even quadruple banking of certain frequencies was that, for cryptographic reasons which I will explain later, it was absolutely essential to have a 100% accurate text of any message that might he used for crib purposes. In our experience it was most unwise to believe in the accuracy of single text even if it was transmitted with Q.S.A. 5, and so we asked automatically for double banking on frequencies likely to be used for cribbing. Especially on the Mediterranean keys and Bonito interception was extremely unreliable and quadruple banking on crib frequencies was often necessary. On Bonito the assistance of R.S.S., who did not normally work on Naval traffic, was enlisted and for some Bounce traffic originating from weak transmitters in Northern Italy, we relied on R.A.F. station in Italy.
Requests for double banking for cryptographic reasons became extremely numerous by the end of the war. In 1941, with only one key, special cover on a small group of frequencies was sufficient but by late 1944 we were normally breaking some 9 or 10 keys for each of which special cover on some frequencies was required. On the whole the policy was to ask for double banking if it was at all likely that it might be useful and accept the fact that a certain amount of unnecessary work was being done by intercept stations. There were at any rate sufficient hazards involved in breaking keys and it was felt that it would be foolhardy policy to take risks in the matter of interception when these could be obviated by double banking.
Most traffic arrived from the intercept stations by teleprinter, being duplicated by carbon. Retransmissions, dupes as we celled them, were also teleprinted in full after 1942; before this German preamble and differing groups only of dupes had been teleprinted but we found that we were unable to rely on the intercept stations to notice all differences.
Shortage of teleprinters was a perennial problem as the traffic constantly increased while those responsible for teleprinters persisted in believing that it would decrease. The effect of this shortage was that traffic got delayed at Scarborough for considerable lengths of time before teleprinting and was not in fact cleared until there was a lull in the traffic. It was usually true that there were sufficient teleprinters to cope over a period of 24 hours with the traffic sent in that period but they were quite insufficient for the rush hours on the evening and early night shift. In the spring of 1944 the teleprinter situation was reviewed and considerably improved in anticipation of the Second Front and possible heavy increases of traffic. Experiments were carried out with a priority teleprinting system for certain frequencies but the list tended to be so large (having to cover cryptographic and intelligence needs)