The Ancestors of Aleph/B: Synoptics

Now that we've collated all the probable accidental errors for the Synoptics, and grouped them by column-width, we can arrange them also by probable age based on the same factor, and compare the independent early history of each gospel in the hands of the Alexandrians, and in particular the Caesareans.

The following chart-pieces can be clicked on to enlarge, and you can navigate back using the back-button in your browser.

Luke (20 - 25 chars per line)

It appears that Luke took a uniquely bad beating during the earliest copying phase, when wider columns were popular.   These omissions were not so much the result of bad copying (all copyists make these errors), but rather a result of little or no systematic proof-reading techniques, and lack of experience with the kind of errors most commonly found.    This discovery is not particularly surprising, given that Luke would be most popular among Gentile converts and Jews of the Diaspora, where professional scribal technique would be lacking, and a learning-curve would be in operation.

Matthew and Mark have only suffered half as many losses in this period, probably due to the greater experience and care Jewish-Christian copyists were able to exert on the task.

Matthew (17-19 chars per line)

Next, in an intermediate period, while the Gospels were still being copied independently in double or triple columns, Matthew now takes a savage beating from the same loose environment lacking adequate error-correction techniques.   The others suffer moderate losses at the hands of different copyists.   All of this could have been accomplished in a single manuscript in which either different sources were used for each gospel, or different scribes copied the sections.

Mark and Luke (14-16 cpl)

Mark now gets hit hard in this later period, involving MSS with 3 to 4 columns per page.  Luke also suffers significant damage at this time, probably early in the 3rd century.  Matthew, probably due to better error-checking, suffers only moderate damage during the same phase.   These differences must now be traced to either individual scribes or correctors responsible for each gospel, since they will now most likely be bound together in single books containing all four gospels.

Mark (11 -13 chars per line)

Mark now takes one last pounding, at the hands of some scribe or lazy corrector, while being copied from a MS having 4 columns per page, like Codex Sinaiticus.  Luke's Gospel also takes significant collateral damage in this line of transmission.

It may be significant that these two gospels are usually physically adjacent and in the middle of the book, in copies of the four gospels.  One can picture a corner-cutting corrector or scribe only proof-reading the first part of the book (Matthew) and possibly the last (John), so that an overseer might be fooled into believing the entire MS has been properly corrected.

Finally, minor omissions of unknown date and circumstance also accumulate in the gospels, on a smaller scale.

The picture is interesting, but tells a sad story, in which Gospels like Luke and Matthew, immensely popular once they were published, took an early beating at the hands of inexperienced scribes using inadequate error-correction techniques.

Conversely, and perhaps more importantly, the damage that Mark suffered seems to have mostly occurred in the later period of multiple-column manuscripts.

It seems that someone in Caesarea, or perhaps Alexandria,  mistakenly took an older copy of Mark for a more accurate text.  But this text was already badly mutilated from its own recent history of inadequate correction.   The reputation of this copy seems to have been acquired from a misplaced trust in its source, perhaps Origen or some other bishop.  This lapse in judgment led to several manuscripts like Sinaiticus and Vaticanus being produced with an inferior text of Mark. 

This same copy was also probably the source of the missing ending of Mark, which had likely lost its last page.

It is somewhat ironic that many textual critics imagine that the Alexandrian text of Mark is a relatively older and purer form of the text, when it appears to have suffered most of its damage in the latest era of its peculiar textual transmission history.   This text apparently originated sometime in the early 3rd century, when MSS were being copied in narrower columns.   As far as accidental omissions of significant parts of the text, the Aleph/B text of Mark is probably the worst surviving text known.

Luke has also suffered serious damage in the Alexandrian stream of transmission, but being a much larger book, this damage is perhaps less on a percentage basis, than that suffered by Mark. 


mr.scrivener

Text manipulation in Codex Sinaiticus (Mark's Ending)

The following is a chart tabulating the letter-counts for the columns preceding and following (Scribe A) and those actually of the four pages on the 'cancel-sheet' (two folios, 4 page surfaces) which were inserted by 'Scribe D'  to replace the original pages:

Click to Enlarge, Backbutton to return.

