• To participate in the 911Metallurgist Forums, be sure to be Logged-IN
  • Use Add New Topic to ask a New Question/Discussion about GeoMetallurgy or Geology.
  • OR Select a Topic that Interests you.
  • Use Add Reply = to Reply/Participate in a Topic/Discussion (most frequent).
    Using Add Reply allows you to Attach Images or PDF files and provide a more complete input.
  • Use Add Comment = to comment on someone else’s Reply in an already active Topic/Discussion.

Interpreting A Hard Plot For Gold Deposit Duplicates (2 replies)

Helena Russell
8 months ago
Helena Russell 8 months ago

Can anyone help me understand how to pin down possible reasons for getting only 37% of my original assays vs. pulp duplicates data plotting within 10% when I generate a HARD plot? I have over 400 pairs of data from three different labs. At each of the 3 labs the reproducibility according this plot is between 30% and 40%. However when I generate a normal regression correlation plot my R2 is over 0.9 at all the Labs.
I imagined that my regression correlation graph would coincide with the HARD plot. Are these two graphs different?

How can I improve my HARD plot from 37% to 90% plotting within 10%? What is the implication of the 37%?

8 months ago
Sturmbann 8 months ago

It's hard to comment without seeing the chart, but are you looking at data that is well away from the detection limit? If you are close to DL, you will see variability that is due to analytical noise and not the result of variability in the duplicates as such.

Ace Levy
8 months ago
Ace Levy 8 months ago

Unless your gold is very fine then you are likely to get a fair bit of noise from the pulp duplicates. The first thing to check is how is the duplicate pulp collected - if is taken from jus the top of the bag this is a problem. Check the procedure.

Next think to check is for coarse gold - have you done any screen fire assaying? If you have some coarse gold you may need to increase the charge of your fire assay to get better repeat precision. Also check what proportion of the charge is actually rock - if you have a lot of sulphide you may have an issue there.

Suggest if you are getting 90% of results with a hard difference of less than 40%, then that would be pretty typical of a lot of gold deposits around the world. Generate a ranked hard plot for this.

I'm not saying that this is a good value but you need to have a think about how critical this is for your ore-waste decisions.
Previous comment about including detection limit data is a good one. Perhaps exclude all results either result is not at least 5x detection limit.