Warning to anyone planning a crime: don’t drink the water.

Scientists are reporting this evening that an analysis of hydrogen and oxygen isotopes in human hair reveals where a person drank water, which will allow investigators to figure out if a suspect was in the vicinity of a crime—and also to track past movements of unidentified murder victims.

The new technique analyzes isotopes of hydrogen and oxygen, forms of the elements that have an extra neutron in its atomic nucleus. The ratio of oxygen-18 to oxygen-16 in air is the same everywhere, and ratios of both hydrogen and oxygen ratios in food are also basically the same across the country because the American food supply is so uniform. But water is still like a microbrew, containing different constituents depending on where it comes from.

As a result, geochemist Thure Cerling and ecologist Jim Ehleringer of the University of Utah report in the journal Proceedings of the National Academy of Sciences, a single hair bears clues to where someone has been over the last few weeks or even years, depending on the length of the hair. “You are what you eat and drink—and that is recorded in your hair,” said Cerling in a statement from the university.

Salt Lake County Sheriff's Detective Todd Park is using the hair-isotope method to try to identify a murdered woman whose remains were found by hunters near Interstate-80 along the south end of the Great Salt Lake on October 8, 2000. Detectives recovered 26 bones, some hair, a T-shirt and a necklace. Despite creating a facial reconstruction and publicizing it nationally, the police have been unable to ID the woman, who stood about 5 feet tall and was 17 to 20 years old when she was killed. But when Park arranged with Ehleringer for an isotope analysis of the victim’s hair, it showed that for the last two years of her life she lived in the Northwest, mostly in the Idaho-Montana-Wyoming area, and maybe into Oregon and Washington. Park hopes that examining missing persons records from those areas will lead to an ID.

Ehleringer and Cerling are co-founders of IsoForensics, Inc., which sells isotopic analysis of forensic substances. A method Ehleringer developed to trace the origins of cocaine or heroin is now used by the a method now used by the U.S. Drug Enforcement Administration: variations in the amount of carbon, nitrogen, oxygen and hydrogen isotopes in soil and water show up in coca and poppy plants.

For the hair analysis, says Ehleringer, “You can tell the difference between Utah and Texas,” but probably not Chicago and Kansas City. In general, the amount of the heavy isotopes oxygen-18 and hydrogen-2 levels in drinking water decreases as you move inland from the West Coast: as rainstorms move off the Pacific Ocean, rain drops containing oxygen-18 and hydrogen-2 tends to fall first because it is heavier. But cloud temperatures and season also affect which isotopes rain drops contain, with the result that heavy isotopes are also relatively more common in water inland from the Gulf and southern Atlantic coasts. The lowest concentrations of hydrogen-2 and oxygen-18 are found in water in northern and western Montana, north-central Idaho and northwest Wyoming (because the heavy isotopes drop out of clouds before they reach these inland areas), while the highest amounts of hydrogen-2 and oxygen-18 in drinking water and hair are in southern Oklahoma, north-central Texas, Florida, south Georgia and southern South Carolina (in warm regions with more evaporation, lighter-isotope water evaporates from surface sources first, leaving heavier-isotope water behind to find its way into drinking water.

Just for the record, reporters take no pleasure in questioning the power of drugs to treat depression. To the contrary: journalism is notorious for attracting curmudgeons, grumps and depressives—some of my best friends are one or more of the above—so we wish with all our hearts that antidepressants would work.

And that scientists wouldn't keep finding evidence that they do not.

In January I reported on the file-drawer effect in studies of antidepressants. The file-drawer effect refers to the fact that scientifically-sound studies on the efficacy of antidepressants are not published, as The New England Journal of Medicine article described. Most of those studies were negative—that is, the drugs did not help patients much more than a sugar pill (placebo) did, if they helped at all. That skews the perception of doctors, scientists and you and me about these drugs; basing our assessment of antidepressants on published studies alone is like evaluating the prowess of a baseball team when only its wins and not its losses are reported.

Now a team of scientists has examined many of those unpublished studies, obtained through a Freedom of Information Act request for the U.S. Food and Drug Administration. As many people feared, once you include the deep-sixed studies, antidepressants look hardly more effective than a placebo at lifting patients’ black cloud of despair.

For their analysis, scientists led by Irving Kirsch of Britain’s University of Hull started with the data dump they got from the FDA on fluoxetine (Prozac), venlafaxine (Effexor), nefazodone (Serzone), and paroxetine (Seroxat /Paxil). They zeroed in on differences between the improvement reported by patients receiving the drug and those receiving a placebo. As is standard in such clinical trials, neither the patients nor the scientists running the study knew which patients were receiving real drugs and which were receiving placebos.

In short, there was virtually no difference in the response to drug vs. placebo of patients who suffered moderate levels of depression, and a small difference for patients with very severe depression, they report in the study published this evening in the journal PLoS Medicine. That small difference was, however, clinically insignificant—that is, the difference was so small that government health authorities do not recognize it as a meaningful improvement: on a standard scale of depression, patients should improve by 3 points, but the spread between placebo and drug was only 1.8. The difference between drug and placebo was clinically meaningful only for patients at the upper end of the very severely depressed category.

The reason for the tiny, or nonexistent, differences? Patients respond so well to placebo—to the mere thought that something might be helping them—that there was little room for an actual drug to do more. Across all groups, response to placebo accounted for more than 80 percent of any improvement. (In contrast, the placebo response to pain drugs is estimated at about 50 percent.) That suggests that even when patients are taking and benefiting from, say, Zoloft, the vast majority of the improvement is due to what their minds are telling them—that is, the belief that they would be helped. Only the most depressed patients showed little placebo response.

The scientists conclude that there is little reason to prescribe the new antidepressants to any but the most severely depressed patients except as a last resort. Kirsch summarized the findings this way: “Although patients get better when they take antidepressants, they also get better when they take a placebo, and the difference in improvement is not very great. This means that depressed people can improve without chemical treatments.”

But it seems that there is a larger message here. The placebo response—the belief that treatment will make you better—is enormously powerful. Surely it’s time to investigate further how it works and how it can be harnessed.