Near infra-red analysis: The use of multivariate statistics for investigation of variables in sample preparation and presentation of tea leaf

A data analysis procedure involving principal component and canonical variate analyses was used to investigate the effects of three sample variables (particle size, temperature and sample rotation) on the near infra‐red spectra of six teas of different geographical origin. The effects of milling method were shown to be much larger than those of sample temperature or rotation. The analysis provided a rational basis for the choice of a milling technique and sample presentation for a study of the possible relationship between tea quality and near infra‐red spectroscopic data.