blocks of wood at the base, so that the pieces would easily stand upright. The middle piece, which is rectangular and higher than the rest, was placed a little in front of the rest of the figure. The students were asked to describe precisely what they saw; and with one exception they all described, in different words, a semicircular piece of cardboard with a rectangular piece in front of it. In reality there was no half-circle of cardboard, but only portions of two quarter-circles with the portion back of the middle piece omitted. The students, of course, were well aware that their physical eyes could not see what was behind the middle cardboard, but they inferred, quite naturally, that the two side pieces were parts of one continuous semicircle. This they saw, so far as they saw it at all, with their mind's eye. III [Illustration: FIG. 4.--The black and white portions of this design are precisely alike; but the effect of looking at the figure as a pattern in black upon a white background, or as a pattern in white upon a black background is quite different, although the difference is not easily described.] There is a further interesting class of illustrations in which a single outward impression changes its character according as it is viewed as representing one thing or another. In a general way we see the same thing all the time, and the image on the retina does not change. But as we shift the attention from one portion of the view to another, or as we view it with a different mental conception of what the figure represents, it assumes a different aspect, and to our mental eye becomes quite a different thing. A slight but interesting change takes place if we view Fig. 4 first with the conception that the black is the pattern to be seen and the white the background, and again try to see the white as the pattern against a black background. I give a further illustration of such a change in Fig. 5. In our first and natural view of this we focus the attention upon the black lines and observe the familiar illusion, that the four vertical black bands seem far from parallel. That they are parallel can be verified by measurement, or by covering up all of the diagram except the four main bands. But if the white part of the diagram be conceived as the design against a black background, then the design is no longer the same, and with this change the illusion disappears, and the four bands seem parallel, as they really are. It may require a little effort to bring about this change, but it is marked when once realized. [Illustration: FIG. 5.--When this figure is viewed as a black pattern on a white background, the four main vertical black bands seem far from parallel; when it is viewed as a white pattern on a black background the pattern is different and the illusion disappears (or nearly so), and the four black bands as well as the five white ones seem more nearly parallel.]