Fernandez ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 2011 Tomás Fernández Remarks on Editing a Byzantine Anthology Tomás Fernández HESE OBSERVATIONS, although intended to be general in scope, arose from the preparation of a critical edition of a particular Greek anthology of the late ninth or early tenth century, the Florilegium Coislinianum (FC).1 They may help to bridge a gap in scholarly literature because, while there is plenty of excellent material concerning general editorial technique, none specifically addresses the issues of textual crit- icism associated with compilations, such as the FC.2 Therefore, 1 The basic reference on this anthology is still M. Richard, “Florilèges spirituels grecs,” in Dictionnaire de spiritualité 5 (Paris 1962–1964) 484–486, reprinted in M. Richard, Opera minora I (Turnhout/Leuven 1976) no. 1. See also I. De Vos, E. Gielen, C. Macé, P. Van Deun, “L’art de compiler à Byzance : La lettre Γ du Florilège Coislin,” Byzantion 78 (2008) 159–223, with full bibliography. A second article, by the same authors, is already finished: “La lettre B du Florilège Coislin: editio princeps,” Byzantion 80 (2010) forthcoming. For a succinct description of the FC cf. also T. Fernández, “Un auteur inconnu dans le Florilège Coislin: Léonce de Damas,” Sacris Erudiri 47 (2008) 209–221, and “The Florilegium Coislinianum and Byzantine Encyclopaedism,” in S. Neocleous (ed.), Papers from the First and Second Post- graduate Forums in Byzantine Studies. Sailing to Byzantium (Newcastle upon Tyne 2009) 127–144. For a more general overview see H. Chadwick, “Flori- legium,” RAC 7 (1969) 1131–1160, with a selective bibliography of older scholarship. 2 See however the wise observations of P. Odorico, Il prato e l’ape. Il sapere sentenzioso del monaco Giovanni (Vienna 1986), esp. 55–58. Francesca Mal- tomini was kind to send me her “Selezione e organizzazione della poesia epigrammatica fra IX e X secolo: la perduta antologia di Costantino Cefala e l’Antologia Palatina,” Encyclopaedic Trends in Byzantium? (forthcoming). Her methodological remarks about the use of the sources when dealing with a T 168 REMARKS ON EDITING A BYZANTINE ANTHOLOGY ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 this contribution may be of some service to editors and readers of anthologies, especially in what concerns the relationship between the primary source and the compilation that quotes it. This paper is concerned with the edition of anthologies proper, not with the use of anthologies in the preparation of critical editions of the fragments quoted therein. In what follows, I use the words “anthology,” “florilegium,” and “compilation” synonymously. 1. The editorial process for an anthology differs from that of a non-compilatory work in many respects. One of the most fundamental is that the editor will often have at his or her disposal, in addition to the manuscript tradition of the an- thology itself, the source used by the compiler.3 The source will allow the editor to perform the task with greater accuracy. The editor, however, must be careful and avoid the risk of relying on the source excessively, as though the reconstruction of its text, rather than that of the anthology, were the real goal of the edition. The two basic principles that will be developed in the present contribution are that, on the one hand, the text of the source has to be taken into account—which of course does not imply one must necessarily follow it—at all times, even when it contradicts the whole manuscript tradition of the anthology;4 ___ poetical compilation like the Anthologia Palatina have been very instructive to me. And then of course much can be learnt by looking closely at the actual practice of the editors of anthologies, like Hense and Wachmuth’s edition of John Stobaeus’ Anthologion. 3 As far as editing is concerned, not only a compiler but any writer, in- asmuch as he reproduces more or less faithfully the text of others, can be put on a level with a compiler. 