332 
Journal of Agricultural Research 
Vol, V, No. 8 
to end and joined with silver solder. The wires were then bent at the 
junctions into a grid, with the parallel lines about 5 mm. apart and with 
copper and constantan alternating. Each constantan wire and three of 
the five copper wires are about 7 cm. long, so that the distance between 
the two opposing sets of junctions is the same as that between the copper 
wall and the zinc wall. The two other copper wires, which are at opposite 
ends of the series, are longer, to form leads for the element, as explained 
below. 
Wire of the size stated was used chiefly because it was most readily 
available and seemed quite well adapted to the type of element con¬ 
structed. Theoretically, a small wire would be preferable, because of 
smaller thermal conductance, but the support in which each element is 
mounted probably greatly delays change in temperature of the wires 
between the junctions, while affording opportunity for rapid change at 
the junctions. This support consists of a hard maple rod or spindle 
about 10 cm. in length and 15 mm. in diameter. A recess 8 mm. wide 
and 2 mm. deep is cut around the spindle 3 cm. from one end, and in the 
surface are 10 equally spaced longitudinal slots, each nearly 1 mm. wide 
and 2 mm. deep. The five copper and four constantan wires which, 
joined alternately in series in a grid, as described above, comprise the 
four differential thermocouples of an element, were forced into these 
slots until they were about a millimeter below the surface of the wood 
and to that extent were protected against contact with the metal sleeve 
and thimble by which the thermoelement is supported in the walls, as 
explained below. By means of a cut between two adjoining slots near 
the center of the spindle the copper wire at one end of the series is doubled 
back and extends parallel with the copper wire at the other end of the 
series, the two projecting from one end of the spindle and providing 
terminals for the element. The spindles with the wires thus embedded 
were boiled in paraffin for two or three hours, so that they would not 
swell or shrink with changes in the humidity of the air. 
The temperature of the wires thus embedded in the spindle is probably 
that of the spindle and therefore changes slowly—i. e., the temperature 
gradient in each wire is quite like that of the others in the element and 
is relatively constant for considerable periods. On the other hand, the 
junctions between the copper and the constantan wires are not embedded, 
one series of four alternate junctions projecting into the air at one end 
of the spindle, while the series of opposing junctions projects into the air in 
the recess near the other end of the spindle, so that changes in the tempera¬ 
ture of the air surrounding them affect the junctions quickly. 
To keep each element in place between the two metal walls a short 
copper sleeve is passed through a hole in the zinc wall, the sleeve being 
soldered to the zinc at the edge of the hole to insure good thermal con¬ 
ductance; and directly opposite, with its open end facing that of the 
sleeve, a short copper thimble is firmly soldered to the outer surface of 
