28 ROYAL SOCIETY OF CANADA 
Reichsanstalt, Berlin. With a little care the temperature of the bath 
could be kept at 18° C., as the temperature of the room was generally 
not far from that. A change of ‘04 degree would cause a perceptible 
displacement of the meniscus of the solution relative to the mark on the 
stem of the pyknometer. After the pyknometer had been for some time 
in the bath, the liquid was gradually withdrawn until the meniscus co- 
incided with the mark on the stem. A few minutes later, if the coin- 
cidence still existed, the pyknometer was taken out, the outside washed 
by means of a jet of water from a wash bottle, dried with a soft cloth 
and weighed. 
The balance used was made by A. Collot of Paris. As the weight 
of the pyknometer and solution was approximately 48 grams, a body of 
about 50 grams was weighed several times, the weighings being per- 
formed on several days, but in the same manner as was employed in 
weighing the pyknometer, which will be described further on. The 
maximum deviation from the arithmetic mean was ‘0002 grams. Re- 
garding this as the possible error of a weighing, it can be determined 
to what degree of accuracy the density of the air should be known, in 
order to correct for the buoyancy of the air. In these corrections the 
air in the balance case was regarded as kept quite dry by the calcium 
chloride placed there. 
On making the calculation it is found that the density of the air 
should be known to 5 in the sixth place of decimals. From this can be 
calculated how accurately the barometer and thermometer should be 
read at the time of a weighing. The barometer by means of a vernier 
could be read to ‘002 inch; but as the instrument had been in use 
for some time and its error was not known, I thought it advisable to 
read the thermometer in the balance case as accuratély as possible, and 
deduce the necessary accuracy in the reading of the height of the baro- 
meter. The thermometer in the balance case was compared with the 
thermometer already mentioned, and its errors noted. It could be read 
to + of a degree centigrade. Hence the height of the barometer should 
be known to 2 millimetres, or for convenience in reading ‘05 inches. 
In the correction for buoyancy the density of the air was taken from 
Table 6, Kohlrausch’s Physical Measurements (1883). 
The operation of weighing the pyknometer when filled was con- 
ducted as follows: A counterpoise was placed in the left pan, and in 
the right the pyknometer together with weights necessary to cause the 
pointer to oscillate about a point near the centre of the scale. The 
pyknometer was then removed, and weights added until the pointer 
oscillated about some point near the former. The sensitiveness of the 
balance for the weights employed being known, the apparent weight 
of the solution and pyknometer could be calculated. The point of 
oscillation ot the beam when loaded with counterpoise and weights was 
