CHEMICAL AND PHYSICAL NOTES. 119 
convenient to use the 0°05-erm. weight near the top or bottom of the 
stem. Care is taken that the stem of the hydrometer is wetted for a 
distance of 1 or 2 mm., but not more, above the division where the 
hydrometer is going to float. This is an essential precaution for 
ensuring precision. When the last observation, which must be the 
one nearest the upper extremity of the stem, has been made, the 
small weights are removed, and the hydrometer lifted out and put in 
safety, and the temperature again taken with the thermometer. It 
should not differ from the temperature found at the beginning by 
more than 0°°3 C., and in work making any pretentions to accuracy, 
it should not exceed 0°'1 C. Ifa difference of temperature amount- 
ing to 0°°3 C. is observed, and the temperature itself is above 20° C., 
then the mean temperature must not be taken and used for the nine 
observations, but the 0°°3 C. must be distributed over them, and the 
temperature, which the water had at the time of each observation, 
used. The difference affects the fifth place of decimals. 
“Whether at sea or on land, I always log the time in my 
laboratory work. A series of observations with the hydrometer 
as above described takes on an average about twelve minutes; but 
in that time at least nine quite independent observations have been 
made of the density of the water. If a sufficient number of obser- 
vations have already been made with the hydrometer in distilled 
water of the same or nearly the same temperature, they may be used 
for giving the specific gravity of the water. If they have not, or if 
the determination is of especial importance, then a precisely similar 
series of observations must be made in distilled water of the same 
temperature, and variations of temperature amounting to 0°-3 C. are 
then inadmissible. 
“When the corresponding series of observations has been made in 
distilled. water, and they have had the small stem correction apphed 
so as to give the displacing weight in distilled water at the exact 
stem divisions observed in the sample water, we have nine pairs of 
readings, each pair giving the weights of equal volumes of distilled 
water and of the sample, and therefore each pair giving by their ratio 
an independent determination of the specific gravity of the sample, 
referred to that of distilled water of the same temperature as 
unity. The mean of the nine observations gives a result which, 
according to the doctrine of probabilities, should have a precision 
three times greater than that of a single observation, Although 
much may be done to avoid a large range of temperature of observa- 
tion, there will always be some difference in the temperatures at 
which the specific gravity of the various samples is observed, and as 
