Volume 46, Number 5  (September - October 2001)

Key words:  error propagation, response variance, gas chromatography

The propagation of random error in GC analysis has been examined using an example of the volatile compounds in human blood. The following analytes were determined: acetone, ethanal, ethanol, methanol, 1-propanol,  2-propanol. The concentration of each analyte was calculated from the corresponding calibration graph: the ratio of chromatographic peak areas of the analyte and an internal standard (1,1-dimethylethanol) vs analyte concentration. It was shown how the error of determination of peak area affects the error of the analyte determination. The response (the response of peaks` areas) error was estimated in various ways: 1) from the replicated determinations of the response on a given analyte concentration level, 2) on the bases of the error propagation principle, and 3) from the calibration data. The structure of the experimental data obtained has been also analysed and visualised with the use of the standard chemometric methods: PCA and CA.