4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 8/ 



of solar variation, after thus evaluating and removing these four 

 periodicities of 8, ii, 25, and 45 months, is compared with the original 

 data shown in curve A. One other periodicity above 8 months ap- 

 pears conspicuously in curve C, namely, one of 68 months, as indi- 

 cated by the smoothed curve. No other periodicity of importance 

 remaining, the computation was concluded at that point. It had dis- 

 closed that the solar variation since 1918, as depicted by the march 

 of monthly mean values, is adequately represented as the sum of five 

 regular periodicities of 8, 11, 25, 45, and 68 months, whose sum is 

 given in curve B of Figure i. Curve B is drawn below curve A to 

 avoid confusion, but will be seen to be a very close copy of curve A, 

 except in the early years when the observations were least satisfactory. 



I would like to emphasize that none of these five periodicities is 

 of sine form, though the 68-month residual is not far from it. I would 

 like also to remark that these five necessary and sufficient constituents 

 of the solar variation since 1918 are not related to each other in length 

 in the ratios i : | : ^ : :^ : |^ as would have been the periodicities used in 

 Fourier analyses. It seems to me that the method which I have used 

 leads more directly to true and significant relations than the arbitrary 

 fitting of a curve by the classical methods based on Fourier analysis. 



It occurred to me that the various steps used in computing might 

 be done by a machine. Having suggested its design, I was so fortunate 

 as to receive a grant of $1,000 from the Research Corporation of 

 New York to aid in the construction. The work was done mainly by 

 Mr. A. Kramer, instrument maker of the Smithsonian Astrophysical 

 Observatory. Finding some difficulty, however, with the two large 

 grooved barrels, each equipped with 152 sliders and a clamping de- 

 vice, these parts were very accurately made to my order by the 

 Gaertner Scientific Corporation of Chicago. 



Plate I gives a photograph of the completed instrument. A steel 

 scale, a, with double graduation into millimeters and half millimeters, 

 respectively, enables the observer to set up the data on the right-hand 

 drum. This he does by rotating the knurled wheel, b, which, through 

 gearing, engages a rack at o, on which is carried a vertical displaceable 

 pawl. This pawl is adapted to engage successively the sliders, d, d, 

 152 in number, and push them along their grooves to proper settings, 

 as measured by the scale, a. Check screw-clamps are provided to stop 

 the rack at zero of the scale, a, on each return motion, whether to 

 left as just indicated or to right as mentioned below. Thus a long 

 curve, determined by the original data, is set up on the right-hand 

 drum, and its sliders are clamped and fixed immovably by the screw 

 and band, e, e. A small vice-clamp, /, operated by a knurled head now 



