- accumulate

`N`

uniformly distributed numbers, say in range `0`

..`R`

.- substract the mean,

`NR/2`

.- divide by the standard deviation,

`Sqrt(NR^2/12)`

.The larger

`N`

, the better the approximation. Use `N=10`

.Also see http://en.wikipedia.org/wiki/Box-Muller_transformation[^].