1. Introductory Remarks: Schmid, Hernández, and Beyond
In Josef Schmid's influential study of the textual history of Revelation, Codex Sinaiticus figures prominently as the main (and the only complete) representative of one of the two early text-forms. Unlike the superior text-form attested by the codices Alexandrinus and Ephraemi, as well as the text of Oecumenius’ commentary, the one attested by Sinaiticus, 𝔓47 and Origen already contains a fair number of improvements (‘Korrekturen’) to the ‘original text’.Footnote 1 Sinaiticus was probably produced around the middle of the fourth century (or possibly earlier),Footnote 2 so Schmid's claim that, already in the fourth century, the text of Revelation in Sinaiticus was corrected towards another text-form, known from the later commentary of Andreas, is remarkable. Now, the fact that a number of Sinaiticus’ corrections reflect a Vorlage akin to the Andreas text was not Schmid's discovery. Rather, Schmid builds on an earlier work of W. Bousset, who had identified a number of corrections related to what is now referred to as the Andreas tradition.Footnote 3 Significantly, Schmid reinterprets Bousset's conclusions in light of H. J. M. Milne and T. C. Skeat's investigation of scribes and correctors in Sinaiticus,Footnote 4 identifying Bousset's (and Tischendorf's) seventh-century ℵc corrections with scribal corrections made in the scriptorium, some three centuries earlier. Thus, Schmid concluded that ‘Bousset zu dem klaren und sicheren Ergebnis gelangt, daß wenigstens der überwiegende Teil der aus dem 4.Jh. stammenden Korrekturen (= Sa) einer zu Αν gehörenden Hs entnommen wurde’.Footnote 5 But Bousset never reached such a conclusion. In fact, Juan Hernández has recently shownFootnote 6 that, unlike Schmid, Bousset never makes fourth-century claims with respect to these corrections, but simply follows Tischendorf, who dated the C-class corrections to the seventh century.Footnote 7 Schmid's fourth-century dating, argues Hernández, seems to stem from his misreading of Milne and Skeat – precisely the authority to which Schmid appeals in re-dating the corrections.Footnote 8
Such a confusion has, naturally, some implications, and Hernández spells out the following:
Every occurrence of Sa must now be read in light of its re-established seventh-century status. The alignment of Sa to particular witnesses is unlikely to change in most, if not all, cases; the conception of it as a fourth-century witness will change in every case. The most consequential revisions will occur where Sa is marshalled in support of a fourth-century date for the Andreas text type … The impact on other parts of Schmid's work will vary commensurate with the arguments advanced for particular cases. Again, textual realignments are unlikely, but the evidentiary weight of Sa will shift. Its value as a witness … will fluctuate on a case-by-case basis. The textual history of select readings will also appear in a new light … The current investigation will also inform contemporary discussions over text types. Footnote 9
Hernández is correct in identifying a serious dating error in Schmid's argument, and Sinaiticus’ many corrections to the text of Revelation definitely warrant reinvestigation.Footnote 10 However, his claim of a seventh-century date for Schmid's Sa corrections may appear, in this article at least, overconfident. Indeed, Milne and Skeat were themselves hesitant to ascribe a definitive date to the C-class corrections, allowing for some leeway anywhere between the fifth- and seventh-century dates.Footnote 11 Later on, Skeat would give a more specific judgement concerning the Ca corrector in particular, dating him to the sixth century.Footnote 12 More recently still, Amy C. Myshrall's palaeographical analysis led her to similar conclusions.Footnote 13 And even in his latest article, Hernández has invoked Milne and Skeat's more cautious stance, calling for fresh palaeographical investigations.Footnote 14 If indeed Ca worked in (roughly) the sixth century, then his corrections still predate, by almost a century, the composition of Andreas’ commentary, not to mention later minuscules with the Andreas-type text. Since the text of Ca's exemplar must have predated his correcting activity, it could theoretically still be viewed as a sixth-century – and possibly even earlier – witness to the Andreas text. The dating of these corrections, however, cannot, as such, settle the matter. Indeed, as will be seen, further complexities are involved in this line of enquiry, complexities which must be dealt with on a case-by-case basis.
