438 W. LeConte Stevens — Experimental 



sponding deflections at the usual intervals. To find out what 

 would be the residual deflection, if this temperature-difference 

 as indicated by the thermometers were reduced to zero, advan- 

 tage is taken of the fact that for such small differences the 

 curve expressing the radiation is approximately a straight line. 

 Taking therefore the recorded small temperature-differences 

 as abscissas and the corresponding deflections as ordinates, a 

 straight line is drawn through the points thus found. Pro- 

 longing the line in a negative direction until it cuts the axis 

 of ordinates, the distance of the point of intersection above 

 or below the origin gives the correction to be applied to these 

 final readings. The correction to the initial readings having 

 been taken before heating the disk, and the slight variation of 

 temperature in the box having been recorded, we have the 

 means of constructing at least an approximate curve of cor- 

 rections for all intermediate readings. 



It is scarcely necessary to add that all thermometer readings 

 were corrected for the difference of temperature between bulb 

 and tube, and that all scale readings were reduced to those 

 which would have applied to a circular arc. 



The disk employed in the majority of the experiments was 

 made of iron, though a number of measurements were made 

 also with one of copper of the same dimensions. The distance 

 from disk to thermopile was kept constant for each single series 

 of observations, but this distance was varied from 30 cni to 

 40 cm according to necessity, and the sensibility of the galvan- 

 ometer likewise varied, for different ranges of fall of tempera- 

 ture. The superior conductivity of copper ensures quicker 

 equalization of temperature throughout the disk of this metal, 

 but that of iron is amply sufficient, while its higher coefficient 

 of emission makes the experiments less tedious and less subject 

 to error than when a disk of copper is employed. The disk con- 

 stitutes a short cylinder whose radius is r — 7*3 em and whose 

 length is 2 I — 14 cm or the distance from center to flat surface 

 of the cylinder is I = - 7 cm . It can be shown* that if a ther- 

 mometer bulb be placed within such a cylinder of heated iron at 

 its center, and if it be assumed that a brief interval of time, such 

 as one minute, has elapsed, then between the center of the 

 disk and the nearest point on its face, at a distance of 0'7 cm the 

 absolute temperature does not vary more than 0*0002 of itself, 

 or 0*02 of 1 per cent; also that the temperature-difference 

 between center and circumference of the flat surface, an inter- 

 val of 7-'6 cm , does not exceed 0*18 of 1 per cent. It is safe to 

 assume therefore that no error so great as the necessary errors 

 of observation can be introduced by accepting the temperature 



* Lectures of H. F. Weber on the Theory of Cylinder Functions and their appli- 

 cation to the Problems of Physics. 



