Error Propagation when dividing histogram (by a constant, by bin width and by another histogram)

Hello everyone,
I’m quite new to root, I hope my question is not really obvious. Let’s say I have two histograms, h1 and h2. I get their error using Sumw2().
But how should I handle error propagation and how should I do the following operations?:

  1. Dividing them by a constant N, h1/N and h2/N.
  2. Dividing each entry by its bin width, because I have variable bin width.
  3. Doing the ratio h1/h2 (this one is the one I know how to do, I have seen that I should do h1->Divide(h2).
    Thank you very much for your help.

Hi Òscar,


the relative bin errors should remain unchanged. The absolute errors should scale down by N.


the relative bin errors in the resulting histogram should equal the square root of the sum of squares of errors in bins in h1 and h2.