above article by J. Edward Miller, Journal for the Study of the New Testament 26 (2003) 217-36.
In 1995, New Testament Studies published a provocative piece by Philip Payne entitled, “Fuldensis, Sigla for Variants in Vaticanus, and 1 Cor. 14:34-5” in which the author argued that, in codex Vaticanus, a particular siglum indicated knowledge of textual variants by the scribe. That siglum was a horizontal bar in the margin of the text with two dots above it, which Payne coined as the “bar-umlaut” (an unfortunate term but one which has stuck due to this article). The article made quite a splash for it suggested that this famous fourth-century manuscript indirectly commented on dozens of textual variants known at that time, and thus that it contained valuable information about the state of the text of the New Testament in the fourth century. Payne took this one step further: he noted that the “bar-umlaut” at the line above 1 Cor 14:34-35 indicated that the scribe was aware that these verses were textually suspect. To Payne, this was a sufficient external basis to argue that, even though these two verses are found in all relevant extant manuscripts (though the Western witnesses place them after v. 40), they should probably be athetized.
This controversial conclusion led to a spate of responses—both pro and con—as well as a reexamination of codex Vaticanus (a.k.a. codex B).
The most thorough examination of Payne’s thesis, as well as an even bolder interpretation of the data, is Miller’s JSNT article. Inter alia, Miller argues “for the disjunction of the bar and the umlaut sigla by demonstrating their mutually exclusive functions: the bar is a section marker and the umlaut is a text-critical indicator” (219). What is remarkable about this conclusion is that instead of just a few dozen text-critical indicators in Vaticanus, Miller believes there are over 750 such indicators! If he is right, then Vaticanus is, in Miller’s words, “an early UBS text!” (219, n. 6) for it has about half as many text-critical notations as the modern UBS Greek New Testament, and more by far than any other ancient source, be it manuscript or father.
Miller demonstrates his point via four routes. (1) Old Testament “umlauts” are normally next to lines which modern texts of the LXX reveal as having textual variation (though Miller does not tell us which critical texts he is using). (2) Statistical probability in Matthew, used as a randomly selected control group, demonstrates that 59% of the “umlaut” lines correspond to variants listed in Nestle-Aland27 while the unmarked lines corresponded to variants only 27% of the time.1 In other words, the “umlaut” lines were more than twice as likely as unmarked lines to find textual variants in NA27. (3) Parallel passages provide significant evidence that the “umlauts” are indeed text-critical markers. For example, at Luke 10:1 and 10:17 B has ἑβδομήκοντα δύο while other witnesses have ἑβδομήκοντα. At both lines the “umlaut” appears, yet it does not appear in the seventy-eight intervening lines. As Miller notes, this is “convincing evidence that the scribe was employing the umlaut for text-critical purposes.” (4) Finally, Miller compares the “umlauts” in B against known textual variants. Although this method involves several shortcomings (as Miller notes), it is helpful to confirm that the “umlaut” in B does indeed indicate textual variation. The “umlaut” is found next to lines that involve well-known variants in Matt 5:22; Mark 1:2; John 7:39; 1 Thess 1:1; 2:7; etc. In sum, the evidence provided by these four tests “demonstrates that the hundreds of umlauts in the New Testament portion of Vaticanus were intended to signal the reader to textually uncertain lines. Furthermore, there is no so-called ‘bar-umlaut’ siglum. Rather, instances where the bar and umlaut accompany the same line of text are best regarded as coincidental” (231-2).
This is a remarkable and ably defended conclusion. By examining the data from four different angles, Miller has shown that, in all probability, codex B contains over 750 text-critical notations. These lines in Vaticanus now will need to be examined for what they might tell us about the state of the text in the fourth century.
Miller then revisits 1 Cor 14:34-35. He argues against Payne’s view that the siglum was placed beside the line preceding a multi-line variant. Instead, Miller argues, the “umlaut” was placed next to the line in which a variant was known to exist, regardless of whether that variant was a single word or several lines. Miller demonstrates this with illustrations from John 12:7; 16:14-15; Rom 11:6; and Jude 22-23, among others. He notes that “It is unlikely that the scribe would abandon this habit only in the case of 1 Cor. 14:34-35” (233). Payne’s lone example, apart from the siglum preceding 1 Cor 14:34-35, is in John 7:52. Payne argues that the scribe was aware of the Pericope Adulterae, yet an ancient variant (listed in Tischendorf’s 8th edition but not in NA27) occurs in the same line of John 7:52, involving a change from εγειρεται to εγηγερται. In light of this consistent practice in B, the “umlaut” at 1 Cor 14:33 does not indicate any knowledge of a variant in vv. 34-35.
