1910-11.] Specific Gravity and Displacement of Saline Solutions. 639 
agreement between the results obtained in the case of the salts quoted in 
the first twenty-four tables was so good that a complete set of observations 
with one hydrometer seemed to be more valuable than an incomplete set 
with two hydrometers. The mean number of serial means on which the 
arithmetical mean specific gravity of each of these solutions depends is 
2-97, so that the mean probable error of the mean specific gravity of any of 
these solutions may be taken as ±4*5 in the sixth place. 
We have seen that the probable error of the mean of a single series is 
zh7’75 in the sixth decimal place, and the number of independent observa- 
tions included in each series is 9 ; consequently the probable error of any 
one of these observations is 3 x 7‘75 = ±23 - 25 in the sixth place, or ±2*3 
in the fifth place. 
The standard temperatures of the solutions and of the distilled water 
at which our observations have been made are 15°, 19‘5°, and 23° C. The 
majority of the determinations have been made at 19'5° because it has been 
found to be comparatively easy to keep this temperature constant in my 
laboratory at most seasons of the year. In cold winter weather it has 
sometimes been necessary to use 15° C. as the standard temperature, and in 
hot anticyclonic weather in summer it has been necessary at times to adopt 
23° C. as the standard temperature. 
In the experiments which I made before 1902 the variation of the 
temperature of the liquid during the experiment was allowed to reach an 
amplitude of 0‘3° C. without being rejected, but the amplitude did not 
usually exceed 0T° C. A series, consisting of nine separate observations, 
takes from 12 to 15 minutes to complete. 
After 1902, when I began systematic work in a room of the Davy-Faraday 
laboratory, which had a single window to the north, and was otherwise very 
well suited to the maintenance of a constant temperature, the permissible 
amplitude of variation was rapidly reduced so as to be usually insensible. 
Indeed, in the work done during the last year by Mr S. M. Bosworth, he 
has been so successful in the management of the climate of the room that 
cases in which the temperature of the liquid showed any variation have 
been so few that it has been possible to reject them and use only 
observations made at absolutely constant temperature. It is true, however, 
that the climate of that year exhibited no extremes of either heat or cold. 
In the following table the statistics of the variations of the temperature 
of the liquid, while a total of 1316 series of observations was made with 
hydrometers Nos. 17 and 21, namely, 837 with No. 17, and 479 with No. 21, 
are given. In 68 per cent, of the series made with No. 17 there was no 
sensible variation of temperature, and the same was the case in 55’2 per cent. 
