676 PROCEEDINGS OP THE AMERICAN ACADEMY. 



amounted to 110 X 0.2 — 22, or 92 raayers ; about one-sixteenth of the 

 total heat capacity of the mixture. In most cases this was negligible. 



This simple apparatus is so efficient iu its operation as to be worthy of 

 use also in other thermochemical operations where great accuracy and 

 small range of temperature is needed. The method in which the mixing 

 is accomplished is certainly jireferable to any method which demands the 

 pouring of one of the liquids from one vessel into another. 



Results. 



A {"ew actual results with the apparatus may now be discussed in order 

 to show the accuracy which has been obtained. It is possible to calculate 

 the specific heat of a solution by mixing water in this apparatus in two 

 ways. The value for the solution over the range 0° to 17° may be based 

 eitlier upon the heat communicated to it by the warm water as calculated 

 from the average specific heat of the warm water, and its fall of tempera- 

 ture in mixing, or else upon the value for water between 0° and 17° as 

 found by mixing this cold water with exactly the same amount of hot 

 water in parallel experiments. The two methods of calculating give 

 almost the same value, the difference being less than 0.05 per cent, as was 

 implied iu previous statements concerning the quantitative mixing of hot 

 and cold water made on page 674. 



The cause of such slight difference as exists need not be at present dis- 

 cussed — it is enough to emphasize again the fact that reference of the 

 cold solution to cold water measured in parallel experiments over the 

 same range of temperature is by far the best, since thereby any possible 

 constant errors of experimentation are eliminated, being canceled by 

 appearing in each series. The results below with cupric sulphate and 

 argentic nitrate are computed in this way. The heats of dilution were 

 subtracted before the calculations were made. The calculations for the 

 concentration CUSO4 -f 200 HjO were made on the assumption tliat very 

 slight dilution evolves only a negligible quantity of heat, which was 

 shown to be true. 



The " probable error " of this result, computed by the method of least 

 squares, is only about 0.01 per cent. Constant errors must have been 

 eliminated by the method of determination which compared the solution 

 with water under exactly the same conditions, in parallel experiments. 

 Hence the result is probably not far from the truth, although the series 

 is claimed to be only preliminary, made to demonstrate the practicability 

 of the method. Because there are no other determinations of such a 

 solution between 0° and 17°, comparison with others is impossible. By 



