This article is rated C-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects:

Cryptography: Computer science High‑importance This article is within the scope of WikiProject Cryptography, a collaborative effort to improve the coverage of Cryptography on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.CryptographyWikipedia:WikiProject CryptographyTemplate:WikiProject CryptographyCryptography High This article has been rated as High-importance on the importance scale. This article is supported by WikiProject Computer science (assessed as High-importance). Military history: Technology / North America / United States / World War II Start‑class This article is within the scope of the Military history WikiProject. If you would like to participate, please visit the project page, where you can join the project and see a list of open tasks. To use this banner, please see the full instructions.Military historyWikipedia:WikiProject Military historyTemplate:WikiProject Military historymilitary history Start This article has been rated as Start-class on the project's quality scale. This article has been checked against the following criteria for B-class status: Referencing and citation: criterion not met Coverage and accuracy: criterion not met Structure: criterion met Grammar and style: criterion met Supporting materials: criterion met Associated task forces: /  Military science, technology, and theory task force North American military history task force United States military history task force World War II task force

There is a company doing business as Sigaba (Secure Data in Motion, dba Sigaba). Maybe this company (www.sigaba.com) ought to get a mention?

Is the company particularly notable? — Matt Crypto 18:00, 19 Dec 2004 (UTC)

It ought to. It is a very significant player in the secure email market, but I'm biased. DoomBringer 06:44, 18 Jun 2005 (UTC)

The disambiguation link in the header seems a good approach. — Matt Crypto 22:20, 22 Jun 2005 (UTC)

Thanks. I'd add more to the Sigaba corp page, but I'm affialated. DoomBringer 02:56, 12 July 2005 (UTC)[reply]

Additional rotors do provide more security even when a message's length isn't great enough to cause them to advance. A static rotor (such as the fourth one found on some Enigmas) causes a 26-fold increase in the number of possible decryptions.

I've removed that comment. — Matt Crypto 20:09, 30 November 2005 (UTC)[reply]

The SIGABA was the US Army designation, and CSP 888 was the Navy designation. A "neutral" version appears to be ECM Mark II, and I propose we rename this article and use that term throughout. — Matt Crypto 20:09, 30 November 2005 (UTC)[reply]

This is Michael Lee. Since I am no longer a student, the CS Department will soon take my old site http://www.cs.ucsb.edu/~kirbysdl/ offline. I have set up a permanent archival mirror at http://ucsb.curby.net/ that includes copies of the thesis, and I have changed the article's links to point to the new site. To show that I'm not some random wacko, I've included a link on the original cs.ucsb.edu site pointing to ucsb.curby.net. The new site is entirely noncommercial/nonprofit.--Kirbysdl 09:39, 8 August 2006 (UTC)[reply]

Cheers Michael! (And congratulations on graduating.) — Matt Crypto 16:11, 8 August 2006 (UTC)[reply]

1) History should contain when the machine was taken out of service. Which was likely quite a long life (hard to destroy, e.g., US embassies and the USS Pueblo and other ships). 2) production numbers would give a sense of scale. 3) While cyphertext was not cracked, machines were briefly stolen during WWII (Polmar's book), and others later compromised. 4) I seem to recall that John Walker was caught with a simple device to deduce rotor internal wiring (photo in Keith Melton's book). The page isn't quite the same quality and depth as the Enigma page (or as a Fort Historian calls it "Enigma fetishism"). 143.232.210.150 (talk) 17:26, 9 April 2010 (UTC)[reply]

My grandfather was in the Army Signal Corps and recalls a story where two guys were to transport a SIGABA device in a truck, with instructions to never let it out of their sight. They parked it outside of a restaurant/bar, and when they came out, the truck was gone. It became such a big deal that Gen. Eisenhower came into the town (somewhere in France) to address the situation. He offered a week leave in Paris to anyone who recovered the truck with SIGABA in it. Eventually someone found the truck abandoned by a river, someone had stolen the truck without knowing or caring about the SIGABA in the back.

