Summary

Lecture 20 covers historical use of codes and code systems. We will trace their development and look at some examples of famous code systems. We will develop our subject with the help of several expert references. [FR8 ], [OAKL], ^[KAHN], ^[DEVO], ^[WEBE], ^[DEV3], ^[ELLI], ^[ACME], [LINC] ^[SIG2], ^[DAGA], ^[TRIT], ^[MACB], ^[COLE], ^[NICH], ^[MANS], ^[MAN1], ^[WEBE]

Code Systems

A code system is a highly specialized form of substitution. The basic principle underling code systems is the replacement of entire words, long phrases, or complete sentences constituting the plain text of a message by arbitrarily selected equivalents having little or no relation to the elements they replace; these equivalents may be other words, groups of letters, groups of figures, or combinations thereof. ^[FR8]

This replacement process is rarely applied to elements smaller than whole words and when this is done the elements are single letters, groups of letters, or syllables. In a codebook, the words, phrases, and sentences are listed in a systematic manner and accompanied by their code equivalents; correspondents must possess identical copies of the document in order to be able to communicate with each other. An ordinary dictionary may serve the purpose of code communication, so far as single words are concerned, but as a rule a specially prepared document containing the words, phrases, and sentences suited to particular types of correspondence is used. In the U.S. they are called codebooks or codes. Other names come from different locations: repertories, word books, and cipher dictionaries.

Tritheim Code Book

One of the earliest code books was developed by the Benedictine Abbot, John Tritheim. He collected many of the ciphers used in the European courts. He was familiar with the occult and proposed a code based on cabalistic words wherein he tried to hide the real meaning under cover of a mysterious language. The courts burned his book "Polygraphia" in great pomp and ceremony. John was lucky that he didn't go with the fire. The first edition was published in Latin in 1518, a French translation in 1541, followed by a German translation.

Part one of Polygraphia consisted of a number of code words for each letter of the alphabet, but arranged in such a manner that if each letter of the message was replaced by a code word, the result was a complete sentence having an innocent meaning. (Something akin to the three column management techno-babble matrix that was popular in the 80's - pick a word from columns A, B and C, put them together and you have a clever sounding and totally bogus phrase like "computer redundancy equivalents'.) Table 20-1 shows the fourteen coded alphabets illustrating the way they were meant to be used:

Example:

Plain text: 'Do not use bearer.'

Cipher text:

Look under D in the first alphabet = King, 'O' under the 2nd alphabet = Triumphant, etc.

Note the interesting and rich language in the above 14 alphabets. The unfortunate thing about Tritheim's codes was that coded messages required as many words as there were letters in the plain text, which made for a long cryptogram. Note also that some of the words were duplicated which might have caused some confusion.

From Lloyd To Marconi

In 1688 Edward Lloyd ran a coffee-house in Tower Street, London. An enterprising man, he found that several brokers used to discuss their business over coffee. To sell more coffee, he decided he must make things easier for them. He instituted a blackboard, and then a weekly bulletin of shipping information. More independent brokers came and consumed his coffee while doing their business. He later moved his coffee house to Lombard street, in the very center of the old city of London frequented by merchants of the highest class. It was not until 1774, with the rapid increase of marine insurance business, a committee was set up and a constitution formed which has remained practically unaltered to the present day. There is no longer a Lloyds' coffee-house, yet the name is preserved, and Lloyds'is known all over the world as the center of the Marine Insurance business.

Lloyds devised a method of signalling between sea and shore, so that advance news of ships and cargoes might be received. A primitive projector was set up and a system of light signals based on the Polybius' system was started. It was this that gave rise later on to the use of codes for commercial purposes; and apart from the Venetian merchants in the eighteenth century, Lloyds signals were the first to come into general use.

In 1794 in Europe, a system of rapid communications known as 'aerial telegraphy', employing semaphores on high towers visible at considerable distances, was instituted. Whole phrases or sentences could be expressed by one group of signals.

In 1825 codes employing figure groups were in common use. In 1845 the Telegraphic Vocabulary Code was used between Liverpool and Holyhead for the semaphore telegraph. In this code there appear words, phrases, long sentences, each represented by groups of one to four digits.

In England the earliest practical trial of electric telegraphy was made in 1837 on the London and North Western Railway, and the first public line, under Wheatstone and Coke Patents, was laid from Paddington to Slough on the Great Western railway in 1843. ^[DAGA]

In New York, in 1860, Brewell published his Mercantile Cipher for condensing telegrams, in which English dictionary words were employed, and in which we find a fairly complete vocabulary, arranged under captions.

