In the cases of the more suitable subjects I was able to indicate not only the number they had in mind, but also the divisions in which the number was thought, thus 12 as 5 and 5 and 2, or the same number as 2 and 5 and 5, and I was also able to determine the addends in the addition--i. e., whether the problem had been conceived as 3+2=5 or as 2+3=5. It frequently happened that in the beginning I would sometimes mistake these subdivisions, which were recognizable by the less pronounced jerks, for the final number. Thus I would often respond with 4 instead of 8, or 3 instead of 9, or with 3 when the problem was 3+2, just as Hans had so often done. In these tests, too, the difficulty of getting the number 1, as well as the larger numbers, came to light. Thus three times in succession 17 was indicated as 4, as 9, and as 17. But after some practice I was able to give numbers as high as 58 and 96. The frequency of the errors of one unit too many and of one unit too few is also noticeable in these tests.
We also found desirable corroboration, by trustworthy subjects, of the introspective observations of the author, which were reported in Chapter III, with regard to the significance of concentration and the curve of attention. It is hardly necessary to mention that no attempts were made to influence the subjects in their accounts by asking suggestive questions. The most valuable feature about these tests was that the mute horse had now been replaced, as it were, by an animal capable of speech, and that it was now possible to follow the same process both from within and from without. Two illustrations may be welcome. The one who took the part of the horse gave three taps and made the following entry: "At 3 I saw a slight upward jerk of the head on the part of the questioner". The questioner however had thought of 4, and made the following note, without knowledge of the other's entry: "I was aware of extreme tension, so that it was impossible for me to get beyond 3". Or again, the 'horse', reacting to a movement on the part of the questioner, stopped at 3, but the latter, having intended to obtain 2, made the following entry: "I noted clearly that I ceased thinking of the number too late, and did not put on the brakes, as it were, until I had arrived at 3". We see that errors here were entirely the fault of the questioner, just as had been the case in the tests with Hans. (See page 151f.).