Version classiqueVersion mobile

B C, Before Computers

Stephen Robertson

12. Ciphers

Texte intégral

1It is easy to think of computers as giant calculators, and indeed the task of calculation and its mechanisation contributed both to the idea of constructing such a machine and to the conception of the tasks to which it might be addressed. Difficult calculation tasks such as those involved in ballistics (particularly in wartime) provided some of the stimulus towards the post-Second-World-War development of computers. But we have just seen, in the previous chapter, how a really rather different kind of task stimulated a form of mechanisation that brought us close to the computer era. The major stimulus for the actual invention of computers came from another domain again.

2The challenge was that of breaking the codes used by enemies in order to be able to read their supposedly secret messages, in the technological hothouse that was the Second World War. In order to see how this came about, we again start much earlier.

Codes and ciphers

3Throughout history, people have felt the need to write messages (point-to-point messages, in terms of our previous discussion) that would be unreadable to anyone other than the intended recipient, specifically to anyone who might intercept it en route. Military commands, intelligence reports, instructions to agents, love letters, arrangements for meetings, plans for any kind of action or activity that could prompt counter-measures of any kind by any third party—all these and many more might be deemed by the sender to need encryption.

4Since the word code is somewhat overloaded in present-day usage, I will use the word encryption to indicate putting some message into code, in such a way that it can only be read by someone who has the key to the code, and cipher for the method or rules for doing so. The original message is plain text and encryption results in the encrypted or cipher message. Recovering the plain text (given the key) is decryption. Discovering the key, or even the complete cipher system I may still refer to as code-breaking, in deference to popular usage. The whole subject, of designing ciphers and of breaking them, and of studying their properties (such as whether in principle they are breakable) is cryptography.

5A book with a marvellous account of the different kinds of ciphers that have been used through history, and of the efforts of opponents to break them, is Simon Singh’s The Code Book. Much of the rest of this chapter is drawn from Singh’s book.

The alphabet and encryption

6From the beginning and to this day there has been some use of word-based coding systems. A report in a newspaper on my table today describes a case in which some alleged terrorist plotters “used code words” for some possibly suspicious-sounding words, like firearms. But such systems are really intended to disguise or camouflage a coded message, rendering it less suspicious and therefore less likely to attract attention. Another approach is to hide the existence of a message altogether.

7However, most of cryptography addresses the question of how to render a message unreadable even when the adversary is in possession of what he or she suspects or knows to be a cipher message. Once again, it is hard to conceive of much of the history of encryption without the alphabet. Most encryption systems throughout history have been alphabet-based. Ciphers typically involve either or both of: re-arranging the letters of the message, and/or substituting different characters for those in the message. Even in Japan and China, we see evidence of the use of alphabets or alphabet-like symbol sets for encryption. Japanese ciphers tend to be based on one of the phonetic alphabets (kana), while a Chinese cipher might use, for example, either a phonetic alphabet or the so-called Four Corner method of encoding each character into four or five numbers, which is also used as a sort of substitute for alphbetical order, for sorting and then looking up characters.

8Given an alphabet, one of the simplest kinds of encryption is to substitute for each letter in a message the letter three places further on in the alphabet (this was a cipher used by Julius Caesar). If I do this with the heading of this section, I get

9Wkh doskdehw dqg hqfubswlrq.

10Or I could choose a different shift, or I could rearrange the cipher alphabet in some way. My intended recipient needs to know what cipher system I have used, a key that will enable him or her to decrypt the message: both the principle (‘alphabet shift’, for example) and the number of characters shifted.

11But as we shall see in a minute, such ciphers, in which a plaintext e is always represented by the same symbol in the cipher message (in this case an h) are normally very easy to break. To make a stronger cipher, we might use all 26 possible shifts of the alphabet, and a key that tells us which shift to use for which letter. The key is a word, whose first letter tells us which shift to use for the first letter of the message, second for the second, and so on. When we reach the end of the codeword, we return to the beginning. This is the basis for a Vigenère cipher, invented by Blaise de Vigenère in the sixteenth century.

