easily deal with over 100 of these in a shift, since the whole process had then been reduced to a matter of looking things up in a number of different tables and the only arithmetical operations involved were addition and subtraction. I should like to emphasize this point because if these almost perfect beings cryptographers have a fault it lies in disinclination to simplify their mysteries in this way.
31. We found simplicity in scoring most important. When scoring alphabets [See Para. 16] we originally used the actual scores so that a score would be, say, 16xx/183 + 2/98 + 11/115 + ..... . Our first change was to have all scores evaluated in decibans by the Big Room (desire for simple units was not the only reason for this change - the new system was also very much more accurate), working correct to 1/10th of a deciban, so that this score would be, say, 12.3 - 4.8 + 7.4 + . Finally in order to be able to work in whole numbers we changed the unit to "half decibans" and worked to the nearest unit so that the scores would be 25 - 10 + 15 + ... . These changes did not merely save us trouble, they enabled us to succeed where we would otherwise have failed, by making it possible to try so many more alternatives; it was one of the cases where a change of degree produces a change of kind.
32. We found it very profitable to have charts and tables for all calculations that had to be made at all often. For example, there was a process we constantly had to perform, which amounted to working out 3/19ths of the number of letters in a comparison - i.e. 3/19ths of some number between 30 and 400; it paid hands down to have a table of values of 3/19ths of 5, 10, 15, 20 ......... 400 made up once and for all, rather than to spend half a minute making the calculation.
33. A small research section was responsible for keeping all statistics up to date. It is very desirable in a busy section engaged on operational work to keep at least one good cryptographer (not necessarily the same one continually) away from current work; even if