Monday, February 01, 2010

Putting the Distigmai in Their Place: Payne Strikes Back pt. 4

This is the fourt part of the series in which Payne responds to Peter Head's recent SBL presentation "Putting the Distigmai in Their Place." For background, read previous parts here, here and here.

When I heard Head’s paper I received the impression he was asserting that a comparison of the Vaticanus text in every line marked with a distigme in the Gospels to the corresponding text in Erasmus’s Greek NT shows a textual variant distinguishing 92% of these two texts. Apparently Wasserman received a similar impression, for he writes, “Peter had compared the published text of Erasmus reflecting MSS available in his time and had found that in the gospels there was a 92% match between Erasmus edition and the distigmai. If one includes the notes in Erasmus the rate goes up to 98%! This supports Niccums’ [sic] thesis. … Head thinks the 98% match with Erasmus is the death-knell of Payne’s theory.”[1] Even now that I have Head’s paper, I find several statements that seem to imply that Head was referring to variant readings between Erasmus’s Greek text and Vaticanus: “[Sepulveda’s] comparison between Erasmus’s edition and this most ancient manuscript … Vaticanus. … I believe this confirms the late date of the distigmai in the margins of Vaticanus, and even provides us with a name and setting of the person responsible.”

Listening to Head’s paper, I found this assertion of 92% / 98% correspondence between distigmai and variants in Erasmus’s text the most compelling part of his argument. I was puzzled, however, how a single Greek text (Erasmus’s) could have a higher percentage of significant variants[2] than all surviving manuscripts combined. Since Head did not identify which edition of Erasmus’s Greek NT gave these percentages, I used what I have in my library, a reprint of Erasmus’s Greek NT with his Latin translation printed in Basil by Nicolaum Bryling in 1553, to check whether these percentages accurately represent the frequency of textual variants between Vaticanus and Erasmus’s Greek NT. As a test page I used the first Vaticanus page Head displayed in his talk, page 1428, containing nine distigmai. I found that Erasmus’s Greek text varies from Vaticanus in only four of the nine distigmai lines on that page, namely 44% of them.[3] This percentage is only slightly higher than the frequency of textual variants in random lines in Vaticanus and so provides very weak support for Head’s thesis that a comparison with Erasmus’s NT text explains the presence of distigmai. This percentage is far closer to the 35% of random lines[4] in Vaticanus that contain a significant variant[5] than it is to either the percentage of NA27 variants in lines by a distigme adjacent to a bar/obelos (24 out of 28 lines[6] = 86%) or the percentage of NA variants in lines by an apricot color distigme (36[7] of 51 lines = 71%).

Since Head repeatedly associates Erasmus’s text with manuscripts of that period, I also compared how many of the fifty-one apricot color distigmai are by a line where the NA27 lists a variant in the Majority text. Only 23 out of 51 are so listed,[8] so even if Erasmus’s text has a textual variant in every one of these, this would constitute only a 45% correlation, a very low correlation compared to my own tests demonstrating a statistically strong correlation between Vaticanus distigmai and significant textual variants as listed in the NA27. Similarly, Willker writes, “Did Peter say 92% are TR variants? Compared to what? Vaticanus? Vulgate? NA? - I would like to see a table. 
In my count only about 50% are Majority/TR variants (vs. NA).”[9] These comparisons indicate that there is a very weak correlation between distigmai and significant textual variants in Erasmus’s Greek NT text compared with a very strong correlation between distigmai and textual variants as listed in the NA27. I and, apparently, Wasserman misunderstood and, consequently, were mislead by Head’s 92% and 98% figures into thinking that there is this incredibly high correlation between distigmai and variants in Erasmus’s Greek text, when in fact there is not.

Now that I have a copy of Head’s paper, however, I realize that he was not using the term “textual variant” as I was, to refer to different Greek texts, but to differences between Greek and Latin texts: “92% of all the distigmai in the Gospels match passages of variation between that exact line of Vaticanus and the Greek and/or Latin text of Erasmus. If we further take account of variant readings noted by Erasmus in his Annotations (again offering contemporary manuscript evidence) this rate extends to 98%.” Head’s conclusion further broadens the pool of comparison, “Sepulveda carefully compared Vaticanus with other manuscripts in Greek and Latin, and with Erasmus’s edition. Comparison with sixteenth-century witnesses accounts for 98% of the distigmai in the Gospels.”

