This change was preferred to that in Fig. 12, although different reasons were given. One said it was because this change in arrangement made the elements more distinct, hence easier to keep apart, while in Fig. 12 they were made more alike. Moreover, one element seemed as important as the other. He did not class them as major or minor, so he could not compare the relative values of symmetry in principal and alternate units, for in this series he did not feel the distinction. The other answers to this question were rather incoherent, but the series did not seem to suffer much change, either pleasantly or otherwise. Since lack of symmetry in _the element_ was at least tolerated in the examples already given, would it be allowable so to place the units that the two adjacent to any one unit should lie unsymmetrically on either side, that is, may the elements lie unsymmetrically with regard to one another? Suppose a four-group to be repeated at regular intervals, and a three-group likewise; if the two series were combined, must they occur halfway between each other? That is, must they be symmetrically placed as regards the intervening space, or could they be put to one side? [Illustration: Fig. 14] The subject was asked not to group them (as in previous similar arrangements), but to keep them as separate repetitions if possible, and to see if this equal distance was necessary to keep them apart. The result was the same in all cases. The subjects could not help grouping them, and found it impossible to keep them distinct unless so much effort was put into it that no pleasure was left. They said they "_knew_ each unit was as equally distant from the next unit in its _own_ series, as if it did not come at unequal distances from the units in the other, but they could not feel it so, and were obliged to group the two together." For this reason the experiments did not satisfactorily illustrate the point in question. It was necessary to have a series of elements whose unity was more strongly marked, and whose different parts would still remain one _whole_ even after variations, instead of shifting into each other. It was suggested by these imperfect experiments that symmetry was _not_ so important a factor in the different units of a series as the subjects had previously supposed; but that, on the other hand, the different units must be placed at equal distances from each other, if they are to be kept distinct either as two series or as one. Moreover, that _two_ series could not be kept distinctly in mind as separate, _anyway_, without fatigue, the tendency being always to group them into one series with a new repeated element, composed of a combination of the other two. It was necessary, however, to test this more completely. By a simple device the former series was changed radically, so that the difficulties mentioned were overcome. The strings of both the three and four groups were twisted together at the bottom, thus binding them