.'598 Sir James Dewar [Jan. 17, 



tlirongli the release valve was between 55 and 65 grams per hour, 

 while'the delivery was maintained at 425 grams per hour (nearly six 

 litres a minute, sufficient for three persons). Thus the escape through 

 the blow-off amounted to about 15 per cent, of the total delivery 

 for respiration, or 75 to 80 per cent, of the normal loss of the 

 vessel when full of liquid oxygen with the delivery valve closed, 

 and the release valve set to blow off at a plus pressure of half an 

 atmosphere. It is therefore an advantage to lead the oxygen escap- 

 ing from the release valve into the breathing tube circuit. 



ELECTRICAL EVAPORATIOX. 



A commercial vacuum flask adapted for electrical evaporation is 

 shown in Fig. 4(/>). Tlie heating coil A was wound on a cross mica 

 frame. Two asbestos-covered copper leads were sufficiently rigid to 

 keep the coil in position near the bottom of the flask. The two 

 leads passed tightly through holes in a short length of fibre rod B 

 screwed into the metal cap of the flask. A little melted bitumen 

 mixture secured an air-tight insulated joint. A short spiral of 

 copper tube was fixed into the opposite side C of the metal cap 

 to the fibre rod. The blow-off valve was at tlie end D of this 

 delivery tube when necessary. The vessel was filled through a wide 

 fiat shouldered screw stopper S. The dimensions of the beating coil 

 depended on the voltage of the electrical supply available for use 

 with the flask. 



An exceptionally regular flow of oxygen gas is always obtained 

 from this arrangement, and the rate is only slightly affected by 

 external conditions of temperature, etc. It has also the advantage 

 that apart from the small heating coil no fittings or alterations are 

 necessary to the actual vacuum container, as the simple electrical 

 equipment is quite separate, and there is little in the container itself 

 to get out of order. 



METAL CONTACT EVAPOEATIOX. 



The fittings necessary to adapt a similar flask for heating by 

 metallic conduction from the outside aie shown in Fig. 4:{c). A bundle 

 of seven equally spaced aluminium rods A, I in. diameter and 6 in. 

 long, were screwed into a copper plug 8. This was provided above 

 with a slightly coned hollow socket to fit the end of a T-shaped 

 aluminium rod B screwed down through the cap. The co})per 

 socket plug was secured air-tight to the screw cap of the vacuum 

 flask by a german silver tube, easily clearing the inserted aluminium 

 rod B. A copper spiral delivery tube was fixed horizontally into an 

 elbow in the top of the screw cap while the vertical arm of the elbow 

 held the small screw plug through which the flask could be replenished. 



In this form of the apparatus the rate of evaporation depends 



