The Code Battle

On June 20, 1974, a slow, unwieldy vessel filigreed with struts and derricks lumbered out to sea on a top-secret mission. She was the Glomar Explorer. Her secret task was to raise from the depths of the Pacific Ocean a Russian submarine that had sunk. The U.S. Government wanted to obtain the submarine's missile warheads and her codes. For this, it was willing to spend $350,000,000 of the taxpayers' money--an amount equivalent to giving 3,000,000 more people Medicare coverage, sending 20,000 students to college or buying 90 tanks, 60 bombers or a third of a nuclear-powered aircraft carrier.

Why did Washington think it was worth it?

In the spring of 1942, American code breakers, hidden in the basement of a building in the navy yard at Pearl Harbor, broke the main Japanese naval code. Their solutions of Japanese intercepts provided virtually complete information on the size, course and timetable of the Japanese fleet. As a result, wrote a top-ranking officer, "We were able to concentrate our limited forces to meet their naval advance on Midway, when otherwise we almost certainly would have been some 3000 miles out of place." At Midway, the United States smashed the invading armada in a battle that doomed Japan.

A year later, those same cryptanalysts cracked open a moderately long message in a subsequent edition of that same Japanese naval code. It disclosed that the mainspring of Japan's military efforts, Admiral Isoroku Yamamoto, would soon make an inspection trip that would bring him within range of American combat airplanes. Moreover, the message, addressed to subordinate commands, specified that Yamamoto would land at 0800 on April 18, 1943, on Ballale, one of the Solomon Islands. With this information, the Americans dispatched 18 twin-engined P-38s, which ambushed the punctual admiral in his bomber over a tropical jungle, shot it down and gave the U.S. the equivalent of a major victory.

On the other side of the world, British code breakers, working in Quonset huts in the London exurb of Bletchley, intercepted messages of the German army high command during the precarious early hours of the Anzio landing. These revealed, just as the American forces were about to extend themselves from the beachhead, that fresh German units had been ordered into the area. General Mark Clark pulled back and consolidated his forces, repulsed the German counterattack and later advanced into Rome.

On D day, as the Allies stormed the Normandy beaches to breach fortress Europe, the code breakers intercepted a German message ordering a counterattack. Forewarned, General Omar Bradley took measures that helped keep the Americans from being flung back into the sea. Later, at Bastogne, code breakers cracked a radiogram that enabled General George S. Patton, Jr., to inflict heavy losses on a redoubtable German paratroop division.

World War Two had seen dozens, perhaps hundreds, of similar instances in which code breaking had played a vital role. A former director of naval intelligence exclaimed, "It won the war!" Chief of Staff George C. Marshall declared that code breaking was "our main basis of information regarding Hitler's intentions in Europe" and contributed "greatly to the victory and tremendously to the saving in American lives." A high official said that it shortened the war by a year. After it was all over, a Congressman paid high tribute from the floor of the House: "I believe that our cryptographers ... did as much to bring that war to a successful and early conclusion as any other group of men."

During World War Two, code breaking had become the most important means of obtaining secret information. No other source possessed to the same degree the elements of successful intelligence: volume, anticipation and veracity. Reports based on visual observations of the enemy by patrols and on interrogations of prisoners of war were voluminous and accurate but good only for the immediate future. Spies, on the other hand, could reveal enemy plans far in advance, but suspicion permanently blighted their labors: No general would risk his men--or his career--on the radioed word of an informant whom the enemy might have paid more or put under duress. Aerial photographs yielded data as hard as could be, but they were relatively sparse, owing to their snapshot nature, and showed only what was already there or on the march. Code breaking alone could provide the quantity and quality of intelligence necessary to sound military planning.

•

As the hot war congealed into the cold, the U.S. Government wished to preserve the information-gathering capabilities that had proved so effective against the Axis. The trauma of Pearl Harbor, which had led to the centralization of military affairs in the Defense Department and of intelligence in the Central Intelligence Agency, eventually fathered as well a unified code-breaking agency--the Armed Forces Security Agency, established in 1949. The merits of the unified approach soon warranted expanding the role of the Defense Department's AFSA. On November 4, 1952, President Harry S. Truman turned it into the National Security Agency, serving every branch of Government.

