Interestingly,

 ax.plot(np.arange(2000, 2010), range(10))

does not result in any offset with current matplotlib. Hence this needs to change anyways for the whole paragraph to make sense.

So suppose it is

ax.plot(np.arange(20000, 20010), range(10))

instead. Then, using ax.get_xaxis().get_major_formatter().set_useOffset(False) does indeed get rid of the offset, just as currently described in that part.

Concerning the "multiplicative offset", another example can be added. E.g. if

ax.plot(np.linspace(0, 3e8, 301), range(301))

then adding ax.get_xaxis().get_major_formatter().set_scientific(False) will remove the "multiplicative offset".

The difference between the two is not well documented anywhere. So it would in addition make sense to add a respective paragraph in the ticks-and-ticklabels section of the text tutorial.

Thanks a lot. If you fix the matplotlib behavior in order to provide 2 independent concepts for offset and scientific notation, it is far better.

But if you only update the docstring, you only provide documentation which do not match the actual matplotlib behavior. So, do you plan to fix the behavior? Is there a related issue i can follow?

Thanks a lot.

Offsets and scientific notation are two independent concepts, and IMO work as they should. This is only a documentation issue. See #17590, #17592.

Definitely not. Else, disabling scientific notation should not be needed to disable the offset.

This said, i did not checked it again since January.

Oh, yes, in fact the issue was on my PR description: #16232

I've commented in #16232 on that.

I see, there is an offset and a factor. Thanks a lot for the clarification.