Textual Evidence

The "Agreement"
between P75 and B

Excerpt from: Nazaroo, NT Papyri don't support Hort, (TOL forum thread, 2009)

Page Index

Introduction: - Early Papyri

  I. Fragmentary Evidence: - they only cover 30% of NT text

  II. Lack of Support for W/H: - & they differ strongly from each other
    (a) P45:Chester Beatty I - F. Kenyon on P45: - non-Alexandrian
    (b) P66:Bodmer II - E.C. Colwell, J.N. Birdsall on P66: 'impure'
    (c) P75:Bodmer XIV-XV - K.W. Clark on P75: a wild text?

  III. Limited Use: - proper application of papyri still unclear
    (a) P46:Romans - H. Gamble on P46: problems with papyri

P75 and B : - Claims versus Facts:
    (1) Third Century - not "2nd Century"...
    (2) 95% Agreement? - not likely ...
    (3) Sarah Edwards - calculations examined...
    (4) Colwell's Collations - for comparison

Return to Index


The Early Papyri

There are three main observations which must be realistically faced when dealing with the papyri:

Only Partial Texts
Have Survived

(1) The papyri only cover a fragment of the NT text.

Even with newly discovered and published finds, the coverage is scanty.
Pickering sums it up:

"..only about 30% of the NT has early papyrus attestation, and much of that 30% has only one papyrus."

- W.N. Pickering, Identity... (1980), p.77

Hurtado describes the situation lucidly:

"... extant (surviving) MSS that can plausibly be dated to the 2nd century [100-200 A.D.] are lamentably few in number, and none of them gives us a complete text of any NT writing. In fact, the extant 2nd cent. manuscript (MS) evidence consists largely in a handful of MSS.

...the amount of text preserved in the total body of 2nd cent. MS material is still frustratingly small.

- L. W. Hurtado, "The NT in the 2nd Century",
Transmission and Reception:..., (Brill, 2006), p.5 fwd

Lack of Support for Hort

(2) Much of the time, the papyri don't support Hort's text (WH).

Hort's text (along with most modern critical editions of the NT) was based mainly on the shared readings of Codex Vaticanus (B) and Sinaiticus (א).

A lot of the time, the papyri depart from א/B significantly, sometimes drastically, often in support of 'Byzantine' or 'Western' readings, and even readings previously only found in Versions (early translations) as well.

(a) Chester Beatty I - P45

P45 was classified by Streeter and Kenyon as of the Caesarean Text-type:

"...(P45) is very imperfect, consisting of 30 leaves, nearly all incomplete, out of a probable total of 110...

[in] St. Matthew, enough is preserved to show clearly the character of the text. ... It is not identical with any of the main families, the Alexandrian (the ‘Neutral’ of Hort) or the Western, into which the MS evidence has been classified by modern scholars.

In Mark it clearly belongs rather to the 'family' which Streeter identified as ... ‘Caesarean’, though this MS strengthens the probability that the origin of this family was in Egypt, from where it may have been carried by Origen himself to Caesarea.

In Luke & John ... [P45... is] intermediate between the Alexandrian and Western text-types, but slightly nearer to the former.

In Acts it is distinctly of the Alexandrian type, having a few of the minor variants so noticeable in that book."

- Sir Frederic Kenyon,
The Bible and Modern Scholarship (1949)

This classification has since been mostly abandoned, and the character of P45 is now described as 'eclectic, mixed, or unaligned'. But this does nothing to alter its readings, and its obvious differences from the WH, NA, and UBS critical Greek texts. Based on Klijn's work, Pickering notes that

"where Aleph and B disagree, and at least one papyrus (out of P45, P66, P75) joins either, (i.e., in 43 variation units found in the text where they all overlap)...

Against the other two papyri:
Papyrus P45 stands alone with:
א  19 times = (44 % agreement),
B  24 times = (56 % agreement),
TR 32 times = (74 % agreement!).

- W.N. Pickering, Identity..., p.55

If nothing else, this reveals the folly of not including the Byzantine Text (TR) in any attempts to classify papyri. Wherever א and B are proven corrupt (through their own disagreement with each other), P45 more frequently prefers the TR to either one of them!

