HT, A. Hazen—Thermometer Exposure. 371 
between the Stevénson shelter and the Kew thermograph 
records which show the latter 49° higher at 9 A. M. and ‘82° 
owerat 9 P.M. Possibly these differences may be due in part 
should be good size and a free access of air to the interior. 
This shelter is 4x38 feet with single louvre work on a 
sides three inches wide and inclined at an angle of 30° to the 
horizontal. The roof is double and the bottom close. The north 
side is a door which can be removed. By the kindness of Mr. 
Clark, opportunity was granted for conducting the experiments 
upon the roof of one of his buildings in a thickly settled part 
of Washington. This roof is about 60 feet above ground, and 
Is free to air currents save from the southeast. here were 
metallic screen inserted, after several weeks’ comparison this 
Screen with its thermometers was placed in the upper ‘“‘ Pattern,” 
its door having been removed. - 
The following plans and precautions were taken to determine 
a adaptability of these shelters, and to check the thermome- 
rs; 
4 single reading a few times each day will hardly give what we 
Wish except for the mean ; while continuous observation, under 
