IKN

Barkerville Gold MinesHow many ounces does it take to fill the Albert Hall?

On June 28 Barkerville Gold Mines (BGM.TSX-V) published their NI 43-101 compliant resource estimate of an indicated 10.6 million ounces grading 5.26 grams per tonne gold. The Qualified Person (QP) for this report, Peter George, went on to suggest that, based on his interpretation of the geology, the district had the geological potential to contain between 65 and 90 million ounces grading between 4.11 and 5.49 grams per tonne gold. Needless to say, this tweaked our interest because, if accurate, the Cariboo Gold project represented one of the most important gold discoveries of the past two decades and we wanted in.

Our internal review turned up some fundamental red flags that we felt were serious enough to alert readers—we had received numerous queries from subscribers and felt it was important enough to issue an “avoid” commentary. If the BGM resource announcement didn’t immediately raise some red flags, you may want to re-consider speculating in this sector. Quinton’s initial review can be found in the July 2 issue of Exploration Insights. 

On July 11, Barkerville issued a clarification statement that was mandated by the British Columbia Securities Commission (BCSC), the gist of which was, “The Company cautions investors not to rely on either the indicated resource estimate or the “geological potential” disclosed in the Original News Release.” On August 14, BGM filed the Technical Report on SEDAR reporting an indicated resource of 6.605 million ounces grading 0.126 opt plus an inferred 5.729 million ounces grading 0.189 opt; the BCSC immediately issued a cease trade order until the company file a technical report that is acceptable to the commission.  

So, what is the problem with this resource estimate?

First let’s be clear on what a resource estimate really is. It is merely an estimate and that is the critical concept to bear in mind when considering any such report. The objective of a resource estimate is to come as close as possible to representing tonnes, grade, metal content, location, and distribution while accurately honoring the available data. Because the distribution of the data, usually in the form of drill hole and/or underground samples, in relation to geological controls is almost always uneven, levels of confidence in the resource are assigned to the estimate. From most to least confidence it reads like this; Measured, Indicated and, Inferred. The resources theoretically have to exhibit “reasonable prospects of economic extraction” but do not require an economic study to substantiate those criteria. To qualify for the highest level of confidence, Proven and Probable Reserves, which are a subset of Measured and/or Indicated resources, the estimate has to be backed up by an economic evaluation. Detailed explanation is presented here.

Because resource estimators are always interpolating between known data points (samples of variable quality) into areas with no data, there is considerable subjectivity, as is implied by the term “reasonable”. In reality, the actual volume of material that is assayed typically comprises between 1 millionth to 1 billionth of the deposit volume (for instance a 30 gram sample pulled from 5 kilograms of crushed rock can be used to determine the grade and tonnes of rock 50 meters away). Therefore, a stringent and accurate interpretation of the sample quality, geology and geostatistics is critical to extrapolating where mineralization is and at what grade. Got it? 

The Geoex Report on the Cariboo Gold Project, August 12, 2012

Let’s get this out of the way at the start: The technical report is deficient with regards to the inclusion of details of the project database and many aspects of the resource modeling procedures; so much so that a complete understanding of the process and conclusions could not be arrived at. The report also lacks relevant figures and tables that are key to any resource report’s ability to give a full understanding. If Barkerville is serious in their efforts to produce a reliable and widely acceptable resource report they should bring in another independent firm to re-do the entire study in order to clear up this mess. Mine Development Associates and AMEC come to mind; although it is actually the estimator that matters more than the firm.

Probably the most “unique” aspect of the Geoex resource estimate is the treatment of high grade, low grade, and no grade in the model. They are all given equal weighting; meaning, each composite assay whether it is 0.01 opt, 54.0 opt, or not assayed are given equal weighting in the estimation (another problem we’ll leave to the estimators, but note individual assays are being decomposited into multiple assays, thereby manufacturing data that doesn’t exist). Basically, the report describes what can be interpreted as extremely high grade samples being projected (smeared) into rock (blocks) that does not contain high grade mineralization and probably is essentially barren. It also appears to us that grade is being projected beyond drill core that was not assayed (reportedly because only visually mineralized core was being sent out for assay at the time).

It’s as if you threw all the assays, from 0 to 54 opt, into a pot and said they all have equal continuity, when the actual geology tells us that high-grade mineralization is quite poddy and has much more limited continuity than lower-grade samples. The end result is an exaggerated volume of estimated high grade material for which you have no support and we would have little confidence (not an official NI 43-101 designation by the way).

In fact, the entire approach of the resource estimate reported in the Technical Report ignores the known geological reality that gold mineralization at Cariboo occurs in high grade, steeply dipping, erratic pods. Further, within the high grade zones, gold mineralization is extremely variable and nuggetty yet the values are uncapped in the resource estimate, so the high grade “nuggets” are treated with equal influence in the estimation as the lowest-grade samples.

