WI: Enigma machines used a mixiture of Latin, Greek and Cyrillic

If the Greek and Cyrillic characters are only in the encrypted text, there would be no need for such a distinction.
That's not how Engima worked.

Enigma is a symmetric cipher; the input and output require the same character set. To decode an Engima ciphertext, the machine is configured to the same initial setting that produced the Cipher text, and then the ciphertext is entered by the operator on the keyboard. This produces the original plaintext message.

Enigma, like any competent cipher of the era, is polyalphabetic. To make it more complex, one doesn't need to add additional characters to its instruction set, one only needs to add additional rotors or plugboard leads (this is what they did historically).
 

Cook

Banned
The Failure of Enigma was not the mechanic, it was naive beliefe of Nazi And military, that this machine code was undecipherable by the enemy.

That was not actually the case. The major failing was poor operational procedure: the same message was often sent by both high-grade and low-grade encryption, and in some instances not encrypted at all, providing an open door to the cypher. The highest grade code ‘Neptun’ also happened to be used by the most disciplined operators, those of the Kriegsmarines primary surface ships – and this remained unbroken throughout the war.
 
Enigma is a symmetric cipher; the input and output require the same character set. To decode an Engima ciphertext, the machine is configured to the same initial setting that produced the Cipher text, and then the ciphertext is entered by the operator on the keyboard. This produces the original plaintext message.

Actually, the only requirement is that the character set for the ciphertext include the plaintext character set, it can also include characters that are not in the plaintext, but the keyboard must include keys for those extra characters as well as keys for characters in the plaintext.

Enigma, like any competent cipher of the era, is polyalphabetic. To make it more complex, one doesn't need to add additional characters to its instruction set, one only needs to add additional rotors or plugboard leads (this is what they did historically).

But adding additional characters would surely help. Plain English text has only 26 letters, and plain German text also has ß (German ss), and the three umlauts ä, ö and ü. That's quite a small set, having more than 26 characters in the cypher text would surely help make it harder.
 
Actually, the only requirement is that the character set for the ciphertext include the plaintext character set, it can also include characters that are not in the plaintext, but the keyboard must include keys for those extra characters as well as keys for characters in the plaintext.



But adding additional characters would surely help. Plain English text has only 26 letters, and plain German text also has ß (German ss), and the three umlauts ä, ö and ü. That's quite a small set, having more than 26 characters in the cypher text would surely help make it harder.


This would make the machine bigger, increase the errors in transcription, and has previously been mentioned require a totaly new alphabet for transmission. Properly used eg not repeating the setting change from the day settings, allowing a rottor to be used in the same place two days running, allowing a letter to be encyphered as it self ( the instance that no letter could be itself both reduced the number of possible encryptions and assisted the code breakers by rapidly highlighting wrong assumptions) and avoiding rigid message patterns ( eg a weather report sent at the same time each day) and encrypting everything enigma would have been significantly harder to break.

The change you are suggesting would increase the time taken for machines to break a code but wouldn't stop the errors in use that provided the chinks in Enigma that the breakers exploited to break the code, and would make encryption, transmission and deciphering harder and more time consuming so increasing the likelihood of short cuts, or need to ask for retransmission both of which would make the messages more vulnerable.
 
Just using Western Arabic Numerals in addition to the 26 letters of the Roman Alphabet would not require a new transmission code, the existing transmission code incorporates both letters and numerals. The machine wouldn't be that much bigger with 36 position rotors.
 
The thing is, no matter how complicated they make it, it can still be cracked. the British were experts at the sort of quick raids to grab stuff like rotors and codebooks.
 
Well, yes, but the idea of using both letters and numbers is the added apparent randomness of the ciphertext. Sure it could still be cracked, but 36 rather than 26 symbols in the ciphertext may have made it harder.
 
Well, yes, but the idea of using both letters and numbers is the added apparent randomness of the ciphertext. Sure it could still be cracked, but 36 rather than 26 symbols in the ciphertext may have made it harder.
That doesn't help with the Heer and Luftwaffe since they use a system of sending a six-letter word with the last three letters in code, which means you couldn't start the rotors with a number since there is no word that I know of that has a number in it. Thus you limit the number of starting positions, removing most of the need for a 36 setting rotor. It also means replacing absolutely every enigma machine in operation.
 
But what if Enigma machines had 36 position rotors right from the start? The point is that no word has a number in it, but a 36 position rotor allows it to be encoded with a mixture of letters and numbers.
 
There are a few points to consider here.

First is purely mathematical if you have say 2 rotors with 26 possible outputs each you get 676 possible results, replacing them with two 36 output rotors gives you 1,296 possible results, adding an extra rotor instead gives you 17,576 possible results obviously much harder to crack.

The second point is almost all messages will use standard letters so this makes cracking the code much easier since the British can eliminate all rotor settings that would require a non-standard input.

The third point is ease of use, the entire point of the enigma system was not that it was unbreakable but that it was easy to use both encoding and decoding, while breaking would take so long that the message would be useless. It is easy to show that without electrical computing machines, enormous numbers of staff and any cribs decoding would take years (of course the British had all those things and possibly the brightest collection of minds ever under one roof). Adding extra letters would increase the likelihood of operators making mistakes, having to repeat messages or even giving up and sending urgent information in clear.