4 For example, some editors of a florilegium may decide to supply a group of words fallen by a saut du même au même (or any corruption easy to explain palaeographically), even if this was an error of the archetype and, consequently, was present in all the manuscript branches of the anthology. Some others will decide to place a crux. Both options are valid. It must be borne in mind, however, that the editor who supplies the missing words will not do so because the source had them, but because the original anthology also most likely did so—at least in the intention of its compiler. TOMÁS FERNÁNDEZ 169 ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 on the other hand, that the text of the source cannot be used to “correct” the manuscript tradition of the anthology itself, and that it must be used only to attempt to get closer to the original reading of the anthology. In order to define a criterion to determine in which cases, and to what extent, it is valid to utilize the source for the estab- lishment of the text of the compilation, I have endeavoured to place the source in the stemma of the anthology ( fig. 1). This could seem misleading, for it could be understood as an im- plication that the source is placed on the same level as the manuscript tradition of the anthology, or, worse still, “above” it, as if in the case of divergence its reading is to be preferred over that of the anthology. The method used herein is rather to the contrary, for, as we shall see, the source has no autonomy with respect to the anthology, and consequently its readings will be useful only in relation to those of the anthology. Throughout, the perspective of this paper is that of the editor of an anthology, which differs in many regards from the perspective of the editor of the primary source. The latter must apply a different methodology, since for him or her the an- thology will be a hyparchetype whose reconstruction will be of use only to the extent it may help reconstruct the archetype of the source itself. For the editor of the anthology, by contrast, the source used by the compiler5 is what I will call a hyper- archetype or superarchetype. This hyperarchetype is inversely symmetrical to a hyparche- type, because its reading is useless per se, except to the extent that it helps determine the reading of the archetype (of the an- thology). For example, see the model below: the reading of ω or β will only be useful insofar as it helps determine the reading of Ω. Thanks to this hyperarchetype, the editor can on occasion attain the reading of the autograph and therefore get beyond 5 The “source used by the compiler” may coincide with the archetype of the source, with the hyparchetype used by the compiler, or even with the single manuscript he copied. 170 REMARKS ON EDITING A BYZANTINE ANTHOLOGY ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 the archetype, which in these cases, of course, should no longer be called “archetype.” Actually, as we shall see, the source pro- vides information about the autograph directly, and about the archetype by extension. The following model includes a simpli- fied stemma of the FC, embedded in the larger stemma of an imaginary source upon which the FC depends. The source, in its position as a hyperarchetype, can provide a window on the autograph of the compiler (ψ), and of course, a fortiori, also on the archetype of the anthology (Ω). If AT or Cπ share a reading that cannot be explained by polygenesis or contamination with ω, β, or both, then it is certain that this was the reading, not only of the archetype of the anthology (Ω), but also of the autograph (ψ).6 The foremost practical consequence of this distinction is that emending any such reading—which on occasion might be needed—would imply intervention in not only the archetype of the anthology, but also