Why would Head include Erasmus’s Latin text as a basis for identifying Greek textual variants? Head states that he agrees with me that “the distigmai mark places of textual variation between Vaticanus and other texts known to the dotter.” I assumed when I heard this that Head, like me, was referring to textual variants that could help establish the original form of the Greek text or otherwise explain the development of the Greek NT text. Erasmus’s Latin text is not a reproduction of any other Latin text nor does it have any independent value in establishing the original form of the Greek NT text or its subsequent development. Consequently, if one is looking for textual variants between Vaticanus and Erasmus’s text, the only text of Erasmus that is relevant is his Greek text. It makes no sense to compare Erasmus’s Latin text to find textual variants between Erasmus’s Greek text and Vaticanus. Why choose a derivative translated text for a basis of a Greek collation when the directly comparable text is on the facing page?

How does one identify textual variants between a Greek text (Vaticanus) and a Latin text? The only way I can imagine is to look for Latin translations that do not accurately reflect the Greek text and to presume that a differing Greek text caused them. I recently had an experience that shows how unreliable a translation can be for making judgments regarding textual variants. I emailed the chairman of the NIV revision committee a document identifying over 100 instances where the NIV text does not accurately reflect the underlying Greek in passages in Paul’s letters related to the ministry of women in church. If I had concluded that all or most, or even some, of these translation errors indicated the NIV text was based on a Greek text other than the one I used for my critique, I would have been wrong, for I know that the NIV translators used the same NA and UBS Greek texts that I used to make my criticisms. Consequently, one cannot assume that differences in translation, whether English or Latin, necessarily or even usually identify underlying Greek textual variants. Since my initial comparisons of Erasmus’s Greek text do not produce anywhere near a 92% correlation with textual variants in Vaticanus distigmai lines, I have serious doubts about Head’s assertion, “A careful investigation of the Gospel text of Vaticanus with a distagme [sic] (in the Gospels) and the Latin and Greek texts of Erasmus by my colleague Leslie McFall resulted in a 92% match rate.”

In order to be convincing, Head will need to establish that a very high percentage of the lines in Vaticanus next to a distigme have a significant textual variant in that same portion of text in Erasmus’s Greek NT text. Such a tabulation should not include variations in spelling, since if de Sepulveda were including things that minor, there would probably be thousands of distigmai in Vaticanus. If minor variants are included, the percentages could not be fairly compared with the percentages I have found of significant textual variants of the sort that the NA27 identifies. Furthermore, since distigmai occur throughout Vaticanus, if de Sepulveda is the source of all these in Vaticanus as Head’s thesis states, it would mean that de Sepulveda probably compared the entire Greek NT text of Erasmus with Vaticanus. Do the distigmai in Vaticanus mark all or virtually all of the locations where there are textual variants in Erasmus’s text? To the degree that differences between Erasmus’s text and Vaticanus are not marked by distigmai, Head’s thesis is weakened.

One should expect a higher frequency of textual variants in Erasmus’s Greek text corresponding to text following a distigme in Vaticanus than in random Erasmus text since the NA27 identifies textual variants in 𝔐 in 45% of such text compared to only 35% of random text in Vaticanus. Since McFall may have included as textual variants many minor variants that the NA27 does not list, I would not be surprised if he can identify significantly more than 45% correspondence between distigmai text and Erasmus’s Greek text. If McFall adds to this anything he construes as a textual variant in Erasmus’ Latin text, then, of course, that percentage will rise further.

Unless Head clearly defines what he means by “textual variant” his figures of 92% or 98% are meaningless. How minor can differences be and still fit his definition of “textual variant”? Does his definition include spelling variants? Does it include the absence or presence of nomina sacra? Does it include differences that do not affect the meaning or message of the text? Does his definition include textual variants in other Greek texts available in Erasmus’s time? Does his definition include differences in Erasmus’s Latin text? Does his definition include textual variants in other Latin texts available in Erasmus’s time? If so, what constitutes a Latin textual variant? What assurance can he provide that he is not including as textual variants the sorts of differences that I regard as errors in translation in the NIV but are not based on any Greek textual variant? I simply cannot believe that there are far more significant textual variants between the distigmai lines in Vaticanus and Erasmus’s Greek NT text than in all Greek manuscripts combined, which is what I originally thought Head meant by 92% and 98% and which is what it should mean if he defines “textual variant” as I and most others have in discussions of the Vaticanus distigmai up until now.

