The Day We Lost the H-Bomb: Cold War, Hot Nukes, and the Worst Nuclear Weapons Disaster in History

Or, even worse, suppose a Soviet submarine slipped into the search area and released a timed nuclear device? The bomb would explode, and everyone would point fingers at the Americans.

The Lotsman stayed on scene until early March, usually cruising between five and eleven miles away from Alfa 1. Then she vanished. Nobody knew what she had learned during her stay.

From Washington, Guest's Technical Advisory Group kept a close eye on the developments in Spain.

Even if the bomb had fallen into the sea, Guest might never find it. If the admiral came up empty-handed, the Navy would have to stand before Congress — and the secretary of defense — and explain why it had spent so much money on an unsuccessful search. Heads would roll.

The TAG understood this clearly. The advisers were not only sending gear to Spain, they were also thinking about the endgame. If the search failed and the Navy brass were hauled before Congress, they would need proof that Guest had done everything possible to find the bomb. Or at least they would need something that seemed like proof-some fancy numbers to wave in front of the politicians. What they needed, they decided, was math.

John Craven of the Technical Advisory Group called Captain Frank Andrews, who had overseen the search for the USS Thresher, and asked for assistance. Andrews had retired from the Navy but was happy to help. He suggested that Craven call Wagner Associates, a small consulting firm outside Philadelphia. Soon Dan Wagner, the owner of the company, was flying to Washington with a member of his staff, a probability expert named Tony Richardson.

In Washington, Craven briefed the two mathematicians on the situation and gave Richardson a rough

“probability map” that he had sketched. The map, which showed the area off the coast of Palomares, resembled a contour map. However, the contours on Craven's map showed not the height of a mountain ridge or the depth of an ocean trench but the probability that the bomb had fallen into certain points in the sea. Craven hadn't had much information when he had drawn the map, so his initial stab basically outlined what everyone already thought: that the bomb lay either right off the beach or somewhere near the fishermen's sighting. Craven gave Tony Richardson a copy and sent him and Frank Andrews to Spain.

On the plane to Madrid, Richardson sat next to Andrews and discussed his strategy, sketching out ideas on graph paper. He knew basically how to run a systematic search — mathematicians had been working on search theory since at least World War II. First he had to develop a probability map laying out where the bomb might be hiding. Second — this was the tough part — he had to find a way to evaluate the search as the Navy carried it out. And not just say “good” or “bad” but quantify the search, evaluate it mathematically. Then, as the search continued and new information came in, he would update the probability map, hopefully narrowing down the search area. Richardson could keep the analysis going until the Navy found the bomb or gave up the search.

On the plane, Richardson explained his system to Andrews. He thought he could call it “search failure probability.” In other words, after the Navy had searched a given area, this was the probability that the bomb was there but the Navy had failed to find it. Andrews shook his head.

Tony, he explained, you have it all backwards. You are dealing with the Navy. You can't talk about failure! You need to talk about success. Richardson objected, showing Andrews a sample probability he had plotted on his graph paper. Andrews looked at it and frowned. The line that Richardson had drawn sloped downward toward the bottom of the page. It looked like a business with a bad quarter or a stock market crash. No, no, no, Andrews explained. In the Navy, graphs need to point up.

Richardson and Andrews reported to Admiral Guest on February 22. Richardson's reputation had preceded him. The ship had prepared for the arrival of the distinguished mathematician, assigning him a generous stateroom with a private sink and stewards. So Guest was a bit taken aback to discover that Dr. Richardson was a baby-faced twenty-seven-year-old who looked even younger than his age.

Eyeballing this new member of his team, Guest asked Dr. Richardson what he could do for the mission. Richardson launched into a description of his plan — now called search effectiveness probability — and an explanation of Gaussian probability distributions. As Guest's eyes glazed over, Andrews stepped in and cut Richardson off. After the admiral escaped, Andrews turned to Richardson. Would this kid ever learn? “Tony,” he said, “don't talk about Gaussian distributions to an admiral!” Later, Guest pulled Frank Andrews aside. “Where the hell did you get this high school kid?”