Japanese Telegraph Codes

The present article is a much abridged version of an article in Japanese. Later additions to the Japanese version may not have not been incorporated herein.

Telegraphy in Japan began when service started between Tokyo and Yokohama in the second year of Meiji (1869), when Japan was embarking on modernization by opening the door to the western world. Telegraphic codes are mainly for reducing the telegraph cost by representing words and phrases with code words (typically consisting of two or three characters). Although some authors highlight confidentiality (from telegraph operators) provided by such telegraph codes, they were not intended for absolute secrecy as diplomatic codes. See another article for early Japanese diplomatic codes and ciphers.

Table of Contents

Variety of Codes (with Some Notes on Japanese Language)

Earliest Japanese Telegraphic Codes

Iroha Codes

Period up to the Sino-Japanese War (1888-1895)

Period up to the Russo-Japanese War (1896-1904)

Late Meiji and Taisho Period (1908-1921)

Late Taisho and Early Showa Period (1922-1936)

Around World War II (1938-1952)

Western-style Codes

Bank Codes

Variety of Codes (with Some Notes on Japanese Language)


Telegraphic codes were called denshin-ango, denshin-fugo, denshin-ryakugo, etc. but these terms were also used with different meanings. In particular, ryakugo may refer to common abbreviations used in telegraphy such as "henma", "hema", "hen" or the like to represent "henji wo matsu" (wait for reply). (Japanese is notorious for turning just about anything into abbreviations such as pasokon (personal computer), tanabota (Tana kara botamochi., a windfall) to Burapi (Brad Pitt).) Some of the codebooks with ryakugo in the title are more directed to such abbreviations than codes. Japan Tourist Bureau's Denpo Ryakugo Shu (1935) distinguishes such abbreviations and code words and use both.

Ango (with a long o) is a broad term, which may mean "cryptography", "code/cipher", or "code word (group)". It should be noted that there is another word "angŏ" (with a short o) for specifically referring to a code word or code group. Other than this, the transcription used in the present article does not generally distinguish long and short vowels. (Thus, ryakugo may represent either of two words, one with a long o and the other with a short o. For this case, the meaning is pretty much the same.)

The earliest codebook described herein uses fucho, the term commonly used to refer to a scheme for obscuring price etc. by substituting kana characters for digits.

Kana Code

Code words in telegraphic code for Japanese typically consist of two or three katakana characters. Often, "wi", "o", "we" were not used because they have the same pronunciation as "i", "wo", "e", respectively, which results in a total of 45 katakana characters. Some codes reserves "wi", "o", "we" (in particular the first two) for special purposes.

(The Japanese writing system consists of hiragana, katakana, and kanji. Hiragana (Wikipedia) encompasses every syllable occurring in Japanese. So is katakana (Wikipedia). (Hiragana and katakana are collectively called kana (Wikipedia).) Theoretically, any Japanese sentence can be expressed with hiragana alone but usually some terms or word stems are written in kanji (Chinese characters) and foreign words/names are written in katakana. It should be remembered that while today hiragana is the basic kana, in the Meiji era, katakana was preferred. Thus, kana used in telegraphy refers to katakana.)

Voicing/Semi-voicing Marks

With kana, voiced consonants are represented by dots (voicing mark), as opposed to having different letters for voiced and voiceless consonants as in the Latin alphabet. That is, a kana for "ka" with a voicing mark (two dots to the upper right) represents "ga".

Voicing of "h" is treated in a peculiar way in Japanese. In Japanese phonology, "b" is treated as a voiced version of "h" rather than "p", which in turn is treated as a semi-voiced version of "h" for historical reasons. (Briefly, the ancient "p" developed into "f", then into "h". Thus, while "ba" is represented with a kana for "ha" with a voicing mark, "pa" is represented with a kana for "ha" with a small circle to the upper right, a semi-voicing mark.)

In orthography, the voicing mark or semi-voicing mark is used with only those kana for which the mark is relevant. Thus, "ma" is never given a voicing or semi-voicing mark. However, some telegraphic codes used such combinations in order to increase the number of possible code words in two-kana code. Such voicing/semi-voicing marks used in combinations not allowed in orthography may be referred to as unorthographical voicing marks herein.

It should be remembered that the voicing or semi-voicing mark is by itself counted as one character for charging.

Three-kana Code without Voicing/Semi-voicing Marks

It appears the first to employ entirely three-kana code without voicing marks was Asano's Denshin Fugo (1899). Tsuji's Denshin Fugo Roku (1891) was basically a two-kana code but included three-kana code words with the second character limited to either "wi" or "o" (which were not used in other code words). More or less similar schemes were also used in Gifuken-yo Denshin Fugoroku (1894), Yoda Kanichi, Shoyo Denshin Ango (1898), Doi Katsuma et al., Teikoku Kumiai Denshin Fugo (1898), and Satomi Otokichi Denshin Ango Jisho (1899). Use of limited characters for the second place allows identification of three-character code words when mixed with two-character code words. One interesting scheme employed by Watanabe (1901) employed both two-kana and three-kana code words but the three-kana code words never began with the two characters of two-kana code words. (While Kobayashi Keishi, Nihon Denshin Ango Jibiki (1896) had 473=103823 three-kana "regular words", it was not a usual code for representing plaintext with code words but was merely a listing of numbered code words.)

It was not that the three-kana system prevailed after Asano (1899). It was only after 1920 that three-kana code became the mainstream. (Even after that, voicing marks were used by some codebooks such as Ichiyanagi (1929).) Considering this, the 1906 edition of the Iroha Code may be said to have been one of the earliest to part with the voicing marks.

Although the three-kana system could provide much larger vocabulary than two-kana code, if a vocabulary as large as 100,000 is not required, two-kana code including some kana with voicing marks could provide shorter code words in average.

Absence of Word Breaks in Japanese Orthography

As noted above, when code words of different lengths (e.g., two-kana/three-kana) are used, there must be some means to identify individual code words because word breaks are not shown in Japanese.

Code Words Employing Figures

Apart from voicing/semi-voicing marks, figures were sometimes used to increase the number of patterns. For the private (i.e., non-government) sector, use of figures in code was allowed in the 1890s. The 1898 edition of the Iroha Code was one of the earliest to adopt code words including figures.

Of the figures 0-9, "2", "3", and "8" were often (but not always) excluded from use in code words because kanji for these figures look similar to katakana "ni", "mi", and "ha", respectively.

I-RO-HA Order and A-I-U-E-O Order

Today, the usual order of kana is the A-I-U-E-O order (Wikipedia), which is based on a systematically constructed syllabary and has its origin in Sanskrit. In the old days, the I-RO-HA order was more commonly used, which is based on a pangram poem, which begins "Iro-ha nihoheto ..." (Wikipedia).

While most early telegraphic codes employed I-RO-HA ordering, occasionally, kana with voicing/semi-voicing marks, which do not occur (in writing) in the Iroha poem, were arranged in the A-I-U-E-O order. For kana without voicing marks, (after somewhat special instances of Iida (1896) and Kobayashi (1896),) Tsuchiya (1909) employed A-I-U-E-O ordering. Some codebooks including Denshin Ango (1908) and Terada's Denshin Ango (1918), employed A-I-U-E-O ordering only for code words and employed I-RO-HA ordering for plaintext words/phrases. Ikeda (1889) is exceptional in that the first character of two-letter code words is arranged in A-I-U-E-O ordering and the second character in I-RO-HA ordering.

