Formerly, we set to project a plane net from one side of an aquarium (Fig. 1), and observed the direction in which a fish-group moves away on passing the free end of the net, after having moved along the side and the net Namely, we measured in many fish-groups the angle, φ, between the produced line of the projecting net and that direction for several values of the angle, θ, between the projecting net and the side of aquarium along which the groups had moved on. The obtained results are as follows: The value of θ differs with fish-groups for a constant value of φ, and the frequency distribution curve of θ has a normal form. The mode of φ decreases as θ increases, but the frequency at the mode increases. This relation holds only when another group does not exist near the free end, when a group moves away on passing that end. When another group exists near the free end, it attracts to some extent a group which goes on along the projecting net. In this case, the attracting effect increases as the angle, φ', between the line of sight of attracting group at the free end and the produced line of the net decreases and as the distance, s, of attracting agent from the free end decreases. A group in motion has a stronger effect than at rest.
The present study was made to know how a fish-group would be attracted. The method taken in this study was the same as stated above, i.e., the frequency distribution of φ. was obtained in any case whatever an attracting agent was put near the free end of projecting net or not. An example of the distribution is given in Fig. 2. In this figure, curve I represents the distribution of control, i.e., the case where any attracting agent was not placed. Curve II is the distribution in the case where two attracting agents were placed near the end of projecting net. Curve II1 is the distribution selected in such a manner that it is similar to I and it coincides with II in the direction which deviates for from the agents. Curve II2 represents the resulting distribution in consequence of the subtraction II1 from II. Curve II2 thus obtained was of normal form in most cases where a single agent was placed. When two agents were placed at the same time subtending a large angle at the free end of projecting net, curve II2 could be divided into two normal distributions II2' and II2" The two distributions may be supposed to be due to those two attracting agents.
Assuming that the distribution II1 belongs to the groups which were not influenced by the agent and that the distribution II2 to the attracted groups, we measured the strength of attracting effect of the agent in terms of the percentage of the area beneath the curve II2 to that beneath the curve II.
The strength of att ?? acting effect of various agents thus obtained, and the mode and the quartile deviation of the distributions I, II1 and II2 (II2' and II2" separable if possible) are given in Table 1.