Un système de signaux maritimes: Saussure’s Example of a Visual Code

Saussure regarded nautical flag signalling codes as a prime example of semiological systems. They have something in common with language (e.g., the principle of arbitrariness) but also some distinctive features (for instance, visual signifiers are unlike auditory ones in being able to deploy different features at the same time, such as shape and colour). The best-known example is the International Code of Signals (ICS). The system as it was during Saussure’s lifetime is explained with reference to his scattered observations, such as that a signal flag is merely “ un morceau d’étoffe ” until it is brought into use according to the prescriptions of the code shared by its users. Some commentators at the time envisaged this form of visual communication as a “universal language”. However, it is based on a restricted code, lacking the semantic universality of natural language. Each flag represents a letter, and combinations of flags represent predefined wordings in a “code book”. Such visual codes therefore depend on the prior existence of natural languages into and out of which they can be translated.

Saussure's observations regarding visual codes are frequently overlooked, but there are some intriguing references in his lectures and notes to un système de signaux maritimes obtenus au moyen de pavillons 1 de diverses couleurs-"a system of maritime signals based on coloured flags" (Écrits,p. 54;Writings,p. 34).These references are well worth exploring since he chose to refer to this nautical code as a prime example of a semiological system, having something in common with language but also exhibiting some distinctive features of its own.
Flags have been used for centuries as a form of visual communication in daylight beyond shouting distance.There are three systems (Woods, 1968, pp. 76-77).In the first two, signals are made with an extended arm (or arms) at three different angles: the best-known system is two-flag semaphore while the "wig-wag" system uses the positions and movement of a single flag (Myer, 1868).In both of these handheld flag systems the flags serve primarily to highlight the various positions.However, it is the third type, the shipboard flag hoist system, which is the subject of our current concern.Despite their importance in maritime history, flag-hoist codes are noticeable by their absence from most standard histories and encyclopedias of communication.The best-known example of such a system is the International Code of Signals (ICS), which is still in use today among seafarers, and it seems likely that this was the primary example that Saussure had in mind.Since this code has been largely ignored even by semiologists, it will be necessary to provide a little more information about it here than might normally be expected in this journal.
Prior to 1857 (when Saussure was born) there was no uniform system employed by vessels of different nationalities for communicating with each other at sea.Multiple competing non-secret codes appeared between 1814 and 1855, the most famous being Captain Frederick Marryat's Code of Signals for the Merchant Service (1817), based on numbered flags and a corresponding "vocabulary".On 2nd July 1855 the British Board of Trade appointed a committee to report on a uniform system, and a Commercial Code of Signals for All Nations was issued in April 1857 (see Figure 1).The flags agreed upon took the form of one burgee (swallow-tailed), thirteen square (or rectangular) flags, and five pennants (tapered flags, one of which was the Code Signal and Answering Pennant).Many of these were of Marryat's design, though now flags were labelled with letters (from B to W) instead of numbers.Vowels were omitted because of a Victorian concern that "objectionable" four-letter words might otherwise be generated unintentionally or by lewd sailors (Hulme, 1897, pp. 138-139).The Committee wrote that "Too much importance should not . . .be set upon the objection which naturally occurs, that the alphabet is thus incomplete, and that the power of spelling is apparently lost; for it should be understood that the letters are not used as letters, but as signs, characterizing the different flags by the most familiar method and in an order well known" (Cornewall-Jones, 1898, pp. 360-361).The codebook included a "vocabulary" of words, phrases, and sentences, arranged alphabetically by keyword, alongside what I refer to here as the "flag-letter" combinations to be used for each.The code was renamed the International Code of Signals in 1869 or 1870 (Mead, 1953, p. 181).This originally British code was translated by the governments of France, Denmark, Holland, Sweden and Norway, Italy, Germany, Austria, and Portugal and was also adopted by the USA, Denmark, Russia, Greece, Spain, and Brazil (Larkins, 1872, p. xii).
