BY H. S. SALCKO WARDLAW. 623 



pears to exist between the (lepressioii of conductivity due to 1 c.c. 

 of fat in 100 c c. of inilk and the average number of globules 

 into which this amount of fat is divided, or the average size of 

 the globules. The degree of subdivision of the fat thus does not 

 seem to influence its effect on the conductivity in excess of the 

 effect due to the volume displaced. 



4. The relation between the increase of inscosUij due to the fat 

 of milk and its effect on the electrical conductivity. —The electrical 

 conductivity of a solution of electrolytes depends on the velocity 

 and concentration of its ions. One factor which determines the 

 velocity of the ions is the resistance which the liquid of the 

 solution opposes to their movement. In the case of aqueous 

 solutions of salts only, this resistance is measured by the vis- 

 cosity. In the case of systems which are not homogeneous, how- 

 ever, the viscosity of the system as a whole is not a measure of 

 the resistance offered to the passage of ions. The addition of 

 gelatine to a salt solution, for example, may increase the vis- 

 cosity enormously, while the resistance to the movement of ions, 

 as measured by the conductivity, is hardly afiected. The system 

 is no longer homogeneous, but consists of two phases. 



Milk is such a heterogeneous system, and contains at least 

 three phases. No simple relation is to be expected, therefore, 

 between the viscosity of milk as a whole and its electrical con- 

 ductivity. It was thought, however, that these two properties 

 might be connected in some regular manner which would explain 

 the disparity between the volume occupied by the fat of milk 

 and its efiect on the conductivity. 



In the following Table are shown the effect of the removal of 

 measured amounts of fat on the conductivity and viscosity of 

 milk. The viscosities were measured in an Ostwald viscosimeter 

 at 25''C. The values given are those compared with water as 

 unity. The results have a maximum relative error of 1%. 



