ACOUSTICS AND GRAVITATION. 93 



disregarded. Thus, in the curve B', kinks of the value of io 3 /io 5 , or nearly 

 i per cent, due to the solutional effect, are sometimes present, which are tem- 

 porary errors, even if evanescent. 



The apparatus was now left without interference for over 10 days (excepting 

 the changes of room temperature) and then examined on April 6. Curiously 

 enough, the diver B' had lost no air at all, though subsequently losses occur 

 from changes of temperature on succeeding days (see curve B')- Thus it 

 appears clearly that the exhaustion made during observation is one of the 

 chief causes of the loss of air. This unexpected and unfortunate result, so far 

 as the method is concerned, seems to be fatal to any expectation as to its 

 availability for the ultimate purpose in question; but it throws a more definite 

 light on the mechanism of the process than was heretofore obtained. An ex- 

 haustion of but 20 cm. of mercury or less increases the effective gradients 

 enormously, so that the air is removed by suction, as it were. Without exhaus- 

 tion or corresponding interferences (temperature) it is improbable that the 

 rates would conform with the ratio of areas. The same result is obtained 

 in the next experiment. 



The diver F, after the close of the above experiments, was also provided 

 with a neck, by closing the wide mouth of the cylinder with a tube 5 cm. long 

 and about 0.5 cm. in diameter. The ratio of mouth-areas available for dif- 

 fusion is thus 3*/(o.5) J = 36. This was increased by drawing out the end of the 

 tube. The mass of the new balloon-shaped diver F' was 16.552 grams. The 

 results are given in the graph, figure 112, marked F'. The initial rate of in- 

 crease of glm, after a week's observation, may be estimated from the graph as 

 300 per day; or relatively to the initial charge 300/112000 = 0.0027 per day, 

 thus of the same order as in the preceding case. The loss of air per day was, 

 therefore, dm /cfr = 981X3007(1 13,000)2 = 23X10-*, also of the same order as 

 before, though the tube was wider. 



Left to itself for 10 day? (after March 26) it gives evidence of some air-loss 

 on April 6 ; but the diminished rate during the interval of quiescence is 

 none the less striking, as the curve F' shows. Here also, therefore, the ex- 

 haustion during observation is the chief cause of the observed diminished 

 air charge. 



78. Sheathed or inclosed divers. A number of incidental experiments were 

 made, partly in the endeavor to counteract the convection, partly to test 

 smaller and more mobile diver?. It was also thought possible that by removing 

 the free surface of liquid at c, figure no, the loss of air at this surface would be 

 reduced. The apparatus, AA, thus took the form given in figure 113, where 

 the diver, ab, with open mouth below, is surrounded by the cylindrical vessel 

 EE, closed at the top. Water completely fills A A, except at e and a. The 

 divers were small test-tubes, the dimensions in the twc cases being: 



No. A'. . A/4.426 grains; diameter 1.5 cm.; length 8 cm.; h =6.0 cm. 

 No. C . . M 4 grams; diameter 1.5 cm.; length 8 cm.; h *=8.o cm. 



