168 MR, RICHARD ROBERTS ON PROPOSED 
intensity, while the dioptric method perfectly fulfils this 
condition by distributing the rays equally to every point 
of the horizon; and not only does the French apparatus, 
as lately improved, produce as the average effect of the 
same combustion of oil about four times the amount of 
light that is obtained from the catoptric mode, but its 
annual maintenance, including interest on first cost of 
apparatus, is considerably less. In the event, however, of 
the whole horizon not requiring to be illuminated, the 
dioptric light would be more expensive than the reflected 
light; but the greater power and more equal distribution 
of light may be considered of such great importance as to 
outweigh the difference of expense. On the other hand, 
the catoptric system insures a more perfect exhibition of 
the light, not only from the fountain lamps being less 
liable to derangement than the mechanical lamps used in 
dioptric lights, but because the extinction of one lamp in 
a catoptric light leads to much less serious consequences 
than the extinction of the single lamp in a dioptric light. 
Experience, however, goes far to show that in practice the 
risk of extinction of the lamp in dioptric lights is very 
small, and I think that there can be little doubt that the 
more fully the system of Fresnel is understood, the more 
certainly will it be preferred to the catoptric system for 
the illumination of shore lighthouses. It must, however, 
be evident that the great oscillation to which lightships as 
at present constructed are subject, renders all arrange- 
ments of glasses for the dispersion of light in a zone 
inapplicable to them. Our lightships are therefore still 
furnished with twelve-inch reflectors (mounted on gim- 
bols), which, by their great divergence of the light, obviate 
the objections to which the dioptric system would be sub- 
ject; although that divergence involves a corresponding 
loss of intensity. 
The conditions required for a floating light to be effi- 
