Data Verification and Red Pine Exploration

Friday Resource Geology Tip: In this post, I hope to convey a few points: 1) a high-level summary, based on public information and news articles, of what happened at Red Pine, 2) my attempt to convey the importance of QP data verification, and 3) my personal thoughts on what “good practice” in data verification should look like when reporting mineral resources. Please appreciate I have no insider knowledge on Red Pine with all quotes from publicly available sources while my observations and opinions are my own.

Red Pine Exploration – Wawa Gold

Nearly two weeks ago, we read the headlines “Red Pine says former CEO [Quentin Yarie, P.Geo] tampered with Wawa gold assays”. The company is downplaying this mischief and Mr. Yarie stepped down back in February, then took on the role of Chairman at Canadian Metals. The allegations state that he manipulated 532 assays (out of 58,377 samples collected from 2014 to 2022) after receiving data from Activation Labs and prior to forwarding data to the Red Pine team for analyses from 2015 until late January 2024. The result was a roughly 60% share price drop for investors, inclusion as yet another case history of fraud in the junior mining industry, and probably worst of all, Mr. Yarie got himself inducted onto my infamous Erik’s Naughty List.

So here we have a allegations of the former CEO purposely increasing grade reported from the independent laboratory. Well, thank goodness the NI 43-101 regulations require an independent Qualified Person (QP) to perform data verification (Item 12 of NI 43-101F1) to identify and correct errors or malicious behaviors. Golder (now WSP) and Mr. Brian Thomas, P.Geo authored and signed off as QP for the mineral resources on the project in 2021 and again in 2023 (resources effective in 2019). This begs the questions: Did the QP do enough? Could more have been done? Should more have been done? Would another QP have done anything different?

Based on the published Technical Reports, QP data verification involved a small subset of data checked against lab certs, independent logging and sampling of quarter and full-half core sampling, and a review of previous data verification work performed. In 2019, at the Surluga deposit, a whopping total of 341 samples were selected for comparison to original assay data. Of those reviewed, 309 matched the database while 32 did not. Reasons are described by Mr. Thomas with a statement “it is the QP’s opinion that the differences are not material to the project.” As for Minto Mine South deposit, 161 Au assays were checked with all data verified against the laboratory certificates. Given that about 10% of these checked assays failed, one may argue that would warrant additional database checks if not consider validating 100% of laboratory certificates against Red Pine’s internal database.?

Thankfully, data validation did not stop at the desktop stage. I commend Golder/WSP for collecting independent quarter and full remaining half core samples. Core was relogged, sample cut and shipped to a third-party laboratory for independent analysis. This isn’t often done for a variety of reasons. In some cases, companies do not wish to budget this additional step, it can take multiple weeks to receive results, but more often than not, what does the QP do should those independent results not quite match? I think Red Pine, and more generally, high nugget gold systems are challenging and how much data is needed before a QP can confidently state something is not right?

In the figure below, we see two samples that obviously don’t smell right. This begs many questions. Was it due to nugget, fraud, biases in sampling, lab error, or some other unknown. Can any definitive conclusion be drawn? Was Golden even given a sufficient budget to do additional work if warranted? If I was in Mr. Thomas’ shoes as QP, would I have done anything different with those quarter core results? Would any other QP or what about the Golder peer reviewers? I’ll summarize by saying the QP’s job is not always straightforward, results are rarely black and white, and at what point does a QP raise concerns?

In the 2023 Technical Report, during 2022 another 377 samples were selected for comparison to lab certs. In this case, all passed. Eight more independent samples were relogged and collected from core and independent analyses. The results shown in the figure below show material differences between the Red Pine database and the independent lab. Additionally, Golder/WSP reviewed historical independent samples which show a fairly regular bias to increase Au grades from Red Pine (RPX Sample in Figure 3). Should a flag be raised now or is there still insufficient data to support any decisions?

?

Hindsight may suggest these and historical verification results (Figure 3) would raise concerns, but the QP statement is “The QP believes that these differences may be due to the uneven distribution of gold within those sample intervals, differences in sample volumes or possibly the result of differences in analytical procedures”.

One of the key takeaways I have is to ensure open and honest communications are occurring between the QP and the company, that most people do not want to call out fraud unless the data is undisputable, and many variations in data verification can be explained by geological variability, sampling and assay errors, and simply the result of a deposit exhibiting a high nugget effect. Being QP and verifying data is not easy, especially in challenging and complex deposits.

