Sun May 5 10:02:48 1996

From majordom  Sun May  5 10:02:48 1996
Return-Path: 
Received: by scholar.cc.emory.edu (5.0/SMI-SVR4)
	id AA05567; Sun, 5 May 1996 10:02:48 +0500
Message-Id: <9605051358.AA50781@mail.uni-muenster.de>
Mime-Version: 1.0
Content-Type: text/plain
Content-Transfer-Encoding: 7bit
Date: Sun, 05 May 96 16:02:00 +0100
From: schmiul@uni-muenster.de(Ulrich Schmid)
Subject: Re: Textual Theory
To: tc-list@scholar.cc.emory.edu
X-Mailer: SPRY Mail Version: 04.00.06.17
Content-Length: 15621
Sender: owner-tc-list@scholar.cc.emory.edu
Precedence: bulk
Reply-To: tc-list@scholar.cc.emory.edu

On Fri, 3 May 1996, Robert B. Waltz wrote:

[quoting Schmid:]
>>What other scholars state in this regard is simply of no value,
>>_unless_ they prove the results from the data to be inferior when
>>compared to other samples. I do not say that this might be
>>impossible, but I need hard facts in order to know the limits of
>>_Text und Textwert_ and to use it in an appropriate way. I 
>>concede that the handling of _Text und Textwert_ is not very 
>>easy, but from my own experiences it offers more significant 
>>results than most critics expect.

>I will concede the last statement. For example, I think it
>*proves* that there are, in Paul, *no* "Western" witnesses except
>D F G 629 and the Latin versions...

I would like to withdraw the term "prove" in my obove mentioned 
statement, and have to have it replaced by "give examples that". 

>I will also admit that it is hard for me to prove the superiority
>of my sample. *I* have found many more relationships using my
>sample, but that *may* be because the data in _Text und Textwert_
>is so hard to use (its practical utility would probably be doubled
>if all results for the same manuscript were placed on the same 
>page).

I do not want anybody to "prove" something in this peculiar 
instance, for what we are dealing with are only reative samples 
compared to the overall evidence. I was just focussing on the point 
that there are some critics of _Text und Textwert_ who seem to 
simply rule down the validity of that tool, without giving evidence 
that they fully evaluated the data. There are two things I would 
like to say on that. First, I again concede that it is not easy to 
work with _Text und Textwert_. But, I suspect, doubeling its 
practical utility by, for example, placing all the results for the 
same manuscript on one page, would have doubled the space needed. 
Second, I recall my own first steps in getting acquainted with 
_Text und Textwert_. It took me a good deal of time, and I was 
driven down at least one blind alley. But, since it is a 
_published_ tool of referrence, I would like to know about others 
experiences in testing it in order to assess its validity and 
limits, when compared to other samples.  

>But conceding all that, I remain very troubled by the data in
>_Text und Textwert_. I think the readings are biased (too many
>variants which have the "Western" text on one side and everything
>else on the other). I think there are too few variants where the
>Byzantine text divides. This makes it very difficult to establish
>the families within the Byzantine text...

Well, the readings are choosen out of the pool of known variant 
readings which the different textual theories (safe, I think, the 
pro-Byzantine position) up to know considered as significant. I 
admit that it is only a small sample out of that pool, and one 
could argue on some of the readings. On the other hand, the goal of 
_Text und Textwert_ was _not_ to achieve an exhausted survey into 
the depths of the Byzantine text, but to single out the MSS which 
deviate more or less significantly from the Majority text. The 
predominant intention was to get a referrence tool that may be of 
help when choosing MSS for a new _editio critica maior_. And 
therefore, the deviation from the Majority text was singled out as 
a point of referrence in order to detect those MSS which had to 
become part of a critical apparatus.

>But most of all, I think there are too few variants. Ten entries
>in Colossians? That means each variant is responsible for 10%
>of a manuscript's score...

>Which also brings us to the question of statistical significance.
>This is worth remembering: A five percent difference in a sample
>with twenty readings is *not* significant. That represents a
>difference in one reading. If we had instead eighty or a hundred
>readings, and still maintained our five percent difference, *that*
>is significant.

I totally agree with that. The fewer items a sample has, the lesser 
is the significance drawn from percentage alone. To partly 
overcome such a situation it is absolutely necessary to weigh the 
evidence. Therefore, the full collation data are given, so one 
could check every single reading within the conspectus of all 
readings of a variation unit. The _quality_ of a reading 
(transposition, addition, omission, substitution, etc.), especially 
when rare readings or subsingulars are involved, has a high value 
(may be with the exception of the Byzantine text, for in its later 
strata almost no such readings seem to occur). I admit that not all 
of the units of variation are pertinent to such an assessment. But, 
I think, for the sake of balance one also has to include some 
transpositions or minor additions, because they too could turn 
out to be more significant than usually expected.

