1884.] Energy and Radiation in Incandescence Lamps. 171 



production was effected by calculating the position of maxima from 

 the hyperbolas calculated in Table IX. It will be noticed that the 

 maxima rapidly recede towards the lower limit of refrangibility, and 

 at just invisible heat the maximum lies near 18,000. 



When a filament is at a constant temperature the energy required 

 to maintain it at such must of necessity be expended, either by 

 radiation, by heating matter in immediate contact with it, which in 

 this case is gas in a state of extreme tenuity, by conduction, or by 

 an internal loss.* If there be no expenditure of energy by anything 

 except radiation, it is manifest that this would increase in exact 

 proportion to the energy shown in the filament. Now we have seen 

 that the radiation is not thus strictly proportional to the expended 

 electrical energy, but that from the first we have energy unaccounted 

 for by radiation, which gradually increases up to a certain point and 

 then becomes constant. The question arises as to what cause this can 

 be due. A reference to fig. 2 will show that convection currents 

 either inside or outside the lamp are not the cause. We have in this 

 figure the analysis of the radiation for different parts of the spectrum, 

 and it will be seen that the rays of low refrangibility have curves 

 which are concave to the axis of abscissa?, whilst for the rays of 

 higher refrangibility the curves are convex to the same axis. Now 

 if convection currents caused this palpable concavity in the curves 

 from rays of low refrangibility, we ought to be able to measure their 

 radiation as total radiation from the lamp when the direct radiation 

 of the t filament itself is cut off. As a matter of fact the total 

 radiation from the globe and surrounding air was immeasurable on 

 the thermopile. Thus the radiation from the heated gas inside, 

 and the heated air outside the lamp, and from the glass globe itself, 

 were inappreciable. The former might, perhaps, be expected if the 

 lamp-black with which the face of the pile was coated did not absorb 

 the particular radiations emitted from them. If the total radiation 

 from the glass globe and its surroundings was not measurable, much 

 less could it be anticipated that the galvanometer needle would be 

 deflected when such radiation was distributed through the spectrum, 

 more especially as the rays falling on the prism proceeded from a very 

 narrow section of the lamp (including the filament) lying on a plane 

 passing through the slit and along the axis of the collimator. We 

 find, however, that in this case the curves of radiation of the rays of 

 low refrangibility are concave, whilst those of higher are convex to 

 the horizontal axis. This would imply, if the cause of the convexity of 

 the total radiation curve lies outside the filament (since it has been 

 shown that the radiation from the globe and surroundings is inap- 



* Any apparent loss in radiation due to the lamp-black on the thermopile not 

 absorbing radiation may be dismissed, as the loss will be shown to occur in the rays 

 of high refrangibility which it is known are absorbed by lamp-black. 



