98 
very slow when the lower end of the hollow carbon was closed. When a 
gas was blown into the globe the admission of air was prevented by the 
escaping gas. 
By the term carbon bands will be meant those heading near 2—4737 
and 74—4382. The bands heading near 2 — 3883, 2 — 4216 and 2—3590 will 
be called the cyanogen bands, though’ the researches of Augstrém and 
Thalén, Lockyer, Liveing and Dewar, Kayser and Runge, Crew and 
others are very conflicting on this question. 
When the spectrum was photographed immediately after starting the 
are it was identical with the spectrum of the open are. After ten min- 
utes the metallic lines had disappeared almost completely. The cyanogen 
and carbon bands remained strong even when O, CO. and S vapor were 
forced into the are and globe. 
It was thought that the cyanogen bands might have been due to the 
nitrogen of the air contained in the porous carbon poles. To remove the 
air the carbons were placed in a small air-tight iron cylinder, which was 
connected by tubes with an air pump, and a cylinder or CO,. The eylin- 
der containing the carbons was placed in a furnace and kept red hot for 
two hours; the air pump was worked continuously. The carbons were 
then allowed to cool in an atmosphere of CO., in which they remained 
for several days. The operation was then repeated and the carbons cooled 
as before. 
When the treated carbons were placed in the lamp and a continuous 
stream of CO, was passed into the are and globe, the cyanogen bands ap- 
peared to be a trifle weaker and the carbon bands a little stronger than 
with the untreated carbon poles. However, the difference was very slight. 
A like result was obtained when, instead of CO,, O and the vapor of S were 
passed into the are. A continuous stream.of S vapor was obtained by at- 
taching to the end of the hollow carbon a glass tube sealed at the lower 
end and filled with 8S. A Bunsen burner kept the S boiling vigorously, the 
vapor passing into the globe and are through the hollow carbon, which 
was heated by a second burner to prevent condensation of the vapor be- 
fore reaching the are. 
It appears that the weight of evidence favors the view that the three 
bands in question are intimately connected with the presence of N in the 
are. But there are reasons for hesitating to accept the conclusion that 
they are due to cyanogen or to any other compound of C and N. One 
