10 
Walter Stiles 
osmotic pressure as due to the movement of liquid through the 
capillary spaces of the semi-permeable membrane from the liquid 
of lower surface tension to that of higher surface tension. Modifica¬ 
tions of the theory have been proposed by I. Traube (1904) and 
Battelli and Stefanini (1905, 1907 a, b, c) a presentation of which is 
outside the scope of the present work. For an adequate discussion 
of the views of these and other writers the reader is referred to 
Findlay’s monograph already cited. 
Negative Osmosis 
In all the cases so far discussed in which a solution is separated 
from the solvent by a membrane, there results a passage of liquid 
across the membrane from solvent into solution. It is now necessary 
to turn to a phenomenon which appears anomalous and inexplicable 
by the laws so far considered. This is the phenomenon of negative 
osmosis characterised by the flow of liquid across the membrane 
from the solution to the pure solvent. 
Negative osmosis was observed by Dutrochet (1827) and later by 
T. Graham (1854). The phenomenon has recently formed the subject 
of a series of investigations by Girard, working with membranes of 
pig’s bladder, and by Bartell and his collaborators, who employed 
membranes of porcelain and gold-beaters’ skin. The results of these 
investigations seem likely to have important bearings on questions 
of cell permeability, so that some detailed consideration of them 
appears necessary. 
Bartell (1914) first worked with membranes of porcelain about 
5 mm. thick, the pores of which were about 0*2 [i in diameter. This 
diameter of pore is on the border-line as regards osmotic effects; 
with somewhat wider pores osmotic effects are not produced, while 
with narrower pores the osmotic effects are definite. The membranes 
are thus decidedly permeable to dissolved substances. 
Osmotic cells were prepared in which the separating membrane 
was porcelain, and the cells were filled with solutions of the salts 
examined. These included chlorides, nitrates, sulphates and acetates 
of a number of monovalent and divalent kations in concentrations 
of M/10 and, in the case of sulphates, M/$ and M/ 2. In all the experi¬ 
ments in which potassium and ammonium salts were employed 
passage of liquid took place in the normal direction, that is, from the 
solvent to the solution (positive osmosis), with sodium sulphate and 
sodium acetate positive osmosis also occurred, as it did also in experi¬ 
ments with lithium sulphate; on the other hand, the chlorides and 
