ON THE INHERENT INVALIDITY OF ALL CURRENT SYSTEMS OF ODOR CLASSIFICATION* Giraudan-Ddawanna, Inc., New York, N.Y. I N TH •. philosophical examina- tion of the fundamentals of classi- fication and its applicability to the field under consideration, namely the domain of odor, the writer is guided by a scientific truism which has been expressed by many, but never more pointedly than by W. S. Jevons (1): I "To mark out those fields of ex- perience which, on the basis of the present state of our knowledge, are still inaccessible to scientific under- standing, often constitutes a valu- able, though negative, contribution to scientific progress." In order to delimit the scope of our discussion, let us define our terms. "Classification," states Broadfield (2), "consists in dif- ferentiation of qualities." This dif- ferentiation may or may not have quantitative differences within the qualitative. In this respect, Sayers (3) points out that "classification arranges things according to their degrees of likeness." By odor classification, I mean the * Presented at the December 8, 1949, Meeting, New York City. grouping together of different odor- ous substances or stimuli according to a qualitative similarity or af- finity among them. Thus, for the purpose of this paper, I am not evaluating systems of quantitative odor classification, based for example on threshold strengths nor am I considering the hedonistic or affectire classifications, based on whether a stimulus is con- sidered pleasant or unpleasant by 'the observer. Although I deny the validity of current systems of odor classifica- tion, I do not deny the desirability of such classification. The value of establishing relationships among in- dividual pieces of knowledge has been emphasized by researchers in many fields, and particularly by eminent philosophers of science, as Poincar•, Ostwald, Pearson, and others. Poincar• (4) stated that science "is before all a classifica- tion, a manner of bringing together facts which appearances separate, though they were bound together by some natural and hidden kin- ship. Science, in other words, is a system of relations."

:26 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS Wilhelm Ostwald (5), among whose many contributions was his system of color classification, wrote: "Thus we see that the ordering of facts and their relationships in each individual science is the first and mostimportantfunction in its de- velopment." And Karl Pearson (6): "The classification of facts and the forma- tion of absolute judgments upon the basis of this classification... es- sentially sum up the aim and method of modern science." (Emphasis in originaL) The value of classification has likewise been stressed by research workers in the field of odor. The distinguished chemist, ¾ves-Ren• Naves (7), stated recently: "This problem of classification has not been solved, and it is one of the most disconcerting of our industry." But if the chemist has expressed the need and importance of a classi- fication of odors, he has not failed to underline the futility of at- tempting such a classification at this stage of our knowledge. This was the substance of the thinking of Marston T. Bogerr (8), who said: "An exact and impersonal scien- tific classification of odors, compar- able to that available for colors, seems clearly unattainable in the light of our present knowledge. The best that can be done is a more or less superficial grouping into types or classes which to the ob- server seem to bear a sort of family resemblance to one another, and we therefore speak of (for example) lily type, rose type, musk type, of odor, but the boundaries of these groups are both vague and vari- able." It is my purpose here to bring forth the theoretical basis that makes this conclusion of Dr. Bo- gerr inescapable, and to show that it is furthermore a conclusion sub- stantiated by considerable experi- mental data. II Any scientific system, whether or not it deals with classification, must be objective to be valid, and ,ob- jectivity involves reproducibility. By what standard, by whose mind, shall validity be determined? Quoting again from Karl Pearson. (9), we find: "In order that a conception may have scientific validity, it must be self-consistent, and deducible from the perceptions of the normal hu- man being... But even if an in- dividual mind has reached a con- ception, which at any rate for that mind is perfectly self-consistent, it does not follow that such a con- ception must have scientific validity, except as far as science may be con- cerned with the analysis of that individual mind." The meaning of objectivity was defined by Poincar• (10): "To understand the meaning of this new question (can science teach us the true relations of things?) it is needful to refer to what was said before on the conditions of objec- tivity. Have these relations an ob- jective value? That means: are