ON THE ANEMOMETERS OP PLYMOUTH. 255 



late and change the position of the plate is another source of error, and there 

 is also from this cause a considerable oscillation in the traces of direction. 

 In violent storms, the vane often whirls ronnd altogether and throws the 

 pinion out of the range of the rack-work of the bar ej\ and unless consider- 

 able attention be given in replacing it the direction is in error. From these 

 and other contingencies, the registers which liave come into my hands are far 

 from perfect; nevertheless I have thought it desirable to go patiently over 

 them with a view of obtaining such general deductions as the number of the 

 clearly recorded observations are competent to furnish. 



As this machine does not register integral results, but merely the pressure 

 of the wind on a given area, it is requisite to find the mean velocities due to 

 the several mean pressures, and multiply these into the time during which 

 each wind blew. 



The recorded observations being entered under prepared forms containing 

 the directions of the wind to sixteen points of the compass for every twenty- 

 four hours, together with the corresponding pressures at each hour; a mean 

 force was obtained from several heights of the pencil on each side of the 

 hour-line, not exceeding half an hour on either side, and the mean force 

 thus determined was taken as the mean force of that hour. From these 

 forces the mean pressure and total number of hours which each of the se- 

 veral winds blew were deduced. In order to turn these mean pressures into 

 corresponding velocities, extensive tables of pressure and velocity were cal- 

 culated from the labours of Rous, Smeaton, Hutton and others, as well as 

 by reference to experiment, and from these the velocity of the wind in feet 

 per second, or miles per hour, to any given pressure, could be readily found. 



The registers however of Osier's anemometer which have come into my 

 hands, inclusive of some lately received, are all very defective in continuous 

 and perfect observation, and are beside occasionally interrupted by the 

 damage done to the machinery of the pressure-plate during very violent 

 gales. It became requisite therefore to resort to some approximate method 

 of discussing such observations as the registers contain, so as to make them 

 available for scientific deduction. The general method pursued in this case 

 is as follows. 



First. The forces clearly marked for the different directions of the wind 

 were taken out and tabulated for a given period, together with the number 

 of corresponding observations. From the numbers thus deduced, a mean 

 pressure was deduced for each direction in dividing the sums of the recorded 

 pressures by the number of observations. 



Secondly. The mean pressures thus obtained were taken as an approxi- 

 mative value of the mean force of each wind, supposing the record had been 

 complete in all its detail. 



Thirdly. The total number of hours of each wind, in respect of direction 

 only, were taken out and tabulated ; the direction being pretty generally re- 

 corded when the pressure-plate was not acted on. 



Lastly. The velocities in miles per hour due to the mean pressures were 

 multiplied into the total number of hours of each wind, and the product taken 

 as the integral effect, or the distance in miles a particle of air would have 

 passed over in either of the given directions to sixteen points of the compass. 



From these integrals the mean direction and velocity of the wind were 

 deduced as before by geometrical construction. 



In Table XIV, will be found the results of the registers of Osier's anemo- 

 meter at Devonport for the years 184'1 and 1842, as thus deduced, and from 

 the last line of which we obtain the following effective forces in velocity and 

 direction for these years. 