(Among dictionaries of Japanese, Otsuki Fumihiko's Genkai (1889-1891) (Digital Library from the Meiji Era) and Jirin from Sanseido (1907) (Digital Library from the Meiji Era) already employed A-I-U-E-O ordering.)

A beer company's Denshin Ryakugo (1938), Nakayama's Wao Shoji Denpo Ango (1938), Nihon Chisso Denpo Ango Cho (1941), Nihon Denshin Ryakugo Sho (1943), and Hayashi's Anzen Denshin Ango (1944) all employ A-I-U-E-O ordering. These coincide with the period when Latin-letter code words were beginning to be given along with the kana code words.

Either way, for the plaintext, the arrangement took account of the written form in kanji. For example, ones with the first character being kana came before those with the first character written in kanji, with those having the least strokes being the first among them.

Some codebooks used alphabetical ordering of the plaintext according to its romanized form in Latin letters (Japan Tourist Bureau (1935), Ichida (1941)). Such ordering of Japanese words according to romanization is not limited to those codebooks but is also seen in, e.g., indexes including some (even only a few) English words/names in the field of English studies.

Two-Part Code

Several earliest codes employed a "two-part" system, having separate listings for encoding and decoding. Such a system was common in "secret codes" because of the random arrangement of words/phrases to prevent cryptanalysis. But it was not the case for commercial codes.

One rationale appears to have been that, as noted above, the ordering of the plaintext took into account the written form and was not completely according to the kana ordering. Eventually, it appears to have been recognized that the two-part system did not much contribute to the usability of the code. Those published after the first Iroha Code (1888), which still employed the two-part system, employed a one-part system, consisting of a single listing to be used both for encoding and decoding.

(Today, Japanese dictionaries (for Japanese users) have headwords written in kana, followed by any kanji form. This eliminates the complexity of ordering by taking into account the kanji form. Such headwords in kana were already employed in an early Japanese dictionary Jirin (1907) (Digital Library from the Meiji Era) but there were more complications. In the traditional kana orthography employed in the school education until the end of World War II, the relation between pronunciation and writing in kana was not straightforward. Thus, Jirin had to include an index according to pronunciation.)

Two-letter Differential

The principle of two-letter differential, whereby any two of the code words have at least two letters different from each other, was not employed for many decades after it became common in the 1880s in the West. Among the codebooks listed herein, only Nihon Denshin Ryakugo Sho (1943) and Hayashi's Anzen Denshin Ango (1944) employed the two-letter differential. Nakayama's Wao Shoji Denpo Ango (1938) employed it for alphabetical five-letter code words but not for kana code words. Tsuchiya's Konyo (1910) took care not to use the entire combinations so that if the two characters of a three-kana code word are determined, the remaining character is limited to one of about five candidates. Nakamura (1916) describes a plan of code for preventing transmission errors.

Earliest Japanese Telegraphic Codes

Telegraphy was introduced into Japan at the beginning of the Meiji Era. It covered major cities in about 1880 and the prefectural capitals in about 1890.

There is a manual for telegraphy from 1880, Nihon Teikoku Denshin Raiso Hitsuyo (Digital Library from the Meiji Era).

In 1882, in the aftermath of the Imo Incident (Wikipedia) in Korea, private telegraphs in code were prohibited from 22 August to 4 September. (Prohibition of use of code in time of emergency was also seen during World War I and II in the West.) A newspaper reported business circle's complaint about the ban. Telegraph code was already used commonly enough to arouse such complaints.

As described below, the Bank of Japan prepared telegraphic codes to be used with its business correspondents in 1885. When Nippon Yusen (Japan Mail Shipping Line (Wikipedia)) was established in 1885 by merging, it inherited codes from its predecessor. Thus, telegraphic code was already commonly used among major companies in the mid-1880s.

Aoki Shinkichi, Keiben Hitsuyo Denshin Fucho Go (1881)

(Digital Library from the Meiji Era)

The author knew of Western codebooks. This work was published with the permission of the Home Minister and the Telegraph Office. The author remarks elementary school children may practice encoding as a diversion during a penmanship class.

This is a two-kana code. The ordering appears to be adapted based on I-RO-HA ordering. In the image below, the edge of the book appears to be that of traditional Japanese binding (Wikipedia).

Since the code words do not quite follow the I-RO-HA ordering, the code words are listed in a tabular form in the appendix.

Forestry Office (of the Ministry of Agriculture [and Commerce]), Toin Bunrui Denshin Fugo Hyo(1882)

(Digital Library from the Meiji Era)

A single sheet listing two-kana code words in I-RO-HA ordering.

Construction Office of the Ministry of Agriculture and Commerce, Komukyoku Denshin Yakuji Hyo(1883)

(Digital Library from the Meiji Era)

A single sheet listing two-kana code words in I-RO-HA ordering. The legend is substantially the same as that of the Forestry Office above.

General Affairs Bureau of the Ministry of Justice, Denshin Fugo: reprinted in March 1891 (1891)

(Digital Library from the Meiji Era)

This is placed here because the preface is dated June 1888. It stipulates this code is used "except for cases in which confidential code should be used."

It employs a two-part scheme in I-RO-HA ordering: the first half is for encoding (p.1-31) and the second half is for decoding (p.33-61).

This is a two-kana code, including 2025 words in total. Three kinds of code titled "Exception", "Caution", and "Special" are included as supplements.

"Exception" code represents dates with code words having "1" or "0" as the first character of the code word.

"Caution" code uses "wi" and "o" to represent law courts and place names. It was named "Caution" because "wi" and "o" are prone to errors in transcription because they sound the same as "i" and "wo", respectively.

"Special" code uses "4", "5", "6", "7", and "9" as the first character to represent office and country names. ("2", "3", and "8" are not used because their kanji look similar to katakana "ni", "mi", and "ha".) It was named "Special" because use of digits in code words was only allowed for government telegrams at the time.

The front cover of this code book bears a writing "Abolished on 30 November 1909". What appears to have replaced it is Denshin Ryakufu: revised in November 1909 (1909) (Digital Library from the Meiji Era). Again, this is a two-kana code but had done away with designations such as "Exceptional", "Caution", and "Special."

Also issued by the Ministry of Justice is Denshin Fugo Hasshinnin Ryakufu: revised in June 1907 (1907) (Digital Library from the Meiji Era).

Matsuoka Narazo, Shoka Hitsuyo Jiyu Angŏ (1888.12)

(Digital Library from the Meiji Era)

A small book published in the same month as the Iroha Code below. This is not a codebook but deals with a cipher for substituting a kana for another kana. It claims to provide freedom (jiyu) to write any complicated message in cipher in contrast to telegrams or letters in code. Actually, it is no more than a cipher wheel, which was repeatedly reinvented in history.

Considering the remark that kana with voiced consonants may be represented by use of hiranaga rather than katakana, it may be mainly directed to application in "postcards" (p.4).

Ango Onshin (undated postcard)

Apart from telegraphy, I acquired an old postcard titled ango onshin (cryptographic communication). It is a handwritten cipher wheel. The text explains how to decipher. The sender is to write a starting letter in a field appended to the message field.

Iroha Codes

The Iroha Code went into many editions during the decades after its first publication in late 1888. The 1915 edition claims it had distributed several tens of thousands copies during the 26 years. It was named as a codebook relatively in common use among the few codebooks in Japanese (Nakagawa (1907) p.382).

