The present article describes field codes (trench codes) used by the American Expeditionary Forces (AEF), sent to Europe during World War I under the command of General Pershing (appointed in May 1917) (Wikipedia), relying on Friedman (1942) and materials quoted therein.
What might seem an obvious choice had to be learned by experience by those without hindsight. At first, codes were considered unsuitable for use in the front line and were reserved for communications between headquarters.
The Army Signal School of the AEF taught polyalphabetic substitution cipher with a celluloid cipher disk and Playfair cipher. Record shows that as late as 22 August 1918, new keywords for Playfair cipher were issued for seven intervals in September, the month of the Battle of Saint-Mihiel. (Friedman p.4; Appendices 1-3).
Codebooks, on the other hand, had been considered to be impractical for use in the field. In addition to the burden of production, distribution, and safe keeping (of which production was greatly facilitated by the advance of printing technology), there is always a risk of capture if used by field units (Friedman p.5-6).
How the French adopted code for field use was recorded by Givierge in 1924. At first, notebooks (carnets de chiffre) containing several tables for encoding important words in telephone messages were provided. They contained two-figure groups for letters and some 50 three-figure groups for common expressions. When wireless communication between small units developed, use of these notebooks was authorized as an exception. These developed into a codebook for words and syllables. The Germans, too, followed a similar course, beginning with small encoding tables for replacing common sentences by groups of letters or figures. Some of these tables were small enough to be written on concentric circles (like a cipher disk). By June 1917, the code system had become common. (Friedman p.6-7)
Still, the Code Compilation Section of the Signal Corps was set up in December 1917 at the General Headquarters (GHQ) at Chaumont, a town on the Marne. At the time, substantial American troops were not yet deployed in the front line. The Section, led by Major Howard R. Barnes, served until shortly after the Armistice in November 1918 (Friedman p.9; Kahn p.326).
The Code Compilation Section had no precedents to rely on. Even the British and French allies were at first reluctant to disclose their systems. It was not until later that they provided copies of obsolete editions for study. (Friedman p.9) It is not known exactly when it occurred but record shows that before the March offensive, the British, French, and American codebreaking teams were in cooperation (Friedman p.266) and that in June 1918 their relation was close enough that the Americans had their new code evaluated by the British (Friedman p.14-17). (On the other hand, the Radio Intelligence Section of the General Staff, "G-2, A-6", established in July 1917 under Frank Moorman, had access to British and French cryptanalytic methods to study (ASA2 p.178).)
The first products of the Code Compilation Section include a "Telephone Code" (sometimes called the "Female Code") for representing organizations and officers by names of women for use over telephone (Friedman p.20, 272) and an addenda sheet for the War Department Telegraph Code (see another article), both in March 1918.
The first substantial codebook produced by the Section, the American Trench Code (Friedman p.131, Appendix 11), was a small codebook of some 1600 words and phrases, for each of which a four-figure code group and a three-letter code group were assigned. (The first two figures and the first letter were printed at the top of the page.) The following shows an example of encoding with this code:
This was basically a one-part code, though sections for numbers, dates, etc. precedes the main body which starts with "1332 CEW" representing "A". Some pages had code groups for "ed, ing, ly, ment" printed in blank space.
A table for superencipherment (monoalphabetic substitution), called a "distortion alphabet" or a "distortion table", for letter code groups was also provided. For example, "What" is represented by a code word "ZEP", which is then enciphered into "UWN."
However, this was delivered no further down than regimental headquarters because of danger of capture (Friedman p.9, 241).
A smaller codebook of about 500 words and phrases, the Front Line Code (Friedman p.223, Appendix 19A), was also made for the front line and was issued down to companies (Friedman p.9-10). This assigned two-letter groups AB-ZZ for each word and phrase. (Some combinations such as AA, AE, ... are not used. Apparently, letters J, Q, and X are not used.) The same enciphering table as for the American Trench Code was provided.
A report of 17 May 1918 by 1st Lt. J. Rives Childs of a so-called Enemy Code Solving Section (abbreviated G2-A6) of G-2 (General Staff, Second Section) of GHQ evaluated the American Trench Code and pointed out some defects of the code and demonstrated weakness of the monoalphabetic superencipherment (Friedman p.117, Appendix 10). Moreover, it was realized that a two-part code was more suitable for handling in hard conditions at the front than a superenciphered one-part code (Friedman p.13).