If you want to make the Enigma machine more difficult to crack there is a much easier way.

The designer arranged it so that no letter is ever encoded as its self, that is E never comes out as E, apparently this was supposed to make it harder to break but in fact this simplifies the problem massively. In addition to reducing the possible number of combinations it was easy to test possible cribs by checking to see if any letter matches if it does then you have the wrong crib, try it on a different message.

With out this quick and dirty way of reducing the possible messages that might match a crib decoding would be much harder and slower.
 
1. Yes, the point is that the enigma machine was used to encrypt messages to be transmitted with existing morse code. It puzzles me they only used 26 position rotors, given that Morse code can encodes numbers as well as letters.
2. No comment.
3. How is adding extra characters supposed to leave more room for mistakes?
4. The way the designer arranged it is that each letter could be encoded as one of 25 different symbols other than that letter, if each letter could be encoded as one of 35 different symbols, then this arrangement wouldn't have simplified code breaking as much.
5. Without this "quick and dirty" way of reducing the possible massages that might match, a letter is less likely to be encoded as itself if the range of symbols in the ciphertext is greater.
 
I am not as smart a Turing and company but I can see several ways using numbers might make things easier for a code breaker. Consider what numbers might be used in messages, dates, times, latitude longitude, unit designations , telephone numbers, compass headings, wind speeds. All these can be independently determined by the British and if you know where in a message a number is you only have 10 possible inputs, often less as there are rules over which digits are most likely.

It is possible increasing the number of possible inputs may increase security but the increase is negligible compared to increasing the number possible combinations of the output.

My comment on the likelihood of mistakes was addressing the original poster's question about nonstandard letter sets.

Doing some maths, if you want a 50% chance of spotting a false crib then you need about 16 letters for 25/26 combinations, 35/36 only increases this to about 20 letters. Yes it helps a little but eliminating the design flaw would be much easier and have a massive effect on the speed of decoding.
 
1. But numbers in the ciphertext would not be dates, times or any other examples mentioned.
2. In order to have more possible combinations in the output, there needs to be an increase in the number of technically possible inputs so that it can be decoded with the same machine that encodes if, but that doesn't mean that the cleartext needs to include all of them.
3. When I started this, I didn't think about the enigma cipher as having anything to do with radio transmission, I didn't think about encrypted messages being sent in a radio code that predates the enigma code.
4. Yes, eliminating the design flaw would help. And also extending the range of characters in the ciphertext, without that flaw, would mean the each character isn't encoded as itself as frequently as it would without the flaw but still only a 26 position rotor.
 
2. In order to have more possible combinations in the output, there needs to be an increase in the number of technically possible inputs so that it can be decoded with the same machine that encodes if, but that doesn't mean that the cleartext needs to include all of them.

No this is absolutely wrong, the number of possible combinations in the output depends (almost entirely) on the number of rotors and the design of the plug board not on the number of different possible inputs.

Imagine a binary machine, each digit has only two possible states but to decipher the message we need to know the initial positions of the rotors (and the plug board) increasing the number of rotors makes this exponentially more difficult even though each output digit can still only be 0 or 1. That is, there are many different combinations of rotors that can produce that 1 and having fewer possible states for each digit does not (significantly) help in deducing the set up.
 
...the number of possible combinations in the output depends (almost entirely) on the number of rotors and the design of the plug board not on the number of different possible inputs.

But in order to decode the message, you need to have a key for every possible character in the ciphertext, not just the ones in the plaintext. The Enigma machine was used for decoding as well as encoding.

Imagine a binary machine, each digit has only two possible states but to decipher the message we need to know the initial positions of the rotors (and the plug board) increasing the number of rotors makes this exponentially more difficult even though each output digit can still only be 0 or 1. That is, there are many different combinations of rotors that can produce that 1 and having fewer possible states for each digit does not (significantly) help in deducing the set up.

Okay, does this machine encrypt each bit individually? There are many different rotor combinations that, encode each bit as itself, and many that each bit as the other.
 
I think we have a terminology issue.
When I use the phrase "number of possible outputs for an individual character" I include all those outputs that produce the same character by different routes. This is correct from a cryptographic/mathematical viewpoint, the letter 'A' in the output could have been produced from the letter 'D' in the input by many different combinations of rotors but if you don't know which one that does not really help you decode the next character.

So for my hypothetical digital ENIGMA there are only two possible inputs for any digit but the number of possible outputs for that digit are enormous even though the print out will be "0" or "1".
 
But what if Enigma machines had 36 position rotors right from the start? The point is that no word has a number in it, but a 36 position rotor allows it to be encoded with a mixture of letters and numbers.
The first 'word' of every message gives the cypher code, with the first three letters sent in plain, and the last three referring to rotor position (so I'm given to understand), thus you can't start the rotors on any numerical value because the positions of the rotors have to be readily guessable.
 
Top