<U6 



Sir James Dewar 



[Jan. 1" 



dilt'ereiit sized necks. Stoi)pers of varying' degrees of isolation and 

 tightness were tried as to tlieir effe(;t on the rate of loss of the vessels 

 to which they were fitted. The hest arrangement was a plain hollow 

 stopper, filled with the same liquid air as that in the vessel. This 

 stopper was a cylindrical bulb sealed above to a tu])e held in an 

 indiarul)ber bung fitting into the neck of the vacuum vessel. A 

 small outlet hole was made in the bung by which connection was 

 made to the meter for measuring the rate of evaporation of the liquid 

 in the flask. Small silvered cylindrical vacuum vessels could sometimes 

 be so employed (Fig. 1(1), l)ut the stopper effect was not so complete ; 

 and fui'ther they took a long time to reach a steady reduction of rate. 

 A two-litre silvered glass vessel (referred to in curves 1 and 2 of 

 Fig. 17) had a neck aperture of 37 mm. diameter by 70 mm. long. 

 AVlien this was closed by an easy fitting silvered vacuum vessel coutain- 



FiG. IG. 



ing liquid air (Fig. 10) the I'ate of loss was lowered by 7j per cent. 

 AVith only one silver coating (on the inner shell), the saving 

 effected l)y the same stopper amounted to li^ per cent. The corre- 

 sponding reduction in metal vessels was much greater ; thus a :25-litre 

 vessel, with neck uiqm)tected, lost 1*21 kilos, of liquid air per day 

 (4*84 per cent, per day of the total contents). When the neck was 

 kept at liquid air temi)erature by a wide tube full of li(juid air fixed 

 round it, the loss fell to 1M)0 kilo, per day, or a reduction of 

 17 '3 per cent, on the previous rate: the inner metal neck was 

 1 "6 cm. diameter and 30 cm. long, while its walls were 0*2 cm. thick 

 (much more than was necessary), 0*6 mm. being sufficient with 

 snitable tubes. In a vessel of better construction the value of the 

 reduction would rherefoi'e be relatively less than the value ol)served 

 in this case. 



EFFECT OF EXTERNAL TEMPER A.TUHE. 



Appreciable alteration in the evaporation rate is also caused by 

 changes in the room temperature, while tenqieratures such as are 

 experienced in the upper air have a considerable efi'ect. When a l(t- 

 litre copper container, whose neck tube was made of badly ccmducting 



