In the days of King Arthur, Modred, the king's son, rebelled and raised an army to overthrow his father. Two great hosts of knights met on the field of Camlan, but at the last moment father and son decided not to fight. They called a truce and sent spokesmen forward to seek an agreement. Meanwhile, each army, suspecting a trick, stood poised.

Negotiations were proceeding smoothly until a snake, slithering in the grass, suddenly bit one of the knights. The knight cried out and drew his sword to kill it. The assembled armies mistook this as a signal for battle and sprang to the attack. By day's end, all but two of the 100,000 warriors lay dead. King Arthur and his son fought and killed each other, and with them perished Camelot.

Today the United States and the Soviet Union face a similar danger. Each side has marshaled enormous military forces poised to strike. Fearing total mutual destruction, the two sides have been talking, but suspiciously and sometimes very little. Yet a regional conflict, a terrorist act or an accident could ig- nite a deadly confrontation.

For decades, government officials, military strategists and the public have focused on the danger of deliberate nuclear attacks. "Today," as the late senator Henry Jackson (D-Wash.) said in a speech two years ago, "it is more and more being recognized that a nuclear war could break out even though neither side wanted it. It could break out not by deliberate intent, but by accident or misunderstanding."

The arms talks, which will resume in two weeks in Geneva, focus on reducing nuclear weapons. Reductions are vital, but they cannot stop human error. "Even the most optimistic arms negotiator agrees that for decades ahead we will be living in a world with tens of thousands of uclear weapons, a few hundred of which could destroy us," says Robert S. McNamara, secretary of defense under Presidents Kennedy and Johnson and former president of the World Bank. Arms reduction alone won't keep us secure. "Therefore," McNamara says, "improving crisis management is an absolutely essential step toward reducing the risk of nuclear war."

THE MOST UNEASY days in the nuclear age gave birth to the first crisis control measure -- the so-called hotline. The Cuban missile crisis of October 1962 brought home the lesson that in times of great hostility the leaders of the Soviet Union and the United States must be able to talk to each other. During the crisis the two superpowers had to rely on an American Broadcasting Co. reporter and an open radio broadcast to relay extraordinarily sensitive messages. So on June 20, 1963, Washington and Moscow agreed to install the hotline, a teletype whose Washington terminal is in the Pentagon with an extension in the White House.

Every even hour on the hour an American officer sends a test message, usually light poetry or prose, to his Soviet counterpart in Moscow. Every odd hour on the hour, he receives a message back. The hotline has been used in crises such as the Middle East war of 1967 and the Lebanese conflict of 1982. Last July, American and Soviet negotiators agreed to add to the hotline the capability to transmit documents.

But the hotline is a crude device for sophisticated communications. The time available for leaders to respond to a crisis is shrinking as the flight times of missiles grow shorter, their accuracy increases and each side considers targeting missiles to kill the other's leaders.

There are too many ways an unexpected nuclear crisis could erupt: the superpowers could become embroiled in their allies' wars, a missile might be fired by accident or fly an errant path, a terrorist group might detonate a bomb mistaken for an attack by a superpower.

The growing sensitivity of both Soviet and American warning systems creates a danger in times of crisis that a precautionary alert by one side will set off a similar alert by the other, which will in turn provoke the first side to increase its state of alert, and so on toward war.

Washington and Moscow have instituted safeguards against mistaken use of nuclear weapons, but in time of acute crisis many safeguards may be removed to make weapons more readily usable.

The most worrisome danger of accidental war is the possibility that several unexpected events could occur si- multaneously at a moment of severe U.S.-Soviet tension. Each event could interact to produce effects none could by itself. In his 1983 book, The Command and Control of Nuclear Forces, Yale professor Paul Bracken describes such an instance. In 1956, just as the Hungarians were revolting, the British and French tried to retake the Suez Canal from Egypt, and the Soviets threatened to destroy London and Paris with nuclear missiles.

"The headquarters of the U.S. military command in Europe received a flash message that unidentified jet aircraft were flying over Turkey and that the Turkish Air Force had gone on alert in response," Bracken wrote. "There were additional reports of 100 Soviet MiG15s over Syria and further reports that a British Canberra bomber had been shot down, also over Syria. (In the mid- 1950s only the Soviet MiGs had the ability to shoot down the high-flying Canberras.) Finally, there were reports that a Russian fleet was moving through the Dardanelles . . . The White House reaction to these events is not fully known, but reportedly Gen. Andrew Goodpaster was afraid that the events 'might trigger off all the NATO operations plan.' At this time, the NATO operations plan called for all-out nuclear strikes on the Soviet Union.