the autograph of the compiler. In any such cases, the editor can also be sure that this reading, even if attested in only one branch of the tra- dition, is the correct one. Of course, he or she must have ex- cluded the possibilities of contamination, double readings present in the archetype, coincidences by conjecture, and other such pitfalls.7 6 This situation may be exemplified with the textual tradition of Catullus, ideal for its simplicity: one branch is represented by the manuscript O, the other by the hyparchetype X, in turn constituted by manuscripts GR. If any reading of G or R coincides with O, it is certain that the hyparchetype X had this reading (and also the archetype). 7 An example from the FC is at 19.25, Μωϋσέως <ᾧ> πιστεύειν φασίν; all MSS. but T omit the ᾧ (which the source also had). Some of the manuscripts of the hyparchetype π corrected the genitive to Mωϋσῆ or Mωσέα. Here it is clear that even if T had the same reading as the source, this was not the reading of the archetype; the scribe must have found it by conjecture. Of course, there is no way to determine which the reading of the autograph was. TOMÁS FERNÁNDEZ 171 ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 Figure 1 ψ = autograph of the compiler of the FC; Ω = archetype of the FC; Α, Τ, C = witnesses of the FC (A = Paris. Coislinianus gr. 294 [11th–12th c.], T = Hierosolym. Sancti Sepulcri 15 [10th–11th c.], C = Paris. gr. 924 [10th c.]); π = hyparchetype of the FC (containing what M. Richard called the “third recension”8 of the anthology); ω = hyperarchetype of the FC (= archetype of an imaginary source quoted by the FC); α, β, γ = hyparchetypes of the source (note that β is at the same time hyperarchetype of the FC) 8 Richard (n.1 above). This third recension comprises, among others: Mediolanensis, Ambrosianus Q 74 sup. (10th c.); Argentoratensis, Bibl. Nat. et Univ. gr. 12 (a. 1285–1286); Atheniensis, Bibl. Nat. 329 (13th–14th c.); Athous, Iviron 38 (a. 1281–1282)’ and Vat.gr. 491 (13th c.). Note that the FC can sometimes be proved to belong to a given hyparchetype without, however, being directly issued from any of the extant manuscripts of the tradition. Frequently in fact, all of the manuscripts of the branch of the tradition upon which the FC is dependent are more recent than the FC. C T A π Ω V Ω * * ω ψ α β γ 172 REMARKS ON EDITING A BYZANTINE ANTHOLOGY ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 A note on “original,” “autograph,” and “archetype.” An autograph (thus ψ in the stemma above) is historical as opposed to the ideal “original” the editor might reconstruct by emending the manuscript tradition of the author, not with the help of any independent textual evidence such as a primary source or a statement of the author,9 but on the basis of the rules of Greek grammar, the modus scribendi of the author, etc. The autograph, being historical, is what the author effectively wrote. In most cases, of course, the autograph has been lost forever. For example, consider a mistake in the autograph preserved in the archetype and all subsequent copies: if it is not attested inde- pendently, it will be impossible to show that it stems from the author himself and, following the general rules, it will be emended. In this manner, it cannot be asserted that an error was present in the autograph, even if the unanimous textual tradition points in this direction. Of course, we cannot know for sure which mistakes ancient and medieval authors made, which is not so regrettable; yet not only mistakes, but many other peculiarities of the autograph are necessarily lost in the leap from the archetype—or the text that we can reconstruct from the textual evidence—to the corrected text that the editor prints. This corrected text witnesses the attempt of the editor to reconstruct the original of the author, and this original, despite always being ideal to a certain degree, is even further