438 PROFESSOR W. WATSON ON A DETER.MINATION OF THE VALP'E OF 
also tested by using the magnetometer which carries the coils to measure H at Kew 
while the coils were in place. The difference between the value obtained and that 
given by the Kew Observatory magnetograph curves was the same as the difference 
when the coils were removed. 
When measuring the dimensions of the coils two micrometer microscojDes were 
employed. These microscopes are attached to two carriages Avhich move along a 
V-groove in a heavy iron bed-plate and can be clamped at any distance apart. By 
means of two brackets attached to the same bed-plate, a standard metre could be 
supported in front of the microscopes. Readings on the standard metre were taken 
before and after each set of measurements made on the coils, and so the screws 
of the microscopes were only depended upon to measure a fraction of a millimetre at 
each end. The runs of the micrometer screws were compared with the graduations 
of the metre. The metre actually employed was a brass one, with the scale divided on 
silver. It is of the standard H section and was made bv the Geneva Instrument 
»/ 
Company. The errors of this metre for the divisions actually emjffoyed were 
determined by comparison with a nickel-steel standard metre by the same makers, 
which has been compared with the International Standards at the Bureau at 
Sevres. The temperature coefficient and the correction to each reading having been 
determined at the Bureau, the error of the brass metre could be calculated. The 
temperature coefficient of the brass metre does not come into the measurements, as 
the measurements on the coil and the comparison with the nickel-steel metre were 
made at the same temperature. 
Since when the measurements were actually juade, the nickel-steel metre had not 
been returned by the Bureau, the following measurements are given in terms of the 
brass metre, and a correction will have to be applied at the end for the difference 
between the two metres. This difference was found to be proportional to the length 
measured, so that it is j^i’obably due to the fact that the temperature coefficient of 
the brass metre vdiich was assumed in reducing the results is not quite correct. The 
final result, however, is free from any error on this account. 
In order to determine the distance between the mean planes of the coils, the 
distance between the outside edges and between the inside edges was measured for 
twelve positions equi-distant round the circumference. To allow of the setting ot 
the cross wires of the microscopes being made vdth precision, the edges of the 
grooves were turned quite sharp. 
Three independent sets of measurements were made in this way, one set before 
the magnetic measurements and the other two after. The results obtained, reduced 
to a temperature of 16° C., are given in the following table :— 
