816 
DR. T. R. ROBINSON ON THE DETERMINATION OF 
(45.) I was surprised at finding the cylinders so much more consistent than the 
hemispheres. I attribute it to two causes : 1, the cylinders are 18 oz. heavier than 
the 9-inch hemispheres, and their greater momentum tends to neutralise any fluc- 
tuations of friction or of the vortex’s current ; 2, any radial currents will probably 
have more power to increase the resistance by acting on the convex hemispheres than 
they can have on the planes which terminate the cylinders. It must be observed that 
in this form of anemometer the lateral friction is greater than with the hemispheres. 
D = 7 5 gave A" ' = 22 ‘5, while with No. 1 it was 27 '6.' 
I shall now briefly recapitulate the conclusions which, in my opinion, may be drawn 
from these experiments. 
(46.) The equation (III.) represents the observations from V== 5 to \ T/ =40 and 
through a wide range of friction. The discordances follow no law which might 
indicate a relation between V' and v different from that expressed by (III.); they are 
evidently such as would be produced by casual disturbances of W and F. 
(47.) If the equation (I.) contains any other functions of V' and v than squares and 
products, their effect is so small that it is masked by these disturbances. 
(48.) The coefficient of the impelling forces a appears to be as the area of the cups, 
and to be the same for radii of 24 and 12 inches. 
(49.) There is reason to believe that /3 and y are proportional to a, or that x and 2 
are constant in all anemometers whose arms are of sufficient length to prevent the 
wake of one cup from interfering with the cup which follows it. For Nos. I. and III. 
this is highly probable ; it is not disproved by No. IV., which is not worse represented 
by it than by any other system of constants. No. II. requires a larger x and z than 
the others, but I think I have assigned a sufficient cause for this. If this opinion be 
well founded, the difference between the indications of different anemometers would 
F ... V' 
depend only on the fraction and the limiting value of — given by the constants of 
paragraph 39 would =2’286 instead of 3, which I originally assigned to it. More 
exact observations may show that x and z are not rigorously constant, and diminished 
section of the arms may alter this limit ; but probably the changes will not be of 
much practical importance. 
(50.) If these observations were repeated in a quiescent locality, if* the frictions 
were measured at each experiment and by the means described in the note to 
paragraph 20, more accurate results would certainly be obtained ; but there would 
remain the inevitable difficulty arising from the impossibility of rightly estimating the 
effect of W. Any measures of it which can be taken give only an average of the 
whole circumference, and one outside the anemometer’s track, while the experiments 
with the balloons and air-meter show that it varies notably in various parts of the 
circumference, and with the distance from the centre. There is also an outward 
current of considerable magnitude, but very irregular. The rotation of the anemo- 
meter itself, whether vertical or horizontal, produces a secondary vortex, which must 