The ABC code, also based on dictionary words, first appeared in 1874. (Refer to Table 20-2.) Up to 1872 the telegraph companies, by international agreement, charged pronounceable code language words as plain text; the higher tariff applied only to cipher or numeral language. These were charged for at a rate of five characters per word; and in 1875, at St. Petersburg, the maximum length was fixed for either plain text or code words at seven syllables. This led to abuse, such as words as Chinesiskslutningsdon - 21 letters, but only 6 syllables - were used by coders. ^[DAGA]

The rule was changed to apply to European or Latin words but not artificial words. In 1903, code words of ten characters were allowed. They had to be pronounceable to be authorized for transmission at the cost of plain text words.

ABC Code

In 1904 Whitelaw's Telegraph Ciphers appeared with 400 million pronounceable words. Not really a code book, it was a list of 'artificials' used for private codes. These code words were composed of five letter only, for example FORAB, LUFFA, LOZOJ, etc. as are all words used in commercial codes today. Twenty-thousand words of five letters each were given, and since each was pronounceable, and any two of these words could be joined together to form a chargeable according to telegraph regulations as one word, so 20,000² gave the total of potential words as 400 million.

In 1906 Bentley's code appeared, a compact phrase book based on five-letter groups, applicable to business affairs in general. It cut the cost of international transmissions by half.

Morse Code

Samuel Finley Breese Morse was born in 1791 in Charlestown, Mass. His invention of the electrical telegraphy was second only to the famous 'Morse Code'. He based his Morse Code on the frequencies of letters calculated on quantities of type found in the printing office. Since his frequency tables are an enormous help in deciphering every code, lets compare here the original calculation made by Morse with the Normal Frequency and the Telegraph Frequency. (See Table 20-3)

For the letters which were most frequent he used the simplest combination of dots and dashes, which an automatic contrivance of the electric current alternately transmitted and suspended during longer or shorter intervals and reproduced at the other end of the wire on strips of paper. The experienced operator knew the 'fist' of the sender as well as the differences between the dots and the dashes.

Order       1  2   3 3 3 3 3   4 5 6 7 8   9 9
Morse:      E, T,  A,I,N,O,S,  H,R,D,L,U,  C,M
Telegraph:  E, O, A, N, I, R, S, T, D, H, L, U

Order       10  11   12   13 14 15 16    17  18
Morse:      F, W,Y,  G,P,  B, V, K, Q,  J,X,  Z
Telegraph:  C, M, P, Y, F, G, W, B, V, K, X, J, Q, Z

This comparison is remarkable. The normal frequency order corresponds to LANAKI's data presented in Lecture 1. ^[NICH]

The Morse code was not only used in telegraphy but also in signalling by flags, by flashes of lights, by long and short blasts from a whistle, and for some of us knocks on the wooden cages to fellow prisoners in Viet Nam.

The Army used to allow 10 days for recruit signalmen to learn Morse code. Morse presents a simple method that he invented in Table 20-4. This table presents a short list of words, one for each letter of the alphabet, the long and short syllables indicating dashes and dots.

The famous message SOS = SAVE OUR SHIP = ... --- ... = dit dit dit dah dah dah dit dit dit.

Observe that each of these words contains as many syllables as there are dots and dashes in the corresponding Morse alphabet; but owing to the difficulty of finding suitable words, it was assumed that vowels followed by two or more consonants are long and those by single ones short. In the words Katherine and offensive, for instance, the final syllable must be considered long. Morse put together the following memorization aid:

GALLANTLY and FURIOUSLY he fought AGAINST the foe at WATERLOO.
IVY creeping along the ground suggests HUMILITY.
The JURISDICTION of the NOBLE LEGISLATOR was OFFENSIVE to the BARBARIAN.
A PHOTOGRAPHER saw SEVERAL KANGAROOS on the MOUNTAIN.

U.S. Coast Guard Discontinues Morse Code

Cipher history takes strange turns. It was with some sadness that I read the 31 March 1995 announcement by the USCG that Morse code equipment would shut down after more than 83 years of monitoring telegraph distress calls such as the 1912 Titanic collision with an iceberg. Switchoff occurred at USCG communication centers in Boston, Honolulu, Hawaii, Miami, New Orleans, San Francisco, and Kodiak, Alaska. Even private listening posts will cease service.

Samuel Morse invented the code to carry messages on the telegraph machine he patented in 1840. Morse cipher systems followed soon after that. USCG operators were a breed apart because they could send and receive international language at 20-35 wpm or more. Radio hams know the meaning of "personal touch" as keyed dots and dashes bounce off the atmosphere. WWII vets know that the radioman's "fist" was more identifiable than passwords.

