CHEMICAL ANALYSES OF TOBACCO. 279 



retained on the sieves were severally dried, ignited and weighed, and the finest portion, or that passing through the 

 6.25 mm sieve was then submitted to the following process of separation, which is based upon the method of Knop, 

 as described by Wolff (op. cit., p. 10), but perfected by the employment of the principle of fractional separation. 



The sediment and water passing through the 0.25"" sieve were placed in a glass cylinder 53 centimeters long 

 and 37 mm in internal diameter. The cylinder was closed at the bottom and was provided with a lateral tube 

 inserted 6 centimeters above the bottom. Three other lateral tubes were inserted at intervals of 10 centimeters 

 above the first tube, and a ring was etched into the cylinder 10 centimeters above the uppermost tube. The lateral 

 tubes were closed with rubber tubes compressed by spring clips. The sediment being placed in the cylinder, water 

 was added to the mark or ring, the cylinder closed with a rubber stopper, and vigorously shaken until the contents 

 were thoroughly mixed. It was then placed upright, the stopper removed, and after standing undisturbed for five 

 minutes the clip on the uppermost tube was opened and the water allowed to flow into a beaker. After five minutes 

 further standing, the second clip was opened and the water drawn off into the same beaker; in the same manner 

 the water was diawu off from the other tubes* at intervals of five minutes until the level of the lowest tube was 

 reached. The cylinder was then refilled with water to the mark, thoroughly shaken after inserting the stopper 

 and the water again drawn off at intervals of five minutes, as before ; the operation being repeated until the water 

 drawn off was almost free from turbidity. The sediment remaining in the cylinder from this process of washing 

 by subsidence is termed by Kiiop "fine sand", the material flowing off in suspension in the washing waters 

 " dust", and the process of separation by Knop's original method ends here. 



A little reflection will show that as far as a separation into particles of definite size is concerned, the foregoing 

 method is very imperfect. While the repeated agitation of the sediment from each washing with fresh additions 

 of water tends to remove the objection noted by E. W. Hilgard (Silliman's Journal, 1873, [3] vi, 288, 333) in his 

 admirable paper on silt analysis, and to which all methods of continuous washing known previous to the invention 

 of the methoM of Hilgard were subject, viz, the tendency of the particles of clay to ball together and sink with the 

 (parser particles even in opposition to upward currents of water, it is nevertheless attended with the following 

 detect to which Hilgard's method is also liable, and which, as far as the author is aware, has not hitherto received 

 sufficient attention. In the case of a quiescent body of water containing suspended sediment, the particles of the 

 latter sink to the bottom with varying degrees of rapidity, proportional to their size and specific gravity, or, as the 

 mineral constituents of soils do not, for the most part, differ very greatly among themselves in specific gravity, 

 chiefly proportional to their size. In course of their descent, however, the more swiftly descending particles collide 

 with the more slowly descending, or, as we may say, " lighter" particles. The rate of descent is therefore influenced 

 not only by the relative size of the particles of suspended matter, but also by the number and force of these 

 collisions. These two last-named conditions vary according to the difference in size (and specific gravity) of the 

 particles and also according to the relative number of the particles of each different size. It is evident, therefore, 

 that the error arising from the cause just mentioned is not constant, but that it must vary in amount according to 

 the specific character of each sediment, and that the results of separations by simple subsidence can never afford 

 a basis for an exact comparison in the matter of the state of subdivision of sediments of diverse origin and differing 

 in the relative proportions of the particles of the different sizes. The conditions above instanced as obtaining in 

 the case of a subsidence from a liquid in a quiescent state, obtain with equal force in the case of subsidence from 

 slowly moving upward currents, and undoubtedly constitute a source of material error in all methods of mechanical 

 analysis by continuous subsidence. 



In the method pursued by the author the defect above noted was obviated by the following means : 



The " fine sand " from the first series of subsidences was placed in a separate vessel, the washings were allowed 

 to remain undisturbed for twelve hours, the turbid liquid decanted off, and the sediment returned to the cylinder. 

 Water was then added to the mark, the whole shaken up, and the liquid drawn off at intervals of five minutes, as 

 in the first series. The sediment from this operation was placed in a separate beaker, the liquid "washings" 

 returned to the cylinder, and a<rain subsided as before; the sediment from this second subsidence was added to 

 that from the preceding operation and the washings again returned to the cylinder, the operation being repeated as 

 long as any sediment could be obtained from renewed treatment of the washings ; the final washings were then 

 placed in a separate vessel for subsequent microscopic measurements. 



The collective sediments from the last series of operations were then returned to the cylinder and subsided with 

 fresh additions of water, as in the case of the first series ; the " fine sand " thus obtained being added to that from 

 the first series, and the washings being collected in a large beaker. The latter were left at rest for twelve hours, and 

 the sediment returned to the cylinder and treated as before until no further separation could be effected. The " fine 

 sand " resulting from all of these operations was then dried, ignited, and weighed ; the weight of the portion removed 

 by the washing being determined by difference, as it was, owing to its excessively slow rate of subsidence, found 

 impracticable to collect it for direct weighing. The size of the particles of " fine sand" was then determined by 

 micrometric measurement, and was found to vary from 0.25 mm to 0.009" average diameter. Similar measurements 

 were made on the material obtained by long subsidence from the washings from the foregoing operations, with the 

 result of showing that the average diameter of the largest particles did not exceed 0.01 m . 



