Ed Lorentz, author of perhaps the most famous meteorological paper ever, died yesterday. He found that perfect weather forecasts were impossible, because even small errors in observations led to dramatic changes in model results. How great is it that the most cited meteorological work is a negative result, one that proves that the perfect forecast is impossible? You do know of his work, even if you haven't heard it described in these terms before -- he was the author of the notion that a flap of a butterfly wing in Brazil could cause a tornado in Texas, a notion that led to chaos theory and to need for probabilistic forecasts.
Ed Lorentz gave a general lecuture at the lab (or was it the university?) a few years ago, to a very crowded room. He provided a simple introduction to Lorentz attractors, the starting point for chaos theory. In these days of fancy multi-media presentations, he used transparencies (someone had to go dig out one of those overhead projectors) with computer printouts of simple graphs. And his talk was no less lucid for that.