"As it turned out, the 'jets' over Turkey were actually a flock of swans picked up on radar and incorrectly identified and the 100 Soviet MiGs over Syria were really a much smaller routine escort returning the president of Syria from a state visit to Moscow. The British Canberra bomber was downed by a mechanical difficulty, and the Soviet fleet was engaging in a long-scheduled exercise."

If this coincidence had been suggested as a "scenario," it might have been dismissed as too improbable.

Two and a half years ago, the U.S. Arms Control and Disarmament Agency asked a group of scholars at Harvard, including Dr. Richard Smoke and myself, to study the problem of improving U.S.-Soviet control in such crises. We began by examining past tense moments and we asked policymakers in Washington and Moscow these questions: "If you were a leader on the verge of a serious crisis with the other superpower, what might you wish you had discussed beforehand with your counterparts on the other side? What joint institutional arrangements might you wish you had in place?"

Our study had its real beginnings in 1981, when Sen. Sam Nunn (D- Ga.) asked the Strategic Air Command if the United States could recognize a "diguised third-country attack," a nuclear strike by a third nation that deceptively appeared to have come from the Soviet Union. The answer from SAC was not encouraging: it recommended major improve in detecting such a ruse.

Then Nunn, Jackson and Sen. John Warner (R-Va.) proposed establishing a crisis control center. The idea drew support from such well-known defense specialists as Bobby Inman, Brent Scowcroft and James Schlesinger.

Recent events have pushed the concept closer to reality:

* In March 1984, Secretary Konstantin Chern powers "not to allow situations fraught with danger of nuclear conflict," adding, "if such a danger exists, [they should] hold urgent consultations to prevent a nuclear conflagration."

* In June, the Senate urged the president to negotiate with the Soviets to establish crisis control centers in Washington and Moscow.

* In July, agreements were reached to improve the hotline.

* In September, President Reagan, speaking to the United Nations, proposed regular Cabinet-level meetings as well as "periodic consultations at policy level about regional problems . . . to help avoid miscalculations [and] reduce the potential risk of U.S.-Soviet confrontation."

* In November, the Aspen Institute International Group, which included former European heads of state, called for creating a network of crisis control centers.

THE MOST TANGIBLE way to go beyond the hotline is to establish two centers, one in Washington and one in Moscow, each jointly staffed by American and Soviet crisis control specialists. The centers would be electronically linked by telephone, computer, facsimile transmitters and teleconferencing. A dozen military and diplomatic officers from each side would work together at each center, a skilled staff ready on an instant's notice to engage in intensive problem-solving.

Imagine the centers' possible use in a future Middle East crisis. War, let us say, breaks out between Israel and Syria. As in October 1973, the war escalates and Soviet forces prepare to enter the fray. The next day Washington calls a worldwide nuclear alert. That night a nuclear explosion devastates downtown San Francisco.

The president and his advisers face extraordinary un it the Soviets? The act of a terrorist group? A U.S. weapons accident?

The hotline prints out a message from Moscow. The general secretary disavows any responsibility and offers his sympathies. He says he has directed the Soviet staff at the crisis control centers to cooperate in providing proof it was not a Soviet attack.

Many in Washington are suspicious. Soviet submarines are known to be in waters close to the Pacific Coast. The Pentagon insists on a higher alert. The centers swing into action. A teleconference begins among staff officers in Moscow and Washington. The American officers in Washington and the Soviet officers in Moscow have many lines of communication into their respective military services and intelligence agencies, through which questions and information now pour. The meeting goes on for hours.

Is the San Francisco scenario likely? Not at all. Is it possible? Yes.

EVEN MORE USEFUL than such crisis management is crisis prevention. At crisis control centers, the United States and the Soviet Union could act jointly to prevent nuclear proliferation, nuclear terrorism or a nuclear attack from a third nation. A precedent exists. In August 1977, the Soviets tipped off the U.S. government that South Africa was planning to test a nuclear device. A strong but quiet American protest followed, and no test took place.

There is, of course, another side to this. Crisis centers could be misused for intelligence gathering or deception. There is no clear protection for misinformation but there are precedents to guide the wary. For example, the American delegation to the U.S.-Soviet Standing Consultative Commission, the monitoring body for SALT I, independently checks Soviet information for accuracy and uses information filters to protect against intelligence leaks. And in any case, the risk of deception is not significantly greater in a staffed center than it is in any com- munication medium, includ- ing the hotline. Face-to-face communication may offer the opportunity to more effectively challenge statements.

