Abstract
In this study, a pattern recognition technique which can calculate such various molecular properties as a molecular weight, a molecular surface area, a sigma charge, a moment of inertia, a molecular refraction and so on, was used together with the sensory experimental data presented by Amoore which contain the values of odorous similarity of 97 compounds to ethereal, camphoric, musky, floral, pepperminty, pungent and putrid odor.
The relationships between molecular properties and the odor quality were investigated using a multiple regression analysis. Five regression formulas with 5 descriptors were formulated for 5 odors, that is ethereal, camphoric, musky, floral and pepperminty odor. The high correlation was found between experimental similarity of ethereal odor and calculated ones. These formulas were employed to predict the odor quality of 27 ethereal odorants and 44 musky odorants, previously unused, and the results were satisfactory.
