ULTRAMICROSCOPY 79 



puscles, Perrin obtained a liquid in which the particles were 

 ultramicroscopic, and could no longer be seen with ordinary 

 illumination. The particles do not show any scintillations, 

 and, as will be seen later, are very probably spherical. After 

 dilution a small quantity of the emulsion was placed in a 

 microscopic preparation, the thickness of which was fixed and 

 equal to about o'i2 mm. 



Distribution of the Granules. — The distribution of the granules 

 in this liquid was carefully studied, an interval being allowed 

 in order that a steady state might be reached ; and Perrin 

 found that the weights of the particles caused them to con- 

 centrate in the lower horizontal layers ; and as we ascend in 

 the liquid the concentration of the granules diminishes. 



In a freshly prepared solution the particles were observed 

 to fall towards the bottom of the solution at a measurable 

 rate. The concentration of the granules in these lower layers 

 increased until, eventually, a steady state was arrived at, when 

 the concentration varied with the level in the liquid in a very 

 striking manner. 



The number of particles present in a given volume — that 

 of the field of the microscope — was determined at various 

 levels, differing from one another by about 25 microns. This 

 process of counting is extremely tedious, and Perrin found it 

 necessary to make several thousands of observations. Owing 

 to the movements of the particles the chance of any obser- 

 vation being correct is not very great ; and it is only by 

 taking a large number that an accurate value can be ob- 

 tained. 1 



Below are given the concentrations, in arbitrary units, of 

 the granules at heights differing from one another by 25 p, 

 the concentrations being given in the second line. In the 

 third line are given the values these would have if the variation 

 occurred in geometrical progression. 



Height in n 



Number of particles . 



Number, if variation were in geometrical ) 

 progression ) 



It will be observed that the experimental values agree, 



1 During the process of counting the number present in the given volume 

 may change, some particles entering or leaving ; however, by taking a large 

 number of observations this can be allowed for. 



