604 PEOI’ESSOE BUNSEN AND DE. H. E. EOSCOE’S PHOTO-CHEAIICAL EESEAECHES. 
and similarly for IJ I* and 
Hence 
r=(i- 
-■■0 
—hxO. 
( 3 .) 

f=l— 'Y/ (numlog = A«)^ (4.) 
The values for a, and § employed in all our experiments were calculated from the equa- 
tions, by determining the amount of chemical rays which passed through two plates of 
crown glass, one of which was very thin, and the other much thicker. The following 
experiments gave the requisite data. !„ signifies the amount of light, measured by our 
instrument * before transmission through the glass plate ; I signifies the amount of light 
measured after transmission ; and h the thickness in milhmetres of the plates employed. 
Series of Experiments II. 
Exp. 1. 
Exp. 2. 
Exp. 3. I 
lo 
19-85 
18-42 
19-09 
I 
13-96 
13-14 
17-21 1 
h 
17-1 mm. 
17-1 mm. 
0-30 mm. 
From experiments I and 3 the value of a, is found to be 0'00642 ; from experiments 
2 and 3, 0'00605, or a mean of 0'00623. From experiment I we get ^=0’0520, from 
experiment 2 g>=0-0452, from experiment 3 g>=0'0485, or a mean of 0-0486. 
The following conclusions may be drawn from this experiment : — 
1. That 4-86 per cent, of the chemical rays from a flame of coal-gas which fall per- 
pendicularly on a surface of crown glass, are lost by the fii’st reflexion. 
2. That nine-tenths of the chemical rays, issuing from the same soince of light, are 
extinguished by passing through 160-5 millimetres of crown glass. 
The thinnest of the plates employed by us had a thickness of 0-30 milhmeti-e ; 
the unit amount of light in passing through this plate will be reduced from I to 
^Q-o-3.o-oo623_Q.gggy_ passing through this plate, therefore, 0-43 per cent, of the light 
is absorbed. This is so small an amount that it falls within the limits of the unavoid- 
able errors of observation, and we may, without diminishing the accuracy of the result, 
write h=() instead of A=0-3. Hence formula (4.) becomes 
oi- e=i-\/r 
^0 ■‘o 
The value of ^ calculated according to this formula, from No. 3 of Series I., gives 
^=0-0506, which corresponds very nearly to the mean value 0-0486 preHously foimd. 
This value is probably the more correct, as it is independent of the errors ensuing from 
the introduction of a. 
* Philosophical Transactions, 1857, p. 359, Plate XVII. 
