This report gives the results of a cross-testing experiment on the determination of particle size distribution involving 17 laboratories from 9 European countries. Three sands with differing gradings were used.
Generally, the reproducibility standard deviation of the determination of particle size distribution by sieving satisfies SR < 0.05 (X (100.0 - X))½, when either 200 g or 30 g test portions are used, so that the reproducibility will be satisfactory for the purpose of assessing the compliance of aggregates with specifications, provided that tolerances T in specifications for particle size distribution satisfy T > 0.30 (X (100.0 - X))½.
The number of mistakes made in the calculations, and the number of stragglers and outliers in the data, suggest that several of the laboratories are not familiar with the washing and sieving test. As this is the preferred method in cases of dispute according to the draft CEN method, it is clearly important that laboratories should be able to obtain reliable results when using it. It is therefore recommended that laboratories should consider taking part in proficiency tests regularly.
30 g test portions gave higher repeatability and reproducibility standard deviations than the 200 g test portions. Nevertheless, the use of small test portions would make the washing and sieving operations quicker, and reduce the risk of overloading of sieves, so it is recommended that their use should be considered further.
Further work is required to establish a functional relation for the repeatability of the sieve test. This would allow the factors that influence the repeatability to be understood properly so that the test could be designed to give adequate precision at minimum cost. Also, laboratories would be able to use it to check that they are achieving a satisfactory level of repeatability.
The results of the experiment allow the precision of the determination of particle size distribution by sieving of 200 g test portions to be stated (in units of % passing) as:
r1 = 0.042 (X (100.0 - X))½
R1 = 0.086 (X (100.0 - X))½
where "X" represents the average of the test results. The results of the experiment indicate that the reproducibility of the test will be much worse than this if sieves are overloaded, and both repeatability and reproducibility will be better than this if sieves are lightly loaded.