PART II. 



EXPERIMENTS WITH THE USE OF SPECIAL APPARATUS. 



Iii the ordinary use of the anemometer, (let us suppose it to be a Robinson's 

 anemometer, for illustration,) the registry is seldom taken as often as once a minute ; 

 thus, in the ordinary practice of the United States Weather Bureau, the registration 

 is made at the completion of the passage of each mile of wind. If there be very 

 rapid fluctuations of the wind, it is obviously desirable, in order to detect them, to 

 observe the instrument at very brief intervals, e. g., at least every second, instead 

 of every minute or every hour, and it is equally obvious that in order to take up 

 and indicate the changes which occur in these brief intervals, the instrument 

 should have as little inertia as possible, its momentum tending to falsify the facts, 

 by rendering the record more uniform than would otherwise be the case. 



In 1887 I made use of the only apparatus at command, an ordinary small 

 Robinson's anemometer, having cups 3 inches (7.5 cm.) in diameter, the centre of 

 the cups being 6| inches (16| cm.) from the centre of rotation. This was placed 

 at the top of a mast 53 feet (16.2 metres) in height, which was planted in the 

 grounds of the Allegheny Observatory, on the flat summit of a hill which rises 

 nearly 400 feet (122. metres) above the valley of the Ohio River. It was, accord- 

 ingly, in a situation exceptionally free from those irregularities of the wind 

 which are introduced by the presence of trees and of houses, or of inequalities of 

 surface. 



Every twenty-fifth re volution of the cups, was registered by closing an electric 

 circuit, and the registry was made on the chronograph of the Observatory by a 

 suitable electric connection, and these chronograph sheets were measured, and the 

 results tabulated. A portion of the record obtained on Jnly 16, 1887, is given on 

 Plate I., the abscissa? representing time, and the ordinates wind velocities. The 

 observed points represent the wind's velocities as computed from the intervals 

 between each successive electrical contact,' as measured on the chronograph sheets, 

 and for convenience in following the succession of observed points they are here 

 joined by straight lines, though it is hardly necessary to remark that the change in 

 velocity is in fact, though quite sharp, yet not in general discontinuous, and tin- 

 straight lines here used for convenience do not imply that the rate of change of 

 velocity is uniform. 



7 