from his- torical reality in the case of an author who cannot be assumed to consistently have chosen the lectio optima,10 as is often the case 9 A classical example is Cicero’s remark in an epistle to Atticus (6.2.3): he states that he indeed made a mistake writing Phliuntii, but that he immedi- ately corrected it into Phliasii. The only extant manuscript, a palimpsest, has the “wrong” reading. Thus, and thanks to Cicero’s observation, we can be sure that the “error” was the actual first reading of the autograph, and consequently, however mistaken it may be, is by no means a corruption. 10 This assertion follows the usual principle that textual certainty can be attained only by authors whose literary abilities are trusted; as Fränkel pointed out, “ ‘Der Autor (so lautet unsere Arbeitshypothese) weiß immer am besten was zu sagen ist und wie es zu sagen ist.’ Die Annahme gilt aber klärlich nur für Texte von einer relativ hohen literarischen Qualität. Alle TOMÁS FERNÁNDEZ 173 ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 with compilers. As for the distinction between the autograph and the arche- type, it may not be of great practical use, especially in works whose autograph is very close to the archetype. Still, the editor can emend the latter more freely than the former. 2a. “Error” of the anthology also present in the source. The source may show with absolute certainty that a given error was pres- ent already in the autograph of the anthology. For my pur- poses, it does not matter if the mistake was in the archetype of the source, or only in the branch of the manuscript tradition used by the compiler of the anthology. I quote an example from the FC (22.7–20): […] ἕκϰαστος τῶν γενοµένων ἀπὸ µοναδικϰῆς ἦν οὐσίας […] τὰ ἄνω πάντα ἀσώµατα, τὰ κϰάτω πάντα σώµατα· τὰ ἄνω νοερϱά, τὰ κϰάτω αἰσθητά· τὰ ἄνω ἀόρϱατα, τὰ κϰάτω ὁρϱατά. πάντα σώµατα τὰ κϰάτω, κϰαὶ οὐρϱανός, κϰαὶ ἥλιος, κϰαὶ σελήνη […], πάντα σώµατα, πάντα ἁφῇ ὑποβαλλόµενα κϰαὶ ὄψει ὁρϱώµενα. τὰ ἄνω πάντα ἀόρϱατα, νοερϱά, ἄγγελοι, ἀρϱχάγγελοι, θρϱόνοι, κϰυρϱιότητες, ἀρϱχαί, ἐξουσίαι, δυνάµεις, τὰ Χερϱουβίµ, τὰ Σερϱαφίµ. οὔτε ἐκϰεῖνα ὁρϱατά, οὔτε ταῦτα ἀόρϱατα· οὔτε ἐκϰεῖνα σώµατα, ἀλλ᾽᾿ εἰ κϰαὶ ταῦτα σώµατα,11 πάντα ἐκϰ µοναδικϰῆς ἐγένετο οὐσίας, κϰαὶ τὰ ___ anderen entziehen sich ihr, und entziehen sich damit auch der Möglichkeit einer einigermaßen verläßlichen Textkritik, ebenso wie der Möglichkeit einer verläßlichen Interpretation”: H. Fränkel, Einleitung zur kritischen Ausgabe der Argonautika des Apollonios (Göttingen 1964), 139. Nevertheless, the prin- ciple that the author always knows what the best way to say something is, remains wholly unhistorical. Furthermore, the idea of what is “best” is anything but objective. The dictum si melius est, Catullianum est may be convenient for the editor, but has no claim to historical accuracy. (For this dictum see J. M. Trappes-Lomax, Catullus. A Textual Reappraisal [Swansea 2007], passim.) However, it is even less historically tenable in the case of authors who do not necessarily write in the way that seems best to the editor schooled in the rules of Classical and Patristic Greek. 11 Instead of ἀλλ᾽᾿ εἰ κϰαὶ ταῦτα σώµατα, the source reads οὔτε ταῦτα ἀσώµατα, ἀλλὰ κϰαὶ ταῦτα σώµατα, κϰαὶ ἐκϰεῖνα ἀσώµατα. It is easy to see that its text is no better than that of the FC. The compiler of the FC possibly corrected its exemplar here—something which he rarely does. In this and similar cases, however, the fact that no critical edition is available makes 174 REMARKS ON EDITING A BYZANTINE ANTHOLOGY ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 κϰάτω κϰαὶ τὰ ἄνω· κϰαὶ τὰ µὲν ἀπὸ σωµάτων, τὰ δὲ ἐξ ἀσωµάτων· τὰ µὲν σώµατα, τὰ δὲ ἀσώµατα. µόνος ὁ ἄνθρϱωπος ἀπὸ διπλῆς ἐγένετο οὐσίας […].12 The fragment states that everything, despite being of a unitary essence, falls into one of two classes: either that of the in- telligible, incorporeal, etc.; or that of the