330 Prof. Spencer U. Pickering on Delicate 



error, therefore, in experiments where one thermometer is 

 nsed will have to be increased by half this amount, thus be- 

 coming + 18 # 86 cal. This number, then, represents the 

 probable divergence from the truth of any one experiment ; 

 and, from the manner in which the various thermometers 

 were standardized (by comparison with each other) , this error 

 will probably be as great when two experiments performed at 

 different temperatures are compared, whether they were per- 

 formed with the same or with different instruments. 



Perhaps, for the purpose of determining this probable error, 

 the calculations ought to be based on the difference between 

 duplicate experiments performed not on the same day, in 

 which case the error would be somewhat larger. I have not 

 done so, however, chiefly owing to uncertainty in the number 

 given for this quantity (see above). Many causes also which 

 tend to increase the difference when the experiments are not 

 performed consecutively might not apply in general. Amongst 

 these I may mention the very hygroscopic nature of many of 

 the salts taken, where a difference in the state of moisture of 

 the atmosphere would affect the extent to which traces of 

 water were absorbed during their exposure while being trans- 

 ferred to the weighing-tubes : these substances, again, were 

 often kept in sealed tubes containing sufficient for two or 

 three determinations only, and the contents of the various 

 tubes may not have been dried to precisely the same degree. 



XLV. On Delicate Calorimetric Thermometers. By Spencer 

 Umfeeville Pickering, M.A., Professor of Chemistry at 

 Bedford College*. 



IN the foregoing communication it has been shown that the 

 average error of a determination of the heat of dissolu- 

 tion of a salt in 400 H 2 amounts to + 18*86 cal.; but this is 

 only an average, and it must not be overlooked that the error 

 in some cases may exceed this very considerably. It is not 

 uncommon to find a difference of 40 or 50 cal. between two 

 duplicate experiments (although the average error in such cases 

 is only +13*3 cal.), and a proportionately larger error must 

 therefore be expected occasionally in the comparison of two 

 experiments at different temperatures. 



The determinations from which these conclusions were 

 drawn consisted of a series of experiments with various salts 

 at temperatures ranging from 1° to 26° C, where it was im- 

 perative that the greatest possible accuracy should be attained, 

 and also that the experiments at the different temperatures 

 should all be strictly comparable with each other. The ther- 



* Communicated by the Physical Society : read February 13, 1886. 



