sides, as in the diagrams A, B, C (Fig. 1), though with a feebler outline. A picture thrown off quite out of drawing by such a lens, in which straight lines turn out as curves, is evidently inaccurate. The inaccuracy may not be felt by many, but it exists. It may perhaps be expected that this defect disappears in the case of what are called correct lenses, but let the attempt be made to obtain a view with these correct lenses of lofty buildings taken from a low position. The lines that ought to be perpendicular commonly converge upward. This is caused by the photographer being obliged to direct his instrument at an acute angle upward, in order to be able to take in a view of the whole building. In doing this, perpendicular lines project themselves, converging upward. To avoid this defect, lenses have been made with a very large field of view. These are called pantoscopes. But these reproduce distant objects apparently on a very
Fig. 2.
small scale, and objects near at hand on a very large scale—peculiarities unnoticed by unprofessional persons, but detected by close observers of Nature.
A remarkable phenomenon, exciting the wonder of the uninitiated, is the distortion of spheres in photography. Let the reader imagine a row of cannon-balls; these will always appear balls to us, and the artist will always draw them as a circle. But, if they are taken through a lens with a large field of view, the balls situated near the rim of the lens no longer appear circular, but elliptical.
To explain this phenomenon, we must attend once more to the mode in which the picture is produced. Let it be conceived that there are three balls, A B C, in front of a camera, K, with the lens o (Fig. 2). Each ball projects a cone of rays on the optical centre of the lens. This is continued within the camera, and cuts the surface of the picture, if its axis falls obliquely upon it, in the form of an ellipse, such as A C. Only, if the axis of the cone of rays is perpen-
