Traditionally, when clinical trials of a drug produced negative results--that is, failed to show that treatment A is better than treatment B--they simply weren't published. Conventional wisdom was that only positive findings could tell us anything. No longer. A new technique called "meta-analysis," which collates and analyzes all the different studies of a given subject--including the duds--is rapidly supplanting the old method of considering just positive results. According to New Scientist, 100 leading science journals announced a global "trial amnesty" on unpublished findings last month, encouraging researchers to submit studies--no questions asked--that they might otherwise be embarrassed to own up to. Experts say there may be 50,000 such studies languishing in obscurity that could be put into a central database and collectively mined for useful findings.