My point is that you have to take positive action to control for a factor. Since there's a practically infinite number of factors that could have an effect, you

My point is that you have to take positive action to control for a factor. Since there's a practically infinite number of factors that could have an effect, you can't assume that only the experiment variable is changing, because that would require first enumerating an infinitude of factors. You can list the things that you think are most likely to be confounding and make sure that they're either controlled for or randomized, but eventually you have to make a judgment call of when the groups are similar enough. Here's a couple more silly but vaguely possible examples: what if one cage was nearer the window, and UV exposure had a significant effect? Or what if the big-pen rats were in the corner where the air was more stagnant, and rats don't like to get high when it's humid? Nobody would have predicted those as a possible hypothesis, so there would be no effort taken to account for them. That's not poor science, it's a necessary part of the process.