232 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 70 



correct there must plainly be great and sudden changes in the daily 

 and yearly temperature amplitude at different parts of the earth 

 and particularly we must expect that these will be most strongly 

 marked in the tropics. But the investigations which we have 

 summarized of the daily temperature amplitudes in several tropical 

 stations show no secure indication that this is the case. 



Only at Antananarivo and Fort de France the combined curves 

 of daily amplitude which we have collected show considerable varia- 

 tions. The curve for the first named station (fig. 71, IV, T-A) 

 shows no marked similarity with the sun spot curve or the promi- 

 nence curve. To be sure it has a maximum between 1892 and 

 1895 that may have a certain similarity to the sun spot maximum, 

 but its most conspicuous minima in 1891 and 1897, as well as the 

 rise from 1897 to a maximum in the year 1908 and 1909, have little 

 similarity to either the curves of sun spots or of prominences and 

 just as little with the magnetic curves. The whole appearance of the 

 curve is indeed very exceptional. 



The curve for the temperature amplitude in Fort de France 

 (fig. 71, VI, T-A) has more similarity with the curves of sun spots 

 and prominences, having a minimum in the year 1900 and a rise in 

 the years after. The maximum comes in the year 1907, that is, in 

 the last year of sun spot maximum and exactly in that year when 

 the prominences reached their maximum. In the earlier sun spot 

 period the maximum of temperature amplitude falls between 1893 

 and 1894 very well with the sun spot maximum, but in this period 

 there is a secondary maximum in the year 1897, and the sun spot 

 period is therefore divided into two parts, a phenomenon which 

 we have already often observed. A corresponding minimum we 

 find also in the precipitation curve for 1897. It has the appear- 

 ance as if in these cases there is really an increase in the daily 

 temperature amplitude with simultaneous increase of sun spots. 



At the other tropical stations which we have investigated, we 

 can find, however, no well marked dependence between the sun 

 spot curve and the curve for the daily amplitude. We have already 

 spoken of this in regard to Batavia (see fig. 68). We find there 

 that the daily amplitude increases with decreasing cloudiness, as is 

 natural. The less the prevailing cloudiness the greater is the out- 

 going radiation and consequently the greater the amplitude of the 

 temperature. We find also that the curve for the daily amplitude 

 rises and falls about simultaneously with the temperature curve 

 and the curve for the air pressure. That the latter would be the 