(b) Bodmer II - P66

P66 is a similar case. E.C. Colwell describes his own findings:

"The Bodmer John (P66) is also a witness to the early existence of many of the readings found in the Alpha text-type (i.e. the Byzantine Text, Hort's "Syrian Recension").

"Strangely enough, to our previous ideas, the contemporary corrections in that papyrus frequently change an Alpha-type reading to a Beta-type reading (Hort's "Neutral", i.e., the Alexandrian).

"This indicates that at this early period readings of both kinds were known, and the Beta-type were supplanting the Alpha-type -- at least as far as this witness is concerned."

- E.C. Colwell,
Origin of Text-types, p.135

E. F. Hills describes P66 as follows:

P66 contains 13% of all the alleged late readings of the Byzantine Text in the area which it covers (18 of 138 readings).

13% of the 'Byzantine' readings which most critics have regarded as late have now been proved by P66 to be early readings."

- E.F. Hills,
The King James Version Defended, (DM, 1956) p .54

Pickering notes again that,

"where Aleph and B disagree, and at least one papyrus (out of P45, P66, P75) joins either, (i.e., in 43 variation units found in the text where they all overlap)...

Against the other two papyri:
Papyrus P66 stands alone with:
א  14 times = (32 % agreement),
B  29 times = (67 % agreement),
TR 33 times = (77 % agreement!).

- W.N. Pickering, Identity..., p.55

Although it favours B over א ( 2:1 ), the oldest known copy of John even more solidly prefers the Byzantine Text ( 4:1 ) in all testable cases (i.e., when א/B conflict), even when it is at variance with both of the other early papyri.

G. D. Fee is no supporter of the Byzantine Text. He and various other critics want to continue to class these early papyri as "Alexandrian". Pickering contends with him on this point, citing Fee's own data:

Both P66 and P75 have been generally affirmed to belong to the "Alexandrian text-type".58 ...

G.D. Fee goes to considerable lengths to interpret the evidence in such a way as to support his conclusion that,

"P66 is basically a member of the Neutral tradition" 61

- but the evidence itself as he records it, for Jn ch. 1-14 is as follows:

P66 agrees with:
P75 : - 280 / 547 (51.2%)
B  : - 334 / 663 (50.4%)
TR : - 315 / 663 (47.5%)
א  : - 295 / 662 (44.6%)...

Does this evidence really suggest "two clear textual streams"?

- Pickering, , Identity..., p.56

Original Footnotes:
58. cf. Metzger, Textual Commentary..., (Lond. 1971) p. xviii.
61. G.D. Fee, Pap.Bod.II (P66): Its Textual Relationships..., (Utah 1968) p. 56.

Statistics like these have forced many to call these papyrus texts wild.
J.N. Birdsall described P45 and P66 as follows:

"In these 3rd century MSS, whose evidence takes us back into the mid 2nd century at least, we find no pristine purity, no unsullied ancestors of Vaticanus, but marred and fallen representatives of the original text. Features of all the main texts isolated by Hort or von Soden are here found -- 'mingled' very differently in P45 and P66."

- J.N. Birdsall,
The Bod. Pap. of the G.of John (1960) p.17

(c) Bodmer XIV-XV - P75

"...the papyrus (P75) vividly portrays a fluid state of the text at about A.D. 200.

Such a scribal freedom suggests that the gospel text was little more stable than the oral tradition, and that we may be pursuing the retreating image of the 'original text'.

- K.W. Clark, "Theological Relevance of Textual Variation...", JBL 85 (1966), p.15

Lets pause for a moment and consider this remarkable statement.

If P75 were really in "95% agreement" with Codex B, how could any expert make such an incredible claim, and survive the subsequent scorn and ridicule? If the text has not significantly changed from the proposed time of P75 (c.200 A.D.) to that of B (c. 330 A.D.), what is he talking about?

The answer is novel, but not unexpected.

P75 IS NOT "in 95% agreement" with Codex B. This is utter nonsense. The two MSS differ in hundreds of important readings, and other significant details, even in the short span where they overlap.

And its not just the findings of one researcher, like G.D. Fee, quoted above, in passing (where P75 and P66 were found to disagree a startling 51% of the time). Obviously, BOTH P66 and P75 cannot be "in 95% agreement" with Codex B. In fact, neither of them are.