2. Sa Corrections: AnalysisFootnote 15
In what follows, then, we shall inspect all the Sa corrections cited by Schmid, inquiring into their origin and textual significance. The results of our analysis shall be reflected upon in the final section.
As for the presentation, the readings are grouped according to sections of Schmid's work wherein they appear. It will occasionally be observed that Schmid only cites one part of a larger correction and even divides it into several variation-units. Since such a procedure is not without problems, both the initial and the corrected readings are listed in their entirety, with Schmid's delineation of variation-units underlined.Footnote 16 The corrected reading is, at each point, preceded by a siglum indicating the corrector. Unless otherwise noted, the readings of Sinaiticus and identifications of correctors are derived from the Codex Sinaiticus Project (CSP) website,Footnote 17 while the textual evidence cited comes primarily from NA27–28, supplemented (where appropriate) by M. Lembke's collation of 2846Footnote 18 and H. C. Hoskier's edition.Footnote 19
2.1. Major Textual Groupings
Schmid lists seven corrections agreeing with the Andreas text:Footnote 20
At 4.1, Ca substituted the masculine λεγων for the more appropriate feminine λεγουϲα. This shift more likely reflects Ca's exemplar, although it could also betray his own grammatical concerns,Footnote 21 as the initial (and better attested) reading λεγων is syntactically incongruous.Footnote 22
Similarly, at 7.9 Ca replaces περιβεβλημενουϲ with περιβεβλημενοι, thus bringing the participle into syntactical congruence with the rest of the clause (cf. εϲτωτεϲ earlier in v. 9).Footnote 23
At 9.12, the singular ερχεται is followed by the feminine plural δυο ουαι. By shifting to the plural, Ca resolves the syntactical mismatch. While this correction could, again, reflect Ca's concern for the correct Greek, the external support suggests that it probably stems from the corrector's exemplar.Footnote 24
Of a different kind is the correction at 13.6. Considering that a Greek copulative clause typically takes και before every conjoined constituent,Footnote 25 Ca's insertion of και between την ϲκηνην αυτου and τουϲ εν τω ουρανω may at one level seem expected.Footnote 26 Even so, the corrected reading obscures an appositive relationship between the two phrases.Footnote 27
At 16.17, a scribal error went unnoticed during the scriptorium correction process.Footnote 28 Ca altered the initial reading οτε to ο ζ̅ [= εβδομοϲ] αγγελοϲ, the reading of 𝔐A. Interestingly, 𝔐A adds αγγελοϲ at each point of the present narrative (16.3, 4, 8, 10, 12, 17), save for the first instance (16.2). Whether Ca's exemplar, too, read αγγελοϲ consistently is a moot point, as the correction itself was probably not triggered by the absence of αγγελοϲ, but rather by the change in meaning occasioned by the initial replacement of an explicit subject – whether ο εβδομοϲ αγγελοϲ or ο εβδομοϲ – with the temporal particle οτε.
Schmid refers to the correction at 18.7 as Sa, but it does not come from the same class of Ca corrections. Rather, it was made by the Cc* (Schmid's Sc) hand.Footnote 29 The corrected reading is relatively widely attested and hence probably genetic, although an attentive reader could also have made it without an exemplar.Footnote 30 In any case, the correction comes from a different, most likely later and independent, hand, and thus adds little to Schmid's cumulative argument concerning the relationship between Sa corrections and Andreas.
At 22.20, Ca added χριϲτε, thus creating a formula wholly unknown to Revelation, though not uncommon in the patristic literature.Footnote 31 Considering the external attestation of the reading, it seems less likely that Ca would have expanded the original address on his own initiative, even if such a possibility cannot be excluded with certainty.Footnote 32
Two corrections appear in Schmid's collation of the Koine (𝔐K) text:Footnote 33
As it is, Schmid's reference to 11.9 is misleading, as Ca never actually completed the correction.Footnote 35 It seems that he initially began to alter the reading, but, having written the left vertical of η, abandoned the correction – for whatever reason. As Ca's final judgement on (or a careless correction of) the reading favours the original form, the corrected reading cannot be (contra Schmid) cited as agreeing with 𝔐K.
