THE REGULATION OF TEMPERATURE. 55 



of the solvent. Evidently the magnitude of the so-called extra excur- 

 sion will depend very much upon the rate at which the solvent can 

 pass through the membrane in either direction. 



The magnitude, and therefore the importance, and the peculiarities 

 of thermometer effects depend upon several conditions which will be 

 briefly recapitulated. They are : 



1. The relative volumes of the liquids (solution and mercury) and 

 of the gas in the cell. Since the latter is always very small as compared 

 with the former, slight disturbances of temperature must always pro- 

 duce large thermometer effects. 



2. The degree of the "lag" in the passage of solvent through the 

 membrane, which, in turn, depends upon temperature and the area 

 and age of the membrane. 



3. The rapidity with which the changes in temperature are accom- 

 plished. The relation of 3 to 2 is self-evident. 



4. The "lag" in the distribution of solvent through the solution by 

 diffusion, which produces temporary conditions of non-homogeneity 

 in respect to concentration. 



It is obvious that, in one sense, we could have no thermometer effects 

 if the passage of solvent through the membrane and its subsequent 

 uniform distribution by diffusion were instantaneous, since, in that 

 case, there could never develop in the cell a condition of over or under 

 pressure. In other words, we should then have at all times simply 

 the osmotic pressure of a solution whose concentration varies with the 

 temperature. The fact that diffusion does not quite keep pace with 

 transferences of solvent through the membrane is not a source of 

 serious trouble, but in the lag of such transferences behind fluctuations 

 in temperature we have a most formidable obstacle in the way of the 

 accurate measurement of osmotic pressure. The only remedy for this 

 unfortunate situation is to be found in the most perfect means which 

 can be devised for the automatic maintenance of constant temperature. 



It will be gathered from what has already been said that the duration 

 of thermometer effects also depends principally upon the rate at which 

 the solvent is able to diffuse through the membrane. In practice it 

 is found that, according to the age of the membranes, they may last 

 from 12 hours to 4 days after the bath has recovered its normal tem- 

 perature. As regards the minimum temperature change which will 

 give a sensible thermometer effect, it may be said that a fluctuation 

 of 0.01° produces a movement of the mercury meniscus which can be 

 detected. A change in temperature amounting to 0.05° gives a large 

 thermometer effect, even when the membrane is new. It has not been 

 found practicable to regulate the temperature of the large baths which 

 are in use to within less than 0.01°; accordingly the meniscus is con- 

 stantly moving within narrow limits, with the result that two successive 

 readings, several hours apart, are rarely quite identical. Fluctuations 



