Increase in coarser plucking standards significantly (P<0.05) depressed levels of theaflavin (TF; μmol/g), total colour and brightness (spectrophotometer) of black tea. But total thearubigins (TR) TRSI, TRSII and brightness (taster) levels did not significantly (P<0.05) change with plucking standards. Increasing fermentation duration led to rise in TF (μmol/g), total TR, total colour, TRSI and TRSII, liquor brightness (taster) but not liquor brightness determined by spectrophotometer method (P<0.05). Correlation analysis showed that total TR and TRSII had negative correlations (r=−0.66 and −0.77, P<0.01), respectively, while total TF levels had positive correlation with taster brightness (r=0.57, P<0.01). Total TF level had highest correlation with spectrophotometer liquor brightness (r=0.87, P<0.01) for a single substance. TRSII and total TR gave an r of 0.86 indicating that the two groups of substances were strongly correlated to each other. Regression analysis showed that the direct linear model gave the best fit for the sample data studied. The coefficient of multiple determination (R2) was 0.606 in linear model I for the tasters’ liquor brightness. Thus, the independent variables TF and total TR explained 60.6% of the total variation in liquor brightness scores observed. When TRSI and TRSII were included in the linear model II instead of total TR, but with TF maintained, the coefficient of multiple determination improved to 78.9%. This confirmed that the brightness attribute of black tea, assessed by the taster, could best be explained by the combination of TF and TRSII and that TRSI had a lesser role in the tasters’ evaluation of liquor brightness. Indeed, the test statistic in linear model I showed that the coefficient of TF positively and significantly influenced liquor brightness at P<0.01 whereas the coefficient of total TR negatively and significantly influenced liquor brightness (P<0.0001). However, in linear model II the effect of total TR on the taster brightness was clearly unmasked and the influence of each individual component well elucidated. The coefficient of TRSII was negative and significantly explained liquor brightness at 0.01% level (P<0.0001). The coefficient of TRSI was negative but insignificant. Hence it has no discernible influence on the taster liquor brightness. The coefficient of TF was positive and significantly explained taster liquor brightness (P<0.01). Thus, TF and TRSII explain taster liquor brightness and the lower the level of TRSII the higher the score for brightness. The situation for spectrophotometer brightness was somewhat different. The coefficient of multiple determination was 0.896 in linear model I. Thus, TF and total TR explained 89.6% of the variation in spectrophotometer brightness. The coefficient of TF was positive whereas that of total TR was negative. Both coefficients were significant (P<0.0001). When TRSI and TRSII were included in the linear model II instead of total TR, R2 improved to 91%. Unlike for the tasters’ brightness, the coefficient of TRSI was now significant (P<0.05). The coefficient of TRSII was negative and significant (P<0.01) whereas the coefficient of TF was positive and significant (P<0.0001). The differences in the contributions of TRSI and TRSII to black tea liquor brightness and the observed discrepancy between the test methods, due to variations in plucking standards and fermentation duration, are discussed.
Read full abstract