The initial reading at 20.8 is part of the larger omission, probably occasioned by the scribe's distraction at the line-ending.Footnote 36 Thus, Ca probably at first noticed the omission of τηϲ γηϲ and, as he checked his exemplar, noticed the absence of τον not only before γωγ, but also before μαγωγ, and effected the correction accordingly.
In discussing the relationship between the Andreas and the Koine text-forms, Schmid cites following corrections:Footnote 37
While the insertion of αμην at 1.18 could be viewed as an arbitrary doxological addition, here it disrupts the flow of the sentence, and hence probably reflects the corrector's exemplar.
At 3.5, Ca shifts – not unexpectedly – ουτωϲ to ουτοϲ.Footnote 38 Though the former reading is preferred by Schmid (as well as by the editors of NA28),Footnote 39 the latter is read by the vast majority of the Greek tradition and significantly alters the meaning of the clause.
The initial reading at 19.6 is probably a scribal error. It is actually possible that Ca restored a reading that had also stood in the exemplar of Sinaiticus. If so, both Sinaiticus’ and Ca's exemplar support the majority reading, thus corroborating its external attestation further still.
At 20.9–10, Ca corrects a lengthy singular omission, but even the corrected reading has rather meagre support. It seems possible, then, that the correction was not carried out accurately, so that the corrector's exemplar may have contained one of the majority readings.
The initial reading at 21.20 is supported by two important minuscules with the Oecumenius-type text.Footnote 40 It is unfortunate that 𝔓47 is not extant at this point, as it would be of great value in determining whether the reading of Sinaiticus is scribal or genetic in origin. In any event, the Ca reading is, inter alia, attested by both 𝔐A and 𝔐K.
Eight corrections are noted in Schmid's discussion of the A C Oecumenius text-form:Footnote 41
At 1.19, both the initial and the corrected reading have significant external support. One could possibly argue that this could be just an arbitrary grammatical improvement, because, in the New Testament, μέλλειν is usually followed by the present infinitive.Footnote 43 Nevertheless, since Revelation exhibits much greater fluctuation in this respect,Footnote 44 the genetic origin seems more plausible.
The initial, somewhat pleonastic, reading at 6.4 is supported by the vast majority of witnesses. The correction smooths the clause out by deleting the superfluous αυτω. Notwithstanding the support of A 2344, one cannot exclude a possibility that such a correction may have occurred independently.
At 11.16b, the initial reading is well attested. Unlike most other witnesses, however, Sinaiticus (followed by 1006 1841 al) also inserts και before επεϲαν, making the relative construction even more awkward. By expunging οι, Ca attempted to improve the construction – whether based on the exemplar or not.
The initial reading at 14.8 is also shared by 𝔓47, Sinaiticus’ closest ally, and is unlikely to be a scribal error. By inserting αγγελοϲ, Cc* alters it to a more widely attested reading, preferred by Schmid as well as the editors of NA28.Footnote 45 The correction could be due either to the corrector's exemplar or to the influence of the context (cf. 14.6,9).
As noted, at 14.8–9a Ca restores a lengthy scribal omission. The present variation-unit concerns omission of the second επεϲεν within the correction. Since the external evidence for the omission is by no means meagre, we probably have a genetic agreement, although the possibility of accidental haplography cannot be excluded.
The variation-unit at 14.8–9b concerns addition/omission of the relative pronoun η (misleadingly accented as ἢ in Schmid's discussionFootnote 46). The reading is widely attested.
The initial reading at 21.6 is attested in both later text-forms and reflects a misunderstanding of the syntax.Footnote 47 The correction itself is puzzling: Ca initially altered γεγονα to γεγοναν. Because Ca did not supply ειμι in the following part of the verse, however, the result is a singular conflation of the majority reading with that of A 1678 1778. Oddly, γεγοναν was subsequently cancelled. Perhaps the corrector's Vorlage was furnished with a correction or a reader's note that confused him somehow. Alternatively, the corrector may have – initially or later – departed from his Vorlage. All the same, the problematical nature of this correction precludes any further judgement on the precise wording of the corrector's exemplar at this point.
