A multidimensional statistical study of similarities between 74 notes used in perfumery

Abstract

From a data bank of 2467 odoriferous products, the similarities between 74 notes used in perfumery were calculated. The similarity matrix (74,74) contains – 63% of zero values and shows that only a few pairs of notes present high similarity coefficients. A fine analysis using ascending hierarchical taxonomy with the complete linkage procedure shows that 14 notes are isolated while 60 notes are regrouped in 27 groups containing two to four notes. The isolated notes correspond to well-defined structural particularities. Some pairs of notes or groups present similarities in their chemical structures but some groups are built on the basis of semantic processes. This study shows that the notes are generally independent, with no strict hierarchy among them, and rules out the existence of primary odors.