?possible missing line?
so the 11-rule would admit ??% of this sample).
(b) If X5 is in question. Assume the following 12 letters are expected
to be above their X5- pairs :-
/,9,M,G,L,P,A,U,5,8,D,P. If in the supposed Delta D fewer than 8 go the
right way, again there is reason for suspicion.
(The corresponding results were :-
5 6 7 8 9 10 11 12
1 3 6 12 12 12 6 4
The 8-rule would admit ?30%? of the sample).
Two cautions should be given
(i) Many of the counts included in the sample which were rejected by
the rules were unquestionably essentially correct on examination of a
songle pair. This rule is far from being sufficient grounds for final
rejection.
(ii). Often the doubt about a setting arises from a good slide; in
this case it can hardly be expected that the rule will discriminate. The
method for this case will be described later.
When the last impulse set has been found satisfactory, look
at the other impulses. Take some dominant letter (say 5) and compare its
score, with the scores of letters differing from it in a single impulse.
For instance, G may seem unduly high and this casts suspicion on X1, if,
however, U is high and I is on the low side, it will probably be per-
missible to accept X1. In case of doubt short runs can be done.
Decibanning of settings when rivals are good slides.
Suppose we have 3 rival settings of X3, - 01, 03 and 05 -
and that there is a good slide of 2 on Delta X3, Suppose also that we are
sure one is correct, then a fairly accurate assessment can be made.
We do all three 32-letter counts; they will be very similar and
they can be used as a sample to guess the approximate score in the right
place of each letter. (Say, take the average of the three scores). If
the estimated number of J's is 143 and of K's is 106, then a theory about
?possible missing line?