Every picture may be considered as a section of this cone, by a plane perpendicular to the horizon, and, therefore, to the central ray of the cone. Then the question is, should the intersecting plane include more than the area of the cone at the point of intersection, or exactly its area, or less than its area? If it include more than its area, we shall not be able to see the edge of the picture, if we stand in the proper point for seeing the rest of it. All the artist's labour on the edge will, therefore, be lost on those who know where to stand; and its effect on those who do not, will be to make them stand in a wrong place. If it include exactly the area, the edge of the picture becomes a substitute for the natural limit of sight, and everything takes its true position. If it include less than its area, we feel that we could see, and should naturally see, more than the artist has given us; the edge of the picture becomes a cutting, interfering, distinct termination, just as the edge of a window is, when the spectator is kept twelve feet back into the room. We wish to get the edge out of our way, and to see what is behind it: and the ease, beauty, and propriety of the painting is entirely disguised or destroyed. According to this reasoning, then, our pictures should all be circular, and of such a size that the distance of the eye from their centre should equal their diameter. But we see that all artists, as a general principle, make their pictures parallelograms of varied proportion. This is a proof that such a form is desirable, and something very near a proof that it is proper. We have, therefore, to investigate three questions:-I. What are the causes which render such a form desirable? II. What are the principles on which such a form is admissible? III. What are the limitations under which such a form is to be given?