An 18 month follow-up observation was made on 10 normal children (ages 2: 7-4: 0 at the beginning of follow-up), and 30 debile children (ages then 6: 0-12: 0). Figures used are: Vertical line, Horizontal line, Oblique line (inclined at 45 degree), Greek cross, X-shape (St. Andrew's cross), Square, Tilted square (standing on one of its apex), and Diamond. Children were to draw them, compose them with sticks, and represent them by dotting ends or corner points. Results and discussions. (1) It was assumed from preparatory observations and children's confessions that they pe rceived the figures correctly enough, but they met difficulties in composing or arranging lines to make the figure after a model. In other words, they were not agnostical, but apraxical. For example, a 2: 6 child, who could draw oblique lines in his spontaneous drawing, was puzzled with an oblique line as model. He laboured to represent it with vertical orhorizontal lines, or with some curved lines, saying “I cannot do it.” This constructive power seemed much more to do with general intellectual development, rather than with manual dexterity. But we discussed a case with some symptoms of developmental agraphia and constructive apraxia, showing selective defects in dealing with letters and geometrical figures. In such a case the special deficiency did not tell his general ability level. Such special deficiencies have been reported as the mark of “brain damage”. But we have on the other hand a motor aphasic case which shows selective defect in language activities, not in perceiving or constructing visual figures. Moreover, in dealing with visual figures, defect of perception and that of construction do not always go together. Symptoms of “brain damage” should be observed in a more integrated vision of clinical study on aphasic, apractic, agnostic cases. (2) On the basis of the % of successful children, we could arrange from the easiest form to the most difficult one as follows: Vertical and Horizontal lines, Greek cross, Oblique lines and Square, X-shape, Tilted square, and Diamond. Constructing with sticks, or by dotting the cornerpoints, was a little easier than copying with pencil, with some exceptions in easier forms. After all, the dominant factor of the difficulties seemed to be in the properties of the figures, not in the means of reproducing them. Diamond was distinctly more difficult than Tilted square, in that the latter was easily grasped as a whole, while the former was not. Many children who could make the tilted square with sticks failed in the diamond. Evan after having arranged three sticks correctly, they often were quite puzzled as to how to set the last side to complete the diamond. (3) Following up the normal children, we could distinguish several stages of copying power. 1) 2: 6-3: 6, in this stage they went as far as to succeed in copying the square, but often failed in the oblique forms. 2) 3: 6-4: 6, they succeeded in composing with stiks or dots the tilted square or sometimes the diamond, but failed in copying the diamond. 3) 4: 6-, they succeeded in copying the diamond. Mentally relarded children showed a retardation in attaining those stages as we had expected ; e. g. those who could do well with the diamond were to be seen after the age 9: 6. (4) Can the copying ability be improved by instructions? We know that it is much easier for a child to “imitate” the drawing behavior of an instructor than to “copy” after a model. And we can train a child to draw a diamond even at the age of 3: 6. But to accomplish a form somehow or othe r is one thing, and to copy it after a model is another. The latter being our problem, the two methods were adopted here. 1) When a child succeeds by sticks and fails in copying, he is told to compare the two results and given chance to try ag ain. 2) Children trace a model, and then copy it again.