Clarification on Error ysis Question?

I would expect that the errors would be distributed normally. So, for +/- 1 sigma error bars, I'd expect to see the line p through at least +/- 1 sigma of the samples. That's about 68%.