19. For our early breaking we were very dependent on re-encodements from Werft; there were also weather R.E's from Dolphin but those were difficult to do correctly owing to changes in phraseology, punctuation and so on and therefore were liable to use a lot of bombe time. When Werft changed, as it did every month or so, we had to break a few days on the Dolphin R.E's to enable the Werft cryptographers to get going again by using the Plaice - Werft R.E's themselves. Once they had got under way they could again help us and all would continue smoothly until the next Werft change.
20. Later on, in Summer 1944, a group of operational U-boats was formed in the Baltic which re-encoded Shark messages. A separate key, Narwhal, had by this time (June 1944) been formed for Arctic U-boats and for some months Narwhal, Shark and Plaice were so linked by R.E's that a break in any one gave an immediate break into the other two. [See graph of the re-encodement network].
21. An analysis was made in January of the average bombe time spent per break made on the various keys; this revealed that a quite uneconomic amount of time had been spent on Turtle. This was partly due to the inherent difficulty of Turtle R.E's but more to the fact that we had rather prematurely handed over control of the key to Op. 20 G while their cribbing section was still very inexperienced; the result was an interesting example of how much experience is needed to develop a really sound judgment of the merits of a crib or re-encodement. Following this investigation bombe time spent on Turtle was drastically reduced and we began to take a more active interest in it ourselves. As time went on the standard of the Op. 20 G cribbing steadily rose and all Shark cribbing (except R.E's from other keys) was ultimately controlled by them with excellent results.
22. On February 1st Porpoise became a 4 wheel key and partially abandoned the throw-on system of indication; instead of encyphering the complete message setting twice i.e. encyphering ABCABC, ABCDAB was