The Special Fish Report
Albert W. Small (December 1944)
Codes and Ciphers
TOP SECRET Special Fish Report Page 97
to make each run in two parts, looking at even offsets of one tape
with reference to the other on the first run (by plugging Robinson to
count only where Level 5 of tape 1 (control level) plus level five of
tape 2 (control level) = dot), and looking at odd offsets of one tape
with reference to the other on the second run (by plugging Robinson
to count only where level 5 of tape 1 plus level 5 of tape 2 = cross.)
This does not take twice as long, because the tapes are in effect
slid 2 positions at a time with reference to each other in each run,
and the over all time is the same as if only me one run were made and the
tapes slid one position at a time.
The two tapes are placed on Robinson and the Robinson
set to record all positions wherein simultaneously:
P tape level 1 + DELTA of (Z tape level 1) = dot.
P tape level 2 + DELTA of (Z tape level 2) = dot.
P tape level 3 + DELTA of (Z tape level 3) = dot.
P tape level 4 + DELTA of (Z tape level 4) = dot.
P tape level 5 + Z tape level 5 = dot for even runs
or cross for odd runs.
In other words, Robinson is plugged up to record positions wherein:
Delta/93 (DeltaP2+P5twoback+DeltaZ2) = Delta/124 (DeltaP2+P5twoback+DeltaZ2) =
Delta/279 (DeltaP2+P5twoback+DeltaZ2) = Delta/341 (DeltaP2+P5twoback+DeltaZ2) = dot
and the causal probability for such is b^5 + (1-b)^5 whereas the random
probability is 1/16. Such a run is much stronger than a run on a single
impulse of the special tapes, wherein causal probability is b^2 + (1-b)^2
as against a random of 1/2.
A sample Robinson crib run as described above is shown next page,
and described in the pages to follow.