THE TEMPERATURE OF LEAVES OF PINUS IN WINTER 47 
If these values are plotted against the temperatures as ordinates, a 
curve is obtained which, for all practical purposes is a straight line, 
the deviation at — 5° C. being negligible. From this curve (figure 2), 
by interpolation, values can be obtained for any temperature between 
— 20° and 20° C. The two columns of figures at the right of figure 2 
contain these empirical numbers with the computed numbers inter- 
polated. 
8. Sensitiveness of Apparatus. — The galvanometer was found 
readily sensitive to a displacement on the bridge wire of two milli- 
meters. Taking the factor obtained for 0° C. as an example, the 
apparatus is sensitive to .2 8.30 = 0.024° C, which is amply 
sufficient for the purpose of this investigation. The apparatus may 
be made still more sensitive by decreasing further the current from 
the cell and increasing the sensitiveness of the galvanometer corre- 
spondingly. 
9. Accuracy of the Method. — ^To determine whether the theoretical 
sensitiveness of the apparatus would hold in practice, tests were 
made in December and January, and again in March at the close of 
the experiments. The method consisted in placing the junctions in 
baths of known temperatures, finding the corresponding displacement 
on the bridge wire, computing the temperature by applying the proper 
factor obtained from the curve, and then comparing the calculated 
with the actual temperature difference. The results are given in 
Table II. 
Table II shows representative readings taken from a large number. 
The average error of the December and January readings is only 
0.0155°; the greatest for any one reading 0.055°. For March the 
average error is 0.0233°; the greatest 0.056°. 
10. Uniformity of the Junctions. — Table III gives the results of 
an experiment to test the uniformity of the junctions used. Two 
sets of junctions were placed in the same baths. By throwing a 
switch, changes from one set to the other could be made in rapid 
succession. A number of other tests gave results similar to those 
in the table. 
The greatest difference in displacement is 2 mm. This, at the 
temperature difference given (5.04°) represents an error of 0.023° C, 
or .4 per cent. 
1 1 . Sources of Error. — The possible sources of error in this method 
are: (i) Variation in the current due to weakening of the cell; (2) the 