I took my radio training via USCGA SAR. I know the feeling of listening to 11 radios concurrently. Sometimes a May Day could be heard only once and action had to be decisive. Morse has been replaced by automatic equipment to link with Global Maritime Distress and Safety Systems via satellite relayed signals with location fixes.

The ending of the Chesapeake USCG Atlantic COM Center for Morse at 1919 hours EST is the ending of a great era.

I went to my closet today. My USCG web belt still fits. I couldn't bring myself to pull out my old uniform. I just don't look like Arnold Schwartznegger anymore.

Commercial Codes

Historically, commercial codes were used not so much used for secrecy as for saving money on long telegrams. Authorized, pronounceable words of maximum length of ten letters being used to cover several sentences. The code words used were entirely fictitious, and followed each other in alphabetical order, being made up of five letters each, so that two codewords can be sent by telegraph for the price of a half word. [Note that modern day E-mail on the Net has completely made this a non-issue. In any one day, I may write to classmates in England, Germany, Italy, Japan and Spain and in less than 30 minutes have answers, with attachments, and be charged a flat rate for the service on this end!] Other codes constructed on these principles were Bentley's and Webster's. They allow two words, or even short sentences, to be formed into one telegraph word of ten letters. There are commercial codes today with equivalent translations into every European language, so that English, German, or Italian business men, without knowing each other's languages, can exchange telegrams (or FAXS).

Marconi Code

Senator Guglielmo Marconi was devoted to an idea - the sending and receiving of wireless signals through space. His wireless inventions are legendary. Marconi also invented and perfected the Marconi Codes. The complete Marconi code consists of four volumes comprising English, Spanish, Japanese, Russian, Italian, Portuguese, German and Dutch equivalents. The English text is alphabetical, and every other language has a complete index of all the words. The code is divided into two parts - one containing general phrases and the other a numerical system.

Again, the chief aim of standard code was to save cost of cable charges and the cost of time required to code the messages. Upwards of 17,050 combinations could be obtained by the Marconi code. A checking system was used to ensure accuracy.

The code words were composed of five letters each, corresponding to a word or sentence used in trade or business. The codewords could be combined to for a telegraph word of ten letters by the International Telegraph regulations.

There were some differences with codes such as the ABC code. Each code word has a two-letter difference from each other code word. This two-letter difference ensured that no two words would have the same four letters in the same position. A code word like BOPEZ would eliminate codewords like COPEZ, DOPEZ and also such forms as BAPEZ and BEPEZ. (see below)

The Marconi Numerical System was arranged so that a range of figures in combination with some of the most commonly used qualifying phrases, together with an efficiency check, could be transmitted in one complete pronounceable word of ten letters. The first syllable in this section consisted of two consonants, thereby distinguishing it from a phrase section in which none of the code words began win two consonants. As the code words in the numerical section were only two letters long, five words or phrases could be included in one telegraph word of ten letters.

The Marconi arrangement was as follows (Refer to Table 20-5):

1. 1st Syllable provided for a variety of phrases which were employed in combination with the figures or phrases in the following syllables, describing as 'qualifying phrases'; e.g. 'TH' = remit by cable, 'TW' = ship immediately.
2. 2nd Syllable provides for an extensive variety of phrases descriptive of the following weights and measures; i.e. 'OM' =pounds, 'WG' = tons.
3. 3rd Syllable provides for figures from factions to 100.
4. 4th Syllable provides for more figures to be used in conjunction with the third syllable. If unnecessary a blank must be used here, or short phrase to qualify, such as 'ZA' = per month.
5. 5th Syllable provides for a further series of phrases to be used in conjunction with the foregoing; e.g. 'AL' = for immediate shipment. It also supplies a check for the whole coded word.

The checking system is very simple. The check numbers given in brackets on each code syllable are added together for the four syllables used; tens are disregarded, and for the fifth syllable the letters are chosen from the column bearing the same number as the total arrived at from the addition of the first four syllables.

Compare the ABC code Table 20-2 with the Marconi code in Table 20-5.

Non-Secret Codes

Various codes are suited to particular types of correspondence. Many large commercial firms have their own private codes. For example, an early commercial codebook was made by ACME Commercial Code Company in the 1930's. (See Tables 20-2a&b) Most industries have highly specialized technical language (part of the defense of mystique in every industry or profession -Latin for doctors and lawyers, female terms and mathematics for engineers, ISO 9000 terms for quality managers, snake oil terms for computer types, plus a whole bevy of terms for cryptographers, etc). The purposes of many these codebooks are brevity and compression not secrecy. The military and diplomatic applications call for security, and speed of communications, especially for front-line communications.

