( 569 ) 
this tube was blown out in the middle to a small bulb, and was 
hermetically fixed to the stump with a piece of india rubber tubing. 
It was then half filled with water, and momentarily pumped empty 
that we may inject the cut vessels. I then left it open for half an 
hour and finally closed it with the perforated rubber stopper, through 
which the manometer was stuck. Once a day the bulb-tubes had to 
be replenished, for the wood always leaks a little from the inter¬ 
cellular spaces. The bark leaks still more and for this reason I always 
removed it at the place where the rubber tube was to come. 
As long as the tree was alive, no regularity could be perceived 
in the indications of the manometers: they all showed a pressure, 
smaller than that of the atmosphere, but sometimes one “sucked” 
more, sometimes another. After a few days I killed the portion of 
the branch bearing the manometers over its whole length by means 
of steam. At once the manometers followed the rule indicated above, 
and did not depart from it. The differences of pressure became very 
considerable towards midday, showing that the dead portion offered 
a great resistance to the strong current. 
The crown and the base of the branch remained intact during 
this treatment. The leaves showed only after 3 weeks, that they had 
suffered from the operation; up to that time they remained perfectly 
fresh. When at last they began to change, they gave the impression 
that they were diseased, rather than that they suffered from want 
of water. 
Two manometers were attached to the small trunk of a Comus 
and fixed to almost equal stumps of branches, the one 66 cm. above 
the other. The whole tree was 2 metres high. Before I cut off the 
branches, which were to yield the stumps, I killed the trunk at 
these two places with steam over a length of 10 12 cm. The 
manometers were thus attached to dead branch stumps on dead 
pieces of the trunk, separated by a living portion. 
I wished to investigate whether the living intermediate portion did 
pump or not. If so, it would always be occupied in diminishing the 
difference of pressure between the two dead pieces of the trunk. If 
it was then suddenly cooled with ice, the manometers would have 
to diverge suddenly and would once more approach each other if 
the tree was left to itself. Finally if it was killed, the well-known 
regularity would be bound to appear. 
The result was different, however. The intermediate portion evi¬ 
dently did not pump, for the manometers behaved exactly as in a 
dead tree. At midday they sometimes differed by 24 cm. of mer¬ 
cury. However — on the fifth day their behaviour changed fairly 
