LIEUT. DRUMMOND ON THE ILLUMINATION OF LIGHT-HOUSES. 385 
should not be less than-^ = 21: if it be less than this, there must be dark 
spaces diverging from the light-house as a centre, in which nothing but the 
unassisted light of a single Argand will be visible. 
In revolving lights there are five, seven, and even ten reflectors on a side, 
the number of sides being usually three or four. In the light-houses lately 
erected on Beachy Head, and on the Perch Rock at the mouth of the river 
Mersey, there are thirty reflectors in each, disposed on three sides, each bearing 
ten reflectors. These are the latest, and may be considered as the best speci- 
mens of this method of illumination; being about ten times more powerful than 
the ordinary fixed lights. In some few instances oil gas has been introduced, 
but the intensity of the flame being very little, if at all, superior to that of an 
Argand lamp supplied with the best spermaceti oil, little or no advantage can 
be expected from this introduction, as far as regards the brilliancy of the light, 
when reflectors are used. 
About thirty-eight years ago the experiment was tried, in three or four light- 
houses, of substituting glass lenses instead of metallic parabolic reflectors. 
These lenses were 20 inches in diameter, of 19 inches focal length, and about 
5 inches thick ; but from the imperfection of form and the badness of material, 
the light transmitted by them appears by our late experiments to be about 
one-third of that of the reflectors now in use, while their divergence is so 
small, that at 1^° on each side of the axis they cease to be visible. With a 
view probably to remedy these defects, a somewhat extraordinary arrange- 
ment was adopted, viz. — the addition of parabolic reflectors behind the lenses. 
It is true that by this means some addition is made to the direct light of the 
lens, and, what is of more consequence, the divergent light is increased so 
that at an angle of about 3° with the axis, it is equal to about thirteen times 
the light of an Argand. So far therefore the reflector, though but a small 
portion of it comes into use, contributes to the effect of the lens ; but the 
converse experiment does not appear to have been tried, viz. — how far 
the reflector was improved by the lens placed before it ; otherwise it would 
quickly have been perceived that the effect of the reflector alone was about 
double the united effects of the reflector and lens ; while at the same time its 
effective divergence was also greater, being about eight times that of the com- 
bined lens and reflector, at an angle of 3° on either side of the axis. 
3 d 2 
