CONSIDERATIONS ON THE OBJECT AND METHODS. 5 



The test that the distribution in space of the points of observation fulfil this 

 condition will be, that it turns out to be possible to draw synoptical maps, by use of 

 the observations; for such maps give continuous representations of the fields of the 

 observed elements. The distances to be allowed in the net of observations will 

 therefore depend upon the space- variations of the elements. The network must be 

 satisfactory for the element having the strongest space- variations. But nothing 

 hinders elements which have less irregular distribution in space from being observed 

 at a smaller number of points in the network. 



A suitable time-differential must be determined by a comparison of synoptic 

 charts representing the field of the same element at successive epochs. The changes 

 which the element has undergone from epoch to epoch must be small enough to 

 allow us to form satisfactory approximate values of the time-derivative of the element. 

 The time-differential must therefore be chosen so as to suit the element which has 

 the most rapid time- variation. But nothing hinders elements having slower time- 

 variations from being observed, only, for instance, at every second or every third of 

 the eoochs of observation, which have thus been chosen. 



93. Special Remarks on Meteorological Observations. In passing to concrete 

 meteorological observations, we shall first make some remarks regarding the principle 

 of simultaneity. 



The ideal is of course the use of self-recording instruments having sufficiently 

 large time-scale. But whichever instruments or methods of observation be used, 

 it will be neither possible nor required to realize simultaneity in the mathematical 

 sense of the word. Most meteorological elements will under ordinary circumstances 

 change very little during as small an interval of time as, for instance, half an hour. 

 Departures of this magnitude from the precise epoch of observation will therefore 

 not usually produce errors of greater importance, though exceptions are not excluded, 



The general slowness of the variations makes it possible to use averages 

 registered during suitable intervals of time instead of true instantaneous values. 

 For one element, wind, the use of averages, as we shall see, will be unavoidable, and 

 it will have certain advantages also in connection with other elements, especially 

 inasmuch as time-integrations should be performed afterwards. But if averages 

 be used, they should be used at all the cooperating stations, and taken according to 

 the same rules at all. These conditions have been excellently fulfilled by hourly 

 averages which we have obtained from the U. S. Weather Bureau. 



Observations obtained from the higher strata by meteorological ascents will 

 cause certain difficulties inasmuch as the records taken by the same instrument at 

 different levels are not taken simultaneously. But the departures will be reduced 

 according as we increase the velocity of the ascent. A registering balloon can be 

 made to mount from the ground to the lower limits of the isothermal layer in less 

 than an hour. Departures up to half an hour from the true epochs of observation 

 being considered allowable, we are entitled to consider the observations obtained 

 by such a balloon in different levels as simultaneous with observations taken near 

 the ground half an hour after its launching. 



