388 LIEUT. DRUMMOND ON THE ILLUMINATION OF LIGHT-HOUSES. 
the two jets, and at the same time to diffuse the heat more equably, it is mad^ 
to revolve once in a minute, by means of a movement placed underneath thh 
plate m , and with which the wire f, carrying the ball and passing through the 
stem, is connected. Notwithstanding, however, this arrangement, the effect of 
the heat is such as gradually to cut a deep groove in the ball, so that at the 
end of about 45 minutes it becomes necessary to change it*. In a light-house 
where it is of essential consequence to maintain a constant light, it would be 
unsafe to entrust this to an attendant, and hence the necessity of devising 
some means for remedying this inconvenience. The apparatus represented by 
fig. 3. is designed for this purpose, and is drawn in the manner in which it is 
applied to a reflector, the dotted outline of which is shown. 
The wire a b passes through the focus of the reflector, and upon it are placed 
the number of balls at A, required for any given time ; these, by means of 
the shears s, as shown in fig. 4, are admitted between the plates p p, and 
thence permitted to fall in succession to the focus. No. 1. represents the focal 
ball ; about two minutes before the change, the ball 3. falls into the position 2, 
where it becomes gradually heated. At the end of that time, the curved sup- 
port t, moving on a pivot, is thrown into the position represented by the dotted 
line, by the momentary descent of the ringr, which, receiving an impulse from 
the weight iv, acts upon the extremity u of the support. No. 1. falls, but is 
prevented fi om descending more than its own diameter by the loop /, and 
No. 2. following it, occupies the focus. The support t, being immediately re- 
leased, returns by the action of a spring to its former position, retains No. 2, 
and suffers No. 1. to escape through the loop into the cistern. 
The wire a b and the support t revolve together, and carry round the focal 
ball, which is ignited as in fig. 1. by the two jets 2 z. These jets, which are 
moveable round the joints d d, enter through small apertures cut in the sides 
of the reflector, and are easily adjusted to the proper distance from the ball. 
Wherever the light is required to be diffused equally around, the renewal of 
the lime may be effected still more easily, by using a cylinder as represented 
in fig. 5, instead of a ball, which being gradually raised while revolving, brings 
fresh portions in succession opposite the jets. In a reflector, a cylinder occa- 
• When a cylinder is used instead of a ball, a ring of minute crystals is found adhering to the sur- 
face above and below this groove. 
