MEASUREMENT OF OSMOTIC PRESSURE 561 



from "thermometer effects" and "barometer effects" due to 

 small changes of temperature (rarely exceeding 0*02° C.) and to 

 variations of atmospheric pressure ; these effects were very 

 serious in the case of dilute solutions, which could only be 

 examined in cells provided with the newest and most active 

 membranes and in periods of steady barometric pressure. The 

 problem of constructing a manostat for use in these experiments 

 is under consideration. 



The most serious disaster in the whole course of the work 

 was an infection of the laboratory with Penicillium glaucum dur- 

 ing rebuilding operations on a lower floor, which necessitatedjthe 

 constant use of antiseptics during the whole of the subsequent 

 four years. The precautions used suggest the practice of a 

 bacteriological rather than of a chemical laboratory. The two 

 germicides which were most effective in destroying the spores 

 without injuring the membranes were thymol and gaseous 

 prussic acid ; all the solutions used in the measurements were 

 sterilised by the addition of thymol to o'ooi normal concen- 

 tration, saturated solutions being used for storing the cells 

 during the vacation. The mould seems to feed upon the 

 membranes. The first evidence of infection is usually the fact that 

 membranes which were previously rendering satisfactory service 

 show signs of leaking and fail to recover their fully semi-perme- 

 able character when resubjected to the membrane-forming process. 



5. The Measurements. — The main results of the measurements 

 are summarised in Tables I. and II. 



The upper part of each table shows the final measurements of 

 the osmotic pressure of cane-sugar solutions between o° and 25 

 as carried out in the years preceding 191 1. In this range the 

 ratio of osmotic pressure to " gas-pressure " is absolutely steady, 

 so that Gay Lussac's law may be applied rigidly. The observed 

 osmotic pressures exceed, however, the corresponding gas pres- 

 sures by an amount that ranges from 6 to 1 1*4 per cent. The ratios 

 in the case of the decinormal solutions rise in a somewhat 

 surprising manner and there is a further remarkable rise when 

 solutions of this concentration are examined at o c ; as this is 

 within o - 2° of the freezing-point of the solution it is possible 

 that the effect is in some way due to the polymerisation of the 

 solvent. 



The lower part of the two tables shows the measurements 

 that have been made during 191 1 at temperatures above 25 