As Pickering tells us, using Klijn's data, P75 agrees with Aleph 9 times (against B); and with B 33 times; -- but, P75 also disagrees with both Aleph and/or B in the same group of readings a stunning 29 times in favour of the Textus Receptus (i.e., the Byzantine text).

If the large number of "agreements" between P75 and B is due to B's agreement with the TR against Aleph, this "agreement" is meaningless. It only confirms the text of the TR against Aleph. All these readings must be discounted, certainly as original, but not as Alexandrian readings. Only the unique readings of B can be of any value in assessing the agreement between B and P75.

Regarding "agreement", Fee himself shines a light on the true state of affairs. He plots the % agreement between P66 and P75, B, Aleph, A, C, D, W chapter by chapter.

For each document, the graph bounces up and down chapter by chapter in a unique erratic fashion. All show a range of variation in agreement of over 30%: Codex B for instance goes from 71% agreement with P66 in ch. 5 to a low of 32% agreement in ch. 7. With no correlation found between the given comparisons, a similar erratic plot between B and P75 must be assumed confirmed. (Pickering, ibid., p.55-57)

Colwell found independantly that P75 had 27 leaps forward and 10 leaps backward, all unique (no other MS support, including B!). How then with 37 significant differences in a mere 14 chapters, can anyone say P75 and B are in "95% agreement"?

The Papyri:
Questionable Value

(3) They aren't all they are cracked up to be.

(a) They still aren't old enough to significantly advance our knowledge of the state of the text in the critical period between 80 - 200 A.D.

Example: Letter to the Romans

The text of the Pauline Letters, particularly Romans is a case in hand: Although P46 is a substantial early papyrus witness for Romans, its text must be evaluated very carefully, and its use is limited. One of the most even-handed assessments of P46 comes from H. Gamble Jr., one of the world's leading experts on the text of Romans:

P46 and Romans

"The discovery of the Chester Beatty Papyrus of the Pauline letters (P46) was a matter of great interest for the question of the textula history of Romans, as it provides the doxology not at the ent of ch. 16, which the papyrus does contain, but between 15:33 and 16:1. The intrinsic interest of this unique reading is enhanced by the fact that P46 is by far our oldest MS of the Corpus Paulinum, dating from around the beginning of the 3rd century. Thus it antedates our oldest codices, B and Aleph, by a century and a half.

In it we have a form of the Pauline text which was current prior to the great normalizing recensions.

Although the value of the MS is great in these and other respects, it had not worked a revolution in our understanding of the history of the Pauline text, either in general or in respect of particular readings.

In textual complexion the affinities of P46 lie mainly with B and 1739. But the papyrus also exhibits numerous agreements with Western witnesses over against the Alexandrians, and so indicates that many variants previously termed "Western" are by no means either strictly Western or late, and in some cases my very well represent the correct text (see Zuntz). Thus Zuntz has denominated the type of text seen in P46 as "proto-Alexandrian".

It cannot generally be said that the papyrus uniquely preserves important readings with claims to originality. Of course many special readings are to be found in it, but the vast majority of these must be set down to scribal error and alleviative conjectural emendation."

"...Far from solving the problem of the textual history of Romans, as some have too readily assumed, P46 has only complicated the evidence."

(- H. Gamble Jr.,
The Textual History of
the Letter to the Romans
Studies & Documents 42, p.33-34)

The results are similar with other newly discovered early papyri. They add to our knowledge, but on the one hand, most of the textual problems are not solved by their discovery, and on the other, many new problems have arisen from their evidence instead.

Return to Index

P75 and B
Their Agreement


The alleged agreement between

Taken from:
Nazaroo, NT Papyri don't support Hort, (TOL forum thread, 2009)

Headings have been added for clarity and navigation purposes.

P75 and B

Their Alleged
'Agreement' Unmasked

A typical poster on another forum made the following claim:

"...compared to the VERIFIED STUDY that
P75 (2nd cent.) and Codex B
have 94% affinity"

There are two key falsehoods to take note of in this rhetorical double-talk:

(1) P75 is Likely THIRD century, not "2nd century":

I quote from the latest listing of all NT manuscripts given at the huge Muenster textual-critical center in Germany:

Shelf number (name): P. Bodmer XIV-XV

GA number: P75

Century: III = 3rd century (A)

Content: L 3,18-4,2+; 4,34-5,10; 5,37-18,18+; 22,4-24, 53; J 1,1-11,45. 48-57; 12,3-13, 10; 14,8-15,10

Address: Bibliotheca Bodmeriana, Cologny/Genf (Schweiz)


But lets get a second opinion:

"Victor Martin originally dated P75 to the early 3rd century on the basis of comparability to MSS such as P. Oxy 2293, 2322, 2362, 2363, and 2370. All these mss, as with P75 display the defined hand of the early 3rd century.

