This is a much abridged version of articles in Japanese (army, navy) and describes codes and ciphers of the Imperial Japanese Army and Navy mainly from the Meiji Era.
Table of Contents
In October 1874 (the year of Meiji 7), the following checkerboard substitution cipher was issued. It represents kana with two digits from 1-7. Kana are arranged in I-RO-HA order vertically starting from the bottom-left corner.
In 1875, there was a similar cipher with kana arranged vertically starting from the top-left. Another version starts from the top-left and proceeds horizontally. Still another version starts at the top-right corner and proceeds diagonally in a boustrophedon manner.
A revised cipher came into use in January 1877, which included 22 substitution tables. Each substitution table juxtaposes 44 kana in I-RO-HA order with those in reverse I-RO-HA order with one of 22 different starting positions, whereby paired kana are transposed with each other. For example, with No. 1, I is enciphered into SU and SU is enciphered into I. Actually, the 22 substitution patterns appear to have been created with a cipher wheel similar to those of Mishima Michitsune (issued in November 1879) and Iwakura Tomomi (used in 1877).
This cipher was used until after the outbreak of the Sino-Japanese War (1894.7-1895.3). The latest use known to the present author is on 11 October 1894.
Numerical cipher was used for communication with China, Korea, etc. The following cipher can be reconstructed from existing telegrams. (Characters in blue are conjectures.)
Overseas communication may be conducted with figures or Latin alphabetical letters. The Foreign Ministry preferred alphabetical code because of error resiliency. When the Army set up military telegraph lines in Korea at about the time of the Sino-Japanese War, only kana and figures were accepted. The Foreign Ministry found it inconvenient and explained that it had mainly used alphabetical letters, which would help deciphering in case of some errors, and used kana only for messages not particularly confidential or for exceptional emergency. The Foreign Ministry warned that errors in numerical cipher often made a message ununderstandable. The Minister of War explained that some of their telegraph operators were common soldiers and could not handle telegrams in alphabetical letters. In October 1894, however, the army made arrangements such that alphabetical letters could be accepted.
In July, a military attaché at the Consulate in Seoul reported that new cipher was necessary. At the same time, the diplomats in consulates in Korea sent similar reports to the Foreign Ministry.
At noon on 15 October 1894 (Meiji 27), the old cipher in use from 1877 was replaced by OTSU Cipher, which was again a polyalphabetic substitution cipher as follows (with kana in A-I-U-E-O order this time).
A similar substitution cipher based on A-I-U-E-O order had already been used in August 1893. The cipher is called Cipher No. 2 of the Accounting Bureau.
General Kawakami Soroku (1848-1899) left a substitution table in I-RO-HA order. It is clearly different from the OTSU Cipher or earlier ciphers in that it had an irregular arrangement of kana and included all the 48 kana. The present author has not seen an actual use of this cipher.
From the time of the Sino-Japanese War, the Japanese Army revised its cipher system more regularly.
The OTSU Cipher was replaced on 15 September 1896 (Meiji 29) by the KEN Cipher, which appears to have been triggered by a report of loss of a copy of the OTSU Cipher.
The KEN Cipher was replaced on 25 March 1897 (Meiji 30) by the GEN Cipher, again apparently triggered by a loss of cipher.
Within the year, the GEN Cipher was replaced by the MAN Cipher. Its manual revised in December 1897 changed designation of the substitution tables from numbers to words.
The MAN Cipher was replaced on 1 August 1898 (Meiji 31) by the BUTSU Cipher.
The BUTSU Cipher was replaced on 10 January 1901 (Meiji 34) by the YAKU Cipher. Although the specific structure of the cipher is not known, a record shows it was pointed out that it was too complicated. The YAKU Cipher appears to have been still in use in the first half of 1904 but was discontinued by June.
Apparently, the KON Cipher, introduced with the KEN Cipher, was a numerical cipher for overseas communication. It appears to have been discontinued with the BUTSU Cipher in 1901. The TOKU Cipher, which was in use with the YAKU Cipher, appears to have replaced the KON Cipher (yet to be confirmed). It appears to have been discontinued with the YAKU Cipher in June 1904.
The revisions of kana cipher after the Sino-Japanese War made the system too complicated to be practical. At the time of the Russo-Japanese War, the army reverted to simpler systems.
The Japanese Telegraph Cipher (Wabun Denshin Ango) (Military Secret No. 4) issued on 19 April 1904 was probably the first of this name. It was revised every two or three months during the Russo-Japanese War and occasionally after the war.