Note that: The letter-counts are based on the text as transcribed on the Codex Sinaiticus Homepage (British Museum), but we have subtracted all marginal notes (which were added independently later), and also the peculiar symbols of Scribe D (the ">" arrow used to fill rows to make the page look fuller.).   Still counted as a character was the "dots" indicating verse and paragraph divisions, because these are assumed to have been copied from the original master-copy, and are not the invention of either scribe.

The behavior of Scribe D (the person who substituted the pages) is fairly straightforward.    For the first three columns, he simply copies the pages he has removed, rightly noting that the previous Scribe A (author of the rest of the NT minus a couple of pages) has already done some of the work of compressing the text for the purpose of creating nearly a whole blank column of space.

He has already carefully calculated that the best course for handling Luke's verses will be to squeeze them into SIX of the last columns on the 'cancel-sheet' (folio 77-5 recto last 2 columns, and the 4 columns of the other side, 77-5 verso).  This will gain him about a half a column.

Now he calculates the remaining approximately SIX columns of Mark, and to leave the last column nearly empty (but avoiding leaving a whole column), he crams about 60 letters into the last column of folio 77-4 recto.  He has actually been too extreme here, and now must write the final FIVE columns of Mark rather sparsely to make sure he spills over a little into the last blank column before Luke.

A glance at the column heights explains why he chose this route.  He did not have enough material to stretch Luke to fill an entire SEVENTH column, and even if he had done this by some real spreading,  He would have only had just enough space for Mark, filling all the previous columns, without the Long Ending (Mark 16:9-20).   This may indeed have been exactly what Scribe A had originally done.

However, Scribe D, likely the Overseer and Corrector of the scriptorium, knew this would be highly unsatisfactory, and wished to at least leave a nearly blank column to tip off future users and enable them to copy in the Ending if they chose to.   Scribe D then, was aware of the Long Ending.  Even though he did not allow sufficient space for it, due to his desire to keep the look of the manuscript professional and standardize the book-seams, he did feel compelled to make sure that at least the option was available and the problem highlighted.

In this strategy, Scribe D mimicks almost exactly the behavior of the scribe of Codex Vaticanus, who also leaves just enough space to allow for Mark's Ending, but in that case he also left a rather awkward completely blank column between Mark and Luke.  It seems clear that our Scribe D had seen that solution before and wanted to improve it by extending Mark into the blank column, and give the appearance of a normal manuscript.

We only need try to reconstruct what the original Scribe A had done on the pages now lost, which ought to explain how the 'need' for a substitution arose.

Ancestors of Aleph/B: Mark Revisited

We posted our analysis of the omissions common to Aleph/B in Mark previously here:

Ancestors of Mark

We continue that discussion with a look at the actual formats which generated them, and where we tentatively place them in the textual history with the following diagram:

Click to Enlarge and backbutton to return

If we compare our findings of probable column widths in terms of characters per line with actual manuscripts from the same and earlier periods, we find a startling, but not really surprising agreement.

We may list a few of the more important manuscripts and their column-widths below:

P66  -  (225 A.D.)   23-27 characters per line, typically  25-26.  (left & right justified)

P75  -  (250 A.D.)  28-31 cpl, with the average being 27-30 cpl. (left & right justified)

W     -  (5th cent.)  28-30 cpl, with the average being 28-29 cpl.  (left & right justified)

A      -  (4th cent.) 19-27 cpl, but rigorously 20 cpl exactly for most lines. (left & right justified)

D      -  (4th-5th.) 12-38 cpl, with lines fluctuating wildly, but on average 25-28 for a full line.