>Turn to the Catholics for a minute. We've both alluded to the
>"harklean" family (which I follow Duplacy in calling family 1611).
>This family in fact has a large number of subfamilies -- 614+2412,
>1505+2495, 630+1799+206+429, etc. Perhaps *you* could find them
>by examining _Text und Textwert_. I couldn't. 

The Catholics are a good example, because my friend Klaus Wachtel 
-who wrote his dissertation on _Der Byzantinische Text der 
Katholischen Briefe_- and I got our training on _Text und Textwert_
from the General epistles. Before we started to work with T&T we 
had been focussing on Claremont-profile, because we could not 
imagine that T&T would be particularly helpful when assessing the 
Byz. text; T&T simply was not designed for that. Then we started 
with MSS groupings from T&T and related the findings to von Soden's 
groupings. Klaus carefully examined all the MSS groups he singled 
out on the bases of T&T (for details consult his book published in: 
Arbeiten zur Neutestamentlichen Textforschung 25, deGruyter: 1995). 
I should add that Klaus not only counted the evidence, but also 
gives a 215 pp. textual commentary on the 98 _Teststellen_. Here I 
may give only few results based on counting.

a.) With respect to the Harklensis Vorlage (1505-1611-2138-2495) Klaus not only 
could determine the intra-group relationships, but he also could classify the 
related MSS  (206-429-522-614-630- 1292-2200-2412) in their relationship to one 
another and to the primary witnesses. For example the differences are mainly due 
to the supplementation of the related MSS in the last three epistles (2John 
-Jude) from at least two different Vorlagen. And lastly another partly close MS 
to 2138 could be identified: In James and 1Peter 1890 is very close to 2138, 
_and_ 1890 turned out to be the missing 1522.

b.) The first interesting fact with respect to the Majority text was that NO 
single MS completely AGREES with the Majority of witnesses at all the given 98 
Teststellen (orthographicals or itacims are not counted as deviations from the 
Majority text!). So by consequence, a Majority text theory built solely on mere 
counting would result in an eclectic text! NO single Majority-text manuscript in 
the proper sense of the term can be found within the bulk of 522 MSS for the 
General Epistles. (This is not contra Maurice Robinson. I am well aware that he 
holds other theoretical standards. But may be there are some who want to go an 
easier way.)

c.)To assess the Byz. MSS Klaus singled out _Leitlesarten_ -i.e. readings that 
depart from the Majority text, but are attested 
by at least 100 Byz. MSS. These readings are very interesting 
because the could help finding out even slighter changes within the 
Majority text. 
It turned out that von Soden's Kr group was identified by means 
of a distinct pattern at the _Leitlesarten_, shurely more 
complete and more precise than von Soden's. In the General epistles 
23 MSS belong to Kr -i.e. they give at 98 units of variation 
identical readings and share the distinct pattern. (Complete 
collation samples of two of these MSS in Jamens and 1Peter prove 
this striking feature even from the corrections in these MSS.) 
Another 15 MSS depart from that characteristic pattern only once, 
and another 10 only twice.   

>I repeat, I think _Text und Textwert_ is a *tremendous* 
>accomplishment.
>But it is *not* a sufficient manuscript classification tool.
>Probability theory makes this claim, and -- while I can't prove
>my statement otherwise -- unless firm COUNTERevidence can be
>offered, I will trust the mathematics.

There is no need to distrust the mathematics at this point. From 
T&T of the General epistles it is possible to classify most of the 
150 MSS that depart more than 10% from the Majority-text by means 
of grouping them. From the remaining 372 MSS roughly 18% could be 
settled by referrence to Kr. This, I would say, is indeed a 
tremendous accomplishment, especially when compared to von Soden. But, I never 
claimed T&T to be a sufficient manuscript classification tool. Exactly therefore 
I would like to compare the results drawn therefrom with the results drawn from 
other samples. 

>Now *that* surprises me. Not that 6 has a number of near-singular
>agreements with 1739 (since that is a statistic I keep), but
>that you could work through the data enough to find it.

>But that's just because of the way _Text und Textwert_ is 
>presented.
>Would you have discovered that fact if you had not been looking?

Well, after a few bad experiences as a novice, I developped a 
certain manner to work with T&T. Since the information are very 
condensed in only few statistics, one normally has to look at least 
twice at the _Hauptliste_. This is, as I am used to it, no major 
problem, if the parameters are clear. 

>The question -- which I do not claim to have the answer to -- is,
>"Was there ever a purer text behind family 2138?" This family, in
>the general epistles, is clearly distinct from the Byzantine text.

I do not have an answer to this question either. But it is 
interesting that Thomas of Harkel, who used in 616 AD a text very 
similar to family 2138 as Vorlage, chose one Greek MS for Acta and 
the General epistles, two for the Pauline Epistles and three for 
the Gospels. As far as I can see up to now, Thomas' text in the 
Pauline epistles tends to be slightly more Byzantine (as do 1505-1611-2495), but 
could by no means be called Byzantine. So may be this points to different dates 
the Byz. text of the various parts of NT developped. 

