With but few exceptions, the errors were, as a rule, of one place. The series with an odd number of sheets (5, 7, 9) gave better results than those with an even number (6, 8, 10). In the tests with the odd number of sheets the experimenter (K.) stood in front of the middle sheet, so that it was at the apex of a right angle made by the series of papers and the median plane of the subject's body; whereas in the case of the even number of papers the subject stood opposite the space between the two middle sheets, thus making the position of the sheets less favorable. In the preceding tests the distance between the centers of the neighboring sheets was always 50 centimeters, so that the angle through which the median plane of the experimenter's body would have to turn in order to pass from one sheet to the next, was about 3-3/4 degrees. In the following tests these distances were gradually decreased. The sheets, always five in number, were replaced by ever narrower white strips of paper mounted on dark cardboard and illumined by a Nernst lamp. The following table shows the decrease in correct inferences running parallel with the decrease of the angle through which the subject would have to turn in order to be in line with the several pieces of a series successively. The percentage in each case is based upon at least 100 tests. Angle: 3-3/4 deg. 3 deg. 2-1/2 deg. 2 deg. 1-1/2 deg. 1 deg. Distance between the centres of two neighboring papers: 50cm. 39cm. 33cm. 26cm. 20cm. 13cm. No. of correct inferences: 77% 73% 71% 68% 66% 61%