Application of bivariate statistics to full wine bottle diamagnetic screening data

A bivariate correlated Student distribution is applied to full wine bottle diamagnetic screening measurements. Previous work involving a limited number of rare wines indicated that like wines cluster in a plot of the first two principal component scores derived from a covariance matrix of the diamagnetic screening measurements. This study extends the approach to a much larger, statistically meaningful sixty bottle wine library where bivariate statistics are used to comment on the measured data. The full bottle diamagnetic screening of thirty-six identically labeled, sealed bottles of wine obtained from four different sources combined with principal component analysis data reduction followed by treatment with a bivariate distribution permit the effect of wine transport and storage to be observed. The usefulness and future success of the method towards the identification of counterfeit wines is mentioned.