Cut half core, take one half and crush and split 50:50 and send both supposed duplicates to laboratory. Means turning your core shed into a sample prep facility but gives better control over your duplicate samples.

I find this fascinating that a lot of people are regularly using quarter core as a test of the half core sent in for analysis. This suggests you have no faith in the sampling method you are using? If so should I considering investing in your project or not? How do your management team and Board of Directors sell this to potential investors? Do investors leave a technical review meeting with the strong impression of ever decreasing circles attempting to improve results by another 0.1% when they have just accepted a feasibility study at an accuracy of ± 15%.

One comment above suggested that if the half core and quarter core showed wide variance then hole core should be sent in. I did not know that Brex was still about. If the mineralization is such that the drilling and sampling method will have a substantive effect on the sample results then either a form of RC drilling is used with large sample size bordering on bulk sampling or direct bulk sampling. For example well weathered and altered ore bodies would probably need this treatment. Or are the other contributors suggesting that quarter core is required as they do not trust either the drilling, sampling or analytical teams?

The point I am trying to put over is that regularly using quarter core as a check on the half core suggests that you, the resource geologist, have no faith in your drilling, sampling and analytical method or the staff as a tool to analyses an ore body. If this is the case then find a different sampling method. 

As pointed out by some of the precedent colleagues, core duplicates may provide useful information about various aspects. However, the main purpose of core duplicates is to assess the sampling precision, when their results are evaluated in combination with coarse-reject duplicates and pulp duplicates, each of them referring to different portions of the sampling-preparation-assaying process.

Assessing sampling precision may appear to be a rhetorical issue, but it is relevant for answering to a very practical question: are the half-core samples reasonably representative of the entire core segments from which they were collected? As mentioned above by various colleagues, we can only evaluate the representativity of a sample by repeating the sampling/preparation/assaying process in similar conditions, or at least in conditions as similar as possible to the original conditions; in other words, in repeatability conditions, which are only ensured when both the original and the duplicate samples have undergone the same processes from collection to assaying.

When "duplicate samples" are collected, prepared and/or assayed using different procedures, equipment or methods, at different moments, by different personnel, on different labs, etc., the repeatability conditions are not respected. It does not mean that such actions are completely useless, but they are not valid for assessing precision on repeatability conditions (or sample representativity). We usually collect various types of duplicates during due-diligence or audit exercises, not mentioning the check samples submitted to secondary laboratories as part of the customary QC protocols, but their purpose is not certainly to assess precision.

If you want to assess how representative your half-core samples are, I strongly recommend not using anything else than the remaining half-core, submitted to the same lab in the same batch with the original half-core samples. Otherwise, you will not be able to determine if your half-core samples are sufficiently representative of the entire core segments. When your ordinary sampling protocol consists of half-core sampling, collecting quarter-core "duplicates" of half-core samples, or quarter-core sample pairs, or crushed splits of the entire core or the half-core, or the like, will not certainly allow you to assess how representative your original half-core samples are, simply because such irregular procedures are not part of your regular sampling protocol.

There is an additional issue, though, raised by some colleagues. If we collect half-core duplicates, those intervals will remain with no core in the box. Nevertheless, nobody is talking about collecting 100% core duplicates. I personally recommend the insertion of 2% core duplicates, which means that only two samples in 100 would lack witness material in the box.

If we keep an adequate photographic record of the original core, having 2% of the samples with no remaining core could not be a serious issue. At least, it’s not for me as a regular auditor in tens of projects elsewhere. Are we afraid of some auditor complaining for the lack of remaining material in two samples out of 100? Then we can accommodate the coarse-reject bag into the box, after receiving it back from the lab, in the place corresponding to the sampled interval. Of course, we would not have coherent core there, but RC drilling does not produce coherent core either, and it is widely used and accepted. In fact, we can explain the auditor that we had to use RC for that interval.

PS: If anybody is interested in the recommended control-sample insertion frequency, I could submit a paper on the subject.

