90 DISPLACEMENT INTERFEROMETRY APPLIED TO 



. , _ RMT _ 



g ~MHp m (i/ P ,-i/p s } 



The corrections are then found by logarithmic differentiation (considering 

 TT a correction of H), as follows : 



(3) dg' = gd*/H; dg"=-gdp m /p m - y dg'" = 



Unless the temperature is very variable, this would usually suffice; but it is 

 probably just as easy to construct tables for dir, pjp*, and Pw/P s and use them 

 directly. If the temperature rises, dg' and dg'" are positive, dg" negative, and 

 a temperature, say 18 C., may be taken from which to reckon all corrections. 

 In the work below TT is always deducted from H. 



77. Observations. The work was begun with cylindrical swimmers, wide- 

 mouthed below, and not constricted as in figure no, the object being to find 

 the maximum variation in a room of variable temperature, in case of the 

 presence of a porous porcelain cup c, below. Two vessels were selected with 

 appurtenances of the dimensions following: 



B. F. 



M =37.425 grams 10.939 grams 



Pe= 2.47 2-47 



2r= 3.8 cm. 3.0 cm. 



length= 5.8 cm. 6.7 cm. 



&= 5.0 cm. 6.0 cm. 



The vessel B was about 6 cm. in diameter and 26 cm. high, with a clear 

 space of 12 cm. between the levels c and e- F was about 5 cm. in diameter, 

 with the other dimensions similar. Observations were made at midday, 

 daily, with the exception of the gap after February 18 due to illness. The 

 data were carefully tabulated in detail,* together with the values of g/nt com- 

 puted therefrom by equation (i) . All apparatus being near together, the tem- 

 perature t is about the same in each. The barometer minus the manometer 

 reading is the quantity H in the equation. 



Both series of data for g/m are constructed in the B and F graph, figure 112, 

 with the ordinates in ten thousands, in addition to the temperatures t, in the 

 lapse of time. 



The results in each case are disappointingly irregular, owing to the inevitable 

 marked variations of the temperature of the room. The work with B (heavy 

 swimmer), extended between February 14 and March 2, shows an average rate 

 of w-loss of 0.75 per cent per day. In view of the irregularity of the curve this 

 may be regarded as identical with the corresponding rate of the earlier experi- 

 ments in the absence of a porous cup, as will presently appear. It follows, 

 therefore, that the porous cup has had no effect whatever; or that air does not 

 here diffuse appreciably through the porcelain walls as was anticipated. In 

 fact, throughout a large part of the experiment the mouth of the swimmer 



* It seemed advisable to remove this bulky table, as much of the essential information 

 is given by the graphs. 