B      -  (4th cent.) 16-19 cpl,   consistently 16-17 for majority of lines. (left & right justified)

א       -    (Aleph, 4th) 10-15 cpl with a fairly tight average of 16-17 cpl.   (left & right justified)

Codex A:   Remarkably, just as we found an unusual number of omissions in Mark having an exactly 20 chars in length, so we find in Codex Alexandrinus not only a 20 cpl line length, but one rigidly enforced throughout the codex.  This surviving manuscript indicates that this was indeed a standard or popular format in some quarters, with its own rules.  Even with Aleph/B we don't see the letter count so strictly enforced as with Codex A.  Yet this two-column format most likely predates the 3 and 4 column formats that became popular in the late 4th century.  The general trend was to go narrower for readability and error reduction.  Codex A is probably a copy of an early 4th century text, made in the same format as the master.

P66 , P75, and W show the earlier, more primitive format of a single wide column.  This very style led to a large number of errors, and columns were deliberately narrowed in the early 3rd century to combat this problem.   It is easy to imagine an early Caesarean or Egyptian copy of Mark with such a wide format, perhaps slightly narrower, such as we suspect caused the errors grouped with the 'First Ancestor' of Aleph/B.

Codex B's three column format must have been popular in its time, but of course most manuscripts from the late 3rd and early 4th century have been lost or destroyed.   This was possibly a reduction from the format of Sinaiticus,  with a reversion to 3 columns and 15-17 cpl.  Since the largest number of omissions are multiples of this length, we must suspect a similar format in an ancestor of the nearest ancestor to Aleph/B.   These errors, while obviously prior to the divergence of these two texts, cannot then be much older than the time the final common ancestor was produced.

Codex Aleph is important, for it shows that many of the Great Bibles of the 4th century must have had this very narrow, 4-column format.    Such a manuscript again was likely to have been in the copying-line between the early copies of Mark and the last common ancestor of Aleph/B.    This again dates the errors of this length to this intermediary period when 3 and 4 column narrow formats were popular. 

---

It should be noted that  Aleph and B are hardly the first manuscripts to have been written in 3 or 4 columns, and their own near-ancestors must have had similar formats.  While the texts chosen by the copyists appear to have been old, and eclectically selected, the gospels seem to have come from the same 'nearest common ancestor', that fathered the majority of omissions shared by Aleph and B.

We can see from a comparison of the grouped readings with similar line-lengths, that each matches very closely with known manuscript-formats circulating from the late 2nd to the late 3rd century.   This suggests quite rightly that while many of the omissions in Aleph/B are perhaps more ancient than Aleph and B themselves (even by 100 years or more), they can hardly be older than the formats that generated them.

There is nothing requiring the proposed order of formats in our chart above to be fixed, other than basic probability, and the known development of same.  It is possible that one or two formats in the chain could have been reversed in their historical sequence.   The outcome would have been the same, only the dates of the omissions would be shifted and interchanged by a few decades. 

At the same time, without any good reason to change the sequence, we might as well stick with the most probable order.   When we come to merge our reconstructions for each Gospel in the Alexandrian stream of transmission,   It will be prudent to keep this order unless further evidence suggests otherwise.

mr.scrivener

The Many Ancestors of Aleph/B

With all the many discussions of a 'common ancestor' for Aleph/B, one can be pardoned for beginning to believe there actually was such a thing, in the manner described.

Here we present for the first time a realistic picture of the Ancestor(s) of Aleph and B:

Click to Enlarge this picture

The first problem that confronts us is that Aleph and B are whole Bibles, each having a rather full set of NT books. 

These NT books of course originated from diverse sources.   More importantly for the purposes of Textual Criticism, each of these books had its own textual history and a long period of separate circulation as independent works.

Then came an equally long period, in which the Gospels were collected into a single manuscript, probably post-Marcion (c. 200 A.D.).   Likewise, the letters of Paul circulated first separately, then were early gathered into a collection, which, with a few later additions, also circulated separately for a long time.   Also, it seems likely that Luke/Acts were originally circulated together, then split up, with Luke being gathered with the other Gospels, and Acts being collected with the Catholic Epistles (James, Peter, John etc.).   Revelation seems to have had the longest independent circulation.

Finally, these collections were gathered into a single volume (a New Testament) for the use of churches, sometime in the early 4th century, and that seems to be where Sinaiticus (Aleph) and Vaticanus (B) originate.

