I don't know about rusting steel nails, but in any experiment, the control is an experimental subject that is exposed to all of the same conditions as the main subject except for the one thing that you think will matter.
For example, lets say, you like to make pancakes, and one day you change up your recipie by adding a pinch of salt. The pancakes taste better. Was it the salt? The way to find out is to do an experiment: Make two batches, one with salt, and one without. Try to keep everything else about the two batches as much alike as possible. Mix them up at the same time (or as close as possible). Scoop the flour from the same bin with the same scoop. Get the milk from the same carton. Don't let the batter for one batch sit any longer than the other. Cook them on the same griddle at the same temperature. The batch without the pinch of salt is your control batch.
Do the experiment several times. If the batch with the salt is consistently better than the control batch, then you can reasonably conclude that it was the salt that made your pancakes better (i.e., your hypothesis was valid), but if the salt batch is not consistently better than the control batch, then you'll probably want to go back and think about what else might have been different on that day when you first changed the recipie, because it probably wasn't the salt.