visible, corporeal, etc. The underlined section can be translated: “Neither are those things visible, nor these invisible. And the former are not bodies, but even though the latter are bodies, they all consist of a unitary essence.” The idea is clear. The problem, however, is that ἐκϰεῖνα should refer to the corporeal entities, which were mentioned in the first place, and the ταῦτα to the intelligible ones. This was felt by the scribe of the hyparchetype of EK, who, instead of οὔτε ἐκϰεῖνα ὁρϱατά κϰτλ., wrote οὔτε ἐκϰεῖνα ἀόρϱατα, οὔτε ταῦτα ὁρϱατά· οὔτε ἐκϰεῖνα ἀσώµατα, οὔτε ταῦτα σώµατα […]. In the source, ἐκϰεῖνα instead of ταῦτα and vice versa (or ὁρϱατά instead of ἀόρϱατα and vice versa) possibly resulted from a genuine mistake. Therefore, the editor of the source could, possibly, correct it legitimately. The compiler of the FC, by contrast, transcribed the “mistake” instead of correcting it. This does not imply that he fully decided his text should be like this, but by leaving the mistake unchanged, he confirmed it. The distraction, if it was a distraction, had to be greater. Since it was already present in the autograph itself, emending this reading would imply correcting the autograph of the compiler himself—not only the text of any given scribe. A similar case is that of “corruptions” detected in the primary source which are to be found tel quel in the anthology. In his edition of Nemesius, for example, Morani followed an Armenian and a Latin translation and corrected πρϱὸς θερϱαπείαν τῆς ἐξ αὐτῶν ἐκϰείνων βλάβης κϰαὶ τῆς τῶν ἄλλων ἀρϱρϱωστηµάτων ἰάσεως (FC 29.388–390 = ed. Morani p.14.10– ___ any conclusion provisional. 12 = Ps.-Io. Chrys. In illud: Sufficit tibi gratia mea, PG 59 508.46–509.2. TOMÁS FERNÁNDEZ 175 ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 11), turning τῆς … ἰάσεως into πρϱὸς τῶν ἄλλων ἀρϱρϱωστηµάτων ἴασιν. This might be accepted in an edition of Nemesius, if the editor is convinced it is a genuine error.13 The editor of the FC, on the contrary, could never emend the text in this way, even if it improves the syntax of the sentence, for it is clear that the autograph had the same reading as the manuscript tradition of Nemesius (whereas the original of Nemesius might have had the reading as corrected by the editor). Instances of such errors, present both in the anthology and in the source, are rare, because they are usually corrected either by the scribes or by the editors.14 The situation described below —when the compilation and the source coincide in a good reading—is much more frequent. 2b. Correct reading of the anthology also present in the source. One example may suffice to show that, if a good reading was pres- ent both in the source and in the manuscript tradition of the anthology (or a branch of it), it can be assumed that this reading was already present in the autograph of the compiler: κϰαὶ ταύτην ἀφθόνως εἰς τὰ δεύτερϱα δευτερϱοφανῶς […].15 Only manuscripts AT have βαθύτερϱα instead of δεύτερϱα. If, as I argue, AT are one branch of the tradition, against Cπ, which 13 It is not even clear that in an edition of Nemesius the text should be emended. Anyone translating would possibly correct what in Greek, taken too literally, may be wrong; the evidence of the translations might thus not reflect any primary reading. Since the text is never incomprehensible, emendation is not altogether necessary. 14 In a critical edition, this would not be a problem because the diver- gences would be presented in the critical apparatus. 15 Quoted in full: ἀλλὰ κϰαὶ νεφέλης αὐτοῖς ἰδέαν ἡ θεολογία περϱι- πλάττει, σηµαίνουσα διὰ τούτου τοὺς ἱερϱοὺς νόας τοῦ µὲν κϰρϱυφίου φωτὸς ὑπερϱκϰοσµίως ἀποπληρϱουµένους, τὴν πρϱωτοφαῆ δὲ φωτοφάνειαν ἀνεκϰποµ- πεύτως εἰσδεχοµένους, κϰαὶ ταύτην ἀφθόνως εἰς τὰ δεύτερϱα δευτερϱοφανῶς κϰαὶ ἀναλόγως διαπορϱθµεύοντας κϰαὶ µὴν ὅτι τὸ γόνιµον αὐτοῖς κϰαὶ ζωοποιὸν κϰαὶ αὐξητικϰὸν κϰαὶ τελειωτικϰὸν ἐνυπάρϱχει κϰατὰ τὴν νοητὴν ὀµβρϱοτοκϰίαν τὴν τὸν ἐκϰδόχιον κϰόλπον πιαλέοις ὑετοῖς ἐπὶ ζωτικϰὰς ὠδῖνας ἐκϰκϰαλου- µένην. FC 5.165–172 = Ps.