Why should a control group differ from an experimental group in only one variable?

That way you know if there was a difference, you can tell what caused the difference. If you were to vary say 10 things, you wouldn't be able to say what of the 10 variables caused the effect.