DISTRIBUTION OF TEMPERATURE ON THE EARTH, 239 
fore the variations are produced at all times of the year by causes 
which exert a general influence, and that for so long a space of 
time as a month, the local perturbations, as caused by a clouded 
sky or a precipitate, may be regarded as subordinate elements. 
In constructing monthly isothermals, however, it is necessary 
to determine this subordinate element, in order to correct the 
observations for a few years of one place by the long-continued 
ones of a neighbouring station. To determine the magnitude of 
the local perturbations, it is not advisable to choose places which 
are distributed over a large extent of country, but rather groups 
of stations situated close together ; the places of observation in 
the state of New York, on the one hand, and in Suabia, Bavaria, 
Belgium, Bohemia, and Saxony, on the other, furnish two groups 
well suited for this purpose. 
A second question, which was only partially answered in the 
first part is, In what direction does any abnormal temperature 
propagate itself over the surface of the earth? In examining 
this point the monthly means could not be employed, because 
extremes of short duration are frequently obliterated during that 
time by deviations in an opposite sense. The extension of un- 
usual warmth in winter, by the action of south-west winds, gene- 
rally takes place too rapidly for its advance from station to sta- 
tion, to be traced by the monthly means; it was, therefore, ad- 
visable to employ five-day means, and for this purpose the cal- 
culations of Brandes at Leipzig, and Suckow at Jena, and also 
thoseof Schrénn were taken. Brandes’ calculations are made from 
the observations of 9 years at Petersburg, 
6 , » Sagan, 
9 5 wy “Rochelle, 
12 ., 5, Manheim, 
10. 3 » St. Gothard, 
104 3 -y) Bome; 
5 5, », Zwanenburg. 
and also for 27 years at Jena. 
Here follow the tables, of which the annexed is a specimen, 
containing the five-day means at Petersburg during the months 
of August and September. 
