The present study was mainly intended to inquire into the effects of preceding trials over the performance of following problem series when the preceding series was diferent from the subsequent one about the cue for the solution. Simple anagram series were used as problems, and were given to junior high school students. Experimental series were divided into the preceding and the subsequent series with respect to the letter-orders used as cues. Though two series were arranged with different letter-orders, these letter-orders were, generally, constant through each series. Each anagram was presented one after another to the subjects at the rate of 10 seconds per one. 1) Experiment I was designed to find the effects of trial number of preceding series. Subjects were divided into groups in terms of the number of trials in the preceding series-6, 12, 24, 42, and 60 trials, five groups in all. There was no transfer effect in the group which performed 6 trials in the preceding series. In the group which performed 12 trials, rather negative transfer was observed in terms of the median on trials to reach the learning criterion though there was a positive transfer with respect to the total number of correct responses in the subsequent series. More marked negative transfer was found in the group which performed 24 trials. As the number of preceding trials, however, became as large as 42 or 66 significant positive transfer appeared. In order to explain the non-monotonous variation of transfer effects indicated by these results, it was hypothesized that a response tendency at each trial was determined by the composition or algebraic summation of the two tendencies existed simultaneously at the trial, viz., the facilitation and the inhibition, and, that these two tendencies were different in the process of development though they were both functions of the number of trials. In addition, inquiry was made about the case where the letter-order of preceding series was not constant but was varied at random every trial. Under this condition, only the group which performed 24 preceding trials showed negative transfer, and the other four groups showed no transfer. 2) Experiment II was designed to investigate the case where there were differences between the preceding and subsequent series with respect to the difficulty of solution. In the case where the problem had similar difficulty in the both series, the results were the same as in Experiment I. In the case where the preceding series was easier than the subsequent, the phase of negative transfer was hard to appear while there was no great difference about the phase of positive transfer. In the case where the preceding series was more difficult, the phase of negative transfer was, inversely, hard to appear while there was no great difference about the phase of positive transfer. These results were also explained by the above-mentioned two-factor hypothesis. 3) The relation between two cases where the letter-orders were constant and random through the preceding series was examined in Experiment III. In Experiment IV, the case where the shift of cues between two series was made gradually was tested, and moreover, in Experiment V, the case where cues were changed one after another every block of trials was investigated. But these results were not necessarily clear.
In the previous investigations we analyzed the relationships between three R's (relations) included in the triad (the social situation composed of three person: the chooser p, the chosen o and the third person q). As a consequence, it was suggested that our initial hypothesis derived from F. Heider's hypothesis should be corrected to adjust itself better to the results. This is the first report of the research designed to inquire more deeply into the consistent discrepancy between Heider's hypothesis and our findings. Presumably there are two possible roots to cause such discrepancy. One of them is inadequacy of his hypothesis, some evidence for which has recently been presented by J. P. Runkel. The other is the peculiarity of the method of collecting and analyzing data employed in our investigations. This latter problem is to be discussed in the succeeding paper. In the present investigation we examine Heider's hypothesis once more from the viewpoint of perception of mutual relations. Up to the last investigation we dealt with only three R's, as does Heider: p's attitude toward q (R1), p's attitude toward o (R2) and p's perception of q's attitude toward o (R3). In the triad, however, there exist three more R's which are inverse of R1, R2 and R3 respectively. Heider does not mention them, for, we guess, he assumes implicitly that two mutually inverse R's have the identical sign. This implicit assumption is reasonable enough from Heider's standpoint, but our past findings cast a doubt upon the absolute validity of it. In the present investigation, then, we analyze the relationship between R3 and its inverse: p's perception of o's attitude toward q (R4). Subjects of the investigation were some two hundred pupils of sixth grade in an elementary school. They were given nearsociometric tests and four kinds of relation perception test concerning their homo-sexual classmates. The major findings are as follows: In the first place, a remarkable tendency was found that the signs of R3 and R4 become identical. This means that Heider's assumption is valid in general. The strength of the tendency, however, was never uniform: it was found that it varied with two conditions. First, it depends on the very signs of R3 and R4: they tend to converge into  less remarkably than into [+] or [-]. This might mean that  is less powerful than other two signs in affecting the state of the configuration. Another condition which influences the strength of the tendency is whether the signs of R1 and R2 are identical or not: the tendency is stronger when they are identical than when they are different. This means that the signs of R3 and R4 depend not only on each other but also on the signs of R1 and R2. Moreover, the fuller analysis showed that the way of dependence of the former on the latter is consitent with the results obtained in our past investigation. It may be concluded, then, that the present investigation gave an additional evidence for the inadequacy of Heider's hypothesis within the limits of such procedures as we employed.
The purpose of this study is to analyse the personality traits of a single person by means of P-technique factorization and to compare them with the common personality traits which have been revealed by the ordinary R-technique studies. A normal nineteen-year-old female student sat as the subject for forty-eight days successively. Thirteen variables were obtained from experimental objective tests, and twelve variables from self-rating procedures. The centroid method of factorization and the oblique rotations of axes were applied to the matrices of correlations based upon the 48 day-to-day variations of the intra-individual scores. Four significant factors were analysed from both groups of data, and the results were compared with those which Cattell and others had already obtained chiefly by means of R-technique analyses. The first factor of the objective tests is to be clearly identified with the source trait “C” of R-technique studies, while the fourth of the self-rating data, too, seems to correspond to it. At the positive sides of these factors, therefore, are found the variables associated with mental stability, emotional balance, volitional control, and so forth. Next, on the third factor of objective tests are loaded the variables of PGR deflections; hence it is interpreted to be similar to the factor of “emotional abundance” which has been previously revealed as an objective test factor of a normal individual and been principally associated with the source trait “H”. Finally, the third factor of self-rating data is considered to be common to the source trait “A”, for it clearly shows the symptoms of cyclothymia and schizothymia. These results tell that most of the P-technique factors obtained from our subject have correspondences to some of the R-technique ones already found and recognized in the previous factorial analyses. Not all the factors analysed here, however, correspond to the common source traits, for there are left some which seem to be peculiar to the subject. So far as this study is concerned, these probable unique factors are not clearly explained, but if we are provided with abundant case-by-case P-technique data not only of normal persons but also of abnormal ones, it may be possible to develop more diagnostic or clinical interpretations of the results.