1 -6 Journal of Agricultural Research \o\. xvii, .\o. 4 



It will be noted that duplicate check counts were made generally in 

 cells of varying capacity, giving the maximum numerical and percentage 

 variation and indicating as well the supporting value or colloidal capac- 

 ity of each type of bitumen, based on that of refined Trinidad 'asphalt 

 considered as 100. All results were computed on a basis of i to 5,000 

 dilution of the original colloidal portion of the sample. (See p. 175.) 

 These bitmuens have been separated into groups containing clay as col- 

 loidal material (No. 1-19) and into others in which this mineral matter 

 was replaced by carbonates, sulphates and acetates of copper, iron, 

 zinc, and lead (No. 20-47). The samples included in the first class, 

 except untreated refined Trinidad asphalt (No. 1-5), were prepared by 

 incorporating 33 per cent sandy clay in each type of bitumen by Richard- 

 son's method of heating an aqueous emulsion of clay and bitumen until 

 all moisture and gas had been expelled.^ In the case of refined Trinidad 

 asphalt and clay (No. 6-10) all insoluble mineral and organic matter 

 originally present in the bitumen was removed before emulsifying by 

 dissolving in benzol and adding about 2 per cent shellac dissolved in 

 alcohol and evaporating to constant weight, redissolving in benzol and 

 filtering through an alundum tube until the solution was essentially 

 void of colloidal particles. 



In the second group of bitumens (No. 20-47) the salts were intro- 

 duced in an anhydrous condition and the mixtures were heated to about 

 170° C. under constant stirring until all evolution of gas had ceased. 

 On examining solutions of this kind under the ultra-microscope it was 

 found that the copper carbonate salts had been largely reduced to red 

 cuprous oxid, accompanied by an enormous colloidal dispersion (No. 

 20-30), while with the remaining salts the reduction had been much 

 less complete (No. 31-39) or entirely lacking (No. 40-47), and the de- 

 velopment of colloids correspondingly less. It may be stated, there- 

 fore, that the colloidal capacity of the second group of materials taken 

 as a whole was dependent largely upon the degree of chemical reaction 

 between the bitumen and the salts employed, while in the first group 

 this supporting value was related more directly to the physical char- 

 acter of the bitumen. 



A comparison of the duplicate counts recorded in columns 3 and 4 of 

 the table indicated that a maximum variation of less than 10 per cent 

 was attained in samples of the first group (No. 1-19), while in the second 

 group (No. 20-47) the results were, on the whole, less concordant, owing 

 largely to the greater dispersion of colloidal matter. 



In conclusion, it may be stated that the accuracy of this method for 

 counting colloidal particles in bituminous solutions depends chiefly 

 upon accuracy in construction and calibration of the cell employed, as 

 well as upon the proper consistency and optical purity of the support- 

 ing liquid. 



' Richardson, Clifford. 1917. op. cit. 