The Importance of Data Verification

Most geologists appreciate that all our interpretations, inferences, and decisions assume the data we rely upon is sound. I often call this fundamental data (drilling, assay, mapping, geochem, etc.) the “foundation of Resources”, because when cracks and weaknesses appear in foundations, the entire structure, refinements, and fancy refinements are no longer stable. Over the past few weeks, we’ve witnessed this firsthand when this seemingly straightforward section of any mineral resource disclosure goes sideways at Red Pine and the resulting stock price and reputational damage that ensued.

Data verification is explicitly outlined in NI 43-101 as Item 12, SEC under Chapter 9, and JORC Table 1 buried down in Section 1. Data must be verified by the QP and deemed suitable for use. The codes do not necessarily outline how this is done, just that the QP must disclose how it was done and any findings and options on the topic. The topic is purposely left nonprescriptive as there are many ways to assess data verification based on what types of data are being verified. The most common is checking for consistency between independent laboratory certificates and company databases but there are others described in the next section of this post.

Why does this warrant its own item or chapter is disclosure regulations? Simply put, it’s damn important! Consider all interpretations, statistics, estimation, domaining, and geological decisions are highly driven by the fundamental data collected. If this data is not verified by the QP, all further decisions and interpretations have reduced or no confidence. When it comes to mineral resources and the potential to extract value from a deposit, the value of the asset is not yet realized and all we have is confidence (or lack thereof) in the work completed.

Where things get rather tricky is how does a QP verify historical analytical, metallurgical, physical, or other data if there are not lab certs or other independent documentation accompanying the data. The majority of the projects I work on have some portion of their database which is legacy with virtually no certs, QA/QC, or supporting documentation. Can a QP verify this legacy data? Historical and non-verifiable data may be okay to use but the QP must tread carefully in how the data is treated and the resulting confidence in any interpretations or outcomes derived from that data. Re-drilling or at least twinning is recommended in most cases. The key takeaway here for legacy data is to be transparent and clear in how non-verifiable data is used, what if anything was done to verify the data and how this additional risk is accounted for in modeling, estimation, and classification.

Good Practices in Data Verification

This is by no means an exhaustive list but merely my summary of what I’ve either performed, observed, or directly reviewed that I would deem “good practice” in Data Verification.

• A good start to any data verification is by checking original laboratory certificates against a property database. Digital files can be altered so I recommend obtaining these directly from the laboratory, if possible. Ideally, 100% of assays should be checked and this should be a straightforward exercise using scripts but perhaps there’s a flashier tool or AI out there that QPs could share below.
• Historically, I’ve seen many QPs check some subset of data, say 5% or 10%. I’d state this is certainly better than 0% but a lot can be missed and perhaps a better approach would be to check all assays above a higher-grade threshold, within a pit shell, or other targeted method of selection to really test the volumes most driving project economics.
• The “best practice” would be to collect independent samples through quarter or half core. These independent samples should be conducted with no company personnel and all samples analyzed at an independent external laboratory. The problem with independent samples, as mentioned above, is how does the QP collect a sufficient number of data to be statistically relevant to make a decision on bias, raise flags, or call out fraud? The QP is never given an unlimited budget and time for this verification. Additionally, for due diligence exercises, there is rarely the time for the collection, transport, analyses, and interpretation of results.
• Data verification is about more than just chemical data. Often geologists and QPs get the blinders on and overly focus on chemical assay data. There is no argument that Au assays in a gold project are important, we must realize that often mineralogy, bulk density, rock type, metallurgy, and other properties are just as critical to provide confidence in tonnage, grade, recoverability, and economic estimates. This is especially true in most industrial minerals and iron ore properties.
• Perform checks on drill collar survey, downhole survey, and depths units. This can often be as simple as an independent GPS or survey coordinate check on a few collars, checking general downhole orientations from capped drill holes, along with 3D visual checks of data for reasonableness.
• Performing exploratory data analyses (EDA) by various categories (year, method, lab, company, domain, etc.) should provide insight into biases, errors, or outliers.
• Re-assaying rejects and pulps is a great way to increase an independent analysis without the high costs of core cutting, but the assumption is that pulps were not previously salted or manipulated.
• Twin drilling. Rarely do I see companies willing to take the time and money to conduct twin drilling for QP data verification or due diligence, but twins are a good way to validate legacy data. Twin drilling provides insight into variations due to different methods, sampling, prep, & analyses, assuming low nugget. Good practice includes a bit of it all with detailed discussions on findings with actions disclosed in the DV section of Reports (NI 43-101, SEC, etc.).
• Verification of raw, composited, and block data using production reconciliation, if available.

This article covered a lot, but I hope you found parts of it helpful or insightful for consideration on data verification. As always, I hope readers can add their own thoughts, experiences, disagreements, and observations on data verification. Happy Friday!