What this means is that within the conceptual pit, which was eyeballed in by the estimator, the estimate reported is unconstrained and with no cutoff grade applied—everything is included and co-mingled in the estimate and reported as the resource! Yet, more than half the assay composites used in the resource model are described by the estimator as having no significant values. Therefore in the absence of extremely closely spaced drilling to tell you where the high grade is, the results of such an unconstrained estimation will most likely be a total lack of correspondence between the resource model and the actual in situ mineralization. This applies to the entire model, but particularly so for the higher grades. Again, the important point here is that if this were to be mined, most of us would have virtually no idea what grade to expect and little confidence that ore is even there—a very tough way to go if you are serious about making money digging holes in the ground.

The resource tonnage estimate is also rather “unique”. An eyeballed open pit was used to capture all the modeled mineralization. Since every bit of rock within the estimator’s mineralized limit is included in the hand-drawn pit and would have to be mined anyway, it was therefore included in the tabulation of the resource tonnage (no cutoff grade was applied to the resource model blocks). This methodology is indeed unique—that is, an NI-43-101 resource in which no cutoff grade is applied. Blocks with 0 grade were compiled along with all mineralized blocks.

The “mineralized zone” was interpreted on cross sections spaced at 400’ intervals across a 3,200’ strike length and brought down to level plan (horizontal sections) representing (in the report’s terminology) mine benches 100-feet in height. Again, this methodology completely ignores the geological reality of the mineralization, and fabricates horizontal controls to gold mineralization that in reality do not seem to actually exist.

Another unusual aspect of the bench level estimates is that 48% of the resource ounces lie within one 100-foot bench (#3950). For the non-geostatisticians in the audience, the coefficient of variation (CV) for this bench is over 7. The CV is a fancy way of expressing the variability of the assay (or composite) grade population used in the estimation.  A CV of about 1 or less means the global variability of the data used in the estimation is at a level that, assuming all other methodologies used are appropriate, introduces little risk of the estimate being wrong. Progressively higher CVs indicate there are increasing risks of producing an unrealistic estimate due to excessive variability in the underlying data. A CV of 7, with no other mitigating constraints applied to the modeling (as is the case at Cariboo) demonstrates that the risk in the estimate is extreme and significantly outside of the geologic reality. Again a major bummer if your mine plan relies on the mineralization to be there and it isn’t.

Corroborating evidence for our lack of confidence in the estimated resource comes from the calculation that, 73% of the gold in this single 100-foot bench (#3950) vanishes if you apply the top cap used by the estimator in a test case to evaluate its effects. Surprisingly, this extreme result did not prompt a reconsideration of the use of an essentially unconstrained estimate. 

I’d love to tell you more 

Before leaving the NI 43-101 report here are some fun cocktail party facts:

• The resource tonnage of 88.7 million tons is six and a half times higher than the previous 2006 tonnage of 13.8 million tons.

• The 13.1 million resource ounces is 22 times more than the previous 0.59 million ounces. Interesting that 6.5 times the tons produces 22 times the ounces, eh?

• The indicated resource is based on only 22,355 feet of core sample data.

• 48% of the indicated and inferred resource lies within one bench (#3950).  This bench has a Coefficient of Variation of 7.

• A total of 4.58 million ounces are removed from the 3950 bench if a grade cap is applied.

• 78% of the resources are estimated in benches with a CV of greater than 3.

• More than half the composites used in the estimation have a grade of 0 opt.

• Of the 16,076 composite assays that exceed a grade of “0” only 794 have grades that are not equal to those immediately above or below them in the drill hole.

• There are 266 composites that equal or exceed 1opt. Only 44 have values that are unique from the two surrounding composites from the same hole.

• There are two consecutive composites grading 56.44 opt. 

• The inferred grade is 50% higher than the indicated grade.

• The tonnage factor (density) was not measured, it was calculated.

• There was no evaluation of QA/QC (quality assurances and control) other than Barkerville’s assurance that no problems were experienced with the standards and blanks submitted to the lab.

• The possible effects of core recovery and percussion drilling on assays are not addressed other than the estimator’s confidence that they are generally good.

So, long, complicated, and boring story short: In our opinion, the 43-101 compliant Technical Report is crap, the estimated tonnage of mineralized material, particularly the high grade, is bloated and meaningless and, high grade has been smeared and is flying all over the place. We don’t buy into this estimate, and Barkerville Gold Mines would do well to hire a recognized resource estimator to complete a full-blown independent resource estimate and put all questions to rest. Quinton and I don’t doubt that there is gold there or that there may be a decent deposit at Cariboo, we just have no means of evaluating what it might look like or be worth.