I did some work early this semester for a new physics course I've been teaching trying to distinguish between scientific laws, theories, hypotheses, etc.
I found that my ideas as to what these words mean do not exactly match what other science educators think. I'm still working on it. Here are some ideas, however.
A scientific theory is not just an intelligent guess like common people imply when they say "it's only a theory" when they speak of General Relativity, evolution, or such things. What common people call their "theory" about something (like that the dancing light over the tree is an alien spacecraft) probably should be called speculation, not even a hypothesis.
For many science educators, a scientific theory is a model useful for prediction. For example, the molecular model or theory of matter treats molecules as little balls that bounce against each other. If the theory/model is good science then it should make correct predictions about experimental results that have never been tested before.
A scientific law is a mathematical statement about the way things are observed to be. It's not subject to debate or further confirmation. It's what all competent observers agree to. It's not as interesting as a theory since it doesn't make remarkable predictions about the unobvious. It only predicts obvious things.
For example, there is a law of Newton's Gravity that objects under normal conditions accelerate according to the equation a = F/m, where F = Gm1m2/d^2. All competent observers agree that this is the observed acceleration to within experimental error. It doesn't make any remarkable predictions about what happens when near Jupiter (you just change the values in the equation and get expected results).
One could also consider a theory of Newton's Gravity that every mass in the universe pulls with a force as if there is an invisible string between them. It's an "action-at-a-distance". On the other hand, there is a theory of GR that mass curves space-time which causes objects to move along geodesics. The theory of GR predicts that light will bend around the Sun a specific amount. If you alter Newton's equation to accomodate light then you can also predict some curving, but it's smaller than what GR predicts and GR predicts what is observed (to within experimental error), so the theory of GR is superior to the theory of Newton. Newton's law, however, still stands correct in the regime in which it was formulated, otherwise it could never have been considered a law. The law of GR (the GR equations) must reproduce Newton's law in the regime where Newton's equation works or GR cannot be true. Newton's equation becomes invalid when you consider exotic objects like neutron stars or black holes.
What do the rest of you think about this distinction between scientific law and theory?
I'll add more in the next post.