Unfinished Business: The Ending of Mark in Two Catena Manuscripts

Abstract

Two Greek gospel manuscripts with an exegetical commentary in catena form present a text of Mark which ends in the middle of Mark 16.8. One is GA 304, a twelfth-century codex which is often adduced as a witness to the Short Ending. The other is the eleventh-century GA 239, which has not previously featured in discussions of the conclusion of Mark. In each case, it is shown that considerations of scribal practice, codicology and the broader traditions of text and catena mean that neither witness should be treated as evidence for the Short Ending as found in Codices Sinaiticus and Vaticanus.

1. Introduction

The ending of the Gospel according to Mark is one of the best known textual variants in the New Testament.¹ The various conclusions to this gospel have been so frequently described that there is no need to rehearse them here.² The editorial consensus that the earliest attainable text of Mark ends abruptly, after ἐφοβοῦντο γάρ in 16.8, has now been reaffirmed in the latest volume of the Editio Critica Maior (ECM Mark).³ This, the so-called Short Ending, is attested by the two famous fourth-century Greek pandects Codex Sinaiticus and Codex Vaticanus, a variety of early translations, and observations in early Christian writers including Eusebius, Hesychius of Jerusalem and Severus of Antioch. Extracts from such authors are often included in New Testament catenae, scriptural commentaries first compiled in the sixth or seventh century consisting of exegetical scholia from Christian writers presented alongside the biblical text. In addition to the two ancient codices, ECM Mark also cites a twelfth-century catena manuscript, Gregory-Aland 304, as a third Greek manuscript which attests the Short Ending. This witness has been adduced for this reading since the third edition of the United Bible Societies’ Greek New Testament (1975).⁴ In their introduction to the text of the New Testament, the Alands point to the manuscript as an example of the tenacity of the textual tradition, demonstrating that the Short Ending ‘persisted stubbornly’ for centuries.⁵ A recent detailed study of GA 304 by Monier reiterated this conclusion and urged ‘further research into the understudied legacy of Greek catenae’.⁶ In responding to Monier's appeal, the present article identifies another Greek catena manuscript in which Mark concludes at 16.8 (GA 239). However, by attending to the scribal practice and codicology of these manuscripts and comparing them with other catena traditions, it demonstrates that neither GA 239 nor GA 304 has a strong claim to be a genuine witness to the Short Ending of Mark, and that the latter should be removed from current editions of the biblical text.

2. GA 239

The manuscript Moscow State Historical Museum, Syn. gr. 46 (Vlad. 84) is an eleventh-century gospel codex which previously belonged to the Pantokrator Monastery on Mount Athos (Diktyon 43671, GA 239).⁷ It consists of 277 parchment leaves written in the format of a frame catena: the biblical text is set in the middle portion of each opening in a single column of thirteen lines, with the commentary in the outer margins. In his catalogue of Greek New Testament catena manuscripts, Parpulov assigns it to group 2.viii along with GA 146 and 556.⁸ The majority of the document consists of the C132 catena on Luke (ff. 3r–157v) and the C141.1 catena on John (ff. 159r–276v), each preceded by a single page of kephalaia. Despite the presence of a final folio (fol. 277), blank apart from some lection indications and a brief homiletic extract, the Johannine catena breaks off at the end of folio 276 after ἐκɛῖνος in John 21.23. The beginning of the manuscript also appears to be incomplete. Folio 1 contains seven verses from the final chapter of Mark, beginning with ἔρχονται in Mark 16.2 and ending with ἐφοβοῦντο γάρ in Mark 16.8. Thus, as it currently stands, in this codex the Short Ending of Mark is followed by the chapter titles at the beginning of Luke.

