Mon Jun 17 08:05:03 1996
From owner-tc-list Mon Jun 17 08:05:03 1996
Return-Path:
Received: by scholar.cc.emory.edu (SMI-8.6/SMI-SVR4)
id IAA23256; Mon, 17 Jun 1996 08:04:00 -0400
Date: Mon, 17 Jun 96 07:00 CDT
X-Sender: waltzmn@popmail.skypoint.com
Message-Id:
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
To: tc-list@scholar.cc.emory.edu
From: waltzmn@skypoint.com (Robert B. Waltz)
Subject: Re: On "examples" and "proof"...
Sender: owner-tc-list@scholar.cc.emory.edu
Precedence: bulk
Reply-To: tc-list@scholar.cc.emory.edu
content-length: 10368
On Sun, 16 Jun 1996, wlp1@psu.edu (William L. Petersen) wrote:
>I agree with Waltz's comment totally. If he reads the portion of my post
>which he himself quotes, he will see why.
All right, my apologies. I guess we *are* on the same side in this
particular sub-debate.
I think it's just that those of us with scientific training get
subjected to so much hearsay evidence that we get hypersensitive.
At least, I do.
So let me say that I am *not* in disagreement with Peterson. But I
would clear up a few points. (This is probably going to be incredibly
boring, but I don't want to be accused of fuzzy language.)
>He draws this quote from my post:
>
>>>To my way of thinking, an example "proves" a lot more than rhetoric.
>
>He then takes me to task for failing to note that "examples can be used to
>prove anything." That is not what I am saying, and Waltz must know from
>statistics (which he praises in his post) that an absolute claim needs only
>ONE exception to be invalidated.
True -- but a good scientist will never make an absolute claim. Oh, we
are often guilty of loose language (I've done it on this list), but a
true scientist would not say "33 is an Alexandrian text," but rather
"33 agrees with the Alexandrian witnesses in 86 of 110 instances observed
instances" (or some such).
A scientific approach is to make a hypothesis about data. This
hypothesis *must* include a mathematical measure, and have a way of
measuring "goodness of fit." Observations are then made, and the
experimenter examines the data. If the data comes close enough to
the model, then the model is, tentatively, accepted.
How much correlation we need, of course, depends on the subject under
examination. In the physics lab, we considered ourselves to be doing
a lousy job if our correlation was less than 97% (a correlation
coefficient cannot exceed 100%). In medicine, a 50% correlation
is miraculous.
Determining correlation for textual relationships might be harder;
we have a finite data set. I'll have to think about that.
Most scientists believe in something. I myself believe in the
Law of Conservation of Mass/Energy (more strongly than I believe
in God, I fear). But a true scientist expresses all things in
probabilities. It's just that Conservation of Mass/Energy is
99.9% sure....
It's a good thing, too. If you ever have the opportunity to study
quantum mechanics (not something I recommend :-), you'll learn
that *anything* can happen. It's just that some things are *highly*
unlikely. (Almost as unlikely as anyone caring about this post.)
So scientists will not let *one* example destroy a theory. It
just reduces the correlation coefficient....
>In the case at issue here, I have seen lots of RHETORIC (which is, after
>all, what I contrast "example" with; I did not contrast it with statistics,
>as Waltz's post would lead one to think...) being used to make all sorts of
>claims. We all recall examples from the list's short history: the fathers
>cite inaccurately; the fathers' MS tradition has been corrupted in the
>history of their transmission; the Byzantine (or Alexandrian) text MUST
>have been the earliest text; discussion of the history of the text (in most
>posts on TC) commences in the fourth century, ignoring the second and third.
>
>The question in my original post was: Why are we talking this way when the
>empirical evidence from the second, third and fourth cents. either
>invalidates or "brackets" (i.e., severely restricts and limits) the claims
>being made?
>
>I have capitalized my reservations about the fathers/versions/AND NT MSS
>throughout my posts. I make no absolue claims as to what I "know" we will
>find--for I do not. All I have said--and I have presented the TEXTUAL
>evidence for it--is that we CANNOT presume that the fathers cite carelessly
>ALL THE TIME, or that their MS tradition has been corrupted ALL THE TIME,
>for the EVIDENCE EMPIRICALLY DOES NOT SUPPORT THESE DOGMATIC, ABSOLUTE
>ASSERTIONS. And only ONE example is necessary to falsify these claims. If
>those statements are modified to reflect the empirical facts (e.g., that
>Justin's text has many "Western" elements in it, and few [if any?]
