ON STANDARDS FOR USE IN ELECTRICAL MEASUREMENTS. 133 
the steam was driven off through the mercury. When the water in flask 
A was reduced to about a tablespoonful, the boiling was stopped, and 
_ the level of the mercury was raised until it flowed back first into flask 
B and thence into the barometer tube, as flask A cooled. 
The open end of the barometer tube was then sealed, the flask B 
replaced by a small cup of dry mercury, and the end of the tube opened 
below the surface. The water remaining on the top of 
the column was driven back into the flask by pouring Fic. 11. 
hot water over the tube. 
During our experiments, water occasionally collected 
on the mercury, but by means of a concave mirror it 
was driven back into the flesk; the mirror was of course 
removed some time before an observation was taken. 
The tank, filled with water, was maintained at any 
required temperature by means of a gas regulator. The 
lower parts of the barometer tubes were screened by 
sheets of asbestos, and the two cups were connected by 
a small siphon. The glass sides of the tank were covered 
with white paper to prevent radiation; openings were 
left for observations, during which the water in the 
tank was kept in a continual state of agitation by the 
oscillation of a large paddle driven by a water motor. 
The paddle, fixed in one corner of the lid, swept across 
the tank, driving the water before it, and lifting it at 
the same time. We have tried several forms of stirrers, 
and we believe this to be a more effective form than a 
screw or a plunger. 
The difference in the height of the mercury in the 
two barometer tubes was ascertained by the katheto- 
meter G. 35, in the Cavendish Laboratory, and by 
means of it readings could be taken to ‘50mm. Care 
was taken to bring both levels horizontal before each 
observation. 
As the coefficient of expansion of the kathetometer scale was unknown 
and the temperature of the room usually about 20°C., we decided to 
compare it with the standard scale R, whose coefficient of expansion and 
scale errors had been determined by the Standards Department of the 
Board of Trade.' 
Twenty-one comparisons were made (greatest divergence from the 
mean ‘10 mm.), and the result was as follows :—300°35 mm. on katheto- 
meter scale at 20°=300°35489 of Board of Trade Standard (S.S.) at 0°. 
Thus no scale correction was necessary. 
The difference (D) of the mercury columns was corrected for tempera- 
ture, pressure of mercury vapour and latitude, and the resulting length 
_ denoted by Dy: the temperature corresponding to Dy was deduced from 
the very full table given in Part 3 of Carnelley’s ‘ Melting and Boiling 
Point Tables.’ 
The extremities of the curve (at 0° and 100°) having been determined, 
it was only necessary to get points between 30° and 80°. 
Ninety observations were taken, and although occasional divergences 
presented themselves, the mean path gives a curve which ws believe to be 
— oe, 
* Standard metre, verified June 1882, designated R in Mr. Chaney’s report. 