The Oxyrhynchus editors ...however said the date should be late 2nd or possibly early 3rd century for 3 of these mss.: P.Oxy 2293, 2363, and 2370."

"Seider dates P75 "2nd/3rd century" (Palao.der griech.Pap. vol.2, Stutt. 1970, p.132).

I [Comfort & Barrett] would date it late 2nd century, possibly early 3rd."

Text of the Earliest NT Greek MSS,
Comfort & Barrett, (Tynd. 2001) p.501

It seems that even among those who would like to date it 2nd century, there is no real commitment to be found.

The poster appears to be about 50 years "off" in a direction that favours his own position, while being rather coy about the wide range of possible dates already assigned to this manuscript.

Typically most NT manuscripts can only be dated "plus or minus 100 years" with any reasonable accuracy, even when their discovery location is a well-known and well-researched site. Of course, being an expert, our poster should KNOW this.

(2) 94% "affinity" between P75 and B?

Not really:

Comfort & Barrett kindly give the accurate calculations and their source:

"It is also well-known that the text of P75 was of the sort used in formulating Codex Vaticanus (B);

the texts of P75 and B are remarkably similar, demonstrating about 85% agreement.

(see C. Porter,
"Pap.Bod XV (P75) and the Text of Cod. Vat.",
JBL 81 (1962) 363-376)."

Thus the normal understanding is about an 85% agreement, based on a thorough collation by C. Porter in 1962.

It seems the "94% affinity" (with a little exaggerating by our poster) can actually be tracked down to yet another less than ideal source:

"What does this examination of P75 under the magnifying glass add up to?

...It reveals an astonishingly high 92 % agreement between B and P75. This figure was obtained by collating B and the other manuscripts noted in the first chart against P75, with these new readings, as a base.

By comparison, P66 is shown to read with P75 about 83 %, and א only 79 %."

- Sarah Alexander Edwards,
"P75 Under the Magnifying Glass",
Nov.Test. Vol 18, (Jul /76), pp.190-212

First we notice its 92%, not 94%. Things just seem to get better and better when posters put a little extra 'spin' on sloppy quotes.

But wait: What "new readings"?

Sarah Edwards' Calculations

Lets be clear exactly what happened. Normally direct access to P75 is simply not allowed. It resides in a very exclusive repository, and even the most prestigious scholars have only been allowed to look at them briefly between glass plates.

As Edwards confesses,

"Most of my work on P75 for Under the Magnifying Glass was done at home from the published photographs and transliterations of the papyrus.

After many futile attempts over the years to obtain permission for a working visit in the Bodmer library, through the persistence of several Swiss friends I was finally allowed to come in April /74.

The Foundation Martin Bodmer arranged to have the Johannine leaves of P75 removed from the vault and placed between sheets of plastic for my use, but I was not to be permitted to touch them.

The Director, Dr. Hans Braun, carefully laid the first page in front of me - but before I'd read ten lines he re-appeared with a big smile and a great stack of papyrus and put fourteen chapters of P75 into my hands.

For two wonderful days I checked and rechecked every line. ... "

- Edwards, ibid, p 194

Even Edwards only had two days to examine 14 whole chapters of John (and those only).

Whatever the value of a mere two days (qualified scholars of Aleph have had months and even years to study it in the British museum, by contrast), what Edwards did was take her transcriptions and notes, and work at home for several months, reconstructing her own version of the many lacuna in the textual pages.

She came up with 18 'probable readings' differing completely from all previous collations and reconstructions, and on the basis of this limited sample (of John), made her own 'calculations' (not presented to the public) and definitions regarding "% agreement"

In order to boost the estimated (and we underline estimated) "% agreement" from 'about 85%' to '92% agreement', Edwards invented 18 new alternate reconstructions of places no longer legible or extant!