Furthermore, there must be a control group using the same definition of “textual variant” in order to assess the significance of percentages of correlation. If “textual variant” is defined so broadly that 92% of distigmai lines have one, but a similarly high percentage of non-distigmai lines also have such a “textual variant,” the 92% is not credible evidence, for it has no statistical significance. In light of the already established higher correspondence between Vaticanus distigme lines and textual variants in the Majority text (𝔐) than in random lines of Vaticanus and the relationship between Erasmus’s text and the Majority text, one should expect a higher percentage of textual variants in Erasmus’s Greek text corresponding to Vaticanus distigme text than in random Erasmus Greek text. One should take this into account in any conclusions drawn from comparisons of Erasmus’s “distigme text” to Erasmus’s “control text.” My own use of a control group of 540 random lines in Vaticanus was essential for getting significant chi-square probability results[10] confirming the correlation between significant textual variations and lines marked by distigmai.

Head writes with apparent approval that “Niccum noted that in 1533 J.G. Sepulveda had written to Erasmus about the results of a comparison between Erasmus’s edition and this most ancient manuscript ‘most diligently and accurately copied out in uncials’. De Sepulveda had, according to this letter, been comparing the text of Vaticanus both with Greek and Latin manuscripts extant in his time and with Erasmus’s edition, and on the basis of this study sent Erasmus a list of 365 readings, apparently where Vaticanus and the Vulgate agreed against the Greek text published by Erasmus.” Head’s paper lacks documentation that de Sepulveda added distigmai to Vaticanus or actually sent such a list to Erasmus.[11] The most obvious way for Head to establish the thesis that de Sepulveda penned all the Vaticanus distigmai in the process of comparing Erasmus’s edition to Vaticanus, would be to compare the Greek NT text of Erasmus to Codex Vaticanus and demonstrate the following two statements to be true:

1. Wherever there is a textual variant between these two texts, there is a distigme.

2. Wherever there is a distigme, there is a textual variant between these two texts.

My own preliminary comparisons of Erasmus’s Greek NT text to Vaticanus distigme lines shows that neither of these is true, nor is either anywhere close to being true.

Furthermore, if de Sepulveda himself penned the distigmai in order to identify locations in Vaticanus that differed from Erasmus’s Greek NT text, as Head’s thesis seems to postulate (“a comparison between Erasmus’s edition and this most ancient manuscript”), why when he wrote to Erasmus did he speak of only 365 variants instead of 825?[12] If Head’s explanation of the 365 is that this is limited to those distigmai passages “where Vaticanus and the Vulgate agreed against the Greek text published by Erasmus,” one might attempt to establish this by showing that 365 of the 825 distigmai lines contain textual variants differing from both Erasmus’s Greek NT text and the Vulgate text. Identifying textual variants based on a translation, however, is, as explained above, subjective and prone to error.

More fundamentally, if de Sepulveda were comparing multiple manuscripts to Erasmus’s Greek NT, wouldn’t it make far more sense for him to mark up a copy of Erasmus’s Greek NT for this purpose than to mark up irreplaceable manuscripts? If someone noted variants directly on multiple original manuscripts, he or she would have to go through each manuscript to tabulate a total. But if that person noted the variants directly in a copy of Erasmus’s Greek NT, that single source would hold all the suspect readings and would permit that person to tabulate those with relative ease. Furthermore, since according to Head, de Sepulveda’s concern was to establish errors in Erasmus’s Greek NT, that is not only the most logical place to note them, Erasmus’s text is the only text that would include all the suspect readings in question.

By Head’s view de Sepulveda had the audacity to pen “perhaps 825” distigmai in Codex Vaticanus, the NT manuscript with the reputation for being more carefully guarded than any other. Furthermore, Head’s thesis requires that de Sepulveda not only wrote on virtually every leaf of Vaticanus, he turned pages containing “more than fifty” of them while the ink was so wet these distigmai left mirror impressions on the facing page! It is hard to imagine someone in de Sepulveda’s position treating Vaticanus in such a careless manner to note variants with Erasmus’s or other texts.

Nor is it likely that a sixteenth-century scribe would mark so many other Vaticanus readings as textual variants that were standard in his day. Nor does Head’s conjecture explain the distigmai that occur where no known manuscript has a significant variant. Such occurrences are natural, however, if the original scribe was noting variants in the fourth century since most, if not all, of the manuscripts available to the scribe of Vaticanus are no longer extant.

