Well I try to be concise I presume you know
what a Quincunx is, particles falling through
an array of pins into columns at the base. Lets
have 40+ columns for a smooth normal distribution
which repeats every operation. Then any mass to
the side could bias the particles slightly one
way causing a noticable change in the curve. I
envisage a device that can be reset quickly by
rotating backwards so that the particles fall
out of the columns and back into the hopper at
the top ready to fall through the pins again.
Hope this is clearer!