In 1887 a committee was again set up to bring the Code up to date.A revised edition was issued in January 1889 but after extensive international consultation the committee's final report in April 1897 proposed a radical revision (Cornewall-Jones, 1898, p. 365).Time was allowed for it to be translated into other languages, and it was brought into force, as scheduled, on 1st January 1901, the old one being used concurrently until the end of that year (Perrin, 1922, p. 188).The new Code consisted of twenty-six flags-one for each letter of the alphabet (fear of vowels now seemed quaintly old-fashioned), plus the Code Flag/Answering Pennant (see Figure 1).Combinations of two flags were reserved for "urgent and important" signals; "general" signals involved three flags; four flags covered information such as vessel identification numbers and place names (Hydrographic Office, 1907, p. 3).Much was made of the fact that the number of signals that could now be made using up to 4 flags in the same hoist was now 375,076, compared to 78,660 in the old Code (Cornewall-Jones, 1898, p. 366).Although most of the flag-letter combinations of the earlier Code were retained, the revised system meant that the pairings of flag combinations and messages in the 1901 codebook were entirely different from those in the earlier codebook.For instance, for the signal "cannot understand the message", the corresponding flag-letter sequence in 1857 was DPN, whereas in 1901 it was YHU Daniel Chandler (Larkins, 1866, Part II, p.46;Hydrographic Office, 1907, p. 423).(1857 and 1901).In this illustration of the "flag alphabet" what each flag stands for appears on its immediate left.The accompanying dates refer to the publication year(s) of the codebook in which these particular correspondences applied.The key to the actual colours used is on the bottom right.
Ferdinand's maritime references might have been prompted by conversations with his younger brother Léopold, who served as a French naval officer from 1882 to 1899 (landlocked Switzerland being unable to afford him such an opportunity).Ferdinand's first known reference to maritime flag-hoist codes is in his own notes for a lecture in March 1885, where he mentioned "the language of the signals in use in the navy" as an example of a "system of signs"-a reference that he then crossed out (Joseph, 2012, p. 322, citing AdS 374/1).Saussure's manuscript "On the Dual Essence of Language" is dated 1891 by the French editors and 1895 by Tullio De Mauro, the editor of the Italian edition-though John Joseph (2012, pp. 380ff.)notes that it incorporates ideas from a decade earlier.In it, Saussure declared that "a comparison between the language system and a system of maritime signals based on coloured flags, although rather approximate, may be fruitful in a number of ways" (Écrits,p. 54;Writings,p. 34).On 16th November 1908 he suggested to his students that the system of maritime signals was analogous to the language system, according to the notes of both François Bouchardy and Albert Riedlinger, and both Bouchardy and Émile Constantin note that these systems are of the "same order" (Engler 1, p. 47; 2ème Cours, p. 9).By the time of the Third Course, according to Constantin and George Dégallier on 4th November 1910, maritime signals (noted as an instance of visual signs), were the prime example Saussure gave of one of the "semiological institutions" among which language "must be classed" (Engler 1, p. 47; 3ème Cours, p. 9).On this limited basis, Saussure's semiological perspective appears to have become increasingly inclusive.
Saussure was interested in "elements that are common to langue and to other semiological systems" as well as in "whatever differentiates langue from other semiological systems" (Écrits p. 288;Writings,.He insisted that "language is merely a specific case of the theory of Signs" (Écrits,p. 220;Writings,p. 154).Like language, the maritime signal code is "a system of signs that express ideas" (Cours, p. 33; Course, p. 16).All "systems of expression" are based on convention (Cours, p. 101; Course, p. 68).The fundamental properties of language-most notably the principle of arbitrariness-are shared with other semiological systems based on social conventions, including maritime signalling codes.Following the American linguist William Whitney (1827-94), Saussure noted that "language is a convention, and the nature of the sign that is agreed upon does not matter" (Cours, p. 26; Course, p. 10; cf.Whitney, 1887, p. 19).In 1907, Saussure likened the indifference of the material in the language system to that in the system of maritime signals, noting, according to Riedlinger, that the system would not change if the colours of the flags faded (1er Cours, p. 23).He wrote that "The phonatory act appears to be a necessary instrument, but in itself remains as inessential as the act of dyeing flags so as to give the impression of green, red, black, etc., in the case of maritime signals" (Écrits,p. 248;Writings,p. 176).Perhaps he should have paid more attention to his brother since green is not used in such flags, on the principle that "the colours must be quite unlike, so that they do not 'merge' at a distance" (Perrin, 1922, p. 166)-a differential principle that is of course foundational for Saussure.
