QUANTUM ERROR CORRECTION has been experimentally demonstrated for the first time, greatly advancing the promise of carrying out interesting calculations with quantum computers (Updates 310 and 367). Skeptics have maintained that quantum computers would crash before carrying out a useful calculation since the devices rely on fragile, easily corrupted quantum states. Proposed in 1995 and developed unceasingly since then, quantum error correction has been all theory up until now. Aiming radio-frequency pulses at a liquid solution of alanine or trichloroethylene molecules, researchers at Los Alamos and MIT (Raymond Laflamme, 505-665-3394) spread a single bit of quantum information onto three nuclear spins in each molecule. Spreading out the information made it harder to corrupt. The bit of information was a combination or "superposition" of the values 0 and 1, so that it represented a little amount of 0 and a little amount of 1 at the same time. Measuring the spins directly would destroy this superposition and force the bit to become a 0 or a 1. So, the researchers instead "entangled" or interlinked the properties of the three spins. This allowed them to compare the spins to see if any new differences arose between them without learning the bit of information itself. With this technique, they were able to detect and fix errors in a bit's "phase coherence," the phase relationship between the quantum waves corresponding to the 0 and 1 states.(D.G. Cory et al., Physical Review Letters, 7 Sept 1998.)