METEOROLOGY. 321 



220. Dr. V. Kremzer has stinlied the variability oT rainfall in Europe, 

 following the path of HaDn, vou Danckelmann, and others. Among 

 the general conclusions to be drawn from his tables we note the follow- 

 ing: 



1. The mean variability in quantity of rainl'all increases with dimin- 

 ishing latitude, and the more rapidly in proportion as we penetrate the 

 region of subtropical rains. 



2. From the records for Padua for 150 years he finds that the error 

 of the mean annual rainfall as computed for any 30 years will be 1 per 

 cent, and the largest probable error 2.5 per cent. 



3. The sources of variation in rainfall measures due to changes in the 

 observers, in the exposure of gauges, and in the general changes of the 

 regions surrounding the gauges, are all included in the above figures 

 for the actual rainfall ; he makes a special illustration of the effect of 

 these sources of error in the case of the records at Breslau, where the 

 great mean variability and annual variability is largely traceable to 

 Instrumental changes. 



4. As to stations that lie in the rain shadow of a mountain chain, they 

 also have a higher variability because of this location than do those 

 stations that lie on the opposite side of the mountain and receive an 

 excess of rain. In this respect he examines especially the records from 

 Italy and the Hartz Mountains. 



5. In reference to elevation above sea level, the variability of rain is 

 much less for an entirely plane surface considerably elevated above the 

 sea, but the contrary if mountains intersect an elevated plain. As re- 

 gards the relation between coast and interior stations, the former have 

 the greater variability. 



G. As regards the seasons, the greatest variability belongs to the cold 

 season of the year, while the least variability or greater uniformity 

 belongs to the season of vegetation. 



Kremzer gives convenient general tables for deducing the number 

 of years of observations necessary in order to obtain an annual mean of 

 a given accuracy; thus for a station whose annual variability is 20 per 

 cent., 11.8 years will be required to obtain a mean rainfall correct to 

 within o per cent. The probable errors of the monthly rainfall for any 

 one station as based on observations for 10 years are as follows : For 

 Algiers 22 per cent., southern Spain 19, lower Italy 18, Central Italy 16, 

 upper Italy 15, Switzerland 14, Germany 12 and 11. 



The above remarks refer to percentages of rainfall, taking the annual 

 and monthly mean total quantities as the basis; in a subsequent section 

 on the actual rainfall he discusses the occurrence of dry and wet periods. 

 By combining together all his rainfall records for Europe, he obtains a 

 general mean which must have some cosmic or extended terrestrial rela- 

 tion. Combining each year with its neighboring years ho obtains a 

 smoothed curve, which is expressed in the second column of the following 

 table in percentages of the departure of the quantity of rain from its 

 S. Mis. 33 21 



