242 The Marquis of Salisbury on Spectral 



the electric action on the thermometer is of the same kind as 

 the action upon the conductor. 



But whatever the process of electric excitement in the ther- 

 mometer may be — whether by induction or by conduction 

 through leakage — the exhibition of light without any but 

 the minutest development of heat is worthy of notice. The 

 heat caused by a full discharge through Geissler's tubes is 

 well known. It will, as Wu'llner notices, disintegrate the glass, 

 producing sodium-, and at last even calcium-lines. If increased 

 by resistance of gas of 500 millims. pressure, it will soften and 

 bend aluminium electrodes. But in the thermometer, where the 

 resistance is that of the thickness of a tube of glass, scarcely any, 

 if any, rise of temperature is produced. With thermometers 

 of the ordinary bore, I have failed, after many trials, to dis- 

 cover any alteration of reading at the end of five minutes' expo- 

 sure to the discharge of the coil in the manner I have described. 

 Four experiments with a thermometer of very fine bore, gradu- 

 ated to tenths of a degree Fahr., have given an average rise of 

 three quarters of a degree Fahr. in the course of five minutes. 

 I am doubtful whether even this rise is to be attributed to heat, 

 and for this reason. In two other thermometers a small portion 

 of the column of mercury was separated from the rest. The 

 action of the coil after a short time made the interval wider, 

 pushing up the separated fragment without expanding either 

 portion ; and in the interval so created a brilliant light ap- 

 peared. It is evident, therefore, that the coil tends to produce 

 a motion in the mercury of the thermometer analogous to that 

 observed by Poggendorff in larger tubes ; and the slight appa- 

 rent rise of three quarters of a degree in five minutes, produced 

 by the coil in a thermometer of very small bore, may be due to 

 an action of this kind. But even if there be a real rise of tem- 

 perature to this extent, it is so minute as to be practically unim- 

 portant. In effect, therefore, the light was produced at a tem- 

 perature of less than 60° F. 



This light is strong enough to be examined in the spectro- 

 scope. One prism, of course, only must be used; and the room 

 must be perfectly dark. It is also necessary to use cross-wires 

 in the telescope, as cross-hairs are too slender to be seen in so 

 dim a light. Of course much the easiest mode of examination 

 is to dispense with a slit altogether ; but on that plan the lines 

 of the spectrum, in some thermometers at least, are so rugged 

 that it is difficult to identify them. I have therefore been com- 

 pelled to use the slit. To my surprise, I found that different 

 thermometers gave different lines ; or rather, some gave more lines 

 than others. Instruments made by the best makers, such as 

 Casella or Elliott, give only the three following lines. All the 



