6o 



THE DIFFUSION OF GASES THROUGH 



obtained in the earlier experiments. One should have anticipated a slower 

 rate of diffusion if any change of rate were to be found. The reverse is 

 the case. 



The only explanation of the large rate encountered may possibly be found 

 in the fact that the partition on which the diver descended or rested during 

 quiescence was a sheet of clean copper. It is not improbable that the oxi- 

 dation of this sheet in the lapse of time increases the effective gradient as 

 the metal becomes a sink for oxygen. In other words, the normal diffusion 

 gradient is enhanced by the chemical effect introduced. If this proves to 

 be the case, the experiment presents a rather sensitive test for such action, 

 as the diffusion coeflicient has been increased nearly three times. 



Fig. 17. — Chart showing loss of standard volumes of gas in diver 

 in lapse of days. Diffusion of air through water. 



The diffusion coefhcients by volume k computed from table 1 6 for tlie two 

 rates specified are 

 10 = 0.0445 c.c./day, or kX 10'" =2.86 I'o = 0.0539 c.c./day kX 10'" = 3.59 



showing the mean value io^\- = 3.22 ato^C. and 76 cm. of mercury. In 

 view of these discrepant results it is necessary to consider the case of air 

 under varying conditions, as will now be done for the present apparatus. 

 In §41, moreover, swimmers of dift'ercnt dimensions will he substituted. 



I may note in conclusion that a gradually rising temperature would 

 decrease the apparent diffusion owing to the gradual rejection of gas from 

 the water within the diver, whereas a gradually falling temperature would 

 have a reverse effect. It would be very difficult to discriminate, in such a 

 case, between the true dift'usion and the solution discrepancy. 



The double-tube apparatus A was eventually removed to a vault of more 

 constant temperature and taken out for observation only, leaving the 

 copper partition in place. These new results are included in table 16, after 



