THE 73° CALORIE. 



383 



of mercury present, any lack of complete symmetry might, by keeping 

 the mercury from taking on the true mean temperature, affect the 

 heat loss. A new coupling was therefore made, in which all of the 

 outer turns were cold. This certainly resulted in a temperature 

 distribution much more dependent upon the quantity of mercury 

 present. No influence on the data was found. A separate test 

 showed the heat loss from the Dewar flask to be so small that the 

 entire contents of the flask might have changed in temperature by five 

 degrees without afl^ecting the final result, the 73° calorie, by one part 

 in a thousand. 



Two series of consecutive runs are given. A third series, tfiken 

 between the other two, is omitted, having been taken with a different 

 arrangement of the calorimeter, under unsteady conditions resulting 

 therefrom. Of the first series only the results are given. The last, 

 figures 5 to 10, is plotted in terms of the bridge readings. In this 

 series it was attempted to obtain readings for two flows, with two 

 temperature dift'erences for each flow, resulting in four groups of points. 

 No two points in any one group were. taken in succession. The im- 

 portance of this is shown especially in the data for figure 7, where the 

 points scatter widely. Reference to the data shows this to be due not 

 to erratic readings but to a gradual drift, leaving the result practically 

 unaffected. A is given in centimeters of bridge wire, this unit being 

 equivalent to about .016 degrees. The flows were 7.5 c.c./sec. and 

 15 c.c./sec, marked s and / respectively. 



Data for figure 7. 



I 