The PKZIP program, which is used so widely on the net, is a compression 'codebook'. It provides economy of transmission and minimal crypto-security. The power of the program lies in the ability to delineate and hold entire directories and then to create an indexed tree of the coagulated sum of files. PKZIP is an example of a non-secret code. Compression is more valuable than secrecy. The condensing power of a code is dependent on its vocabulary. When we add the goal of secrecy to economy, we then have a secret code. Actually, code transmissions save money because of the lowers number of characters to be transmitted over the channel.

ACME Seven Digit Codebook

In 1934, the ACME Code Company, with offices in London, New York and San Francisco, developed a codebook for condensation 7 figures into 5 figure groups for international business cables. The transoceanic standard for code language was issued January 1^st 1934, and superseded the category 'B' regulation (CDE) five letter code words without vowel restrictions. Category B service was cheaper on short coded messages that the category cable A intercontinental transmissions.

I have codebook number 6015. It is laid out in three tables on each page. Table 1 is for 1^st and 2^nd Figures, First and Second Letters; Table 2 is for 3^rd, 4^th and 5^th Figures, Third and Fourth letters; Table 3 is for 6^th and 7^th Figures, fifth letters. The conversion (condensation) of 7 Figures into a five letter code word and vice versa is accomplished on one page, in a single operation. Numbers 0000000 to 9999999 are included in the codebook. The condenser is used for :

14 Figure codes (2 Five Letter code words)
21 Figure codes (3 Five Letter code words)
28 Figure codes (4 Five Letter code words)
etc.

It can be used in conjunction with any numbered code, catalogue, parts list, steamer list, etc.

Encoding

To encode the figures 4732651 into a five letter code word, we divide the 7 figures into three groups:

The 3^rd, 4^th, and 5^th figures determine the page from which we apply the condensation codes. 326 is found on page 3 of the codebook. (Table 20-6a&b reproduces part of the pages.)

Alongside the figures 326 are the two letter groups YM, YN, YO, YP, which gives us the 3^rd and 4^th letters of the codeword we will form.

To determine the group to use, we look at the 6^th and 7^th figures table and find the fifth letter. The figures 51 are in the same column as YP and the letter alongside of 51 is M. Thus we have the 3^rd, 4^th, and 5^th letters of our codeword YPM.

To get the 1^st and 2^nd letters of the codeword, we refer to the table covering the 1^st and 2^nd figures, on the same page and is found that for 47 are FR. The entire codeword is then FRYPM.

The codeword then is FRYPM.

Decoding

To decode the codeword STROW we break it down:

The first and second letters determine the page from which you will decode your full seven figures. In this instance the First and Second letters ST, they will be found on page 6, alongside of which we find the 1st and 2nd figures "87".

You then look on the same page for RO, in the table covering the third and fourth letters. It is found that RO means "782," thus giving you the 3rd, 4th and 5th figures.

W being the final letter, we refer to the table for 6th and 7^th figures. In the same column as RO appear and on the same line that the letter W is found, we find the 6th and 7^th figures 84. Our final product is 87-782-84. STROW = 8778284.

ACME also produced a Commodity and Phrase Code Supplement which was just as much fun? ^[ACME]

Brevity Codes

In military cryptography, the greatest degree of condensation is afforded by 'prearranged-message codes,' or 'brevity codes.' A prearranged-message code is a tactical code adapted to the use of units requiring special or technical vocabularies; it is comprised almost exclusively of groups representing complete or nearly complete messages and is intended for shortening messages and concealing their content. The police '10' codes fall into this category. A brevity code has as its sole purpose the shortening of messages. A field code is a small tactical code which contains a large number of code groups representing words and a few common short phrases, from which sentences can be composed; a syllabary, which is a list of code groups representing individual letters, combinations of letters, or syllables, is used for spelling out proper names and . numerical tables, or list of code groups representing numbers, dates, and jargon. The Army Special Forces Codes fall into this category. A jargon code is a very short code in which bona fide dictionary words, baptismal names, rivers, lakes, etc are used as code groups. Lincoln's war time codes fall into this category. [LINC] A voice code or recognition code is used for transmission by small radio-telephone sets used in combat. Other names are combat code or operations code. [TEC] The Navy has a special brand of codes used for protection of marine traffic. An example of this code system is the International Code of Signals (1969 edition, revised 1981 INTERCO) ^[SIG2]

International Code Of Signals For Visual, Sound And Radio Communications (INTERCO)

The Defense Mapping Agency, Hydrographic/Topographic Center issued in 1969 and again in 1981, their Publication No. 102, "International Code of Signals For Visual, Sound, and Radio Communications," United States Edition. This code was adopted by the Fourth Assembly of the intergovernmental Maritime Consultative Organization in 1965. The document was prepared in nine languages: English, French, Italian, German, Japanese, Spanish, Norwegian, Russian and Greek.

