Researchers used a mix-match algorithm correlating ultraviolet (UV) visible spectra with the library spectra of pure dyes in ionized water to identify two, three, and five component mixtures of food and fabric dyes in water. The algorithm performs principal component regression (PCR) on a library of spectra of pure components, treating each spectrum as a standard mixture featuring a unit concentration for that component. Projecting the scores of an unknown spectrum onto the principal component bass set and multiplying by the regression coefficient approximates the concentrations. An adaptive filter removes all similarities of the first target component from the mixture and library spectra. PCR applied to the modified spectra identifies a second target component and the process is repeated until the residuals approach zero. Searches of 107 sample mixtures of food and fabric dyes produced 100 percent correct identifications in deionized water, 92 percent in stream water samples, 89 percent in river water samples, and 74 percent in ocean water samples. 21 refs.