The presnt study was planned to investigate the perceptual solidity in children, namely the younger children are apt to fail in finding simple figures hidden in complex figures; and the possibility of it's alternationby using emdedded fiigures. To provide the distances among figures, 3 types of embedded figures were made by adding surplus lines inside of simple figures (C1), outside of them (C2), and combining C1 and C2 (C3). In Experiment I, children of four different ages (4, 5, 6 and 7 years old) were repuired to perform 3 different types of responces in each figure: tracing with red-pencil, with finger and putting plastic figures. The main results were: (1) There were significant differeces in difficulties of 3types of embedded figures in the all grbups and under all responce patterns: C1 (easy)-C2 (middle)-C3 (dificult). (2) The right responces were increasing with ages under all responce patterns: putting plastic figures (easy)-tracing either red-pencil (middle)- tracing with finger (difficult). The main purpose of Experiment II was to examine the effect of preceding perceptual experiences (training) on the following perceptual situation (test). To be concrete,(1) In the perceptual learning situation where the types of C1 er C2 are used as training stimuli and C3 as test stimuli, it is supposed that the different types of training figures contribute differently to the following test.(2) It is also supposed that the improvement of right responces by training appear differently concerning ages.(3) Is there any tranofer of training to the not-trained stimuli? If there is some transfer, how it appear concerning the types of training stimuli and ages? Materials used were identical to in Experiment I. Subjects were kindergarten children (5.6 years old) and elementary scool children (7years old). Subjects of each age group were divided into 4 subgroups: 3, experimental groups (trained with C1, C2 or Simple figure respectively) and control group. The results were: (1) The increase of right responces from Pre-Test to Post-Test were found in all age groups and all expermental groups. The rate of this increase enlarged progressively with ages. It enlarged also differently concerning the training conditions: C2 (most effective)-Simple figures, C1 (middling)-control (less). (2) The transfer of training was found, but it appeared differently as to the age groupe: elementary school children showed a high degree of increase in both trained figures and not-trained figures, elder kindergarten children showed high degree of increase for trained figures but less for nottrained figures and the younger kindergarten children showed less in both.