There's a great story of a histogram of heights in Napoleon's army having two peaks, eliciting all sorts of theories. Reality was that height in the army was normally distributed, but that the data was collected in centimeters and people were looking at a histogram binned by inches. The middle bin contained only two centimeter counts while the bins on either side contained three each, thus having dramatically more counts.
Comments
Is this basically just an effect of quantization aliasing?
In other words, rounding error.
There's a great story of a histogram of heights in Napoleon's army having two peaks, eliciting all sorts of theories. Reality was that height in the army was normally distributed, but that the data was collected in centimeters and people were looking at a histogram binned by inches. The middle bin contained only two centimeter counts while the bins on either side contained three each, thus having dramatically more counts.
Would love to find the citation for that one...