NO. 4 PERIODIC SOLAR VARIATION — ABBOT 



Table 2. — Continued 



Turning to the 68|-month period, instead of 51 repetitions of it, 

 as in the tabulation of the 7-month period, there are only 5. Yet, as 

 monthly means are used instead of lo-day means, the individual 

 values are probably more precise in proportion to V3. Combining 

 these two considerations, the probable error of the mean values in the 



68^ 



j-month tabulation is A/ — — =1.8 times as great as of those in 

 K 5x3 

 curve F of figure i. Hence we are not to expect as smooth a curve 



for 68:1: months as curve F of the 7-month period. But on the other 

 hand, as we shall see, the range of the 68^-month solar period is 

 nearly three times as great as that of the 7-month period. There is 

 therefore no difficulty in recognizing the 68:5- period as well evidenced. 

 In figure 2, AA is the direct mean of five repetitions of the monthly 

 values composing a period of 68:J: months. The curve AA is very 

 rough, but shows some indication of a superrider of 68^-4-3, or 22| 

 months. Extracting this short period, curve BE results. This curve 

 discloses a period of -^ X 68^ or 34^ months. Extracting it, curve CC 

 results. This shows very plainly a period of 1/7x68:!: months. Ex- 

 tracting it, curve DD results. Here, somewhat less distinctly, a period 

 of i/ii of 68:3: months is seen. Extracting it, curve FF results. 

 Here, by careful analysis, a period of 1/19x68:^ months is found. 

 Removing it, there remains curve FF , which, though it may contain 

 shorter subperiods, is smooth enough to plainly exhibit the trend of the 

 68:|:-month period, originally so much obscured in curve A A. The 

 smooth curve GG has a range of 0.18 percent of the solar constant. 



