Illumination of small Arcs of the Horizon. 275 



This instrument, therefore, fulfils the condition of collecting 

 the entire sphere of diverging rays into one parallel beam. 

 Let c c, figs. 1 and 2, represent a series of straight prisms 

 placed vertically and in front of the apparatus just described, 

 and suppose their horizontal cross sections to be similar to 

 those of the lens I, and their length to be equal to the greatest 

 diameter of the paraboloid. It is obvious that the parallel 

 rays impinging upon these prisms will be refracted in the 

 horizontal, but will suffer no deviation in the vertical plane. 

 The horizontal refraction will be similar to what takes place 

 at the lens I, but will lie in the opposite direction, so that the 

 rays will converge to a focus, /, in front, and will again 

 diverge from that focus in the same angle in which they were 

 made to converge towards it. An observer at any point in the 

 azimuthal angle, G, /', H, will therefore have the benefit of a 

 vertical strip of light whose height will be equal to the diame- 

 ter of the paraboloid, a a, and whose width will be propor- 

 tioned to the breadth of the flame employed. The principal 

 objection to this arrangement is the inequality which must 

 obviously exist in the intensity of the light as the observer 

 passes from the middle of the azimuthal angle where the light 

 will be brightest, towards the limits of that angle on either side 

 where it will be weakest ; for the lateral prisms do not inter- 

 cept so large a portion of light as those which are nearer the 

 centre of the beam of parallel rays. 



In order to remove this objection, as well as to reduce the 

 loss of light caused by absorption, I propose to use several 

 sets of straight prisms, instead of having a single large one 

 embracing the whole width of the holophotal apparatus. Such 

 an arrangement is shown in figs. 3 and 4, in which is also re- 

 presented the most perfect form of holophotal apparatus, in 

 which totally reflecting prisms p and b are substituted for 

 metallic reflection. The loss by absorption, which in metallic 

 reflection amounts to about one-half of the whole incident 

 light, is thus saved. It is unnecessary here to give a detailed 

 description of this apparatus, as I have published it in th 

 Transactions of the Royal Scottish Society of Arts, vol. iv., 

 and as I have already in this paper fully described the nature 

 of the holophotal action in the case where metallic reflection 

 is employed. 