The publisher Hayashi Tamiji was a relative (like a nephew in age) to Hayashi Yuteki, who established Maruzen for importing Western books and goods. At 15, Tamiji came to Yokohama in 1872 to help the business. He was engaged in compiling a catalogue of Western books in 1880 and co-authored Eiwa Gogaku Hitori Annai (Modern Conversations in English and Japanese for Those Who are Beginning to Learn the English and Japanese Languages without the Aid of a Teacher) (1887, 4th ed.) (Digital Library from the Meiji Era). Tamiji was credited as "editor" in the 1915 edition.

I-ro-ha refers to the first three characters of Japanese syllabary in the traditional ordering. Considering Tamiji's knowledge of Western books, he may have got his inspiration for the name from the ABC Code commonly used in the West.

Iroha-biki Denshin Ango (1888.12)

(Digital Library from the Meiji Era)

The code words are made of two kana, including those with voicing marks. A two-part system is employed, whereby separate listings are used for encoding and decoding. In the encoding part, dates, times, numbers and percentages, prefectures and place names, banks, companies, and merchants are provided in their groups for ease of looking up.

Zoho Saihan: Iroha-biki Denshin Ango (1895)

(Digital Library from the Meiji Era)

This is an enlarged edition published soon after the Sino-Japanese War (July 1894-March 1895), which brought about increase of the traffic of telegrams. It added more than 1000 terms and code words by using "o", "wi", "pa", "po", "pe", "pu", "pi", "vu", and "ko with a semi-voicing mark", which had not been used in the first edition.

It did not change the code of the first edition but simply made additions. Thus, the users of the first edition could continue its use. The addition could be distinguished by the dash as opposed to dots for separating the plaintext and code words.

Teisei-zoho: Iroha-biki Denshin Ango (1898)

(Digital Library from the Meiji Era)

This further made over 1000 additions by using "1", "3", "4", "5", "6", "7", "9", and "0".

Again, it made only slight corrections to the previous editions (mainly regarding ships, companies, banks, and stocks), which were marked with "*".

Shinpen Denshin Iroha Ango (1906)

(Digital Library from the Meiji Era, Google)

This completely new edition, now renamed "Denshin Iroha Ango" from "Iroha-biki Denshin Ango", switched from the previous two-kana code to three-kana code. It abandoned the rather old-styled two-part system and employed the one-part system as with many other codebooks.

Hayashi Tamiji, Zoho Shinpen Denshin Iroha Ango (1915)

(Digital Library from the Meiji Era)

An addition of more than ten thousand brought the total vocabulary to over thirty thousand.

It had a thesaurus-like index of terms classified by their meaning.

In late 1935, Maruzen entrusted compilation of a new codebook to a new editor. The result was Hayashi's Anzen Denshin Ango (1944) described below. (The new editor's surname Hayashi is the same as that of Hayashi Tamiji in sound (kana) but different in kanji.)

Period up to the Sino-Japanese War (1888-1895)

Ikeda Chugoro, Denshin Ryakujirin (1889)

(Digital Library from the Meiji Era)

The code words are made of two kana, of which the first is in A-I-U-E-O ordering and the second is in I-RO-HA ordering. The plaintext is in I-RO-HA ordering.

Tsuji Chufu, Denshin Fugoroku (1891)

(Digital Library from the Meiji Era)

Tsuji is also the author of Nisshin-jihen Denshin Fugo (1894) and Denshin Ango (1895, coauthor).

The code words are either two-kana (in which "o", "wi", and "we" are not used) or three-kana with the second being "wi" or "o." Voicing/semi-voicing marks are used even for kana not eligible according to orthography (referred to as "unorthographical voicing marks" herein).

Igarashi Mitsuaki, Denshin Fugo Jisho: Tokyo Tsushinsha Senyo (1892)

(Digital Library from the Meiji Era)

Two-kana code in I-RO-HA ordering. Unorthographical voicing marks are used.

What appears to have succeeded this is Kuroishi Tota, Denshin Fugo Jisho: Tokyo Tsushinsha Senyo (1896) (Digital Library from the Meiji Era).

The introduction is substantially the same but the new author appears to have made a complete revision. In the image below, the left and the right show the 1892 and 1896 editions, respectively.

Ehime Prefecture Police Department, Keisatsu Denshin Fugo (1894)

(Digital Library from the Meiji Era)

The code words are made of two or three kana.

The introduction is very similar to that of Tsuji (1891).

Gifu Prefecture, Gifuken-yo Denshin Fugoroku (1894)

(Digital Library from the Meiji Era)

The code words are either in the form of (i) three kana consisting of "0" and two kana or (ii) four kana with "wi" inserted between the two kana characters. The arrangement is in I-RO-HA ordering.

Use of "0" allows mixing the code words with plaintext. The "0" may be omitted if the code words are written between parentheses ( ). That is, "0" is an indicator of code words and this is substantially the same as the usual two/three-kana code.

Tsuji Chufu, Nisshin Jihen Denshin Fugo (1894)

This was compiled for reporting events about Japan, China, and Korea during the time of the Sino-Japanese War (1894-1895). The preface says it was compiled in a few days, taking entries from directories of government officials, maps, lists of war ships, etc.

The code words are made of two kana in I-RO-HA ordering. The section under each first character is divided into subsections of "government offices and places", "names", "place names", "ships", "miscellaneous and verbs." At the end, sections for "numbers and quantities" and "Chinese and Korean place names" are provided. The latter includes many entries in I-RO-HA ordering.

Code words in two kana are followed by those in the forms of "n+kana", "kana+n", "kana with semi-voicing mark + kana", "kana + kana with semi-voicing mark", "kana with voicing mark + kana", and "kana + kana with voicing mark". (In Japanese, the character "n" is somewhat special in that no word begins with it (as opposed to the sound "n" occurring in "na", "ni", "nu", "ne", "no") and that it does not appear in the I-RO-HA pangram poem.)

Shorin Benran (Hakubunkan) (1895.3)

(Digital Library from the Meiji Era)

The handbook for sales of books published by Hakubunkan includes a small telegraphic code (revised in March 1895) for representing some thirty words with one kana (p.78-79).

Osaka Meteorological Observatory, Osaka-fu Kisho Hokoku. Osaka Sokkojo Kisho Zappo (1895.8)

(Digital Library from the Meiji Era)

The booklet includes three pages of one- to three-kana code.

Tsuji Chufu, Katagiri Katsuhiko, Denshin Ango (1895.12)

(Digital Library from the Meiji Era)

This is similar to the same author's Denshin Fugoroku (1891) above in its vocabulary for journalism.

The code words are made of two or three kana in I-RO-HA ordering. The second character of a three-kana code word is either "wi" or "o". ("O", "wi", and "we" are not used in other contexts.) Unorthographical voicing marks are used. At the end, two-kana code words of the form "n+kana" or "n with voicing mark + kana" represent expressions about numbers.

Separate from the general code, two-kana code for appointments and another two-kana code for parliamentary affairs are provided. Their use is to be identified by the indicator "i" or "ro", respectively, at the beginning of a telegram.

Period up to the Russo-Japanese War (1896-1904)

In this period, code words including numbers and entirely three-kana code appeared.

Kashiwagi Kichisaburo, Jitsuyo Denshin Angŏ Hayabiki (1896)

Digital Library from the Meiji Era

This is a simple two-kana code (I-RO-HA ordering). The first character of the code word includes those with voicing marks (A-I-U-E-O ordering).