Clearly the arrangement should not depend on goodwill, but as Nunn said: "You don't have to trust the Russians to do this; you only need to make an assumption that they are not madmen, and they will act in their own interests."

THERE IS MORE that can be done beyond crisis control centers. The United States and the Soviet Union could, for example, negotiate an agreement on incidents involving aircraft.

When a Soviet fighter shot down Korean Air Lines Flight 007 18 months ago, killing all 269 passengers, a wave of outrage spread throughnsions between the superpowers.

Although no serious escalation took place, the sobering "What ifs?" remained: What if the plane had been American with 269 U.S. citizens aboard? What if the attack had come during severe international tension?

The Korean airliner was by no means an isolated incident. Five months ago an American aircraft carrying 200 passengers drifted 500 miles off course and came within 15 minutes of overflying the Soviet Union's Kola Peninsula near Norway, where the Soviets have supersecret military installations. In December, a Soviet cruise missile went astray, flew over northern Norway and finally crashed in Finnish Lapland.

These two recent examples indicate that as hostile superpowers confront each other around the globe, the triggers for accidental war are many.

There is a promising precedent for creating safety mechanisms against these triggers: the Incidents at Sea Agreement reached in 1972.

The accord was reached because Soviet and American naval vessels and tracker planes follow each other, sometimes dangerously closely, all over the world. In the late 1960s and early 1970s this practice had turned into a game of chicken on the high seas. The intimidating behavior included the aiming of ship guns and missiles at an offending vessel. Since 1972, however, the encounters are much rarer.

Captains in both navies now have rule books that tell them how to communicate with each other and how to avoid accidents. Every six months high-ranking American and Soviet officers meet to review the process. This low-profile professional communication goes on even at times of high tension. Many fear we cannot work with the Soviets. The truth is we already do.

Had an Incidents in the Air Agreement existed 18 months ago, when KAL 007 strayed into Soviet airspace, the Soviet Air Defense Command might have immediately queried American experts at the Moscow crisis control center about the nature and mission of the Korean airliner.

Working rapidly the American experts in Moscow might have identified the plane as a civilian 747. They could then have worked together with the Soviets to make radio contact with the plane and direct it to a landing spot for Soviet inspection.

ANOTHER STEP to take beyond crisis centers would be to brief the president extensively on crisis control techniques.

Just before 8 a.m. on June 5, 1967, the telephone rang in Lyndon Johnson's bedroom at the White House. Secretary of Defense Robert Mcssage never heard before by an Amerian president. "Mr. President," he said, "the hotline is up. The Soviet premier wants to speak with you."

"Well," mumbled the groggy president, "What should I say?"

The fate of millions might rest on a president's ability to communicate effectively with his counterpart in the Soviet Union during a crisis. There is, moreover, no second chance. Yet presidents receive little preparation for such a role.

A body of knowledge about crisis management does exist. In part it has been captured in the memoirs of participants in past crises. In part it lies as yet untapped in participants' minds.

On entering office, each new president is briefed on the nuclear options at his disposal. President Reagan in addition once observed a war game that pitted the United States against the Soviet Union in an acute confrontation over Central Europe. But a president receives no detailed briefing on how to defuse an intense crisis.

The idea of a crisis control briefing is not new. After the shock of the Cuban missile crisis, Walt W. Rostow and McGeorge Bundy decided to organize a "crisis game" for President Kennedy. A date for the game was finally set for early December 1963, but Kennedy was killed a month earlier.

In past crises, American participants have often wished they knew more about the Soviets. A presidential crisis control briefing could discuss the Soviets in detail: the strengths and weaknesses of the leaders, their systems for making decisions in crisis and their response to high-risk situations.

The crisis centers, the Incidents in the Air Agreement, and the presidential briefing, together with the hotline and other measures, would constitute key elements of a "crisis control system" much as fire stations, hydrants, emergency exits and smoke detectors make up a fire control system. Like fires, crises are endemic. Some are accidental, others are deliberate. In either case they can be stopped before they are out of control.

A runaway crisis is the greatest nuclear danger. Like the passengers aboard KAL Flight 007, we are all passengers on a fragile craft vulnerable to the consequences of miscalculation, fear, miscommunication and blunder. Crisis control is one way to step back from the brink.