B back to A, but not completely, and the negative deflection will be less than the previous positive deflection. Owing to this "damping" effect, the oscillation curve will approximate to a logarithmic decrement curve. After a series of reversals the oscillation dies away, and the substance becomes almost inert. A glance at the hypothetical curve to the right shows that at the inert stage, b, the substance as a whole ought to become more conducting the fresh specimen, a.
The opposite should be the case with negative substances (see fig. 4(α).
Fig. 5 exhibits the actual curve obtained with a (compound) positive substance.
Fig. 5.—"Damped" Oscillation Curve for a Positive Substance.
It is remarkable for its regularity. The next figure (fig. 6) gives the curve for iron. The first diminution of resistance is too great to be properly represented in the diagram. Here we have the same type as in the previous case; the inert stage, b, is also more conducting than a.
IV. I will now consider the case of a negative substance exhibiting damping; arsenic will be taken as an example where the damping is not so great as in the case of iron.
Fig. 7 represents the actual curve obtained with arsenic (compare with the hypothetical curve for a negative substance, fig. 4 (α).
It will be observed that the substance in the fatigued state is, on the whole, less conducting than in the fresh condition, as we were led to expect from the hypothetical curve. It will also be seen that the oscillations are very regular towards the end. The curves given in figs. 6 and 7 are those obtained with specimens immediately after they were set up. Had I allowed them a period of rest to allow the particles to get properly settled, I would have got curves even more regular. It is, however, evident that in substances exhibiting damping, two opposite electric conditions are induced in fatigued specimens; the positive becomes on the whole more conducting and the negative less conducting than in the fresh specimens. At the inert stage the rate of mutual conversion from one state to the other probably becomes equal, and the apparent fatigue is thus not due to the absolute want of sensi-