NSA reigns today as the supreme arbiter of all matters cryptologic in the United States. It promulgates cryptologic doctrine, establishing the rules by which, say, the State Department will encipher its dispatches. It coordinates the code-breaking agencies of the Army, the Navy and the Air Force in their specialized missions against their foreign counterparts. It issues specifications to manufacturers for components of cipher machines, which it, for security's sake, then assembles on its own premises. It analyzes foreign radar emissions, so that U.S. nuclear bombers will be able to jam or trick them in war. But, most of all, it cracks the codes of foreign governments and daily submits the solutions to U.S. officials as high as the President. This function has made it the biggest intelligence agency in the free world--bigger even than the CIA--and, within the U.S. Government, the most secret.

Counting the military personnel assigned to it, about 100,000 people work for NSA--about five times as many as for the CIA. It spends several billion dollars a year. In sharp contrast to the head of the CIA, the NSA director, normally a three-star general or admiral, never makes statements to the press and rarely appears before Congressional committees in public hearings.

Security is as tight if not tighter at NSA than anywhere else in the Government. Its headquarters--two boxy modern buildings at Fort Meade, Maryland, just visible from the Baltimore-Washington Parkway--is surrounded by three fences, two topped with barbed wire and one electrified. It is protected by U.S. Marines. Inside, Marines escort visitors everywhere, including to the men's rooms. NSA employees must meet some of the Government's strictest security standards. They can be fired if the director merely finds it "to be in the interest of the U.S."

All this secrecy enshrouds work sometimes far from Fort Meade. Much of it begins in lonely monitoring posts scattered about the globe, especially along the borders of the Soviet Union. There, in Quonset huts on the wind-swept Eritrean plateau in Ethiopia or in the dusty foothills of the Hindu Kush in Afghanistan, far from prying eyes and the electrical interference of cities, radiomen lean forward, straining to pick up every dot and dash or every syllable of a foreign radio transmission through the static that crackles in their earphones. Their antenna fields sometimes cobweb whole mountainsides. Other monitors fly in airplanes or sail in ships as close as they dare to foreign coasts or frontiers to pick up every possible scrap of text. Sometimes their sedentary work becomes dangerous. In 1960, the Russians shot down Francis Gary Powers' U-2, which was carrying not only cameras but also "black boxes" whose magnetic tape recorded Soviet radar signals. Israeli planes strafed the U.S.S. Liberty as it cruised the eastern Mediterranean during the Six-Day War, its electronic ears wide open. And the U.S.S. Pueblo became a cause célèbre when the North Koreans captured it, packed with eavesdropping gear, early in 1968.

Some of the interception is automated. Satellites moving slowly above the Soviet Union receive, process and retransmit Russian radio signals. (NSA's share of the cost of lofting these squat cylindrical spies in the sky constitutes a major portion of its vast budget.) It was such a satellite with a sophisticated antenna system that reportedly eavesdropped on Kremlin leaders as they talked over the radiotelephones in their cars.

In West Berlin, in a hidden U.S. intercept post, a $3,000,000 machine by Ampex, filling a space equivalent to two living rooms, can tape-record 2000 channels of communication simultaneously. The tapes are burned after they are used once, because erasing them for reuse would destroy their superhigh quality. Other machines, which record everything sent on a given frequency, continue to print out periods on six-ply carbon paper when the circuit is "up" but nothing is being sent. They keep on tapping for hours, days, weeks, even months, at two minutes and 15 cents a page, just waiting for some message to come across.

In the United States, NSA is reported to have monitored most cable and telex messages into and out of the country. Computers scan messages for trigger words, such as oil and Mideast, and have texts containing them printed out. Such economic intelligence could help the Government make decisions on such matters as oil imports and grain sales, vitally affecting the cost of living. But the questionable legality of this activity is one (continued on page 224) Code Battle (continued from page 136) reason that the House and Senate intelligence committees are looking into NSA.

Another reputed agency achievement was a spymaster's dream. The Soviet Union allegedly eavesdropped, from its embassy in Washington, on hundreds of thousands of domestic American telephone calls, including those to and from Congressmen. NSA then intercepted the Soviet transmissions of the results back to Russia.

