Abstract
Educable retarded (n =71) and intellectually normal (n =71) boys were measured for both vertical and horizontal errors on an overarm throwing task. The accuracy of throwing errors in terms of the total variance about the target of the intellectually normal boys was superior to that of the retarded boys. Variable errors and constant errors (absolute values) of the retarded boys were greater than those measures for the normals. The relative contributions of variable error to the total variance of the retarded boys were not substantially greater on either horizontal or vertical measures than were those for the normals. These findings did not support the variability hypothesis (Baumeister, 1968) which states that variable error contributes more to the performance levels of retarded than of normals.