12The Vigenère cipher makes use of the Vigenère square, showing all possible shifts of the alphabet (see Figure 18). Suppose that we again want to encrypt the heading of this section, and the codeword is revolution. We write out the plain text, and underneath it the codeword, repeated as many times as are necessary to match every letter of the plain text. Then we look up each plain text letter in the top row of the Vigenère square, and encrypt it with the corresponding letter in the row identified by the codeword letter. The first lookup (column T row R) is circled in the figure. Given the codeword, decryption is equally simple—but you need the codeword.


plain text


repeated codeword


cipher text

13The Vigenère cipher is much stronger than the simple substitution of the alphabet shift, and was thought to be unbreakable. In the example, you can see that the two As in alphabet are represented by different letters in the cipher text. But it can be broken—the man who established this fact is someone we have already encountered in Chapter 10: the nineteenth century mathematician and inventor Charles Babbage.

Figure 18: Vigenère square. Diagram: the author.

Figure 18: Vigenère square. Diagram: the author.

Code breaking

14Suppose that I am in possession of a cipher message, or a set of such messages from a single source—but that I am not the intended recipient, and do not know the cipher. If I have any reason to believe that the cipher is a simple alphabet shift, or indeed any simple one-for-one substitution, then it should be easy for me to discover the key and thus decrypt it. In particular, the number of occurrences of each letter will provide a clear clue as to which letters might have been substituted for, say, E or T or A (the most common letters in English). The longer the message the easier this is, but in the above short message I have three each of E and T, and also N, and only two As

15Even so, we see immediately that breaking is a different kind of task from encryption and decryption. Encryption, and decryption for the recipient in possession of the key, both involve following a very simple set of rules. Breaking the cipher, however, is a little more complex. The code-breaker may have to do some counting and statistics, and then try out a number of possibilities.

16Babbage’s method of breaking the Vigenère cipher involves looking for repeated sequences of characters in the cipher message. The distances between such sequences will give good clues as to the length of the keyword used, after which an extended form of the analysis of the statistics of letter occurrence, as used to break simple substitution ciphers, is likely to be effective.

17However, using a longer key (for example a phrase or an entire poem) makes it more difficult to break. The final stage of this development was to construct a whole series of long random keys, each printed on a separate sheet of paper, forming a pad, of which sender and receiver would each have a copy. The sender would encrypt a message using the first sheet, and would then discard the first sheet so that it would never be used again. The receiver would decrypt it also using the first sheet, and then discard the sheet. This cipher, the one-time pad, was invented by Joseph Mauborgne for the USArmy at the end of the First World War, and is known to be unbreakable by anyone not in possession of the one-time pad. Its major limitation is the necessity for producing and securely distributing the pad.

18In fact the process of inventing better ciphers (by those trying to send and receive secure messages) and devising ways of breaking them (by their enemies) is a game people have played for millennia.

Methods and machines

19Given that the processes of encryption and decryption are normally based on well-defined rules, it’s a little surprising that the use of mechanical aids was relatively slow to get going. Simple substitution ciphers require no more than a two-row table: plain-text letters on the top row and substitutes on the bottom. The Vigenère cipher requires a square table, with each of the 26 possible alphabet shifts on its own row. Even the one-time pad is essentially paper-based.

20However, it is also possible to make a simple mechanical device to help with either the simple substitution or Vigenère-style encryption and decryption, in the form of a pair of disks, one inside the other. The letters of the alphabet are written around the edge of each disk, and the inner disk is rotated relative to the outer disk to set up a single substitution table. If it is further rotated during encryption, a Vigenère-style cipher is produced.

21Such a disk was invented by Leon Alberti in the fifteenth century, and similar devices were in use for a long time, including during the American Civil War. Perhaps surprisingly, it was not until the twentieth century that the use of machinery for encryption and decryption advanced much further. However, the application of a complex cipher system really does suggest or even demand machinery: the more complex the rules to be applied, the more important it is to delegate their operation to a machine, which might be expected not to make mistakes.

22Mechanisation of encryption and decryption did not really take off until the invention of the Enigma machine. The German military famously used Enigma as their preferred cipher device during the Second World War, both for encryption and decryption, with daily changing keys; and the British, equally famously, had at Bletchley Park an establishment devoted to reading German cipher messages, which did in fact repeatedly and successfully break these daily ciphers.