Many intercepts that pour into the Fort Meade headquarters are, like these, in clear language. The communications of Soviet air force pilots with one another is another example. Analysts listen to their chat, recognize individuals by speech peculiarities, index names and other details. From this, NSA can build up a good picture of a squadron--its commander, its men, its morale, its equipment, its transfers. Many such analyses join to create a picture of the Soviet air force as a whole.

But most of the intercepts are in code, and these go to the code breakers, a rare and peculiar breed of men. "Back-room boys," the British call them. Most of them today are mathematically inclined, in contrast to those of pre--World War Two vintage, who were primarily linguists. The change reflects the world-wide shift in cryptographic systems.

They are highly intellectual, lovers of word games, puzzles and chess. Indeed, the late British chess champion C. H. O'D. Alexander was a star of the British code-breaking establishment. Once, when he was playing the Russian grand master David Bronstein, he learned during conversation that Bronshtain did the same kind of work. Curiously, many of the great cryptanalysts have been fine musicians. The greatest code breaker of World War One, the Frenchman Georges-Jean Painvin, had won a prize in cello at the Nantes conservatory. After Pearl Harbor, the Navy's code breakers, needing more men, commandeered the band of the sunken battleship California. Nearly all the members proved above average and some were outstanding.

The work requires, for success, a rare and peculiar turn of mind, sometimes termed cipher brains. It is not surprising that many cryptanalysts are magnificent eccentrics. Take, as an example, Britain's finest cryptanalyst, Dilwyn Knox, who in World War One reputedly cracked the German submarine code in his bath and in World War Two helped solve the several versions of the German cipher machine Enigma. Day after day, he would try to leave his office through the cupboard. The girls there waited to see if once, just by chance, he would go out through the door. He never did. Whenever he ruled a line, he ruled his thumb in. Yet another part of his brain so illuminated the complex mechanism of the Enigma that it greatly aided Britain in staving off defeat and later in winning the war. Knox worked intuitively. A certain movement of the mechanism had been called a crab. "Where there's a crab, there's a lobster," reasoned Knox--and he found the corresponding movement.

Another of the Bletchley originals was Alan Turing. One of the greatest mathematicians of the century, he is widely known as the creator of the Turing machine, an idealization of the computer. During World War Two, Turing bicycled the three miles from his rooms to Bletchley every day on a rickety contraption whose chain regularly fell off. Instead of just fixing it, Turing noted that this event occurred every so many revolutions of the pedal. He then correlated a bent wheel spoke with a damaged link in the chain. Only then did he attack the repair. He sometimes set his watch by making some complex preliminary calculations and then observing from a fixed point the occultation of a particular star by a certain building. He took his love of exercise to extremes, preferring to jog 14 miles across London to rushing for trains and waiting in smelly underground stations. Tall, broad-shouldered, blue-eyed, he paid not the least attention to the 100 or so girls in his department at Bletchley. Instead, he devised a telephone scrambler that baffled the Germans, who had been listening to the transatlantic conversations of Roosevelt and Churchill. And he pioneered in developing for code-breaking purposes one of the world's first programmable computers, called the Colossus. It enabled Bletchley to read many German cipher-machine messages that otherwise would never have gotten to Allied commanders in time to be of use.

But this is the era of the corporate man and of teamwork. As in science, where dozens may research a problem, whole teams of cryptanalysts may attack a foreign cipher system, and most are about as colorful as dentists or engineers. At NSA, code breakers work in offices like those of an insurance company. In large rooms, each devoted to a particular world region, country or foreign-government branch, stand rows of flat-topped, gray steel desks. The cryptanalysts bend over them, scanning print-outs, testing with colored pencils solutions on square-ruled paper, flipping the pages of some reference book. They confer, stare distractedly out the windows, scribble furiously and sometimes yelp with joy. Each man is constantly scrutinizing the intercepts for some quirk, some irregularity, some pattern that constitutes the chink in the armor of the cryptogram. During World War Two, an English woman cryptanalyst sensed something odd about an Italian intercept. She quickly spotted it: The page-long cryptogram had no Ls. She knew that the Italians had been transmitting fake messages in an attempt to deceive the English. She knew, too, that this particular cryptosystem precluded any letter from representing itself--in other words, an A in the original message could not become an A in the cryptogram, though it could become any other letter. The fact that this intercept had every letter except L therefore meant with a high degree of probability that the original message was a dummy consisting only of Ls. On this basis, she broke into the system.

