104 
Each gas had its own tube of P 2 O 6 , so that each tube of 
P 2 O 6 did not become contaminated with more than one gas. 
Experiments were tried with 
(a) Nitrogen in hydrogen. 
(b) Hydrogen in nitrogen. 
(c) Nitrogen in oxygen. 
(d) Oxygen in nitrogen. 
(a) Nitrogen in hydrogen. 
The hydrogen was generated by the action of sulphuric 
acid on zinc, washed and dried by passing through water 
and strong sulphuric acid, and after coming off for some 
time was collected over mercury. 
The eudiometer is then placed in front of the spectroscope 
slit (at a distance of J metre), the spark focussed on to the 
slit by means of a small lens, and the spectrum examined. 
If nitrogen or oxygen are not visible, and only the three 
hydrogen bands are present, then air (or, in some experi- 
ments, pure nitrogen) is added to the hydrogen in small 
quantities (about 1 c.c.) at a time, and the spectrum observed 
between each addition. 
Thus, in one case, to 143 c.c. hydrogen was added to 1 c.c. 
air and the nitrogen lines were not visible, so another 1 c.c. 
air was added ; but now the nitrogen green line wave length 
5004 was seen very faintly, and on the addition of another 
1 C.C. air it could easily be seen. Therefore the amount of 
nitrogen that must be present in hydrogen in order to be 
detected in the spectroscope is IT per cent. 
Again, to 130 c.c. hydrogen 1 c.c. pure nitrogen (made by 
passing air over thin sticks of phosphorus in Hempel’s gas 
analysis bulbs) was added, from a Hempel’s burette, but the 
nitrogen line was not seen; so *5 c.c. more nitrogen was 
added, and now the nitrogen green line was seen, but it was 
rather faint. This gives IT per cent. 
To another 130 c.c. hydrogen, not dried at all, 1 c.c. nitro= 
gen was added, but no nitrogen lines appeared ; *6 c.c. more 
