﻿Calcium Carbonate in Water. 963 



necessary since, on several occasions, single results were not 

 confirmad on repetition ; indeed, the general experience 

 throughout was that, at any time, and for no apparent 

 reason, a false result was liable to be obtained. Further, 

 when results in agreement were obtained, the degree of 

 accuracy was usually not closer than within 2 per cent., 

 whereas the limits of error should not amount in all to 

 1 per cent. 



The causes of this are, I think, as follows. Firstly, there 

 is the extreme difficulty of keeping the silica vessels and 

 filtering apparatus chemically clean. The amount of surface 

 is necessarily very large, a great deal of handling has to be 

 done in the course of an experiment, and the slightest trace 

 of acid or alkali introduced on the tube or the interior of 

 the vessels would be sufficient to spoil a result. The surface 

 of the silica vessels is very rough, and particles of solid are 

 washed out only with great difficulty. Thus, in using a 

 vessel for arragonite which has been employed for calcite, it 

 is not sufficient to wash out, or even to boil out, with water ; 

 the vessel must be boiled out first with dilute acid for a long 

 time, and then with successive quantities of water. 



As an illustration of the reality of danger of error froin 

 this source, it niay be pointed out that the introduction of 

 *001 gr. of alkali (calcium carbonate^ is sufficient to cause 

 an error of from 5 to 10 per cent. 



The silica vessels were therefore most scrupulously cleaned 

 out before each set of experiments, the filtering apparatus 

 before each single experiment. The filter-paper and muslin 

 squares were changed after each cjeterminatipn. 



A second source of error is the presence of a trace of 

 carbon dioxide in the water used. As this reappears, however, 

 in a much greater degree, in the course of subsequent ex- 

 periments, it will be discussed later. 



The results for calcite and arragonite at 25° C., 50° Q., 

 and 100° C. are given below, both in tabular and in graphic 

 form. At the two lower temperatures the thermostat was 

 kept constant within 0°'l C; the figures given for 100° C. 

 are those obtained at the boiling-point, the barometric pres=- 

 sure being observed at the time of the experiment. The 

 true temperature was usually a few tenths of a degree lower 

 than 100° C, but, as will be seen frprn the anagram, the 

 temperature coefficient is so small that the results are sub- 

 stantially correct for that temperature. The diagram also 

 shows the solubility curve for calcite under ordinary atmo- 

 spheric conditions (see page 973), 



3 R 2 



