Answer:
Basically one, with variations according to circumstances and subject matter.
We need to start with something, usually a novel observation, but sometimes just a conjecture. Then we need two main components.
A mathematical model of some set of phenomena.
A set of observations or experiments, chosen as far as possible to rule out that model and all of its competitors.
Then we go with what survives, until we have reason to replace it with something even better. For example, there was an early theory of aerodynamics which explained airplane and bird flight, but not bee flight and many other insects. Then an improved theory added vorticity, which is essential for bees to fly. Problem solved. (Contrary to legend, at no time did aerodynamics say that bees cannot fly. Scientists are not accustomed to speak in such inexactitudes.)
Getting rid of ancient superstitions took centuries, once we started to get hold of the method, but it has been getting much better since.
Quantum mechanics began with novel observations showing that the classical understanding of black body radiation was impossible, and revealing several puzzles about the photoelectric effect. The first was known as the Ultraviolet Catastrophe. Planck, Einstein, Bohr, and dozens of others supplied pieces of the math that eventually became the Standard Model, working to incorporate many other novel observations along the way (quarks and gluons, dark matter) and making a multitude of predictions in various versions, many verified (notable the Higgs boson) and many falsified (the decay of the proton).
The Copernican heliocentric model of what we now call the solar system began with a number of unexplained observations and a number of contradictions with observations in the geocentric model. However, even though the Copernican mathematical model gave more accurate results than previous systems, it was not accurate enough. Kepler determined the correct orbits, and Newton worked out the derivation of those orbits from inverse square gravity. He and others were then able to explain far more phenomena, such as tides and comet orbits, and predict the existence and orbit of Neptune.
Darwin came out with two driving forces for evolution (which was already accepted as fact by scientists at the time). They were Natural Selection and Sexual Selection, both having to do with the probability of having offspring. But he had no idea how genetic inheritance worked at the cellular level. Mendel worked out some of the mathematical laws for genetics, but had no idea how they worked. Molecular biology answered those questions, and raised many more. We are making amazing progress on a multitude of them. We have a mathematical model now for inheritance and for protein coding and synthesis. We do not have an adequate model for protein folding and function.
And so it continues. The models continue to improve, the observations continue to improve, and the experiments continue to improve. More and more incorrect theories fail, and the survivors get better and better.
The biggest problem is the conflict between Quantum Mechanics, defined only in flat Minkowski spacetime, and General Relativity, defined only in curved spacetime, based on the Riemann tensor. This is only a major issue inside black holes and in the earliest Big Bang epoch, before cosmic inflation at 10−32 sec. General Relativity by itself predicts singularities in both cases, and Quantum Mechanics, specifically the Uncertainty Principle, says Nuh-uh. No particle can be confined in a space smaller than its wavelength.
Explanation: