THE SUN-SPOT PERIOD. 141 



in general emplo}' data other than those g-iven hj a single station, 

 instead of emploj'ing the onlj' rational method which could distinguish 

 a general influence of solar origin from causes purely local and tem- 

 porary, nameh^, the study of the contemporaneous records of numer- 

 ous stations. Furthermore, the authors did not employ a sufficiently 

 long period of observations, for these ought at the very least to extend 

 over a complete sun-spot C3'cle. Some writers even ventured to draw 

 conclusions from the observations continued only a few months at a 

 single station. rinall3% for the most part these earl}- investigators 

 studied the records of stations in the temperate zones, where, as 

 Koppen has shown, the local and accidental variations are so great as 

 to mask completely such minute changes of mean temperature as are 

 here in question. 



In 18T3 there appeared the well-known memoir of Koppen, who 

 concluded, from an able discussion of the thermometric observations at 

 numerous stations during the period from 1820 to 1870, that the pres- 

 ence o| sun spots was attended by a slight diminution of the terres- 

 trial temperature." Since the appearance of this memoir, which 

 constituted the first trustworthy results reached in this direction, no 

 extended work on the subject has been published. 



II. 



Encouraged by the friendh' counsel of M. H. Poincare, I have 

 undertaken to continue the study of this important subject for the 

 period 1870 to 1900, for it seemed to me very desirable to throw addi- 

 tional light, if possible, upon a point so important to ph3'sical astron- 

 omy and the physics of the earth. 



The work of Koppen esta1)lished that the curve of variation of mean 

 annual temperature is reasonably regular only for tropical stations, 

 and that in the regions exterior to the Tropics the curve of variation 

 becomes so irregular that it is impossible to recognize in it any perio- 

 dicity whatever. This result was perhaps to be expected, for the 

 tropical regions are characterized by a very even climate, whereas 

 for stations nearer the poles the accidental variations of temperature 

 are very great, and indeed enormousl}' greater than the slight varia- 

 tion of temperature which will be found below to attend the sunspot 

 c^xle. 



Accordingly 1 have made use of thermometric o])servations from 

 tropical stations exclusively in this study, but since the meteorological 

 observations of the past thirty years have been greatly extended and 

 sj^stematized, I have been able to employ material nuich more exten- 

 sive and trustworthy than was at Koppen's disposal. Thus the series 

 of observations for separate stations are generally longer than he 



« Koppen: Zeitschrift der r)sterreicliis('he Gesellschaft fiir Meteorologie, YdI. VIII, 

 1873, pp. 241, 273. 