The reading ην εν δωμαϲι at 21.18 seems to be a scribal misreading of ην η ενδωμηϲιϲ, a secondary variant shared by most witnesses. The correction exhibits a genuine textual shift towards the superior reading η ενδωμηϲιϲ.Footnote 48
Three corrections appear in Schmid's collation of the 𝔓47 ℵ Origen text-form:Footnote 49
The και read initially at 11.8 is well attested and overall a superior reading. On the other hand, Ca cancelled it, thus bringing the reading into conformity with 𝔓47 and the Andreas tradition. Taking into account Ca's addition of αυτων after the following κϲ̅, the entire corrected reading aligns neatly with 𝔐A at this point.
Schmid's reference to Sa at 11.12 seems overconfident: the correction is comprised of an erasure that cannot be attributed to a specific corrector with any certainty.Footnote 50 Since the corrected reading is, inter alia, supported by 𝔓47, the erasure could be due to the original hand, who had initially used the plural under the influence of the surrounding context.Footnote 51 As the correction cannot be assigned with certainty, however, its value is, for Schmid's purposes, rather limited.
At 19.13a, the singular initial reading περιρεραμμενον is most likely due to the scribe.Footnote 52 Cc*'s (also singular) shift to περιρεραντιϲμενον may suggest that his exemplar read ρεραντιϲμνενον (P 2320 al) or ερραντιϲμενον (1006 1841 pc), but he failed to cancel the prepositional prefix περι.
In examining the relationship between Sinaiticus and 𝔐K, Schmid cites one Sa correction:Footnote 53
At 21.27, Ca rectifies the nonsensical reading ποιωϲει and cancels the article, thereby shifting the reading to that of Alexandrinus and others. Although the initial reading is a scribal error ‘whose occurrence def[ies] easy explanation’,Footnote 54 it seems clear that his Vorlage included an article. The corrected reading is most likely genetic.
In his treatment of the problematical variation-unit at 18.3, Schmid remarks on the parallel at 14.8 that also involves one Sa correction:Footnote 55
Schmid's discussion concerns variation of πεποτικεν/πεπτωκαν, initially absent from Sinaiticus. The Ca correction reads πεπτωκαν, a secondary reading that probably owes its origin to the influence of the preceding επεϲεν βαβυλων.Footnote 57 The reading's slim support might be non-genetic, as harmonisation to the immediate context was a pervasive tendency across the tradition,Footnote 58 and morphologically similar words such as πεποτικεν/πεπτωκαν lend themselves well to such a visual confusion.
2.2. Patristic Citations
Three variation-units discussed under Origen's text involve corrections:Footnote 59
Although the corrected reading at 11.8 is, as noted, genetic, it adds little to Schmid's discussion of Origen's text, as Origen's citation is rather free and hence ‘ohne Beweiskraft’.Footnote 60
At 14.3, Sinaiticus has a well-attested anaphoric article before εκατον. Surprisingly, Ca deleted the article, altering the reading to that with very sparse attestation. It seems impossible definitively to account for this intervention: while we could invoke the corrector's exemplar, such an explanation is rendered problematic by the reading's meagre support. Other conceivable explanations such as harmonisation to Rev 14.1Footnote 61 or incongruity of grammatical genders (the following participle is in the masculine) do not seem to be particularly impressive either. In any event, the Ca reading is most likely non-genetic.
The initial reading at 19.13b is probably a conflation of κεκληται and the following article το, resulting, effectively, in a pluperfect form κεκλητο.Footnote 62 Probably, Ca simply restores the intended reading of Sinaiticus.
Three Sa readings appear in Schmid's discussion of Hippolytus of Rome:Footnote 63
The initial reading at 11.3, though relatively well attested, is ‘offenkundig sinnlos’.Footnote 64 The corrected nominative reading is strongly supported and thus likely genetic.
The well-attested masculine participle at 11.4 is syntactically incongruent, as it belongs to a construction governed by the feminine article. It is interesting that the reading was corrected by the later Cc* rather than Ca, whose (surely different) exemplar has been said to be related to the Andreas text.
At 18.8, the initial reading is widely attested. Surprisingly, Ca altered it to a reading with extremely weak attestation – a reading, incidentally, of the Textus Receptus. Importantly for Schmid's purposes, the correction (with which the text of Hippolytus disagrees) most likely does not betray a genuine genetic relationship.
