4 BULLETIN 420, U. S. DEPARTMENT' OF AGRICULTURE. 
of Standards. A Weston cell certified by that bureau furnished the 
known potential. A switch, essentially the same as that described 
by White, 1 was used to correct for parasitic electromotive forces. 
In the set-up the hot junction of each element was placed where 
desired and the cold one packed in crushed ice within a thermos 
bottle. The terminals were led to mercury pools in a block of par- 
affined wood protected from air currents. Connection with the 
potentiometer was made through copper leads whose bared and amal- 
gamated ends could be dipped in the successive pairs of mercury 
pools. In this way any element could be quickly and easily brought 
into circuit. 
Differences of one microvolt could be detected by the system. This 
corresponds to approximately 0.018° F. (0.01° C). The experimental 
errors incident to rapid readings resulted in ah accuracy of about 
0.09° F. (0.05° C). 
In determining the difference in temperature between the top and 
bottom of a bottle a differential thermo element was used ; this con- 
sisted of two pairs of copper-constantin couples in a thin glass tube. 
The hot and cold junctions were placed so that one would be about 
one-half inch above the bottom and the other just beneath the 
surface of the liquid when cool. This differential couple was not 
calibrated, and its design was not all that could be desired, but it 
doubtless was sufficiently accurate to furnish a fairly good measure 
of the differences in temperature between the top and bottom layers. 
In this connection it may be noted that the thermometers used in this 
work are numbered from 1 to 10, which numbers are used to identify 
them with the curves in the drawings showing the results of the various 
experiments. 
Air measurements were taken with a Taylor-Pitot tube placed in 
the center of the straight part of the discharge pipe from the box, 
about 4 feet from the nearest bend. The tube was arranged to give 
both the impact and static pressure, their difference, or the velocity 
head, being indicated on a graduated water column. Only one tube, 
situated in the center of the pipe, was used. The average velocity 
head across the entire section of the pipe was determined by multi- 
plying this center reading by the factor 0.91. 
The velocities obtained with the Pitot tube were checked with an 
anemometer which had been calibrated by the Bureau of Standards. 
The anemometer was moved across the. end of the outlet pipe in such 
a way as to give what was considered an average reading. As the 
temperature of the air in the outlet pipe from the box was, of course, 
higher than that of the incoming air, the volumes were corrected in 
the ratio of their absolute temperatures to give the inlet volume. 
i White, Walter P. Journal of American Chemical Society, 36, p. 1S56, 1914. 
