final result and decision of policy to adopt is often extremely difficult. One might have a crib with two main forms one considerably better than the other; identification has proved difficult and there are two candidates - one decidedly better than the other but not certain. On the better candidate the better form crashes and one must either take the better form with the worse candidate or the worse form with the better candidate. There is also another crib with one pretty good form and a number of feeble alternatives - useless individually but having quite a fair chance collectively; there is an excellent identification for this crib and the good form goes in without a crash - the intercepted text however is poor quality and may easily contain a corruption. In four hours time an absolutely cast iron crib which never fails is due to come in. Naval Section are pretty anxious to have the solution as soon as possible however and do not want to wait an unnecessary four hours; at the same time Hut 6 are hard pressed and one must not waste bombe time. These are the circumstances in which the difference between a first class and a second class cribster appear.
13. R.E's may be either from one Enigma key into another or from a hand cypher into Enigma. In the latter category we had R.E's from Met. trigram substitutions (W W's and Archery), from R.H.V. and Henno (reserve hand cyphers used when the Enigma machine was out of action and also used regularly by certain small ships), from Werft (hand cypher used by small ships in the Baltic) and from Auka (G.A.F. hand cypher). Identification is done in the same way as identification of straight cribs - by external characteristics. The difficulty of getting the text of the R.E. absolutely correct depends on how much it has been altered before being re-encyphered. Some R.E's were always quite easy because few changes were made and those were of a standard kind - others were almost impossible; very often it would not be possible to do an R.E. completely but some chunk could be found which was most unlikely to have been altered at all and this could be correctly placed by using the non-crashing property of the machine, giving enough material for the bombe.