In August 2003, a massive power failure plunged much of the eastern United States into darkness, disrupting traffic, emergency services, food preparation, medical care delivery, and life in general for millions of people. A crisis of such enormous proportions hardly seemed the time to think about butterflies, yet the August 15, 2003 front-page story about the crisis in the San Francisco Chronicle was entitled “How a butterfly’s wing can bring down Goliath.” Keay Davidson, a science writer for the paper, was of course referring to chaos theory— the idea that, in a complex system, such as an overloaded and antiquated power grid, an infinitesimal change can bring about a total collapse of the system. As for the butterfly effect itself, Davidson explained, “In the 1960s, MIT meteorologist Edward Lorenz popularized the notion of the butterfly effect. An infinitesimal shift in the weather—say, the turbulence caused by a butterfly flapping its wing—can set in motion atmospheric events that climax in a hurricane. Such events are for all practical purposes unpredictable.” (h t tp : / /www.s fga te . com/cg i -b in / a r t i c l e . c g i ? f i l e = / c / a / 2 0 0 3 / 0 8 / 1 5 / MN191082.DTL). As a lepidopterist of sorts, I was intrigued by the butterfly reference and by Professor Lorenz, so I thought the metaphor warranted further investigation. As it turns out, Lorenz was a meteorologist in the early 1960s who experimented with computer simulations of weather (on, entomologically enough, an early computer called a “Royal McBee”) (http://www.zeuscat.com/andrew/chaos/ lorenz.html). He devised a series of twelve differential equations to account for various meteorological phenomena; for his model, he entered a series of variables and then ran recursive equations to generate different outcomes. One eventful day, in an effort to recreate a particular weather pattern, Lorenz entered the values recorded on a printout from the middle of the earlier run, but once the next run had completed its course, he obtained a different outcome. The difference was due to the fact that, while the program calculated values to six significant digits, the printout displayed values with only three significant digits. Although the difference between the two runs was tiny (one part in one thousand due to rounding error), because of the iterative nature of the calculations, the tiny error had been amplified until the ultimate outcome was completely different. Lorenz recognized that this “sensitive dependence on initial conditions” might have broader implications. He presented the concept rather obliquely in a paper delivered in 1963 to the New York Academy of Sciences, in which he quoted a fellow meteorologist as remarking that “if the theory were correct, one flap of a seagull’s wings would be OK LISTEN