The human cryptanalyst--especially one with a cipher brain--remains even today the basis of code breaking. Solving something still often comes down to the fitting of a half-remembered name to an incompletely solved message, and only a human being can do that. But computers help greatly. They can count torrents of letters at high speed, tirelessly compare one message with another in a search for repeated groups of letters, generate all possible solutions of a cryptogram to let the cryptanalyst find the one that's not gibberish.

Consequently, NSA has assembled more computers under a single roof than probably any other institution in the world. They number in the scores, if not the hundreds. And these machines are among the fastest and most sophisticated in the world. Not content with buying the biggest and best computers it can find, NSA expands and upgrades them. Some years ago, it acquired the IBM Stretch, a machine so huge that only a few other Government agencies, such as the National Weather Service and the Atomic Energy Commission, had use for one. But it was not good enough for NSA. The agency added a portion, called the bump, that was larger than the original computer. At that time, the typical magnetic tape had only 100 "bits" to the inch. NSA squeezed in 3000--and then streamed the tape past the reading heads at 275 inches per second.

NSA's extraordinary computer capacity accounts for much of its success in the world of cryptanalysis, where success is partly a function of available computer time. Britain's present code-breaking agency, Government Communication Headquarters, at the western edge of the flower-bedecked spa of Cheltenham, is, in the view of one observer, falling farther and farther behind NSA, because it cannot keep up with this country's computer capability.

•

Although they use the most modern of marvels, the code breakers do not disdain the most ancient of tricks: simply getting hold of the other fellow's code. This is what the CIA tried to do with the Russian submarine. Maritime seizures of this kind have figured frequently in cryptologic history, and it is curious to note that the Russians were also involved in one of the most famous of these cases. Just after the start of World War One, the German light cruiser Magdeburg was wrecked in the Baltic. A few hours later, the Russians picked up the rigid body of a drowned German officer, whose arms still gripped the lead-bound codebook of the Imperial German Navy. The Russians at once passed it to the British, the chief maritime power, who used it to master the German codes, first naval and then diplomatic. Later, the British sent divers down into sunken U-boats to salvage new editions of codebooks. During World War Two, the land forces of both sides frequently captured cryptographic documents from the enemy.

Secret agents often steal codes outright. In August 1941, Mussolini's Military Information Service got a wax impression of a key from an Italian employee of the American embassy in Rome. The Italians made their own key and stole the military attaché's copy of the Black Code. They could then read not only his messages but, because the code was used throughout the world, all American military-attaché messages that they could intercept. The most valuable came from the man in Cairo. He was in close contact with the British in North Africa and daily radioed back to Washington detailed reports on British experiences, reinforcements and plans. The Italians picked these up, read them with their stolen code and used the information to foil British moves. Once, the British planned a commando-style attack on Axis airfields in the Mediterranean to reduce air strikes while they pushed a convoy through to besieged Malta. The Germans, forewarned, repelled the British attacks and forced the convoy to turn back.

But sometimes espionage backfires. In 1943, the American Office of Strategic Services (OSS) rifled the offices of the Japanese military attaché in Lisbon. The Japanese detected this and changed their attaché code, depriving the Allies for more than a year of a valuable source of information. So governments turn to juicier means.

An American girl working for British intelligence in Washington let herself be seduced in 1941 by an Italian, obtaining the Italian naval code, and in 1942 by a Frenchman, obtaining the French. The Russians today exert great efforts to entrap code clerks, sometimes setting 20 men on one to discover his weaknesses and exploit them. The Russians do this not only in Moscow, and one can imagine the dilemma of a young and ill-paid Syrian code clerk in an expensive Western capital when approached by a slender blonde who promises him money and herself for a few inconsequential pieces of paper. Such clerks will seldom betray an actual cipher, in whose secrecy they have been drilled, but will often pass over messages in plaintext. A comparison of these with their coded versions will, in many cases, permit a reconstruction of the cipher system and a consequent reading of future messages.