The ICS was created to enable ships of all nations to communicate with each other using "certain arbitrary Signs having a universal signification" (Larkins, 1872, p. xii).Its signal flags are visual forms which have no intrinsic relation to what they signify (either flag-letters or wordings).It is a code in which "a particular sign has been chosen in preference to another by convention" (Vendryes, 1921, p. 7).In relation to signalling systems in general, Albert J. Myer, the founder of the U.S. Army Signal Corps, noted: "Any other signal symbols would have answered just as well for the letters, if we were as well trained to recognize them" (Myer, 1868, p. 36).Like written letters, the signal flags are arbitrary graphic signs which signify only as part of a system.While distinct configurations are a fundamental requirement in such systems (ibid., p. 16), in accordance with Saussure's principle of arbitrariness no configuration is any more appropriate than another for what it signifies.The presence or absence of the same shape or colour in any two different flags in the same code does not carry any interpretive significance.Any connotations of shapes or colours are entirely irrelevant, such as where the design incorporates the shape of a cross or happens to be a reversal of the colours of the tricolore.The various configurations serve simply to differentiate between forms, unlike modern road traffic signs (Krampen, 1983), where Saussure would have noted in the pictograms a "rudiment of a natural bond between the signifier and the signified" (Cours, p. 100; Course, p. 68).Such a bond can actually hamper the effectiveness of visual signs as what are widely termed (in unSaussurean usage) "general symbols" (Price, 1969, Chapter 9).We read in the Cours that "Signs that are wholly arbitrary realize better than the others the ideal of the semiological process" (Cours, p. 101; Course, p. 68).It is not difficult to see why Saussure might have instanced maritime signalling codes.
Communication based on discrete, arbitrary signs (as with both language and flag signalling) depends upon a high degree of systematization, though "language is not a mechanism created and arranged with a view to the concepts to be expressed" (Cours, p. 122; Course, p. 85), whereas the ICS is of course deliberately designed, its rules being wholly explicit.Even "conversational turn-taking" is highly formalized.
In making a signal, a ship first hoists her ensign with the code flag under it, and if necessary the distinguishing signal of the vessel or station with which she desires to communicate.On seeing this signal the ship (or station) addressed then hoists the "Answering Pendant [pennant]" (i.e. the Code flag) at the "Dip", that is, some little distance below its position when hoisted "close up" to the block at the masthead or yardarm through which the signal halliards are rove.The first ship then hoists her own distinguishing signal, consisting of the four letters appropriated to her name, and then proceeds with the signal she wishes to make.When the first hoist is noted down and translated in the ship receiving the signal, this ship hauls the answering pendant "close up" to show that the signal is understood and keeps it there until the signalling ship has hauled that hoist down; the answering pendant is then again lowered to the "Dip" until the next hoist is disposed of, and when the ship signalling has finished, she hauls down her ensign to indicate that the message is at an end.(Perrin, 1922, p. 188) Systematization requires prior agreement.It was suggested in the Cours that la langue "exists only as a sort of contract" for members of a community (Cours,p. 31;Course,p. 14;cf. Écrits,p. 103,p. 288;Writings,p. 68,p. 202).The conventions of the ICS were established by explicit agreement among representatives of its users.The construction or selection of messages is governed by the rules of the code and the "vocabulary" which it provides.As with language, making sense of flag signals depends on the users' knowledge of the conventions; users must learn the code.To interpret a signal, individual flags have to be correlated with a flag-letter, and code letter groups then have to be matched with the corresponding entry (a word, phrase, or sentence) in the codebook (see Figures 2 and 3).Because of the radically arbitrary character of signal flags, those without access to the relevant codebook cannot infer messages by simply observing the signals.The presence of particular combinations of flags in a signal offers no direct semantic clues.