Hort himself acknowledged that a single "common ancestor" for Aleph/B was actually an artificial construct, meant more as a convenience in discussing common readings, than as an actual single ancestor:

 "Whatever be the mutual relation of א and B, each of them separately (א in the apocalypse excepted), is found...to be in fact essentially a text of the 2nd century or early 3rd century.  This fact, which is independent of the coincidences of אB, so that it would remain true of  א if B were unknown, and of B if א were unknown, suggest the most natural explanation of their coincidences.   They are due...to the extreme antiquity of the common original from which ...the two MSS have diverged, the date of which cannot be later than the early...2nd century.   ...So high an antiquity would of course be impossible if it were necessary to suppose that the 'common original' was a single archetypal MS comprising all the books as they now stand.    But, ...there is reason to suspect that the great MSS of the Christian empire were directly or indirectly transcribed from smaller exemplars which contained only portions of the NT; so that the general term 'common original', which we have used for the sake of simplicity, must in strictness be understood to denote the several common originals of the different books or groups of books."

(Hort, Introduction, ¶ 301., p. 222-223). 

---

Many Ancestors:

We know then for a fact that Aleph and B did not have a single ancient ancestor, but several. 

It is acknowledged for instance that the book of Revelation in Sinaiticus is unlike any other version.

 "The Apocalypse in codex Sinaiticus is a striking example of a fourth-century text that differs substantially from modern critical editions. It exhibits dozens of differences at key points, reflecting the concerns, interests, and idiosyncrasies of its earliest copyists and readers. Taken as a whole, Sinaiticus’s text of Revelation may constitute one of our earliest Christian commentaries on the book, disclosing its fourth-century milieu and anticipating the later concerns of Oecumenius and Andrew of Caesarea." (Juan Hernandez Jr., 2009)

   In John 1:1-8:38  (the first 8 chapters) Codex Sinaiticus differs from Vaticanus and all other Alexandrian manuscripts. It is in closer agreement with Codex Bezae (D) in supporting the Western Text-type.  Vaticanus on the other hand, offers a version of the Western Text-type for the Pauline Epistles.

This means that for various books, or collections of books, the two manuscripts have completely different ancestors. (at least three ancestors are involved in just the three sections mentioned).

---

Common Readings:

On the other hand, the agreements, especially the 'agreements in error', that is unique agreements between Aleph/B against almost all other authorities, does indeed seem to point to common sources. 

This is no surprise in fact, when we know that the Four Gospels began to be collected together and produced as a single book,  beginning sometime in the early 3rd century.   It must have been after this time, that most of the common readings (especially the errors, unusual minority readings) between Aleph and B arose. 

There are however several possible sources for the agreements between Aleph and B, and they must all be carefully considered in any given case:

(a)   readings common to the original copies.  These would be expected to be rather common readings also found elsewhere,  particularly in independent lines of transmission, such as other text-types.

(b)   readings (errors) which arose by coincidence because of an unfortunate feature built into the text, such as multiple instances of homoeoteleuton from varying layouts.

(c)   readings consciously modified by editors attempting to correct or improve the text, including spelling conventions and grammar corrections,  and especially omissions of parts of speech and phrases perceived to be ambiguous, wordy or clumsy.

(d)   readings deliberately chosen among already available variants, popular or preferred readings involving theological or doctrinal issues, or historical perceptions.

(e)    readings due to a mistake or selection made in a previous common ancestor.   Many readings may well be the results of accumulated errors in a common transmission-line.

---

Striking Differences:

There are also however, many differences between Aleph and B, even within books that are assumed to have a common ancestor (e.g. the Gospels);  Hoskier for instance enumerated some 3036 differences between Sinaiticus and Vaticanus. 

Those who promote the text of Aleph/B draw attention to these differences and claim that this proves the 'ancestor' of Aleph/B goes back "several generations", perhaps as many as "ten" (this however is probably exaggerated).  The purpose of this notice is to drive back the proposed date of this 'ancestor' to the 2nd century:

"..their common ancestor must have been copied several generations before. ... Aleph and B are distant cousins from long after their common ancestor, which itself must go back several generations. Indeed, when they agree, their common reading usually is from the early 2nd century." (Wallace, Komoszewski, Sawyer, Reinventing Jesus (Kregel, 2006) p. 78. 