-Dion. Areop. De caelesti hierarchia p.56.7–13 (ed. Heil-Ritter). 176 REMARKS ON EDITING A BYZANTINE ANTHOLOGY ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 are the second branch, how could it be explained that the second branch had δεύτερϱα, just as the source did, if this were not also the reading of the autograph and the archetype of the FC? The hyparchetype of both A and T already had βαθύτερϱα, as either mistake or innovation16—unless it is postulated, as Morani did,17 that there are alternative variants which can be traced to the archetype; this suggestion, however, is entirely unlikely. 2c. Shared readings. As these examples have made clear, it does not matter whether the reading shared by the source and the anthology is correct or mistaken; be that as it may, it can be proved to have been present in the autograph of the anthology. I offer three final examples of lections which may be taken indifferently to be an error or a correct reading. FC 28.2–4: γνῶθι σαυτὸν τίς εἶ – διπλοῦς ἄνθρϱωπος κϰαθέ- στηκϰας, ἐκϰ ψυχῆς κϰαὶ σώµατος συγκϰείµενος –, κϰαὶ ὅτι […] ὁ αὐτὸς θεὸς κϰαὶ ψυχῆς κϰαὶ σώµατός ἐστι δηµιουρϱγός. The source has: γνῶθι λοιπὸν ὅστις εἶ, ὅτι διπλοῦς ἄνθρϱωπος κϰαθέστηκϰας, ἐκϰ ψυχῆς τε κϰαὶ σώµατος συγκϰείµενος, κϰαὶ ὅτι […] ὁ αὐτὸς θεὸς κϰαὶ τῆς ψυχῆς κϰαὶ τοῦ σώµατός ἐστι δηµιουρϱγός.18 The 16 It would be a different matter if we believed that A and T are two independent branches, as do De Vos et al., Byzantion 78 (2008) 166. In that case, it would be necessary to postulate that the reading of Cπ was due to contamination or conjecture (most unlikely), or that A and T independently committed the same error or innovation (very unlikely as well). At any rate, the basic principle would not be affected because either the common error of AT would have been due to polygenesis, or the “good reading” of Cπ (which would actually be a mistake from the perspective of the FC ) would have been gotten by contamination or diuinatio. In any of these cases, co- incidence of any witness of the anthology with the source does not prove anything. To explain the bizarre corruption of δεύτερϱα into βαθύτερϱα, Jacques Noret has suggested to me that possibly some scribe wrote βʹ´τερϱα (βʹ´ being a numeral), which could easily have been wrongly interpreted as βαθύτερϱα. 17 M. Morani, La tradizione manoscritta del “De natura hominis” di Nemesio (Milan 1981) 31. 18 Cyrillus Hierosolym. Catecheses ad illuminuandos 4.18.2–5 (ed. Reischl). TOMÁS FERNÁNDEZ 177 ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 first ὅτι is omitted, not only by the FC, but also by the branch of the manuscript tradition of Cyrillus to which the FC is re- lated. The latter observation shows that the FC did not have this ὅτι at any stage of its composition. It could be argued that the omission of ὅτι in a branch of the manuscript tradition of Cyrillus does not make any difference, since also without this there would have been no need to emend the text; indeed, it was comprehensible without the ὅτι. The difference, however, lies in what the omission of the ὅτι makes clear, that even from a genetic point of view, the FC never had it, and in fact its autograph did not have it. If the source had had the ὅτι in all its branches, this would not be so clear. FC 102.3–4: κϰαὶ τὴν ληστρϱικϰὴν τῶν Ἰουδαίων ἔφοδον FC, κϰαὶ τὴν ληστρϱικϰὴν ἔφοδον τῶν Ἰουδαίων source.19 Three wit- nesses of the source have the same reading as the FC. Since the FC is probably related to the same family as these three wit- nesses, it can be established that the reading shared by these three MSS. and the FC originated beyond the archetype of the FC, and thus that it was already present in the autograph of the FC. It was neither an error nor an innovation of the compiler; the text he copied