I agree with much in your post, but a straight 2% for any duplicate programme is not what i use. It really depends on the commodity and the inherent variability, plus the type of programme. An early-stage scouting programme for Cu-porphyry i would lower the frequency, a resource drill-out for coarse gold, i would increase it.

I'm personally not too worried about sampling an interval to destruction but the filleting options caters for this appropriately. Not many people commenting on this. And i still agree with earlier comment that the sample support is too small anyway so your results sub-optimal regardless.

And yes, i am interested about your thoughts on recommended control-sample frequency and your reasoning behind it.

Just in response to earlier comments

“I find this fascinating that a lot of people are regularly using quarter core as a test of the half core sent in for analysis. This suggests you have no faith in the sampling method you are using?”

There is a difference between having faith in something and the scientific method of collecting data to prove that something is so. Every time I implement a sampling protocol I do have faith that it will work, however how do I prove that it is working to my management team and Board of Directors without collecting data to show it is working? (I am also pragmatic, and know that in a perfect world my sampling protocol will work perfectly, all the time, without the need to check; but, alas and with much hand-wringing, I have a workforce made up of imperfect humans who are prone to making mistakes. The combination of “humans” AND trying to evaluate heterogeneous mineralisation on a planet where no two ore-bodies are exactly the same is a recipe for disaster, so a geologist just has to collect some data to make sure everything is okay).

Additionally, if I am doing a due diligence review on a project for a possible M&A, I will be very suspect of the Project that has an absence of field duplicates. I am sorry, but a “I have every faith in the sampling method we are using” from their Geology Manger will just not wash with me (just imagine telling your own shareholders that you are buying into a Project, and the due diligence is fine because their Geology Manager said “I have every faith in”!).

"One comment above suggested that if the half core and quarter core showed wide variance then hole core should be sent in. I did not know that Brex was still about. If the mineralisation is such that the drilling and sampling method will have a substantive effect on the sample results then either a form of RC drilling is used with large sample size bordering on bulk sampling or direct bulk sampling”

If you are referring to the remark made in my previous contribution, then you are taking it out of context a little. I was merely stating that a significant difference *may* indicate that the sample volume is not appropriate for the style of mineralisation being evaluated, in which case larger samples may be required – whole core or larger sized core or bulk samples or large diameter RC holes. A significant variation would not automatically trigger a “larger sample!” response but an “hmmm something is awry here, it requires further investigation” response. Using Occam's razor I would first start looking at simple things that may account for the variation, before jumping to the more expensive “larger sample” solution.

As far as the “Bre-X” jibe - in the coal industry usually the whole seam core sample is taken for coal quality testing, so it is not unknown to have no reference sample remaining after sampling. Of course, best practice is to retain some core, and in those cases where it was deemed that the whole sample will be submitted, then I would try to at least retain a fillet of core in the core tray.

“The point I am trying to put over is that regularly using quarter core as a check on the half core suggests that you, the resource geologist, have no faith in your drilling, sampling and analytical method or the staff as a tool to analyses an ore body. If this is the case then find a different sampling method.

Again there is a difference between “having faith” and being able to prove that your faith is well placed. 

There appears to be diverse beliefs amongst us true believers of TOS. I have known some non-believers of TOS just dismiss it as a only a theory to justify their view to reduce direct costs and perform token QC to tick a box and not consider the consequential costs when things go pear-shape.
May be the practice of 1/2 v 1/4 quarter we are confusing the purpose.

If the purpose of the sampling is for a resource estimate as part of institutional funding for project development then the sampling will need to be at a very high standard as it will be closely scrutinised by a CP or QP acting in the interests of the institution to justify their interest charges to cover the perceived risks. If the development of a project is self funded by the company then a lower standard of compliance maybe acceptable by the company board.

The main purpose is to establish the sampling and assaying protocol is adequate for the purpose of which the above is only two of many. This protocol should be adopted at the commencement of a project by submitting duplicate samples at each stage of sample reduction to develop a balanced design sampling tree and once a protocol has been established submit for each "laboratory batch" a duplicate at each stage to develop an unbalanced design sampling tree to monitor each batch of assays complies with the DQOs for acceptance or reinvestigation or rejection. This can only be achieved if identical samples are submitted at the same time and undergo the same process as commented on above several times.

