144 Shiro Tashiro 



In summarizing, I may emphasize the following points: 



1. Particular care must be taken to test the air- tightness of the 

 apparatus. 



2. Purifying the air must be done with greatest care, as this is 

 essential. 



3. The apparatus must be perfectly dry. 



4. A weak suction pump cannot be compensated by frequency of 

 washing. 



5. As long as the ratio between the c.c. taken from the chamber 

 and the original volume of the chamber is needed, it is most 

 important to have the pressure in A and B equal to the atmos- 

 pheric. If this is accomplished we can neglect any caution against 

 pressure and temperature variations a correction which is always 

 necessary for ordinary methods of analysis of exceedingly minute 

 quantities of any gas. 



In devising this method and in constructing this apparati, I am 

 under great obligation to Professors McCoy and A. P. Mathews and 

 to Dr. F. C. Koch. 



In order to test the accuracy with which an estimate of concen- 

 tration of carbon dioxide could be made, many determinations were 

 carried out with samples of air which contained accurately known 

 concentrations of carbon dioxide prepared by Dr. F. C. Koch. The 

 experimenter did not learn the concentrations of the samples until 

 after the analysis had been completed. In making up the test sam- 

 ples, pure carbon dioxide, made by heating sodium bicarbonate was 

 diluted with the carbon dioxide free air several times in succession, 

 as illustrated by the following example: 5.5 c.c. of pure carbon diox- 

 ide was diluted to 52.0 c.c. over mercury and thoroughly mixed; 

 5.5 c.c. of the first mixture was diluted to 52.0 c.c.; i.i c.c. of 

 the second was diluted to 50.7 c.c.; of this third mixture 5.6 c.c. 

 was received from Dr. Koch. I diluted this a fourth time to 

 255.6 c.c. to form a mixture to be analyzed. The following observa- 

 tions were made: 0.5 c.c. was introduced into the apparatus and pro- 

 duced no precipitate in ten minutes; 0.5 c.c. more of the same sample, 

 gave no precipitation in another interval of ten minutes; 0.5 c.c. more, 

 a total of 1.5 c.c., was run into the bulb. In six minutes the first 

 evidence of a precipitate appeared on the surface of the drop at 

 d of apparatus 2 and in eight minutes was well developed. Since 



