Commercial Telegraphic Code Books

I am compiling a list of as many commercial telegraphic code books as I can find. My data base is continually changing; if you have any suggestions, additions, or corrections, please let me know by email to reedsj@dtc.umn.edu

What is a ``code book''

Code books were used in the era of telegraphs (from 1845 until well into the second half of the 20th century) to shorten telegrams, which were paid for by the word. These books, arranged like dictionaries, would list many useful phrases or even sentences, each with its corresponding code word. One sent the code words, and the recipient of the telegram would have to look up their meanings in his copy of the code book. This could save quite a bit of money on intercontinental telegrams, since the price per word on undersea cable connections was very high. (The word cable means both the actual telegraph cable layed on the ocean bed, and to a cablegram sent over via the ``submarine telegraph,'' and then, as a verb, to send a cablegram, as in ``the arrest warrant for Mr Fogg was cabled ahead.'') I think one could expect to spend about a quarter of a dollar per word on a trans-Atlantic cable message in 1900. Code books were as much a part of ordinary business life as fax machines are nowadays. Secrecy was only a subsidiary motive for using code books, although a few code books of the era were specifically (if their titles are to be believed) written to provide privacy.

Many hundreds of different code books were published. The Library of Congress has about 1,000 different code books in its collection, and there are possibly as many again not in its collection. There were general-purpose code books, sold to the public. There were trade-specific code books, whose vocabularies were especially rich in terms and phrases useful in some particular trade. Some of these, too, were sold to the public, but others were kept for the private use of particular firms. Some code books were hardly different from a modern mail-order catalogue, offering a separate code word for each item available for sale. (And governments used code books to encode their official diplomatic, military, and naval correspondence with what might be regarded as extremely private specialized ``trade-specific'' code books.)

Until about 1905 the vast majority of code books supplied actual dictionary words (or artificial words) as their code words, such as ``Snatch = Sutter Packing Co., Yuba City, Cal.'' in the private code of the California Fruit Canners' Association, or ``Pascoela = Natives have plundered everything from the wreck,'' in the very popular general-purpose ABC Code, some supplied number equivalents -- sometimes instead of, and sometimes in addition to-- the code word equivalents.

But in the first decade of this century code books began appearing with code words which were meaningless and often unpronounceable fixed length groups of letters, usually of five letters. Thus, in a later edition of the ABC Code we have ``ewvgl = Pascoela = Natives have plundered everything from the wreck'' and so on.

These code books were compiled by specialists, who of course vied with each other to produce the better product, either from the point of view of resulting compression of telegrams, pertinance of vocabulary, ease of use, and (starting in about 1910) resistance to errors: how likely was it that one code word or code group could be mistaken for another.

Sometime between 1910 and 1915 code books began to appear (and after 1920, commonly appeared) with ``mutilation tables:'' Latin square charts expressing the formula for deciding which five letter groups were actually used in a given code.

The proliferation of code books is reminiscent of the current proliferation of computer software products. Like out-of-date software, when a given code book fell out of use it was likely to be thrown away.

Are telegraph codes still in use?

Codes are still in use to convey very specialized and formulaic information in such fields as weather reporting, stock market reporting, aviation flight control, and epidemiological field reporting. They are not in general use in Europe or America by the public, nor used for general (non specialized) communications. I have been informed that they were used in Telex communications between branch offices and the head office of at least one large bank, as late as the early 1970's.

In China, however, telegraphic codes are (or have recently been) widely used, since the usual writing system for Chinese is not directly usable with the telegraph. But as the telephone, the fax, and email supplant the telegraph in China, telegraph codes will fall out of use there too. See my brief web page on the subject, and my attempt at a Postscript reproduction of the Chinese telegraph code.

I have also heard of the use of code books for use with telephone pagers, where messages should be compressed to a few digits.

What is a ``code condenser'' and a ``cable register''

Code condensers were also used during the era of code books. If ordinary code books are methods of compressing linguistic text, then code condensers are methods of compressing digit sequences into code words. Using charts, formulae, and book-sized lists of numbers and words they were schemes for re-expressing sequences of digits as code words. These digit sequences might be code numbers looked up in a numerical code book, or actual numbers (such as prices and amounts) occurring in an otherwise uncoded message. I have a very poor idea about the actual use of code condensers in practice.

Cable registers or cable directories listed ``cable addresses'' much as a phone book lists phone numbers. Entries would also list code books held by particular addressees, so one could avoid sending code telegrams to recipients lacking the proper code book. In the first decades of this century there was a number of such cable registers, published annually, usually covering a the whole world or at least a whole country, but as the century progressed the number of cable register publications decreased, through mergers and through attrition. The Marconi's International Register, still published annually, stopped listing codes after the 1972 edition.

What is my method? How good is my coverage?

My list is based on library catalogues, which are often searchable via the Internet. The Library of Congress classification for code books is ``Cipher and Telegraph Codes,'' which is sometimes nowadays mistakenly used as a synonym for ``Cryptography'' and for ``Morse code.''

I exclude books whose titles are listed in Oriental languages, because I cannot read them. I exclude books about Morse code, and modern specialized codes, mostly aviation and weather codes. I include a few old specialized code books, but make no attempt to be inclusive.