Many telegrams during the Russo-Japanese War are in several ciphers, of which the following substitution cipher would correspond to this series of cipher in view of the timing of revision.
Although it was no longer polyalphabetic, the arrangement of kana became irregular (in particular from the version of October 1904). Further, kana with voiceless consonant and kana with voiced consonant (e.g., ka and ga) were now to be assigned their substitutes independently. (With kana, voicing of consonant is represented with dots. Previously, cipher of kana with voiced consonant had been simply the cipher of kana with voiceless consonant plus dots.)
Moreover, single kana with a voicing mark was used to represent numbers or military terms.
In 1925 (Taisho 14), distinction was made between a version distributed in areas of heavy use and to be updated frequently and a version distributed in areas with infrequent use.
Some telegrams during the Russo-Japanese War were encoded in two-kana code. This appears to be the Army Telegraph Code (Rikugun Denshin Fugo) revised in early 1904 (because it has a similar form with the existing 1901 version (see an article in Japanese)). The Army Telegraph Code was to reduce the number of letters in a telegram to save telegraph cost and it was distinguished from a secret code or cipher, though the distinction was not always observed.
The following are code words used in encoded telegrams from May 1904 to April 1905.
The Army Telegraph Code adopted a three-kana system without voicing dots in 1922 (Taisho 11). This is about the time most commercial codes adopted similar three-kana code (see another article).
In February 1904 (Meiji 37), the Army Telegraph Secret Code (Rikugun Denshin Ango) was distributed. There is evidence that it was a voluminous codebook of more than 1000 pages and employed figures to represent words. A report that "305" is missing in the margin of p.613-614 suggests these pages included 100 codes from 30500 to 30599. In brief, the Army Telegraph Code appears to have been a five-digit code.
There are many telegrams encoded in five-digit code. The same code was used by military attaché resident in Europe as well as the commander and the staff in Manchuria and Korea.
From the extant telegrams, the code can be partially reconstructed as below. It indicates that words and phrases are assigned to code numbers in the alphabetical order of the romanized plaintext.
From 15 November 1904 on, the code numbers were augmented by 3 for military attachés in Europe, but not for other districts. In the above table, green records show code numbers calculated by subtracting 3 from such augmented code numbers.
The Army Telegraph Secret Code of February 1904 (Meiji 37) was revised on 1 October 1912 (Meiji 45). The next revision came into use on 1 January 1918 (Taisho 7) and discontinued in 1926 (Taisho 15).
In May 1902, the military attaché in London was given an alphabetical code in place of the one he had made for correspondence with Japan. There is also evidence that an alphabetical code of May 1902 was given to the commander of the army in Manchuria and several officers and was returned in 1906 after the Russo-Japanese War.
The Alphabetical Telegraph Code of May 1910 (Meiji 43) was revised in 1912 (Meiji 45). Probably, this was a code of some 600 entries for representing kana/words/phrases with two alphabetical letters, which is known to have come into use around 1912.
The military attaché code is known to have adopted four-letter code after World War I (e.g., RNES for Germany, GAWD for negotiation, NBSJ for kana u) and a polyalphabetical system. Such versions would correspond to the polyalphabetical codes Jr, Jn, and Jq as designated by Yardley's Cipher Bureau (see another article).
In 1923, Polish Captain Jan Kowalewski was invited to Japan and gave a lecture on cryptology to army officers. Although the polyalphabetic system of Jr had been introduced before this, later revisions may have owed something to his lecture.
Specific alphabetical telegraph codes were prepared for military attachés in China (in 1914 or earlier) as well as for those in Russia (in 1925 or earlier), while the same code was used for both Europe and China/Korea during the Russo-Japanese War (1904-1905). In 1927, the Alphabetical Telegraph Codes (type KO and type OTSU) were re-designated for use by military attachés residing in Europe and America.
It had been known that as of January 1876 the incipient Japanese Navy used code patterns such as "26", "15", "no 9", "98", "to 9", which appear to be a checkerboard cipher. It was similar to the cipher used in 1874 during the Taiwan Expedition (another article in Japanese).
Right afterwards, in March 1876, the navy issued a similar telegraph cipher. It indeed employed a checkerboard system, which had columns headed by one- or two-digit figures or one or two kana and rows numbered 1-9. For foreign communication, the one or two kana heading a column were accompanied by an alphabetical letter or alphabetical letter+Arabic figure. (The Arabic figure was used presumably because the number of kana is about twice that of alphabetic letters.) The total number of combinations (including blank places) was 1278.
