Sometimes it’s difficult to eliminate external factors like vibration and surface contamination from the experiment… that’s how my physical experiments “go wrong” — they’re simple experiments so there’s not a lot that can actually go wrong.
As Michael said, there are also numerical experiments (using computer code), which can go wrong. Luckily, this helps us write better code in the future.
Too many times to count! I’ve made mistakes with collecting data for experiments; once spent an entire afternoon taking measurements of skulls and somehow managed to not save the data file – not a happy bunny! Things also go wrong when I’m writing code in a computer program called R to try and analyse my data; I usually spend days struggling with error messages only to find that the problem was a comma missing somewhere in the code.
Before I started my PhD I did various lab and field-based experiments which also had their fair share of things going wrong. The worst was probably when I was using sticky traps to try and catch lizards in Greece. We set the traps in a shaded corner of a wall and left them for half an hour before checking them. On one of the days within the space of that half hour a lizard got stuck on the trap, the sun moved so the trap was no longer in the shade and by the time we came back to check the trap the lizard was not moving very much. Luckily we managed to cool it down in the shade and release it but it was still a fairly dazed and sun-weary lizard!
As Michael said, sometimes it’s the things that go wrong which ultimately make your experiments better. So the moral of the tale is don’t leave lizards out in the sun!