Recently, we reviewed the data accumulated from Proactive Monitoring (PAM) and Good Delivery List (GDL) application assay results from April 2005 to May 2019. For gold, there have been 2,400 assay comparisons, and for silver, there were 530. These include assay comparisons from laboratories which failed the tests (450 for gold and 50 for silver).
What Does the Data Tell Us?
Each participating assay laboratory has received feedback on their individual assays. But what can we learn from the data collected?
A key indicator is the difference or divergence between the assay laboratory reported fineness values and the LBMA reference fineness values (from the Referee Laboratories) for the same test samples. Each Referee Lab produced a set of test samples and verified the homogeneity by assaying many sub-samples. LBMA conducted blind cross checks of their assays in comparison to two or more other Referee Labs to arrive at a reference assay value.
Analysing the Assay Divergences
The divergence is calculated by:
∆ = LabAssay – LBMA
For gold samples < 999.5‰, 2087 assays were compiled. Overall, 93% of assay divergences were within ± 0.15‰ which is the limit for this material according to the GDL Rules. A frequency distribution histogram is shown in Figure 1. The distribution peak is at 0.00‰ divergence and the divergences are normally distributed. There is very slight tailing towards the positive divergence side which indicates some laboratory assays are high biased perhaps from proof over correction using fire assay analysis.
Figure 1. Divergence frequency distribution histogram for Au < 999.5‰ for data collected from April 2005 to May 2019.
For higher purity gold > 999.5‰, 313 assays were compiled. Overall, 86% of assay divergences were within ± 0.05‰ which is the limit for this material according to the GDL Rules. A frequency distribution histogram is shown in Figure 2. The distribution peak is centred at -0.01‰ divergence and the divergence distribution is slightly skewed towards the negative divergence side. This indicates that there is a slight tendency for low bias reporting. For high purity gold, assays may be done by fire assay or by difference. More data must be collected to better understand this tendency especially with lower number of comparisons.
Figure 2. Divergence frequency distribution histogram for Au > 999.5‰ for data collected from April 2005 to May 2019.
