410 
FA1RLKY : BLOWING WELLS, 
I have also on various occasions analyzed the air proceeding 
from the well both with a portable eudiometer on the spot, and 
also of a sample taken at the beginning of these experiments by 
Mr. Hutton. In each case the analysis did not differ appreciably 
from that of ordinary air, and gave 20-9 per cent of oxygen. 
There was no excess of carbonic acid. I was informed, however, 
that when the shaft of the well was first sunk there was a great 
and sudden outrush of choke damp, so that the well sinkers 
barely escaped with their lives. These facts, and the information 
I had as to the great force and quantity of the current following on 
the changes of the barometer, led me to suggest the existence of 
a large cavity in the strata communicating with the shaft of the 
well. In February, 1879, when the well was opened in my 
presence we found the fissures in the sandstone near the water 
line at a depth of 15 yards from the surface. The well then con- 
tained about 6 feet of watei . The out-current shown by a lighted 
candle held to the fissure was very powerful. 
Observations were made dependent on the application of 
Boyles' law to measure the out or in current, so as to get an 
approximation to the size of the cavity, first with a vane anemo- 
metor, afterwards with large gas meters. Only these readings 
are of use for this purpose where a current due to a change of 
the barometer has had full time to expend itself. 
In many cases the current is checked or modified by subse- 
quent barometric changes. 
Considering the difficulty of closing a well shaft covered with 
stone flags so as to be air tight, and that readings could be taken 
only at intervals, generally in the morning and afternoon, the 
measurements can only be regarded as approximate. Of course 
the assumption was made that no other openings or fissures led 
into this cavity. 
Readings were also made of the temperature of the out-cur- 
rents, and of the temperature of the atmosphere. I fear, however, 
that the temperature of the out-currents was more dependent on 