The simplest way to obtain another nation's codes is to sell it one's old code machines. The United States has sold obsolete cipher machines to Turkey, for example, after carefully noting such key elements as the wiring of the code wheels. Turkey accepts this because, wanting mainly to keep her messages secret from Russia and Greece, she gets machines that will do this at a price that she can afford. She either doesn't care about American eavesdropping or talks herself into believing that some changes in the machines will prevent it. After World War Two, Britain rounded up the thousands of Enigma machines that Germany had used and sold many of them to one of the emerging nations. Since she had read the machines in the Forties, she could read them in the Fifties and Sixties--and so could keep tabs on what that country was planning.

•

Does all this mean that no code secret is safe any longer? Have these three factors--the brilliance of cryptanalysts, the power of computers and the assistance of espionage--at long last validated Edgar Allan Poe's famous dictum: "It may well be doubted whether human ingenuity can construct a [cipher] of the kind which human ingenuity may not, by proper application, resolve."

By no means. Of the hundreds or thousands of coded intercepts that flow daily into NSA headquarters, perhaps only four percent are broken. The explanation for this apparent anomaly lies in the development of secret writing.

This development may be viewed as the latest in the ceaseless struggle between the makers and the breakers of codes. The makers, of course, came first. Apparently, when a culture achieves a certain level of literacy, the need for secrecy in writing reaches a critical point. Cryptography thus sprang up spontaneously and independently in the four great civilizations of antiquity and later in many other societies.

Some of the early methods were bizarre. A Persian shaved the head of a slave, tattooed onto the bald pate a message urging his son-in-law, a local governor, to revolt, waited for the hair to grow back and sent the slave off down the road. Some methods were ingenious. The ephors of Sparta wrapped a strip of leather around a wooden staff, wrote their orders down its length, took off the leather, thus jumbling the letters, and dispatched it to their general in the field. He wound it around a baton of the same diameter and read the message. And some early methods were simple. Julius Caesar replaced each letter of his message with one three places down the alphabet, so that A became D, B became E, and so on.

Caesar's elementary cipher sufficed for his day, because the first code breakers did not appear until several centuries later. It was the Arabs who discovered the principles of cryptanalysis. But their knowledge contracted as their civilization declined, and not until the Renaissance did the West rediscover cryptanalysis.

The new nation-states used it to read the messages that the foreign ambassadors in their capitals were sending to their home countries. By the 1700s, clandestine mail-opening and cipher-solving centers called black chambers existed in most of the monarchies of Europe. Often located in curtained, candlelit rooms of post offices, they employed specialists in a variety of dark arts. Some deft fellows opened letters tracelessly, usually by softening the wax seal and then passing a hot wire under it. Engravers took impressions of seals and then forged them. Batteries of secretaries took down letters dictated at high speed so that they could be returned to the mail without missing a delivery. Translators interpreted exotic tongues and cryptanalysts cracked foreign ciphers.

The code breakers bashed away on the principle of letter frequency. In English, for example, the letter E is used more often than any other. So if you have a cryptogram of Caesar's type--A becomes, say, P; B becomes W; and so on--in which X is the most common cipher letter, you can presume that X stands for E. With this as a start, you can proceed rather as if you were solving a crossword puzzle. You fill in what you know and guess at the rest. The word e?e? might be ever or even.

Other aspects of letter frequency help. The second most frequent letter in English is T. Three common letters that rarely contact one another are A, O and I. A high-frequency letter that follows vowels in four fifths of its appearances is N. The letter that most often precedes them is H. Pairs of letters have distinct frequencies as well: The most common is TH. The experienced cryptanalyst picks out letters under their cipher-text disguises as easily as you spot friends at a masquerade party.

This principle placed a mighty weapon in the hands of the code breakers. The code makers soon blunted it with devices of their own. The tussle went back and forth, with cipher inventors tinkering with their systems to fill the chinks probed by the cryptanalysts, but with the code breakers usually on top.

The cryptographers' quest for their Holy Grail, the unbreakable cipher, led at one point to a system that amateurs to this day believe to be the only one that cannot be solved: the book code.

During the American Revolution, when Benedict Arnold was negotiating to betray West Point to the British, he at first encoded his messages by means of volume one of the fifth Oxford edition of Blackstone's famed legal classic, Commentaries on the Laws of England. Arnold searched for the plaintext words in the book and then, when he found each one, wrote down its page number, line number and word number in the line. General, for example, was 35.12.8. But some words took a lot of hunting. Arnold did not turn up militia until page 337. Others he could not find at all but had to spell out, using the same system for letters. The code proved so cumbersome and timeconsuming that the conspirators abandoned it after sending only a single message from each side. Undoubtedly it was, as amateurs believe, unbreakable. But it was also impractical.

