MECHANICAL ANALYSES 



75 



method of certain Carnegie bottom samples 



(Size grades in microns) 



15.6-7.81 7.81-3.91 3.91-1.95 



1. 95-. 977 



.977-. 488 



.488-. 244 



Remainder 



Total 



4.56 



10.84 



17.02 



19.60 



15.22 



8.7 

 14.7 



12.4 

 (14.2) 



13.7 



7.6 



12.4 



12.06 



19.32 



100.03 



"3. Constant and uniform temperatures so that no 

 convection current or other disturbances occur during 

 sedimentation. 



"4. A sufficiently dilute concentration so that the 

 particles may fall independently and that the density of 

 the fluid displaced by the particles is not appreciably 

 different from that of water." 



To the above assumptions should be added two more, 

 namely: 5. that the sediments consist of "ultimate par- 

 ticles of definite size, no matter what their actual di- 

 mensions, and that analysis of the suspension dispersed 

 into these ultimate sizes leads to one and only one dis- 

 tribution curve" (Gripenberg, p. 56). 



6. The density of the settling particles is known. 



There are several sourcesof error which affect the ac- 

 curacy of the mechanical analyses. These are as follows: 



1. Errors of collection and preservation of samples. 

 The random nature of the bottom samples obtained with 

 snappers, and the great distance between the samples, 

 means that the representativeness of such materials 

 with regard to the bottom deposits of any given area is 

 unknown. Besides this essential limitation of the sam- 

 pling technique, the mechanical composition of the ma- 

 terial collected may be altered by washing while it is 

 being hauled to the surface and by alterations during 

 storage. 



2. Errors of dispersion. The purpose of dispersion 

 is to destroy all aggregates of ultimate particles without 

 affecting the size of the particles themselves, but no 

 criteria exist to indicate when this process has been ac- 

 complished. Although it is often possible to obtain re- 

 producible results with any one method of dispersion, it 

 is known that the addition of various peptizing agents re- 

 sults in the production of differing sedimentation curves 

 (see Galliher, 1933 and Richter, 1931). The subject is 

 exceedingly complex, but it is probable that the concept 

 of ultimate particle size should be abandoned and that a 

 statement of the distribution curve of colloidal particles 

 in a sediment is valid only for a given set of base ex- 

 change ions in the outer layer of the colloidal micelles, 

 and of dissolved ions in the dispersing medium, as sug- 

 gested by the work of Gedroiz (1931) and Mattson (1929). 

 The most satisfactory procedure of dispersion, then. 



would be that which maintained as closely as possible 

 the original set of base exchange ions in the colloidal 

 micelles. 



3. Errors caused by the initial current resulting 

 from stirring in the settling tube. That this error is 

 rather pronounced is shown by the relatively large vari- 

 ations found in the 62.5- to 3l.3-micron size grades of 

 the multiple analyses given in table 25. According to 

 Fisher and Oden, the error due to such initial currents 

 becomes negligible only after 53 minutes. 



4. Convection currents in the settling tube, due to 

 changes of temperature during prolonged settling, result 

 in mixing and therefore in an apparent increase of the 

 very fine grades at the expense of the maximum sizes ' 

 affected by the mixing. The magnitude of this error is 

 unknown, but it probably amounts to several per cent, as 

 differences in temperature between the walls and the 

 center of the settling cylinder of even a few hundredths 



of a degree probably give rise to currents equal in mag- 

 nitude to the settling velocities of particles less than a 

 fraction of a micron in diameter. 



5. The difficulties attending the calculation of par- 

 ticle size from observed settling velocites already have 

 been discussed in part. In addition to the fact that the 

 particles are not spherical, and hence do not exactly 

 obey Stokes' law, differences in the densities of the set- 

 tling micelles due to (1) variations in the specific gravi- 

 ty of the mineral particles and (2) to the varying amounts 

 of water adhering to them, give rise to inaccuracies in 

 the determination of equivalent diameters. An examina- 

 tion of the data given in Dr. Nutting's report on the 

 mean grain densities of the samples shows that these 

 vary about 0.1 unit in any given type of deposit. From 

 figure 10 it may be calculated that at 20° and at a densi- 

 ty of 2.7, an error of 0.2 in the determination of density 

 results in a 6 per cent difference in calculated diameter 

 and a percentage error of 1.5 per cent in the distribution 

 curve of the samples analyzed. Changes of temperature 

 result in changes in the viscosity term of Stokes' law 

 and hence also affect the determination of particle size. 

 It may be calculated from figure 10 that with a grain 

 density of 2.7, a change in temperature from 25° to 20° 

 results in an error exactly similar to an error of 0.2 in 



