HT, A. Hazen—Thermometer Exposure. 367 
bility of more extended experiment. Thermometers suspen- 
ded in free air, in the shade of a dwelling or wall during the 
desired to critically study the past records, or what is more 
Important to compare observations, whether mean or daily, at 
different stations it will manifestly be necessary to eliminate 
from them all effects of improper exposure, ore 
ttmay be argued that the most important consideration is 
that of uniformity and that constant errors may be neglected, 
provided they are the same in all the exposures. If, however, 
varying atmospheric conditions diminish or intensify constant 
sources of error, it is wise to avoid these as much as possible. 
The essential point to be regarded is that a shelter shall at 
any and all times give an air temperature influenced as little 
as possible by harmful causes. 
fulfills the first condition above. The east an 7 
Pieces however, employed for screening from the morning and 
afternoon sun, would seem to check very light winds on those 
sides, and there does not seem to be sufficient provision against 
soil or sod radiation. A shelter similar in plan to this has been 
adopted as a standard in Melbourne, Australia. This shed on 
144 square feet of horizontal surface, two roofs, of pelvaot . 
iron, nine inches apart, the ridge of the outer roof eight fee 
above ground, the thermometers suspended in a wire cage one 