23Enigma generates a letter-by-letter substitution of the clear message, but the substitution table effectively changes with every letter. But unlike the original keyword-based Vigenère system, the table does not repeat itself every few letters. It is more comparable to the one-time pad.

24It is a fascinating machine in its own right. Developed in 1918 by Arthur Scherbius, it looks very much like a typewriter—in fact the keyboard is closely based on Sholes’ keyboard described in Chapter 5 (which by 1918 was well established as the standard form of keyboard for typewriters). But instead of paper, the back of the machine has a replica of the keyboard in a lampboard, an arrangement of lettered disks each with a lamp behind it—see Figure 19. (You might note in passing that this keyboard differs a little from the Scholes typewriter keyboard (Chapter 5), although obviously derived from it. In particular, the offsets differ—having only three rows, the offsets are one-third of the key width. Experienced touch-typists would have noticed this!)

Figure 19: Enigma machine

Figure 19: Enigma machine

The letter L is pressed, and the D lamp is on.​wiki/​File:Enigma_Machine_A16672_open,_letter_L_pressed.agr.jpg
CC BY-SA 4.0.

25The clear message is typed in, as it might be on a typewriter, but at each keystroke, instead of printing, one of these lamps is illuminated, indicating a new letter—the cipher code to be used for the letter just typed in. The illuminated letter then has to be recorded somehow—written down or typed or transmitted directly.

26The mechanisms that allow the continually-changing table of substitutions are several and ingenious, and I will not attempt to describe them here. They depended on initial settings, which were changed daily; once into a message, the settings were changed automatically by the process of typing the message. That is, every keystroke resulted not only in the coding of one letter of the message, but also in re-arranging the table of correspondences for the next keystroke.

27The resulting cipher was extremely complex and difficult to break, but the complexity arose not so much from complex rules, as from a combination of many applications of simple rules. This is exactly the province of the machinery of the time, and it is no surprise that encryption and decryption should have succumbed to some such form of mechanisation, not long after the typewriter and the comptometer.

Breaking Enigma

28As I have indicated, code-breaking is a different order of task altogether. Almost inevitably, given a series of cipher messages, breaking the cipher system involves a combination of knowledge or guesswork as to the mechanisms involved in the cipher (or rather the rules which they mechanise), knowledge or guesswork about some key settings, and trial and error. There is a very strong sense in which code-breaking is an art-form. Like any art form it has its supporting technology (both in the form of machinery and in the form of know-how, methods and ways of doing things, sets of rules that may be applied), but it needs inspiration as well. This is certainly not true of encryption or decryption. It could be said to be true of devising new cipher systems—and indeed there are a couple of late-twentieth-century inventions here that look truly inspired—but as a hothouse for developing new ways of thinking, it is hard to beat Bletchley Park.

29Bletchley Park was the Second World War UK Government establishment in charge of attempts to read any intercepted cipher enemy messages. Many messages to or from military units of all kinds, of the German or other Axis powers, were intercepted and sent to Bletchley Park. And for much of the war many of these messages were successfully decrypted. There was always a challenge at the beginning of the day, because the keys were changed each day and the new key had to be discovered from some of the early messages intercepted. Then for some longer intervals, weeks or months, a particular cipher might become unreadable because of a change in some part of the encryption procedure by the Germans—until the Bletchley Park people had discovered how to deal with this new variant.

Post-war cryptography

30In the subsequent history of cryptography, following the end of the Second World War, the computer has loomed large. Most modern cipher systems are computer-based, in the sense that computer programs are used for encryption and decryption as well as by the code-breakers. In fact most systems make use of the fact that a message in a computer (necessarily in one of the binary codes discussed in Chapter 5) looks very much like a number, to which arithmetic operations can be applied. Of course it isn’t really a number, but with certain safeguards we can pretend that it is. We can encrypt by applying arithmetic operations to it such as addition and multiplication; decryption then means reversing these operations.

