COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Thesis
C00004 ENDMK
C⊗;
Thesis
I have started to read your thesis, and I like it very much so far.
On page 13, however, there is a problem that would make any statistician
unhappy with the thesis.  You simply can't avoid discussing the a priori
probability of the screw being present and errors of the first and
second kind, i.e. the probability that a method will find a screw when it
is absent and the probability that it will miss a screw when it is
present.  Given these and the a priori probability that the screw will
not be picked up, one can get estimates of the a posteriori probability
of a screw.

	In discussing combining several methods, you can't escape
considering their statistical independence.