2.3. Early Fragments
In discussing early fragmentary manuscripts of Revelation, Schmid refers to two corrections. The first one appears in his collation of 0169:Footnote 65
At 4.2, 0169* lacks και before ευθεωϲ. Schmid's collation indicates that the corrected text of 0169 includes και, thus agreeing with the corrected reading of Sinaiticus, as well as 𝔐A. However, the correction clearly reads ευθεωϲ δε, hence Schmid's postulated agreement with 0169c – and, more importantly, with Andreas – is illusory. Moreover, it is uncertain whether the Vorlage of 0169 contained a shorter text, as the correction might have been made by the original scribe, using the same exemplar. Since the omission took place at a line break, it could easily have been accidental. As for the correction, the sole support of 0169* 2027 most likely suggests a non-genetic origin, perhaps reflecting Ca's avoidance of asyndeton.
Another Sa correction is noted in Schmid's collation of 0207:Footnote 66
Here the scribe initially omitted the entire phrase μιαν εκ των [τεϲϲαρων] κερατων. Schmid cites this correction in agreement with the omission of τεϲϲαρων in 0207. Though omission by homoioteleuton cannot be entirely excluded, the reading is more likely genetic.
2.4. The Use of Language
Schmid cites three corrections in discussing morphological issues:Footnote 67
The form αφειϲ at 2.20 is attested by the vast majority of the tradition. Conversely, the reading introduced by Ca is found in only a handful of late manuscripts. This scant support, then, may possibly be non-genetic, so that Ca made the correction on his own accord – perhaps under the influence of εδωκα in v. 21 – although the aorist is, admittedly, not well-suited for the present context.Footnote 68
At 16.10, we have a well-attested initial reading, which Cc* altered to a reading whose support is overall scant and much later. While it is possible that Cc*'s exemplar contained a reading attested by 046 and a few minuscules, it seems just as likely that he altered the form on the basis of his personal preference.
At 20.12, the scribe probably made an omission at a line break (after βιβλια ηνεωχθη). Schmid cites the correction as an example of the double augment in ηνεωχθη, which is, in fact, read by most witnesses in 𝔐K and 𝔐A. The widespread attestation of this orthography may thus suggest a genetic origin, though it could also be a harmonisation to the preceding ηνεωχθη, found in the original text of Sinaiticus.
Two corrections appear in Schmid's discussion of the use of article:Footnote 69
Though the omission at 11.16a is not singular, it might still be a scribal error, as omission of small words occurs frequently in Sinaiticus.Footnote 70 Moreover, the article does not carry much of its anaphoric force, as the previous mention of the twenty-four elders (5.8) is rather removed from the present context. Importantly, Schmid's reference to the omission of οι by Sa is thus incorrect.Footnote 71
As discussed, the initial reading at 21.27 is nonsensical, though the Vorlage of Sinaiticus most likely contained an article. By deleting the article, Ca shifts the reading to that of Alexandrinus (inter alia).
In discussing the use of grammatical cases, Schmid cites one correction:Footnote 72
Since the shift to the genitive at 9.13a occurs within Ca's larger correction, it could be genetic, although the corrector's preference for the genitive after ακουειν is perhaps a more likely possibility, especially in view of the reading's sparse attestation.
Schmid's discussion of pronouns involves an Sa correction at 18.7:Footnote 73
Schmid's case for the reading αυτην seems corroborated still further, since the correction, despite its weighty attestation, need not have been genetic, as only a minor orthographical matter is involved. Besides, the correction is in any case quite late.
Schmid's discussion of the use of prepositions involves the following correction:Footnote 74
At 17.6, Ca shifts the sub-singular dative to the better-attested genitive. The initial reading may have been phonetically conditioned,Footnote 75 so that the scribe merely deviated from the reading of his exemplar. In any event, Schmid cites the correction among witnesses to the (probably secondary) omission of εκ. Whether Ca's exemplar had εκ or not is a moot point, as his intervention was limited to the shift of the grammatical case.
Two corrections cited by Schmid involve stereotypical expressions:Footnote 76
At 16.18, the scribe initially leapt forward to copy βρονται, but caught his error immediately and copied the phrase in a different sequence.Footnote 77 In making this ad hoc correction, however, he failed to skip over the superfluous και βρονται.Footnote 78 It was this meaningless repetition that probably triggered the Cc* correction whose execution, however, seems equally incompetent, as it only cancels βρονται, resulting in a nonsensical sequence of two consecutive και.Footnote 79 Even if one overlooks the repetition of και, the resultant reading lacks firm genetic support.
