102 REPORTS ON THE STATE OF SCIENCE. 
temperature gradient will remain nearly adiabatic, and will therefore 
allow a constant flow downwards. But when the air enters the region 
between the surface and 5 km. it will begin to be cooled by radiation, and 
the cooling will increase with approach te the surface, although in the 
surface layers themselves convection may reverse the process. Such 
cooling would be an effective bar to further direct downward convection 
and would allow only a gradual oblique convection by which the descending 
air would be transferred to the earth’s surface, being cooled sufficiently by 
radiation in its progress to enable the convection to take place. 
IV. (c) The Advective and Convective Regions. 
Perhaps the most remarkable phenomenon revealed by the investiga- 
tion of the upper air with balloons carrying self-recording instruments 
is the comparatively sudden cessation of the fall of temperature at a 
height varying with the time and the latitude. Above this height, which 
may be regarded as the height of an irregular but roughly horizontal surface 
dividing the atmosphere into two regions, the temperature at any time 
varies very little in a vertical direction, showing on the average a slight 
tendency to increase. This comparative absence of regular vertical 
variation of temperature in the upper region led to the name ‘ isothermal 
layer or region ’ to distinguish it from the lower atmosphere, in which the 
vertical variation of temperature is about 6° C. per 1,000 m. The first 
indication of a considerable falling off in the gradient appears to be con- 
tained in a paper by M. Pomortzeff' referred to by Berson in the dis- 
cussion of the Berlin results. Pomortzefi tried to explain on theoretical 
grounds the diminution he found. ; 
The actual cessation of the fall of temperature was first noticed by 
M. L. Teisserenc de Bort * in June 1899, and again in March 1902. It was 
also discussed shortly afterwards by Assmann.®* 
Teisserenc de Bort found the average height at which the change 
occurred to be about 11 km. He discovered also that the height was 
greater near centres of high pressure than near centres of low pressure, the 
average heights for the two cases being 12°5 and 10 km. respectively. 
Later observations agree on the whole with these results. 
It may be asked if this is due to the slope of the isobaric surface, which 
would be lower over a cyclone than over an anticyclone. This is not the 
case. The difference of pressure over the two regions at a height of 10 km. 
does not amount to more than 10 mm., while the difference of pressure 
between 10 and 12 km. is 50 mm. 
This excludes the hypothesis that the air in the upper region is an 
inert isothermal mass consisting always of the same air, lifted up and down 
by the disturbances in the lower part of the atmosphere. There must be 
interchange of air in the upper region itself or between the two regions. 
The absence of vertical temperature fall implies that general direct 
convection in the upper region is also absent, but the occurrence of irregu- 
\ Wozduchoplanawiji i Izsledonaniji Atmosfery, vol. iii. 1897. 
The results obtained by Hermite and Besancon in March 1893 showed a tem- 
perature of 21° C. just below 16 km., but the balloon floated for some hours at that 
height, and in no sense can they be said to have anticipated Teisserenc de Bort’s 
discovery. C.R. cxvi. p. 767. 
* Séances, 1899, &c., Annuaire de la Société Météorologique, 1902. 
° Ergebnisse Aéronautischen Obs., Berlin, May 1902; Berl. Ber., 1902. 
* But see Shaw’s deduction, infra, p. 108. 
