92 DISPLACEMENT INTERFEROMETRY APPLIED TO 



be 1,220 per degree per day. The mean g/m rise due to diffusion is but 1,800 

 per day, but the temperature effect is reversed in heating, so that a steady 

 mean rate appears in the lapse of time. A net solutional g/m discrepancy of 

 830, per degree, per day, i. e., about 0.3 per cent, must, however, be kept in view. 

 To carry out the comparison of the present and the earlier results, just re- 

 ferred to, the computation of dm/dt may be made. If the observed datum 

 g/m = x, it follows that dm/dt=(g/y?)(dx/di), where dx/dt is obtainable 

 graphically. Thus, for vessel B the datum is 



dm /dt = 42 5 (98 1 7(67000) 2 J =93X10-* gram /day 



In case of vessel F, the results during the first 20 days are on the average 

 dm /dt = (g& 1 7(24 1 500) 2 Ji,85o = 3iXio- gram/day.* 



The swimmer in case of this vessel was of the same form as that formerly 

 used, when after the lapse of ovei two months the mean rate dm/dt = 34.9 X io~* 

 was computed. In view of the temperature irregularity and relative shortness, 

 of the new graph, the two rates may be considered of the same order. At best 

 the porous cup can not have produced a reduction of rate of more than 4/35. 

 which is quite inadequate. Again, the ratio of coefficients in B and F is 3.0. 

 The ratio of areas, about (3-8) 2 /(3.o) 2 = 1.6, is thus out of proportion and some 

 other explanation of the excessive rate for B will have to be found. 



On March 2 the cylindrical swimmer in the B apparatus was replaced by 

 a balloon-shaped vessel having the following dimensions : 



^=17.869 grams; diameter of bulb, 4.5 cm.; /t = 4-8 cm. 



The stem was about 4 cm. long and about 2 mm. in internal diameter. It 

 was placed in the same vessel B, with the porous cup removed, and the obser- 

 vations begun on March 3 are constructed from the tables in the graph B', 

 figure 112, the ordinates being g/m in ten thousands, as above. 



The mean rate of increase of g/m during a period of about 20 days is now 

 smaller, but by no means in proportion with the reduced area and by no 

 means negligible. Taken fiom the curve graphically, g/m increases about 225 

 per day or 0.00217 f the initial value per day. Referred to the loss of air, 

 this becomes dm /dt = gSi X225/(io6,ooo) 2 = 2oX io~* giam per day. Thus 

 this loss as compared with the preceding case (without tube) is reduced 93 X 

 io~*/2oXio~ 6 = 4.7 times, while the reduction of areas is (n.5) 2 (o.o3) 2 , roughly 

 several hundred times. Hence, even if we leave the length of the neck quite 

 out of consideration, the phenomenon before us can not in its larger aspects be 

 one of diffusion. The gas leaves the diver at a rate which is only secondarily 

 dependent on the length and section of the efflux pipe below and virtually 

 moves more rapidly as this neck is narrower. Furthermore, although the 

 irregularities or temperature effects are now less marked, they can not be 



*The rate eventually diminishes and during the month its average value would be about 

 25X io~* gram per day. 