While the American Trench Code provided both number groups and letter groups, the subsequent field codes provided either letters or numbers. The British used a three-figure system (Friedman p.9; p.114, Appendix 9). Unlike commercial codes (see another article), telegraph cost would not have been the deciding factor for field codes. An extensive study as to relative advantages of both systems was made but no definite conclusion could be drawn because opinions were almost equally divided. The Code Compilation Section often received requests to provide both letters and numbers. For voice communication over telephone, it was recognized that numbers were preferable. (Friedman p.26-27) The Urbain Code of the French provides both three-figure groups and three-letter groups, the former for telephone and the latter for wireless and wired telegraphy (Friedman p.45-46).
In a post-war meeting in February 1921, it was agreed that letters were preferable to digits (ASA3 p.21).
After the decision to adopt two-part code, a series of codes were prepared in five months from June to November 1918.
The first was the Potomac Code (Friedman p.153, Appendix 16A), the first of the so-called River Series, issued on 24 June 1918. (Use of such an epithet may have been learnt from the French, whose early codebooks had names such as Olive or Urbain (Friedman p.37 Appendix 4, p.43 Appendix 5), though the indicator system of these French codes were not adopted until later.) On this occasion, sample messages encoded in this code were submitted to the British for evaluation and favorable comments were received (Friedman p.145, Appendix 14).
It was a two-part code of 47 pages in total, with each page containing 100 entries in two columns, and contained about 1800 words and phrases. Variant groups were provided for common expressions (the lack of this feature was one defect pointed out for the Army Trench Code).
Possibly in consideration of Childs' report, the Instructions warned the user that "Words spelled out, letter by letter, ... are one of the favorite points of attack by enemy code men" and advised avoiding use of words not in code book when other words with the same significance are provided. In order to conceal double letters, different groups were to be used for them and a null should be placed between them.
In blank space of the columns, nulls and code groups for "ed, en, er, es, ing, ion, ll, ly, nd, re, s, st, th" were printed (Friedman p.17). The assigned groups may vary from page to page (again as advised by Childs' report). Thirty-five code groups were defined as nulls and it was prescribed that at least one null should be used every ten groups at irregular intervals.
Three editions were prepared: one distributed down to regiments; another to army headquarters, and a third held in reserve at GHQ, meaning that two sets were in reserve. When the codebook was captured on 20 July, replacement was complete within two days (Friedman p.271, 17; cf. p.21).
The Suwanee Code (Friedman p.160, Appendix 16B) followed on 15 July. It was similar to the predecessor, the Potomac.
The Wabash Code followed on 31 July. It was slightly smaller.
The Mohawk Code (Friedman p.167, Appendix 16D) followed on 3 August. Instead of the three-letter groups, four-figure code groups were provided. It contained about 2500 groups (numbered 2500 to 5000) for 1600 words and phrases. This codebook was captured in October and was thus discontinued (that is, it was not replaced by the following codes).
The Allegheny Code (Friedman p.173, Appendix 16E) followed on 12 August. The number groups were selected from 1500 to 5000. This was captured in October.
The Hudson Code (Friedman p.176, Appendix 16F) followed on 2 September. The number "2222", meaning "Code Lost", printed on the cover in red, was to be used from memory in case the codebook was lost.
The Colorado Code (Friedman p.179, Appendix 16G) (Internet Archive) followed on 24 September. This reverted to letter groups instead of numbers. Sixteen groups "com, con, dis, ed, en, er, es, ing, ion, ly, nd, nt, re, s, st, th" for spelling were printed at the bottom of every two pages. The code group "DAM", printed on the cover, was to be used from memory in case the codebook was lost. (The same code group "DAM" would be used in subsequent codebooks including at least Champlain, Huron, Osage, Seneca, Field Code No.1.)
When the Second Army was created in October (Wikipedia), the Lake Series of codes, with the cover printed in red, was begun with the Champlain Code (Friedman p.189, Appendix 16L) on 7 October, while the River Series was continued for the first army.
