Evaluation of inter-substitution among tobacco leaves with multivariate statistical test
In order to evaluate the inter-substitution among tobacco leaves, nine chemical components, including total sugar, total alkaloids, total nitrogen, cellulose, starch, polyphenols, ash, petrol-ether extract and total volatile bases, in 2482 flue-cured tobacco leaf samples belonging in 80 categories were determined, then the data were divided into 80 object sets and 30 validation sets, the approximate T2 test values between the two data sets were calculated, and the inter-substitution among the leaves was evaluated by the ratio of T2 to c2, where c2 was the critical values at a given confidence level. The results showed that: 1) Chlorine was not applicable to be included in T2 test, because the determined chlorine contents in a number of samples were not conformed to a normal distribution. 2) In the two data sets, the samples were of higher inter-substitution (T2/c2 <1) if they were of the same grown area, grade and variety, while the samples were of lower inter-substitution (T2/c2 >1) if they were of different properties. It was concluded that this method features higher accuracy and recognition ability and is applicable to the evaluation of inter-substitution among tobacco leaves.