NO. 7 ASTROPHYSICAL OBSERVATORY, I9O4-I953 — ABBOT II 



of it, 10 or even 50 years in advance, or backwards. Recently, how- 

 ever, some meteorologists seem to be more favorably impressed. 



Some confusion has been caused by over-riding periods. 



It must be clear to all that if there actually is such a large harmonic 



family in weather periods, ranging in lengths from 4 months to the 



master period of 273 months, then a tabulation at individual stations 



273 

 of weather periods, e.g., of — j-=68:|: months must be overlaid by 



273 273 

 shorter periods, such as ~"o~'"^o~' ^t cetera. So the primary tabulation 



for a 68:|^-month period must necessarily look very rough. It may not 

 even suggest a "hidden" period of 6S^ months unless those overlying 

 shorter harmonics can first be removed. 



I now refer the reader to the accompanying figures 2, 3, and 4, pre- 

 pared from Jon Wexler's electronic tabulations of monthly "World 

 Weather Records," 1870 to 1940. He obtained monthly normals of 

 precipitation at Rome, Italy, Kief, U.S.S.R., Capetown, South Africa, 

 and other stations, separately for years when Wolf sunspot numbers 

 ^ 20. From these he obtained the departures in percentage of normal 

 for all months, 1870-1949. These are divided into four groups which 

 we call Category 1, Divisions 1 and 2; Category 2, Divisions 1 and 2. 

 "Division" refers to first and second half of the interval 1870-1949, 

 "Category" refers to Wolf numbers ^ 20 sunspot numbers. 



Figure 2, for the 68^-month period in Category 2, combines Divi- 

 sions 1 and 2 at Rome, Italy. It shows five starting dates : June 1902, 

 February 1908, October 1913, June 1919, and March 1925. A lack 

 of recorded observations in later years before 1949 prevented using 

 a sixth column. The mean of these 5 columns is plotted in the bottom 

 full curve. No one seeing this curve could suppose it would easily 

 be resolved, and would disclose a smooth sine curve of 68^ months. 



^ ,^ .J. . J 273 273 273 273 273 273 



But when the over-ndmg periods —^, -ry, -^, -j^-' -^f -jj; are 



successively computed and removed, as shown in figure 2, there re- 

 sults the beautiful smooth sine curve shown at the top of figure 3, 



273 

 representing only -r-or 68-^ months. Its amplitude is over 30 percent 



of normal precipitation. 



I am sure it will not be necessary to so particularly describe figures 3 



and 4 relating to Kiev, U.S.S.R. and Cape Town, South Africa, which 



each include one group of Jon Wexler's electronic tabulations, being 



273 273 

 for the periods -^— and -5— months, respectively. They display over- 