31One of the great discoveries of cryptography in this period is the principle of asymmetry. This is based on the fact that some arithmetic operations are easy to perform in one direction, but much harder in the other (it’s easy to multiply two large prime numbers; it’s much harder to factor the product and rediscover the original two primes). The resulting cipher system is known as Public Key Cryptography. It allows the person who wants to receive a message in cipher to make public an encryption key; anyone who wants to send him/her a message can use this key to encrypt it. However, the decryption key is different. Only the recipient of the message knows this decryption key—it need never be made available to anyone else. In principle, this makes for a much more secure setup—in almost all previous cipher systems, sender and recipient would have to share a key, and the necessity for sharing is a major source of insecurity.

Bletchley Park and its legacy

32Despite the fact that cryptography really entered the machine age only after the First World War, the challenge of cryptanalysis and code-breaking must really be credited with kick-starting the IT revolution of the second half of the twentieth century. In the end, we did not invent computers in order to control machinery, as Jacquard might have done; we did not invent computers in order to do repetitive numerical calculation, as Babbage tried to do. We did not invent them to analyse censuses; nor to organise our accounts or do payroll; nor to do weather forecasting; nor to do word processing; nor to facilitate telecommunications; nor to play our music or look after our photographs—though they are very useful for all of these things and more. We invented computers in order to break codes.

33The operation of Bletchley Park depended very heavily on people: collecting, transcribing, analysing the intercepted cipher messages. Initially, all analysis was entirely by people, using essentially pencil and paper, and human effort remained central to the code-breaking task. However, early in the war the great Alan Turing designed a machine called a bombe, which greatly helped in eliminating many possible initial settings (given a crib, a human guess as to the plaintext version of a particular section of the cipher text). This invention allowed Bletchley Park, for much of the war, to discover the day’s new key settings early in the day, enabling the decryption of any further messages that day as soon as they were received.

34Later in the war, the Bletchley Park effort had serious difficulties with another German system, the Lorentz cipher. This was similar to Enigma but more complex, and it typically took weeks to break one day’s messages. Max Newman, another Bletchley Park mathematician, started developing plans for a new machine that would be much more adaptable than the bombe—in fact, it was what we now describe as programmable. This was much more difficult to build than the bombe, but eventually in late 1943 the engineer Tommy Flowers designed and constructed a working version, using thermionic valves (as used in early radios). It was called the Colossus, and with its help, the keys for Lorentz-ciphered messages could be discovered quickly.

35Colossus was the clear forerunner of the modern computer. It was electronic, digital, and in some sense programmable, and used many of the ideas and principles and methods that a modern computer scientist would regard as essentially those of a computer.

An act of vandalism

36Then, at the end of the war, the entirety of what had been the Bletchley Park operation was eliminated. Winston Churchill, who had been the chief backer of Bletchley Park, ensuring funding for it against opposition from some quarters, demanded that all evidence of the UK’s cryptographic abilities should be utterly erased. Not only was Colossus itself destroyed, but all the blueprints for it were burnt. All Bletchley staff were required to keep silent about anything at all that went on there.

37Despite my heading, vandalism is a poor word to describe Churchill’s action. It was a 2000-year throwback to the first emperor of China, in the second century BCE—burning the library, in order to suppress the subversive knowledge held therein.

The next phase

38But it’s hard to kill an idea like that. In the world of the 1940s, outside Bletchley Park, some of the necessary ideas were already coming together. A project between IBM and Harvard University, masterminded by Howard Aiken, developed the Harvard Mark 1, a giant programmable calculator with many computer-like features, which first ran in 1943. The destruction of Bletchley Park left behind, in addition to the handful of eccentrics who believed in the possibility of building a computer, another handful who had actually seen one in operation. Within a year or two immediately following the war, academics in the UK (at Manchester and Cambridge) and in the US (in Pennsylvania and elsewhere) started building computers. Within a very few years, the computer age had taken off.

39But that’s another story.

Table des illustrations

Titre Figure 18: Vigenère square. Diagram: the author.
Fichier image/jpeg, 88k
Titre Figure 19: Enigma machine
Légende The letter L is pressed, and the D lamp is on.​wiki/​File:Enigma_Machine_A16672_open,_letter_L_pressed.agr.jpgCC BY-SA 4.0.
Fichier image/jpeg, 35k


Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search