1886.] The Harmonic Analyser at the Meteorological Office. 385 



Fig. 2. Fig. 3. 



<cated to the second and third pairs of cranks throngh sets of toothed 

 wheels, so arranged that they may, if desired, be changed for others 

 of different ratios to the cylinder carrying the cnrves, and in this way 

 the terms of other orders of the expansion may be obtained, shonld 

 they be required, with the same instrnment, merely going over the 

 curves afresh, and using wheels of the proper ratios in place of those 

 used for the first, second, and third pairs of terms. 



" The cylinder over which the curves have to pass is provided with 

 an ingenious arrangement by which it can be expanded or contracted 

 to alter its circumferential measurement by about four-tenths of an 

 inch, so that within certain limits variations in the length of the time 

 ordinates of the curves can be very readily allowed for. To effect 

 this the cylinder (see figs. 4, 5, and 6) is made in three sections, each 

 provided with an eccentric movement ; of course, except when these 

 are at their normal positions the " cylinder " is not truly cylindrical, 

 but still, even when moved to their extreme limits, the deviation is 

 not so large as to cause any inconvenience in its use. Adjustments 

 are also supplied to the pointers, of which two are used at once, the 

 one to follow the outline of the curve dealt with, and the other the 

 ■zero line from which the ordinates are measured ; and throughout the 

 machine all racks and toothed wheels are skew cut to further lessen 

 the risk of error from backlash. 



" The height of the machine is 3 feet 8 inches, and the length of the 

 steel bar, which is rather longer than the bed of the machine, 9 feet. 



" The machine has now for some time been in regular use at the 

 Meteorological Office, and notwithstanding the weight and solidity 

 of some of its parts, the whole is so nicely balanced and fitted that it 

 works with the utmost case and smoothness." 



The machine was delivered at the Office in December, 1879, and a 

 lengthened series of trials was at once commenced, to determine its 

 constants, and thoroughly test the accuracy of its working, for which 

 purpose systems of straight lines and curves, of which the values were 

 known, were first nsed. A few small unforeseen difficulties were early 

 met with, necessitating slight modifications in some portions of the 

 instrument. 



The chief of these faults was a slight turning of the cylinders upon 

 their axes, when the balls were moved to and fro along the disks, 



2 d 2 



