicius and Arnobius however are not typical for the work done at the CSEL. Especially with the big names, i.e. Ambrose, Jerome, Augustine and Gregory the Great, but even with others such as Juvencus' biblical epos, an editor has to deal with innumerable text witnesses. This is one of the major challenges we have to face. I will also return to this. In sum, it has become evident, I think, that the methods of the so-called New Philology 14 are not adequate for our texts, which in most instances are not texts that underwent systematic rescription, quite the contrary: they were held in esteem as authoritative and were thus not adapted freely. As a consequence, we generally apply stemmatic methods in our editions, even when it is impossible to design a stemma, as happens with text traditions that remain hopelessly contaminated.

Glimpses of the Work done at CSEL
I am now going to address some problems we have encountered in our most recent editing projects, some experiences we had and some solutions we found. I will do this without any systematic order and, of course, without the intention of completeness. Rather, every edition has challenges and problems of its own. Whereas nowadays editors have at their disposal digital images of almost all manuscripts, our predecessors had to content themselves with microfilms. About 50 years ago, however, Rudolf Hanslik, at that time head of the CSEL, was in a far better situation since, when he ordered a microfilm of a Regula Benedicti codex from a library in Southern Italy, after a few weeks he received the manuscript itself in the post. In any case, the editors of Augustine, at that time the focus of the CSEL, in particular had to invest much time and energy in investigating the pertinent manuscripts. Therefore, Hanslik founded a series of publications intended to increase the speed of production and also the reliability of the editions. The series offers catalogues of manuscripts that transmit works attributed to St. Augustine. The series thus goes so far as to include writings falsely ascribed to Augustine. Starting with Italy, the series now comprises eleven volumes, thereby covering almost all European countries except France. 15 18 To make an editio princeps of a sermon by an author as famous for his preaching as Augustine is a fascinating task. We have tried to reconstruct when, where, and under what circumstances the sermons were delivered, and to analyse their characteristics, the lines of thought and the structure of the arguments. We have had to deal with problems of textual criticism, it is true, but at the same time we had to help the readers to understand the spirit and temperament of the text by punctuating and organising it in paragraphs.
Remarkably enough, new texts have been discovered not only in very old manuscripts. To give two examples, Brian Møller Jensen discovered a new sermon by Augustine in a manuscript in Piacenza from the twelfth century, 19 and the above-mentioned codex found in Marseille with the Epistulae Divjak originated as late as the fifteenth century. Thus, it becomes evident that we might miss new texts or at least good variants if we confine ourselves to the oldest text witnesses alone. Moreover, with pre-Carolingian manuscripts it may be the case that the texts they contain are even more corrupted than those that underwent correction in accordance with the Carolingian reforms. Even the famous Codex Petripolitanus from the early fifth century transmitting Augustine's De Schiller, Vienna 2009 (ibid. 791 doctrina Christiana has some evident blunders, although it was written during Augustine's lifetime. 20 Yet with the works transmitted to us in some old as well as in many younger manuscripts, it may prove impossible to collate all of them. In such cases we take sample collations in order to identify those manuscripts that transmit variants not attested by the older ones, since the younger manuscripts may be copied from older ones now lost. To establish text families, it may be useful not only to look at variant readings but also at the corpora of texts contained in the manuscripts, since the arrangement of texts was often retained. But even indirect text transmission may be very valuable. For example, when the Venerable Bede has a variant reading in a quotation from Augustine, we know that this variant is pre-Carolingian that may be lost in the extant Augustine manuscripts. At the same time texts which we are working on may cite others so that they attest variants not preserved by direct text transmission. Handling facts of this sort requires complex methodology and can lead to interesting results. I will present examples taken from three of our projects and which differ in their sets of problems and in the appropriate strategies for resolving them. I have deliberately chosen examples on a macro scale and others that pertain to single words only. The most spectacular example is the recent discovery of the commentary on the Gospels by Fortunatianus, 21 bishop of Aquileia in Northern Italy around the middle of the fourth century. Until then only three quotations of the text were known, two of them stemming from an exegetic compilation handed down in a manuscript from around 1100 in Troyes, the other from an exegetic anthology in a ninth century manuscript in Angers. In each of them the quotation is attributed to Fortunatus or Fortunatianus respectively. 22  the digital library of the Dombibliothek at Cologne, he stumbled over a Gospel commentary lacking any indication of its author in codex 17 from the early ninth century. The text however could easily be dated to Late Antiquity, because quotations from the bible had the pre-Vulgate wording. Dorfbauer spotted the three quotations mentioned (cf. note 22), each of them in its appropriate context. Since the language of the text and its wording have striking characteristics, he was able to attribute the text beyond any doubt to Fortunatianus. Interestingly enough, the text seems to have been very much appreciated by later writers: it was cited quite frequently, but because the citations do not give any author, only now could they be identified. 23 The constitutio textus becomes a really intricate task when the text to be edited consists of excerpts taken from other and earlier texts, as is the case of the monastic rule for nuns written by Donatus in the middle of the seventh century. This rule consists mostly of excerpts from the monastic rules of Caesarius of Arles, Columbanus and Benedict, which, apart from slight adaptations in the grammatical gender or number, are cited ad verbum. In the case of Donatus' rule, the main line of textual transmission, which consists in a single manuscript, the so-called Codex regularum, develops alongside with an extensive secondary transmission. However, the editor, Victoria Zimmerl-Panagl, 24 took not only into account the wording of the preceding rules from the available editions but also all their variant readings, because it was impossible to identify the manuscripts used by Donatus. One textual problem which, though inconspicuous, can serve as an example of intertextuality as well as paratextuality is found in ch. 64 of Donatus's rule which consists of a single sentence by which, it seems, headgear is regulated. Or the other way round, did the scribe correct the word in the rule of Caesarius according to the rule of Donatus? Or does incatum occur in both rules only because this was the form the scribe was accustomed to, and incaustum is correct in both rules? In this case, the editor indicated with an asterisk in the critical apparatus that the decision to print incatum is still doubtful. 26 Besides, some manuscripts display a line in the margin of that paragraph. We can guess that it was meant to indicate the height of the headdress. Alas, the length of the lines differs from manuscript to manuscript. In the Codex Regularum, it is 3.25 cm in the Donatus text and 5 cm in the Caesarius text. In another Caesarius manuscript in Tours which is now lost, it is said to have been 9.5 cm. 27 During my own work on Augustine's Contra Iulianum, I encountered a somewhat similar problem. Augustine wrote this work in the last decade of his life, as part of the discussion he had with Julian of Aeclanum, a follower of Pelagius, on original sin and on whether men can be completely free from sin or not. The first two books contain a collection of quotations from patristic writings that are meant to support the doctrine of original sin. In some of these quotations, the manuscripts of Contra Iulianum have the very same variant readings that can be found in the manuscripts transmitting the cited works. Some of these accordances might have come about by chance, it is true. For example, in Contra Iulianum 2,10 a quotation from Ambrose, In Lucam 7,142 is inserted that does not match exactly what seems to be the original text: 25 'They, i.e., the nuns, are not allowed to bind their heads higher than is indicated here with ink.' 26 In the preface, the editor discussed this problem at p. 114f. 27 Ibid., p. 114.
