mcvey home
telegraphic codes and message practice



Is there anything in it
No, not anything in it
Or anything near it

ex John Hartfield, New Leviathan Cable Code (New York, 1896)

Telegraphic codes were among the technologies and practices that emerged to get electric communication to work from around 1870 well into the 1930s, and even beyond. In form, they were dictionaries whose design enabled:

  1. efficient assembly and encoding of messages at source — the coded version being significantly shorter than a verbatim or natural language message could be;
  2. packaging: safe transmission of the code (or sometimes mixed code/plaintext) message through an error-prone system thanks to built-in redundancy and resort to rhyming dictionaries and terminals lists, mutilation tables, or check-sum or other operations;
  3. secure transmission of sensitive information (secrecy); and
  4. efficient decoding of messages at destination.

The relative importance of these objectives would depend on the user. It might be said, however, that secrecy was ordinarily a secondary objective; it was message compression and protection that were crucial to getting cable telegraphy to work in the first place. Telegraphic communication, particularly overseas, was expensive. Messages were charged by the word — 25 cents per word between London and New York, by 1890, and higher between points further afield than those two communications hubs. The economies afforded by letting one code or cipher word stand for many, made it possible for the telegraphic transmission of time-sensitive market and business intelligence and correspondence to institute itself deeply among a broad range of users.

Users of telegraphy fell into four categories: governments (including military); the press; business; and individuals. Roughly speaking, each category of user paid different rates — governments the most preferential rates, businesses the least. Press rates were subsidized by other users, and indirectly by the government (in England, for example). Among the few means available to business users for reducing their communication costs was to reduce the number of chargeable words in their messages (they might also send messages at off hours, or take advantage of deferred service, of course). The cable codes were the primary means to reduce wordcount. The telegraphic code dictionaries served trade and industry above all, although personal social codes were also used.

The codes did not appear out of nowhere. Their immediate tributaries were nautical and coastal signal systems (typically involving flaghoist combinations), and visual telegraphs such as the Chappé optical telegraph system that operated across France from the mid-1790s well into the 1840s. With all of these systems, each sign or combination of signs represented a word or phrase in a code dictionary. (I might add that the expression telegraph generally — but not always — referred to land or coastal communication, with cable referring to long-distance communication through ocean cables.)

Early telegraph codes like Rogers (1846) derived directly from these signal codes (Rogers was designed to work both with visual and electro-magnetic telegraphy). But as electro-magnetic telegraphy spread geographically and into diverse types of business, new codes appeared at an accelerating pace. During the period 1870-1900 in particular, as cable routes expanded and telegraphy permeated further and further, the codes left their signal code predecessors behind in terms of powers of expression and compression.

The subject domains of the codes ranged from the highly general (even encyclopedic) at one end, to the highly specialized at the other — goldsmithing, jewelry, watchmaking and allied trades; fingerprint description; meteorological and astronomical observation; chess; horse racing; india rubber and gutta percha; shoemaking; sodas (including ginger ale and effervescing kola; shipping and nautical; finance; produce; social codes for domestic communications (e.g., office to home via sixpenny telegrams); coal; chemicals; lumber; police; travel; patent law; general legal; railroad; metaphysicians; motor manufacturing; insurance; hops; sugar; farm machinery; cotton; Black Sea and Mediterranean trade; fireworks; ostrich feathers; butter and general provisions; theatrical; cinema; oil well tools and accessories; missionary and ecclesiastical; medical; dental supply; biscuits; dairy; eggs; wine; magazine distribution; bookdealing; textiles; grain; paper; tailoring; scrap metal; mining; steel; engineering; and indigo.

A reminder: the codes are thematic dictionaries of phrasal and other selections; each code, therefore, presents what its compiler believed to be at least the minimum set of expressions that would be required by its specialist user.

The codes competed on various mixes of convenience, depth and/or breadth, error-resistance, reputation of compiler, novel system, even naturalness of their expressions (for the so-called verbatim codes).

Telegraphic communication offered speed — better than the posts, at least until airmail — and offered a written record — unavailable from telephony. But it could be frustrating. Rates were high; garbling (mutilation) of messages in transmission common; messages passed through too many hands, increasing the likelihood of error. It would take many decades for the technologies to become available that would ultimately increase capacity and dependability, and put telegraphy at the businessman’s fingertips (to borrow an expression from Donald Murray). These included loaded (inductive) cables; time-division and carrier current (harmonic or frequency) multiplexing; thermionic valve (vacuum tube) and other devices for signal regeneration, shaping and processing; synchronous and start-stop (asynchronous) printing telegraphy; and automated switching systems. Each of these and other pertinent technologies followed its own timeline. There were bottlenecks; advances in one had to wait on or were necessitated by advances or halts in another. But eventually, improvements in these other technologies — and most importantly, their integration in a record communication system — led to the retirement of the codes. But eventually, improvements in these other technologies and, most importantly, their integration in a record communication system, led to the retirement of the codes. Like old telephone directories and software manuals — like the Morse operator himself — the codes vanished.