Another striking feature in the Harklensis Corpus Paulinum seems to be that 
Thomas' second MS he used did not display a more Byzantine text compared to the 
other Vorlage (1505-1611-2495), but a text that displays readings found in MSS 
like aleph B D F G 1739. If we take into consideration the goal for and the time 
in which the Harklensis was made, the whole thing becomes disturbing for a 
pro-Byzantine position. The Harklensis was made in 616 AD and designed to be an 
exact (!) translation from Greek MSS for the Syriac monophysite church to 
improve on and supersede the (Peshitta and) Philoxeniana in order to get an 
adequate referrence tool for theological debates with the main Greek speaking 
churches at that time. For this task good MSS had been chosen, presumably ones 
that reflected not only mere local texts, unless one presumes this local texts 
to be recognized (and spread) from Egypt to Syria and probably Byzanz too. But 
if so, then the label "local text" has to be applied to the Byz. text too, and 
explains virtually nothing with respect to an overall theory of textual 
transmission from a pro-Byz. viewpoint.   

Concerning the ending of Romans, Bob further wrote:

>I think that, if the fourteen chapter form circulated widely, it
>would explain the inclusion of the doxology after chapter 14. But >apparently 
the fourteen chapter form was not common.

I think it depends on how early the cross-contamination took place and what 
other evidence can be added. 
a.) The fourteen chapter form evidently was part of the pre-Marcionite 10 letter 
edition. There is evidence to this edition even outside Marcionite circles not 
only with respect to the shorter form of Romans, but also with respect to the 
distinct letter sequenzes.
b.) There are various other readings in Gal, 1.Cor, Eph, and 1.Thess, that the 
Marcionite text testifies to, which are found  exclusively either in the Old 
Latin ("Western"), or Old Syriac, or Alexandrian or Byzantine textual tradition. 
To one of these readings the "Western" text testifies by giving a conflate 
reading (Marcionite text on the one side and Alex. + Byz. texts on the other 
side), to another of these readings the Byz. text testifies by giving a conflate 
reading (Marcionite text + 3 minuscules on the one side and most of the Alex. 
witnesses on the other side).
c.) To conclude therefrom: The Marcionite text seems to antedate all our known 
text-types at least in their actual state of appearance. Since it seems to be 
more plausible to assume a pre-Marcionite text to have spread these readings, 
than the Marcionite text himself as a distinct _Marcionite_ entity, there is 
more evidence to a pre-Marcionite 10 letter edition that affected the rest of 
tradition.
d.) I agree with presupposing that the 14 chapter form of Romans once must have 
played a more prominent role, but I think, this has to be assessed within the 
conspectus of the pre-Marcionite 10 letter edition. This edition indeed must 
have played a prominent role, since some of its readings (including the doxology 
of Romans) are now to be found spread all over the textual tradition. But on the 
other hand, since Marcion himself charged others on the base of the 10 letter 
edition of having corrupted Pauls letters, and was charged the same by others 
mainly on the base of the 14 (13) letter edition, it is easily explanable why 
the 10 letter edition was singled out and no longer transmitted outside 
Marcionite circles.
Some remainders of that edition, nevertheless, survived within NT textual 
transmission (MSS, versions, patristic testimonies).  

[quoting Schmid Bob further wrote:]

>>Second, the patristic evidence only points to two larger, major
>> editions of the Corpus Paulinum: The 10 letter edition of
>>Marcion (which is to my mind of pre-Marcionite origin) and the >>canonical >14 
(13) letter edition, with the somehow ambigous role >>of Hebrews.

>I would agree, with one caveat: p46 apparently omitted the >Pastorals. So there 
may have been *three* editions.

I am really not shure on P46 omitting the Pastorals. The crucial point is that 
P46 apparently lacks 14 pages at the end, but the missing 2Thess and Pastorals 
would have required (cf. Kenyon) 23 pages. Kenyon assumes that P46 originally 
contained 2Thess but lacked the rest leaving the last 10 pages blank. Others 
think Philemon was originally contained too leaving the last 8 1/2 pages or so 
blank. Others opted for a shortened form of the Pastorals included. 
There are three points which to my mind have to be taken into consideration:
a.) For a single-quire codex as P46, due to its nature, one has to calculate the 
space required for the text _before_ starting to write.
b.) The stichoi numbers in P46 are somehow misleading (Eph, though longer than 
Gal, was counted as having 59 stichoi less than Gal).
c.) From the middle of the codex the numbers of letters per line and the numbers 
of lines per page tend to increase.

To conclude therefrom: P46 gives at least some evidence that there must have 
been initial miscalculation taken place. So, this MS cannot be used as firm 
evidence with respect to a third and different edition of Pauls letters.

Ulrich Schmid. Muenster






Back