ALTER: RAINFALL AND Sl'N-SPOT PERIODS. 31 



small, mountainous island, where, as Professor Pickering says, ''The 

 rainfall is very unequal in different portions of the island." It 

 varies from 33 inches west of the mountains to 248 on the eastern 

 end of the island. For ^Madagascar there is but one station, with 

 a record over only 21 cycles, so that the correlation is all that one 

 could expect. In the case of South Australia, however, we have a 

 long, homogeneous record from fifty stations. The effect of the 

 period is evidently much less certain there than in the region of 

 groups A and B. In this it reminds one of the results obtained from 

 the central third of the United States, a region located between the 

 two types represented by groups A and B. Data are not at hand to 

 show whether such a reversal, as in the United States, would be found 

 between the northern and southern parts of South Australia. An 

 investigation of this character would. I venture to predict, show the 

 reversal. I hope to secure data to examine this region more 

 thoroughly. 



GENERAL DISCUSSION. 



In group A. which consists of interiors or eastern coasts of large 

 continents, we find the minimnm of our curves coming exactly at 

 phase 1 in each case. This is the phase, as told above, which every 

 ninth cycle contains the sun-spot maximutyi. Each of these curves 

 shows also the effect of a second harmonic of this period with one 

 minimum at this same phase, the other neutralizing the maximum, 

 which would normally fall at phase 8. This much can safely be ac- 

 cepted as true features of purely continental cur^-es. 



In group B we find more variation in cur\-es from one section to 

 another. For the Pacific coast we find the minimum at phase 7 and 

 the maximum at phase 13; for northern Europe the minimum at 7, 

 if we smooth our curve, and the maximum at 14. The small amount 

 of data from Chile does not give any very definite results, almost 

 equal minima at 2 and 12, with maxima at 10 and 14. The marine 

 type seems, then, with considerable uncertainty, to give a minimum 

 of rainfall at time of sun-spot minimimi and a maximum shortly 

 before the sim-spot maximum. 



The halves or thirds of the curves at any one place will differ 

 from each other for one or more, probably all. of the following 

 reasons: 



(a) Accidental errors and other periodicities are not entirely 

 damped out. 



(5) The epochs of sun-spot maxima and minima are uncertain, 

 and consequently some data are incorrectly placed by one or more 