In physical form, the codes ranged from small cards and pamphlets, to enormous compilations of thousands of pages. Their conceptual and typographic arrangements embraced both simple lists and elaborate sequences of cipher-building tables with five or more semantic orientations. These made it possible to express a commercial transaction, or specifications in an order for butter or a locomotive — or both — in one or a few codewords or five- or ten-letter ciphers.

So, for example, from John Hinrichs’s The Baltimore Export Cable Code (1896), one might select a two-part code word from a table for grain, as follows:

addag / at shillings 35 / 1/2
iduce / 17,000 quarters No. 2 Red Winter Wheat, ship next month

yielding the 10-letter codeword addagiduce, meaning:

Firm Offer 17,000 Quarters No. 2 Red Winter Wheat at sh. 35 / 1/2 full cargo per steamer to Cork for orders, ship next month.

The code dictionaries were designed to support both the (1) decoding and (2) assembly and encoding of messages.

For their decoding function, the codes were a specialized kind of bilingual dictionary. For the most part, they arranged words in the unknown language alphabetically, with plaintext meanings arrayed to the right. The recipient of a code message would look up the coded words, or deconstruct its cipher component from tables, and rebuild the plaintext message.

For message assembly, the codes organized phrase selections in thematic groupings. The sender of a linguistic message would have an intention, not too rigidly expressed. It need only be a list of items that needed to be communicated. The sender might start at an alphabetically arranged thematic index, and move on to the most promising thematic section. He would then search through the selections for those that most closely and efficiently expressed the intended message, pursuing cross-referenced leads where these promised a better selection. In a sense, the soft message would take or stiffen into its specific form, from out of the available selections. Next to each expression would be a codeword or cipher component; it was the assembly of these that would yield the message that was actually sent, for decoding at the receiving end.

The transaction, the trading relationship, and what the author knew the receiver could understand, were crucial contextual aspects of the process. That context was a shared encyclopaedic background knowledge without which telegraphic communication could not work. [Bear in mind too, that the total all-in cost of telegraphic communication included the preparation and decoding of messages, so that the arrangements of the codes needed to convenient.]

With the codes, one wrote (translations) to read, and read (from the available phrases) to write.

Code and cipher messages, then, are quite different from the style one associates with press dispatches — short, subject-verb-object assertions where each sentence stands out as distinct as a brick. The skeletal dispatches themselves would be fleshed out by copyeditors at the head office. It is indisputable that press telegraph style worked its way into literature and popular speech; its most prominent exemplar is probably Hemingway, whose eliptical constructs of simple words and pure existence suggests a rhetoric-degree-zero ethic. Individuals’ telegraphic style, to the degree they did not resort to codebooks, also was eliptical and artificial in this way. But building code and cipher messages by selection was a different kind of writing.

Instructions for using codes frequently encouraged users to thoroughly acquaint themselves with their arrangements. A typical suggestion is found in the preface to the 280-page private Leoni Code (1920) —

We wish to point out that in order to understand and be able to use intelligently any code-book, it is necessary to give some time to careful study. We find that ten minutes a day given to the study of our code-books helps materially in the construction of cables and telegrams, as new descriptions and phrases are continually being brought to light.

I wonder how one might have studied a code. Ten minutes a day, perusing such a dictionary to no purpose other than to become familiar with its selections, seems almost license to daydream or write poetry.

With this in mind, go back to 1846 and Henry J. Rogers his Telegraph Dictionary, and Seamen’s Signal Book

This is the place
This is
This place
This port
This was
This will

Did a bored signal officer ever, on some long uneventful voyage, read Rogers or a signal code predecessor aloud, in the wrong way, and strangely realize that it seemed, somehow, newly right?

On the code side, one encounters pronounceable word codes in English and various mixes of other European and pseudo-European languages; root-and-terminal and cipher-building codes; two, three, four, five and eight-letter codes; six, ten, twelve and thirteen figure codes; and combinations of these. Included among or related to the codes are various mutilation tables and error-detection tools, condensors for rendering multiple-digit figure codes into single pronounceable or euphonious words, prospectuses of codes, root-and-terminal selections, official vocabularies, lists of words suited and unsuited to code use, and even works of literature disguised as codes and as compilations of (uncoded) telegrams.

