vapor, and (5) on the rate and manner of the subject's breathing. The great incidental difficulties in olfactometric work are (1) the variability of the organ through obstruction by mucus or (2) exhaustion, (3) the adhesion of the odorous matter to parts of the apparatus, and (4) the presence of compensating smells. The.freedom of the nasal passages may be tested, but exhaustion can neither be prevented nor measured, nor can adhesion and the presence of compensating odors be absolutely excluded. We employed Zwaardemaker's olfactometric method in which (I) the measure is the amount of odorous surface exposed, (2) the time of exposure may be disregarded, (3) the diffusion-rate of vapor is under control, and (4) the subject's breathing is supposed to be self-regulating. We did not (5) succeed in regulating the temperature of our laboratory, but its variability was not of primary importance in difference-determinations. Adhesion makes the method of minimal changes impracticable for most substances with Zwaardemaker's method of smell-measurement, and exhaustion contributes to make the method of right and wrong cases very difficult. We therefore used the method of just noticeable differences. This psychophysical method involves an error from the subject's tendency to judge in terms of hand-movement. Another occasional source of error, incidental to our apparatus, was the escape of some odors between the inhaling-tube and cylinder. Both of these circumstances tend to make the values of Δr⁄r smaller for the larger standards. Adhesion and the ordinary time-error tend to balance exhaustion. In spite of the four most serious sources of error, (1) exhaustion, (2) adhesion, (3) the movement-error, and (4) the unmeasured increment to some stimuli, we found Δr⁄r to be about 1⁄3 in 36% and about 1⁄4 in 26% of our determinations. It was about 1⁄2 in 12%, about 1⁄5 in 12%, about 1⁄6 in 4%, greater than 1⁄2 in 5% and less than 1⁄6 in 5% of the determinations. The slight use we made of the other gradation-methods confirms the general result. There is no great variation from one substance to another, or from one of Zwaardemaker's classes to another.
There is much yet to be done and said in olfactometric work—"of making of books there might be no end"—but we believe that enough has been said and done to offer some evidence that Weber's law applies to smell and that the value of Δr⁄r lies between one-third and one-fourth.