 However, like the common readings, these differences may also have multiple causes, some of the most obvious being:

(a)   mistakes by the scribes themselves who made Aleph and B.  Each of these (two or more) scribes must have contributed many errors to both manuscripts.  Singular readings are the prime suspects.

(b)    differences in policy or preference regarding grammatical constructions, or how to handle suspicious readings in the master-copy.

(c)   the work of correctors, who may not have left clear traces of their work.  In particular, B has been overwritten and re-inked throughout, and Sinaiticus suffers from the hands of at least 10 correctors.

(d)   inherited differences from previous generations of copying separating either manuscript from the common ancestor or master-copy.

(e)   differing usage of cross-references or secondary sources used for textual comparison.  Both manuscirpts show signs of editorial activity of various kinds in the compilation of their texts.  

From all this, and each possible cause involves real variants, it is clear that neither similarities nor differences can be easily categorized as to how they arose, or dated as to when.  

---

 Three Important Eras:

Nonetheless, a natural and straightforward procedure presents itself, namely considering the three stages of transmission,

(1)  First Era:  Separate Works  -  Individual and independent circulation of the NT books, and the differences arising from varying circumstances between them in their transmission.

(2)   Second Era:   Early Collections  -  Circulation of the groups of books as collections,  and the features they would then share, and the errors arising from this format.

(3)  Third Era:  Whole NT Volumes  -  Circulation of the groups as complete copies of the NT, and the features and errors arising in this period.

It should be clear that the common readings shared between Aleph and B must belong to the first two periods of transmission, while the readings in which they differ will likely have arisen in the third period of transmission. 

We strongly doubt that "ten generations" of errors can be identified among the differences between these two manuscripts,  and prefer to identify the date of the nearest 'common ancestor' independently, based upon what is known about when these collections were popularized.   It is reasonable then, to date the nearest common ancestor of Aleph/B to about the beginning  or middle of the 3rd century, when it is known that such collections of the Gospels circulated, as P75 demonstrates.  

Consequently, we assign common readings between Aleph/B to this time and earlier, when they would have come together to form a common text spanning all four gospels.

Differences between Aleph and B we will primarily assign to the period from about 250 A.D. (c. P75) to 320 A.D.  (the inside date of manufacture for Aleph/B).

We have already analyzed the pre-history for Mark and Matthew (see our previous posts on this), and shown it to be quite different for each book, as expected.  On the other hand, we can also combine our findings, and propose that there was a common ancestor in the line after the gospels were gathered into a collection, in which the columns were about 15 characters per line.   This narrow width suggests that the copy that generated this large group of shared errors spanning Mark and Matthew was rather recent, since narrow columns were a practice that came much later than the time of P75.    The final result of this analysis will naturally date this large group of errors to this late period, namely the late 3rd century.

We will post a list shortly.

mr.scrivener

Breaking out the Ancestors of Aleph/B: Mark

Taking the convenient list of Markan omissions from the last post in hand, one immediately notices that the five largest chunks of text uncannily break into lines with 15 characters per line.  This is a strong correlation, and not likely to be a mere coincidence.  It is further corroborated by other shorter omissions, also breaking down into lines of 14-15 characters.

Five Big Ones:  Click to Enlarge

In the following fuller list, we mark those with yellow to indicate that they may belong to an ancestor with a different line-width, since the text can be divided up several ways.   Note that the three single lines of 15 chars per line:

Ancestor 3: click to enlarge

For reasons that will shortly become apparent, we call this group of omissions "Ancestor 3".   This long list of 31 lines is a strong indication of line-width, since almost half of these Variation Units can be identified as homoeoteleuton, regardless of what width we select from the likely options.  These omissions were certainly accidental, and so were caused by layout, not content.  The lack of theological, historical, or linguistic reasons for the omissions/additions also corroborates an accidental cause.

