MR. .1. H. BRTNKWORTH ON SPECIFIC HEAT OF STEAM. 



391 



3 in 1000 in flow 1, 2, or 3, will produce an error in S of 8, 6, or 1 in 1000 respectively. 

 The flows could generally be measured to O'OOOl gr./sec., so that the error due to any 

 one flow could not in general exceed 1 in 1500 on the value of S, and would be much 

 less for the smallest flow, which was the most difficult to measure. 



Since the values of h and_& depend on relatively small differences, the values of S, 

 calculated by this method from each single set of three flows, are very susceptible to 

 small accidental errors in X, due to slight variations in the conditions, especially with 

 flows taken on different days. Much better agreement may be obtained by taking a 

 large number of observations under identical conditions, and employing constant mean 

 values of h and k in the reduction of the series. This method has been adopted with 

 most satisfactory results by Mr. BRINKWORTH, but could not be applied to the 

 observations given in the following table, because the conditions were varied widely from 

 day to day. The number of observations taken with each calorimetric arrangement was 

 insufficient to enable the variations of // and k to be interpreted with certainty at the 

 time. But Mr. BRINKWORTH has since shown that the values of heat -loss observed in 

 these experiments, when plotted according to his method in fig. 10, show a most 

 remarkable agreement with the results of his analysis. This is a very severe test, 

 because the maximum heat-loss is less than 4 per cent, and each individual result is 

 plotted separately. The agreement is so close that it leaves little doubt of the 

 substantial accuracy of the observations. The only difference is that, owing to the 

 arrangement of the leads, the heat-loss in my apparatus was somewhat greater than 

 in his under similar conditions. 



SUMMARY of Observations with Vacuum Calorimeters. 



VOL. CCXV. A. 



3 G 