To lend some credibility to her work on areas of the text which don't actually exist, Edwards adopted a supposedly 'improved tracing technique' called the 'Huston method':

"The matter of spacing is of crucial importance in reconstructing the text of a damaged page. The longer the lacuna, the more imperative it becomes to exercise as much control as possible over the length of the reading proposed to fill in the missing portion.

When a scribe writes with a regular hand such control can be exerted by graphing. In this method, the perceivable text is first written in, one letter to a squareas in a cross-word puzzle, taking care to place these letters in the proper relationship one to another. ...

"A second technique involves tracing letters in the scribe's own hand. As the center examples show, it is far to easy to change the text by stretching out the proposed reading, or conversely, by squeezing in an extra word. ...

The most reliable procedure is an improved tracing technique developed by Hollis Huston. The missing letters are traced in combination so as to reproduce the spacing and other peculiarities of the scribe's hand."

- Edwards, ibid.

Regardless of the real or imagined 'improvement' in accuracy or guesswork for conjectural stretches of text across missing pieces of page, one must by force make a few critical observations:

(1) Any scribe can write any size, and scribes often have more than one 'hand' or style. The result is often far more uneven than supposed, even with a large sample of text surrounding the missing sections.

(2) Variations in the form and space of certain letters are large and frequent in the most carefully written handwriting.

(3) Conjectural proposals by their very nature are extremely subjective and open to bias by the reconstructor, both by what is hoped or expected, and by simple lack of complete knowledge of possible variants available to someone writing 1,800 years earlier from an unknown exemplar.

(4) Such work must be open to a peer review process of scholars with competing theories and diverse interests, in order to catch any possible bias or error in technique or assumption or result.

(5) Such work must be fully disclosed, as to both purpose and funding of the researchers involved.

As to this specific case, several suspicious elements and claims are involved, and questions need to be asked:

(1) Who is Edwards, and how did she get access to P75?

(2) Who financed and coordinated the project?

(3) How did 2 days access to 14 chapters work out? We have no doubt she was monitored while inspecting them. But who can confirm exactly how much collating was actually done, and even whether or not it had any impact on her results? These after all are conjectural reconstructions, so how did access to key areas of the papyrus affect them? Did access even make any difference to her result? It is already admitted that much of the work was done "offsite", probably long before she examined P75. What was changed if anything?

(4) How many other scholars have accepted all her suggestions, or even some of them? This area is a notoriously controversial one. Are we to think her article was accepted without any modifications or doubts?

(5) Which groups of scholars checked her work, approved of her methods, and accepted or reproduced her findings?

(6) Who benefits if the "% agreement" between P75 and B is increased or decreased? What theories are strengthened and which weakened, and which did the researcher hold before beginning?

(7) Where are the full open disclosure of definitions of "agreement" and calculations for same? They are not in the article. Edwards provides some sample diagrams, and a brief discussion. But where is the safety check?

(8) Agreement in John most certainly does not mean agreement in Luke or any other Gospel. It is reasonable to assume a single Gospel might be homogenous in text-type , but it is well known that this doesn't hold at all between gospels, and often fails within a single gospel too.

These are just a few of the reasonable scientific limitations on any exaggerated claims of "92% agreement", especially when that "agreement" is plainly ill-defined,

...and most especially when there is a spread of a 7-8% descrepancy between Edwards' findings and that of independant researchers doing more exhaustive comparisons over larger portions of text and more gospels (e.g. Luke/John).

This limited range of text where P75, B, Aleph and P66 overlap, which was used for the measurement is simply the 1st 14 chapters of John (with many lacuna).

But surely the most stunning revelation is that the "92% agreement" is not between P75 and B at all, but between the conjectured text of Edwards and that of B, where P75 is no longer extant.

Colwell's Collations

Is the papyrus P75 really in such close agreement with B?

Previously we showed that contrary to our poster's exaggerated claims of "93% agreement" between P75 and B, the real position of most scholars has been in the range of "85% agreement".

I return to that discussion here, in order to address some additional factors which bring that "85% agreement" claim even lower, by at least an order of magnitude:

Now, I'd like to bring the reader's attention to what Colwell had to say about this aspect of the issue, that of how this "agreement" was calculated:

First, in discussing ITACISMS ("the interchange of similar sounds in diverse spellings") Colwell has this to say about the three papyri, P45, P66, P75:

"These "itacistic spellings" are numerous.