Matsushita Matsunosuke, Shokai-no Shishin (1896)

(Digital Library from the Meiji Era)

This is a booklet provided to customers by Kyuhosha, the first modern news agency in Japan established in 1886. It includes code for the rice market section, code for the stock section, and code for the section for money, foreign exchange, etc. Code words are made of one or two kana.

Iida Konosuke, Shin'an Ango-ho(1896)

(Digital Library from the Meiji Era)

The author explains various ciphers for keeping secrecy in mail or telegraphy. They are no more than primitive ones such as converting a kana into another, representing a kana with a number 1-47, representing a kana with a combination of two digits, or representing a kana with a symbol.

Among them is a sample of a telegraphic code (p.21ff.) Words beginning with "a" are represented with code words of the form "a + one or two kana", and so on.

Yoda Kanichi, Shoyo Denshin Ango (1898.3)

(Digital Library from the Meiji Era)

Two- or three-kana code (in I-RO-HA ordering). The first part includes phrases, conjunctions, names of goods, place names, ships, etc. The second part includes banks and companies, government bonds and stocks, time, percentages.

Three-kana code words have "wi", "o", "ru", "nu" in their second or third place. (There are few words beginning with "ru" or "nu.") These characters are not used in the beginning of code words or in two-kana code words, which allows distinguishing the border from the following code word. (In written Japanese, words are run into each other without any spacing between the words.) Code words beginning with "re" are reserved for dates in the year. Code words beginning with "n" represent prices in yen.

The code words with the first character "i", "ro", "ha", ..., "n" is followed by code words with the first character having a voicing mark, which represent banks and companies, government bonds and stocks, time, percentages, etc. Although the fist character with a voicing mark is in A-I-U-E-O ordering, the second character without a voicing mark is in I-RO-HA ordering.

Doi Katsuma et al., Teikoku Kumiai Denshin Fugo(1898)

(Digital Library from the Meiji Era (1898.10, Enlarged Second Edition)First Edition (1898.9, without front matter))

Two- or three-kana code (in I-RO-HA ordering). Unorthographical voicing marks are used.

Nihon Denshin Fugo Domei-kai, Nihon Denshin Fugo (1899.8)

(Digital Library from the Meiji Era)

The introductory guide is very similar to that of Yoda Kanichi, Shoyo Denshin Ango from the previous year.

The English title "The commercial telegraphic code" is a literal translation of "Shoyo Denshin Ango". The Preface in English includes an acknowledgement for "Mr. Yoshida", who appears to be Yoshida Masahide of the Ministry of the Communications, who is credited as consultant in Shoyo.

There is also an Enlarged Second Edition (1901) (Digital Library from the Meiji Era).

Satomi Otokichi, Denshin Ango Jisho (1899.5)

(Digital Library from the Meiji Era)

This is a two- to three-character code. Three-character code words include either of "0", "1", "4", "5", "7", "9", or a dot in the second place.

Use of numbers in code, which had not been allowed for private telegrams, was now allowable.

Asano Sugimatsu, Denshin Fugo (1899.7)

(Digital Library from the Meiji Era)

The author points out that use of voicing/semi-voicing marks to increase the number of code words leads to errors and complication in decoding and thus employs entirely three-kana code without voicing/semi-voicing marks. Both code words and the plaintext are in I-RO-HA ordering.

Roughly, the three-kana code includes 20 entries times 4 columns times 528 pages=42420 entries, which is much more than 14 entries times 3 columns time 344 pages =14448 entries in Yoda (1898) above.

The National Diet Library also has a revised edition (1903). The author no longer needed to explain the rationale for the three-kana code and simply states all the codes use three kana. This, as well as the first edition, avoids repetition of a character in code words. Thus, the code words "i-na-ne" is succeeded by "i-na-ra", rather than "i-na-na" that would have come in I-RO-HA ordering.

Tsuchiya Kenjiro, Denshin Ango (1899.8)

(Digital Library from the Meiji Era)

This is a relatively small two-kana code in 44 pages. The code words are arranged in I-RO-HA ordering, in which the first character varies for a specific second character assigned for particular categories. (The second characters "i" and "ro" are assigned to expressions about specimens, "ho" and "he" are assigned to expressions about order, etc.)

At the end, a simple cipher for substituting figures 0-9 in prices is provided. This is to prevent a customer from seeing the wholesale price.

Watanabe Sansaku, Jitsugyo Denshin Angŏ: Ryakusho: Jitsugyo Angŏ (1901)

(Digital Library from the Meiji Era)

The code words are made of two or three kana. Frequently used words are given two-kana code words. No voicing/semi-voicing marks are used. The two-kana combinations used for two-kana code words are not used for the beginning of three-kana code words, allowing clear distinction between two-kana and three-kana code words.

At the end, code words made of "one figure + one or two kana" are provided for representing numerals with one to three figures beginning with the figure.

Sanshi Denshin Ango (1903)

This is a code dedicated to silk trade, which was an important industry for Japan during the Meiji and Taisho periods (1868-1926). This may be said to be a Japanese counterpart for cotton codes in the West.

The code words are made of either (i) two kana; (ii) three kana including "0" at the second or third place, or (iii) three kana with the second having a voicing/semi-voicing mark. I-RO-HA ordering is observed.

Asano Sugimatsu, Seiro-eki Senji Denshin Fugo (1904.5)

(Digital Library from the Meiji Era)

This is a codebook compiled for wartime communication during the Russo-Japanese War (1904-1905). I-RO-HA ordering is observed.

As with the same author's Denshin Fugo (1899), code words are made of three kana without voicing marks.

Komatsu Keigo, Jiyu Tsushin Denshin Ango (1904)

(Digital Library from the Meiji Era)

This was published by Jiyu Tsushin Sha, a news agency.

The code words are made of three characters including kana and figures. Kana in I-RO-HA ordering follows figures "1"-"9".

Uetsuka Otokuma, Denpo Fugo (1904)

(Digital Library from the Meiji Era)

This provides three kinds of code: military code, common code, and gazette code, all in two characters.

The code words for the common code represent two or three alternative plaintext words but the author claims the context would provide disambiguation. The common code and the gazette code use figures as well as kana.

Late Meiji and Taisho Period (1908-1921)

Nihon Denshin Ango Kyokai, Denshin Ango(1908)

(Digital Library from the Meiji Era)

The code words are made of two kana or two kana plus a figure. Unorthographical voicing marks are used.

The code words are in A-I-U-E-O ordering, while the plaintext is in I-RO-HA ordering.

Denshin Ango Kyokai ("Telegraphic Code Association") is instituted. Members share a list of members to facilitate use of this common code among the members.

Tsuchiya Yoshiro, Konyo Denshin Ango (1909)

(Digital Library from the Meiji Era)

Konyo ("mixed use") means that the code words can be used with plaintext or code words from another codebook without need of parentheses or the like. Because of the limitation of patterns caused by this, there are three- and four-character code words.

In order to be distinguished from plaintext words, use is made of voicing/semi-voicing marks solely combined with those kana orthographically not allowed. Further, characters "ka", "so", "wa", "-" (length mark), "ro", "tsu" serve for distinguishing four-character code words from three-character ones. Code words containing figures belong to Part 1 (words and phrases related to date, time, etc.) and others belong to Part 2 (general expressions).

The patterns of code words are as follows. (In the following, "voiced kana" is a shorthand for kana with a voicing/semi-voicing mark.)