In flag-hoist codes it is intended that the receiver's understanding of the signal should directly reflect that of the sender.This echoes a reductionist understanding of communication as identified in the "conduit metaphor" underlying everyday usage (Reddy, 1979).In contrast, despite the famous diagram of the circuit de la parole in the Cours (designed to distinguish parole from langue), Saussure does not claim that linguistic communication is like employing a code book with clearly defined rules for encoding and decoding messages.For Saussure, "There is no such thing as a form and a corresponding idea; nor any such thing as a meaning and a corresponding sign" (Écrits, p. 42; Writings, p. 24).In the Cours it is noted that "[a] trait that distinguishes language from all other semiological institutions' is that its entities are not immediately 'given'" (Cours, p. 149; Course, p. 107).Language is far more than a communication system (Ellis, 1993).Maritime signalling codes were not mentioned by Saussure in this context, but as Robert Godel notes, the contrast is that "language must be coextensive with thought, which entails the negative limitation of the signified" within the language system (Godel, 1957, p. 225).What the signal flag "represents for the mind", is based not on "what it is" (a piece of coloured fabric) but on what it is not: that is its difference both from a) "the other signs displayed at the same time" and from b) those "that might have taken its place, and the place of the accompanying signs" (Écrits,p. 54;Writings,p. 34).This general semiological point was later developed in the Third Course in relation to language.Famously, dans la langue il n'y a que des differences ("in language there are only differences") (Cours,p. 166;Course,p. 120).The significance of linguistic units and maritime signal flags alike depends on both syntagmatic relations in praesentia and associative relations in absentia (Cours,p. 171;Course,p. 123; see also Constantin's notes, 27th June 1911, in 3ème Cours, p. 131).The shapes and colours of the patterns signify only with reference to the system of which they are part and the context of the signs with which they are combined.
The flag is merely "un morceau d'étoffe" until it is brought into use according to the prescriptions of the code shared by its users.Thus, according to the Saussure of the Écrits, signs are created in an act of parole.The idea of a "form divorced from its use", he writes, would be a "complete absurdity" (Écrits,p. 31,Writings,p. 15).For those whose only encounter with Saussure is via the published Cours, often criticized for failing to account for language in use, this emphasis may come as a surprise.The "Dual Essence" drafts of 1891-1895 had begun to account for parole in a way which he promised to undertake in the Course-a promise sadly frustrated by his untimely death.As Harris puts it, in the Écrits, he "comes a step closer to Wittgenstein's view of use as a more fundamental semiological notion than either form or meaning" (Harris, 2003, p. 245;cf. Chandler, 2014 re. Peirce).
In his Third Course Saussure suggested some key differences between maritime signals and language.For instance, maritime signal flags are visual signifiers which are unlike auditory ones in being able to deploy different features at the same time (such as shape and colour).They are not dependent on the distinctive linguistic principle of the linearity of the acoustic signifier, according to which auditory units are presented in temporal sequences, as a "chain" (Cours, p. 103; Course, p. 70; see also Constantin's notes, 2nd May 1911, in 3ème Cours, p. 77, andDégallier's notes, Godel, 1957, p. 129).As Roman Jakobson notes (1963, p. 336), the perception of auditory signs is successive while that of visual signs is simultaneous.This is related to the affordances of the medium: "Signals are of two kinds: transient and permanent.They are transient when each sign disappears as soon as it has been completed; as in signals by Motions or by Sounds.They are permanent when the signs are long in view; as when flags are kept hoisted to be read; or the symbols are written on paper" (Myer, 1868, p. 15).Flag signals are in this respect closer to a writing system than to speech.
Another point of contrast with language made in the Third Course is that whereas everyday language is "open to all" (Constantin, 4th November 1910, in 3ème Cours, p. 8), the maritime signalling code is a restricted code-established by a particular group for its own purposes.The ICS was often described in this period as a "universal language".However, Saussure was reported as saying that "The prescriptions of codes such as maritime signals are restricted to a limited group of people who utilise them for only a limited period" (Cours, p. 107; Course, pp.73-74; see also Dégallier's notes, in Engler 1, p. 164).This point should be taken in conjunction with the observation that "for the realization of language, a community of speakers [masse parlante] is necessary" (Cours,p. 112;Course,p. 77;cf. Écrits,p. 334;Writings,p. 238).One mariner treated this issue not as an unavoidable difference but as a rallying call for promoting an international code, readily accepting that "the value of any [non-secret] code is, like that of a language, enhanced precisely in proportion to the number who use it" (Maury, 1860, pp. 741-742).This was indeed a vision shared by the secretary of the Board of Trade committee, the editor of the original published codebook: "The Commercial Code can, by translation, be made an Universal Language of Signals" (Larkins, 1866, p. x).