When the 1879 revision of the international telegraph rules disallowed letter-figure combinations, provisional revision of the cipher appears to have been made, thus deferring a full-scale revision.
At the time of the Imo Incident in 1882, a Telegraph Code was made, which represented words and phrases with two or three kana or figures in order to save telegraph cost, which was in contrast to a telegraph secret code above meant for secrecy. While a similar distinction was also made in the army, the secret code of the navy included not only substitution of kana but also code patterns representing words and phrases, which made the distinction confusing. Thus, it was determined that the enlarged and revised edition of March 1883 (Meiji 16) was the one to be kept. It had both Japanese and alphabetical codes. Its specific nature may be ascertained from the supplement of the next year.
At the time of the Sino-French War of 1884 (Meiji 17), insufficiency of the current code led to production of a supplement, which allows ascertaining the nature of the enlarged and revised edition of 1883. Each column was assigned two kana and each row was assigned a number from 1-9. For foreign communications, both columns and rows had alphabetical letters accompanying the kana or figures. Thus, letter+figure combinations of the 1876 code were eliminated. The following is the structure of the code, inferred from this supplement. (The green squares are in the supplement, with all the other squares being conjectures.)
In December 1888 (Meiji 21), the I-RO-HA Telegraph Code was published for commercial use (see another article). I, RO, and HA were the first three kana in the traditional Japanese syllabary. In a way, it was a Japanese counterpart for the well-known ABC Code (see another article) in English. The I-RO-HA was a code for representing words and phrases with two kana.
In 1889, the navy purchased copies of the I-RO-HA for its use. Actual telegrams encoded with the I-RO-HA are extant.
The navy used the I-RO-HA Code at least until May 1890 (Meiji 23), while an extant telegram shows that the navy made its own three-kana code in September 1890 at the latest. As with many commercial codes, the first kana of a three-kana code word matches the first kana of the plaintext. On the other hand, as discussed in another article, three-kana code (without voicing marks) became common among commercial codes only in the 1920s. The 1913 edition of the Army Telegraph Code still used an old-style two-kana code. The navy's three-kana code as early as 1890 appears to have been novel. (Because of military secrets, it would never have influenced commercial codes.)
There was an attempt to prepare a common code/cipher between the army and the navy and, for the time being, the army provided its own cipher in March 1894 for common use between the army and the navy. The example given shows it was the army's old polyalphabetic cipher of 1877. Although what came out of this plan of common cipher/code is not clear, at least, there is evidence that the army continued providing its revised codes/ciphers to the navy.
In August 1894, right after the outbreak of the Sino-Japanese War (1894.7-1895.3), the navy's secret telegraph code was revised.
JACAR preserves many telegrams encoded in three-kana code, the oldest of which belongs to September 1894 as far as the present author knows. The following is the partially reconstructed code (portions in blue are conjectures). This is different from the code used in 1890-1892 because the plaintext or ciphertext marked with a star in the above is differently assigned.
As with the 1890 edition, the first kana of the three-kana code word matches the first kana of the plaintext.
The kana code of 1894 continued in use after the Sino-Japanese War and appears to have been used in 1897, 1898, and 1900 (judging from use of a common code word).
At the time of the Russo-Japanese War (1904.2-1905.9), it has been known that three-kana code was used, with its second kana, not the first one, matching the first kana of the plaintext. Apart from this, two-kana code words represented ships or units, of which the first kana matches the first kana of the plaintext. This kind of code was in use at least from 1903 to 1907.
From a telegram in 1911 (Meiji 44), it appears that the system of representing the beginning of the plaintext by part of the three-kana code had been abandoned by this time in favor of random assignment.
Telegraphy in kana was not available in the Western world. The 1876 cipher was adapted for such a situation by providing encryption in alphabetical letters and figures. Yet, in 1878 (when a navy vessel was dispatched to Europe) as well as in 1885 and 1886, the navy borrowed a code from the Foreign Ministry.
At least since 1894, the navy had its own alphabetical code for communication with military attachés abroad.
The following is the code partially reconstructed from existing telegrams. The left shows the version used at least from January 1894 up to early September 1894. The right shows the version used since September 1894, which was used as late as January 1897 and October 1898.
A telegram from 1899 uses codes such as "equissimo kwanja", "proscrever sudeni", "alcovitar nite" and "acarreto isaiyubin". Judging from the alphabetical position of the plaintext sudeni, which is near the end of I-RO-HA order, this is the version between the September 1894 version above and the 1904 version below.
The following is the version used in 1904.
JACAR (Japan Center for Asian Historical Records National Archives of Japan)