The addition at 22.20 also appears on Schmid's list of inauthentic occurrences of the ‘Doppelname’ ιηϲουϲ χριϲτοϲ.Footnote 80 Whether deriving from the exemplar or not – and the former seems somewhat more likelyFootnote 81 – the correction appears to be the earliest evidence in support of the longer reading.
One correction is cited in Schmid's discussion of the constructio ad sensum:Footnote 82
The sub-singular omission at 4.5–6 is most likely due to scribal leap. Ca's correction involves a plural constructio ad sensum, agreeing – against AlexandrinusFootnote 83 – with 𝔐A. Schmid apparently prefers the Ca reading, since he notes it among those instances where the original plural was subsequently altered to the singular.
Among several variation-units involving language irregularities discussed by Schmid, the one at 9.12 involves an Sa correction:Footnote 84
As discussed, this correction rectifies an obvious incongruence, which, however, is most likely part of the earliest attainable text.Footnote 85 While it is uncertain whether the correction reflects Ca's exemplar or his own grammatical concerns, the corrected reading appears to predate all other known witnesses to this variant.
3. Synthesis and Concluding Reflections
Throughout our discussion we observed that, as Hernández recently argued, none of Schmid's Sa corrections originated in the scriptorium, and thus they cannot be dated to the fourth century. Moreover, even Schmid's monolithic Sa label seems misleading, as not all the corrections so designated were made by the same hand. If we follow CSP's assignment, thirty-five of these corrections were made by Ca, six by Cc*,Footnote 86 and one correction seems impossible to assign. It follows, then, that textual affinities of these corrections cannot be studied together as if they constituted one class of readings, as in both cases the correctors must have used different Vorlagen. Furthermore, there is a good chance that the correctors occasionally altered readings on the basis of their preference rather than their exemplars, resulting in coincidental agreements.
Let us first consider the six Cc* corrections. In two cases (16.18, 19.13a), Cc* created a singular reading. At 16.10, the corrected reading is weakly attested, and could also indicate the corrector's own judgement rather than his exemplar. In the remaining three instances (11.4, 14.8, 18.7), both initial and corrected readings are relatively well attested, the corrected readings agreeing, in each case, with 𝔐A. Clearly, the witness of Sinaiticus’ Cc* corrections must be weighed on a case-by-case basis and should only be cited with due caution. One must especially keep in mind that, as Milne and Skeat argued, this class of corrections are significantly later than, and most likely unrelated to, the remaining C-corrections.Footnote 87
Secondly, and more importantly, we turn to the thirty-five Ca corrections. Although Schmid cites Sa corrections throughout his analysis, he specifically discusses them only in relation to the Andreas text. It is all the more striking, therefore, that only six of them (4.1, 7.9, 9.12, 13.6, 16.17, 22.20) appear in his collation of Andreas. Taking all of Schmid's citations into account, however, we come to more robust (and certainly more revealing) figures. In fact, only four corrections (4.5–6, 13.6, 16.17, 22.20) cited by Schmid accord (virtually) solely with 𝔐A, and, at five further places (1.19, 4.1, 7.9, 9.12, 11.8), 𝔐A is joined by some of the witnesses to older text-forms.Footnote 88 Nine of these corrections, then, exhibit some affinity with the Andreas tradition; in at least four cases (4.1, 7.9, 9.12, 13.6), though, it is possible that the agreement is non-genetic. One correction is supported by the 𝔐K tradition (17.6) – again, virtually solely, and possibly non-genetically – and, in three further cases (14.8–9a, 19.13b, 20.8), 𝔐K is joined by some of the earlier witnesses. Both later textual streams join together at three places (1.18, 20.12, 21.20), and at five further places (3.5, 11.3, 11.16a, 14.8–9b, 19.6) with some earlier support. In eight cases (4.2, 6.4, 9.13b, 11.16b, 14.8–9c, 21.6, 21.18, 21.27), the corrected readings are supported by early witnesses, but as many as five of these agreements (4.2, 6.4, 11.16b, 14.8–9c, 21.6) could be non-genetic. Notably, five corrections (2.20, 9.13a, 14.3, 18.8, 20.9–10) cited by Schmid have rather scant support and are probably non-genetic. And at 11.9, Ca attempted a shift that he abandoned in the process, reverting the reading back to its original form.