However, the Code is restricted in another sense.Whereas language is a generalpurpose system serving diverse functions, signalling systems are designed as pragmatic, technical communication systems, operating within a restricted universe of discourse: a semantic field limited to "stereotyped sentences such as are generally wanted in maritime discourse" (Anon, 1875, p. 239)."A signal-code . . .applies only to a small number of precise and technical ideas, that is unchanging ideas, which have been established by agreement among men of the same calling" (Vendryes, 1921, pp. 321-322).Signalling systems focus on information, and primarily represent commands, questions, and responses."A signal-code cannot be generalized" (ibid., p. 322).It is not designed to have the "semantic universality" of language (Greenberg, 1968).
Saussure noted that language involves "never-ending change from one moment to the next" (Écrits, p. 88; Writings, p. 60).The system is modified over time by parole (Cours,p. 138;Course,p. 98;cf. Écrits,p. 95;Writings,.To some extent, so is the Code: a contemporary commentator noted that "The International Code of Signals . . .has been continually undergoing revision in a small way . . .; but in the process of time the blanks in the Signal Book became filled up" (Cornewall-Jones, 1898, p. 365).The codebook is thus like a dictionary.Linguistic change is not based on any "act of will" (Constantin, 4th November 1910, in 3ème Cours, p. 9), whereas the ICS is, of course, regulated.Saussure noted, according to Constantin, that maritime signal codes can be (and are) changed by governments (Second Course 1908-9, Engler 1, p. 47).Whereas there is "no sudden general change" in language (Constantin, 19th May 1911, in 3ème Cours, p. 95), the change in the ICS in 1901 constituted a radical disjuncture.
As we have seen, a signal is a sequence of flags which stands for a sequence of letters which in turn stands for a wording (typically phrasal), at which point natural language enables the system to evoke a concept.This raises the issue of "articulation" which later became a key concern for structuralist theorists who were inspired by Saussure's theories (Martinet, 1949).In 1933, the German psychologist Karl Bühler (1879-1963) chose to illustrate linguistic articulation using the example of maritime flag signalling, asserting that whereas language is a "two-class system", a flag code is a "one-class instrument" (Bühler, 1933, p. 139), and that "no articulation whatever of the signal meaning appears in the perceptible signs" (Bühler, 1934, p. 83).Bühler based his example on a version of a special subcode of "Distant Signals by Shape", but his points can be applied to the main Code.He noted that the flags have no meaning in themselves but only in combinations.Although in the Code a few individual flags can be used as signals in their own right, almost all signals do consist of a combination of flags (bearing no relation to linguistic syntax): these "cannot be analysed into smaller elements having a meaning" (Cantineau, 1952, p. 15).Within this system, individual code letters have no semantic significance.
Bühler notes that in the special subcode, precoded sentences constitute the only class of significant units.The editors of the 1931 edition noted that translation into seven languages would have been easier if the Code had consisted of "complete sentences only" (Hydrographic Office 1933, vol. 2, p. ix).A sequence of word-based signals is highly inefficient and poses challenges for translation because of "the varying way in which words forming a sentence are arranged in different languages" (ibid., p. xi).Indeed, when it was used for communication between allied ships in the First World War, "It was found that, when coding signals word by word, the occasions upon which signalling failed were more numerous than those when the result was successful" (ibid., vol. 1, ix).As we have seen, messages in the Code are selected from a set of predefined alternatives which the system is used to transmit.However voluminous the codebook, a flag code does not have the linguistic potential of generating an unlimited number of messages.Consequently, unlike language, such "one-class" systems are not well-adapted to novel situations.The Code-makers were well aware of this: "It is impossible to anticipate and include more than a very small part of the sentences and phrases that are required in actual practice" (ibid.vol.2, p. ix), The ICS has been compared to a writing system (e.g. by Joseph, 2012, p. 387).In the Cours it is noted that there are two types: phonetic and ideographic.Phonetic writing systems include those which are alphabetic: (loosely) "based on the irreducible elements used in speaking" (Cours,p. 47,Course,.Bühler argued that individual signal flags are "elementary marks precisely in the way that the phonemes of a language are" (Bühler, 1933, pp. 138-139;cf. 1934, p. 83).Their function is similarly differential (Cantineau, 1952, p. 15). Harold E. Palmer (1877-1949), an English linguist, wrote rather fancifully: "Let the groupings of this code of maritime signals be made pronounceable-surely an easy task for the expert-and the result will be a complete artificial language" (Jacob, 1947, p. 13)-an assertion that Saussure might have challenged since he was sceptical about artificial languages (Écrits,Writings,.The ICS has been described as primarily ideographic (e.g.Pettingell, 1875, p. 11;Couturat, 1912, p. 488;Vendryes, 1921, pp. 321-322).In ideographic writing systems, "Each written sign stands for a whole word and, consequently, for the idea expressed by the word" (Cours,p. 47,Course,.The flag signals represent wordings rather than sounds.This leads to substantial codebooks, so it is hardly surprising that mariners have never been expected to memorize the "general vocabulary" (Anon. 1908).Saussure's scepticism was warranted above all because, unlike a language, "flaggish" is not only unpronounceable, it is "unthinkable": it cannot be used for the organization of ideas but only for communicating messages.
