"Representative" sampling - What does it mean?
Recently, I came under pressure from a project manager who demanded that I insert the word "representative" in describing samples in my report. Several of the samples in question had been supplied by the client and were taken from a site about 30 km away, albeit in the same geological terrane and similar stratigraphy.
Our clients and colleagues in other professions often consider all rocks of the same geological unit to be uniform in composition. Therefore, one sample tells you all you need to know about that unit, right? Definitely not. Just as a rugby team may be all men, but have different heights; so all samples of the same rock will have slightly different chemical composition. This is why ore has a range of grades.
To be "representative" in the scientific sense, a number of samples would need to be tested so that the variation in geochemistry can be estimated. Just as you would need to measure several players on the rugby team to understand the average height and the variation around the average. Speaking of rugby teams, 15 samples is what I would call the minimum to characterise a rock type to the degree I could call "representative".
In my experience, it is a rare project (or client) that can allocate sufficient budget to analyse a statistically representative sample set. This is strange, since the need for representative samples of the ore body is generally accepted. Obviously, the project return on investment still has a higher priority than characterising an acid mine drainage or water quality risk...