in 



mitted without allowing contact with the general atmo- 

 sphere. This means is found, conveniently, in a thin 

 rubber gas bag attached to one of the gas outlet tubes of 

 the pot, as shown in figure 5. It is convenient to use 

 for these gas bags the rubber bladders made for basket 

 balls or, if desired, a still larger bladder of the 

 same form which can be obtained from the rubber trade. 

 The walls of the gas bag must be of light weight and very 

 elastic. If such a bag is filled only about half full, 

 it will expand as the temperature rises and contract^ 

 as it falls, keeping the pressure in the pot nearly con- 

 stant. As long as there is any gas in the gas bag there 

 will be a slight positive pressure in the pot, due to 

 the elastic tension of the rubber. This pressure must 

 not be high enough to blow out the seal, hence the re- 

 quirement that the walls of the gas bag be thin. This 

 slight internal pressure provides that the minor leakage 

 which it is nearly impossible to prevent, will be out- 

 ward and without effect on the experiment. The plump- 

 ness of the gas bag is a convenient index that the seal 

 is holding satisfactorily, any leakage being indicated 

 immediately by the flattening of the bag. 



In beginning an experiment a stream of the 

 desired gas is run through the pot until the normal soil 

 atmosphere is believed to be displaced. This may be 

 done conveniently by attaching successively to one out- 

 let of the pot several gas bags filled with the desired 

 gas and allowing the gas to escape through the pot, 