This column width is relatively narrow, and so these omissions arose later in the copy-stream, when narrower columns were selected for the very purpose of cutting down on accidental copy-errors, i.e, 3rd century.   It reflects the style of Codex Sinaiticus in terms of page format.

Earlier MSS had wider columns, before it was realised that errors could be minimised by better choices of format.    Other groups of equal-length omissions are also found among the Aleph/B readings.

The following group, which we have labeled "Ancestor 1", conveniently groups line-lengths of 20 - 23 together.  Although the 2nd half-chart may belong to a different (later) exemplar, the wider format is variable enough that all the omissions could have been generated from one master-copy.

Ancestor 1: Click to Enlarge

Again the yellow coding indicates other possible placements for individual variants.   Relatively few omissions would be from the older layers, especially obvious homoeoteleuton, because some would be caught in subsequent generational copying and proofreading.  The omissions without homoeoteleuton features would be most difficult for correctors to find or properly identify.  We should not then be surprised that the early candidates would be fewer in number.

Candidates for Ancestor 2 are more certain, and hence more attractive, by the nature of the numbers, which offer less alternative line-lengths to choose from:

The Green color-coding here indicates more confidence both in the identification as homoeoteleuton, and the proposed line-length in the master-copy.   Although the list is shorter, the existance of the exemplar is more certain.

Ancestor 3 has been glanced at already above.  Ancestor 4 is below:

Ancestor 4:  Click to Enlarge

As the line-length decreases, the alternative options also decrease, making the identification of column-width more certain.  One or two of these omissions could have been generated in an older, wider-column exemplar, much earlier in the transmission-stream, but more manuscripts would have been manufactured in the narrower format, as copying spread rapidly in the later centuries.  The high line-count here (13 lines) lends probability to an exemplar of this width.

With Ancestor 5, we again have an encouraging number of lines of exact lengths.  With 10 and 20 letters per line, other column widths are unlikely and a poor fit, which increases the confidence level in the choice of 10. 

Ancestor 5:  Click to Enlarge

With 19 lines of about 10 characters, this seems like a solid choice for an exemplar somewhere in the copying generations, that would have generated these accidental omissions.  Alternately, 20 characters is an option for many of these readings, but would also suggest a vulnerable ancestor, such as Ancestor 1 (or 1b). 

Manuscripts with columns narrower than 10 letters would be very rare, if they existed at all.   Ancestor 6, then, really represents the 'left-overs'.  Smaller omissions that are more likely to have been generated along the same line, or in the process of memorizing a long clause.   They could have taken place at any time or generation of copy, and must be left as 'floating omissions'.

These proposed "Ancestors" must not be taken to imply that Aleph/B (the common ancestor of the two 4th century Uncials) is only a '7th generation copy' of the original Mark.  This is far from likely.

It should be remembered that good copyists and their generations will be almost undetectable, since they will have reproduced their master-copy accurately.   Only "bad" copies will be identifiable in a sorting process like the above.   

What the following shows, in comparison with what we found regarding Matthew, is that Mark has suffered at least "7 bad generations" of copying, whereas Matthew has only seemed to have suffered about 3 or 4. 

This is precisely what we should expect, since Mark is a much older document, and would have been subjected to more generations of corruption than Matthew.   What we see then, is two different lines of bad copying converging in the 'ancestor' of Aleph/B.  Originally, these gospels would have been copied as separate books, but were then brought together in the 2nd century.

The Omissions have been examined elsewhere (see our Homoioteleuton Blogsite for reconstructed layouts and textual evidences for these Variation Units.).

Our concern at the moment here is to show that:

a) Hort was wrong about the 'purity' of Aleph/B.

b) The textual history of the Aleph/B line of transmission can be reconstructed very adequately, with a proper consideration of all the available evidence and technique.

c) The Omissions of Aleph/B should never be adopted into the Christian NT text, except in certain extremely unusual circumstances. 

This is an isolated, unreliable line of transmission, and all but useless for textual correction.

mr.scrivener