Although subsequent structuralist discourse made much of language as a "code" (Jakobson 1952, p. 224), only two sentences in the Cours refer to language as a "code" (Cours,p. 31,p. 47;Course,p. 14,p. 25), and there are no such references in the Écrits.As we noted earlier, in an unpublished manuscript Saussure retracted his only known reference to maritime signalling codes as a "language".They are "codes" in the usual everyday sense, and although they exhibit some features in common with language, they depend on features that are not shared with natural languages.Foremost amongst these is the prior existence of natural languages into and out of which they can be translated (raising the issue of the role of language in sustaining other sign systems)-a dependency to which Saussure chose not to refer, despite the related point that writing "exists for the sole purpose of representing" language (Cours, p. 45; Course, p. 23; see also Constantin's notes, 6th December 1910, in 3ème Cours, p. 41).Saussure clearly wished to emphasize that the visual code of maritime signals is a semiological system having both features in common with language and distinctive features of its own-a stance which distances him from the subsequent structuralist quest to subordinate all modalities to that of language.His own priorities as a linguist did lead him to overlook the relevance of material features in such a "non-linguistic" system, and he did not mention all of the differences noted here, but (lest this assemblage of his scattered references might suggest otherwise) he was of course seeking to shed light on the "incalculably more complex" system of language (Écrits,p. 220;Writings,p. 154) rather than setting out to offer a semiological account of this intriguing but academically neglected system of visual communication.

Figure 1 .
Figure 1.Code Flags and Pennants in the International Code of Signals(1857 and 1901).In this illustration of the "flag alphabet" what each flag stands for appears on its immediate left.The accompanying dates refer to the publication year(s) of the codebook in which these particular correspondences applied.The key to the actual colours used is on the bottom right.

Figure 2 .
Figure 2.An example of a signal and an extract from the ICS codebook (1901).The signal (L), read top-to-bottom, represents a code-letter sequence (see Figure1) that corresponds to a codebook entry (R).

Figure
Figure 3. Encoding and decoding signals Saussure's references to a maritime signal code are most fully developed in his "Dual Essence" manuscript.Here he wrote that in language, "the vocal figure in itself signifies nothing"(Écrits, p. 73; Writings, p. 48), and that "a form is a vocal figure which is defined in the mind of the speaking subject"(Écrits, p. 49; Writings, p. 29).He illustrated how a configuration becomes a form in relation to maritime signals."Aflagfluttering on the ...mast amidst many others can be said to exist in two ways: firstly, as a piece of red or blue fabric [another reflection of Saussure's limited acquaintance with the code-or limited interest in materiality], and secondly, as a sign or object, taken to be invested with a meaning by those who perceive it".As the latter, a maritime signal flag "depends wholly on the thought attached to it" (Écrits, p. 54; Writings, p. 34; original ellipsis).A vocal figure becomes a form from the crucial moment at which it is introduced into the interplay of signs called langue ["le jeu de signes appelé langue"], in the same way that a piece of material lying motionless at the bottom of the hold becomes a signal at the moment that it is hoisted (1) along with other signs which are hoisted at the same time and contribute to some meaning; (2) from among a hundred other signs which could have been hoisted and the memory of which contributes no less to the . ..(Écrits, p. 38; Writings, p.  21; original ellipsis)