SCIENTIFIC BALLOON ASCENSIONS— VON BEZOLD 3I9 



water, and that even for complete saturation under the average 

 distribution of temperature this quantity can at the most rise only 

 a little more than 2.5 kilograms. Therefore in the neighborhood of 

 Berlin the whole atmosphere contains on a average only 1.6 per thous- 

 and (or one-sixth of one per cent) of water, an amount which can 

 only be increased to 2.5 per thousand for complete saturation under 

 the average distribution of temperature. 



A graphic interpolation shows at once that on the average we find 

 one-half of this total quantity of water in the stratum between the sur- 

 face of the ocean and 1600 meters altitude, so that, for instance, at 

 the summit of the Schneekoppe we already have one-half of the 

 total aqueous vapor of the atmosphere below us. 



If, on the other hand, we consider that 1 kilogram of water spread 

 over 1 square meter of ground covers it exactly i mm deep, we 

 can get a standard for determining how rapidly the ascending air 

 must be renewed over such a surface in order to furnish the quanti- 

 ties of precipitation actually observed and which may still be very 

 considerable even at altitudes of 1600 meters. 



However, we must not forget that we can obtain from the psy- 

 chrometric measurements only the water that is present as vapor; 

 how large the quantities of water may be that are present in con- 

 densed form in the clouds has up to the present time completely 

 eluded our observations. 



The considerations just set forth enable us not only to follow the 

 average-content of the atmosphere for each season but also the 

 increase and loss of heat during great intervals of time. To this 

 end the curves of average condition as to temperature and mois- 

 ture as they are already given in fig. 44 and fig. 45 are transformed 

 into the new system of coordinates, and thus we obtain figures 48 

 and 49. From these diagrams by measuring the areas of the surfaces 

 we obtain the numbers given in the following tables 4a and 4b just 

 as previously we had done for the values Y m and Y s in table 3 . 



As we can deal only with rough approximations, therefore the 

 following table 4 is arranged only for large intervals. 



