104 PRINCIPAL FORBES ON AN EXPERIMENTAL INQUIRY INTO 



110. Uncertainty, I have already said, attends the determinations of con- 

 ductivity for the higher as well as those at the lowest temperatures. In fact, the 

 former are (as will have been seen from the details already given) the results of 

 analogies rather than of direct experiments. The experiments, whether Statical 

 or Dynamical, rarely extended beyond a temperature of 200°, or at most 220° 

 Cent. The results have been here carried out by the analogies afforded by the 

 equations to the curves to nearly 300°. Nevertheless, the continuity of the law 

 of conductivity diminishing with temperature, is consistently brought out by these 

 approximations. 



111. In the preceding Tables the conductivity is expressed in terms of the 

 amount of heat as unity, which is required to raise the temperature of one cubic 

 foot of iron, by one degree Cent. It expresses the amount of heat reckoned 

 in such units which would traverse in one minute across an area of one square 

 foot, a plate of iron one foot thick, with the two surfaces maintained at tem- 

 peratures differing by 1° Cent. 



112. If we now project the values of the conductivity of iron found in the last 

 column but one of the three preceding Tables in terms of the thermometric tem- 

 peratures (Centigrade) in the last columns, Ave are enabled to trace easily the 

 connected results of the whole inquity. (See Plate V. fig. 4.) 



113. We tind that in each case the conductivity diminishes as the temperature 

 increases; and that, for the next part, in a progressive manner. The variation 

 with temperature is clearly most rapid at the low r er temperatures. 



114. The two first series agree very closely in their numerical results, with 

 the exception of certain irregularities in the part of the curve where the tempera- 

 tures are lowest ; which have already been in part accounted for (Arts. 55, 65, 

 109). These two series belong to one and the same bar, though cooling under 

 very different circumstances, owing to the largely increased radiating power 

 conferred upon it by coating it with paper. And the value of the striking 

 coincidence in the numerical results in Tables I. and II. is enhanced by the 

 consideration, that the numbers expressing the conductivity are obtained bv 

 taking the ratios of two different columns (7 and 8), which in the two Tables 

 differ most widely, and the result cannot be even guessed at until the ratio is 

 actually taken. 



115. The third series (Table X.) leads to numbers very sensibly differing from 

 the two first series, yet following the same general law, the conductivity decreas- 

 ing with temperature (excepting at the lowest part of the scale, where we find an 

 anomaly corresponding to that noted in an early part of this paper (Art. 65), 

 showing that the lowest portion of the statical curve has not in this instance 

 been satisfactorily determined). The conductivity in Table X. is smaller 

 throughout than in the tw r o former cases. It is believed that this can be 

 satisfactorily accounted for by the different quality of the iron of which this 