Since the above figures apply solely to Schmid's selection of readings rather than to the whole of Ca's correcting activity in Revelation, a summary of textual affinities of all the Ca corrections in Revelation is in order. Setting Ca's retouchings of scribal corrections and other textually irrelevant interventions aside,Footnote 89 we come to the following figures. Eight corrected readings (1.17b, 2.13, 4.5–6, 6.15, 13.6b, 16.17, 17.8, 22.20b) are peculiar to 𝔐A, and at sixteen further places (1.19, 1.20, 2.10a, 4.1, 4.3–4, 4.8c, 7.9b, 7.9c, 9.12b, 10.1, 11.2d, 11.8b, 14.13, 19.1, 20.12a, 22.18) 𝔐A is joined by some of the older witnesses. Two corrected readings (17.6, 21.14) are supported by 𝔐K, and in eight cases (1.8b, 15.7, 16.14, 18.13a, 19.9b, 19.13, 20.8b, 22.3b) 𝔐K is joined by some of the earlier witnesses. In four cases (1.18b, 17.3, 20.12c, 21.20c), the corrected reading is supported by both 𝔐A and 𝔐K. More significantly, both later streams are joined by earlier witnesses at 169 corrected readings (1.1a, 1.1b, 1.5a, 1.5b, 1.7a, 1.7b, 1.7c, 1.8a, 1.9, 1.11a, 1.11b, 1.11c, 1.13, 1.16, 1.17c, 1.18a, 2.1, 2.9, 2.10b, 2.14a, 2.17, 2.19a, 2.20b, 2.20d, 2.21, 2.23, 2.24a, 2.26a, 2.26b, 3.2, 3.3, 3.5, 3.12a, 3.12b, 3.14a, 3.14b, 3.15, 3.16a, 3.16b, 3.17, 4.8a, 4.8b, 4.8d, 4.9, 4.10, 5.1a, 5.4a, 5.4b, 5.7, 5.8, 5.13b, 6.1, 6.9a, 6.9b, 6.9c, 6.11, 6.12, 6.14, 6.16, 7.2a, 7.2b, 7.9a, 7.10a, 7.10c, 7.11, 7.12, 7.17, 8.6, 9.1, 9.2, 9.2–3, 10.3a, 10.3b, 10.3c, 10.6a, 10.6b, 10.8, 11.1, 11.2a, 11.2b, 11.2c, 11.2e, 11.3, 11.4, 11.5, 11.6, 11.7, 11.8c, 11.10, 11.11, 11.15, 11.16a, 11.17a, 11.7b, 11.18, 11.19a, 11.9b, 12.1, 12.4, 12.6, 12.14b, 12.14c, 13.1, 13.2, 13.6a, 13.7, 13.8a, 13.16a, 13.16c, 13.17, 14.1, 14.2, 14.3b, 14.4, 14.6, 14.14, 14.18, 14.20, 15.3a, 16.2a, 16.13a, 16.13b, 16.15, 16.18a, 16.18b, 16.19, 17.2, 17.4, 17.9, 17.12a, 17.12b, 17.15, 17.17a, 18.8a, 18.9a, 18.9b, 18.14, 18.16, 18.23, 19.2, 19.4, 19.5, 19.6, 19.7b, 19.9a, 19.9c, 19.10, 19.19, 19.20, 20.1b, 20.1c, 20.8a, 20.15, 21.3a, 21.3b, 21.3c, 21.4a, 21.4b, 21.6b, 21.20a, 21.21a, 21.21b, 21.21c, 21.25, 22.3a, 22.6, 22.12, 22.14, 22.20a). Finally, twenty-six singular (1.6a, 2.19b, 2.20c, 2.24b, 7.10b, 7.15, 8.4, 10.9, 11.8a, 12.11, 12.13a, 12.13b, 17.10, 17.17b, 19.12a, 19.12b, 20.12b, 21.6a, 21.9, 21.12, 21.13a, 21.13b, 21.20b, 21.21d, 22.2, 22.7) and twenty-eight weakly attested (2.14c, 2.20a, 2.22, 2.24c, 3.7, 3.12c, 5.1b, 6.6, 7.13, 8.11a, 9.13,Footnote 90 9.14, 9.19, 11.9b, 11.14, 13.5, 13.8b, 13.8c, 13.16b, 14.3a, 14.8–9, 18.8b, 19.9d, 20.1a, 20.9–10, 21.15, 21.16, 21.23) readings were found among Ca's corrections. Besides several reverted corrections, retouchings of nomina sacra, and other textually insignificant interventions whose origin does not require an exemplar, these readings furnish us with the best evidence that, occasionally, Ca made corrections irrespective of his exemplar. Naturally, at some places, we may have imperfectly executed corrections or inherited singulars, but this type of explanation can scarcely account for all such cases.
