1707multiple determination (R2) exceeding 0.97. For the analysis of simulated unknown mixtures within the calibration range (set A), the mean deviation was less than 1.10 mg/dL for any of the components. Folr unknown mixtures outside of the calibration range (set EN), errors are still tolerable, with the mean deviation not exceeding 1.71 mg/dL for the least accurate component, phospholipids. The mean deviation for triglycerides (0.68 mg/dIJ and cholesteryl esters (1.24 mg/dL) was much better.
CONCLUSIONThe first known simultaneous determination of triglycerides, phospholipids, and cholesteryl esters has been made possible by using a multiple least-squares regression method in P-matrix form. The analysis matrix was prepared from calibration of 85 standards, using absorbance data measured at 15 evenly spaced wavelengths over the absorption region of interest. A linear model in the form C = PA + Po gave optimum results for the analysis of mixtures within calibration limits.This method of analyzing the predominant serum lipids by class is presently being applied to the analysis of native serum and results will be reported in a future paper. The methods developed are applicable to the analysife of other complex mixtures, especially systems involving components with highly overlapping spectra.
ACKNOWLEDGMENTWe wish to thank R. J. Obremski for many helpful sug-Registry No. Tripalmitin, 555-44-2; dipalmitoyl-DL-a-phos-gestions and M. A. Maris for computer assistance.phatidylcholine, 2797-68-4; cholesteryl palmitate, 601-34-3.