>Byzantine elements; see M. Mees' conclusions on a book-by-book basis for
>Clement of Al's text; etc.), then I have no objection.
Amen to that. Not having had time to go over *all* available evidence
(who of us has?), I was content to accept others' hasty statements.
It would appear that I (among others) was wrong.
>Like anyone, I can produce examples where the patristic citations DO appear
>"loose" and where the MS tradition probably HAS been corrupted. But--as I
>have--I can ALSO produce instances where it does NOT appear the citation is
>loose or the transmission has been corrupted. All I ask is that the same,
>empirical textual evidence be presented to back up the claims, statements,
>and assertions often being made solely on the basis of rhetoric.
Something I have always tried to do -- though it has often been difficult
in the case of the Fathers. In *my* library, at least, the evidence is
inadequate.
And I will admit to a possibly irrational prejudice in favour of
manuscripts over patristic citations. Probably because I can examime
the manuscript for any reading I want....
>Indeed, that is the "scientific" way, isn't it? Rhetoric and pet
>theories--whether ecclesiastically espoused, or propounded on the basis of
>personal convictions--just don't make it. Evidence does. And when absolute
>claims are made, only ONE example is needed to falsify it.
See comments above.
>I am more than happy to use statistics, which I often do in my work. By the
>way, how many Byzantine or Alexandrian readings are there in Justin, or
>Tertullian, or the Didache??? Statistics anyone??? Indeed, let's use
>examples and statistics, for then, at least, we will all be speaking the
>same language, examining the same evidence.
I can't offer much; my databases are built around continuous text
manuscripts. The only place where I've examined the fathers is in Paul,
and even there I've only done two: Ambrosiaster and Origen. And the
evidence for Origen is rather iffy, since some comes from Greek fragments
and some from the Latin. And, of course, Origen sometimes cited a passage
more than once and in more than one form. (In a lot of instances, I just
had to say that I don't *know* Origen's reading.) And -- before people
ask -- yes, I *did* examine readings section by section; there was no
statistically significant difference. So I'm giving statistics only
for Paul as a whole, because a larger sample has more statistical
significance (and because that saves me having to retype the whole
thing, as well as re-extract the data from my database).
Note that p46 and C have significant lacune, and so their number of
near-singular agreements will be slightly smaller. Note also that
33 is Byzantine in Romans, and so their *is* a shift in its relations
between Romans and other books which I have not noted here. Similarly
with 1175 in Romans and Thessalonians and 330 in the latter part of
Hebrews. These affect the overall statistics only slightly.
In any case, here are the relations of two fathers to assorted major
manuscripts. To save time, agreements (except near-singular agreements)
are expressed as percentages.
Origen (144 readings checked; of these, 64 were Byzantine)
% agreement in
Manuscript % agreement non-Byzantines near-singulars
p46 59% 81% 7
Aleph 58% 87% 7
A 53% 78% 3
B 62% 89% 10
C 53% 79% 2
D 44% 67% 5
G 43% 68% 6
L 47% 67%* 0
33 54% 79% 1
81 56% 74% 1
330 42% 69%** 0
1175 47% 80% 0
1505 46% 55%** 0
1739 73% 90% 7
2127 50% 63% 0
* sample involves only 6 readings
** sample involves fewer than 14 readings (10% of overall sample)
It is evident that Origen has a text closely related to family 1739
(whatever *that* is).
Ambrosiaster (378 readings checked; of these, 130 were Byzantine)
% agreement in
Manuscript % agreement non-Byzantines near-singulars
p46 43% 64% 25
Aleph 37% 68% 5
A 36% 63% 5
B 40% 70% 24
C 41% 66% 2
D 65% 87% 62
G 66% 84% 59
L 37% 75%* 1
33 40% 68% 11
81 38% 62% 3
330 37% 64% 4
1175 42% 78% 2
1505 41% 70% 3
1739 42% 70% 6
2127 40% 62% 0
b 74% 93% 67
WW Vulgate 54% 82% 27
* sample involves only 8 readings
It will be observed that Ambrosiaster is strongly "Western" --
but also has a tendency toward p46/B (this is *not* just the
so-called "Western" tendency of p46 and B; a reading supported
by p46 B D G is almost certainly not going to show up as a
near-singular).
I realize these are relatively late fathers, dating from the
period *after* text-types had become established. But I offer
the data for whatever it may be worth.
Bob Waltz
waltzmn@skypoint.com
Back