Instructions were given out such that all messages in these codes should be preceded by a three-letter indicator: HUD for the Hudson, COL for the Colorado, etc. Such a practice must have been learned by experience. At one time, different organizations were using Playfair cipher, the Hudson, and the Mohawk as well as various private codes and the Code Compilation Section was often asked to determine what particular code was used in certain messages (Friedman p.20).
The Champlain of the Lake Series was followed by the Huron Code (Friedman p.192, Appendix 16J) on 15 October (indicator: HUR). This now contained a page of concise two-letter code called "Emergency Code List" (Friedman p.211) in the beginning. (An "Emergency Code List" had been prepared in September for use in the front line with no access to field codes and 6000 copies were issued down to companies. Thereafter, each new field code was accompanied by a new edition of such an Emergency Code List (Friedman p.20, 258, 272; cf. p.25, mentioning proposal for "small tactical code books" on 17 September).) At the end of the Huron codebook, an easily detachable double receipt was provided for the convenience of officers receiving and delivering the codebooks. The instruction at the beginning now included a telephone alphabet for communicating letters over telephone (e.g., A - able, B- Boy, etc. (Friedman p.186)). This was to address the desire for number groups for the purpose of accurate transmission and possibly clear impression on mind (compare, e.g., "2632" with "ABZQ") (Friedman p.26). Further, the cover had a direction for using an indicator: "Precede every message in this code by HUR." and also called for attention for new entries by an indication: "Note. -- The * indicates new word or phrase." (These features, except for "*", are the same for Field Code No.1 below.)
The Osage Code (Friedman p.195, Appendix 16K) came on 28 October (indicator: OSA) This appears to be a River Series code paired with the Huron (Friedman p.271; compare Appendix 16K and 16J).
The Seneca Code (Friedman p.184, Appendix 16H) of the Lake series followed on November 6 (indicator: SEN), which in substance differed from the Osage only on the first page of the Encoding Section having variants for "minutes", "o'clock", "battalion", "regiment", etc. and the ordinals from "first" to "tenth" in the blank space in the columns. (This feature is also seen in Field Codes No.2, No.3 below.) Typographically, the initial capital of the entries was abandoned. Compared with the early Trench Codes, the nearly 1900 words and phrases were only moderately enlarged from 1750 but, due to revisions of necessary entries, only 1045 remained unchanged from the original.
At the time of the Armistice on 11 November 1918, the Niagara Code was in press and the Michigan Code and The Rio Grande Code were in manuscript.
Apart from these, there were Field Code No.1 (Internet Archive) (indicator: FC1), Field Code No.2 (Friedman p.202, Appendix 16M) (indicator: FC2), and Field Code No.3 (Friedman p.206, Appendix 16N) (indicator: FC3). These were printed but were never issued. After being kept in reserve for about three years in the United States, they were destroyed as obsolete.
In the meantime, the Code Compilation Section also compiled "Staff Code", a comprehensive code of about 30000 entries, for headquarters (see another article). It was basically a one-part code but a series of distortion tables for superencipherment were made in July. (Presumably, the advice of Childs' report of 17 May was not in time for this codebook. But anyway this was not for use in the front line.)
In October, a service code for radio stations was compiled and printed in six days. Up to this time, a French code had been used but the language had been a constant source of trouble (Friedman p.272, 20; p.237 Appendix 19 D.). The American Radio Service Code No.1 (indicator: RAD) is a three-figure two-part code and had entries such as "304 ... Antenna was damaged", "450 ... I am obliged to stop sending until ... o'clock."
After the war, Division Field Codes (initially called Army Field Codes) were made mainly for communications within the division down to companies, while Army Field Code (initially called Army Tactical Code) was planned for communication between divisions (ASA3 p.9-10, 22).
William F. Friedman, "American Army Field Codes in the American Expeditionary Forces during the First World War" (pdf) (1942) The following lists its appendices.
