Error in Calorimetric Work, 327 



It was found that with the apparatus employed, and with a 

 difference between the temperature of the calorimetric liquid 

 and that of the surrounding air not exceeding 2°*5 C, the 

 rate of cooling was practically constant throughout the 20-30 

 minutes (and even for a considerably longer period) during 

 which the experiment lasted ; whereas for accurate measure- 

 ments it was never safe to trust to the results of one experi- 

 ment to determine the rate of cooling in another experiment, 

 even though the two were performed within 10 minutes of 

 each other, and apparently under exactly similar conditions. 

 To determine the rate of cooling, therefore, all the operations 

 (pounding, stirring, &c), performed during the first 5 

 minutes to effect the dissolution of the salt, were imitated as 

 nearly as possible during a second, third, and sometimes even 

 fourth and fifth similar intervals, the temperature being read 

 at the end of each interval. The actual rate of cooling (or 

 heating) rarely exceeded o, 04 C. for 5 minutes, and was 

 generally much less. 



In these experiments the proportion of salt to water taken 

 was 1 : 400 molecules*, the actual volume of the latter being 

 about 600 cub. centim. The chief salts operated on were the 

 sulphates of sodium, potassium, lithium, copper, magnesium, 

 potassium-magnesium, and potassium-copper, both hydrated 

 and anhydrous. 



The thermometers were all read by estimation by means of 

 a lens to the twentieth part of the marked divisions, i. e. 

 to 0*05 millim.f ; and one such estimation-figure, being equi- 

 valent to *002° C, makes a difference of about 16 cal. in 

 the molecular heat of dissolution. I should not put the esti- 

 mation-figure at less than one tenth of the marked divisions 

 were it not that these experiments afford ample proof that 

 some reliance can be placed on smaller amounts than this. 



The heat-disturbance in the calorimeter varied between nil 

 and 2°*3 C, the average being rather less than 1° C. 



Out of some 600 determinations of this nature, I may 



* With double salts 1 : 800 ; but, for the sake of comparison, these are 

 here reduced so as to be comparable with the other experiments. 



t The divisions in some of the instruments measured rather more than 

 1 millim., namely 1*2 milium ; but, provided sufficient attention be paid 

 to a certain proportion between the breadth of the mercurial column and 

 the distance between the divisions, there is no appreciable difference be- 

 tween estimating the fractional portions of these two quantities. Besides 

 practice, and uniformity in the lines themselves, nothing conduces so 

 much to accuracy in estimation as a just proportion between the breadth 

 of the mercurial column and the distance between the divisions. I find 

 the best proportion to be 1 : 10 j 1 : 15 is too great, and 1 : 5 too small. 

 Different ratios, however, might suit different observers. 



