THE CONSTANTS OF THE CUP ANEMOMETER. 
817 
modify the primary one ; to what degree we cannot say. The effect of these variations 
of W is shown by the changes of v in individual revolutions, while Y is perfectly 
uniform. 
(51.) For this reason, combined with the difficulty of obtaining satisfactory measures 
of the centrifugal friction, I have given up the idea of repeating these observations, 
and I propose instead to try the method described in the end of my paper. 
I have established on the flat roof of the dwelling house here the anemometer 
No. I., with its axis vertical and 22 feet due east of that of the Kew instrument 
already there. Both of them are about 16 feet above the roof. The chronograph 
and its battery of six Leclanche’s cells are placed in the wooden house which 
supports it, and are connected with it and the other instrument by insulated wire 
and contact makers. From the altered position of the axis it is necessary to apply 
the brake weights by means of a right-angled lever ; the counterpoises already 
described have been replaced by springs to prevent the chance of accidental disturb- 
ance of their position. The axle now rests on a toe of hard steel, 0'25 inch in diameter, 
turning in oil ; and to prevent it from cutting under the pressure of the axle and its 
appendages, the rollers formerly used to lighten the centrifugal friction have been 
altered by Mr. H. Grubb so as to relieve the toe of any required portion of the 
weight. This arrangement works so well that the friction is little more than in the 
Kew one, though its weight is l - 6 times as great. The arms are now of plate steel -|th 
of an inch thick, with a sharp edge, strengthened by wire stays, and with means of 
placing both the 9-incli and 4-incli cups at 24, 12, and 8 inches from the centre. 
(52.) The mode of experimenting which I intend to follow is this. Considering 
the Kew as the standard S, and the other E, and arranging one of the chronograph 
points to register the v of S, the other of E, apply a brake friction to the latter which 
will diminish v. When this has lasted long enough to make it tolerably certain that 
each instrument has been acted on by the same amount of wind, apply a different 
brake friction to E and take another set. We have hence three equations, but four 
unknown quantities, Y, a, (3, and y ; and since a appears as a coefficient of V 2 , it 
cannot be obtained separately. But if, as I believe, it is given with tolerable pre- 
cision by the measures given in Table YI., the others may be determined. 
The formula (VI.) enables us to estimate for the Kew anemometer the error in Y 
caused by Aa=l. When v=l it is — 0’20, 0'059 of the computed V; for ^=20 it 
is — 3'30, 0'072 of Y. The highest v I ever observed was in a squall in March, 1849, 
when during five minutes it was 41 - 2 ; this would give AY=— 8 14, 0'072 of 112 - 80. 
It is possible that more exact measures of a might be obtained by moving the 
measuring apparatus in a right line, and in an enclosed space of sufficient dimensions, 
but the requisite conditions for such an experiment will not be easily realised. 
(53.) If these constants give for any other pair taken with a different wind V s = V E 
they are right ; if not, the formula (YI.) will show us whether the error arises from 
using a wrong a, or whether the constants vary with v. 
5 M 2 
