THE SOURCES OF THE NITROGEN OF VEGETATION, ETC. 479 
just referred to, then the water must flow out of A through the safety tube g, 7, s before 
it can pass through the tube ¢, d, e, into the bottle B. In accordance with this con- 
sideration, the safety-tubes for the apparatus of 1857 were arranged as shown in 
Plate XIII. (g, 7, s); but, owing to occasional leakage of the joints at the top of the 
vessel A, this principle could not be relied upon; and hence the arrangement shown in 
fig. 1, Plate XIV. was adopted in the experiments of 1858. 
The height from the top of the vessel A to 7’ (fig. 1, Plate XIV.) was 12 inches, which 
is sufficient to allow the whole of the air to pass out of the vessel A, whilst the great 
height of d’, of the tube ¢’, d’, e, entirely prevented the water from passing over into the 
bottle B,—an accident which unfortunately happened on a few occasions with the appa- 
ratus of 1857. 
When the vessel A was full of water, it was drawn off by a cork-hole at the bottom, 
air being at the same time admitted by the tube # at the top of the vessel. 
The minimum pressure upon the glass shade F would be 
vox 7S? 0-136 inch, 
in which 1-0 is the difference between the height of the lowest and the highest level of 
the sulphuric acid in the bulb-apparatus M. Experiments showed, however, that owing 
to friction, &c., the maximum pressure on the inside of the glass shade would be raised 
to double the above estimated minimum. 
The plants were supplied with water, as already said, through the tube wv, shown best 
in fig. 2, Plate XIV. ‘At first fresh distilled water was supplied; but as soon as a suffi- 
cient quantity of condensed water had run through the tube m'o’ and collected in the 
bottle O, this was drawn off by means of the tube 7’ to water the plant when required. 
In the experiments of 1857, the condensed water was drawn off from time to time, from 
the surface of the slate and mercury, by means of the tube no. 
All the Woulfe’s bottles were made as air-tight as possible by means of very good 
corks. Those of the bottles E, Plate XIII., and fig. 1, Plate XIV., and also those of O, 
fig. 1, Plate XIV., were, however, covered with a cement, composed of eight parts gutta 
percha, twelve parts common rosin, and one part Venice turpentine, well melted toge- 
ther. The glass tube mn! o' was also fixed into the lute-vessel ww, at mn’, with this 
cement. 
In the experiments of 1857, tubes of unvulcanized caoutchouc, made by ourselves 
from the sheet, were used for the various joints indicated in the figures; but as these 
soon became unsound under the influence of the atmospheric changes to which they 
were exposed, tubes of vulcanized caoutchouc were substituted in 1858. The ends of 
the glass tubes o and w, Plate XIII., and of the tubes ¢’ and wu, fig. 1, Plate XIV., were 
fitted with pieces of caoutchouc tubing into which pieces of solid glass rod were fixed 
as stoppers. 
In 1857, twelve such sets of apparatus, and in 1858 a larger number, were employed. 
The whole were arranged side by side, on stands of brickwork erected for the purpose, 
MDCCCLXI. 3 U 
