“How large should a placer sample be to insure an accurate determination of the grade?” This has been the subject of placer miners and evaluators for years. The geo-statisticians have weighed in with possible answers. Unfortunately their statistical standard is not realized until after the sample has been cut, concentrated, weighed, and the colors counted. Is there a better way? Yes, possibly. This essay delves into placer gold sample size and how to do better sampling of gold placers.

First let’s review the principles of placer sampling. A sample attempts to take a small portion of something much larger in hopes of fairly representing the entire body, or in this case an orebody.  Sampling produces best results if the mineral particles are consistently sized and evenly distributed throughout the material. Unfortunately this is never the case in gold placer deposits. In the real world the gold is in various sizes that range from large nuggets to flour size within one placer. Nuggets are much rarer than finer particles. Fine gold is much easier to find than are nuggets, but when a nugget is captured; it can make a huge impact of the grade of the sample a.k.a. the nugget effect. A large nugget will affect the grade of a small sample much more than on a large sample. Small samples have a much better chance of missing any gold due to gold’s scarcity. As a result the statisticians have come up with the concept that for statistical reliability the samples need to be large enough to contain at least 20 gold particles in each mesh size. This is a short order for flour gold, but a very tall order for gold nuggets. The rule of thumb is the larger the gold particle size expected, the larger the sample should be.

Figure 1 is a statistician’s logarithmic graph of gold particle size versus sample size. Twenty particles are the barest minimum number to have a 95% certainty. Not every mesh size must have 20 gold particles retained; as there are 40 Tyler screens between their coarsest, at about 1 inch openings, to their finest at 400 mesh or 38 microns. Rather only the economically important gold sizes need to have a minimum of 20 particles. In most placer mines those sizes are retained on 10, 35 and 100 mesh screens.

Size-of-sample-expected-to-contain-20-gold-particles

Figure 1 shows the relationship between the number of gold particles per kilogram (and pounds in orange) sample, particle size

[assuming all particles are uniform in size], and grade of the sample in parts per million. The parts per million lines are the diagonal ones. The orange lines point out how the difference in particle size affects the sample sizes. From Clifton et al., 1969 and Wiltse, 1983.

Placer samples must be large enough to fairly represent the most important or influential constituent sizes. Figure 2 graphically depicts how large sample sizes or diameters have better odds of containing more particles and larger particles. Typically placers close to their lode source have coarse enough gold that nuggets are far more important than the fines. Other placers, much farther downstream from the lode, may have no nuggets that would influence the sample. In this later case, nuggets are not influential, so huge samples trying to collect 20 of them would be a waste of time and money.

the-idealized-gold-particle-interception-by-sample-size.-After-McCulloch,-2003

Figure 2 shows the idealized gold particle interception by sample size. After McCulloch, 2003

 

Wiltse provided an example of how to use figure 1 even if you only a 16” pan and a 10 mesh sieve. Multiple pan samples are capable of quantitative analysis in the following example. Stream gravel in a cut bank has a coarse cobble layer where the gold is believed to be hidden in the fines amongst the cobbles. Sixteen inch pans vary in pan factors from 150 to 200 pans per bank cubic yard; assume a pan factor of 180. If you estimate the percentage the fine sized gravel (1/4 inch) in the overall exposure to be 30% and you sample only the small grains in the cobble layer. Your sample is 30% of what you would be sampling if you had the equipment to wash the cobbles too. You take six pans and sieve them through a 10 mesh screen to reduce the volume. Any gold too coarse to pass a 10 mesh screen is easily seen on the screen and saved. The minus 10 mesh sand is panned down to down to just gold particles. Sieve the gold particles through the 35 mesh screen and count the colors larger and smaller than 35 mesh. If they amount to 20 or more in each, then each sample is statistically up to the standard. The math would look like this:

 

6/180 (pans/bcy) = .11 bcy in the sample
30%

 

The sample represents an original volume of .11 bcy and it has been reduced to about 50 pounds of minus 10 mesh. If the cutoff grade is an unrealistically high of 1 part per million (ppm) for the sake of this example, then the sample would need to have at least 20 minus 10 mesh gold particles. If there are less than 20 particles, the ground does not meet the statistical standard. The amount of gold could still weigh up to be an adequate grade, but it would not meet the statistical standard for number of particles. You will need to add more pans for better statistics. The gold should be weighed after the particles are counted. Notice this multi pan sample was not split, it was sieved and panned which reduced the volume without reducing the number of gold particles. It was not sent out to be chemically assayed.

