THE ART OF SAMPLEING

One example put forward at the meeting demonstrated the difficulties in preparing adequate samples for analysis The following assumptions were made:

* There was a random distribution of equal-sized gold grains

* A 10-cm sample of bq core was crushed and 25% pulverized to 100 mesh

* The gold particles passing through 100 mesh screen each weighed 0 05 g

* A 20-g sub-sample was taken for assay

* The grade of the ore was 0 5 oz gold per ton

The material was then hypothetically sampled based on the Poisson Statistic and the distribution of the gold values was plotted Only 6% of the sub-samples can be expected to return the true value and, more significantly, almost 20% of the sub- samples are expected to be zero Of course, this is a simplified situation, as many gold explorationists will confirm that gold is rarely distributed randomly, thus compounding the problem

Fire assays present problems of their own A sample submitted for fire assay analysis in a commercial laboratory goes through a number of well- documented steps (Documentation for this technique has been found dating back as far as 2600 BC ) A number of these steps are critical and warrant special attention First, a decision must be made as to how much sample will be assayed The most common sample weights used are a half- assay ton (14 6 g) or one-assay ton (29 2 g) As little as 5 g is used in some Canadian laboratories, whereas 50 gm is reportedly the accepted weight in Australia The sample weight should be maximized, although the appropriate weight will depend on the distribution of gold in the sample material

The pulp is then mixed with a flux, which is adjusted according to the oxidizing strength of the sample material This means that unusual types of ore, which can be quartz veins, sulphide-rich or chromite-rich ores, require adjustments to the routine flux used in a production laboratory The flux may not be carefully adjusted in a production laboratory if the ore type is not indicated when the samples are submitted

The sample, flux, litharge (lead oxide) and a few milligrams of silver are fused in clay pots at 2,000 degrees C for approximately one hour The molten material is then poured into cone- shaped iron molds The silicate slag floats on top of the noble metals collected in the lead button The cooled brittle slag is hammered from the lead button, which is shaped into a cube The weight of the lead button is the assayer’s best clue to the effectiveness of the fusion Adjustments can be made to the flux and the assay can be repeated to improve gold collection

The lead button is placed in a bone ash cupel and fired again About 98% of the lead is absorbed into the cupel and the rest volatilized All that remains at the end of this complicated procedure is a very small gold (and often silver) bead Until recently, the only method to estimate the amount of gold in the bead was to anneal the bead, primarily to remove the silver with acid and to physically weigh it for a gravimetric finish

Detection limits have been lowered by dissolving the entire bead and analysing the resultant solution Atomic absorption (aa) and inductively coupled plasma (icp) spectroscopy are commonly used to determine the concentration of gold in these solutions (This should not be confused with an aa analysis wherein the solution derived from an acid digestion/mibk extraction procedure is analysed by atomic absorption spectroscopy to determine the concentration of gold in a sample )

Fire assay is not a simple technique It is often referred to as an art and qualified assayers are hard to find Some sources of error are standard for all types of analyses, such as calibration and contamination of laboratory reagents A relatively unique problem for the assayer is that the sample is transferred at least five times during the procedure and the pots, cupels and certainly the lead buttons cannot be labelled There is plenty of room for human error Most errors can be controlled by the use of internal control samples, duplicates and blanks The commercial laboratories have well-developed quality control programs It is interesting that company-controlled laboratories, the mine assay labs, are often more neglectful of these requirements It is frustrating for the assayers who perform quality control tests to discover that turn- around time is usually more important to a geologist than data quality

A geologist can’t be expected to have an analyst’s deep knowledge of the assaying process and, therefore, it is the responsibility of the analyst to ensure that the geologist is aware of all options But the geologist should know the method of sample treatment, the best method to use for a given sample type, and the ramifications of any changes made to the procedures

In addition to understanding where errors may occur, it is critical to know the size of the errors being incorporated into the calculations for ore reserve estimates, said Ron Deptuk, chief geologist for bp Resources Canada’s engineering and mining division Minimizing errors is important because the sampling and analytical errors for each sample are cumulative It is possible to measure the precision of gold assays with quality control systems — specifically, duplicate analysis Precision is defined as “the expected difference between measured values and real values for a given confidence limit ” Thus for the confidence limit of 68% (one standard deviation) that is used at bp, a precision value of 20% would mean that the gold assays from the same block of ore would agree within 20% of the mean of the samples in 68% of the cases studied Gold values which differ by more than 20% are still acceptable if they are restricted to 32% of the cases

To demonstrate the importance of measuring precision, a reserve simulation by C J Carter of bp Minerals America was presented The average grade was set at 2% with a standard deviation of 1% Using a cut-off grade of 4%, as the precision increases to 100% the grade estimates will increase gently to 134% However, tonnage estimates will increase dramatically to 817% of the correct value Where the cut-off value is equal to the average ore grade, even at a precision value of 100%, the tonnage estimate is 86% of the true value Using a cut-off value near the mean ore grade, measuring precision and limiting precision errors are obviously critical when evaluating ore reserves

According to another speaker, Bill Roscoe of Roscoe Postle and Associates, critical decisions to proceed in the early stages of exploration are often based on inadequate data or evaluation factors The quantitative data for a project should be as representative and reliable as possible so that exploration funds are not needlessly diverted to uneconomic showings but are used effectively to evaluate potential mines

Generally speaking, the following precautions should be taken in the early stages:

* Collect large samples with a minimum of 10 gol
d grains or about 10 kg

* Pulverize large sub-samples of the 20 mesh material and analyze maximum-sized pulps

* In special cases where coarse gold is expected, a metallics assay should be requested Studies that may be postponed to later stages of exploration include the determination of specific gravity, grain shapes, particle s ize distribution, liberation factors and mineralogy All these recommendations will add costs at a time when it is more difficult to raise funds Lynda Bloom, a Toronto-based geochemical consultant, organized and chaired the meeting on sampling

]]>