Sampling nuts

I was watching QI recently and host Stephen Fry asked “Why do Brazil Nuts always rise to the top in a packet of mixed nuts?“

(For those who don’t know QI is a BBC quiz show, the quiz is about facts that are not widely known, or widely accepted facts that are in fact not true. Contestants are rewarded for answers that are Quite Interesting, thus the name QI.)

The answer to this question, according to QI, was ‘no-one knows’. Regular contestant Alan Davies received a bonus for this answer, a rare occurrence for Alan, but I was even more surprised since in my mind, I knew why.

Let me explain using a couple of slides from a presentation on sample preparation

As you can see when the bag of nuts is shaken in transit the brazil nuts would part and smaller nuts would fall through, having the effect of raising the brazil nuts to the top, in the same way that larger particles part and smaller particles slip through in the slides shown.

This is why when preparing a sample for measurement we need to think about our sampling. How is the initial sample taken, is it just spooned off the top? Then, when preparing a sample for image analysis, are we sure the sample is well mixed and we are getting a cross-section of all the sizes involved? Perhaps I’m wrong about the Brazils, but when making a measurement, if you don’t consider your sampling you may end with a result that is just nuts!