Part 1

Voiced kana + one or two figures

Kana + Figure +"n"

Part 2

Voiced kana + kana

Voiced kana + "ka"/"so"/"wa"/"-" + kana

Kana +"0" + kana

Kana +"0" + "ro"/"tsu" + kana

That is, the code words of this code can be recognized with the first character "voiced kana", the second character "0", the second and third characters figure + "n", with inserted "ka"/"so"/"wa"/"-"/"ro"/"tsu" indicating four-character code words. This seems to be quite complicated for many applications.

The arrangement is largely in A-I-U-E-O ordering.

For foreign communication, the code words may be transcribed with Latin alphabet, wherein the voicing and semi-voicing marks are to be represented by "b" and "p", respectively. (Otherwise, there is no way to represent "voiced m", "semi-voiced m" with the Latin alphabet or any phonetic symbols for that matter, because these combinations do not represent any sound in Japanese.)

Code words can also be represented with a five-figure number by combining a three-figure page number with a two-figure number assigned to each code word. Addition of one or two checking letters is also proposed.

This work was followed by the same author's Tokubetsu Shiyo Ango: Konyo Denshin Ango Heiyo (1910) the next year. Since the code words of the Konyo had distinctive forms such that they could be distinguished from plaintext, they can also be distinguished from code words from other codes as long as such other codes do not employ the same distinctive form. This 1910 work is to provide such a special private code ("Tokubetsu Shiyo Ango") with specialized vocabulary to be used with the Konyo. It is a three-kana code. The code words are basically in A-I-U-E-O ordering but plaintext words are arranged in classes.

While many codes simply use just any combinations of kana, this codebook appears to use only about five kana given specific first two characters. For example, it appears the only code words beginning with "a-e" are "a-e-ra", "a-e-a", "a-e-sa", "a-e-ni", and "a-e-ma."

For foreign communication, again, the code words may be represented with a five-figure number and it was proposed to combine two such five-figure numbers into a ten-figure group, which is to be sent as one word, as was the practice in the world since 1904.

Tetsudo Koshukai Henshubu, Tetsudo Gengyoin Hikkei (1912)

(Digital Library from the Meiji Era)

This handbook for railway staff includes some abbreviations ("ryakugo") in I-RO-HA ordering such as "i-ka-ku" for "ittojokyaku" (first class passenger) and "i-fu-da" for "itto josha kippu" (first class ticket) ("fuda" means "ticket") (p.312-324).

Sato Sonosuke, Denpo Ryakufu Annai-shu (1917)

(Digital Library from the Meiji Era)

The author points out that telegraph code is too complicated and its cost saving is not enough. He proposes here an alternative by representing some 4000 short sentences with both three kana (no voicing marks) and numbers. If either the kana or the number is mutilated, the other will serve the purpose. Although this increases the length of a telegram, its entries in the form of complete sentences will allow transmission within the initial charge (provided that the necessary sentence is found).

Terada Shigeteru, Tei-shiki Denshin Ango (1918)

(Digital Library from the Meiji Era)

This is a three-character code. A code word consists of an "initial symbol" (figure or kana with a voicing/semi-voicing mark) plus a "sound symbol" (kana in A-I-U-E-O ordering). When the initial symbol is made of a single figure, the sound symbol consists of two kana. When the initial symbol is made of two figures or kana with voicing/semi-voicing mark (counted as two characters in telegraphy), the sound symbol consists of one kana.

The plaintext is in I-RO-HA ordering. General vocabulary is followed by sections of classified vocabulary. Among the latter, Latin letters are assigned code words such as "9-9-a" (A), "9-9-i" (B), "9-9-u" (C), etc. When several code words with the same initial symbol follow one another, the initial symbol for the code words other than the first may be omitted.

Koizumi Kiyoshi, Tetsuzai Yoran (Koizumi's Handbook for Iron Dealers) (1920)

(Digital Library from the Meiji Era)

This handbook contains a three-kana code for article names and a two-kana code (consisting of a kana with a voicing mark (counted as two characters in telegraphy) + another kana) for common words.

The code words for articles are scattered in the data tables throughout the volume. The arrangement appears to be somehow based on I-RO-HA ordering.

The code for common words is in the appendix (TELEGRAPHIC CODE p.119-137 and TELEGRAPHIC CODE FOR NUMBERS p.138-139). The first character (kana with a voicing mark) is in A-I-U-E-O ordering, the other kana being in I-RO-HA ordering. The plaintext is also in I-RO-HA ordering.

Tsukahara Takashi, Teikoku Denshin Ryakugo (1921)

(Digital Library from the Meiji Era)

This includes a two-kana code for dates and a three-kana code for common vocabulary, both in I-RO-HA ordering.

Late Taisho and Early Showa Period (1922-1936)

From about 1922, a kana code without voicing marks appear to have become the mainstream.

Nippon Yusen Kabushiki Kaisha, Denshin Ango (1922)

Digital Library from the Meiji Era

This is a fifth edition of a code Nippon Yusen (Japan Mail Shipping Line (Wikipedia)) took over from its predecessor in 1885. It is a three-kana code in I-RO-HA ordering.

Kudo Hayato, Nihon Denshin Ango (1922)

(Digital Library from the Meiji Era)

The code words are made of three characters consisting of kana (I-RO-HA ordering) as well as figures and symbols (punctuation and length mark).

Each entry is also given a five-letter code word in Latin alphabet, which was then the mainstream of western codes. Since the international telegraph regulation stipulated the length of a code word should be at most 10, two such five-letter code words could be combined together for transmission as one word.

On the other hand, the two-letter differential is not implemented and there are many pairs of code words with only a single letter difference.

Mitsubishi, Wabun Denshin Ango (1929)

Mitsubishi's codebook for Japanese text. (Codebook switching indicators to and from a separate codebook for English text as well as other codebooks are provided in the front matter, as seen in the image below.)

The code words are made of three characters consisting of kana (I-RO-HA ordering) as well as figures. Each entry is also given a five-letter code word in Latin alphabet.

The code words are constructed according to four mutilation tables adapted for the new telegraph regulations adoped in 1928 such that two five-letter code words form a 10-letter telegraph word conforming to the "Category A" of the regulations (see another article). See another article for the mutilation tables of this codebook.

Ichiyanagi Seizo, Denshin Ango (1929)

This is a small two-character code that was originally made for personal use. The code words are made of kana (including those with unorthographical voicing marks) and figures.

Nakayama Hisami, Jitsuyo Denshin Ryakugo (1930)

The author had obtained a preface by the Undersecretary of the Communications Ministry as early as 1921 but almost all was lost in the Great Kanto Earthquake in 1923. The author points out secrecy as well as cost saving as the merits of a telegraph code and quotes a newspaper article of an accident whereby a post-office clerk leaked a secret telegram to a competitor.

The code words are made of three characters (kana, figures) in I-RO-HA ordering.

For foreign communications, it is proposed to convert the code words in kana to Latin letters with a table.

The author continued to be engaged in compilation of codes such as Wao Shoji Denpo Ango (1938), Nihon Tsusho Ango (1950), and Shoji Denpo Ryakugo: Jisa-shiki Niji-sei (1952).

Japan Tourist Bureau, Denpo Ryakugo Shu (1935)

(Digital Library from the Meiji Era)

This is a code for branches of Japan Tourist Bureau. It distinguishes but uses both abbreviations ("ha-ma" for Yokohama etc.) and code words. It incorporated entries from several railway abbreviation glossaries.

A code word is made of three characters: "5" or "6" + kana (the same as the first sound in the plaintext) + kana (I-RO-HA ordering). The arrangement is in alphabetical order of the plaintext in its romanized form.

