This Distribution was made using the profile of the Standard normal distribution revolved about the Z axis. It is an approximation, that I drew free hand (tracing a screen capture from https://www.desmos.com/calculator/2kmx0enkkz) to get a rough idea of how the percent per sigma differs from the percent per sigma by volume. I used Blender to measure the volumes (before making space between each cylinder so that they can nest). The running percent totals (by volume) are roughly as follows:
1 sigma - 36%
1.25 sigma ~ 50%
2 sigma - 81
3 sigma - 97%
4 sigma ~ 100%
Individually they are:
1 sigma - 36%
1 sigma to 1.25 sigma - 14%
1.25 sigma to 2 sigma - 31%
2 sigma to 3 sigma - 16%
3 sigma to 4 sigma - 3%

I made it so that you could arrange 50% of the total distribution by volume in two parts by nesting the first piece inside the second piece. Some things that I learned:
-for a 2D Gaussian the 50%, by volume, mark is at .6745sigma, but for a 3D Gaussian it is closer to around 1.25sigma, which I did not expect.
-like the previous, 1 sigma for a 2D contains 68% while for a 3D it only contains roughly 36%
This wasn't meant to be an exact calculation, nor was it generated using a formula. It does make a decent approximation, and a nice demo piece. I think this helps to demonstrate that distributions are different when going from 2D to 3D.

I would print it at a scale where the biggest ring is about 9cm OD.