194 



MR J. y. BUCHANAN ON THE 



of a whole degree was 12 millimetres. On such an instrument variations of one- 

 hundredth of a degree are easily appreciated by the practised eye. It must, however, 

 never be forgotten that ivliat is directly observed is, at the best, the most probable 

 value of the temperature of the bulb of the thermometer. The legitimacy of the 

 conclusion that this is the temperature of the medium in ivhich the thermometer is 

 immersed depends on the expertness and experience of the experimenter. The use 

 of an instrument of the degree of delicacy above specified is justified only when all 

 the precautions have been taken which are required in order to justify the experi- 

 menter in concluding that he has the temperature of the system under such control 

 that its uncertainty is not greater than ±0*005° C. Nothing short of first-rate 

 work secures this. 



In the work which Mr S. M. Bos worth has been doing under my direction, upwards 

 of 3000 hydrometer observations have been made during the last twelve months, and 

 the temperature conditions have been controlled with such skill that only three of the 

 series showed a sensible variation of the temperature of the solution from the standard 

 temperature during the time the experiment lasted. Their results were rejected, not 

 because they were not very good, but because in this respect the others were perfect. 



We will now proceed to a comparison of the figures in Tables IIa. and IIb. 



§ 112. It would be useless to take the means of each series of specific gravity 

 results in Table IIa. and compare them with the mean results given in Table IIb. 

 But these numbers show the progressive character of the alteration of the specific 

 gravity in each consecutive series, as well as the considerable diff"erences of these 

 numbers inter se, when contrasted with the agreement which holds among the mean 

 results set out in the other table, IIb. 



The mean results are given in the following table : — 



Table IV. 

 Giving the Mean Specific Gravities calculated from the Series in Table IIa. 



Series. 



Hydrometer. 



Mean Specific Gravity. 



1 

 2 

 3 

 4 

 5 



A 

 B 

 A 

 B 



A 



1-446019 

 1-445917 

 1-445750 

 1-445826 

 1-445330 



Taking the five mean results given above, we find that the maximum amplitude of 

 variation is 689 units in the sixth decimal place. If we turn to Table IIb., we find 

 the maximum amplitude of variation to be only 26 such units. 



Although the consideration of the mean specific gravities shows clearly that there 

 is an unstable condition in the supersaturated solution as contrasted with the stable 

 condition of the 6 "3 CaClg solution, this is made more evident if we compare, as in 



