This reminds me of Feynman's story of the researcher who eliminated possible variables in rat experiments one by one, to find a surprising result (EDIT: To find that rats didn't need smell, spatial relationships, taste, or vision to find the cheese - the sound of the floor was enough) https://www.realclearscience.com/blog/2014/02/the_rat_experi...
I never did manage to trace Feynman's story back to a source, but nevertheless, also something Derek Lowe repeatedly covers for biology/chemistry on "In the Pipeline": a lot of the 'cell lines' or 'reagents' used in lab work turn out to just not be what they're supposed to be, either because they never were or they degraded into something else in storage or something. So to give a recent paper he blogged, there's som…
"This kind of mouse we raised under these conditions, and we changed nothing, but they varied in these ways over time"
I often wonder whether the variables we check against are actually as independent and subject to the central limit theorem as we assume.