Calibration of the - 5 - D. M. Poole and 



Emeiy Settling Tube W. S. Butcher 



for Sand /malysis 



tlriis the probable error (P.E. = 0.6745 <r ) . For further comparison between 

 different tests it is necessary to have the probable error expressed non- 

 dimensionally. The non-dimensional form was obtained by expressing the proba- 

 ble error as a percentage of the median diameter. 



As shown by Kirunbein (1934-), the probable error may be separ- 

 ated into any niimber of component errors. This separation is shown by the re- 



yo 2 2. 2 



e-] + eo + eo •••Sn ' Since the probable errors are ex- 

 pressed in per cent (see above), they are non-dimensional ajid apply to any 

 test. The total error (E) in the experiments cjaalyzed in this po.per corres- 

 ponds to the "laboratoiy error" of Knimbein's paper. The "sampling error" of 

 the cited paper has been eliminated by using one bulk sample split into the 

 desired number of portions. The total error (E) has been divided into split- 

 ting error and settling tube error. 



Settling Tube Error ; If the same sample is run through the settling tube 

 a nuiiiber of times, the probable error of these runs is due to the errors in 

 running the saiaple through the tube, observational errors, and errors in 

 timing. These may be considered as the error of the settling tube itself or 

 of the method, since splitting and weight-of-sample errors do not enter. Hence, 

 the same sample was run through the settling tube 10 times and the results 

 gave a percentage probable error of 0.8. For comparison the same sample was 

 sieved 6 times, using a different split of the sand tested in the settling tube, 

 Here the percentage probable error was 0.7, and thus is approximately the same 

 as that obtained in the settling tube. It should be noted that the third 

 quartile shows a greater spread for the sieve analysis thrji for the settling 

 tube (see figs. 2 and 3). The median diameters from the settling tube average 

 0.131 mm. and from sieving, 0.136 mm. Although the sand was a reasonably clean. 