In his catalogue, Treu dismisses the initial page as ‘a discarded leaf’ presumably added as part of the binding process.⁹ Instead, he prefers to consider Luke as the start of the manuscript. In keeping with this, the ownership note of the Pantokrator Monastery appears at the bottom of folio 2r, as if this were the first page of the book (even though it describes the contents as a commentary on Matthew and Luke, possibly reflecting an original two-volume work). There are several large blotches on folio 1r resembling traces of glue from a pastedown, although they could be mould or other deterioration.¹⁰ On folio 1v, a later hand has written out the first three lines of this page again in the lower left margin in a manner typical of probationes pennae on spare pages at the beginning or end of a volume. Treu notes that the manuscript consists of thirty-five quires, with the fourth and fifth leaves of the second quire missing. Quire signatures are visible on ff. 7r, 21r, 29r and 37r, indicating regular quaternions. Nevertheless, the arrangement of the initial quire is unclear: is it a quire of six pages including the Marcan leaf, a defective quire of five pages (ff. 2–6), or a quaternion from which two pages have been excised (perhaps for an illustration)? There seems to be continuity between the page ruling on folios 1v and 2r, suggesting that the first two pages may be part of the same quire, but the exact distribution of leaves is impossible to establish from the microfilm images.

In contrast to Treu, the Kurzgefasste Liste describes GA 239 as a three-gospel manuscript, including the page of Mark as part of the original production.¹¹ Certain features of its presentation offer support for this. The biblical text appears to have been copied by the same hand as in the rest of the manuscript, and its format matches the catenae on Luke and John, where the lemma is written in thirteen lines of around twenty-two characters with wide outside margins left for the commentary.

On this page of Mark, however, there are no exegetical scholia. Although there are catena types without any commentary at the beginning of Mark 16, this lack of scholia is surprising given that they are present throughout this chapter in the other two members of Parpulov's group 2.viii.¹² One of these, GA 146, not only has an identical layout to GA 239, with thirteen lines of biblical text in frame catena format, but also has the same distribution of text on its folio 263 as GA 239 has on folio 1, beginning with ἔρχονται and ending with ἐφοβοῦντο γάρ (before continuing with Mark 16.9–20 on the next page). These manuscripts display other similarities: the scriptural text on the final page of Mark in GA 146 (fol. 265v) is written in a cruciform manner like that of Luke in GA 239 (fol. 157v); the gospels begin with identical gold lettering and decoration underneath a blue and gold headpiece (compare the opening of Mark in GA 146 with that of Luke in GA 239); and the lists of kephalaia are arranged in the same fashion. What is significant for the first folio of GA 239 is that the colour images of GA 146 indicate that, on many pages, the biblical lemma was written in advance of the commentary, as shown by the different colour of ink. This is difficult to determine on the monochrome microfilm of GA 239, although certain folios suggest that it was also the case (e.g. ff. 205r, 258v). The practice of copying the scholia at a later stage of production provides a context in which a folio with only the biblical text could have been surplus to requirements and redeployed as a pastedown in another volume. A further indication of a limitation to the amount of Mark in GA 239 is the quire numbering which, unlike the folio numbering, is original to the production context. The first visible number is for quire 2 on folio 7r, indicating that this would only have been preceded by a single quire, five of whose folios contain Luke (ff. 2–6). GA 146 demonstrates that, if the first quire were a quaternion, its two missing pages would have been sufficient to contain the Longer Ending of Mark in this catena format. Nevertheless, as observed above, the constitution of this quire cannot be established from the microfilm, and Treu does not comment on any missing folios here.

The key feature which indicates that folio 1 of GA 239 is not an instance of the Short Ending of Mark is the lack of concluding decoration. Although the phrase ἐφοβοῦντο γάρ stops two to four characters short of the right-hand margin of the text block, the resulting blank space is not exceptional and is consistent with the continuation of the text on the following page. There is no indication that this is the end of the gospel, unlike in Luke where the final page of biblical text is arranged in a cruciform pattern, as mentioned above. In addition, the final words both of the catena and biblical text in Luke are followed by a small cross, a very common device for indicating the end of the text (compare also the ending of Mark in GA 146). Despite the loss of the final page of John from GA 239, the frequency and consistency of this practice means that, as a cross is present in Luke, it would be expected in the other texts in the codex.

