276 Thomas Stevenson on the 



Let us suppose, then, that the parallel rays proceeding from 

 this apparatus are required to be deflected through a larger 

 horizontal angle than in the former case, and that it is farther 

 desirable to spread the rays more uniformly over the arc to be 

 illuminated. Instead of one set of the straight refractors which 

 are used in the former case, let there be four sets of such refrac- 

 tors, c c, figs. 3 and 4, each having two totally reflecting 

 straight prisms, whose horizontal cross section is similar to 

 that of the totally reflecting prisms pp, which are used in the 

 holophotal apparatus of glass. Each set of refracting and 

 totally reflecting straight prisms will have its own focus /' 

 in front, from which the rays will diverge through the azi- 

 muthal angle due to the number of the prisms in each set, so 

 that the denser, as well as the weaker portions of the light, 

 will thus be destributed with sufficient uniformity over the 

 whole arc. The same principle will also be found very useful 

 for Apparent lights,* where the light near the confines of the 

 illuminated arc has been found to become faint. 



In order to save the loss of light due to the absorption and 

 superficial reflection resulting from the use of two optical 

 agents, I propose, in certain cases, to substitute for the lens 

 commonly employed one on a new principle, having one side 

 ground into the form of the straight refracting prisms c, c, c, c, 

 shown at figs 5, 6, 7, while the other side remains convex, but 

 of different radius. Fig. 5 shows the elevation of the outer sur- 

 face of a lens on this principle, while fig. 6 shows the elevation 

 of its inner surface, and fig. 7 its middle horizontal cross section. 

 In the lens which I have represented in the figures, the rays con- 

 tained in the larger conical angle a?, /, x, are spread over the 

 smaller horizontal angle x f x . The want of divergence which 

 has often been complained of in the large polyzonal lenses in 

 our revolving lights might easily be remedied by adopting this 

 principle of construction. The convexity of the straight 

 prisms in the horizontal plane would of course in such a case 

 be very small. The bull's eye lenses used in hand lanterns, 

 or in railway signal lights, might also with advantage be made 

 on a similar principle. 



* Vide, Description of the Stornoway Apparent light, erected on a sunk rock 

 in Stornoway Bay, the illumination of which is derived from a distant lamp 

 situated on the shore. — Trans. Roy. Scott. Soc. of Arts for 1854. 



