286 
and one Grove was used. The discharge did not pass till 
the gauge was 0°70. The light was pale, greenish blue ; the 
envelopes of the negative ball not blue, but reddish, yet no 
trace of fluorescence. The fine dark bands were seen as before, 
and were visible when the contact was broken by hand, so as 
to give a single flash, which seems to show they do not depend 
on the succession of discharges. At 010 the bands were 
curved and broad, and the stream trumpet-shaped ; the nega- 
tive ball had its three envelopes, and round them a wide, faint, 
blue haze. At 0:05, the lowest which the pump could then 
give, the light was faint and wide, but still blue; and each of 
the menisci, which are now fully developed, was, with its in- 
terval, about 0°25 across. This vacuum conducted so well, 
that, though the negative ball was connected with the pump 
plate by copper 0'-2 by 0°15 section, a sheet of light passed 
round the sulphuric acid holder, and covered the plate with 
green sparks. 
‘<The total absence of fluorescence here is very striking, and 
I rather hastily concluded, that this property depends on the 
presence of free oxygen in the vacuum. 
“5. Oxygen, procured from chlorate of potassa and per- 
oxide of manganese, and passed through solution of potassa, 
was next tried; gauge at 008. The appearance differed little 
from common air; the light was equally fluorescent, and the 
only difference noticed was, that the green sparks at the nega- 
tive ball were more numerous and intense. After a while the 
light became greenish and the fluorescence less, then it got 
a peculiar copper colour, which soon passed off, and it became 
as at first, except that the dark bands became much more dis- 
tinct. Was this owing to the formation of ozone? On in- 
creasing the distance from 3! to 4'*75, the bands were scarcely 
visible in the middle, but reappeared towards the negative 
ball. 
‘«©6. I tried nitrogen prepared by leaving for four days in 
a confined portion of air a paste of equal parts of sulphur and 
