Rapid Spectrophotometric Analysis of the Chemical Composition of Tobacco: Part 3: Polyphenols

Two hundred thirty-eight ground samples of tobacco were scanned with a computerized near-infra-red (NIR) spectrophotometer to study the relationship of NIR spectra to the polyphenol content of the samples. A multiple linear regression model was used to select the most appropriate wavelengths for making the measurements. The equation is a valid equation for predicting polyphenols in tobacco. If the coefficients are validated on the same kind of tobacco and in the same year, the standard error of prediction of the NIR method (± 0.747 %) approaches that of the wet chemistry method (± 0.70 %) with coefficients of variation of 21.75 % and 23.0 %, respectively. The largest standard error of prediction across years was ± 1.353 %.

P = polyphenols in %

D₁² = 2^nd derivative parameter

i = wavelength in micrometers at which the parameter is measured.