Clearly a series of pans of the finer sediments does not represent the entire mining section and should only be used to represent the cobble layer. This data combined with other samples of the adjacent layers could represent the mining section.

It is unlikely that one would foresee before actually processing the sample what size the gold particles will be beforehand. Few of us can see into the future or into the ground with much clarity. However we know from experience that large size gold grains are deposited with large rock sizes. This is related to hydraulic equivalency. It is easy to see the size of the gravel pieces as you are digging them. The rock sizes can be used on following table from the Bureau of Land Management training materials as a handy guide for the size of your placer samples.

LARGEST ROCK SIZE SAMPLE SIZE
+10 inches (small boulders) 3 tons or 1 cubic yard
+ 4 inches 1 ton or 1/3 to ¼ cubic yard
+ 2 inches 400 pounds or 1/10 cubic yard
+ 1 inch 100 pounds
+ ½ inch 50 pounds

Table 1 shows the statistically adequate sample sizes for placer gold sampling programs (BLM).

This table is useful for transported (alluvial and beach) placers, but not residual and colluvial placers. The size of the rocks in residual placers is not related to gold size because residual and colluvial placers are not transported by running water. Their rocks can be any size, or completely without rocks; while their gold particles are often coarsest of all placers. Coarse rocks and coarse gold are usually found together in only transported placers.

It is common in steeply sloping hillside areas and glaciated landscapes to find huge boulders, the size of chairs, desks and even small foreign cars. These are erratic boulders that should not be confused with stream carried gravel. Huge boulders like these were not rolled by the stream; rather they rolled down the hillside, or dropped by a long gone glacier.

The Bureau of Land Management has another useful and relevant table relating to sample size. Table 2 shows the effect gold in three different particle sizes have on the sample grades at $1600 gold. It is obvious from this table how much a nugget can affect small samples and why large samples are always more statistically believable.

Sample Size 20 Mesh (6.5 mg) 40 Mesh (1 mg) 60 Mesh (.3 mg)
7.5” by 12” drill hole $26 $4 $1.33
5.5” by 12” drill hole $59.28 $9.12 $2.72
6” by 12” channel spl. $18.56 $2.52 $.96
12” by 12” channel spl. $9.24 $1.32 $.36
16” pan @ 180 pans/cy $61.56 $8.52 $2.52
12” pan @ 400 pans/cy $136.80 $18.93 $5.60

Table 2 presents the effects of gold particles on various size samples. Adapted from Wells and the BLM.

In summary samples should be large enough to be representative of the important influential components of a sample. Often nuggets are the most influential component, but not always. If nuggets are scarce, while +20 mesh flakes are what pays the mining costs, then taking huge samples in an effort to get 20 +4 mesh nuggets may be a wasteful expense. As a matter of fact, this logic has been applied for years in the placer industry by discounting nuggets in small samples for fear of the nugget effect skewing the sample data.

In spite of what is written above gold pan size samples are still important and worthwhile for prospecting. Pans can determine the presence or absence of gold. Pan samples can even be used for quantitative sampling if multiple pans are washed or the gold particles are fine.

Statistics is a numbers exercise. It is a way to mathematically set a standard to measure up to. But employing statistics does not guarantee the placer ground contiguous (laterally or vertically) with a sample will be a similar grade or even guarantee any gold at all. This is true even though the adjacent gravel may be identical in every other respect. Statistics work well where the material sampled is homogenous with random distribution of the constituents. Gold particles in stream placers are definitely not randomly distributed in homogenous materials. In order to minimize the statistical risk, many samples and large samples are the best recommendation.

References

B.L.M. (U.S. Bureau of Land Management, National Training Center) Advanced Placers (Course 3000-76)

Clifton, E., and others, 1969, Sample size and meaningful gold analysis: U.S. Geological Survey Professional paper 625-C

McCulloch, Robin, Lewis, Bob, Keill, Don and Shumaker, Matthew, 2003, Applied Gold Placer Exploration and Evaluation Techniques: Montana Bureau of Mines and Geology, Special Publication 115

Wells, J.H. 1969, Placer Examination: Principles and Practices: U.S. Bureau of Land Management, Technical Bull. 4

Wiltse, M.A., 1983, Preparing Placer Samples for Assay; Alaska Placer Mining, Bruce W. Campbell, James A. Madonna and M. Susan Husted, eds.

MIRL Report No. 68