As I understand it -- and I know professional geologists will...

As I understand it -- and I know professional geologists will correct me if I'm wrong -- you get a bag of gravel collected over an an interval. You get good separation between intervals because you can clear the debris out of the drill rods at the end of an interval before starting the next. Within a single interval you don't get any separation; the chips are all mixed together. So you have to set the intervals at the finest level you want to analyse at. If you take 1m samples in the field you can't do 0.5m assays at the lab.

Doing a 4m analysis involves taking an equal weight of chips from each interval over the 4m. Statistically we hope that the chips from each bag are representative of the whole interval, giving you a weighted average of mineral content over the whole 4m. But with gold this is very subject to sampling errors because of the veiny, inconsistent mineralisation.

The results from hole HFRC022 illustrate this randomness perfectly!

Dec 22: HFRC022: 12m @ 0.57 g/t gold from 56m (including 4m @ 1.24 g/t gold)
Jan 30: HFRC022: 11m @ 9.58 g/t gold from 57m (includes 1m @ 98.7 g/t)

The discrepancy is frankly astonishing. Clearly the samples taken for 4m assay missed the high grade chips. As the lab extrapolate the assay for an interval over the entire interval, grabbing a high-grade chip can have an enormous impact on the result. Hole HFRC022 has gone from bust to bonanza. If we were to assay it again there's every chance of it being a bust again.

We have no way of knowing which set of randomly selected rock chips from each interval were more representative. I guess the lesson is to wait until you have plenty of holes before you can be confident about what size and grade of resource we're looking at.