Abstract: In this paper a new method for achieving predictive models, called latent root estimator, is introduced.The technique was utilized for the simultaneously quantitative measurement of a system containing five components (salicylic acid, thymol, phenol, benzoic acid and resorcin) from their severely overlapping UVspectra and the predictive performance of the resultant calibration model was tested with a separate set of samples.The results indicate that the method possesses an obvious significance to gain stable predictive equations for collinearity systems and to reduce the influence of instrumental noise for concentration estimation.

Key words: Calibration technique, Latent root estimator, Spectrophotometry