We've been looking for some kind of record of this story but nothing has turned up. Does anyone here know anything about this? Harksaw (talk) 12:27, 28 September 2015 (UTC)[reply]

Maybe this: https://www.nsa.gov/public_info/_files/cryptologic_spectrum/Colmar_Incident.pdf The circumstances of the theft are different but it does mention that Eisenhower was personally involved. The week leave in Paris offer is not mentioned, but is easy to believe, given the importance of recovery. --agr (talk) 01:51, 29 September 2015 (UTC)[reply]

This is precisely the story! Thank you very much, my grandfather was very happy to receive this. 68.77.171.20 (talk) 17:03, 29 September 2015 (UTC)[reply]

Perhaps you could write up or video your grandfather's story, preferably with his name and unit, have him sign it and send it as a letter to a Cryptographic journal or maybe the NSA's museum. The one week leave inducement is an interesting bit of cryptographic history that should not be lost and his first person account would be valuable.--agr (talk) 21:11, 29 September 2015 (UTC)[reply]

From the control rotor bank, the model, which receives four voltage signals at the input of this rotor bank, has only nine connections to the input of the index rotor bank (I was recently foolhardy enough to correct this in the article). Input 0 receives no signal and has no connection (n.c.), unlike in the schematic of the machine shown in Timothy J Mucklow's NSA publication "The Sigaba/ECM II Cipher machine "A beautiful idea"", where this contact is connected to ground, which sooner or later inevitably leads to a short circuit. The post-war Sigaba model CSP-2900 actually had ten such connections, but the control rotor bank in it also had six voltage inputs. The outputs were combined into ten groups of 1 to 4 output contacts (2*4, 2*3, 3*2, 3*1), whereby three of the twenty-six output contacts had no connection this time (n.c.) : P, Q and R. In the previous model, all twenty-six outputs of the control rotor bank were grouped into nine groups of 1 to 6 contacts.(1*6, 1*5, 1*4, 2*3, 1*2, 3*1). In the Post-War Sigaba CSP 2900 model, the 2nd and 4th rotors of the Cipher Rotor Bank always rotated backwards. In contrast to the previous model, not only 30 different rotor movement variations could therefore take place, but 31 : with number 31., ALL rotors rotated one step further, just not in the same direction. Permissiveactionlink (talk) 18:21, 30 June 2025 (UTC)[reply]

In the NSA publication "The Sigaba/ECM II cipher machine, "A beautiful idea"", Timothy J. Mucklow also calculates the key space of the Sigaba machine. However, he makes a mistake: the index rotor bank offers 5! * 2^5 * 10^5 = 384,000,000 setting alternatives (permutation of the five index rotors, normal or reversed installation position of the index rotors, basic position of the index rotors), but the index rotor bank only replaces a plug board that can be varied to C(10,2)*C(8,2)*C(6,2)*C(4,2)*C(2,2) = 113,400 types. (C(x,y) = x!/((x - y)! * y!)) The effective key space of the index rotor bank is therefore approx. 12 bits less than the corresponding key space of the Sigaba index-rotor-bank calculated by Mucklow: not 107 bits for the complete machine, but approx. 96 bits. The index rotor bank was suitable for confusing unauthorised persons, but the high number of possible settings has nothing to do with the actual key space : Numerous different settings of the index rotors lead crytographically to identical wiring. The different labelling of the index rotors with 10 to 19, 20 to 29,....., 50 to 59 was very disadvantageous. Uniform labelling with 0 to 9 or 01 to 10 would have been better, whereby the rotors would then have been labelled I, II, III, IV and V on the side. This would have made them indistinguishable for unauthorised persons. Permissiveactionlink (talk) 18:41, 30 June 2025 (UTC)[reply]

