As the results in Table 31 show, the time and number of errors rapidly diminished. Number 7, for example, made no errors in the second test. The chiefly significant fact which appeared in these preliminary experiments, however, was that the mice soon ceased to care whether they got out of the labyrinth or not. After they knew the path perfectly, they would enter the wrong passages repeatedly and wander about indefinitely. It was obvious, therefore, that the labyrinth could not be used to reveal the role of sight unless some sufficiently strong motive for continuous effort to escape from it could be discovered. Naturally I looked to the electric shock for aid. The labyrinth of Figure 23, which for convenience in distinguishing it from several other forms to be described later I have designated as labyrinth B, was placed upon a board 90 cm. long and 30 cm. wide about which had been wound two pieces of phosphor bronze wire after the manner described on p. 94. At _O_, Figure 24, there was an opening closed by a swinging door which led into a box 40 by 24 cm. In one corner of this box was a small nest-box. The significance of this rearrangement of the labyrinth is apparent. As in the preliminary tests, the dancer was started at I, but instead of being allowed to wander about without any other result than delay in escape, it was given a shock each time it made an error. The satisfaction of escaping from the narrow bounds of the labyrinth's passages, which alone was not strong enough to impel a dancer constantly to do its best to escape, was thus supplemented by the powerful and all-controlling tendency to avoid the disagreeable stimulus which resulted from entering certain of the passages. The result of this modification of method is strikingly exhibited by the data of Table 32. [Illustration: Figure 24.--Labyrinth B on an interrupted circuit board. _I_-1-2-3-_O_, labyrinth path; _B_, nest-box; _N_, nest; _EW_, board wound with phosphor bronze wire; _IC_, induction apparatus; _C_ electric cell; _K_, key.]