The midpoint of the axis of bisection in a triangle appears to be displaced toward the apex so that the apical extent seems to be shorter than the basal extent, an effect originally reported in 1895 by Thiéry and later in 1898 by Wundt and, therefore, referred to here as the Thiéry-Wundt illusion. Following a demonstration strongly suggesting that the illusion is yet another version of the Müller-Lyer effect in a stimulus figure with inward-directed angles, four exploratory experiments designed to throw more light on this group of illusions are reported. The first showed that the effect occurs in an open angle, between converging lines that do not meet to form an apex, between converging stepped lines, and when one of the converging lines is removed, leaving a single oblique line. The other three experiments showed that the illusion also occurs in an open semicircle and a rectangular bracket, is weakened by the addition of a complete or partial baseline to form a triangle, and weakly but reliably when the angle is minimally formed from dots marking the ends of oblique lines. On the basis of these data, Judd's version of the conventional Müller-Lyer figure, and informal but easily repeatable observations, it is concluded that the Thiéry-Wundt illusion is, like other variants of the Müller-Lyer group of illusions, due to a process of directional biasing-an extension of the concept of biasing proposed originally by Morgan et al (1990, Vision Research 30 1793-1810).