out already had it. (Needless to say, even if this were not the case, and no witness of the source had the same reading, the reading of the FC should not be emended.) FC 102.31–33: εἶτα ἡ πρϱὸ τοῦ σαββάτου ἡµέρϱα, ἔχεις τὰς τρϱεῖς νύκϰτας κϰαὶ τὰς τρϱεῖς ἡµέρϱας. This is the reading of T. Instead of ἔχεις, A had ἔχει, and all the other manuscripts εἰς. The text of the source is: εἶτα ἡ πρϱὸ τοῦ σαββάτου <νὺξ κϰαὶ µετὰ ταύτην ἡ τοῦ σαββάτου> ἡµέρϱα, ἔχεις τὰς τρϱεῖς νύκϰτας κϰαὶ τὰς τρϱεῖς ἡµέρϱας.20 Having this text at hand, it is easy to tell why T had ἔχεις, why A corrected it into ἔχει (so that it would agree with ἡµέρϱα), and also why the hyparchetype of the other manuscripts conjectured a preposition (εἰς) so as to account for the accusatives that followed it. It is difficult, 19 Greg. Nyss. De tridui spatio p.287.9 (ed. Gebhardt). 20 Greg. Nyss. De tridui spatio p.289.1–3. 178 REMARKS ON EDITING A BYZANTINE ANTHOLOGY ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 however, to see how T could have conjectured the ἔχεις, which makes no particular sense with the omitted group of words, if this were not the reading of the archetype. And this could not have been known for sure if the source had not been available to us. 3. Conclusion. All critics, past and present, are well aware of the principle that they may not neglect the primary source of the anthology they are editing. It is no less obvious that they cannot print any reading merely because it is present in the primary source, though in any case they should take the trouble to find a plausible explanation of why the reading of the anthology differs from that of the source. The scope of this article has not been to re-state these well established metho- dological principles but to explain what may be considered the legitimate uses of a source. To do so, I have attempted to show that the anthology must be placed as accurately as possible within the larger system of the textual tradition of a primary source, even if, as is usually the case, there is a different textual tradition for almost each of the excerpts quoted. The autonomy of the anthology will not be endangered by any such consideration, not even if, for instance, the editor tries to find out why it has omitted or altered part of the text of its primary source. On the contrary, many facts which would have otherwise remained hidden may be illuminated. While, admittedly, the basic principles about editing anthol- ogies are well known on an intuitive and empirical level, they still need to be discussed theoretically, refined, and sys- tematized. The observations in this paper are no more than a first attempt in this direction, for they are neither exhaustive nor definitive, and some may need further elaboration. Yet I expect that they may be of some use to scholars striving, on the one hand, to make explicit and justify some practical rules of their craft, and on the other hand, to refine the ways in which TOMÁS FERNÁNDEZ 179 ————— Greek, Roman, and Byzantine Studies 51 (2011) 167–179 we edit and—more importantly—the ways in which we read ancient and medieval texts.21 September, 2010 Faculty of Arts Katholieke Universiteit Leuven B-3000 Leuven, Belgium Tomas.Fernandez@arts.kuleuven.be 21 I gratefully acknowledge the financial support of FWO-Flanders. The Katholieke Universiteit Leuven kindly hosted my research in Belgium. All my gratitude is due to Jacques Noret, Basile Markesinis, Pablo Cavallero, María Eugenia Steinberg, Reinhart Ceulemans, Elke Verbeke, and Sara Rich, who read successive versions of this paper and provided invaluable comment. In particular, J. Noret and R. Ceulemans devoted a great deal of energy and critical acumen to this contribution, and I am indebted to them for advice on innumerable points. I would also like to thank the anonymous referee, who provided a very insightful comment, from which this paper has greatly benefited.