In sum, there is no reason to change the current scholarly treatment of GA 239. Treu's judgement that the leaf of Mark is surplus to the rest of the codex is supported by the quire numbering and lack of scholia, while the Kurzgefasste Liste is correct in noting the presence of Marcan text which could have come from a continuous-text manuscript, although in this case it seems likely that it is a leaf discarded during the production process. At any rate, despite the current arrangement of the manuscript, there is no compelling reason to consider it as a witness to the Short Ending.¹³

3. GA 304

The manuscript Paris Bibliothèque nationale de France, grec 194 contains a catena of Matthew and Mark copied in the first half of the twelfth century (Diktyon 49763, GA 304).¹⁴ It is written in alternating format, with the biblical lemma in red for the first few pages. The scholia begin with an enlarged initial, and end with an outsize raised dot in the middle of a blank space. Although the lemma initially ends with a colon, from around fol. 30 the outsize dot is also used to indicate the close of each portion of the gospel. In Matthew (ff. 1–166), the patristic sources are consistently indicated by monograms in the margin. These are less common in Mark (ff. 167–241r), and are absent from chapter 7 onwards with five exceptions (ἄλλως on ff. 211r and 214v, Κυρίλλου on fol. 220r, Φωτίου on fol. 230r, ᾽Ωριγένους on fol. 233v). This manuscript is the only member of Parpulov's class 34, and the sole witness to each of its catenae, C114 in Matthew and C126.5 in Mark.¹⁵ Despite this, there are unmistakable parallels between this compilation and that of Theophylact, the eleventh-century archbishop of Ohrid, which is the most widely attested catena on each of these gospels.¹⁶ Many phrases match verbatim and the structure of the comments is similar, yet overall the commentary in GA 304 is shorter and has a tendency to paraphrase. Its dependence on Theophylact may be illustrated by the first sentence of the comment following Mark 16.8:

The text of GA 304 is a clear improvement. It reduces Theophylact's sprawling sentence to three concise clauses, removing several superfluous explanations (κατὰ τὸ ἔθος τῶν Ἰουδαίων, ὡς ἂν μένῃ ɛὐῶδɛς, τὴν ὑγρότητα τοῦ σώματος ἀναπίνοντα). The archbishop's breathless present tense narration, which has the women buying myrrh even as they sit beside the tomb, is replaced by an imperfect and aorist which clearly distinguish the two activities. The dehydrating properties of myrrh are not an adjunct (ἅμα δέ) to the previous statement of purpose, but its rationale (γάρ), in which the key terms are elegantly brought to the front of each clause and the observation generalised, with the plural τὰ σώματα replacing the singular pronoun αὐτό. One might even discern a theological appropriateness in transferring τοῦ Ἰησοῦ to qualify the body, in contrast with the divinity of τοῦ κυρίου. Despite Monier's hesitations, the direction of literary dependence between the two commentaries—at least in this portion of Mark—is evident.¹⁷

