24 4-yr-olds were tested on 2 types of angular figures--a baseline with another line joining at the end at 45, 90, or 135° and a baseline with line joining at the middle at 45 or 90°--to establish whether a perpendicular bias would appear independently of bisection. It is noted that children distort angular figures so that the constituent angles are nearer 90° than they should be, which could be due to a perpendicular bias, a bisection bias, or both. Results show that perpendicular errors were obtained both for end and middle figures, but overall more strongly for middle figures. While 90° middle figures were copied more accurately than 45 or 135° figures, this effect was only obtained for vertical and horizontal presentations of end figures and was reversed for oblique presentations. For end figures, directional errors varied with subtended-line orientation, whereas they varied with baseline orientation for middle figures. It is concluded that, although errors toward the perpendicular do occur with single-angle figures, angle equalization may take place when there are 2 adjacent angles in the figure. An interpretation of the differing orientation effects is that, in middle figures, strong internal relational forces produce a distortion that varies with the angle at which the figure is viewed; in end figures, the absence of relational forces within the figure leads to a stronger influence from external cues. (19 ref) (PsycINFO Database Record (c) 2012 APA, all rights reserved)