Computational chemist here: I agree. I still put in long hours sometimes, but my "lab" is just one ssh away, so hours actually in the office are a bit flexible.
That may certainly be true, but I wasn't just talking about in-office hours - the labs I'm talking about were hardware type labs (instrumentation, laser labs, that sort of thing). It's possible everyone was just going home to do data processing, but I got the impression that people worked maybe 40-60 hours/week mostly depending on what their teaching load was like and whether there were unusal time-sensitive demands on their experiments.