Using Skinner box, we examined the behavior of white rats in the learning of barpressing habit by the partial reinforcement procedure at fixed ratios.Sixty-five rats were divided into 13 groups and trained to learn bar-pressing responses under predetermined schedules (Number of reinforcements / Number of responses : 24/24, 48/48, 96/96, 36/48, 48/64, 96/128, 24/48, 48/96, 96/192, 12/48, 24/96, 48/192, and 0/0), and then extinguished. We analyzed the results in regard to three aspects : number of reinforcements and responses, and ratio of reinforcement.And then we examined the methodological problem of criterion of socalled resistance to extinction, which had hitherto been determined arbitrarily.Findings were as follows :1) The acquisition of the habit was, as Jenkins and Stanley report, more difficult by the procedure of partial reinforcement in the early stage of training, but with the progress of training the difficulty decreased, and at last about the same level of performance was attained as by the procedure of continuous reinforcement.2) The after-effects of the differential acquisition training on the socalled resistance to extinction were :a) In regare to the aspect of the same ratio of reinforcements, the more the responses (therefore reinforcements) in acquisition, the more the responses were emitted in the first extinction session. This suggests the functioning of the principle of effect.b) In regard to the aspect of the same number of reinforcements, the more the responses in acquisition, the more the responses were emitted in the first extinction session. This suggests the functioning of the principle of expectancy or substitution.An interesting fact to us was :c) In regard to the aspect of the same number of responses, the less the number of reinforcements, the responses emitted in the first extinction session showed increasing trend rather than decreasing.It is impossible to interpret this fact by the hypothesis of secondary reinforcement, not to speak of the hypothesis of generalization of after-effect of reinforcement and of response unit.Rather it is supposed that the principle of substitution, particularly the principle functioning in the phenomenon named positive reciprocal induction by Pavlov was powerfully at work.On the problem of the criterion of the socalled resistance to extinction, the following were made clear by computing the coefficients of correlation among the number of responses up to the criteria of 3, 5, and 10 minutes of non-response in two extinction sessions after 24 and 48 hrs. of acquisition training.1) The number of responses emitted in the second extinction session (spontaneous recovery after 24 hrs. of the first extinction session) was little affected by the differential conditions of acquisition training.2) The coefficients of rank correlation of the number of responses emitted in the first extinction session was the higher, the longer the time of the criterion of resistance to extinction. According to these results, it will be necessary and adequate to take the criterion of 10 minutes of non-response to make the number of responses in extinction session the index of effects of differential training conditions.