they exercise a momentary control over the visual impulses from A], Ao, and A 3 , transmitting them through C to B with an intensity con- siderably reduced at first in the case of a bright flash. Perhaps even the apparent delay of T \/' may be due to this initial reduction.
Next suppose a second flash falling on AI DI and A 3 D 3 . Owing to the cross connections from DI and D 3 to Do, the activity of D 2 is aroused again, and the visual impulse from A 2 is blocked once more, so that little or none of it gets through to Bo.
If we suppose with Young that each colour-sensation results from the activity of an independent peripheral system, it is easy to explain the experimental phenomena. Let the second stimulus be white and the first green. DI and D 3 , having been made to act strongly by the repeated excitation, will block all visual impulses of green from AI and A 3 by direct action, and from Ao by virtue of their cross-connections with Do. The result will be that the components of purple, i.e., (W G) = { (K + G + B + V) - G} will reach B t and B 3 , and nothing at all will reach Bo exactly as in Dr. Shelford Bidwell's experiment.
It is evident from a multitude of experiments that we have to deal with two classes of contrast effects, namely, those which are induced gradually and pass off slowly, and others which are momentary and suggest the term spasmodic. The latter require for their production a sharp and sudden stimulus, and one such stimulus may even be followed by a multiple response, as in the recurrent images of Purkinje or Charpentier's bands. To myself these phenomena recall very forcibly the multiple electrical response of a nerve-muscle prepara- tion, or of an electrical organ, to a single strong stimulus, and I suggest that the " recurrent image " is caused by intermittent blocking of the positive after-effect. The failure of Talbot's law for low speeds is easily explained on this hypothesis. The ratio of the magnitude of the response of the " D " structures to the magnitude of the stimulus is greater when the stimulus is repeated a certain number of times per second, i.e., a larger percentage of the total visual impulse from A is blocked by D when there are fifty flashes per second than when there are only five, and consequently the mean brilliancy appears less until the limit is reached, beyond which Talbot's law holds.
The phenomenon is manifested in greatest intensity when the stimulus consists of two or three flashes close together followed by an interval of rest. Possibly the reason of this is partly that the activity of D may be maintained, and partly that A may recover from its fatigue ; however that may be, it is essential in all experiments on the reversal of the image that the eye should be in darkness for one- half of the cycle.
Another phenomenon explicable on this hypothesis is the flickering of certain spectral colours. If a small weight is held out at arm's length, after a time the muscles begin to shake and quiver. And if a
