repeat these experiments a number of times and take more than ten each time. Then let us average all the first ones separately, all the second ones likewise, etc.

Fig. 18 shows lines erected proportional to the successive average times. The decrease from the first average time we may term the "gain by practice "; a line

showing the amount of this gain will be the "curve

of practice."

Let us now average all the results for each day; Fig. 19 shows the result. The decrease in time for each succeeding day may be called the "gain by habit;" a line expressing this will be the "curve of habit."

We now make two sets of experiments under exactly the same conditions, but in one set we warn the subject about two seconds beforehand when to expect the sound whereas in the other we tell him to watch all the time. The results show a somewhat longer reaction in the second set and a far larger average variation; for example, 0.159s and 0.047s as contrasted with 0.142s and 0.008$. We find this to be always the case, and we established the principle that a state of relaxation with a call to attention just before something is to be done enables us to act somewhat more quickly and far more regularly than an attempted state of continuous attention. In the former state we attain maxima of energy by relaxing between times; in the latter we keep at a dull level. It is on a small scale the same principle of intermittent activity and recuperation that is involved in sleep. Distraction of attention likewise produces a longer average time and a far larger average

variation.

Persons may be divided into groups according to their reaction-times. Four types of persons are familiar to the physician. The self-controlled man of abundant vitality reacts quickly and regularly; the phlegmatic or relaxed man responds regularly but slowly; the excitable man of strong vitality gives quick but variable responses; the neurasthenic weak

ling is excessively irregular and his average reactiontimes are slower than normal.

Conditions of disease also show themselves in the reaction-times. In the first stages of intoxication by alcohol the reactions may be quicker and more regular but they soon become slower and very irregular; in an attack of alcoholism where the person's nerves are on edge so that he cannot sleep and feels himself on the verge of delirium tremens, his average reactiontime is very short and his regularity is remarkably good.

The reaction-time depends somewhat on the kind of sensation.

tones.

For noises it is a trifle shorter than for For example, a person who reacts to a noise in 112 will take perhaps 15 for a tone. Even the whistle of a locomotive is not so conducive to a quick jump by the passengers on the platform as a sudden escape of steam.

A particular case of reaction to sound is found in starting a race. In short-distance, or sprint, racing the time required for the

electric contact is put on the end of the starter's

pistol. The arrangement is shown in Fig. 20. The

ε

FIG. 21.-The Runner's Key.

firing of the

pistol causes

the wing to fly back and break an electric circuit, thus making a record. A runner's key of the kind shown in Fig. 21 is attached to the runner by a

thread. The start of the runner jerks and breaks the thread; this moves the lever and makes another record. It is noticeable that long-distance runners are

FIG. 22.-Measuring a Runner's Reaction-time.

very much slower than sprint-runners who practise quick starting; this shows that the reaction-time can

be reduced by practice. The reaction-time seems to be longer where the whole body has to be started than where only a finger is moved; the mass to be moved thus seems to have an influence on the time. In some races the pistol has gone off and the photograph has been made of the runners before they have reacted. The reaction

time to touch can

be found by using the instrument shown in

Fig. 23. The flex

ible conductors

FIG. 23.-The Touch-key.

carry the current through the screws of this touch key and then through the reaction-key. The person experimented upon closes his eyes. Some one takes the touch-key and touches him, whereupon he reacts by moving his finger. Both keys make records on the drum.

A weak touch is answered by a slower reaction than a moderately strong one. As the touch becomes stronger the reaction-time decreases, but when it becomes very strong the time is again lengthened.

To experiment on the reaction-time for temperaturesensations a metal ball is screwed on the touch-key in place of the rubber tip. The ball is heated or cooled as desired. The reaction-time for cold is somewhat shorter than that for hot, and both are longer than for touch. For example, the figures for one experimenter were: touch, 112; cold, 122; hot, 13 Z.

The reaction-time to light may be found by using an electric flash. The intensity of the light has