CRITICAL DISCUSSION OF RESPIRATION APPARATUS. 249 



and at l h 10 m p. m., 4.03 per cent. There was some diffusion, but the 

 rate was so slow that it played no role in experimental periods of 5 

 minutes' duration. Samples were also taken from the larger bag and 

 gave the following results: 12 h 10 m p. m., 3.53 per cent of carbon 

 dioxide; l h 21 m p. m., 3.46 per cent; 2 h 30 ra p. m., 3.36 per cent; 3 h 30 m 

 p. m., 3. 37 per cent. With the larger bag, the experimental periods 

 were about 10 minutes in length and the sampling took place immedi- 

 ately after the period was over, so that this rate of diffusion, did not play 

 a significant role. 



Another possible source of error in the bag method is the difficulty 

 of measuring the air in the bag accurately. It is practically impossible 

 to empty the bag completely, and even when pressed flat and rolled, air 

 still remains and additional air will be sucked back when the bag is 

 again flattened. Douglas recommends using exactly the same pro- 

 cedure before and after the experiment, so as to have the amount 

 of air driven from the bag during measurement the same as that which 

 has actually been added to it. The accuracy of measurement is of 

 special importance, as it is not possible to make long experiments with 

 the bag method. 



The agreement of duplicates in measuring volumes was tested with the 

 larger bag by introducing a known weight of oxygen into the bag and 

 then passing the gas through a 10-liter Bohr meter, noting the tempera- 

 ture, the barometric pressure, and the amount of gas registered by the 

 meter. In one case 15.7 grams of oxygen were used and the meter 

 reading showed that 99.5 per cent of the oxygen had passed through it; 

 in a second case, 63.9 grams were used and the meter reading gave 

 100.2 per cent. In this instance, therefore, the duplicates were within 

 1 per cent; it should be noted that this included not only variations in 

 the bag itself, but also in the weighing of the cylinder and in the reading 

 of the meter. 



As the air in the bag is thoroughly mixed by the kneading process, 

 it is evident that a sample of air taken from the bag represents the 

 average composition very exactly. In this regard the method is su- 

 perior to all other open-circuit methods because of the possibility of 

 thorough mixing. 



One of the advantages of the Douglas apparatus is the fact that it is 

 portable. Furthermore, by using several bags it is possible to carry 

 out several experimental periods in quick succession. On the other 

 hand, there is no control upon the regularity of respiration with this 

 method, as only the total amount of expired air is known, but not the 

 amount for individual portions of time. The periods must also be 

 extremely short and should not be continued so long as to cause the 

 subject to exhale against a noticeable pressure, for it is doubtful if 

 normal respiratory exchange can be obtained under such circumstances. 



The valves used by Douglas are the mica-flap valves of the Siebe- 

 Gorman Company. We have found that these are sometimes unreli- 



