116 Scientific Proceedings, Royal Dublin Society. 



explanation may be easily realized in a simple model. Two 

 similar semipermeable membranes are deposited in two pieces of 

 vegetable parchment. This may be conveniently done by soaking 

 the two pieces of parchment, first in gelatine and afterwards, when 

 the gelatine has set, in a solution of tannin. One of the parch- 

 ments is now spread loosely over the top of an ordinary thistle - 

 head funnel, and its overlapping edge is bound tightly round the 

 rim of the funnel. In order to make this junction water-tight, 

 the outside of the rim is coated with strong glue before putting on 

 the parchment. The second membrane is similarly bound down 

 and glued on the rim, in such a manner as to enclose a space 

 between it and the first. Before closing this space some dry sugar 

 is placed on the lower membrane. After setting, the glue is ren- 

 dered insoluble by an application of tannin. By this arrangement 

 the funnel is closed by a lenticular cell (with more or less semi- 

 permeable walls), containing sugar. The funnel is now filled 

 with water, and set upright, and water is supplied to its lower 

 end. 



Thus arranged it will soon be noticed that the cell above 

 becomes turgid ; and in becoming so, it of course draws up 

 water through the supply tube. Even after the cell has attained 

 its maximum distension, water still continues to rise in the tube, 

 owing to an action which will be explained later. The upward 

 current may be made apparent by supplying water to the fun- 

 nel through a fine capillary tube, and by arranging that the 

 water so supplied will contain a fine sediment in suspension ; its 

 motion may be observed by means of a microscope (see figure, 

 p. 115). Such a model may be kept in action for several days. 

 It will, apparently, only stop when leakage of the sugar back 

 through the lower membrane makes the liquid below isotonic with 

 that of the cell. 



The action of this actual model does not resemble precisely 

 that of the ideal model with the theoretically perfect semiperme- 

 able membrane. This may be seen from the following considera- 

 tions : — At the same time as water is passing into the cell from 

 the funnel, sugar solution is leaking back into the funnel, and 

 exuding on to the upper surface of the upper membrane, owing to 

 the fact that the membranes are not perfectly semipermeable. 

 This leakage may be directly observed in the streams of more 



