In May 1929, the Army Signal Corps, having recently taken responsibility for cryptology from Military Intelligence, held two weeks of training for officers involved in cryptologic work (including secret inks). In addition to military officers, Herbert Yardley, chief of the Cipher Bureau, attended. Some of the training was conducted by William F. Friedman, who served as a civilian consultant on cryptology to the Signal Corps in the period after World War I, and was hired full time to compile codes in 1922.
This conference stimulated a review by the Signal Corps of its cryptologic organization in July 1929. The Signal Corps decided to create units to compile codes, train to solve enemy codes in time of war, and establish intercept services in the field. William Friedman, who was in attendance gave his opinion that it would take up to two years to train cryptanalysts to work independently; the Corps recommended the four men be hired for this training.
The Signal Corps emphasized, however, that "all work of the Signal Intelligence Service is to be organized on the basis of training for war rather than of active operations for immediate interception and solution of the communications of foreign armies or governments." Nevertheless, Signal Corps leaders noted that if information about foreign countries should become available as a by-product of the training, they would accept it.
On April Fool's Day in 1930, Friedman, the U.S. Army's senior (and only) cryptanalyst, hired the first three "junior cryptanalysts" in the government at $2,000 dollars per year. The first to be hired was Frank B. Rowlett, a mathematics teacher from southern Virginia. He was followed later that month by Abraham Sinkov and Solomon Kullback, both former math teachers from New York City. (The salary was good money in the early years of the depression and the new hires were glad to get it).
The Army in mid 1929 had made a decision to establish the "Signal Intelligence Service," with Friedman as chief civilian cryptologist. Although the name suggests that the organization had been established to engage in codebreaking, in actuality, the main duty of its personnel was compiling codes of varying types for U.S. Army use. In accordance with Signal Corps doctrince, personnel were to be trained in cryptanalysis only so they could work against enemy systems in wartime. Furthermore, Friedman believed that skill in cryptanalysis was essential for good cryptographers, and trained his people that way.
William F. Friedman had been born in Russia, and brought to the United States at the age of two. He became involved in cryptology prior to World War I, working at a private think tank. His education at Cornell University had been as a scientist; with a strong mathematical background, he became one of the earliest to realize that the basis of modern cryptology was mathematics not language or area studies. As a lieutenant in World War I, the Army sent him to France to solve German codes.
Friedman, whose forte was training, conducted rigorous education in codes and ciphers for the new employees of SIS. After a period in which they worked on material of his own design, he dramatically introduced them to the records of Yardley's "Black Chamber," and had them go over these materials until they were completely familiar with the foreign systems and the methods used to solve them.
In addition to the work on systems of the past, Friedman put his employees to work studying and solving the rudimentary machine cipher systems of the time. All realized that this was the future of cryptography.
The Army in the mid-late 1930s began rebuilding the intercept services it had had during World War I. It was recognized that the SIS, good as it was at cryptography and cryptanalysis, was composed of civilians and lacked the ability to support combat operations directly.
Training for intercept was to be conducted at Fort Monmouth, New Jersey.
There were five signal detachments -- at Fort Sam Houston, the Presidio of San Francisco, Fort Shafter in Hawaii, Fort McKinley in the Philippines, and Quarry Heights in the Panama Canal Zone -- provided Army intercept. In January 1939, these six detachments were consolidated into the Second Signal Service Battalion.
In addition to an increase in personnel, the Signal Corps was directed in late 1939 to acquire a monitoring station in the Washington, D.C. area. With the coming of war, SIS acquired Arlington Hall Station, a former girls' academy, for its headquarters and established an intercept site at Vint Hill Farms in Virginia, just south of the capital.
In the late 1930s, SIS transferred Abraham Sinkov to Panama and Solomon Kullback to Hawaii. Their tasks were to arrange for more reliable sources of intercept, and to train local Signal Corps personnel in cryptanalysis. Neither move was entirely successful, so they were recalled when it was recognized that both could make more effective contributions at the center.
It should be remembered that SIS was not attached to the Army's intelligence organization. Military Intelligence considered cryptanalysis merely an adjunct of cryptography, which belonged in the Signal Corps. This situation continued until early in the war.
SIS began with purposes other than foreign intelligence; it was to compile systems for Army use and prepare cryptanalysts for wartime operations. Again like its counterpart, SIS produced valuable information almost as a by-product of training and was transformed into a production office for current COMINT.
SIS was renamed the Signal Security Agency in 1943, and in September 1945, became the Army Security Agency.
- William F. Friedman, A Brief History of the Signal Intelligence Service, 29 June 1942, SRH 029, CCH Files.
- Anon. Centralized Control of U.S. Army Signal Intelligence Activities, SRH-276, CCH Files.
- Anon., "Memorandum" re O.C.S.O Conference, 19 July 1929.
- Anon., The Second Signal Service Battalion, SRH-135, CCH Files.