of the Selenium Cell. 31 
marked. Not only were cells containing pure selenium much 
more sensitive than those containing impure selenium, but 
also more steady and prompt in tat. The latter, when 
exposed, required considerable time to reach a constant 
resistance, as was shown by the fact that the galvanometer 
would creep for half a minute or more before finally coming 
to rest. It was quite impossible to obtain concordant results 
by exposing the cell until creeping had ceased. However, 
readings agreeing within about 3 per cent. were obtained by 
exposing the slit at A automatically for 2°5 seconds and 
recording the maximum ballistic throw of the galvanometer. 
When a cell containing pure selenium was exposed to radi- 
ation, the spot of light of the galvanometer rapidly moved to 
a position of steady deflexion, which it maintained for a short 
time ; however, if the exposure continued, creeping would 
set in. As the actual amount of er eeping was small in com- 
parison with the first deflexion, this additional deflexion due 
to creeping was left out of consideration in taking measure- 
ments. Another very important reason why this should be 
done is that, while an exposure of less than 3 seconds was 
required to bring about the first deflexion, it became necessary 
to prolong this exposure eight or ten times until the creeping 
had ceased. When a cell thus exposed was again returned 
to darkness, it would take tive minutes or more for the 
resistance to reach its origina] value and occasionally a per- 
manent change in resistance would have been produced. In 
view of these facts an exposure of two to three seconds was 
given, just long enough to obtain the first deflexion. By 
proceeding i in this manner, the uncertainty of cr eeping was 
eliminated and the cell would return to its original resistance 
in less than one minute. 
Measurements. 
In determining the selective action of a selenium cell in 
different portions of the spectrum, the conductivity of the 
cell in darkness and its change in conductivity upon exposure 
to radiation were measured, By the ratio of the latter to the 
former of these two quantities, I wish to define the term 
“ sensibility.” As it is the purpose to measure the selective 
action of a cell, it is, of course, necessary that the energy of 
radiation falling upon the cell at different times be the same ; 
in other words, in the final plotting of results the only two 
permissible variables are sensibilite y and wave-length: energy 
must be constant. 
The actual procedure in obtaining results was the follow- 
ing :—The thermopile T (fig. 2) was brought into position 