Most of the 261 above-listed corrections rectify what were most likely scribal errors, since, in 137 instances, the initial reading appears to be singular and, in 67 instances, the initial reading is weakly attested. Yet this does not warrant exclusion of such variation-units from consideration, as Ca, working some two centuries after the production of Sinaiticus, would have used a different Vorlage, from which follows that all of his resultant readings are relevant for understanding the textual affinities of that Vorlage.Footnote 91
As seems obvious from this brief summary, the textual direction of the Ca corrections is far from straightforward. Apart from the most obvious problem of non-genetic origin in some cases, unique agreements with Andreas are, in fact, not very numerous. Indeed, one fails to observe a marked tendency towards an Andreas-type text.Footnote 92 Given the further 10 corrections agreeing with 𝔐K and 176 corrections agreeing with both 𝔐A and 𝔐K (and very often also with the rest of the tradition), it seems more likely these Ca corrections reflect – and, perhaps, are themselves part of – a development of the text of Revelation that we see realised more fully in the later text-forms now known as Koine and Andreas (especially the latter).Footnote 93 This is unsurprising, as Schmid himself did not construe Koine and Andreas as completely independent of each other, since they share numerous ‘Korrekturen’ to the more ancient text of Revelation.Footnote 94 This argument gains in plausibility when we take into account the fact that the Ca corrections may well have predated the composition of Andreas' commentary by almost a century, at which time the Andreas text-form was most likely still evolving. If this suspicion proves to be correct – and only further study can determine that – even Schmid's notion of the recensional nature of the Andreas text may be called into question. For Schmid construed the Andreas text as ‘eine Rezension im eigentlichen Sinne, das Werk eines Mannes … der den Text durch alle Kapitel durchkorrigiert hat’.Footnote 95 Nevertheless, we posses no documentary evidence for such a thoroughgoing recension that would predate the composition of Andreas’ commentary – and nor did Schmid, as he did not form his recensional theories on a purely documentary basis, but rather on a large number of ‘Korrekturen’ that were peculiar to the Andreas text.Footnote 96 Thus, we may be on firmer ground in invoking E. C. Colwell's dictum that ‘[a] text-type is a process, not the work of one hand’.Footnote 97 Rather than an incomplete witness to an already completed recension, then, these Ca corrections could possibly be a witness to such a process with respect to the text of Andreas. The notion of a lengthier, slower process of correction and revision, whereby older readings are mixed with the secondary readings, could thus account for the presence of older readings in later text-forms.Footnote 98 One may well wonder whether these ‘Korrekturen’ could not have, in some cases at least, originated as manuscript corrections proper.Footnote 99
An inevitable conclusion of the present discussion is that the Ca corrections, when properly identified and understood, are not what either Schmid or Bousset held them to be. For one thing, as Hernández has demonstrated, they significantly postdate Sinaiticus’ production, and thus cannot be used as the fourth-century evidence for the Andreas text. Secondly, given the textual affinities of the Ca corrections outlined above, Bousset's claim that the corrector used a Vorlage with his K text – akin to that of the text of Revelation in Andreas’ commentary – seems problematic as well. What seems more likely is that our corrections may rather prove to be an important, if indirect, witness to the development of the later forms of the text of Revelation.