More that 635 have been tabulated in the three papyri:

P45 has approximately 90 itacisms,
P75 about 145 itacisms, and
P66 has 400 itacisms.

Exact comparisons as to frequency are difficult, due primarily to the consistently fragmentary nature of P45, and the wide variation in the amount of text per page.

[But] If number of pages be taken as a base,

P45 has 60,
P66 has 148,
P75 has 199.

If number of verses be taken as a base, P45 has 795 (at least in part),
P66 has 808, and
P75 has 1,406.

In general P75 has the fewest itacisms per square inch, and P66 has the most." (p. 110)

"...It the itacisms are to be cited at all in our apparatus criticus, they should all be cited, since comprehensive survey of the evidence is essential... But stop and think what that means.
...If these three papyri, fragmentary as they are, give us more than 600 itacisms to record, the total for all witnesses would crowd hundreds of pages with relatively insignificant lore..."

Let us then remove these itacisms from our list of Singular Readings in P45, P66, and P75. The total of 1,649 Singular Readings is reduced to 1,014:

P66 has 482 singular readings,
P45 has 275 singular readings,
P75 has 257 singular readings, exclusive of itacisms."

(p. 111)

With this discussion by Colwell, we now have a clearer idea of how many variations are actually found in the manuscripts, versus how many are actually taken into account as "significant" by critics.

Colwell for instance, has in the process of simplifying the data, removed over 600 differences between the manuscripts, because they are possible spelling errors, and "probably insignificant".

But he's not done yet:

"Nonsense Readings include words unknown to grammar or lexicon, words that cannot be construed syntactically, or words that do not make sense in the context.

In P66, 2 out of 5 Singular Readings are Nonsense Readings - 40 % [!]

In P75, 1 out of 4 are Nonsense Readings, - 25%.

In P45, less than 1 out of 10, - 10%.

If these Nonsense Readings are removed from consideration, the number of Sensible Singular Readings remaining would be:

in P66, 289 Sensible Singular Readings,
in P45, 250 SSRs,
in P75, 190 SSRs. "

(pp. 111-112)

What has Colwell just told us?

That he has eliminated another 10-40% of variant readings from the count, because he could not identify what the scribe meant in writing them.

Keep in mind that they may appear as "nonsense" to us with our fragmentary Lexicons and grammatical knowledge, but to an ancient Egyptian scribe using an Alexandrian dialect of Koine Greek, they may make perfect sense.

As Deissmann has long ago demonstrated, our knowledge of NT Greek is at best scanty, and can only increase by orders of magnitude over the next century as the many THOUSANDS of papyri are further studied.

But the key point here is that before ever even beginning to collate and calculate "agreement" between NT papyri, just about every researcher is compelled to throw out hundreds of serious differences from consideration, even to compare a small portion of text between only a handful of MSS.

Now when a researcher has narrowed the field of consideration to a manageable size (i.e., less than several dozen "significant readings"), we are compelled to acknowledge that the researcher is no longer comparing "manuscript to manuscript" at all:

The researcher is actually comparing abstract 'lost ancestor texts' underlying the two manuscripts, not the real text of the manuscripts at all.

These two texts are both "conjectural entities" having only a very fragmentary and indirect manuscript support, in the context of some theory about the "lost texts" many generations older than the actual manuscripts.

That is, when someone like Porter or Colwell cite an "85% agreement", they don't actually mean an agreement between the extant texts, but an agreement between the hypothetical documents representing "ideal" ancestor texts that have been reconstructed as best as possible by eliminating and/or ignoring readings which remain either apparently 'insignificant' or insoluble by textual critics.

Just looking at the admitted numbers themselves, of readings which are being consciously ignored and that may be significant (e.g., "nonsense readings"), we can RE-estimate that an "85% agreement" between hypothetical 'ancestor texts' is likely to be as low as a 40-60% agreement between actual manuscripts.

This is more in line with what many independant examiners of these manuscripts have actually found in real life. All "agreements" between MSS are abstract constructs, made by eliminating large portions of text and severely limiting samples for comparison.

Is P75 in "85% agreement" with Codex B?

Not on this planet.

But if pigs could fly...

Return to Index