The other purpose is for independent audits as indicated previously. In this case I think it would be acceptable to submit a half core at a later stage to estimate the limits to the potential sampling and assaying errors.

I would see the submission of 1/2 v 1/4 cores a weaker test than submission of 1/2 1/2 at the same time. The point about the overwhelming rock heterogeneity I agree and geologists recognise it is as scale dependent, hence the requirement for twinned holes and variographic drilling to address the broader heterogeneity.

I do not think we should confuse the two purposes. Your comment on crushing one half core and submitting each crushed at the same time I have seen in consulting company SOPs as a "proxy to half cores" to estimate the sampling error. This is as logical as submitting two half cores as a proxy to duplicate holes. I think this practice is about ticking a box and feeling good. This practice is likely to compromise the original sample. Our sampling and assaying priority should be about minimising the resource risk first.

I could hardly disagree with your implicit concern about a 2% duplicate insertion rate. This is not what I use either. In my comment I wrote specifically about core duplicates, not about all duplicate types. You can find a more elaborate reasoning behind my recommendation in a paper devoted to the subject (Control sample insertion frecquency: is there an industry standard? The Association of Applied Geochemistry, 23rd International Applied Geochemistry Symposium, Oviedo, Spain, 14-19 June 2007). I will be glad to submit a copy of the paper. I tried to paste the file here, but I couldn't do it. Nevertheless, just for additional clarification, I will append the final recommendations of the paper, which I believe addressed your rightful comments:

The QA/QC programs should be tailored to the specific needs of each project. Whereas an overall insertion rate of 20% can be in principle recommended, the individual proportions of the various types of control samples should reflect the problems with higher probability of occurrence. With the advance of the QA/QC program and the identification and correction of those problems, the amounts and relative proportions of control samples can be adjusted accordingly.

It is essential that QA/QC programs are comprehensive, including all types and subtypes of control samples, namely twin samples, coarse and pulp duplicates, coarse and pulp blanks, CRMs corresponding to relevant grade values, and external check samples, so that precision, accuracy and possible contamination at the various points in the sampling-preparation-assaying sequence are properly assessed.

I use quarter core vs. quarter core, with half core retained, not quarter vs. half. This allows a like-with-like comparison. Clearly it is not testing the system perfectly since half core is the usual sample. However I am looking at the material, not the core size, in terms of total precision error. Hence, a small half core sample mass will be similar in mass to a longer quarter core interval. Looking at your FSE equation and monogram we can define minimum sample mass. This will be the mass never mind the core size.
We can then argue whether the quarter core performs differently in preparation, etc.

I have come late into this discussion but I wish to share the following. The 1/4 Core duplicate submission is good practice and does not need to be submitted into the lab at the same time with the half core. You however have to observe that the sampling intervals in the 1/2 and 1/4 core samples are respected and the insertion of the standards and blanks is done at the time of submission to the lab for QAQC. The individual batches can be interrogated for pass or fail criteria. Blanks should show no contamination while standard pulps should yield grade difference of within +-5%.

Pair-wise comparison of the assays from the 1/2 and 1/4 core batches can then be done. The results from this comparison is generally acceptable if 70% or more paired samples fall within the grade difference envelope of +-10%.If the results show greater variability at lower population, high nugget could be deduced.

The challenge in most cases is that the 1/2 core assays could have been done in the initial campaigns dating back 20 years when these QAQC protocols were not so stringent and all you get is a log with assays without any QAQC data accompanying and the duplicate 1/2 core in the shed.

The further splitting of the remaining half core in the shed and submitting the 1/4 core with your updated QAQC protocols will help to validate your data set and improve confidence or lack of it in the overall mineral resource estimation process. The CP or QP will therefore need to scrutinise this data and make relevant comments on the same.