Appendix 1 (p.31; referred to on p.4) ... The U.S. Army cipher disk
Appendix 2 (p.32; referred to on p.4) ... AEF instructions for using the cipher disk and the Playfair cipher
Appendix 3 (p.35; referred to on p.4) ... Use of Playfair System for secret communication in AEF
Appendix 4 (p.37; referred to on p.6) ... A few pages from one of the early French codes: Carnet Réduit "Olive"
Appendix 5 (p.43; referred to on p.8) ... A few pages from the French Carnet Réduit "Urbain"
Appendix 6 (p.51; referred to on p.8) ... Extracts from a two-part French field code
Appendix 7 (p.67; referred to on p.8) ... Extracts from a two-part French field code of approximately 2,300 groups
Appendix 8 (p.73; referred to on p.9) ... Extracts from German field codes (A. (p.75) The Schluesselheft; B. (p.89) An Example of the enciphering and deciphering tables for the Schluesselheft; C. (p.91) The Satzbuch; D. (p.109) An example of the first three pages of instructions and tables for enciphering words not in the Satzbuch)
Appendix 9 (p.114; referred to on p.9) ... Extracts from a British Army field code
Appendix 10 (p.117; referred to on p.10) ... Report of 1st Lt. J. Rives Childs
Appendix 11 (p.131; referred to on p.10) ... The first AEF Field Code [The American Trench Code]
Appendix 12 (p.141; referred to on p.12) ... Enciphering card for first AEF Field Code
Appendix 13 (p.143; referred to on p.14) ... Test messages submitted to British in connection with testing of first AEF two-part field code
Appendix 14 (p.145; referred to on p.14) ... Report of Major Hay, of British General Staff (Code Section)
Appendix 15 (p.147; referred to on p.14) ... Report of Captain Hitchings, of British Army Code Solving Section
Appendix 16 (p.151; referred to on p.17) ... Sample pages from the various field codes prepared and used by the AEF
A. (p.153) Extracts from the "Potomac" Code
B. (p.160) Extracts from the "Suwanee" Code
C. (p.165) Extracts from the "Wabash" Code
D. (p.167) Extracts from the "Mohawk" Code
E. (p.173) Extracts from the "Allegheny" Code
F. (p.176) Extracts from the "Hudson" Code
G. (p.179) Extracts from the "Colorado" Code
H. (p.184) Extracts from the "Seneca" Code
I. (p.189) Extracts from the "Champlain" Code
J. (p.192) Extracts from the "Huron" Code
K. (p.195) Extracts from the "Osage" Code
The Numbered Series
L. (p.198) Extracts from Field Code No.1
M. (p.202) Extracts from Field Code No.2
N. (p.206) Extracts from Field Code No.3
Appendix 17 (p.209; referred to on p.18) ... The Emergency Code Lists [for Fie.d Code No.1, Huron, Seneca]
Appendix 18 (p.213; referred to on p.19) ... The "Staff Code" and sample of enciphering tables
Appendix 19 (p.221; referred to on p.19) ... Miscellaneous AEF Codes (A.(1) (p.223) Extracts from Front Line Code; A.(2) (p.228) Sample of enciphering card for Front Line Code; B. (p.230) GHQ, AEF, Bulletin No. 46; C. (p.232) G.S, 82nd Division, AEF, Memorandum No. 66; D. (p.237) Extracts from American Radio Service Code No. 1: E. (p.244) Code for Designation Organizations, Commanders and Staff Officers; F. (p.247) Pamphlet "Telephone -- T.P.S. -- T.S.F. -- Visual"; G. (p.250) Temporary Code for Transmission of Casualty Data by Telephone and Telegraph)
Appendix 20 (p.253; referred to on p.20) ... The "Baseball Code"
Appendix 21 (p.257; referred to on p.23) ... "Secret instructions for the use of Army Codes"
Appendix 22 (p.261; referred to on p.22) ... Examples of letter, telegrams, and notes concerning violations of cryptographic security
Appendix 23 (p.265; referred to on p.26) ... Lecture delivered by Lt. Col. Moorman before Officers of M.I.D., February 13, 1920
Appendix 24 (p.271; referred to on p.27) ... Extract from the Annual Report of the Chief Signal Officer (1919) [Annual report of the Chief Signal Officer is found at HathiTrust. The Chief Signal Officer was George Owen Squier (Wikipedia) during World War I.]
Appendix 25 (p.273; referred to on p.12) ... Notes by J. Rives Childs after reading draft of this paper