For anal. of multicomponent mixtures using linear regression algorithms, a qual. detection of the anal. sample must be made to ascertain the exact nature of coexisting species in the mixture The most frequently occurring cases in anal. practice is that the range of possible coexisting species is known and a decision has to be made which species actually coexist in the sample. This paper describes a new chemometric method based on bound-branch approach to solve this problem which has been turned to the problem of selecting the optimal subset for regression. The computation time is much less than the enumeration procedure and there is no risk of missing the real combination of the coexisting components in the sample. A program written in True BASIC has been testified for UV and fluorescence spectrophotometric analyses of mixtures containing various possible coexisting organic compounds with satisfactory anal. results.6

Key words: ULTRAVIOLET SPECTROPHOTOMETRY, QUANTITATIVE ANALYSIS, COMPUTATIONAL METHOD, COMPUTER APPLICATIONS, MIXTURES, QUALITATIVE ANALYSIS, STOICHIOMETRY, FUSED RING COMPOUNDS, BARBITAL