AbstractAny smoke component is distributed over both the particulate and the gaseaus phases. The distribution is determined by a) the vapour pressure of the considered substance, b) its concentration in the smoke, and c) its physico-chemical affinity for the particulate phase. The availability of a substance in the gaseaus state, i. e. the condition for its selective removal by an appropriate filter, is therefore dependent on these factors. The experiments designed to study these relationships are based on three variables held each at several Ievels: :1. Test substances distinguishing themselves by the factors a), b), and c): Carbon monoxide, crude (moist) smoke condensate, total water, pyridine, total alkaloids, volatile phenols, catechol, and scopoletin. 2. These substances are determined in the smoke trapped separately per individual puffs. The composition of the smoke leaving the cigarette depends on the puff number, according to the factors a), b), and c). 3· Use of four different smoke filtration devices permitting the study of the availability of a given test substance or of its affinity to the filter material. The following expressions are used for the interpretation of the experimental results: The ratio of the yield of test substance to the yield of crude condensate; the coefficients of filtration of the test substance in both the tobacco rod and the experimental filter; the coefflcient of selectivity, defined as the difference between the respective coefficients of filtration for the crude condensate and for the test .substance. The results of the experimentation tend to confirm the initial hypothesis and give ,sufficient detailed information to permit the tentative establishment of a general scheme which allows to predict the selective behaviour of a particular ·smoke component, if its concentration in the smoke, its vapour pressure and its presumable affinity towards a given filter material are known. lt is further shown that a mathematical expression for the yield of a particular smoke component, as a function of the puff number, can be established on an empirical basis for carbon monoxide, · crude condensate, total alkaloids, catechol and scopoletin. The increase in the yield of carbon monoxide with the puff number is shown to be due to a CO loss through the permeable paper wrapper of the cigarette, this loss being smaller as the butt becomes shorter