In common batch searching of illicit heroin samples--evaluation of data by chemometrics methods

In this study three illicit heroin samples which belonged to three different batches were subdivided into eight samples each. To simulate the dealers' chain, and to check the influence (if any) of diluents on the analytical results, some of the samples were cut with the most frequently used cutting substances, whereas the others were left unchanged. Samples were analysed (within a 10-week period of time) by gas chromatography-mass spectrometry (GC-MS) and characterised by seven variables each. To recover batch links among investigated heroin samples three multivariate methods, i.e. hierarchical clustering (HCA), principal component analysis (PCA) and k-nearest neighbours (k-NN), were applied on to the normalised and scaled analytical dataset. The classification abilities of the HCA, PCA and k-NN were in the range from 95 to 100%. Disturbing effects due to the dilution of samples have not been observed.