This is very good example of the brevity and non-secret codes that had wide distribution for ocean going vessels. Modern day vessels use uplinks to satellites in geo-synchronous orbits to navigate and communicate.

The INTERCO was designed to communicate for situations relating to the safety of navigation and persons, especially when language difficulties arise. It is suitable for transmission by all means of communication including radiotelephony and radiotelegraphy. The INTERCO embodies the principle that each signal has a complete and distinct meaning.

The INTERCO is broken into four parts: 1) Signal Instructions, 2) General Signal Code, 3) Medical Signal Code, and Distress and Lifesaving Signals and Radio Procedures. The appendix includes a National Identity Signals for Ships and Aircraft, plus US/USSR Supplementary Signals for Naval Vessels.

General Signal Code includes sections on: Distress, Emergency, Casualties, Damages, Aids to Navigation, Hydrography, Maneuvers, Cargo, ballast, Meteorology, Communications and Sanitary Regulations. ^[SIG2] See Table 20-7 for sample entries. In Table 20-7, capitalized headings represent major topics, predominantly lower case headings represent subtopics. You can see from the small sample in Table 20-7, that the INTERCO deals with serious situations. I was assigned to a U.S. Coast Guard Radio Room and I can tell you that listening to 11 radios at the same time can be very intense. A MAYDAY maybe heard only once and rarely in calm voice. Sending the cutter is serious business. The USCG does their job exceptionally well.

To indicate DISTRESS:

1. If possible transmit ALARM SIGNAL (i.e. two tone signal) for 30 seconds to one minute, but do not delay the message if there is insufficient time in which to transmit the Alarm Signal.
2. Send the following DISTRESS CALL: MAYDAY MAYDAY MAYDAY. This is ...(name or call sign of ship spoken three times).
3. Then send the DISTRESS MESSAGE composed of:
   1. MAYDAY followed by the name or call sign of the ship;
   2. Position of ship;
   3. Nature of distress;
   4. And if necessary, transmit nature of the aid required and any other information which will help the rescue.

USE PLAIN LANGUAGE WHENEVER POSSIBLE or send the word INTERCO to indicate that the message will be in the International Code of Signals. example:

MAYDAY MAYDAY MAYDAY ... (name of ship spoken three times, or call sign of ship spelled using Phonetic Alphabet in Table 20-8); MAYDAY ... (name or call sign of ship) Position 54 25 North 016 33 West I am on Fire and require immediate assistance.

Basics Of Code Construction

The encoding and reverse procedure of decoding is accomplished by replacing various words, phrases, sentences, and numbers by their code equivalents. The code text is built up from code units each representing the longest possible plaintext unit the code book affords. Encoding the phrase "enemy force estimated at one battalion," and the codebook has phrases "enemy force," and "estimated at," as well as the individual words, we would write down the phrase equivalents.

The elements of which code groups are composed may be one or more of the following:

1. Bona fida words - real words from Dutch, English, French, German, Italian, Latin, Portuguese and Spanish.
2. Artificial words - groups of letters without meaning with vowels and consonants arranged to appear like real words.
3. Random groups of letters.
4. Groups of Arabic figures.
5. Intermix groups, ie. call signs for stations K2KAA, or W5AZZ.
6. All the above.

Parallel Sets

A code may contain two or more parallel sets of code groups of different types. In many commercial codes and some military codes, there is one series of code groups of the bona fide type or artificial word type and another series of the figure-group type, both applying to the same series of words phrases, and sentences of the code. In parts of the world where English letters are used for writing, letters possess greater advantages in accuracy of reading than figures - especially for telegraph or radio transmissions. For communications to China and Russia or obscure ports, Arabic figures are well accepted and code groups composed of figures are used. The main reason for this is assurance of the correct transmission and reception of messages in all parts of the world. Another reason is that certain methods of enciphering code messages for the sake of greater secrecy, figure groups often form the basis for encipherment more readily than do letter groups.

The greatest advantage possessed by letter groups over figure groups lies in the availability of a far greater number of permutations, or interchanges, of letter groups, because there are 26 letters which may be permuted to form letter groups compared to 10 digits for figure groups (assumes base 10 historical use). If code groups of five letters are used, then there are 26⁵ or 11,881,376 groups of five letters versus 10⁵, or 100,000 groups of five figures. Letter code groups are usually constructed to reduce error in transmission.

