42 
Proceedings of the Royal Society 
restored by potash until the end point is closely defined. Each 
burette reading is taken down on a card, and the result copied 
into the Laboratory Book. The cards are preserved for future 
reference if necessary. 
The standard acid and alkali were at first made up to be of the 
strength of normal, i.e ., x 3 6 ’5 grms. of hydrochloric acid, and 
•gL x 56 grms. of potassium hydrate per litre, so that 1 c.c. might 
correspond to 1 milligram of carbonic acid. 
It was, however, found to be more economical of time to make 
up an acid roughly approximating to the desired strength, and to 
determine its chlorine very carefully by titration with silver 
nitrate. The potash was made equivalent to the acid. 
Two solutions of hydrochloric acid have been used for the work 
described here. The first was found by the mean of five concordant 
experiments to have 1 c.c. equivalent to 1*007 c.c. of the potash 
used. More than 5 c.c. of potash is never used, except in very 
exceptional cases, and as the burette is not read closer than 0*05 c.c. 
the difference between 5*00 and 5*03 is considered negligible. The 
mean of the experiments with the second acid solution gave 1 c.c. 
of acid as equivalent to 0*990 c.c. of alkali, and this also is taken 
as equality. The error of considering the two acids as equal is only 
introduced in the titration of the excess, the quantities of acid 
used being reduced separately according to the amount of chlorine 
found in them by analysis. 
Trustworthiness of the Method . — A large number of experiments 
were made on the same samples of water, both for the sake of 
practice and in order to arrive at the probable uncertainty of each 
experiment. 
The following tables (III. and IV.) give particulars of these 
determinations, and of the corresponding deviations from the 
mean. 
If the five cases out of the thirty-five in which the deviation 
exceeds 0*5 be omitted, as probably due to mistakes or misreadings, 
the mean deviation is reduced to 0*24. 
It is probable that 0*2 may be nearer the truth, as it was never 
the practice to filter the water before determining the alkalinity, 
and after the bottle had stood in the laboratory for a day or 
two, the 250 c.c. decanted off was tolerably free from suspended 