In the version of the Sigaba that was used during the war ("CSP-889"), the four voltage inputs of the control bank were at positions F, G, H and I, while NINE lines ran from the control rotor bank output to the input of the index bank. Input 0 of the index bank received no signal (n.c., non-connected, and NOT a connection to ground, as erroneously shown in the circuit diagram of the NSA publication on the Sigaba --> can lead to a short circuit ! ); input 1 was connected to output B of the control bank; input 2 to output C; input 3 to the electrically connected outputs D, E; input 4 to F, G, H; input 5 to I, J, K, input 6 to L, M, N, O; input 7 to P, Q, R, S, T; input 8 to U, V, W, X, Y, Z and input 9 to output A of the control bank. Permissiveactionlink (talk) 15:22, 16 July 2025 (UTC)[reply]

This modified Sigaba model ("CSP-2900") had four significant changes:

1) The cipher rotor bank rotors in positions 2 and 4 generally run in the opposite direction to the other three, i.e. backwards.

2) Unlike the previous model, ALL FIVE cipher rotors can advance one step at the same time, but NOT in the same direction of rotation! However, AT LEAST ONE ROTOR continues to rotate per ciphering step, either forwards or in the opposite direction, depending on the rotor being controlled. In contrast to the previous version, this time there are 31 cipher rotor advancement alternatives per cipher step. However, it is still impossible for not a single rotor to switch on during a ciphering step.

3) The power supply at the input of the control bank now takes place at SIX contacts: D, E, F, G, H, I.

4) There are now actually TEN connections between control bank and index bank, with four to one output contact(s) forming a group and (in the case of more than one contact) being conductively connected. This time, however, three output contacts remain (previously none), which show no electrical continuation, i.e. end blind (n.c.): P, Q and R.

Input 0 of the index bank is connected to outputs U, V of the control bank; input 1 to output B; input 2 to output C; input 3 to outputs D, E; input 4 to outputs F, G, H; input 5 to outputs I, J, K; input 6 to outputs L, M, N, O; input 7 to outputs S, T; input 8 to outputs W, X, Y, Z and input 9 of the index bank to output A of the control bank. Permissiveactionlink (talk) 15:31, 16 July 2025 (UTC)[reply]

If you refer to the post-war model, then you could install a plug board that supplies the control rotor bank with voltage in 26*25*24*23*22*21 = 165,765,600 (key space : + 27.3 bit) different ways at six inputs. An additional non-involutory plug board at the input/output of the cipher rotor bank would add another 26! = 403,291,461,126,605,635,584,000,000 alternatives (key space : + 88.4 bits). And if a plug board is placed in front of the ten connections of the index rotor bank at the output of the control rotor bank, with 26 output sockets, 23 connectors, and 10 groups of four sockets each, which are electrically connected and each have a connection to the inputs of the index rotor bank, then (with groups of the same size as in the CSP-2900 : 4, 4, 3, 3, 2, 2, 2, 1, 1, 1) at least (26!/3!)*(10!/(((2!)^2) * 3!)) different alternatives would be possible. Such a plug board increases the key space of the machine by a factor of 10,162,944,820,390,462,016,716,800,000,000, which corresponds to a further + 103 bit key space! You can rewire these plug boards at longer intervals. This does not have to be done every day. It can also be done by officers in special, lockable additional boxes connected with the machine. Also in this case, this increases the machine's key space from less than 100 to over 300 bits.

In the control rotor bank, only the centre three rotors rotate, while the two outer rotors ("stators") are only adjustable. This means that the period length for the sequence of the various cipher rotor steps is only 26^3 = 17,576. The centre rotor of the control rotor bank rotates with each cipher step and switches the second rotor from the left with only one notch once every 26 cipher steps, which in turn switches the fourth rotor from the left with only one notch once every 676 ciper steps. With two additional stepper motors, it would also have been possible to move the two outer rotors of this bank: the rotors could then have been given 11, 15 or 17 notches each (always at the same letter positions). As a result, the five rotors of the control rotor bank would have continued to move much more irregularly, with an advance period length of 26^5 = 11,881,376. However, no precise statements can be made about the length of the key period of the SIGABA cipher rotor bank, without the internal wiring of the cipher- and control rotors and those of the index rotors anyway. These are still kept secret today. Permissiveactionlink (talk) 12:51, 17 July 2025 (UTC)[reply]