734 PROFESSOR TAIT ON THERMAL AND ELECTRIC CONDUCTIVITY. 
with a point of contrary flexure), while barely affecting the run of the higher 
and much more extensive part of the curve. This remark does not require (as 
will be seen) to be applied to the case of iron, which appears to be a thoroughly 
exceptional one,—though manifest indications of it are to be seen in ForBgEs’ 
diagram of the conductivities of the bar when naked and when covered with 
paper. Another cause may have some effect here. The excesses of tempera- 
ture above that of the air are so small that an inevitable error of even 0°:1 may 
produce a serious effect on the calculated result. 
If I have sufficient leisure, in the course of next session, I hope to settle 
this point by using a cold water bath applied near the middle of each of the 
iron, German silver, and lead bars, the source of heat being kept at as high a 
temperature as in the experiments already made. I now believe from experi- 
ence that in measuring conductivity, at whatever temperature, things ought to 
be arranged so as to avoid any very slow flux of heat. And I also think that, 
especially for very good conductors, such as copper, the bars should be smoked. 
§ 15. With these observations I submit the following values, by no means as 
final even so far as my own work is concerned but, as probably fair approxima- 
tions to the truth. The units are the foot, minute, and degree centigrade, the 
unit of heat being that required to raise the temperature of a cubic foot of the 
substance by 1° C. [See the end of this section. | 
TRon. 
Temperature C. Conductivity. F. 
0 0-:0149 (0:0190) 
50 0°0138 00131 
100 00128 0-0115 
150 00121 00107 
200 00114 00100 
250 00109 00094 
300 *0:0105 0:0089 
350 *0-0102 
The two numbers marked with an asterisk are merely probable deductions from 
the curve representing the others. They are introduced to show the difference 
in character between my results and those of ForsBeEs, due mainly to the differ- 
ence in our estimates of the rate of cooling at high temperatures. The column 
headed F. is (graphically) interpolated from Forses’ table (Zrans. RS L., 
1864, p. 102), which refers to the same bar under the same conditions. This 
table does not extend below 17° C., so that the number in brackets is tosome 
extent conjectural. It is inserted to illustrate what I have said in § 14 above 
as to the rapid change of conductivity indicated when temperature excesses 
are small. 
My numbers seem to point to a temperature, somewhere about red-heat, at 

