APPARATUS AND METHODS USED IN THIS RESEARCH. 41 



rapidly and accurately moistening the air to a suitable degree. For 

 this purpose a Williams bottle (K), containing pure distilled water, is 

 placed in the circuit in such a manner that, by closing a valve in the 

 by-pass, the entire air-current may be forced through the water in this 

 bottle or as little thereof as may be desired. 



Oxygen. — The direct determination of oxygen may be made either 

 by weighing a small cylinder of gas (L) and noting the loss in weight 

 during the experiment or by using an exceedingly delicate and accurate 

 gas meter. Small cylinders of compressed oxygen which can readily 

 be weighed may be secured from the Linde Air Products Company of 

 Buffalo, New York. These cylinders weigh when filled about 3 kilo- 

 grams and contain about 150 grams of oxygen with a purity of about 

 97 per cent. The oxygen supplied by this company is made from liquid 

 air and consequently the residual gas, instead of being nitrogen, as has 

 commonly been supposed, is as a matter of fact in large part argon, 1 

 so that to the volume of oxygen measured, about 1 per cent should be 

 added for the argon. 2 



One of the greatest difficulties in using these cylinders has been the 

 selection of a suitable valve, that furnished on the cylinder by the 

 manufacturer being difficult to utilize owing to the high pressure under 

 which these cylinders are filled. Formerly recourse was had to one of 

 the numerous types of reduction valves, but a thorough test of these 

 did not result in securing such a valve as would functionate properly 

 for any long period of time. One or two types of needle valves have 

 been found which are much less expensive and give a satisfactory 

 closure. Such a needle valve is coupled to the exit of the cylinder, 

 then closed, and the main valve on the cylinder is opened to its fullest 

 extent, The issuing gas may then be very delicately regulated by 

 means of the needle valve. With so high a pressure it is obvious that 

 the packing around the main valve stem should be excellent, so as to 

 give no opportunity for leakage of air. The valves may be tested by 

 immersing the cylinder and valve under water or by weighing the 

 cylinder carefully on a balance and then again an hour later, when any 

 loss of oxygen will be instantly apparent, 



Gas meter.— From many standpoints the use of a small weighable 

 cylinder of oxygen is to be recommended. On the other hand there 

 are certain advantages in favor of using an accurately calibrated gas- 

 meter under such conditions as to preclude excessive temperature fluctua- 

 tions. In our experiments with infants we have almost always employed 

 a large cylinder of oxygen with a valve, conducting the gas through 

 a carefully calibrated meter of the type devised by Bohr and manu- 

 factured by the Dansk Maalerfabrik of Copenhagen. This gas meter 

 registers one liter for each complete revolution of the drum. Being 



Claude, Comptes rendus, 1909, 151, p. 752; Morey, Journ. Am. Chem. Soc, 1912, 34, p. 491. 

 2 For a discussion of this point, see Carnegie Inst. Wash. Pub. No. 187, 1913, p. 74. 



