18 THE UNIVERSITY SCIENCE BULLETIN. 



Lawrence, Kan., where records since 1868 are available. Several hun- 

 dred hours of work showed nothing. Once a stretch of five years was 

 found which resembled another five quite closely after eliminating 

 the seasonal curve. Another time resemblances were found after 

 about twenty -two years. All such were easily explainable as acci- 

 dental. It seemed useless to carry the work further with the data at 

 hand. 



A paper by Professor Turner (3), however, gave me a new sug- 

 gestion, although there was little if any logical reason for any con- 

 nection. In this paper Professor Turner shows plainly the existence 

 of a period in earthquakes with a length between 14.8421 and 14.- 

 8448 months. It occurred to me that this period might be com- 

 mensurable with the sun-spot period. Upon multiplying it by 9, I 

 obtained 11.13 years, which is the mean sun-spot period to the exact 

 hundredth of a year. Such an exact coincidence is very probably 

 not accidental (4a). 



The next move was to examine all sun-spot data in order to find 

 whether such a period also exists in sun spots. The results have 

 been inconclusive, some evidence favoring the existence of the 

 period, but not being definite enough to settle the question either 

 way. The general conclusion seems to be that any relationship of 

 sun spots to weather is not a direct one, and that periodicities which 

 are commensurable may exist in each separately, as might happen if 

 the variations were due to a common cause. This will be more fully 

 developed in the general discussion of results. 



In three preliminary papers (4b) I have investigated the rainfall 

 of the United States, and in them arrived at the conclusion that they 

 afford evidence toward the existence of the rainfall periodicity. 

 When these papers were published it was recognized that they did 

 not constitute proof, that data were needed from all parts of the 

 world and, as Marvin (5) stated in a critical discussion, long rec- 

 ords were needed. Since the publication of the first papers I have 

 been gathering all available data, much of it in unpublished manu- 

 scripts sent me by meteorologists from many countries of the world. 

 The reduction of these data has been a long job, even requiring hun- 

 dreds of hours to prepare a single table. For example, the rainfall 

 of many separate stations were given for Sweden; these had to be 

 combined as one table. The same was true of the Punjab in India, 

 where data from twenty-five stations were copied out of Eliot's 

 book and averaged to give a district record to 1900. After that it 

 was necessary to borrow seventeen large volumes and copy a little 



