RICHARDS. — THE ELIMINATION OF THERMOMETRY LAG. 9 



theless the considerable deviation between the extreme values is greater 

 than we can easily ascribe to uncertainty in the reading of a thermometer, 

 or inaccuracy of weighing. Hence it appeared that the cooling correc- 

 tion, which here amounted to about 5 per cent of the rise of temperature, 

 was probably at fault. 



A series of determinations was next made with the same apparatus, 

 increasing the temperature of the surrounding jacket at the same rate as 

 that of the calorimeter. To accomplish this, a suitable weight of sodic 

 hydroxide was dissolved in the jacket, and just at the instant when the 

 mercury was delivered from its reservoir an equivalent mass of sulphuric 

 acid was poured into the caustic solution from four different outlets, so 

 placed as to make the initial distribution of the acid as uniform as jjossi- 

 ble. A powerful stirrer completed the mixing within a short time, as 

 was proved by the indications of a thermometer hung in the reacting 

 mixture. The total rise of the bath under these circumstances was al- 

 most exactly the same as that observed in the calorimeter, and ran nearly 

 parallel to it ; hence radiation from the sides and bottom of the calorimeter 

 must have been minimized if the influence of the adjacent body of air is 

 negligible. The partial exposure of the top of the liquid still left a slight 

 cooling effect, which had to be corrected arithmetically ; but this could 

 easily have been avoided in more elaborate experiments. 



Three determinations made in this way gave the results 605, 606, 608, 

 average 606 ± 0.06, thus possessing a smaller probable error than any 

 three determinations chosen from among the previous series. Such ir- 

 regularity as exists was probably due to the fact that the bath was not 

 exactly regulated to the proper rate of temperature increase. The re- 

 sult is also distinctly lower than the preceding series, showing that the 

 temperature interval had previously been estimated as too large. The 

 difference of 0.8 per cent in the averages of the two series is a very im- 

 portant one. Possibly the value 606 may have been slightly too low, 

 owing to the large air-space between the calorimeter and jacket, but it 

 could hardly have been as much too low as 0.8 per cent. 



In order to obtain further light upon the matter, another series of de- 

 terminations was made, in which the environment was warmed in such a 

 way as certainly to yield too high a result. The acid and alkali in the 

 jacket of the calorimeter were mixed exactly two minutes before the hot 

 mercury was run into the calorimeter itself, and the final temperature of 

 the environment was made high enough to counterbalance the slight cool- 

 ing effect on top. Therefore during the first two minutes there was un- 

 doubtedly an excessive warming effect. The results thus found were 



