## The Special Fish Report## Albert W. Small (December 1944)## Page 29 |
Tony Sale's Codes and Ciphers |

| previous page | next page | index page |

- CX/MSS

- TOP SECRET Special Fish Report Page 29

legal wheel, must sum to more than 50 decibans:; 3. Individual characters

of the wheels must be at least 20 decibans each, or if lower in spots

they must be so marked and Testery notified for its information in solving

delta psi prime; 4. The 32-letter count of the final Delta-D text

must be satisfactory.

- There is also available a brief significance test to use

- Let R = total number of positions actually looked at in the run.

W = wheel length.

X = sum of the moduli resulting from the run (= Sum|xi|.)

The expected value of X at a wrong setting can be shown to

can be shown to equal R(l - 2/Pi) or approximately .36R. So the

standard deviation is approximately equal to .6(R)^1/2, and two standard

deviations above expected random = .8(R*W)^1/2 + 1.2(R)^1/2. This is graphed for

usable values of R and W. X of an run must exceed this if the run

is to be called significant.

- With these miscellaneous facts gathered together, we are prepared

- Pages 30-34 inclusive are the actual work-sheets used during the

- The first line of the Delta-X1 worksheet (Page 30) and the first line of

rectangle made on Garbo. Message involved, SB 3037 part l.(Original rectangle

not available.) Other messages used during the run: SB 3037 pt 2; SG 917 Roll I

pts 2 and 3.

A "part" is usually 10,168 characters (8x41x31); the last part is what is left.

**
(Note by Tony Sale: pages 30-54 are at present omitted (some are A3 size) until a
way can be found to put these on the web at a reasonable Kbyte size)**

| previous page | next page | index page |