86 REPORTS ON THE STATE OF SCIENCE. 
on the indications of the aneroids was not examined, and it was assumed 
that errors due to this cause would be negligible in comparison with the 
general accuracy of the observations. With this method of testing, 
Teisserenc de Bort found that his maximum barometric heights in ascents 
from 1899-1903 agreed on the average with trigonometrical observations 
to nearly +1 per cent. up to heights of 10 kms., and at 4 kms. the difference 
was negligible! The error was not wholly due to the temperature 
coefficient of the barometer, but partly also to the lag of the instrument. 
Later observations indicated an average error of + 2 per cent. at the 
maximum height and a very considerable lag at the lower altitudes.” 
After the work of Hergesell and Kleinschmidt on the temperature 
coefficients of the barometers*® Teisserenc de Bort began to test this effect, 
but instead of applying corrections for the temperature he replaced all 
tubes having too large a coefficient by new ones whose temperature 
coefficient was negligible. 
During the past two years, in accordance with the suggestions of 
Hergesell, certain of the barometers have been tested at various tem- 
peratures down to —50°C., and it has been found that the majority of the 
barometric heights above 13 to 14 km. are too low by at least 0°5 km. 
At 20 km. the correction is often 2 km. and even more. Many attempts 
have been made to eliminate this effect of temperature, but they have 
not been altogether successful. When the temperature correction is 
applied it is specially noted in publication. It may be pointed out that 
the application of the temperature correction to the Teisserenc de Bort 
barometer is not easy, as the Bourdon tube is enclosed in a cork case and 
its temperature in an ascent is not accurately known. The thermometers 
are compared every two or three months with a standard thermometer 
in an alcohol or petrol bath cooled to —70° C. with carbonic-acid snow, 
the alcohol being continually agitated. 
The methods of testing the instruments employed in the ascents from 
St. Louis, U.S.A., in 1904-07 were similar to those of Teisserenc de Bort. 
The barometer was tested under the receiver of an air-pump, and the 
thermometer by means of a mixture of alcohol and carbonic-acid snow 
down to a temperature of —83°C. No correction was applied for the 
temperature coefficient of the barometer. 
A similar method of testing is adopted at the Observatoire Constantin, 
Russia. 
In Germany the usual method of testing is different, and correction 
is made for the temperature coefficient of the barometer. 
At the Lindenberg Observatory the barometer is tested by placing 
it under the receiver of an air-pump and exhausting to various pressures. 
The air-pump receiver, which is of metal, has triple cavity walls through 
which carbonic-acid gas is allowed to circulate. The temperature is thus 
reduced to various values and the temperature correction of the barometer 
determined for different pressures. The air inside the receiver is kept 
in vigorous circulation by an electrically driven fan, and the temperature 
is indicated by a standard thermometer viewed through a double-glass 
window designed to be free from condensation of moisture on its faces. 
The thermometer of the instrument is calibrated at the same time. The 
exhaustion is carried down to about 50 mm. and the temperature to 
1 Brit. Assoc. Reports, 1903, p. 561. 
2 Compt. rend., 141, pp. 153-155, July 10, 1905. 
3 Bettrdge z. Physik der Freien Atmosphdare, 1, 1905, pp. 108, 208. 
