EXPERIMENTAL METHODS. 15 



METHODS IN GAS- INTERCHANGE DETERMINATIONS. 



The preliminary experiments were made by the writer in the spring of 1912, 

 at which time, also, the scope of this phase of the investigation was planned. 

 During the summer at Tucson charge of these experiments was given to 

 Miss M. E. Latham, who later analyzed the samples and compared the results 

 with analyses carried out by the writer. The series of experiments undertaken 

 in the summer of 1913 were performed by the writer during a brief stay at 

 Tucson and were analyzed by the collaborators during the following winter. 

 With the new apparatus employed during the second summer it was possible to 

 carry through a large number of experiments in a short time and it was found 

 possible to keep the gas samples for an indefinite length of time in tightly stop- 

 pered, mercury -sealed vials. In practice the tubes were about half or two-thirds 

 filled with the gas over a mercury bath. Stoppers were then jammed in very 

 tightly and the tubes kept in an upright position. The only problem was the 

 safe transportation and this was solved by personally carrying the samples so 

 as to insure the integrity of the mercury seal. Two lots were thus brought 

 across the continent, and the writer is absolutely sure that nothing happened in 

 either journey to disturb the seals and thus invalidate the samples. Analyses 

 made both before and after the journey showed no difference; indeed, analyses 

 made in 1914 from samples collected in 1912 gave the same results as those 

 made in 1912. This method of procedure proved itself highly advantageous 

 in many ways. In the first place, it allowed of the collection of a large number 

 of samples during rather brief visits to Tucson, and secondly, the analyses of 

 the samples could be made at leisure under more favorable conditions than 

 possible at the high temperatures found in Arizona during the summer. The 

 making of micro-analyses is exacting work and requires a care in manipulation 

 that is not likely to be attained if the work is done under pressure of time. 

 There is, indeed, the admitted disadvantage that some of the results may be 

 found unsatisfactory after it is too late to repeat the experiments, but con- 

 sidering all circumstances this was by far the best course to adopt. 



The manner first employed for obtaining the required gas samples was 

 somewhat primitive, although efficient. The material under investigation 

 was placed in wide-mouthed flasks (Grignard flasks) of from 600 to 800 c.c. 

 capacity, and the flasks were then rigidly fixed over dishes of mercury to seal 

 them. After a desired lapse of time samples of the atmosphere within the 

 flasks were obtained by means of a bent pipette filled with mercury and oper- 

 ated with a rubber bulb. In order to insure a mixing of the gases before 

 sampling, the mercury in the pipette was several times violently ejected into 

 the flask while it was still sealed. It is of course unnecessary to say that the 

 pipette was so constructed that it was not possible to eject all the mercury 

 on one compression of the bulb and thus contaminate the sample with the air 

 which the latter contained. The gas thus drawn off was transferred to the 

 tube in which it was to be stored without removing the pipette from the 

 mercury bath. The vials into which the samples were run had been filled with 

 mercury with such great care that no bubbles of air adhered to the sides of 

 the vial. No matter how clean the glass may be, and even if the mercury 

 itself is ordinarily pure, bubbles of some size frequently stick closely to the 

 walls. In the filling of these tubes, then, each was minutely inspected, and 

 the greatest care was taken to avoid this possible cause of serious contamina- 

 tion of the gas sample. When the vial full of mercury was in a satisfactory 