The length of code groups used, whether the groups consist of two, three, four, or five elements, depends upon the size of the code. This applies almost exclusively to field military or naval codes, where transmission is through a governmental agency; in commercial messages or governmental communications transmitted over privately operated lines, five-letter or five letter groups are the standard. ^[FR8]

Code groups of modern codes are constructed by the use of tables which permit more-or less automatic and systematic construction in the form desired. These are called permutation tables. Because they may be used to correct most errors made in transmission or writing, such tables are usually included in the code book and are called mutilation tables, garble tables, error detector charts, etc.

Two-Letter Differential

The average telegraph or radio operator did not work without error. One letter different code groups like ABABA and ABABE were easy to mistake and the message could be made unintelligible by only a few transmission errors. If however, every code group in the code book is distinguished from all other code groups in the same code by a difference of at least two letters, then there would have to be two errors in a single group and these two errors would have to produce a code group actually present in the code before a wrong meaning would be conveyed. The principle of making code groups differ by a minimum of two letters is called the two- letter differential. The two-letter differential reduces the possibilities for constructing letter code- groups from 26 ⁵ to 26⁴ (456,976) but considering the advantages, the sacrifice was worthwhile. Permutation tables for construction of figure-code groups are similar in nature and purpose to tables for construction of letter-coded groups. Because of a more limited number of characters available for permutations, the maximum number of 2-figure difference groups possible in a 5-figure code is 10⁴, or 10,000. (This does not account for ASCII code derivations.)

Types

In their construction or arrangement, codes are generally of two types:

1. One-part, or alphabetical codes. The plaintext groups are arranged in alphabetical order accompanied by their code groups in alphabetical or numerical order. Such a code serves for decoding as well as encoding.
2. Two-part or randomized codes. The plaintext groups are arranged in alphabetical order accompanied by their code groups in a non-systematic order. The code groups are assigned to the plaintext groups at random by drawing the code groups out of a box in which they have been thoroughly mixed. Such a list serves for encoding. For decoding, another list must be provided in which the code groups are arranged in alphabetical or numerical order and are accompanied by their meanings as given in the encoding section. Another name for the two-part code is cross-reference codes. Here are extracts from typical one-part and two-part codes. (Tables 20-9 and 20-10.)

Between the two extremes are codes which have features of both; that is complete sections may be arranged in random sequence, but within each section the contents are arranged in some logical order.

When a strict alphabetic arrangement is used in the sequence of the phrases, the code is said to be a strictly alphabetical code. When the phrases are listed under separate headings based upon the principal word or idea in the whole expression, the code is called a caption code. (Tables 20-11 and 20-12)

Assistance

Give assistance

Require assistance

No assistance

Assistance has been sent

Assistance for

Assistance from

Assistance to

Assistant

Assisted

Assistance

Assistance for

Assistance from

Assistance has been sent

Assistance to

Assistant

Assisted

.........................

Give

Give assistance

.........................

No

No assistance required

.........................

Require

Require assistance

More precise and economical coding is possible with a caption code than with an alphabetical code. With a caption code it is easier to assemble an extended variety of expressions and shades of meaning under specific headings than with alphabetical code. On the other hand, the use of a caption code involves more time and labor in encoding.

Two-part codes are used by many governments for their secret diplomatic, military and naval communications because of the advantages they offer over one part codes. Some disadvantages include twice as large in context, printing and distribution costs, compilation is four times greater because of the requirement of accurate cross references. The advantages of two-part codes are greater security and greater accuracy.

In some commercial code messages there is sometimes encountered the practice of mixing plaintext and code text. In governmental and naval communications such intermixtures are rare and present an abysmal ignorance of the fundamental rules of cryptographic security. Because the plaintext words give definite clues to the meaning of the adjacent code groups, even though the former convey no meaning in themselves (such words as and, but, by, comma, for, in, period, stop, that, the, etc) constitutes a fatal danger to the message security.

Enciphered Code Systems

Sometimes the code groups of a code message undergo a further process of encipherment; the resulting crypto- gram constitutes an enciphered code message. Both transposition and substitution may be used to encipher the code. Enciphered code is used under the following circumstances:

1. When the code has a wide distribution and may fall into enemy hands,
2. To improve the security of commercial codes and non-secret codes, and
3. When increased security is necessary for highly classified communications.

Transposition methods are generally used within code groups, such as rearranging or shifting about the letters or figures composing them. A common method is keyed columnar transposition with special matrices with nulls. All the substitution methods previously studied may be used for "super-encipherment" of the code. The most effective methods of enciphering code are arithmetical methods.

If the code groups are numerical, the addition (usually mod 10) of an arbitrarily selected number (called the additive) to each code group message constitutes a simple form of encipherment. The additive may be fixed.

