This report gives the results of a cross-testing experiment on the PSV test. Three aggregates were used in the experiment, and it involved 20 laboratories from 9 countries.
The results show that, when assessed by the criterion that the reproducibility standard deviation should not exceed 8 % of the average, the PSV test satisfies the criterion easily at all three levels. Hence it may be considered to be satisfactory for the purpose of assessing the compliance of aggregates with an upper specification limit. However, it is used in practice, on occasion, to assign aggregates to bands only 2 or 3 units wide. It is not possible for a test method with a reproducibility of 5 or 6 units to do this reliably unless information about individual laboratory biasses is available from proficiency tests. It is therefore recommended that the establishment of a proficiency testing scheme for the PSV test should be considered.
Plots of results for the three levels (S values) against results for the control stone (C values) depart from the pattern that would be expected from the correction formula (PSV = S + 52.5 - C). If the reason for this could be found it would allow the reproducibility of the test method to be improved.
Checks may be applied to the results of the PSV test in addition to those specified in the draft Standard for the range between runs for the control stone. Thus a critical range of 4.0 units may be used to check the range between the results of two individual specimens tested in the same run, or to check the range between two run averages (where a run average is the average result for two specimens).
The results of the experiment allow the precision of the PSV test to be stated (in PSV units) as:
r1 = 0.042 X
R1 = 3.30 + 0.0333 X
where X represents the average of the test results.