The Third Twin Page 0,31
critics haven't failed to make."
"On the other hand, it's not possible to find out about aggression and criminality by studying law-abiding Middle American families. So it was absolutely crucial to my project that I solved the recruitment problem."
"And have you?"
"I think so. It occurred to me that medical information about millions of people is nowadays held on huge databases by insurance companies and government agencies. That includes the kind of data we use to determine whether twins are identical or fraternal: brain waves, electrocardiograms, and so on. If we could search for pairs of similar electrocardiograms, for example, it would be a way of identifying twins. And if the database was big enough, some of those pairs would have been raised apart. And here's the kicker: Some of them might not even know they were twins."
"It's remarkable," Berrington said. "Simple, but original and ingenious." He meant it. Identical twins reared apart were very important to genetics research, and scientists went to great lengths to recruit them. Until now the main way to find them had been through publicity: they read magazine articles about twin studies and volunteered to take part. As Jeannie said, that process gave a sample that was predominantly respectable middle-class, which was a disadvantage in general and a crippling problem to the study of criminality.
But for him personally it was a catastrophe. He looked her in the eye and tried to hide his dismay. This was worse than he had feared. Only last night Preston Barck had said, "We all know this company has secrets." Jim Proust had said no one could find them out. He had not reckoned with Jeannie Ferrami.
Berrington clutched at a straw. "Finding similar entries in a database is not as easy as it sounds."
"True. Graphic images use up many megabytes of space. Searching such records is vastly more difficult than running a spellcheck on your doctoral thesis."
"I believe it's quite a problem in software design. So what did you do?"
"I wrote my own software."
Berrington was surprised. "You did?"
"Sure. I took a master's in computer science at Princeton, as you know. When I was at Minnesota, I worked with my professor on neural network-type software for pattern recognition."
Could she be that smart? "How does it work?"
"It uses fuzzy logic to speed up pattern matching. The pairs we're looking for are similar, but not absolutely identical. For example, x-rays of identical teeth, taken by different technicians on different machinery, are not exactly the same. But the human eye can see that they're the same, and when the x-rays are scanned and digitized and stored electronically, a computer equipped with fuzzy logic can recognize them as a pair."
"I imagine you'd need a computer the size of the Empire State Building."
"I figured out a way to shorten the process of pattern matching by looking at a small portion of the digitized image. Think about it: to recognize a friend, you don't need to scan his whole body - just his face. Automobile enthusiasts can identify most common cars from a photograph of one headlight. My sister can name any Madonna track after listening to about ten seconds of it."
"That's open to error."
She shrugged. "By not scanning the entire image, you risk overlooking some matches. I figured out that you can radically shorten the search process with only a small margin of error. It's a question of statistics and probabilities."
All psychologists studied statistics, of course. "But how can the same program scan x-rays and electrocardiograms and fingerprints?"
"It recognizes electronic patterns. It doesn't care what they represent."
"And your program works?"
"It seems to. I got permission to try it out on a database of dental records held by a large medical insurance company. It produced several hundred pairs. But of course I'm only interested in twins who have been raised apart."
"How do you pick them out?"
"I eliminated all the pairs with the same surname, and all the married women, since most of them have taken the husband's name. The remainder are twins with no apparent reason for having different surnames."
Ingenious, Berrington thought. He was torn between admiration of Jeannie and fear of what she could find out. "How many were left?"
"Three pairs - kind of a disappointment. I was hoping for more. In one case, one of the twins had changed his surname for religious reasons: he had become a Muslim and taken an Arab name. Another pair had disappeared without a trace. Fortunately, the third pair are just what I was looking for: Steven Logan is a law-abiding