The invention of the telegraph a few years later intensified the struggle between the makers and the breakers of codes. But it was radio that brought the struggle to a climax. For radio, theoretically, presents the enemy with a copy of every message that is transmitted. How much this helped the code breakers first became evident in World War One. Battle after battle was decided by the intelligence obtained from cryptanalysis. At 9:05 p.m. on April 28, 1918. for example, American monitors intercepted a coded German message. Cryptanalysts at headquarters quickly broke it, discovering it was an order for an attack at one a.m. Half an hour before the assault, the doughboys were warned--in time to repulse it. On the eastern front, declared a high German staff officer, "We were always warned by the wireless messages of the Russian staff of the positions where troops were being concentrated for any new undertaking. Only once during the whole war were we taken by surprise." Major Joseph O. Mauborgne of the U.S. Army Signal Corps, 36 years old, was himself a cryptanalyst of some expertise. Early in 1918, Mauborgne took a couple of cryptologic ideas that were floating around and combined them. The result was a cipher that could never be solved.

In essence, it works like this:

To begin with, you must have a key. This can be a series of letters, numbers or electric pulses and spaces. The sequence must have two properties (each representing one of the ideas that Mauborgne plucked from the scientific atmosphere). It must be random. In other words, the elements of which it is composed must have absolutely no pattern, no structure whatsoever. And the sequence must be endless. It must have as many elements as are in all the messages you are ever going to send. The key must never repeat; no portions must ever be reused in one message or among several. Both sender and receiver must naturally have copies of the key. For this example, our key will comprise these numbers: 7396407718181563015169....

To encipher, you first transform the letters of your plaintext message into numbers. An easy way of doing this is to let A=01, B=02, etc. Then the message "Attack" will become 01 20 20 01 03 11. You write these numbers under as much of your key as you need. Then you add them to the key numbers. (Using noncarrying addition will reduce errors and permit enciphering from left to right.)

7 3 9 6 4 0 7 7 1 8 1 8

+

0 1 2 0 2 0 0 1 0 3 1 1

___________________

7 4 1 6 6 0 7 8 1 1 2 9

The sum constitutes the cryptogram. When the recipient gets it, he writes it out above the key and subtracts. He comes up with 01 20 20 ..., which he turns into "attack."

But no third party will be able to do that. The lack of pattern and repetition deprives him of any handholds with which he could rip open the cipher. Take the most advantageous case: The cryptanalyst has the actual plaintext of a coded intercept. He could, indeed, recover the key used for that message. But this does him no good whatsoever. Because the key is random and thus entirely unpredictable, the cryptanalyst cannot determine even the next number of the key to use in deciphering other intercepts. It lies forever beyond his ken. The same holds a fortiori for all the other numbers of the key in that message, and in all succeeding messages, since the key never serves twice.

What about trial and error? If the cryptanalyst runs through every possible key, won't he eventually hit upon the right one? He will. And he will also hit upon the right plaintext. But it won't do him any good. For in running through every possible key, he will also be "recovering" every possible message of the same length as the true original in every possible language. For example, with the message 67 83 99 28 01 25 27, key 59 88 79 10 06 24 07 will yield plaintext "retreat," while key 66 89 77 27 97 22 22 yields plaintext "advance." Other keys will yield "sideway," "oranges," "Fuehrer" and "playboy." Because the keys are patternless, nothing permits the cryptanalyst to choose one over another. All he has done through a pointless exercise is to generate a list of all possible seven-letter words in all possible languages that he might simply have taken from a shelf of dictionaries.

The system, then, is truly unbreakable. Mauborgne had achieved the dream of all cryptographers, their version of the philosophers' stone. Code makers in other countries soon reasoned as he had, combined the concepts of randomness and endlessness and independently created the unbreakable cipher. Germany did it early in the Twenties. Her Foreign Office embodied it in its classical form, which has given it its name: the one-time pad. On two sheets of paper were typed a random series of numbers--the key. Many such sheets were then bound into two identical pads, one for Berlin, one for the embassy abroad. After the cipher clerk had used a sheet to encipher a message, he tore it off and threw it away. The decoder did the same.

