Tales From The Black Chambers


Despite the horseplay, the oss developed a code system that, for simplicity and effectiveness, was ideally suited to its needs. During an agent’s training, between glasses of free beer, he was required to memorize a numerical equivalent for each letter of the alphabet. No two agents learned the same “numerical alphabet.” Later, prior to each field mission, he was supplied a copy of what was called a one-time code pad; the only other copy was at his headquarters. The code pad was simply a sheet imprinted with rows of numbers selected at random by machine. The sequence of numbers was of no importance whatever to the system—only that the two pads were identical. When the agent wished to transmit a message, he did it letter by letter. If the first letter of the message was T and his memorized equivalent for T was 16, he added 16 to the first number on the code pad and transmitted the total. The recipient, of course, subtracted the first number on his code pad in deciphering the message. The system was virtually foolproof because no agents had the same numerical alphabet; each code pad was different, with numbers chosen at random; the odds were high that any one letter of the alphabet would have a different numerical value each time it appeared in the message; and even if the code pad was captured, it was useless to the enemy without the memorized numerical equivalents of the letters.

The system, for example, would be used to radio instructions for an airdrop of particular supplies in Burma. The enciphered message further stipulated a recognition code for the pilot of the supply plane. At the appointed hour and spot the agent would lay out recognition panels in a coded pattern. Upon seeing the panels displayed correctly, the pilot buzzed the spot the number of times designated in the message, then watched. The panels had to be rearranged in another coded pattern before he would make the airdrop.

Today the National Security Agency, the supersecret arm of the Department of Defense, is primarily responsible for the important code and cipher work of the United States. Its huge building, a few miles north of the nation’s capital, houses thousands of scientists, engineers, and mathematicians. It constantly seeks men trained in computer work. It seems obvious that computers have taken over at least some of the nation’s cryptography and cryptanalysis. They would appear to be ideally suited to such work. A highly secret message, for example, could be translated into digits by any normal ciphering system. These digits could then be fed into a computer with instructions for processing them. According to the instructions the computer could differentiate them a million or more times within a few seconds and produce a simple mathematical equation—the encoded message. Each differentiation would be the equivalent of reciphering the message into a completely different system. The receiving computer, operating on reverse instructions, would work backward from the equation, integrating as many times as necessary, and produce the numerical equivalent of the plaintext. The theoretical system described here is speculation but well within the abilities of existing computers.

It would appear virtually impossible to break such a computer system. Yet other ciphers and codes, seemingly impregnable in the past, have succumbed to the assaults of the men of the black chambers. If one computer can encode a message, another can break the code. Today, perhaps, computer is pitted against computer, but the secret war of the black chambers continues.