12 REPORT — 1877. 



As the projected researches required that the weight of potassium salt 

 used in each experiment should be known with the greatest possible accurac3 r , 

 it was considered desirable to avoid direct weighing of the solid salt, by 

 employing a definite amount of solution of known strength. 



For this purpose the capacity of a pipette, which nominally held 10 cub. 

 ccntims., was accurately ascertained. The pipette was filled to the mark 

 with distilled water at a temperature of 20° to 21° C. The contents were 

 then allowed to flow into a small accurately tared beaker. The pipette was 

 then allowed to drain for exactly thirty seconds, when the last drop of fluid 

 was expelled by gentle blowing, the nose of the pipette being held in contact 

 with the beaker so as to avoid any chance of loss. This plan was found to 

 result in the delivery of a more constant weight of fluid than spontaneous 

 draining, with or without subsequent contact of the point of the pipette 

 with the main volume of the liquid. The same pipette was always employed, 

 and the contents were delivered in the same manner. All the measurements 

 were made at pretty nearly the same temperature. As a result it was found 

 that in a series of nearly twenty experiments the extreme variation in the 

 weight of distilled water delivered was 8 milligrammes, or about -08 per cent, 

 of the weight, while the great majority of the determinations were within 

 2 milligrammes of the mean. The result of using the pipette for measuring 

 out 10 cub. ccntims. of a 10 per cent, solution of chloride of potassium would 

 be that the maximum deviation from the mean would amount to -04 per cent, 

 of the Weight, though the maximum difference in two successive measure- 

 ments might equal twice this proportion. 



The experiments showed that at 20° C. the pipettes delivered a mean 

 weight of 9-9329 grammes of distilled water. 



The most convenient quantity of chloride of potassium for precipitation 

 with platinic chloride is about *7 gramme, or 10 grains. A solution of pure 

 chloride of potassium was therefore prepared of such strength that the 

 pipette should deliver about that amount. 



The exact amount of chloride of potassium contained in one pipette delivery 

 of the solution was next ascertained. Two determinations were made by 

 precipitating a pipette full with nitrate of silver, and one by direct evapora- 

 tion of the liquid to dryness with subsequent cautious heating of the residue. 



A 1 AgCl 1-3395 = KC1 -G968 gramme. 



A 2 AgCl 1-3400=KC1 -6971 



B 1 .... By evaporation = KC1 *6.970 „ 



The mean of these closely concordant results is -69097 gramme ; -697 

 gramme was therefore considered as the true amount of chloride of potassium 

 in the solution delivered by the pipette. 



At a somewhat advanced period of the investigations some irregularities 

 in the results led to a doubt as to the degree of accuracy attainable by pipette 

 measurements, and it was decided to commence an entirely new series of 

 experiments on a different basis. Recognizing the advantage the em- 

 ployment of solutions has over direct weighing of the solid salt, it was 

 decided to weigh each quantity of solution employed, merely trusting to 

 measurement to obtain approximately the same quantity. By proceeding in 

 this manner all errors due to unequal deliveries of the pipette or accidental 

 alterations of temperature were entirely eliminated. 



For these experiments a fresh solution of chloride of potassium was pre- 

 pared, by dissolving a known weight of the pure potassium chloride in exactly 