Furthermore, neither Niccum nor Head gives any evidence that fifteen or sixteenth century scribes conventionally used distigmai to note textual variants or that de Sepulveda was even aware of this use for distigmai. Nor does Head explain what manuscript source at that time would account for the diversity of textual variants represented by the distigmai in Vaticanus. Willker observes that: “In general there is no CLEAR pattern in the witness support for the various umlauts. We have support from
- D only,
- Byz only,
- D + Byz,
- P46 only,
- some minuscule MSS only.
IMHO this indicates that not one single MS has been used for comparison, but more than one.”[13] How can Erasmus’s text by itself or in combination with other sixteenth century texts account for variants that are attested in, e.g., D alone or 𝔓46 alone?

Furthermore, Curt Niccum told me personally that he does not believe that de Sepulveda penned the distigmai in Codex Vaticanus, in spite of his earlier statement, “Evidence suggests Sepulveda introduced these [distigmai]. … Sepulveda must have shared … the reading καῦδα at Acts 27.16 … attested only in Vaticanus and Sinaiticuscorr.”[14] This reading, however, is also in 𝔓74, 1174, it, etc., cf. UBS4. This error is pivotal since Niccum argued from this reading’s rarity that distigmai “originated with de Sepulveda.” Unless Niccum has changed his view again, it is incorrect to say that it is Niccum’s position that de Sepulveda penned the distigmai. [15]

[1] Cited from

[2] As judged by the variants identified in the NA27.

[3] The four lines in Vaticanus with a different text in Erasmus’s text are in James 3:2-3, 5, 6, and 12b. The five without a variant are in James 3:7, 12a, 15, 17 and 4:4.

[4] Based on the 540 control lines identified in the table in Payne, “Fuldensis,” 253.

[5] As judged by the variants identified in the NA27.

[6] See the table in Payne, “Fuldensis,” 253 plus one I had missed, 1332 C 20 at Luke 14:24.

[7] The NA25 lists a variant in two of these that are not listed in the NA27: 1277 C 19 (Mark 1:5) and 1356 B 24 (John 5:25). Cf. Payne and Canart, “Distigmai.”

[8] This takes into account NA27 convention stated on p. 13* that “𝔐 has the status of a consistently cited witness of the first order. Consequently in instances of a negative apparatus, where support for the text is not given, the reading attested by 𝔐 may safely be inferred: if it is not otherwise explicitly cited, it agrees with txt (=the text).”

[9] Cited from

[10] My calculations include Yate’s correction for continuity. Cf. Payne, Man and Woman, 241-42, and forthcoming, Payne and Canart, “Distigmai.” Cf. the summary above in footnote 1.

[11] This is questioned by Carlo M. Martini, Il problema della recensionalità del codice B alla luce del papiro Bodmer XIV (Analecta biblica 26; Rome: Pontificium Inst. Bibl., 1966), 8, n. 20; who suggests that the existence of these readings was mentioned to Erasmus but that the list was never actually sent to him, cf. Stephen Pisano, “III. The Text of the New Testament,” pages 27-41 in the Prolegomena volume to Bibliorum sacrorum graecorum Codex Vaticanus B: Bibliothecae Apostolicae Vaticanae Codex Vaticanus Graecus 1209 (Rome: Istituto Poligrafico e Zecca dello Stato, 1999), 21. The remaining copies of this set of the Codex Vaticanus B facsimile and its Prolemonena are available at

[12] The number of distigmai by Head’s count.

[13] Willker, “Umlauts: Distribution of the Umlauts,” exactly reproducing Willker’s bold text.

[14] E.g. Niccum, “Voice,” 245, n. 20. “One can only conclude that some scholar after 1400 compared Vaticanus with another text, noting places of variation and/or agreement in the margin.”

[15] I notified Head of this at breakfast the day of his SBL paper.


  1. Thanks to Philip for this. If we recall that the SBL presentation was a 20 minute presentation on the subject of the marginalia this might explain why he hasn't been able to fully grasp an entirely different argument of which only the conclusions and none of the evidence was reported. Once the data is published the force of this argument will be able to be assessed.

  2. Payne, "Why would Head include Erasmus’s Latin text as a basis for identifying Greek textual variants? ... the only text of Erasmus that is relevant is his Greek text. It makes no sense to compare Erasmus’s Latin text to find textual variants between Erasmus’s Greek text and Vaticanus."

    The Vulgate may not be as important to us as the Greek mss, but it was certainly very important to Erasmus and de Sepulveda, and to most Christian scholars of the 16th c.