When we did this the examination of the fits automatically meant that we had to examine the indicator groups of associated messages. We then found the singular phenomenon that if the 1st letters of the indicators, agreed so did the 5th that if the 2nd agreed so did the 6th; even now we were so hypnotized by the idea of bigram tables that we looked at the indicators for another day until someone from Hut 6 (who had had previous experience of a similar indicating system) having a casual look round pointed out that this was just a "throw-on" system, the old original type used in1931 - 7 on Dolphin.
27. This meant that it was quite simple to make a break on 100 messages a day and possible (though very laborious) on 10 or 15. We could certainly have been reading it for some months, possibly since 1941. To find a major key using such a bad indicating system, one that had been abandoned five years earlier on Dolphin, was astounding - nevertheless we should not have been caught napping.