If you have coarse free gold in quartz veins you are probably doing well with ±15% gain/loss per month. Predicting the grade from the pit is not easy and for better reconciliation you might have to look at your basic data that was used to predict mill feed. This could include drill density, method of drilling, sampling procedure, QC of sampling and the method of analysis for coarse gold.

Your description of coarse free gold, presence of electrum, and hessite may tell a portion of the story as well. You mentioned that your circuit was CIL, but with describing the reason for using CIL, such as any preg-rob in the ore. Additionally coarse gold, and electrum both can be, and usually are, very slow leaching, with very minor weight loss to dissolution in a standard 24-48-96 hour leaching circuit. You didn't mention if you have a gravity circuit ahead of the CIL. With respect to the hessite, that is a tellurium mineral (Ag₂Te), and may be giving you fits in fire assay. A quick study of all the mineralogy of the ore is warranted. Lastly, I concur with Karl (above-previous response) that coarse free gold and electrum are exceedingly difficult to sample, with the highest errors seen in mine sampling, and any sampling of coarse particles. I would suggest a true engineering assessment of your sampling practices, even if you didn't result in finding a poorly executed step, you would have built your knowledgebase substantially, and reacted appropriately to the large variance in closure metallurgically.

Hi!
The usual practice is to have a discrepancy of not more than 3% when compiling the metallurgical balance. Also, the discrepancy should not change only in one direction - positive or negative, three months in a row.

The metallurgical balance is affected by taking into account the mass of incoming ore, Dore, tails and gold content in products.
In Dore, the amount of metal is determined exactly.
The mass of ore and tail is determined by the conveyor scales installed in the feed of grinding. It is necessary to check the operation of the conveyor scales, calibrate more often.
More difficult with the definition of gold in the ore and tail. Firstly, the reliability of sampling and sample preparation. Secondly, the method of analyzes. Fire assay gives out more exact result in comparison with AA.

Total, if you can establish the accounting of incoming ore, sampling, sample preparation and analysis of processed products, the estimated content will tend to analysis. Surprises in the form of losses or finds will be less.

Can an auto-sampler solve the problem , because all our samples collected manually by our operators.