# How secure is this modified Bazeries Cylinder?

This device is a modification of the Bazeries cylinder, it has two stacks of 5 rotors which each bear two alphabets plus numbers 0-9, one in correct order and one in random order ( 36 characters in each row ). Plain text in 5 letter groups is entered by setting the rotors as in Fig 3. The shaded character on the bottom rotor is read and the whole rotor stack is then turned ( keeping the individual rotors in alignment ) so that this character appears on the top rotor. The enciphered group is then read from top to bottom as in Fig. 4. This group is then re-enciphered by the same procedure using the lower stack of 5 rotors, producing a doubly-encrypted code message.( Figs. 5 & 6 )

Obviously this would be devilishly difficult to decipher without knowing the rotor stack sequence and the random sequences of the alphabets on each rotor. Frequency analysis, Kasiski tests and Index of Coincidence tests do not reveal any useful information. My question is, could this coding system be broken by other than brute force methods ( which would be very tedious and expensive ) ?

To maintain security, the rotor stack sequence would be changed in accordance with a code book list ( or other protocol ) after every 50 groups.

A specimen of code produced with this device follows :

4FRLO 7W0W9 PL84F 7BITX 9VNBN H0U81 DJ3PU 1R27H 9SCN3 FJTBG 9CIQK D1ZZU
IJB4H YML6O DHTZU IFJBH UABHD L37I8 RYEXE QQ9SU BJX4G YG0KT 7GJAT X0XOW
R9DT1 MEH3A EU8HO 9J70M DO0E4 STXQS 9VS5N 9ICZ9 KN4TR 55AMC ELLGW AQ2IH
AEOC8 E0EMP FM36V OGBEP 8MLO5 0GY00 KSRQN MMJ88 DQYB8 4GU0X 1ZPVU FQTQD
4LVKW 4W4G0


How secure is this modified Bazeries Cylinder?

That depends on what you expect us to compare it to. I doubt you’re looking for an answer along the lines of “it’s more secure than Scytale, but less secure than AES”. Therefore, I will try to narrow it down a bit by saying that it is safe to assume it can not – in any way – provide the same security levels that modern cryptographic algorithms are able to provide. Some attacks are obvious; while others are merely a question of investing some time and effort to practically cryptanalyze the nitbits.

My question is, could this coding system be broken by other than brute force methods ( which would be very tedious and expensive ) ?

Yes, it can. Mentioning only two of many examples: it definitely won’t hold for a second when applying a or a – which would both count as “valid breaks”.

Don’t get me wrong. It’s a nice design and it surely baffled some adversaries back in the days. Yet, in a (let’s just call it) “near-quantum” world, the cryptographic attractiveness of this is nothing more than dust in the wind… a distant reminder of days where things like computers did not exist and security was easier to achieve and uphold.

• And it only gets really interesting when we start considering a chosen-ciphertext-attack, even some modern schemes fail against such attacks (like CTR or CBC). Aug 29, 2015 at 17:02
• @SEJPM $^{+1}$ Indeed, that’s another nice example of an attack other than brute force methods. Aug 29, 2015 at 17:12
• Perhaps I should have specified that the attacker has no knowledge of any plaintext examples, which would exclude such attacks. ( I understand that these are not classed as brute force methods. ) Aug 30, 2015 at 0:09
• Since no-one has offered a decryption of the code message, I would have to assume it is reasonably secure. Or else no-one can be bothered trying ? Sep 12, 2015 at 1:17
• @Kymyst No… the fact no one posted a decryption doesn’t mean it’s secure. It just means no one bothered. The reasons are simple: (1) question answered, (2) decoding snippets like that is off-topic. Quoting the related close-reason text: “Requests for analyzing or deciphering a block of data are off-topic here, as the results are rarely useful to anyone else.” That’s about the same as what is noted in our help-center: “Can I get data analysed here? Can I challenge people to decode something? No. Such questions are not helpful.” – that’s why. Sep 12, 2015 at 2:11