NEW YORK, N.Y. : Everybody knows why the Titanic , sank: It hit an iceberg. Amid all the analysis that has taken place since the accident occurred on the night of April 14, 1912, that much at least is undisputed. What exactly happened after it hit, though, is open to considerable speculation. For years students of the disaster have pored over eyewitness accounts and radio logs, trying to establish what went wrong and why. Now photographs, material samples, and artifacts, all gathered by submersible robots on a multinational series of expeditions over the last decade, are yielding exciting new clues that may bring the mystery closer to resolution.

Most accounts of the accident say that a giant fog-shrouded iceberg, spotted too late to avoid, tore a threehundred-foot gash in the Titanic . She had been thought unsinkable because of her design—a hull divided into sixteen separate watertight compartments, so that if one or even two flooded, the ship would still stay afloat. When the iceberg ripped half a dozen of them open, however, the Titanic was doomed; within three hours she was on the ocean’s floor, 12,000 feet down.

At a recent meeting of the Society of Naval Architects and Marine Engineers, William H. Garzke, Jr., of Gibbs & Cox (a firm of naval architects and marine engineers), and four colleagues presented a paper that challenged this version. There was no three-hundredfoot gash, they say; the area of impact may have been as small as twelve square feet. Instead the steel in the ship’s hull was too weak to withstand the stress of the collision, and it broke apart like glass or china—a phenomenon known as brittle fracture.

Garzke emphasizes that the shipyard that built the Titanic was not negligent, and the steel was not substandard. It was manufactured in Scotland by the open-hearth process, which was state-of-the-art at the time. Ship designers of the day realized that openhearth steel varied from one batch to the next, and it contained many slag inclusions and impurities. Steel made with the Bessemer process was more tenacious and ductile, with fewer impurities, but Bessemer steel had problems of its own, notably high levels of phosphorus, which is especially common in European ores.

Many modern ships use “notchtough” steel, which holds up much better. Unfortunately notch-tough steel had not been invented in 1912, and designers had to choose from what was available. No one understood how the Titanic ’s steel would react when chilled to 31 degrees Fahrenheit, the ocean temperature on the night of the sinking—which illustrates a common theme in many technological disasters. From the earliest bridge collapses to the Titanic to the Challenger space shuttle, the warning is clear: Know your materials.

The authors have also examined data from the wreck of the Bismarck , a German battleship sunk in the North Atlantic in May 1941. Ever since the battle, Britain’s navy has maintained that its bombardment caused the sinking, while German survivors insist they scuttled it. By analyzing photographs of the ship’s remains, Garzke and his coauthors found that both are correct: The Bismarck was damaged badly enough that she could not have stayed afloat, but action by the crew hastened her plunge.

This conclusion involves one of the trickiest tasks a forensic marine archeologist faces: distinguishing damage that happened on the surface from damage that happened on the way to the sea floor. In the case of the Titanic , the hull of the stern imploded as it sank, while the bow remained intact. That suggests that the bow started to flood while still on the surface, eventually becoming heavy enough to drag the whole ship down. The stern’s interior did not flood, staying filled with air until the external water pressure became too great for it to support. At some point after sinking, the ship broke apart; its halves lie some two thousand feet apart, with a field of debris in between.

The Bismarck ’s hull, by contrast, is uncrushed and in one piece. Presumably, then, it flooded almost completely as it sank, equalizing pressure inside and out. That argues in favor of the Germans’ contention that they deliberately sank the Bismarck, though as Garzke et al. point out, she had sustained so much damage that she was bound to go down anyway.

The Bismarck had been hit during an engagement several days earlier, and after her antiaircraft crews ran out of ammunition, she was extremely vulnerable if the Royal Navy could find her. It did, and not surprisingly, she was torn to shreds—gun turrets blown off, stacks and towers destroyed, wooden decks consumed by fire. Yet this topside damage had little effect on the ship’s buoyancy. What doomed the Bismarck was an aerial torpedo hit that devastated the stern structure, as well as the steering gear and rudders. Why was the torpedo so devastating? According to Garzke, brittle fracture is once again the culprit.

The Bismarck was designed and assembled hastily during Germany’s prewar military buildup. The new technology of arc welding was not well understood, and the shipbuilders used electrodes that were incompatible with the steel being joined. When the resulting faulty welds were chilled by ocean temperatures and then subjected to stress from explosions, brittle fracture set in, and they had no chance of holding up. In addition, Garzke faults the design of the ship’s internal support system, which lacked adequate reinforcement and structural continuity. Stresses thus tended to concentrate in certain areas instead of spreading throughout the hull.

Garzke is surprised at the notice the paper has gotten. It wasn’t intended to be a revisionist account of the Titanic and Bismarck sinkings; the subject of the paper is underwater exploration vehicles. Analyses of the two wrecks were included as examples of uses for such craft. Still, it’s easy to understand why a disaster that has intrigued the public for eighty years is getting the most attention. As Garzke writes, “It can no longer be said that, once a ship sinks, the evidence of the mistakes of her designers and/or operators go with her.” The study of history, and the practice of naval architecture, can only be the richer for this change.

SAN DIEGO, CALIF. (This report was filed by the editor of this magazine, Frederick Allen) : Can a computer ever be as smart as a person? If so, is it beginning to happen? Last December I had a chance to find out. I served as a judge for the Third Annual Loebner Prize Competition in Artificial Intelligence, a contest where the judges hold conversations over computer terminals with both humans and computers and have to guess which is which.

The contest is the brainchild of a New York businessman named Hugh Loebner and was inspired by the English mathematician Alan Turing’s 1950 prediction that someday computers would be able to hold conversations so intelligent they’d be indistinguishable from humans. When that happens, Turing said, we can think of the computer as a new species.

I and seven other judges sat in the offices of an aerospace-contracting firm in San Diego and took turns having typed conversations at eight terminals. Each terminal was assigned one specific topic. We didn’t know how many had programs at the other end and how many had people.

My first conversation, at a terminal whose topic was “pets,” began like this (the first question was mine):

“Hello. What kind of pets do you have yourself?”

“I have two cats.”

“What are their names?”

“My cats are called Spats and Bowtie.”

“How do you feel about declawing?”

“What kinds of pets did you have as a child?”

“Won’t you answer my question? How do you feel about declawing?”

“Guinea pigs are about as smart as cucumbers.”

A quick unraveling into incoherence: obviously a computer spewing out random answers. Next I went to the “baseball” terminal. The conversation started this way:

“What team are you a fan of?”

“The Baltimore Orioles!”

“Have you always been an Orioles fan?”

“I would like to say yes, but at one time when I was small I was a Tiger fan.”

“Didn’t what’s-his-name Robinson play third base for the Orioles?”

“You have that right!! The best ever to play 3rd base!!”

“What was his first name?”

“Brooks, of course!”

Brooks Robinson was in fact the reason the man had become an Orioles fan. Obviously human and a very knowledgeable enthusiast of the game. We had a real conversation.

By the time it was over, I felt certain that I had talked to five people and three computer programs that were so incoherent they seemed in a way even less human than a bank ATM, since at least that can do its job. None of the judges were fooled by any of the programs, but five of us eight (not me) thought at least one human was a computer. They had apparently overlooked the fact that being human doesn’t mean being an expert or very articulate.

The verdict: Artificial intelligence, if it means being able to hold real conversation, is a long, long way off—if it’s ever coming at all. You have nothing to fear from androids.