288 
and red, the light much broader and fainter; the menisci were 
seen only in its central part, and disappeared for 1'-5 of its 
middle; the fluorescence scarcely sensible. 
‘8, With carbonic acid, gauge 0-07; the positive ball was 
covered with orange-yellow light; from this sprung a faint 
lilac mass of light for 2', in which bright menisci show occa- 
sionally. Below is 1! dark; and below this, bluish haze in 
which is the negative ball with its red, dark, and blue envelopes. 
Fluorescence extremely feeble. 
‘«¢ Another experiment was made, intending to obtain a 
perfect vacuum on Dr. Andrews’ plan, by absorbing the car- 
bonic acid. The potassa solution, however, became solid when 
the gauge was 0':04. The positive ball was covered with the 
same yellow shade, from which issued a pear-shaped mass of 
livid lilac, with a brighter axis, the whole full of the menisci, 
but faint; it was in rapid rotation, and its point rose and fell, 
sometimes crossing the dark space between it and the steady 
spherical light surrounding the negative one. In this case also 
the fluorescence was scarcely sensible. 
‘* From these facts it appears that nitrogen has, as well as 
oxygen, the power of producing the invisible rays which 
cause fluorescence. Hydrogen, and compounds of it with 
oxygen and carbon, seem totally to want it; carbon is proba- 
bly in the same predicament, and it becomes an interesting 
question whether other highly electro-positive bodies resem- 
ble them. On the other hand, how will chlorine and its con- 
geners comport themselves? I could not try other gases for 
fear of destroying my air-pump, but an apparatus contrived 
by Mr. Bergin will, I hope, enable me to extend my experi- 
ments. It is like a mercurial gasometer, in which the bell- 
glass has a small opening at its top; the circumference of the 
aperture is ground flat, so that it can be covered by a flat 
disc of glass, slightly greased. Through the centre of this 
disc passes a platinum wire. When the bell is pressed under 
the mercury, all air escapes, especially if it be exposed for a 
