86 ON THE RELATION OP OPTICS TO PAIXTINQ. 



the fi.ae gradation and transitions of light and shade 

 on rounded surfaces, which are his chief means of ex- 

 pressing their modelling, with all their fine changes of 

 curvature ; he must take into account the extension or 

 restriction of the sources of light, and the mutual 

 reflection of the surfaces on each other. While the 

 modifications of the lighting on the surface of bodies 

 themselves is often dubious for instance, an intaglio 

 of a medal may, with a particular illumination, pro- 

 duce the impression of reliefs which are only illumi- 

 nated from the other side double shadows, on the 

 contrary, are undoubted indications that the body which 

 throws the shadow is nearer the source of light than 

 that which receives the shadow. This rule is so com- 

 pletely without exception, that even in stereoscopic 

 views a falsely placed double shadow may destroy or 

 confuse the entire illusion. 



The various kinds of illumination are not all equally 

 favourable for obtaining the full effect of shadows. 

 When the observer looks at the objects in the same 

 direction as that in which light falls upon them, he 

 sees only their illuminated sides and nothing of the 

 shadow; the whole relief which the shadows could give 

 then disappears. If the object is between the source 

 of light and the observer he only sees the shadows. 

 Hence we need lateral illumination for a picturesque 

 shading ; and over surfaces which like those of plane 



