1870.] 



Ardurus and a Lyra. 



161 



ten minutes. In order to avoid changes of zero, I have always reduced 

 those readings by comparing a reading with star on face a with the mean of 

 two readings with star on ft taken before and after the reading with star 

 on a, or vice versa. 



With this precaution I have never met with any anomalous results, 

 although in making the observations I have usually joined over the ter- 

 minals, without knowing the direction for heat, and have left this unde- 

 termined until the completion of the observations. I mention this because 

 the differences in the readings for star on a and star on /3 in the state in 

 which I use my galvanometer are small. 



On many nights, when very slight appearances of cloud prevailed, I have 

 not been able to make any satisfactory observations at all. 



The number of divisions over which the spot of light travels on the 

 galvanometer-scale for a given difference of temperature of the faces a and 

 /3 is of course dependent upon many circumstances, and especially upon 

 the position of the sensitiveness-regulation magnet of the galvanometer. 



I have thought it useless, therefore, to publish any results unless obtained 

 upon nights when the state of the galvanometer was eliminated by re- 

 ferring to an independent source of heat. The way in which this has 

 been attempted is as follows : — 



After obtaining the differences in the position of the spot of light on 

 galvanometer-scale for star on a and star on ft I remove the pile from the 

 telescope, leaving all its galvanic connexions untouched, and mount the 

 pile so that of the two halves of the face of a Leslie's cube, containing boiling 

 water, each radiates heat upon one face, a or /3 of the pile, placed at a 

 known distance of about twenty inches from the cube. After some time 

 the deflection of the needle will fall nearly to zero, and become steady 

 enough for observation. A piece of glass, G, is then placed to intercept 

 from p a portion of the heat radiating from one half of the face of the 

 cube, and when the needle has taken up its position, the reading is taken. 

 Next the glass G is placed to intercept a portion of the heat from the 

 face a, and the galvanometer-reading taken, as before, as soon as the 

 needle has assumed its position of rest. 



If, then, 0 is the mean difference of readings for star on face a and face 

 /3, tf, the mean difference for glass before ,3 and a, C the heating-power of 

 each half of the cube at its distance from the faces of the pile, and p the 

 measure of the absorption of the piece of glass G, then the heating-power 

 of star 



=~xCXp. 

 9 



The quantity p has been determined by merely comparing the readings 

 of the galvanometer, obtained by cutting off the whole heat from one-half 

 of the cube, with that obtained by intercepting a portion of this heat by the 

 glass G. A considerable number of accordant results gave p = 0*725. 



To determine the quantity C, I have proceeded as follows : — 