Additive methods may actually be weak cryptographically if the basic code book and code groups embody limitations in construction. Instead of adding a fixed number in encipherment, the latter is subtracted, in which case, in decipherment, the fixed number must be added to the enciphered code groups as received. Such a group (called subtractive or subtractor) in decipher- ment the group becomes an additive. A third method used commonly is the minuend method. It involves the subtraction of the plain code group from the key to yield the enciphered code group in encipherment, and the subtraction of the enciphered code group in from the key in decipherment. Addition and subtraction of a fixed numerical group may be alternated within the same message such as +200, +100 +400 as a cycle or +200, - 100, +400, -200 etc. Instead of a fixed additive, it is possible to employ a repeating large key.

When special tables are employed as the source of the additives or subtractors for enciphered code, a much more secure system is provide. These tables are called a key book or an additive book or a subtractor book. by applying identifying symbols called indicators to the pages, as well as to the rows and columns on each page of the key book, it is possible to provide for secure encipherment of a large volume of traffic. All corespondents must have the same key books. In employing the key book, the indicators tell the recipient of the message what key groups were used and where to begin the decipherment of the enciphered code.

In actual practice, indicators are often disguised or encrypted by a special key or set of keys; this procedure may add considerably to the security of the system.

Table 20-13 shows a page from a typical key book. It contains two sets of 100 4-digit key groups, disposed in numbered blocks each containing 10 rows and 10 columns of groups. To designate a group as the initial one to be employed in encipherment or decipherment, we give the block number, the row and column numbers of the group. For example, 0116 is the indicator for the group 8790. It is usual to take the successive groups in the normal order of reading. Some keys books consist of 50 + pages containing 200 + groups making 10,000 in all. The digits in each block are random numbers. [FR8 ] If the key book is used once and only once, security of the system approaches the one-time pad. The messages are one time system secure even if the enemy has basic code book. Friedman discusses indicators in much more detail in [FR8 ].

Dictionary Codes

Dictionary codes are highly specialized forms of substitution systems. Code books (modified dictionaries) used by the Department of State and military represent a greater condensation of words than comm- ercial systems - a single code group may represent a long phrase. The average condensation of a diplomatic code is 1:5 while a commercial code is only 1:3. ^[DAGA] By way of comparison, modern PKZIP compression is 1:3 - 1:4 on normal text. I recently experimented with PKZIP on the TEA program library for eight words and up and found an average compression of 1:2.5. These groups are all pronounceable artificial words. For example, we might have ABACA in commercial code, EXA in diplomatic code and occasionally syllables as BA in Marconi code.

It is difficult to safeguard against the loss of codebooks which have to be printed in fair numbers. Macbeth reports on an interesting story about Ottoman Field-Marshal Osman Pasha during the Russo-Turkish war in 1877. Pasha entrusted one of his generals, Selim Pasha with a confidential mission. Selim was the officer in charge of ciphers and codes and always kept the code book on his person. Selim departed so promptly on his mission that he forgot to leave the volume with his chief. And the latter, during the whole time of the Adjutant's absence, saw a pile of ciphered messages from Constantinople accumulate on the table without being able to read or reply to them. ^[DAGA]

Codes used in conjuction with ciphers (superencipherment) can be very difficult to break; but the work and time involved in making this combination can be significant (if done by hand in the field.) Computers reduce the legwork significantly.

The typical dictionary code protocol is as follows:

1. Agree with the recipient on the exact edition of the diction to be used, i.e. Concise Oxford Dictionary, current edition, by Fowler and Le Mesurier.
2. Use the number of the page, and the number of the word down the page to encipher:
   Given Plain: " Reunion Berlin Tomorrow"
   Code:
   1006 (page no.), 12(word no) = Reunion 0104 (pages with fewer than four numbers would have a 0 added in front to keep to the uniformity), 17 (word no.) = Berlin 1291 - 08 (on the same principles) = Tomorrow
   Ciphertext:
   100612 010417 129108
   These figures, if greater secrecy is required, could again be enciphered and thus converted into letters by means of an agreed upon cipher.
3. Prepare for superencipherment by dividing the figures into pairs and then convert them into letters by means of a table such as Table 20-14.

The numbers enciphered into letters:

and the cryptogram for transmission:

The suggested cipher can easily be arranged to make pronounceable words suitable for telegraph or radiotelegraph transmission.

Certain dictionaries have been issued which give two columns on each page with words directly opposite to each other. Then it is possible to give the word opposite the one we really mean, or a word which is 5 or 3 or 10 places either above or below the one we want to encode. Codes of this kind can be solved readily.

