An interesting report on bad science in crime labs.

Let us consider something we used to teach our sophomores - lead smelting and refining. Almost all lead occurs as sulfide ores that contain lesser amounts of other metals. Smelting removes the sulfur and refining removes most of minor elements, notably gold, silver and copper. The composition of the refined lead may be easily and inexpensively determined by, for example, spectrographic analysis.

Someone at the FBI decided that if the compositions of two bullets "matched" well enough the two were from the same box of ammunition. Then if the box of ammunition was tied to the defendant, so was the subject bullet.

My reaction to the claim is "HUH?" One crucible of refined lead could make millions of bullets, and the molten lead is not necessarily of uniform composition. A composition match does not prove a darned thing.

This erstwhile expert was clearly working far above his pay grade, but sold the idea to superiors who really should have known better. For the next two decades, thousands of innocent people were convicted on the basis of totally hokum bullet matching. The FBI was the only lab in the country that was using the technique, which should have been a warning that something was wrong.

Finally someone in a high position got the National Academy of Science to address the lead-matching issue. They turned thumbs down and the FBI stopped matching bullets to a particular box.

Recently the Academy performed an extensive study of the nation's crime labs. Law enforcement agencies resented the intervention of mainstream science in the courts and an arm of the Justice Department tried to block the study. It failed and the resulting report decried the lack of science and the use of shoddy practice.

Just round up the usual suspects.

Cross Posted at Power and Control