Once a start had been made with the month's Offizier it was not normally necessary to make any further use of the bombe. The first break gave Stecker for 2 days and 1 setting and the settings for other messages on those two days could now be discovered in the same way as settings for Barnacle, etc. If there then occurred on another pair of days with a different Stecker a message of which the setting was known it was possible to set to work to discover the Stecker - and subsequently, of course, the settings of any further messages --- and so ad infinitum.
To obtain the Stecker when only the setting was known it was necessary to do a Dottery, a job much coveted in early days but whose fascination disappeared as the method was perfected. The first step was to decode the message at the correct set-up with all letters self-steckered: the result naturally was jibberish. However, of the 26 letters assumed in the decode to be self-steckered, 6 were in fact self-steckered and, when one of those was pressed, something happened which was at least comparable to what happened in the machine used by the German operator, i.e., the letter being genuinely self-steckered the current went into the machine in the right way and, if the correct bulb did not itself light up, it must at least have been the Stecker of the right letter which lit up -- see the paragraph on Stecker in Chapter 1. The next step was to record on a square the pairs of letters obtained by writing the decode under the cypher text, the left hand side of the square being the cypher and the top the decode. Of the 26 rows in the square, 20 are of no value but the 6 which belong to the 6 self-steckered letters should have dots distributed over them in accordance with language distribution as distinct from random distribution. Also, all 6 rows should show the same squares to be popular and the same to be unpopular - the most popular square will be the Stecker of E. It was not normally possible to spot all the self-stecker but perhaps 4 would be certain and also the Stecker of E. A second similar process, known as a Yoxhallismus after its inventor, assumes every letter of the message to be an E (the Stecker of which is now known so it can be encoded right through the message) and records the Stecker which each assumption implies. As E is very common, the assumption will often be correct and hence the squares representing true Stecker will fill up much more rapidly than the others. By these methods part of the Stecker was obtained and then began attempts to decode the message with partial Stecker; whenever a word could be guessed which was only partially decoded, further Stecker could be added and the complete Stecker was soon obtained.
On a message of more than 200 letters a dottery was normally successfully completed without much difficulty. In early days much labour was spent on dotteries of insufficient length and little profit resulted from it. At one time or another most Offizier keys were read, and Oyster and Limpet were broken regularly and fairly completely. If traffic was high, we could