What about:
https://en.wikipedia.org/wiki/Gregor_Mendel#Mendelian_parado... Apparently the numbers of the second generation are too good to be true.
https://en.wikipedia.org/wiki/Oil_drop_experiment#Millikan's... Apparently the viscosity was wrong, and then everyone else made corrections to get a similar result.
> What does that have to do with this situation?
The problem is how to document error without overwhelming honest authors. Imagine a nightmare with a DMCA like process, where anyone can can fill a retraction request and the authors have a week to reply. [The data is in an obscure folder in a notebook that is dead since 5 years. Most of the processing was done by a guy that is now working in the industry for x10 salary.] [Assuming you didn't work with mice, and you must resurrect them to fill the additional data asked in the retraction request.]
An alternative is let the editors ask a new reviewer to make the decision, but everyone has horror stories of reviewers that made bad reviews in spite the manuscript was correct. Then what? Ask the authors again to defend the paper?
The current method is that anybody can publish a "comment" if they find a journal that agree to publish it.
