128 CALORIFIC POWER OF FUELS. 



used, no conclusion other than that stated above can be 

 drawn from these analyses, and no possible idea can be 

 deduced as to the actual composition of the gases as a whole. 

 When we try to use laboratory methods of control in practi- 

 cal workings, the first necessity is to obtain correct samples 

 for analysis, that is, average samples. In this respect all the 

 above -quoted authors are deficient. The tests made by 

 Scheurer-Kestner, published in 1868, were the first to con- 

 form to this requirement. His samples were drawn by a 

 system analogous in principle to that described for sampling 

 coal. 



It is not always necessary to resort to such a complicated 

 operation in case of a permanent gas; samples taken from 

 the general current by means of an ordinary aspirator or an 

 oil-aspirator (page 132) will usually do if drawn at a sufficient 

 distance from the fire. If the gases have passed through a 

 long flue, especially one with several bends, they are suffi- 

 ciently mixed, and may be considered as a homogeneous gas. 

 We must remember, however, that as we recede from the 

 fire the infiltration of air, if not prevented, becomes greater. 

 In careful experiments, the method to be described of frac- 

 tionating a large volume is preferable. 



GAS SAMPLER. 



In principle the apparatus consists of a falling-water 

 aspirator, and a second mercury aspirator drawing a small 

 fraction of the gases from the current of the first in a con- 

 stant regular manner and keeping it in a mercury gas-holder, 

 A (Fig. 28), which is a strong glass flask of 3 litres capacity, 

 holding about 40 kilograms (88 Ibs.) of mercury. The 

 gas-holder is connected by the tube a with the tube c for 

 sampling the gas, the flask A and its accessories acting as 

 a Mariotte flask. It is closed at the top by a stopper 

 hollowed out conically below and having holes for two 

 tubes, a and b. This hollowing is to permit filling without 



