282 Unwersity Geological Survey of Kansas. 
that a very sensible amount of combustion of nitrogen often 
takes place under conditions as to mixture, etc., which would 
have given nearly perfect results with the Bunsen apparatus. 
‘Another very fruitful source of error is the rather rapid 
absorption of carbon dioxid in the water over which the gases 
are measured after the explosion. This increases the apparent 
contraction on explosion and decreases the absorption by potas- 
sium hydroxid, and so produces a fictitious appearance of hy- 
drogen. This is the source, without much doubt, of much of 
the hydrogen which is reported in gas analyses. That hydro- 
gen is never present in natural gas we would not like to assert; 
in fact, we believe that it often is present, but only in very 
small quantities. Consequently, we regard with some suspicion 
an analysis of a gas from an ordinary producing gas-well which 
shows more than one per cent. of hydrogen. To this class be- 
long the much-quoted analyses by Mr. S. A. Ford (see table 10, 
p. 245), which have been often referred to as showing that gas 
from a certain well may vary within wide limits. Apparently 
they showed that the gas analyzed rapidly fell off in methane 
and increased in hydrogen, the latter changing from 10 per 
cent. to 36 per cent. in about six weeks. His results, as we see 
them, would point rather to the gradual accumulation of potas- 
sium hydroxid in his measuring apparatus, which would greatly 
accentuate the errors noted above. This is, then, a case of ap- 
parent rather than real variation. 
The calculation of methane and ethane from the explosion 
data is also open to some question. We have made the calcula- 
tion for the gases which we have analyzed, but we do not re- 
gard the results as possessing any very high accuracy so far 
as they represent the actual quantities of the individual paraf- 
fins. In order to make this calculation, the contraction on 
explosion, the carbon dioxid formed during the explosion, and 
the oxygen used up in burning the gas, were determined. Be- 
tween these factors three equations may be set up from which 
the methane may be calculated and three more for the ethane. 
(See page 258.) Each of these equations involves two of the 
factors. They often yield results which differ materially from 
one another, and this difference is due to apparently unavoid- 
able errors in the data from which they were calculated. In 
every case the nitrogenous residue from the gas was actually 
determined, and this furnished a valuable check on the accu- 
racy of the calculated values. It was usually found that the 