Augustinus, Contra Iulianum 2,10: Rursus in eodem opere, cum de spiritali atque incorruptibili loqueretur (scil. Ambrosius) cibo: "Etenim misericors cibus mentis est, inquit, praeclaraque alimonia suavitatis, quae membra non oneret neque in naturae pudenda, sed ornamenta convertat, cum libidinum volutabrum commutatur in dei templum diversoriumque vitiorum sacrarium incipit esse virtutum. Instead of oneret, one Ambrose manuscript as well as several of Augustine's Contra Iulianum have haeret which does not make sense. The error might have its origin in the preceding non or rather in its abbreviation, n , and it may have come about in both text traditions independently (n oneret misread as non eret). This type of explanation, however, is not appropriate for the other textual problem in the same passage. Since all the manuscripts of Augustine's work have etenim misericors cibus mentis est as the first words cited, only a single codex, though one of the oldest extant manuscripts, Orleans, Bibl. mun. 162 from the ninth century, cancelled out misericors and wrote ratio instead, which surely is the original reading in Ambrose. Most of the Ambrose manuscripts have ratio, few others have oratio or rationibus, but none have misericors or any other word which might even faintly resemble misericors. From this evidence we may conclude that misericors in Augustine is a varia lectio already present in Late Antiquity and that the scribe who corrected misericors to ratio in the Orleans manuscript did so because he had a manuscript of Ambrose at hand. In Contra Iulianum the correct text is therefore misericors, 29 though 28 CSEL 32/4: Ambrosius, Expositio evangelii secundum Lucan, by Karl Schenkl (Wien: Österreichische Akademie der Wissenschaften, 1902). 29 Giovanni Paolo Maggioni discussed substantially similar problems in Iacobus' de Voragine Legenda aurea, see: Barbara Crostini -Gunilla Iversen -Brian Jensen (ed.), from the viewpoint of the Ambrose text this word is not what Ambrose wrote.
As already said, a good edition also must make the structure of the text visible. This can, but need not, be part of the constitutio textus in its traditional meaning. In this context, I am going to give examples from a large-scale project of the CSEL, the edition of Augustine's corpus of the Enarrationes in psalmos. 30 I will not, however, deal with problems connected to the orality of the Enarrationes which consist largely of homilies. A preacher may, of course, make anacolutha and mental leaps, or it is possible that during the homily he spontaneously departs from his draft due to an unforeseen event -something that is normally difficult to reconstruct. 31 All of this is a major challenge for an editor, it is true. However, I just want to present one passage from Enarratio in Psalmum 65, 2-3 (commenting on ps. 65, 1-2: [1] Iubilate deo omnis terra [2] psallite autem nomini eius) in order to show the extent to which punctuation and the insertion of paragraphs determines the character of the text. Many Enarrationes follow one and the same pattern: first a verse from the psalm -let it be called verse A -is cited, commented upon, and cited for a second time. Then the same pattern is applied to verse B: citation, commentary, citation, and so on. This is precisely the way the text of the Enarrationes is structured in vol. 36 of the monumental, though uncritical edition the Maurists made in Paris towards the end of the seventeenth century which was, with minor changes, reprinted in CC.SL 39: (2 Again, it is worth noting that only one manuscript from the eleventh century has preserved part of the original reference system. If the editor had considered only the manuscripts predating the year 1000 -about 20 from before the eleventh century -he would not have been able to reconstruct the original arrangement of the text.

The organisation of the CSEL
Up to 2012 the CSEL was funded and the editions were published by the Austrian Academy of Sciences. Since 2012 the CSEL has been part of the Department of Classics at Salzburg University, and the publisher is De Gruyter. The office, however, is still in Vienna. Our focus is on Augustine and Ambrose, but only as a guideline. Thus, within recent years, we have also published editions of an anonymous commentary on Job, two volumes on Prosper of Aquitaine, and the volume on Donatus mentioned above. The CSEL has volumes edited by the staff as well as by external editors. External editors receive advice and supervision whenever needed. Each volume is reviewed first by the CSEL staff, then by the advisory board, before it goes into print. Last but not least, the staff regularly gives courses on palaeography and editorial work in order to hand down the relevant skills and ultimately the pleasure of producing editions to the next generation of researchers.