NO. 7 MONTEZUMA SOLAR-CONSTANT VALUES ABBOT II 



period seemed to be likely. If it seemed so, I arranged the lo-day 

 mean values in groups of tables, each comprising about one-fourth 

 of the total interval 1924-1947. They were never less than six lines 

 long, and with as many columns as there were lo-day intervals in 

 the proposed period. Where periods were not exact multiples of 10 

 days, values were omitted, or columns were omitted, occasionally, to 

 bring the average lengths of the lines to that of the proposed period. 

 The criterion of a true period was always that the several group 

 tables agreed substantially in their means, as to phases and ampli- 

 tudes of the suggested period, throughout the whole 273 months. 

 Such good agreement is shown for the 15^-month period in figure 3, a. 

 In several cases proposed periods failed to meet this test, and were 

 rejected. Sometimes the phases shifted regularly from group to 

 group through the 273-month interval. In such cases the period was 

 shortened or lengthened to give unchanging phases. 



As a result of this branch of the investigation, periodicities of 

 5 2/15, 8.035, 9f, II 3", II 15/16 months were recognized as true, ac- 

 cording to the above criterion. Being incommensurable in length, 

 there was no need to subtract them one by one from the data. They 

 could not materially influence each other. After determining them 

 in the lo-day mean data, they were transformed into monthly means. 

 Then their marches were tabulated throughout the 273 months, their 

 amplitudes added algebraically at each month, and the algebraic total 

 per month was subtracted from the fifth residual list, remaining after 

 removing the longer periodicities named above. This left a sixth 

 list of monthly residuals for further exploration. 



To shorten a tedious story, the methods explained above, when 

 applied to the sixth list of residuals, discovered additional periodici- 

 ties of i4j, 19^, and 24^ months. When all had been removed from 

 the data, no other periodicities seemed worth investigation in the 

 residual plot remaining. It is plotted as curve B of figure i.® The 

 mean of the departures from 1.945 calories in curve B is 0.00189 

 calorie, or 0.097 percent of the solar constant. Many of the larger 

 departures, which materially raise the mean as just given, occur in 



6 One disturbing feature will be noted in figure i, B. Though the year 1947 

 shows no remarkable eccentricity in curve A, it gives a great slump of ^ per- 

 cent in curve B. This is strange, for all the periodicities seem to fit the last 

 year's data, including 1946, as well as the earlier years, as we see from 

 figure I, B. One notes, however, that curve A of figure i is almost entirely 

 below 1.945 calories in 1947. It may be that the Montezuma values of 1947 

 are subject to a yet undiscovered error. Further observations of future years 

 will decide. 