Most of my items come from the catalogue of the Library of Congress, mostly from the L of C's ``PREM'' database of pre-MARC records. The next largest set of items comes from the National Cryptologic Museum. (Unfortunately the NCM collection is as yet only partially catalogued, and to a varying standard of detail. I have several incomplete printouts of the catalogue, but it is sometimes hard to recognize identity of a book appearing in each.) Then from the various large university libraries. A few smaller university libraries have surprisingly good collections of code books: Kent State University and the University of Rochester, as do a number of medium size cities: Seattle, Cleveland, and Kansas City.

I have only started to search the British Library.

What is recorded in my data base?

I record the usual bibliographic information commonly found in American library catalogues (title, author, time and place and agent of publication, language, approximate size and pagination) which in most cases is sufficiently detailed that given a book one can decide which of several descriptions best matches it.

I have also tried to include some information about the code itself: if the code words are ``words'' (i.e., variable length pronounceable letter sequences) or fixed length code groups; in the latter case, their lengths and whether of letters or digits. If a mutilation table is present I note that.

I record locations of copies of the book, with an unsystematic notation.

My database is a flat ascii file. Each book entry consists of a block of contiguous lines in the file. Successive entries are separated by empty lines. Each entry is a sequence of tag / value pairs, which I call items. Each item takes up one or more successive lines in the file. The tag is the first word on the first line, followed by a colon (:) and a tab character. The remaining data on the first line of an item belongs to the value of that item. Subsequent lines start with a tab character, and the data after the tab is continuation value data. The value data used TEX formatting conventions.

Here is a list of all item types:

Entry
Unique identifier for database entry.
Author
Author or compiler of the code book.
Client
In the case of commissioned works, typically private code books, the name of the commissioner. This is often a corporate author and often the publisher of the book.
Title
Title of book.
Ed
Edition.
Pub
Publisher, or, when I'm lazy, all the publication information: place, publisher, publication date.
Place
Place of publication.
Year
Date of publication.
Distrib
Distributor. Often a code book will have a code-book-distributor's label pasted on the title page, obscuring the publisher's name.
Desc
Physical description of book: number of volumes, pagination, and size.
Lang
Language(s) the book is written in.
Code
Notes about the code. Uses abbreviations: MT means ``includes mutilation table,'' W means ``code words,'' 5L means ``5 letter code groups,'' 4F means ``4 figure (i.e., digit) groups'' and so on. Also, prose description of any interesting features about the code itself.
Note
General note about the book.
LCCN
Library of Congress Card Number
CallNo
Call Number.
Micro
Description of a microfilm or microfiche copy of the book.
Loc
Location of a copy of the book. (Name of a library or personal collection.) A notation like NCM:13b means National Cryptologic Museum; the 13b refers to a private notation I made on one of several incomplete NCM catalogue printouts. Locations marked with "\mu" refer to microfilm or microfiche copies.
Nb
Note about a particular copy of the book. Always follows a Loc item which names the book in question.

Problems and Shortcomings

I do not follow my format perfectly.

So far I have only searched libraries in the Anglophone world, and even there am missing many important libraries.

I have lost or garbled data in copying it from library catalogue to my database.

I have seen only a tiny fraction of the listed books, and have no real control over the accuracy of my list.

I am at the mercy of how librarians in the last century have interpreted the ``Cipher and Telegraph Codes'' subject term.

I have not, for instance, come across Baravelli's code book in any of my searches.

Two code books in particular, The Adams Cable Codex and Bentley's Complete Phrase Code appeared in so many editions and reprints, and for different reasons are so poorly described in modern library catalogues, that it is very hard to tell how many actual printings and editions are present. My list currently has dozens of items titled Bentley's Complete Phrase Code which represent something like (this is a wild guess) 15 or 20 actual distinct editions and printings. Until I examine these books I cannot straighten out the bibliographic pigsty they currently occupy in my database.

References

Here is a brief reading list, with annotations.
  1. David Kahn. The Codebreakers, (Macmillan, 1967) chap. 22. This has a very good overview.

  2. William F. Friedman, Report on the history of the use of codes and code language, the international telegraph regulations pertaining thereto, and the bearing of this history on the Cortina report. USGPO, 1928. This is a very detailed treatment of the same material.

  3. William F. Friedman and Charles J. Mendelsohn, ``Notes on Codewords.'' American Math. Monthly, 39 (1932) 394-409. A mathematical study of error detecting capability of telegraph codes; has some examples of mutilation tables.

  4. Rudolf Schauffler, ``Ueber die Bildung von Codewoerter.'' Archiv fuer Elektrischen Uebertragung 10(1956) 303-314. A study of mutilation tables, both theoretically and as applied in a number of (unidentified) commercial code books.

  5. Gustavus Simmons, ``How Good is the Acme Code?'' 4th Joint Swedish-U.S.S.R. International Workshop on Information Theory, Visby, Sweden; Supplementary volume to the Proceedings, 1989. Asserts (misunderstanding Kahn) that the Acme code was the ``earliest example of an error detecting and correcting code.''

  6. Alexander Barg, ``At the Dawn of the Theory of Codes.'' Mathematical Intelligencer 15(1993), No. 10, pp. 20-26. Contains a summary of Simmons's and Friedman and Mendelsohn's papers.

  7. Tom Standage, The Victorian Internet, 1998. This fascinating book contains a very readable synthesis of material from Kahn and Friedman (1928).

Acknowledgments

Many thanks to these people, who have kindly helped me in my search for information:
Last modified 7 May 2007.

Jim Reeds reedsj@dtc.umn.edu