Almost all the values for solids in which Ar exceeds were

obtained with asafoetida, oil of mace, or the home-made cylinder of gutta-percha and gum ammoniac. Thus, out of 9 values in which D. exceeded 1⁄2 for solids, 7 were found with oil of mace, and out of 16 values in which Rob. exceeded 1⁄2, 10 were found with the home-made cylinder of gutta-percha and gum ammoniac, and 4 with oil of mace. We believe that it is perfectly fair to exclude these cylinders from our final results. And if we do so there is little variation from substance to substance. The odor of asafoetida and oil of mace was very perceptible when the instrument was closed, and the mace would

r

CURVES SHOWING THE APPROXIMate Values of IN THE WHOLE COURSE OF EXPERIMENTS BY THE Method of Just NOTICEABLE DIFFERENCES. (See Table VI, Part 3.)

The heavy line gives the values for both solids and liquids; the dotted line gives the values for solids, and the broken line for liquids. The ordinates give the number of cases, and the abscissæ approximate values.

scrape off on the inhaling-tube. While Zwaardemaker's mixture of gum ammoniac and gutta-percha is black and brittle like licorice, ours was yellowish gray, contained strings of gutta-percha, and made the inhaling-tube cloudy and sticky. We did succeed in obtaining stimulus-limina with it when the inhaling-tube was first cleaned, but we believe that the end of the tube was probably soiled most of the time during differencedeterminations. We have not excluded the results for Russian leather because its odor, like that of most of the liquids, was just liminal when the instrument was closed, and the results harmonized with the others. Since most of the liquids had this error of the equal but unmeasured increment, it is not surprising that the values of Ar run higher for them than for

solids. It will be noticed that they run highest for valerianic acid, which was particularly troublesome in escaping from the instrument. Yet as the results for coumarine, heliotropine, and musk show 3 as the most common value, we must conclude that the value of ▲r lies somewhere between 3 and 4.

r

Some of the substances showed an interesting difference of quality with difference of intensity. Thus several subjects thought that oil of camphor smelt like nutmeg when weak, and like turpentine when strong. The slight odor of the paraffine appeared when a strong stimulus was given with coumarine. T. said that heliotropine smelled like heliotrope on the left (the better) side of her nose, and like bitter-almonds on the right. (As a matter of fact the two smells are closely allied.) Se. said that the tallow smelled like onions in his poorer nostril. Fluctuations at the limen were also noted. Coumarine and heliotropine, when weak, were said to come "in whiffs" or waves, and K. always spoke of weak smells as "scattered."

Table VII gives some of the results obtained by the method of just noticeable differences modified in the direction of the method of minimal changes, as described in Chapter I, Section. 4, and shows the agreement of these results with those reached by the ordinary method. C. M. stands for "Combination Method."

We used red rubber with the true method of minimal changes because Zwaardemaker had done so. The cylinder was obtained from Utrecht. The experiments of which the results are given in Table VIII extended through five laboratory-hours. It is needless to say that the instrument was manipulated entirely by the experimenter.

TABLE VII.

Results of the Modified Form of the Method of Just Noticeable

Results obtained for Red Rubber by the True Method of Minimal

r= 20 mm.

Zwaardemaker concluded that for a standard of from 2 to 5 cm., the difference limen was about 1.5 cm., and that for a standard of from 5 to 9 cm., it was about 3.5 cm. This would make the value of Ar run from about 3 to about 34. Our

r

own results agree fairly well with his, and are a very pretty confirmation of the results obtained by the method of just noticeable differences. The writer intends to use the method of minimal changes much farther.

In contrast with these excellent results are those of the next Table :

TABLE IX.

Results obtained by the Method of Right and Wrong Cases.

SUBJECTS-C., D., K., N., ROB., ROG., AND T.

Instrument-Standard Olfactometer. Substances-Black Rubber

The stimuli given were never equal, and the judgment "equal" was counted a mistake. The results of all the subjects are massed.

As we said before, while exhaustion makes the errors nearly all run in one direction, confusion due to the unfamiliarity of olfactometric work is probably most at fault. More experiments should be made with the standard olfactometer and trained subjects. It is difficult to use the large olfactometer with this method, because the intervals between stimuli must be made very long or the subject can guess from the time spent in manipulation how they have been changed.

As a rough method of testing the applicability of the method of right and wrong cases to smell, we blind-folded one subject, stopped his ears with absorbent cotton, and required him to tell which way we had moved from a given standard on the large olfactometer. The results are given in the following Table: