Isn't it extraordinary how difficult it is to get a sample really random? Every possible precaution, as it may seem, sometimes fails to protect one. I remember Greenwood telling me that, in some experiments done by drawing different coloured counters from a bag, there seemed to be a bias against one particular colour. On testing, they concluded that this colour had given the counters a slightly greasy surface, so that it tended to escape the sampler's fingers ...
In a letter to Maurice Kendall
The URL of this page is: