RE: tc-list Collations
- On Sun, 13 Sep 1998, "Perry L. Stepp" <plstepp@...> wrote:
>Both your and Robert W's description of how the computer can do theThere are several responses to this.
>cross-comparisons sound fascinating. But won't itacisms, movable nu's,
>lacunae, visually obscure readings, etc., present problems that will be
>difficult to adequately address without a scholar going through the
>comparisons letter by letter? Handling such phenomena turned out to be a
>significant part of my project.
First, this is a function of the original collations, not the collation
base or the database.
Second, if such a question *does* arise, then of course one must refer
to the original manuscript. But these problems are only a small minority
of variants (or, rather, they are problems only in a small minority
of genetically significant variants).
Third, the original context of this discussion was quantitative
analysis -- determining the genetic relationship between manuscripts.
For this, lacunae do not matter. Visually obscure readings are to
be read if they can, and treated as lacunae if not.
The argument could be made that itacism, movable nu, etc. might
have genetic significance. But the keyword is *might*. This depends
on the scribe and the manuscript. Some scribes copy by letter; these
scribes will usually transcribe spelling and all. Others copy
by syllable; these will be slightly less reliable in this aspect.
Still others copy word by word. With these scribes, itacisms, etc.
mean very little; they will frequently spell the word the way *they*
spell the word. This might even extend to varia such as first or
second aorist ending.
On this subject I suggest that you read the relevant section (pp. 33-43)
in W. L. Richards's _The Classification of the Greek Manuscripts
of the Johannine Epistles_. This is a work marred in all sorts of
places by conceptual defects, but it did do one useful thing: It
initially included every variant in its manuscript base, then divided
them into "questionable" and "non-questionable" variants. The
"questionables" were things like nu-movable. Richards found that
the inclusion of such variants merely decreased the relationship of
closely related manuscripts, without making a significant difference
for less-closely-related manuscripts. I interpret this to mean that
his "questionable" variants are more a product of the scribe than the
Thus, the policy of the IGNTP is, IMHO, correct: Collate against a
standard, then compile the collations into one master file and
work from that. Collating every manuscript against every other is
just a way to increase your work exponentially -- with an actual
*loss* of useful data.
Of course, this does require that you have the right computer program
on the job. :-)
"The one thing we learn from history --
is that no one ever learns from history."