Miller’s article involves an examination of the data that were neglected in Payne’s study. As such, his conclusions rest on a much more certain basis. I found his analysis sober minded and his arguments compelling. In the least, Miller’s study provides four important conclusions: (1) hundreds of text-critical decisions (not just dozens) made by an early professional scribe are accessible to scholarly examination, which gives us a unique window on the principles employed by this scribe. (2) Since many of these variants are found today only in later manuscripts, Miller believes they “can now be dated to the early-fourth century with some measure of confidence” (235). One issue not resolved by the study, however, is which variant is in view by the use of the “umlauts.” For example, if two variants show up in NA27’s apparatus for an “umlaut” line, which is in view? Further, even if only one variant showed up in the apparatus, how can we be sure that that variant is what B’s scribe had in mind? That this second problem is not just theoretical is confirmed by Miller’s third conclusion: (3) “[F]or those marked lines failing to yield extant variants, New Testament scholars must acknowledge the likelihood that some variants known to exist in the early-fourth century have been lost” (ibid.). (4) Ironically, though this same point was Payne’s argument with reference to 1 Cor 14:34-35, Miller has shown that although now-lost variants are noted hundreds of times in Vaticanus, the athetization of 1 Cor 14:34-35 is not among them.
Miller’s study may well mark a bold new chapter in textual research. Codex B needs to be reexamined for the rich data that it contains. Further, textual critics may wish to revisit scores of other ancient witnesses to see if they use a similar device to note textual variation. To date, only Vaticanus is known to have text-critical notations by the first hand. But few manuscripts have been examined with this objective in mind. Further, there are numerous unexplained sigla in the manuscripts of the New Testament. Perhaps among them are a few text-critical notations.
Several questions remain, however. For example, on what basis did this scribe note various variants? Were they simply those known to him or were they serious alternatives in his mind? What kinds of variants in terms of textual affinities are hinted at by the “umlauts”? That there are 140 such sigla in Acts may suggest that the Western readings were known to the scribe of B but rejected by him. But what other text-types can be postulated as underlying the variants that the scribe notes? Is he aware of any variants that show up only in the later Byzantine text for example? Finally, what does B say about several substantial and well-known variants? Does it place a siglum next to Mark 16:8 or John 5:3, for example? On this issue, I checked a handful of places (in a rather non-systematic investigation) in the recently published magisterial Vaticanus facsimile to see whether any “umlauts” appeared next to the line of several disputed texts. The results are tabulated below.
The first group of passages involve additional material that is not found in Vaticanus. Miller notes that “The Vaticanus scribe consistently places the umlaut next to the line supplying the beginning of a questionable reading, whether long or short (and whether the text is included in or omitted from Vaticanus)” (232). In other words, those doubtful passages lacking in B but known to this scribe are indicated by an “umlaut” next to the line in which said variant would have begun had the reading been in B.
There is no “umlaut” at the end of Mark 16:8 to indicate any of the longer endings, nor at Luke 23:34a to indicate recognition of the cry on the cross (“Father, forgive them for they don’t know what they are doing”). John 3:13 has no “umlaut” where the wording “who is in heaven” would have gone. Likewise, John 5:3b-4 (lacking in B) has no “umlaut” where this text would have gone; so also Rom 8:1 (where later manuscripts have “who do not walk according to the flesh but according to the Spirit”). In Eph 1:15, B lacks the wording “the love toward all the saints” as well as any siglum to indicate this variant. (Here, most scholars would assume that B erred via haplography, which would preclude the scribe’s knowledge of any variants because the shorter reading would have been an oversight. But that the shorter reading is in error is not a settled conclusion. Had there been an “umlaut” in the margin here, however, it would perhaps strengthen the case that B’s reading was intentional.) Miller notes a few places in B where the scribe seems to indicate additional material, but the above-mentioned passages are not among them.
Turning to variants which do not involve omissions in B, note the following. In John 5:2, the line which reads βραιστι βηθσαιδα πεν has an “umlaut” in the margin. Most likely, this is to indicate that βηθσαιδα is in doubt (NA27 prints βηθζαθα as the original reading here). In Rom 5:1 no “umlaut” appears next to the εχωμεν, a reading that was changed by a later hand to the indicative εχομεν. I, for one, had hoped for the “umlaut” here since I strongly suspect that the indicative is authentic. The benediction in Rom 16:25-26 (which is also found in some witnesses at the end of chapter 14 or chapter 15) is “umlaut” free—both here and at 14:23 and 15:33 (the other two locations where it is found in the witnesses). The second line for 1 Thess 1:7 (υμας τυπον πασιν) has a siglum, which most likely indicates knowledge of the variant τυπους.
The lack of “umlauts” noted in our brief non-systematic study might not tell us much (but the presence of them in at least two of the passages certainly does). Too many questions still need to be answered before we can make any firm conclusions from an argument from silence. But Miller’s study has provided the stimulus for our thinking, offering a new way to read this magnificent codex.
In sum, Miller’s study is a refreshing piece that takes Payne’s original insights one step further. In so doing, it may well become the catalyst for several other studies that unlock some of the treasures hidden for centuries in this most precious copy of the scriptures.
1 Miller critiqued Payne’s similar approach in which he looked at the “bar-umlaut” in relation to textual variation (p. 226, n. 29). In essence, Payne did not look at a control group which meant that the results of his study were not falsifiable.