An abbreviation is generally in kana but ship names ("... maru") are represented by "kana + kana + 0" (literally "maru" means a circle). Some abbreviations are not limited to three-kana (e.g., "shi-ku-ko-shi-a" for "Tokyo Shinjuku Mitsukoshi Annaijo").

Shionogi Denshin Ryakugo Cho (Shiono's Private Code) (1936)

This is a three-character code for the use of a pharmaceutical company (now Shionogi & Co. Ltd.).

The code words are made of kana and figures. The figures are only used for the first character and represent specific categories: time ("0"), dates ("1", "3"), quantities ("4", "5", "6", "7", "9"). ("2" and "8", which in kanji look similar to katakana "ni" and "ha", are not used. Apparently, the use of "0" below in the second place is an exception.) Apart from these, the second characters are assigned according to the categories: Shionogi's medicines ("i", "ro", "ha", ..., "ru"), Kitazato Lab's sera and vaccines ("n"), general medicine ("we", "hi", ..., "su"), words/phrases ("o", "wa", "ka", "yo"), amount of money ("0").

In the time section, the second character is either of "i", "o", "e" (representing a.m.) or "wi", "wo", "we" (representing p.m.), whose pronunciation is the same as that of "i", "o", "e", respectively.

Fusokyo Hito-no-michi Kyodan, Denpo Ryakufugo (1936)

(Digital Library from the Meiji Era)

This is a three-character code made of "figure + two kana" (I-RO-HA ordering) of a religious body (now PL Kyodan (Wikipedia)).

Around World War II (1938-1952)

From about 1938, codebooks appeared that assigned alphabetic code words as well as Japanese code words. (Kudo (1922) above was an early example of the kind.)

Hosoda Shoten [a trading company], Denpo Ango (1938)

(Digital Library from the Meiji Era)

This company's code dates back to about 1904-1905, when a code was shared between the headquarters in Kyoto and a branch in Tokyo. The copy in Tokyo was lost in the Great Kanto Earthquake of 1923, which caused inconvenience for many years until this code was made.

This is a two-kana code in I-RO-HA ordering.

For each code word, four alternative plaintext words/phrases are given in four columns. Thus, a recipient has to select an appropriate one from four candidates.

Dainihon Beer Co., Denshin Ryakugo (1938)

(Digital Library from the Meiji Era)

This is a three-kana code (in A-I-U-E-O ordering).

For foreign communications, it describes converting Japanese code words into alphabetical code words with a transcription table and transmitting them with five letters to each word.

Nakayama Hisami, Wao Shoji Denpo Ango (Japanese Commercial Code for International Use) (1938)

This includes a completely revised edition of the author's previous work Jitsuyo Denshin Ryakugo (1930) and includes 30,000 more entries.

In addition to three-character code words made of kana and figures, five-letter alphabetical code words as was the mainstream in the western codes were provided. The arrangement of the Japanese code words employed A-I-U-E-O ordering instead of the I-RO-HA ordering of the previous work.

The principle of two-letter differential, whereby every code word differs from any other code word in at least two letters, is employed for the alphabetical code words but not for the Japanese code words. At the end of the volume is included a mutilation table for alphabetical code words, which was then a common device in the Western codebooks for correcting transmission errors.

Hokuren, Hokuren Denshin Ryakugo (1940)

This is a three-character code from Hokuren (an agricultural cooperative) in Hokkaido. The code words are made of kana and figures. Some influences from Nakayama (1938) and Nakayama (1930) are seen.

Ichida Shozaburo, Showa Denshin Ango (Ichida's Japanese Trade Code) (1940)

(NDL Library Search)

This was published by Schofield & Co. in Kobe. Richard Schofield in Kobe published many codebooks in English at least from 1914 to 1936 (see another article). The following is based on the revised edition (Ichida's Improved Japanese Trade Code) (1941).

It provides three-kana code (in A-I-U-E-O ordering) and three-letter alphabetical code words for foreign communication. While five-letter codes were the mainstream at the time, some code compilers in the 1930s employed a three-letter system. The arrangement of the Japanese plaintext is in the alphabetical order of romanized forms.

The sections for ships and for articles use small letters for alphabetical code words (kana has almost twice as many characters than the Latin alphabet). Since Morse codes do not distinguish cases, the alphabetical code words requires an indicator such as "ZZH read the following one word from the Articles." or the like.

Each code word is assigned a checking number 1-26 for error detection. The sum of three checking numbers for three code words is converted to a letter according to a table, which is appended to the code words. In the alphabetical code, three code words plus one checking letter makes ten letters, which is the maximum allowed for code words according to international regulations. In the case of kana code, the check letter is a specific character such as "ku", "chi", "tsu", etc. which are not used in code words.

Mutilation tables are included. That for the kana code allows identification of the third character given the first two characters and a checking number. With this, a correct code word can be inferred for each of the cases in which any of the first to third character is assumed to be in error.

The mutiation table for the alphabetical code is similar to this (and different from the mutilation table for five-letter codes as employed by Nakayama (1938) and many Western codes).

Shokuniku Haikyu Tosei-ni-tsuite (1941)

(Digital Library from the Meiji Era)

This book from a stockbreeders' association lists some abbreviations for telegrams on p.49-57.

Musen Tsushin Gyomu Yogo-shu (1943)

(Digital Library from the Meiji Era)

This is a textbook from a predecessor of what is now The University of Electro-Communications. It includes various abbreviations and their applications.

Nihon Chisso Hiryo Corporation, Nihon Chisso Denpo Ango-cho (1941)

(Digital Library from the Meiji Era)

This is an inside codebook of what is now Chisso Corporation (Wikipedia).

It is a three-character code. Kana code words (in A-I-U-E-O ordering) are followed by "figure + two kana", "kana + figure + kana", and "'wa' + kana + figure" (the latter pattern is reserved for trademarks of Chisso). Code words beginning with "wi", "wo", "n" are reserved for later additions. Figure "2", which looks very similar to katakana "ni" when written in kanji, is also used.

An "error detecting table" is included. It lists for each character similar ones in writing and similar ones in telegraphy.

Nihon Denshin Ryakugo Sho (1943)

(Digital Library from the Meiji Era)

Japan was at war with the allies including Britain and the US since December 1941. The preface explains motivation of compiling the code by deploring there are no match for British Bentley's Second Phrase Code and American Acme Code in Japan and regretting the situation of "relying on British and American codebooks of the enemy nations".

This codebook provides 100,000 entries in three kinds of code: five-figure numbers, five-letter alphabetic code words, and four-letter kana code words. It employs two-letter differential. Although the Introduction refers to a mutilation table in the appendix, it cannot be found in the digital scan above.

Hayashi Atsunobu, Anzen Denshin Ango (1944)

The compilation of this work took nearly ten years after the author was entrusted with the task by Maruzen, the publisher of the Iroha Code, in about the end of 1935. The code employed modern features such as the two-letter differential and a mutilation table (called "analysis table"), though these were anticipated by Nihon Denshin Ryakugo Sho (1943) of the previous year. The word "Anzen" (security) refers to such error protection.

The author explains that telegrams in Japanese, thus far limited for domestic use, had become frequent with Manchuria and China, further extending to the entire "Greater East Asia Co-Prosperity Sphere" including Philippines, Java, and Burma. For use in the Asian countries and also for use with a kana typewriter, the plaintext was written entirely with katakana without the usual mixture with kanji. (Since Japanese has many homonyms, kanji forms are provided in brackets, as in many Japanese dictionaries.)