The unusual feature of GA 304 is that the final biblical lemma comprises Mark 16.1–8, ending with the words ἐφοβοῦντο γάρ. This is followed by a single long scholium, which begins with the text quoted above and ends with a similar observation to that in Theophylact, that the women either feared the Jews or had lost their faculties through fear (ἢ τοὺς Ἰουδαίους φοβοῦμɛναι, ἢ ὑπὸ τοῦ φόβου τὸν νοῦν ἀπολέσασαι).¹⁸ After the outsize dot concluding the scholium, there is a blank space equivalent to about seven lines of text on this folio before the bottom margin (which has been cut off), and a verso which is blank except for several probationes pennae in various hands. In short, if the commentary had continued, there would have been space to copy it. Instead, on the line after the text, there is a version of the well-known ὥσπɛρ ξένοι epigram, comparing the end of copying a book to the return of travellers to their homeland.¹⁹ Not only has this been erased, but another hand has also rewritten some of the words above and below it, as a probatio pennae. Although Monier attributes this epigram to the copyist, the hand differs markedly from that of the main scribe. Not only are these two lines written roughly, slanting upwards in comparison with the rest of the page, but there are numerous differences between many of the letter forms: the sharp apex of the epsilon-rho ligature in ὥσπɛρ contrasts with the smooth rho ligatures in the main text (and all ten instances of ὥσπɛρ in Matthew); the wide loop of xi with the final curve of the descender to the left differs from the tight, straight xi in lines 2 and 14 of this page; rho and phi in the epigram both have a curl to the right at the bottom of the descender, unlike the straight descenders in the main text; the left-slanting ascender of delta is unparalleled in the vertical ascenders or semi-majuscule instances of delta in the commentary text. Perhaps the most obvious discrepancy is in the style of breathing marks: both ὥσπɛρ and οὕτω in the epigram have rounded breathings, whereas the copyist employs square breathings throughout. It appears that a later user, thinking that this was the end of the text, added the epigram, which was then erased by someone else who decided that the text remained incomplete.

A parallel for the addition of this epigram by a later user is seen in a longer version of the verse at the end of Matthew on folio 168v (misnumbered as 166v in the manuscript). These four lines are written in a much fainter ink and a smaller, flatter script which again differs from the main hand. What is notable on this page of Matthew is that the final words of the commentary, which do not occupy a whole line, are centre-justified and followed by a cross, with two further crosses on the line below. This stands in sharp contrast to the last scholium on Mark, where the two words are left-justified and followed by the outsize raised dot. This complete departure from the copyist's practice of ending a work, as exemplified in Matthew, stands as a clear indication that the commentary of Mark is unfinished: the text is not centred, there are no crosses (compare the discussion of GA 239 above), and the punctuation provided is that for the end of a scholium rather than the end of the work. Monier also observes that the text of the commentary on Matthew in this manuscript ends with a formal conclusion and benediction, unlike this final exegetical comment on Mark.²⁰

Why, then, is so much blank space left at the end of Mark? Two other places in the manuscript suggest a possible answer. On folio 150r (numbered 148), a gap of twelve lines is left between the end of a scholium from Origen and the lemma of Matt 26.14–16. This cannot be to avoid a hole in the parchment, as the verso is written without a break. Instead, it appears that part of the exemplar was illegible or damaged at this point, and the copyist left space to provide the missing text later. Six lines have, in fact, been added here by a subsequent user (possibly the one responsible for the epigram after Matthew) but then erased. There is also a blank page in Mark (fol. 233, numbered 231), coming between the commentary on Mark 14.15 and the end of a scholium on Mark 14.22–5.²¹ Each of the texts is incomplete, consistent with a page having fallen out of the present manuscript. The best explanation seems to be that the extra sheet was bound in later in order to repair the damage: a stain visible on the bottom outer corner of fol. 234r (232r) is not present on this leaf, suggesting the page is a subsequent addition, and no ruling lines are visible, which would be expected if this sheet was part of the original quire. If the blank page belonged to the original production, however, it could correspond to a page missing from the exemplar: the loss of a quire's final folio with text from Mark 16 might result in the initial folio of the same quire, perhaps containing part of Mark 14, falling out too. As with the initial quire of GA 239, inspection of the codex itself is the only way to resolve this question.

