Blind Faith-- Samples, Assays, and Estimates

Blind Faith--Samples, Assays, and Estimates
by Brent Cook
November 8, 2013

Here at Exploration Insights Central we tend to devote a seemingly inordinate amount of copy to resource estimates and what goes on behind the scenes en route to building said resource estimates. The reason is simple: a resource estimate is the fundamental building block for everything that comes later in designing a mine. If the estimate is inaccurate or flat out wrong, it can require considerable re-engineering and optimizing to fix-- if it can be fixed at all. These fixes cost time and add to production and capex costs, often taking a deposit from profitable to uneconomic.

Recently, some very high profile resource estimates have shaken the confidence many investors had placed in the Canadian NI 43-101 standard on reporting mineral resources and reserves. For those of you interested in pursuing this topic it is of prime importance to take note of the defining term estimate—the tonnes and grades presented are not hard cold facts. In exploration and mining we are nearly always dealing with insufficient data that can lead to considerable uncertainty and subjectivity regarding what lies hidden beneath the Earth’s surface between data points. Let’s take a step back and consider the underlying data of an estimate--the drill assays.

Here’s how it works

Drill samples are collected and sent off for assay. The results tell us how much metal is in the sample; this information is then assigned to the core interval, which in turn is extrapolated into the rock, and, finally, pushed out into the mineralized body to arrive at a resource. A very simple and straightforward process for anyone with a computer and spreadsheet, right?


One first has to ask, just how representative of the mineralized body is the sample and resultant assay. What was actually sampled and what was not? What was missed or lost during drilling? What was the core recovery and, what portion of the mineralization was recovered (two very different questions).

Consider the image below of cut core from a recent silver discovery in Argentina. The core panel at the top of the box is hard, essentially un-mineralized, rock with ~100% recovery. The central section of core in the image intersected a mineralized vein (dark gray) that is broken, with obvious bits and pieces missing. This is because the gray mineralized material is soft and the core breaks along the small vein. The grey mineralization was broken up as the core barrel turned and was preferentially washed away by the drilling fluids. Although core recovery of this section may be 85%, recovery of the gray mineralization is substantially lower and is therefore underrepresented by the core sample; consequently, grade will probably be under-reported. The third section of core in the box illustrates poor recovery, maybe 60%. How does one sample this material such that the material sent for assay is representative of the section of rock drilled? What is missing, potential ore, or waste?

(Fig. 1: Drill core illustrating variable core recovery and variable recovery of mineralized material. Top panel is competent rock with good recovery. Central panel shows the drill followed a gray vein and some is missing. Lower panel is broken mineralized material with poor recovery.)

And there’s more

Take note that the gray 0.3 cm vein of silver mineralization runs down the middle of the central core section for ~0.5 meter. Let’s say the assay returned 225 grams per tonne silver over that 0.5 meter. But we can clearly see that this is not the true width of the vein, which is 0.3 cm. If sampled perpendicular to the vein, the assay would probably return a much lower grade over 0.5 meter. We are, after all, after true width.

Let’s look at the two illustrations below (Fig. 2) and consider what is being sampled and what is being missed.

(Fig. 2: Left: image of silicified broken rock with clay alteration filling fractures. The open vugs are missing material that is not included in the core and therefore not part of the sample, but still is part of the mineralized body. Right: image illustrates a gold nugget in otherwise barren black rock. Which half of the core goes to the lab makes a big difference in the final assay.)

The core sample on the left shows broken silicified volcanic rock, outlined by brown altered clay that fills the cracks, plus large open vugs. The vugs were presumably filled with brown clay that was washed away by drilling fluids as the core bit was cutting through the rock. The obvious question for those of us wishing to know if this property stands a chance of becoming an economic deposit is, “Where is the gold; in the silicified rock, or the clay?” If it’s in the clay, then any sample from this core would be missing most of the mineralized material. If the gold is in the silicified rock, then the sample is still missing an important component of (un-mineralized) material—the clay. You see the problem here. Either way, this first has to be recognized and then addressed-- before handing the assays off to a numbers wonk with a computer to compile into a resource estimate.

The core photo on the right above presents a very interesting and common dilemma for sampling high grade nuggetty gold deposits.

Core is generally split in half, with one side sent for assay and the other kept as a record of the rock, etc. In this case, where do you split the core, and which half goes to the lab? By splitting it down the middle, it is clear that the right half with the gold will return much higher grades than the left half. Whoever decides where this split occurs and how such a problem is addressed has a direct effect on which assays ultimately go into a resource estimate.

What happens after the sample leaves the project is also critical

Assuming a 1-meter sample interval, the half-core selected for assay weighs about 2 kilograms. The sample is crushed, ground, and pulverized, then re-mixed at the assay lab to form as homogenous a sample as possible. From the homogenized pulp, 30 to 50 grams are split out for the final assay (in the case of precious metals). Again, any mistakes made here (like not cleaning the machines, or analytical errors) can be carried through to the assay database and eventual resource estimate.

There are methods of dealing with the difficulties presented above, although none are perfect. How they are rectified has a huge impact on the resulting resource estimate. Mistakes or improper sample protocol at the very early stages (consider a high school kid with a hangover working for minimum wage randomly cutting core over the summer) are carried and magnified right through the whole feasibility study. It is entirely conceivable that sloppy work is only recognized when what is coming out of the mine is not at all what was supposed to be coming out of the mine. Sound familiar? If not, read this and check the chart.

The point is that the assay can only accurately represent what was sampled and how it was prepared. The resource estimator has to know and be comfortable with every step leading up to him or her receiving the digital database.

Then it gets tricky. . .

A 50 gram assay split from a 1-meter, 4 kg core sample represents 1.25% of the entire core interval—not much. This assay and the other potentially economic assays in the string of core that were analyzed are then assumed to reflect the whole volume of material (mineralization) a certain distance from the sample points (drill holes). The interpolation from a 50 gram sample to a large volume of rock means we are relying on anywhere from one-millionth to one-billionth of the total volume of rock to accurately predict (estimate) the grade of the whole mineralized body. The more complicated the deposit, the more data points that are required, e.g., increased drill density, re-assaying, bulk sampling, etc.

Effectively and accurately addressing this unavoidable deficiency in data requires a good geological understanding of the deposit and appropriate geo-statistical treatment of the assays. It also means there is considerable subjectivity to interpretations, with some resource estimators relying more on old school geology and others relying more on mathematics to interpolate between the gaps in drill holes. Occasionally, the two philosophies clash.

The issues always revolve around how to spatially reflect the mineralization such that a mining engineer knows what to expect to come into the mill, and out of the mill. That knowledge forms the basis of the economic model that is used to finance the mine and project profitability. Knowing where the waste lies is often just as important as knowing where the ore is—a mine is a terrible thing to waste. Correctly estimating where the break is between those two rock types has always made the difference between a profit, or a loss.

As a speculator in this industry one must be aware of the inherent uncertainties that come with resource estimates and the underlying data. Rocks tend to be complex, and always more so when a sequence of events causes them to be broken, sheared, faulted, altered, and mineralized. It is therefore really quite critical to understand uncertainties in the geology, data, and resource estimation methodology to make money in the junior mining and exploration sector. Blind faith just won’t cut it anymore.

That’s the way I see it.

Brent Cook