It had 100,000 entries including 87,841 general words and phrases (in A-I-U-E-O ordering). It provides three kinds of code groups: five-figure numbers, five-letter alphabetical code words, and four-kana code words. As used in this codebook, kana included Arabic figures "0", "1", and "2".

Since the two-letter differential is employed, the fourth character of four-kana code words may be considered as a check character. The analysis table, provided for error correction, may be used to determine a check character given the first three characters of a four-kana code word. One interesting application of such a checking system is sharing of a checking character by three (or more) code words. That is, three four-kana code words may be stripped of their fourth character. Then, the three dropped fourth characters are used as if forming the first three characters of a code word to find a checking character according to the analysis table. Thus, instead of three four-kana code words, one may transmit three three-kana code words plus one checking character. (Finding a check letter by forming a three-letter code word with three check letters of three code words is also used in Yamaguchi's alphabetical code, Oriental Improved Code (1937).

The analysis table may be used for providing a check character for any three-character code word, for example, one from the existing codebook such as the Iroha Code.

Jinjiin Denpo Ryakugo Hyo (1949)

This is a codebook of the National Personnel Authority.

Nakayama Hisami, Nihon Tsusho Ango (Japanese Trade Code for Universal Use) (1950)

This is the author's third codebook following Jitsuyo Denpo Ryakugo (1930) and Wao Shoji Denpo Ango (1938).

The author points out what few codebooks used in Japan had been reduced to ashes in the devastation of war.

The code words are made of three kana. Common words and phrases are in A-I-U-E-O ordering. In the sections from "A" to "TO", the second character begins with the kana "to", which was to prevent a bias in the second character.

The author switched the alphabetical code form the previous five-letter system to a three-letter system. As was typical with three-letter alphabetical codes, three code words were to be combined with one check letter to form a ten-letter code word.

Nakayama Hisami, Shoji Denpo Ryakugo: Jisashiki Niji-sei (1952)

While three-kana code had long since become the mainstream form of Japanese code, this codebook chose a two-kana system. Although the vocabulary is limited, the author points out its economy and ease of use. Further, the author proposes a new "self-checking" scheme.

Western-style Codes

In 1871, soon after the introduction of telegraphy in Japan, inauguration of submarine cable from Vladivostok to Nagasaki opened a way for telecommunication with Europe and further with America via Siberia. Connection between Nagasaki and Tokyo in 1873 made possible overseas telegraphy from Tokyo.

Use of Western Codebooks

Telegraphic codes listed above are generally kana-based codes for domestic use. For overseas communications, codebooks published in Britain or America were used.

As of 1935, the branches of a radio telegraphy service company had one or more of the following codebooks.

ABC (Fifth Edition Improved; Sixth Edition)

Acme Code

Bentley's Code (old and new)

Commercial Telegraph and Cable Code

Duo Code

Lieber's latest Code

Marconi International Code

Paramount Simple Check Three Letter Code

Rudolf Mosse Code (plus Supplement)

Schofield's Eclectic Phrase Code

Schofield's Safe-Check 3-Letter Code

Universal Trade Code

Western Union Telegraph Code

Yamaguchi's Oriental Self-Checking Three-Letter Code [1st ed. 1930, 2nd ed. 1931, Kobe, Japan]

Alphabetical Code Made in Japan

Among these, the Paramount, Schofield's, and Yamaguchi's are alphabetical codes published in Japan.

Apart from an early example of Kudo (1922), it was from about 1938 that Japanese codebooks provided alphabetical code words along with those in kana. Nakayama (1938), Nihon Denshin Ryakugo Sho (1943), and Hayashi (1944) employed five-letter alphabetical code words, while Ichida (1943) and Nakayama (1950) employed a three-letter system. These codebooks allow transmission of messages in Japanese to correspondents in overseas branches, for example.

Before these, Tsuchiya (1909) and Dainihon Beer Co. (1938) advise that kana code words may be written in romanized form for foreign communication. Tsuchiya (1909) also provides numbers for code words. Nakayama (1930) proposes conversion of kana into two-letter syllables with a dedicated table for this purpose.

Of course, apart from published codebooks, there were private codes for the use of specific companies. For example, as described below, the Bank of Japan compiled its own telegraphic code for overseas communication and sent copies to agents in London and New York. This was reportedly an extensive codebook including more than 140,000 entries. Many other private codes were small but efficient figure codes as explained below.

Figure Codes

There is a genre of codes, often called figure codes, which is different from an ordinary codebook that provides alphabetical code words for plaintext words and phrases. Rather than simply using figures instead of letters, figure codes pack various kinds of information by assigning to each digit (or group of digits) of a five- to thirteen-figure number some specific information such as articles, shipping, prices, etc. With such a figure code, just sending one number may convey a message which would require many words in plaintext.

Moreover, since the international regulations allow up to ten letters in one code word, various schemes were devised to convert such numbers into alphabetic code words. There were various kinds of such figure codes and conversion tables devised by Japanese.

Some of the books below describe such figure codes and conversion tables.

Ueda Kotaro, Gaikoku Kawase-to Denshin Ango (1895.8)

(Digital Library from the Meiji Era)

The section of telegraph codes of this book explains that Ager's and Whitelaw's codes, which use dictionary words within ten letters (see another article), were commonly used and gives detailed explanation of Ager's by quoting its tables. It also describes that when there are not so many articles, a simple code as below may suffice.

Abfold ... article A
Abfurm ... article B
Abcald ... article C
Affirm ... article D
am ... Buy 50
as ... Buy 100
ing ... Buy 50 at the quoted price
ous ... Buy 100 at the quoted price

Any one of "Heading" and any one of "Ending" may be combined into one word to be transmitted. For example, "Abfolding" means "Buy 50 of article A".

Ager's is a much complicated version of this. The first two figures of a five-figure number represent articles and the remaining three figures represent numbers/quantities, prices, and shipping (by using a table with a row for combinations of numbers/quantities and shipping and columns for prices). The five-figure number is converted into a code word by a dictionary which associates five-figure numbers and code words.

Although Whitelaw's code is not explained, it is a listing associating code words and numbers.

A checksum is also described, with a proviso "this method was popular before but now rarely used." Indeed, some western codebooks in the 1870s proposed use of check digits but the two-letter differential became the mainstream for error correction since then (see another article).

Kobayashi Keishi, Nihon Denshin Ango Jibiki (1896.4)

(Digital Library from the Meiji Era)

As the author's reference to Whitelaw and Ager suggests, this is not an ordinary codebook that represent words and phrases with code words. Instead, it lists three-kana code words with serial numbers (as did Whitelaw). It was used for translating into code words the number given by a separate codebook ("translation directory"), which gives numbers for plaintext words and phrases.

Although this took after Whitelaw in form, it did not employ the two-letter differential employed by many codebooks of Whitelaw. Use all three-kana combinations means that an error in any single kana turns the code word into another.

In addition to 473=103,823 "regular words" (three-kana combinations) (in A-I-U-E-O ordering), there are 17,672 two-kana auxiliary words including voicing/semi-voicing marks. Of these, 120,000 are numbered.

Tomoziro Imai, Cypher Code (1908)
Tomoziro Imai, Denshin Ango Kogi Yoryo (1910)

Cypher Code is a booklet (in English) of 36 pages and provides figure codes as well as conversion tables for converting numbers into code words. Kogi Yoryo (lecture resume) (Digital Library from the Meiji Era) explains the use of Cypher Code. Apparently, the author used it in the course at Waseda University.

