[electromagnetic]

If you're a fruit-eater the noise would be incredibly high. Think of watermelon, oranges, grapefruits, heck even tomatoes are all frequently eaten by people and contain large amounts of water.

Someone who eats 'healthily' is going to be much harder to geolocate than someone who eats 'carnivore' style.

I suppose you would look for the strongest source and use that. However I can't help but feel that they're measuring differences in identical isotopes and not relying on unique isotopes (IE of a different mineral) to do the locating. More potassium might mean the North American east coast and more calcium could mean the west is how their method reads to me. This means that they're calculating from the sum of the isotopes.

[Groxx]

The article mentions using only hydrogen and oxygen isotopes (I'd assume due to the distortion caused by filtering if you did it by any other elements), and looking at hair samples to detect the isotopes in the proteins / molecules which make it up. Even being carnivorous is going to distort things if you get your meat from multiple locations (say, fish. That comes from everywhere, at least for me), as you'll incorporate the amino acids (which have plenty of H and O) into your own proteins.

After looking closer, it appears that the study makes only the (obvious) claim that bottled water purchased near location A has similar isotopes to water in location A, due to bottling habits. Though I haven't read the study itself. The article may be entirely vague speculation, to drum it up into something more exciting / inflammatory.