K.—BOTANY 185 
2. Resetting —After discharge the trap resets itself after a period of 
from 15 to 30 minutes or more, in U. vulgaris, or as long as 2 hours, more 
or less, in U. purpurea. In this operation water leaves the interior of the 
trap by diffusion through the walls, until an equilibrium has been reached 
and the walls have become concave so much as to press tightly on each 
other (in U. purpurea), or at least to be closely approximated (U. vulgaris, 
U. gibba, U. cornuta). In other words, the trap acts as a cell when sur- 
rounded by a slightly hypertonic solution of a sugar or salt if harmless ; 
but in the case of the trap this condition is not necessary, though with 
sufficiently high solutions of sugar, etc., within, the trap will take up 
water rather than lose it. Czaja (1924) has studied this aspect of the 
physiology of the trap, and tells us that the walls are semi-permeable, 
allowing water to pass but not solutes, but evidently this is not the whole 
explanation. For our present purpose we need not discuss this problem, 
merely recognising the fact that water passes out from the interior of the 
trap, thus producing a reduced pressure within. As a result, the outer 
water presses equally everywhere—on walls and door alike. When, 
therefore, the door is shifted out of its position of equal resistance, the 
water pressing thereon pushes the door in. Thus is furnished a part of 
the energy required to actuate the trap. ‘The remainder (an amount not 
measurable, or at least not yet measured) is supplied by the tensions of 
the (living) walls themselves, which, with an even water pressure within 
and without, still always take on an extreme convexity, when the trap can 
be said to be completely relaxed. From this condition a sound, undamaged 
trap will reset itself in a period considerably longer than that required 
after normal actuation when it is only partially relaxed. 
The energy required for the actuation of the trap is derived from the 
water pressure plus the outward spring of the walls. From now on, this 
will be understood and no further reference made to it. 
3. Watertightness—Since the above is always true, it must also be that 
the door is watertight.2 I have shown (1929) that this watertightness is 
owing to the presence of a membrane, the velum (Figs. 21-25), which 
arises as a cuticular development from the pavement epithelium of the 
threshold, though in certain species other regions contribute to produce 
an accessory velum (Fig. 15), as we shall see. All the living species 
examined conform to these statements. How the door is engaged when 
the trap is in unstable equilibrium is a particular question, along with 
others, as to the extent and proportions of the threshold, origin and 
extent of the velum, and the method of actuation. These points, there- 
fore, are to be considered specifically in what follows. There is, however, 
one underlying fact which may be mentioned at once in this connection 
—namely, that the free edge of the door is always longer than the threshold 
at its inner angles (Fig. 23). The latter can be readily understood when 
the development of the trap is considered and, as Meierhofer (1920) has 
cleared this up, it is not necessary to further amplify. It follows that the 
door edge cannot lie smoothly along the surface of the threshold when 
* When the adjective ‘ watertight ’ is used, I imagine that it must not be taken 
too literally. As long as the inleakage is at a lower rate than the outward 
diffusion through the walls, the trap will work. 