The Soviet Union, whose diplomatic codes had been solved by Great Britain during the trade negotiations of 1920 that led to Britain's first coming to terms with the Bolshevist regime, shifted to the one-time pad by 1930. Since then, no one has solved Russian diplomatic messages. During World War Two, Russian spies, notably Richard Sorge in Japan, hermetically sealed their radioed reports to Moscow with the one-time pad. Britain's Foreign Office was using it by 1943, and thus minimized the damage done by Nazi Germany's most famous spy. This was Cicero, the Albanian valet of the British ambassador to Turkey, Cicero photographed the embassy's most secret documents and sold them--for counterfeit pounds--to the Nazis. In addition to the information itself, the texts of the cablegrams normally, in other cryptosystems, would have provided the German code breakers with cribs to read other British diplomatic messages. But the one-time pad rendered these plaintexts utterly useless to them. And so when Hitler rejected the photographed documents as improbable, Cicero's work, a technical success, proved a substantive failure.

If the one-time pad can thus confer such enormous security, why don't all nations use it for all their messages?

Because they cannot. As with the book code, practical considerations interfere.

The cipher requires that the key be used only once. Yet in network communications, especially in the haste and turmoil of war, inevitably two units will simultaneously select the same portion of key for use. This will lay those messages open to solution. Moreover, the cipher requires that every message letter have its own fresh key element. In practice, it is impossible to produce and distribute sufficient key. During World War Two, the U.S. Army's European-theater headquarters transmitted, even before the Normandy invasion, 2,000,000 five-letter code groups a day. It would therefore have consumed 10,000,000 letters of key every 24 hours--the equivalent of a shelf of 20 average books. The production and distribution of so much material was out of the question.

•

The invention of the ultimate in ciphers did not, therefore, give the code makers complete victory over the code breakers. The cryptanalysts could still attack those ciphers that had to be used where the one-time pad could not be--and, during World War Two and the postwar struggles with Communist countries, sometimes with history-making success.

One of the most important solutions of all time was that of the German Enigma machine, begun by the Poles before World War Two and raised to a mass-production art by the British during actual hostilities. This solution, whose solved and translated intercepts were cover-named Ultra, contributed enormously to the winning of three crucial battles of the war. In August 1940, Ultra forecast hours before radar did where the German bomber squadrons would appear over England. Fighter Command then concentrated its few Hurricanes and Spitfires to deny the Germans air superiority in the Battle of Britain. From 1943 on, during the Battle of the Atlantic, Ultra disclosed locations where the U-boats were refueling from their milch cows. Long-range flying boats did the rest. And Ultra's insight into the German spy apparatus enabled Britain to control every Nazi agent on the island, thereby fooling the Germans about D day--they held an entire army, the 15th, around Calais while the real invasion securely lodged itself in Normandy.

And after that conflict was over and the iron curtain clanged down, code breaking retained its importance. During the Korean War, cryptanalysis helped spot targets for air strikes. In Vietnam, radio intelligence was the only really valuable intelligence that the U.S. Military Assistance Command had.

In diplomatic negotiations, code breaking likewise helped the United States, though peacetime results naturally could not match those of wartime for drama. During the Cold War, NSA cracked some of the codes of more than 40 nations, among them Italy, Turkey, France, Yugoslavia, Indonesia, Uruguay and half a dozen countries of the Near East. "I had in my desk," one former NSA cryptanalyst declared, "all the deciphered communications between Cairo and its embassy in Moscow relating to the visit of the U.A.R. government mission to the U.S.S.R. in 1958 for the purpose of purchasing petroleum in the Soviet Union." Henry Cabot Lodge, then United States Ambassador to the United Nations, once expressed his appreciation to NSA for information about the instructions sent by the Near East governments to their UN missions. (The presence of the United Nations in New York makes it easy for NSA to intercept member nations' cablegrams.) One former high State Department official was always glad to see the man with the locked briefcase who brought around the intercepts. "I got some good clues on how to deal with various countries," he said, "and I quickly learned which ambassadors I could trust and which not." And when, every morning at 7:45, Lieutenant General Brent Scowcraft takes President Ford the latest intelligence, included are solutions from NSA.