In any case, both the lack of scribal indication of the end of the work after the scholium on Mark 16.1–8 and also the dependence of the commentary in GA 304 on Theophylact, are sufficient to establish the incompleteness of this manuscript. Theophylact's catena covers the whole of the Longer Ending, as does the C125.2 catena.²² The dependence of GA 304 on these sources indicate that it does not stem from an ancient text form lacking the Longer Ending. In addition, like GA 304, all of the manuscripts of Theophylact used in ECM Mark present the scriptural text of Mark 16.1–8 as a single unit, ending with ἐφοβοῦντο γάρ.²³ The conclusion of the biblical lemma at Mark 16.8 followed by a scholium derived from Theophylact shows that the C126.5 catena in GA 304 is based on a commentary tradition in which the Longer Ending was present. Despite being the sole surviving source for the two catenae which it contains, the gaps observed above indicate that GA 304 is not the autograph of its commentary in either gospel, although its twelfth-century hand suggests that this revision of Theophylact was made within a century of the archbishop's original composition. Instead, the presentation of this manuscript suggests that the final pages of Mark were missing from its immediate exemplar or the copyist was prevented from completing their work for some other reason. Notwithstanding the current state of its text, GA 304 should no longer be cited in editions as a witness to the Short Ending of this gospel.

In applying this conclusion to ECM Mark, the question arises as to why GA 304 was included in this edition in the first place. The Text und Textwert analysis used to select Greek manuscripts for this edition indicates that GA 304 agrees with the majority reading in 177 of the 192 test passages.²⁴ This gives a percentage of 92.2%, several points above the cut-off of a maximum 90% agreement for inclusion in the Editio Critica Maior. Indeed, the only agreements with the Nestle-Aland editorial text against the majority reading occur in the last seven test passages, comprising the Longer Ending. If these are treated as lacunose rather than omitted, the majority agreement rises to 95.7%, which is the overall figure given at the beginning of Text und Textwert.²⁵ This suggests that, between the collection of the data and the production of the summary tables, the compilers decided that GA 304 was lacunose at the end of Mark. According to the online CBGM, based on 5,321 variation units in ECM Mark, GA 304 has an agreement of 97.8% with the majority text.²⁶ Its gospel text is thus strongly in the Byzantine tradition, yet although GA 304 is listed in the edition among the Codices Byzantini, it is not one of the ‘seven nearly pure Byzantine manuscripts’ selected to represent this text form and its inclusion is not discussed in the volumes of supplementary material or studies.²⁷ It is only a short note in the Introduction which indicates that GA 304 was included because it ‘attests to the short ending of Mark with 01 and 03’.²⁸ Thanks to its presence in this comprehensive edition and associated tools, however, the Byzantine affiliation of this witness's text throughout Mark is confirmed, serving to reinforce the improbability that it preserves an ancient form of the ending. What is more, even though GA 304 does not meet the normal criteria for inclusion, its appearance in ECM Mark enables the identification of several variant readings in its biblical text whose attestation is almost entirely restricted to Theophylact manuscripts, further underlining its connection to this tradition.²⁹

4. Conclusion

The result of this investigation is to discount both GA 239 and 304 as evidence for the preservation of the Short Ending of Mark in Greek manuscripts of the eleventh or twelfth century. This matches the earlier trajectory of GA 2386, an eleventh-century gospel book whose text of Mark breaks off at the end of a page after ἐφοβοῦντο γάρ. This was cited in support of the Short Ending in the first two editions of the United Bible Societies’ Greek New Testament, before being removed following the observation that the lectionary indication on the final line indicated that this was not the end of the gospel: the last page of Mark was likely to have been excised when the evangelist portrait at the beginning of Luke was cut out of the manuscript, and the stub is visible on the microfilm images.³⁰ In sum, there are no known Greek minuscule manuscripts which only preserve the Short Ending of Mark. While the tenacity of the early textual variation at this point continues to be visible in such documents in the form of marginal asterisks, other scribal annotations, and comments from early Christian authors in catenae, claims that this ancient reading is attested by a Byzantine manuscript cannot be sustained unless they are supported by detailed investigation of the witness's codicology, scribal practice, and textual tradition. The present study does not challenge the scholarly consensus on the earliest attainable form of the ending of Mark, but it does demonstrate the imperative to take full account of the context and nature of documents in which an unexpected reading appears before adducing them as evidence for an early form of text.