I was dithering whether I should post this or not. It is an essay I wrote a couple of years ago about the OP. It seems like it might be considered a wall of text even if it is paragraphed, but, what the hell....
------
Science starts out with facts. Facts are observed (which makes "observations" a useful symonym); through the senses, sometimes with equipment to extend the senses, and anyone, with the necessary senses and a non-biased outlook, will come to the same conclusion on the facts. Often times scientists will set up experiments to ascertain and clarify the facts; again, these are replicable, and the entire establishment of scientific publications is there to preserve the manner in which the facts are determined so anyone else can determine them for himself. There is no denying the facts.
[I am excluding mathematics from the sciences for this essay. Mathematics has a completely different logical system, and contradicts a lot of what will be said below.]
A theory is an inductive abstraction drawn on a collection of facts. It is called a hypothesis until some amount of evidence is gathered for its acceptance (though, for example, some theories, like the nebular hypothesis, anachronistically hang in there).
Given a body of facts, the scientist may intuitively draw out a common theme from within them. Newton observed an apple (at least in the abstract), and the moon and planets in their orbits. He knew Kepler's Laws, drawn from the careful pre-telescopic astronomical observations of Tycho Brahe, and Galileo's laws about the motion of objects. All of these can be explained by an attractive force between all bodies of mass. Newton drew the conclusion, and developed the branch of math for doing so rigorously (integral calculus): that this force was proportional to the sum of the two masses and inverse to the distance between them, and then demonstrated that all the known facts, from apples to Mars, indeed fit the theory. As it turns out, later on one fact didn't - the planet Mercury's orbit wasn't quite what it should have been. At first that was thought to be because there was another planet even closer to the sun which was perturbing Mercury (it was even given a name: Vulcan), but Einstein later improved Newton's theory with his "Theory of General Relativity", which triumphantly corrected the anomaly in this case. It did not demolish Newton's theory, but rather corrected it in extreme cases; he increased the range of facts over which their combined theory holds true.
A theory is inductive. There are two kinds of logic available to create new truths from an old one: deductive logic, in which a known general rule is applied to specific instances, and inductive logic, which is the opposite: a number of specific cases are brought together to create a general rule. Deductive logic always creates true results, as long as the premise, the general rule, is always true. If you consider Newton's theory to be true, then it can be used deductively to predict the orbit of any asteroid or comet newly discovered. This happens all the time in astronomy, and the orbit will always be true as long as the assumptions (such as the orbit not being perturbed by a third body) is true, or unless the body to too close to another high mass body, as Mercury is to the sun, in which case Einstein's General Relativity better explains it.
The fact that theories are inductive makes them flexible and open to change. We saw that a single inconvenient observation which should have been worked out by Newton's theory failed to agree. We found that it represented the first of what would be a whole series of differences with Newton's predictions, and they lead to Einstein's correction to the Theory of Gravitation. It is concieveable it may happen again, even to general Relativity, or to another well-regarded theory. Darwin's Theory or Evolution needed much updating as microbiology made strides in understanding the chemical basis of life; in this case there was no new sweeping improved theory, but rather incremnental changes to a theory keeping up to date. The down side is, of course, that no theory can ever be claimed to be 100% proven, because it is never the case that 100% of all facts about a theory can ever be claimed to have been discovered.
An induction is only as true as the facts upon which it is built, over exactly those facts' range. This implies that the scientist that draws the induction looks diligently for facts that don't fit, because a fact brought to light later by someone else which ought to be covered is regarded as a major embarrassment, unless it can be shown to fall validly outside the range of the theory. Sometimes there is no avoiding this; a new experiment or an observation made with new, more powerful equipment than was available to the original theorist brings an uncovered fact to light (such as Mercury's orbital anomaly, which instruments in Newton's time were not good enough to note).
The activity of science has always favored those who can draw a theory out of the greatest possible number of facts; universal application is the ideal. A theory that applies to many facts is always very compelling, possibly an expression of the ultimate elegance of nature. Likewise, it is always compelling to a scientist to find facts that do not fit; these usually lead, in the long run, to better theories, the discovery of which is a prestigious goal. Science can thus always be seen as a competition, often honorable and friendly but sometimes cutthroat and nasty, involving considerable egos and undying (well, only slowly rotting) fame for the winners. Perhaps the highest accolade for a scientist is to be called a theoretician.
Related to a theory is an hypothesis. In science a hypothesis is often what laymen think of as a theory; it is a scientist's guess, or hunch, about how things work. Call it a potential theory in the making. What will make it a theory is, of course, the gathering of facts which are covered, being able to delineate the range, sometimes investigating and creating new facts by observation or experimentation, drawing all that together in a scientific paper giving all the specifics about the hypothesis, and submitting that paper to the scientific world, implicitly by publishing it in a journal and explicitly by that journal consulting other scientists to examine the paper for faults, a process callled "peer review". Once this is done, a hypothesis will begin to cross the line into becoming a theory based on the affirmation of the world of knowledgable scientists. Or it may crash in flames due to bad induction, unexamined facts, poor procedures in gathering facts, or other flaws discovered by reviewers or even later by the inability of other scientists to be able to replicate experiments.
Occasionally language itself can become a barrier here. The idea that planets can condense from a star's dust cloud was developed by Emanuel Swedenborg in 1734. It was called the nebular hypothesis, and the name, which seems to be euphonious, still hangs on, long after it has graduated to the level of a theory. This would only be an oddity were it not for the fact that anti-scientific forces have intentionally tried to use that specific lapse to confuse laymen.
What, then, is a scientific law? A scientific law is a pithy summary of an aspect of a theory, usually expressed in mathematical terms. Kepler's three laws of planetary motion, for example, are aspects of Newton's Theory of Gravitation.
A scientific model is a method used to study a phenomenon by making it replicable in a simplified framework. For example, one can model the motion of the planets based on applying Newton's theory, and be able to predict, for example, events which may occur in the future. A model is always a simplification, and so caution must always be taken that the simplifications don't substantially affect the outcomes. This has become very important with the advent of digital computers, whch often make models easy to create. The fear is that they might not accurately represent the phenomenon, and thus loose the property of significance.
One particular problem with using computer models is that they involve the digitization of what appear to us to be continuous, analog processes. This digitization makes practical shortcuts of analog circumstances which are not altogether true: for example, no computer can multiply a diameter by the constant pi to get an exact circumference, because computers cannot handle the infinte transcendence of pi's value; it can only get as exact as necessary in a given situation. That is worrisome to people who understand how chaos theory (another theory!) can magnify inexactness over time. Much research is being expended to understand this better.