How Ford feels about the work was underlined this past summer. Turkey, peeved at Congress' refusal to give her arms for possible use in Cyprus, ordered the United States to close down its four main intercept posts there. Those posts, located at places bearing such romantic names as Karamursel, had nestled close under the belly of the Soviet Union. The President declared that Congress' "reckless" action had caused "the loss of strategic intelligence data," which "in today's world is absolutely essential to our national security, even our survival."

Valuable as it is, however, code breaking cannot supply perfect and complete intelligence. The enemy does not put everything on the air. Some plans are discussed in conference. Some orders are sent by courier. The Japanese attack on Pearl Harbor came as a surprise because no orders were given by radio.

Of the messages put on the air, not all are intercepted. U.S. monitors did not pick up the messages that might have told them that the North Koreans were going to attack in 1950 because they were targeted instead on the more promising and apparently more vital Russian transmissions. No agency has the manpower to monitor constantly every wave length on the radio spectrum. And, finally, of the messages intercepted, not all are solved. During World War Two, Germany's Army Group North intercepted 46,342 cryptograms opposite Leningrad in the 13 months beginning in May 1943. Cryptanalysts solved only 13,312, or 28.7 percent. Here, too, lack of manpower was undoubtedly a main factor. But the Russians also changed their codes before the Germans had collected the quantity of messages needed to crack them.

In addition to inherent limitations like these, advances in mathematics and electronics are today further reducing the usefulness of code breaking.

Within the past decade, mathematicians have developed powerful new formulas for generating keys. So complex are these formulas that, even given a cryptogram and its plaintext, the cryptanalyst would need centuries--even with all the world's computers of this and the next generation--to reconstruct them and then use the reconstruction to read the next message that comes along. In theory, the ciphers are not unbreakable, as the one-time pad is. In practice, they are. These formulas are embodied in electronic cipher machines, such as the United States' HW-28 and KW-7, which include a further security feature. Each machine has a key element of its own, making it unique. Hence, stealing one of these well-guarded machines would bare only the messages sent to and from that machine.

Moreover, the transistor and large-scale integration of circuits on quarter-inch chips, which have made it practicable to utilize these formulas in cipher machines, are becoming increasingly available to other nations of the world. This means that more and more countries are achieving absolute security for their high-level dispatches.

That is why NSA fails to solve more than 95 percent of the messages it intercepts. Those it does break usually constitute the medium- or low-level traffic of major nations or the top-level traffic of the smaller countries. This is what it circulates to the officials of the State Department and the National Security Council. It continues to intercept and store the major messages of major nations--sometimes in boxcars at Fort Meade, sometimes just in cardboard boxes--in the hope that cipher clerks will err, permitting some kind of break, or that some other fortunate circumstance will arise.

The attempt to recover the Russian submarine was viewed by NSA as just such a chance. In fact, it was a desperation gamble. It is doubtful that the Soviet Union is behind the United States cryptologically, and so it is unlikely that the CIA would get more from the Soviet sub's cipher machine than the Russians would get from an American cipher machine. This means, in effect, little more than a few messages, many of them probably personal, sent to the submarine, and probably none from it, since the essence of its mission is to remain silent and hidden in the depths of the sea.

Why, in view of these generally mediocre results, does NSA persist? Why does it bother to read the systems of these minor countries? One reason, of course, is that cryptanalysts, like other mortals, want to protect their jobs. Their motivation may be even stronger than most, since they cannot readily transfer their skills to the civilian sector. Another is that NSA gives the policy makers a certain assurance that they're not missing anything. These small countries may suddenly become important someday--witness Korea, the Congo, Vietnam. Most of the intercepts are admittedly of little interest or importance, the operation is admittedly a bit of a luxury and a waste, but the Government can afford it and it does provide a margin of safety, so why not? The real question, however, is whether it is worth the billions spent on it.

The answer depends on what the money would otherwise be used for. If the Government were to spend it on some more jet fighters or ICBMs, probably the NSA investment is better. Intelligence is cheap and cost-effective. It can often save more than it costs. But if the Government were actually to spend the money on schools and hospitals and transporation, that investment is probably better. For a nation's strength depends far less upon its secret intelligence than upon its human and material resources. No doubt a balance is best. The problem is to strike that balance, and this depends largely on the wisdom and determination of a country's leaders, and of its people.