General Report on Tunny


23J Page 95


23J     FLOGGING THE EVIDENCE


(a) Impracticability of an exact formula.

No simple formula for weighing the evidence of a run can be exact. Evidence is derivable not only from the sheer magnitude of the bulges but also from having bulges on the right letters, or on a consistent group of letters (e.g. on all language letters); in other words it is unjust to take the message as a fair sample of itself, and necessary to include other messages.

This section gives only a brief crude account with a minimum of mathematics. For a more refined treatment see 23X.


(b) A primitive formula.

If it is assumed that the message is a fair sample of itself, the factor in favour of a setting due to a sigma-age is proportional to eδ2/2 [of 24X.]

It follows that the odds in favour of a settings with a sigma-age S, is about

where Σ refers to rival settings and 2ω allows for random settings.

The decibanage is 10 log10 of this [21(g)]. This is, essentially, the formula used to construct the table for decibanning wheel-setting runs.


(c) Combining the evidence of several runs.

If several runs are used to set the same wheel or wheels.

If there is no competition the decibanage is

.

< previous

next >