Abstract

The use of near infrared spectroscopy (NIRS) is demonstrated to be an important alternative to the empirical and laborious traditional methods for tannin analysis of the bark of black wattle (Acacia mearnsii De Wild.), particularly when large numbers of samples have to be processed rapidly. A set of 98 ground wattle bark samples was used to calibrate the output from a NIRS instrument, taking absorption readings every 2 nm between 1100 nm and 2498 nm. The spectral data from this calibration set was then statistically manipulated using multivariate techniques to generate calibration models. These calibrations were then applied to a separate set of 79 bark samples which, for validation purposes, were also analysed by wet chemistry. The tannin analysis predictions compared favourably with the wet chemistry results on these samples, with standard errors of determination of 1.2%, 0.3% and 1.1% for tans, non-tans and total extractives respectively.