There was, of course, a procedure for indicating the Geheimschreiber key settings. One setting was the inner one, that is the selective connection[5] used for the connection between the ten code wheels and the previously mentioned inversion and transposition relays. The inner setting was in force for three to nine days, starting at 9 o'clock on the first day.
When a message was to be transmitted, the code wheels had first to be set to a given position. These settings had to be the same for both sender and receiver. The transmitting station would select a setting for five consecutive wheels. This setting was transmitted to the receiving station with a three-character so-called ``QEP indicator''. The five remaining wheels were set to a predetermined key value that was valid for all messages during a 24-hour period. This setting was called ``QEK''. The daily key list indicated which wheel would be ``QEP-wheel'' and which setting the ``QEK-wheels'' should have.
It should be pointed out that the number 3 = letter shift, 4 = figure shift and 5 = space in this teleprinter alphabet. This has great importance for the following deliberations.[6]
When the transmission of a cipher message was about to start, the transmitting station would present itself with ``Hier MBZ'' (MBZ here) and would then ask if the message could proceed, ``QRV''. If this was the case, the receiving station answered with ``KK''. The transmitting station then sent ``QEP'' succeeded by five two-digit numbers (e.g. 12 25 18 47 52). Both operators then adjusted the wheels in their respective machines, partly the ``QEK-numbers'' after the key list for the current day, and partly the ``QEP-numbers''. When the transmitting station was ready it would transmit ``UMUM'' (umschalten -- switch over) and when the receiving station was ready it would answer ``VEVE'' (verstanden -- understood). Then they switched over to cipher mode and the transmission of the text itself started. The ciphertexts were consequently always preceded by ``UMUM'' and were therefore easy to retrieve in the large number of signals.
It is possible that Beurling had knowledge of the Siemens & Halske patent, but this is not certain. Borelius recounts that when Beurling visited FRA on 15 November 1976, he reacted strangely to questions about the first break. He evidently did not like the questions to be put. He nevertheless said that he made use of ``threes'' and ``fives'' in the texts.
Telecommunication technical problems were a great help during the breaking. The telegraph lines were long, sometimes bad, and therefore often exposed to interference, which could distort a transmitted character. The readability was nevertheless not disturbed except when the character changed to a ``4'' (= figure shift), because then all succeeding text became an unintelligible sequence of numbers and punctuation characters. If the distortion affected only the receiving station, the transmitting station did not notice anything, and continued the transmission. To reduce the problem, the operators normally used to write ``35'' (= letter shift, space) instead of only ``5'' (= space) between the words. Thus the consequence of a false ``4'' would be restored at the next space between the words.
Beurling discovered that when the plaintext of ``3'' and ``5'' had one pulse the same and four different, this had also to be the case in the enciphered state. For a guessed ``3'' there consequently existed only five possible ``5'' or vice versa. It was therefore relatively easy to establish spaces between words, which would have facilitated further work. It was probably this which Beurling talked about when he alluded to ``threes'' and ``fives''. Hence a guessed ``3'' gave only five possibilities for Q and V in ``QRV'', which asked if the message could proceed. In this way further work was greatly eased.
It also seemed natural to suppose that a part of the encryption process consisted of a transposition of the five-pulse-characters pulse positions. A number of comparisons could give the transposition arrangement.
Beurling tried in this case to trace back the cipher character to its appearance before the transposition. Then he made his next observation. The change from one character in the plaintext to a cipher character after the ``overlaying'' consisted of a change in polarity for some of the pulses of the plaintext character, and for all characters in a column it is always the same character that changes.
We can assume that Beurling now introduced his observations on his ``work
sheets'' (``avvecklings-papper'').
In five rows under the examined text he
placed in every column a dot for pulses with inverted polarity and a circle
for those that did not change. When the emerging five-dot combination
resembled a teleprinter character, it was called an ``overlaying'' character.
Under this character was written the permutation order, that is the
so-called transposition. Gradually it was detected that the pattern of
circles and dots in the five rows of the ``overlaying'' repeated after a
number of characters. Beurling then supposed that the pattern was produced
by pin-wheels like those in the Swedish cipher machine, invented by Boris
Hagelin.
Subsequently he continued the work with the transposition. It turned out that if for example ``pulse 2'' ended up on ``place 3'' then the fourth circuit connection had to be open, otherwise it was not possible. If this circuit connection was controlled by a wheel with even distribution of active and inactive pins, the pulse would end up on ``place 3'' in half of the events, in the rest it would fall on some other place. A converse conclusion should have been possible, using an inverted argument. The complete circuit and hence the details of the remaining wheels were obtained through hypothesis tests. By these means Beurling would have got the break that revealed the Geheimschreiber secret, as Borelius' reconstruction shows.