OF IRON, COPPER, AND GERMAN SILVER. 545 



By adopting this precaution I have verified Prof. Tait's results, and 

 although in some cases the temperature excess was as much as 280°, no such 

 result as that obtained by Forbes was even indicated. 



In addition to the usual cooling experiments, a series was conducted at 

 temperatures as high as it was possible to safely observe with a mercurial 

 thermometer, generally about 330° C. These experiments were of great use in 

 accurately estimating the rates of cooling which are necessary for the higher 

 portions of the statical temperature curve. They were not made until all the 

 normal experimental work was over, as they involved some risk of the safety 

 of the thermometers employed. 



II. Deduction or Conductivity. 



Statical Curve. — Although the manner in which this curve was constructed 

 was similar to that in previous work, it may yet be well to describe it in detail. 



From among all the statical experiments on the metal bar for which the 

 curve was required, that one was chosen which, in its approach to, and 

 maintenance of, the final thermal distribution, appeared to be the most steady, 

 and when considered in all respects, the most successful. This was termed 

 the Standard Experiment. The readings taken during the steady state of the 

 bar were then examined, and one special set of readings, apparently the most 

 trustworthy, was selected. The appropriate corrections having been applied, 

 the true temperature excess at each hole in the bar was obtained by subtracting 

 from the corrected reading the reading taken simultaneously of the temperature 

 of the short bar. The series of numbers thus obtained were then laid down as 

 points on a curve, in which the abscissae represented distance along the ex- 

 perimental bar, and the ordinates denoted temperature excesses. The other 

 experiments on the same bar were then taken in the order of their apparent 

 merits, regarded in the same way as in selecting the standard experiment. 

 The temperature excesses were plotted separately for each experiment, on a 

 sheet of tracing paper superposed upon a blank part of the sheet of divided 

 paper on which had been marked the points obtained from the standard 

 experiment. The axes on both sheets were coincident. After each experiment 

 had been so represented on the tracing paper, it was carefully drawn along, 

 keeping the horizontal axes still coincident, over the sheet beneath, until the 

 points marked upon both lay in one smooth curve. Each point on the super- 

 posed tracing paper was then transferred to that, immediately covered by it, on 

 the sheet below. In this manner a series of points was obtained, through 

 which a smooth curve was drawn, exhibiting the relation existing between 

 temperature excess and position along the bar — what Fokbes termed the 

 statical curve of temperature. 



