92, REPORTS ON THE STATE OF SCIENCE. 
the correction is considerably larger, amounting for some instruments 
to as much as 5 or 6 km. at 25 km.' 
The indications of the thermometers are open to no such objections, and 
from the inter-comparison of different types of instruments it may be 
stated that the temperature 1s known to within +1° C. 
The indications of the hair hygrometer are considerably less accurate, 
as was shown by Kleinschmidt.” The instrument is quite unsatisfactory 
for work at high altitudes, and is unreliable at heights where the tem-— 
perature is below —5° C., 2.e., at heights greater than about 5 kms. 
IV.—(a) Mean Temperatures and Gradients of Temperature. 
The most important meteorological element of which observations 
can be made in the free atmosphere is temperature. Observations of 
pressure furnish practically the only means of estimating heights, and they 
cannot therefore be used to determine directly the distribution of pressure. 
The latter can only be determined indirectly by calculation from the 
observations of temperature and the pressure at the surface. Thus, 
while dynamical meteorology must necessarily be based on a knowledge 
of the pressure and density distributions, it rests ultimately on the distri- 
bution of temperature, and in a lesser degree on that of humidity, in the 
free atmosphere. The calculations are obviously laborious even when 
sufficient observations are obtained ; the difficulty and expense of obtain- 
ing the observations make the task appear almost hopeless. Thus no really 
serious attempt has been made to calculate from observational data the 
actual synchronous distribution of pressure in the upper atmosphere at 
5-10 km. altitude at times when the surface distribution 1s meteorologically 
most interesting. Our knowledge is confined practically to mean values. 
In order to avoid as far as possible negative quantities and to 
facilitate calculation and comparison, temperatures have been usually 
expressed in degrees C. above the absolute zero—273° C. on the ordinary 
scale. Atmospheric temperatures in temperate latitudes lie almost 
invariably between 200° and 300° on this scale, and the initial 2 may be 
generally omitted without risk of confusion. The letter A is used in 
connection with this scale; thus (2)73° Ais 0°C. Further, the vertical 
gradient of temperature is expressed in degrees C. per kilometre and is 
reckoned positive when temperature diminishes with increasing height. 
The most complete contribution hitherto made to the discussion of 
upper air observations is that of Von Bezold, Assmann, Berson, and Siiring * 
who dealt with the observations obtained from manned balloons. The 
following table gives the values they found for the gradient of temperature 
for each kilometre up to 9 km. :— 
Height . F ; 3 © SOS) 6 1=9 223) 4324 426) (5=6' 1 6-So THe sas km 
Gradient , F 5 ~ BON. SiO 2 16°4. . 6:3: — 642.169)... GB 7-2 
Number of Cases . : SIMOOM Oar, AD cea. LO heed 5 2 
Probable Hrrorin Gradient .ae—— 9 — Ie Oe) Oe 
In the surface layer the gradient is affected by inversions, 7.e., exceptional 
cases where the temperature increases with the height. Such cases occur 
most frequently in winter, and as the number of winter ascents in the 
) Ergeb, der Arb. am Aér. Obs. Lindenberg, 1906, p. iii. 
2 Beitr. z. Physik der F'r. Atmos., Bd. ii., Heft i. p. 99, 1906-1907. 
8 Wissenschaftliche Luftfahrten, Braunschweig 1899, 2 vols. 
a te lia sat 
