18 BULLETIN 1059, U. S. DEPARTMENT OF AGRICULTURE. 
a partly open floor, and walls made of slats which overlap to ex- 
clude light from any overhead source without causing complete 
stagnation of the air within. 
Modifications which would still more fully overcome the heating 
of the shelter have been proposed by Koppen (23), who would pro- 
vide artificial air circulation, but such provisions will hardly be 
necessary for any ordinary thermometric work. On the other hand, 
the observer can not be too strongly urged to provide shelters which 
will give the maximum of light protection without preventing the 
natural air currents from coming in contact with the thermometers. 
To obtain true nocturnal temperatures near the ground it may be 
desirable to use a shelter without a floor, so that radiation from 
below the thermometers is retarded as little as possible. In the 
work at the Fremont Experiment Station,. Bates has found that a 
shelter for ground temperatures need be no more than a hood, fully 
open to the north and below the thermometers. If there should be 
considerable reflection from the north at midday, it could be largely 
eliminated b}^ an absorbing screen set a foot or two from the hood.. 
Standaedizing Theemometees. 
The present possibilities of correlation between temperatures and 
plant behavior do not justify the greatest precision in thermometry. 
Units smaller than 1° may be ignored in field work, for all practical 
purposes, though personal taste may dictate that tenths of degrees 
be recorded. The essential thing is that only reliable thermometers 
be used, as the errors in cheap thermometers are not uncommonly as 
great as the difference between two conditions which are being 
studied. Even the standard types of maximum, minimum, and mer- 
curial thermometers may well be critically examined and compared 
with a standard before being used. The Bureau of Standards (37) 
calibrates such instruments at a nominal cost. 
With recording apparatus, such as the air thermograph, adjust- 
ment takes the place of standardization. The use of any such re- 
corder, without thermometers to check its accuracy at frequent in- 
tervals, can not be recommended. 
Maximum, Minimum, and Cueeent Tempeeatuees. 
Where only maximum and minimum thermometers are available, 
of the standard Weather Bureau pattern, the maximum and mini- 
mum temperatures for the preceding 24 hours should be recorded 
once each day, either before 10 a. m. or after 4 p. m., and at the same 
time the current temperature, as indicated by the minimum ther- 
mometer, should be recorded, also the time of the observation. The 
current temperature is principally of value for making a thermo- 
graph correction, and the height of the thermograph pen should 
therefore be recorded at the same time. 