Five figure codes and a five-figure conversion table ("Latin Combination Telegraph Code of 10,000 Cypher Words" over 2 pages) are provided. The conversion table is a simple table, whereby the first three figures of a five-figure number are converted to a root (e.g., 012 into "Abjut") and the last two figures are converted to a termination (e.g., 74 into "itam", 46 into "emus") and the root and termination are combined into a code word (e.g., "abjutitam").

Six figure code is also provided, which allows use of two digits to represent as many as 100 different articles. Although the five-figure conversion table cannot be used, the Berne Official Vocabulary (a numbered list of some officially accepted 200,000 code words compiled by the bureau of the International Telegraph Union; see another article) might serve as such a conversion table by its numbering. However, the most significant figure need to be either "0" or "1" because the vocabulary did not fully provide six figure numbers up to 999,999.

A seven figure conversion table is also explained, whereby "0156" is converted into "Avace" and "400" are converted into "erge", to be combined into "Avaceerge."

A ten figure code is also explained and Mr. Wait's ten-figure conversion table is demonstrated.

Cypher Code also includes "Latin Combination Telegraph Code of 2,000,000 Cypher Words", which has 200 times the number of entries as above but is contained in no less than six pages. The first four figures 0000-1999 are converted into a root and the last three figures 000-999 are converted into terminations. This may be similar to J. C. Siegfried, Latin Combination Telegraph Code of 2,000,000 Words (1894?, San Francisco), mentioned on the Web.

Combination of Latin roots and terminations was used in some Western codes such as Whitelaw's 14,400 Words (see another article). While Whitelaw used authentic combinations of roots and inflectional endings of Latin verbs of the first and third conjugations, Imai's combinations appear to include artificial ones. This was allowed because the revision of the international regulations in 1903 allowed artificial codes as code words as long as they were pronounceable.

Nakagawa Shizuka, Gendai Shogyo Bun Shishin (1907)

(Digital Library from the Meiji Era)

This handbook of commercial writing deals with telegraph ciphers in Chapter 5. It describes various kana codes, ciphers for prices, and Western codebooks.

It outlines the codebooks in Japan and observes that there were few codebooks in Japanese; the Iroha Code from Maruzen was relatively in common use; some of telegraphic codes for news agencies were more intricate but somehow not widely used; telegraphic codes for government offices did not have a large vocabulary and was not publicly available.

Nakagawa Shizuka, Shinsho Seikan (1916)

(Digital Library from the Meiji Era) This was somewhat revised the next year as Shokan Bun Seigi (1917) (Digital Library from the Meiji Era) and later extensively revised.

Volume 6 (p.681) describes telegraphy in Japanese, telegraph networks, telegraphy in Latin letters, etc. A 15-character condensed telegram (p.737) is proposed, which is similar to Western "figure condensers". Fifteen characters are used because the minimum charge for telegrams was 20 sen (0.20 yen) for up to 15 characters. First, various information is packed into 13 figures and a two-kana check is appended. Then, the 13 figures are each converted to kana by a simple substitution cipher. (It should be noted that simple substitution cipher can be used for turning figures into kana because every kana is pronounceable by itself, whereas in Western languages, simple substitution of each figure into an alphabetical letter is likely to result in unpronounceable sequence of letters.)

Nakamura Kanji, Gobyu Sokudan: Wabun Denpo Ango (1916)

(Digital Library from the Meiji Era)

The author points out the few public codebooks available do not pay attention to error protection and declares telegraph codes without regard to errors do more harm than good.

The author studied telegrams and examined which characters were likely to be mistaken for each other and, finding that the twelve characters "e", "o", "chi", "su", "na", "ne", "no", "ho", "mu", "mo", "n", and "1" were not likely to be mistaken for each other, proposed four- or five-character code by using only these twelve characters (or the eleven kana, with "1" reserved for punctuation). Such limitation of letters used for code was also proposed by Marquis de Viaris (see another article).

This book only shows construction of code words and does not provide plaintext assigned to them.

Shirai Kenji, Gaikoku Boeki Jimu Kenkyu; Fu: Ango Denshin Sakurei Oyobi Setsumei (Practical Knowledge of General Office Routine in Export House) (1920)

(Digital Library from the Meiji Era)

This provides (p.93-115) a twelve-figure figure code.

The twelve-figure number is said to be converted to a ten-letter code word with "condensers" such as Voller, Durand, Schofield, etc. for transmission.

Kurihara Ippei, Gaikoku Boeki Jissen (1920)

(Digital Library from the Meiji Era)

The handbook of international trade devotes Chapter 3 to "Cablegrams".

Bank Codes

Ginko Kenkyusha, Naikoku Kawase Jimu-no Kenkyu (1937)

(Digital Library from the Meiji Era)

This devotes a section to telegraph codes (p.101).

A telegraphic remittance code is typically a sheet of paper on which is printed a code table for encrypting the amount of remittance. It is often updated for security. Apart from this, a common telegraph code is typically a volume bound in cloth or full leather and is rarely revised. The author proposes standardization of the code, except for those parts which require absolute secrecy for amounts of money.

Ito Yusaburo, Naikoku Kawase Jitsumu (1926)

(Digital Library from the Meiji Era)

A section is devoted to security of telegraphic remittance codes (p.72). Misuse of of such code may lead to improper remittance. It gives examples of such incidents: acquisition of 30,000 yen by fraud at a Tokyo branch of an Osaka-based bank in 1923 and of 15,000 yen at a Yokohama branch of a Tokyo-based bank in 1925. Apparently, a "replay attack" was done.

In order to prevent such reuse of authentic encoded messages, there were provided ten different substitution tables numbered 1-10 for representing "1", "2", "3", ..., "thousand yen", "ten thousand yen" with a kana. Typically, in order to increase the number of patterns, distinction was made for Rounds 1-5. Thus, fifty kinds of sequence codes such as "Round 1 No. 1", ... "Round 5, No. 10" could be prefixed to the encrypted remittance amount.

Another table further employed redundant encryptions for the amount of money. That is, there was provided a second substitution scheme for numbers and the amount of money was to be encrypted in two ways.

A third table (see the image below) further varies the second substitution table depending on the month.

Bank of Japan, Nihon Ginko Enkaku Shi, Vol. 10 [1913]

(Digital Library from the Meiji Era)

In this voluminous history of the Bank of Japan, a section (p.325ff.) gives an overview of its codes.

Code was introduced soon after the establishment in 1881 of the Bank of Japan. It was enlarged/revised in 1883, 1885, and 1888 to 782 words + 50 blanks. In 1890, an extensive revision added some 1800 entries (in I-RO-HA ordering) and employed two-character code.

Besides correspondence with its own branches, the Bank of Japan used telegraph code with the Eighteenth Bank in Nagasaki. In 1885 introduced telegraphic code with other companies.

The Bank of Japan's telegraphic remittance code (as opposed to common code) was delivered to the Osaka branch in 1883. For other banks, remittance code was introduced in 1885.

After the outbreak of the Russo-Japanese War, overseas agencies increased their importance. Thus, the Bank of Japan compiled its own foreign telegraph code in 1905 and delivered copies to its agents in London and New York. It included a total of some 140,000 entries. The code groups were numbered and the numbers could be used when secrecy was required.

©2013 S.Tomokiyo
First posted on 10 December 2013 as a preliminary version. Last modified on 9 May 2019.
Cryptiana: Articles on Historical Cryptography
inserted by FC2 system