The other embodiment of the codes and of telegraphic practice, of course, is the telegram. Vestiges of code and cipher messages reside in cable register books that were maintained as a matter of course by firms that employed telegraphy, and remain still in business archives.

The codes interest me for several reasons, including these :

  1. link between human communicator, and telegraphic infrastructures
    They were the software that in some ways mediated between the sender and recipient of the message, at one level, and the physical and administrative infrastructure of telegraphy at others. This mediation was manifested in, for example, the different ways they addressed transmission error (long redundant codewords, mutilation tables, avoidance of certain letter sequences that were susceptible to transmission error; their evolution vis-à-vis rulings by the International Telegraphic Union on word counts, allowable dictionary and cipher words, etc; and their evolution from long (redundant) codewords to compact five-letter ciphers, as improved cable technology and terminal equipment (e.g., repeaters, amplifiers) affected error rate and carrying capacity (and thus, rates). The codes shed light on, and can be understood as a reflection of evolutions in, these infrastructures.
  2. window on communication and commercial practices
    The codes permeated trade and commercial practice, and sought in their vocabularies to anticipate any likely message. They are a window into the communications practice of many industries and professions.
  3. lexicographic and coding dimensions
    The codes are a special genre of dictionaries, in which phrase entries must be distributed methodically for ready use. Each dictionary has its own pragmatic norms for arranging its phrases and phrase components and, at a time that alphabetical arrangement of topics was at its height, represent an older, topical, tradition in lexicography. The code dictionaries offer models of semantic coding of terms, and what I call "semantic" data compression. Their use of tables harkened back to John Wilkins his Essay toward a Real Character and a Philosophical Language (1668), and other tabular representations of taxonomic discourse (Slaughter 1982).
  4. cryptography lite
    The codes were designed primarily to compress messages. Some noted cryptographers (Yardley, Mendelsohn) moonlighted on or (Friedman) wrote about the codes, but secrecy was rarely a foremost stated aim.
  5. relationship to other linguistic developments
    By regulating the terms available for telegraphic expression, the codes are analogous to other developments in language in their time. Various attempts by the International Telegraphic Union to regulate telegraphic expression, capped by the Official vocabularies for pre-concerted language that appeared in 1894 and 1899-1900, paralleled various attempts at a linguistic hygiene when language was widely viewed as being in crisis: meanings of words were seen as proliferating and shifting in a state of linguistic anarchy. The codes coincided with the work of C.S. Peirce, Victoria Lady Welby, C. K. Ogden, Saussure, Paul Valéry, Wittgenstein.
  6. writing by selection
    The codes distilled language to telegraphic vocabularies — sequences of related phrases and sentences, made available as linguistic molecules for new uses and (creative) misuses.

With some codes, assembly of a message by selection involved following a set procedure. This might involve an ordering of topic / predicate / object, such as is described by Frederick McCutcheon in the introduction to his Telegram Formula and Code Combiner (1885). Or it might involve a procedure for describing a suspect or fugitive from justice, which is quite similar in the arrangements of five different police codes ca 1895-1911 : instruction (i.e., be on the look out for); name; offense; age; height; weight; complexion; peculiarities, habits and walk, etc., etc.

The BEAMA Code (1922) explicitly draws attention to how its arrangement guides the construction of messages that are thorough and error-resistant:

As regards accuracy the method adopted practically forces both composer and receiver of a message through a sequence of Items in any Section until that Section is finished. As each Section is complete in itself, the inclusion of all essential matter in the final message is ensured. This is of great importance in technical cables, where the omission of an Item would mean the omission of essential data. The Rules will often compel an otherwise careful engineer to consider whether or not such-and-such necessary data have been inadvertently omitted. Few cables arrive without mutilation. Letters omitted from messages sent by this Code can generally be detected by inspection, because the sequence of vowel-consant-vowel consonant is interrupted; but, to assist further, a check system (pages xii. and xiii.) is given...

Their codewords — e.g., abababal in the eight-letter Bauers code; der neue deutsche telegramm-schlüssel (1913) — echoed paralleled dada language. The codes coincided with the writing of Kurt Schwitters and Hugo Ball, August Stramm, Gertrude Stein and, toward their end, Samuel Beckett. It was their accidental poetry — e.g., chaos = But I do not advise it, in the Adams Cable Codex (10th edn., 1896) — that first hooked me on the codes.

a look back

I wrote this several years ago; it shows my early enthusiasm for the codes. I see edits/expansions dated 10 May 2001, 18 June 03, 20 July 06, 24 September 06, and 18 February 07. Another is overdue.