Cryptanalysis Of A Simple Dictionary Code

An Australian criminologist named Mansfield presented some interesting principles for solving dictionary codes. He calculated dictionary progressive lists, giving numbers of words beginning with any two letters in dictionaries of 10,000 - 100,000 words. ^[DAGA]

Given:

We rearrange the list from lowest numbers to highest.

Words beginning with XYZ are seldom used, so we can take it that the highest number indicates a word beginning with a W or a T. [Mansfield made big assumptions about nulls and standardization of the dictionary. Lectures 2 and 3 showed how we can rip this assumption to shreds.] But the list of bigram frequencies (from Lecture 1) gives us the commonest initial group as TH or THE, and if we fix any repetition of such nature, then we may have the T in that dictionary. Naturally, we start with 55381 occurring five times and assume it is THE.

The highest number after that is 61571, so that it could indicate a word beginning with a W. This gives us a clue to the probable number of words in the dictionary used for the code. It cannot be over 65,000 words as XYZ words are very few, seldom more than 3,000. [This part of Manfield's analysis is an extraordinary jump of faith -what is more extraordinary is that it will work more than 60% of the time on simpler dictionary codes.]

According to Mansfield's Progressive Dictionary Lists, we attempt to fix the probable first two letters of each word in the code. For instance the 2nd group 12050 will be between 11646 (terminating words beginning with DA) and 12850 (terminating words beginning with DE), so that it is probable to be a word beginning with DE. ^{[DAGA] [MANS]}, ^[MAN1]

Using Mansfield's lists we obtain:

We locate in the dictionary the word THE (55381) and count back twenty words for 55366 (th). This gives us an area covering words THANE, THANK, THAT, THATCH. We try the most likely THAT. We note the two words starting with letters TO- 56037 and 56442. Words beginning with TO start at 56037 and stop at 56466, so that it is reasonable guess to assume the first is TO and the second (56442), we count twenty words back to find the word TOWN.

The R group is: -RE- (42872) and RE- (44234) and RO- (45174). RE stands 300 words from the end of the RA's which stop at 42573, according to Mansfield's tables. This gives us the following words to select from: RECLINE, RECOMMEND, RECOMPOSE, RECONNAISSANCE, RECOUP, and RECOVER. We choose RECONNAISSANCE. The next look at our cipher is:

We apply the same process to the AE- 00723 and get airplane, while the DE- 12050 occurring one-quarter of the way from the end of the DA to the end of the DE brings us to DEF, limited by DEFACE and DEFY, where only DEFEAT, DEFENSE, DEFEND, and DEFENSIVE are probable. We select airplane defensive us near the mark.

SE- should be sea 46882 and OVER for OV- 36173. The of- is in fact OF, and the HA- is has, and the WA- is was. The complete message reads:

^[MANS] tells us that the real message was off by two words. Instead of AIRPLANE DEFENSIVE, it was AIR DEFENSES, but the meaning was essentially the same.

What Mansfield did show us in 1936 was that the laws of probability work with dictionary codes. The search in the area of possible words will give us the root of the plain text so that we may deduce the whole meaning of the code.

Diplomatic Codes

One of the best references on historical codes (1775-1938) in the United States was written by Professor Ralph Weber. ^[WEBE] He describes one interesting code used in 1867 by the State Department known as WE029. (Refer to Table 20-15) It used a simple substitution masking procedure, eliminated the use of the letter W because it was not used in European or Latin nations, focused on 24 letters of the alphabet and assigned them to the 24 most common parts of speech such as articles and other words (s= plural; a = THE; e = AND, etc.) Other ordinary words were assigned to the approximately 600 combinations of 2 of the letters. Three letters were used for the remainder of the vocabulary required for common diplomatic usage; a fourth letter was added for plurals, participles and genitives. When encoding the plural, genitive, or participle of a 2-letter word, the third letter would be placed apart in order to avoid confusion. Code symbols were prepared for principal countries and cities in the world, for states, major cities, and territories of the United States, and for proper names of men in English. A cipher table was to be used for those words not on the list. The first 74 pages of the code was the encode section, and contained the words in alphabetical order together with the code symbols; for example the very first word was Aaron with the symbol ABA, the last word of the first page was Acknowledge with a symbol of EA. The decode section (3- letter symbols) was not published in one sequential alphabet and was time consuming. Transmission of the code by cable was awkward because number of characters was not standard. It was not until 1876 that the 5 digit form became standard in the American ciphers. This code became the secret communication mask for American ministers in foreign legations in the years to 1876. Table 20-16 is a chart of the number of encoded lines sent from American ministers in seven major nations using this code.