Wed Aug 21 22:27:26 1996

From owner-tc-list  Wed Aug 21 22:27:26 1996
Return-Path: 
Received: by scholar.cc.emory.edu (SMI-8.6/SMI-SVR4)
	id WAA28621; Wed, 21 Aug 1996 22:26:37 -0400
Date: Wed, 21 Aug 1996 22:23:07 -0400 (EDT)
From: Maurice Robinson 
To: tc-list@scholar.cc.emory.edu
Subject: Re: Colwell's 70%
In-Reply-To: <00002958.sm@dts.edu>
Message-ID: 
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-tc-list@scholar.cc.emory.edu
Precedence: bulk
Reply-To: tc-list@scholar.cc.emory.edu
content-length: 2464



On Wed, 21 Aug 1996, Mark OBrien wrote:

> I see this figure quoted a lot in comments regarding quantitative analysis, but
> I'm curious to know why 70% is "the magic number".  

The simplest answer is because Ernest Colwell of Chicago said so.  I
personally think 72.3% is somewhat more accurate *;-)

Seriously, there is _no_ real cutoff point, and all such suggestions are
of necessity arbitrary.  However, it also is obvious that as support for a
texttype approaches the 50% mark, the relative strength of that texttype
is seriously called into question; conversely, as the support for a
texttype approaches 100%, the relative strength is increased.  Merely
splitting the difference would suggest 75% as a reasonable boundary for
establishing a texttype relationship among MSS.  Colwell chose to slightly
broaden the definition by going to 70%; Hodges and Farstad chose to narrow
the definition by going to 80%.  I am personally quite comfortable with
70% (making small allowances on either side of that figure), since such a
"majority" relationship among MSS should be adequately sufficient to
establish a working texttype identification of relationship.

> Surely this percentage is
> going to vary (to some extent) according to the number of manuscripts being
> compared?  

The percentages are based upon quantitative analysis of pair of MSS -- one
compared directly to another in significant places of variation.  No
matter how many MSS might be compared in such a situation, they still will
be counted in pairs which cover the entire spectrum, and only those MS
pairs with high interrelationships will be considered primary members of a
texttype.  Once a texttype relationship has been determined, a reading
would be considered as "belonging" to the archetype of the texttype if it
occurred in 70% (or whatever other figure is settled upon) of the MSS which
were previously assigned to that texttype.  This is why in the
"loipoi"/"polloi" reading with a 53%/47% split, _neither_ reading is in
any way texttype-specific to the Byzantine Textform, although one could
correctly claim that the "polloi" reading _is_ specific to the Alexandrian
texttype.   


_________________________________________________________________________
Maurice A. Robinson, Ph.D.           Assoc. Prof./Greek and New Testament
Southeastern Baptist Theological Seminary     Wake Forest, North Carolina
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Back