ACOUSTICS AND GRAVITATION. 95 



The rate of C remains nearly constant, but that of A' increases again to 

 about r = 0.003. 



It is difficult to account for this curious behavior, which is alike in the two 

 similar apparatus. Of these, C had been set up about 10 days before A'. 

 The inference is plausible that some change in the gac (air) inside the diver 

 must have been the cause, but the gas inside, in the two cases, behaves alike. 

 Hence there may have been some change in the air outside. I have therefore 

 inserted the corresponding graph p, fig. 112, showing the barometric pressure 

 in centimeters for the time. It will be noticed that the minima of the A' and 

 C curves coincide with the remarkably low barometer of the same day. The 

 coincidence would be convincing if it recurred at the next exceptionally low 

 barometer about April 28, but this is not the case. Correspondences with 

 the temperature curve are here less evident, though many similarities are 

 apparent. As to availabilities for the g work, the large rates and irregular 

 curves are of course out of the question. Moreover, the rate of change of g/m 

 depends essentially on the area of the mouth of the diver and not on the 

 mass of air contained, so that the meaning of the rates in the present cases is 

 rather involved. The more appropriate datum would be d(g/m) /dt. 



The divers A 1 and C were now kept without interference other than temper- 

 ature for about a month and then reexamined on June 9. Quite contrary to 

 expectations, the loss of air continued without interruption, at about the same 

 mean rate as before, if the loss of air only and not the change of its mass is 

 taken in question. (See fig. 112.) The importance attached in 7 7 to the exhaus- 

 tions in promoting loss is here contraverted by this new evidence, and the 

 above conclusions must therefore be modified, at least for a sheathed diver 

 like the present. 



79. The swimmer under pressure. Inasmuch as the preceding method (in 

 which the air contents of the swimmer were always observed by the aid of 

 exhaustion) have proved to be unavailable, the question arises whether any 

 better results may be obtained if the swimmer is kept permanently under a 

 very slight pressure excess and observed by increased pressures. From the 

 experience gained, the results should be more favorable, since the sponging 

 action of the water is in a measure removed, particularly if very small pressures 

 are used. 



The modified apparatus is shown in figure 114, the swimmer being now 

 lighter than water. A A is the stand-glass, wholly rilled with water, and ab 

 the (floating) swimmer held down at a definite level by the thermometer T. 

 M is a water manometer for registering the pressure excess at which flotation 

 just occurs at the given level. Pressure is applied by aid of the small water 

 reservoir R, communicating with AA by the tube t with stop-cock F and the 

 flexible pipe p, the whole system tFpR being full of water. To make an 

 observation, R is raised sufficiently and clamped. The cock F is now slightly 

 opened, until the swimmer just begins to sink. The head of water, h, is then 

 read off at M. This adjustment may be made to the fraction of a millimeter of 



