Key Words:
Gas chromatography Signal to noise ratio Linear calibration Signal integrationThe use of linear calibration according to the following equation is very widespread in gas chromatography.where Y, is the integrated instrument response or ratio a is the estimated slope of the calibration curve XI is the concentration or ratio b is the estimated ordinate intercept q is the error termThe estimation is made with the Yi, Xi pairs by the method of least squares, which is a method of maximum likelihood, if the distribution of data is normal. The precision and accuracy of the calibration have been extensively discussed in references [ 11 and [2]. There are two assumptions concerning the properties of the error term, which have not been discussed explicitly in the analytical literature.These are:W, and u are constant in time.where E(E) is the expected value of E V*(E,) is the variance of E, W, is the weighting factorIn our laboratory the ethanol content of 50000 blood samples per year is measured in the concentration range 0.05 gll to 5.0 gll, with 1 -propano1 as internal standard by headspace gas chromatography. Our experience has shown that an essential condition for unbiased estimation ofthe parametersaand b isaspecifiedsignal to noise ratio. where H is the height of the chromatographic peak