FREDERICK GUTHRIE ON THE THERMAL RESISTANCE OE LIQUIDS. 619 
and sperm-oil (as measured by the number of units of heat equal thicknesses of them 
arrest in a given time), are in the order just written, the resistance of water being least. 
I have not at present sufficient data for establishing the relation between the time t and 
the resistance. 
§ 46. If we had found that the time t required for penetration had been the same for 
heat of all temperatures, we should of course expect to find the time t proportional to 
the thickness That this is not the case appears from the following experiment 
with water. The interval between the plates was raised successively from 1 millim. to 
6 millims. by degrees of 0 - 5 millim. ; at each half millimetre the time t was measured 
under precisely similar circumstances. 
T =26-5 
T 1= 36-5 
AT =10 
*. 
t. 
X 
Mean. 
t. 
At. 
millims. 
r 
s 
S 
s 
s 
— \ 
1 
4 
3 
3 
4 
3 
3-4 
T2 
1-5 
5 
4 
5 
5 
4 
4-6 
1-8 
2 
6 
7 
7 
5 
7 
6-4 
2-0 
2-5 
8 
9 
'8 
9 
8 
8-4 
2-0 
3 
10 
10 
11 
11 
10 
10-4 
2-8 
3-5 
13 
14 
13 
13 
13 
13-2 
2-2 
4 
15 
16 
16 
16 
15 
15-4 
3-0 
4-5 
18 
19 
19 
18 
18 
18-4 
3'6 
5 
20 
21 
21 
21 
22 
21-0 
2-8 
5-5 
24 
24 
24 
24 
23 
23-8 
3-2 
6 
25 0 ) 
26 
27 
28 
27 
27-0 
§ 47. It hence appears that the thickness and time of penetration are not directly pro- 
portional to one another, but that if one disk of water be twice as thick as another, 
the thicker one will delay the heat more than twice as long as the thinner one. And 
this fact may well arise from the one previously established, that heat of a high tem- 
perature penetrates more quickly than heat of a lower temperature. For if we imagine 
two simultaneous experiments in which the heat reaches the middle of the thicker layer 
exactly at the moment when it has just completely penetrated the thinner one, then 
the second half of the thicker liquid will have to resist the passage of heat from a source 
of lower temperature, and this, as we have seen, implies a slower passage. Neither the 
case where the liquid is very thin, nor that in which it is very thick, is the most favour- 
able for determining the time t. In the first the margin of unavoidable error forms 
a large fraction of the observed time t; in the second the commencement of the motion 
in Q is so gradual that some little time may elapse before it is recognized. The instant 
of the commencement of a slow motion cannot be so accurately determined as that of a 
