The scientific process is one of making careful observations, describing and classifying those observations, and then theorizing a mechanism to explain it. Biologists, for example, spent a long time in the describing and classifying phase, organizing plants and animals into species, genus, etc., but they had no theory to explain this structure until Darwin came along.

A scientific theory attempts to explain, or at least to make some sense out of the observations. For example, we notice that objects tend to fall toward the ground. If we make careful measurements we will find that they fall in a manner that can be described by certain mathematical formulas, but we still won’t know why they do this. The explanation lies in a theory of gravity.

Any explanation is not necessarily a scientific theory, however. Just as in a court of law, there are rules of evidence regarding what is admissible and what is not. Furthermore, the theory itself must have certain properties in order to be even considered. The most important one is that the theory must be testable. If a theory makes no testable predictions, we have no use for it. It must predict what will be seen in an experiment or an observation. Then, that experiment or observation can be carried out to support (or tear down) the theory. For instance, Newton’s theory of gravity predicts that the Moon will have less gravity that the Earth (because it is smaller), and therefore objects will not fall as fast on the Moon. The astronauts who visited the moon have verified this.

A test, however, can never absolutely prove a theory. A theory by its very nature attempts to explain all such phenomena. A theory of gravity, for example, explains all falling objects, not just the one that I am measuring right now. But I can’t measure all falling objects. Even if I test a million of them, I won’t know for sure that the next one will behave as the theory predicts. After a million tests, though, I can be pretty darn sure, if not strictly 100% certain.

On the other hand, a theory can be readily disproved. If one measurement gives a result that disagrees with the theory, then the theory is wrong. This assumes, of course, that we are confident of the reliability of that measurement—we don’t want to throw away a perfectly good theory because somebody did a sloppy experiment. That is why we usually require experiments to be repeated by other scientists before we take it seriously.

Another interesting property of scientific theories is that they will never be complete. The answer to one question always raises new questions, which means that we will never know everything. More fundamentally, we don’t even believe that our theories are exact descriptions of nature. A theory is merely a model, or metaphor. We are saying “the universe behaves like this” not “the universe is this.” As theories evolve, the model gets more accurate. Newton’s theory of gravity was as accurate as anything we could measure before the 20th century, but then Einstein gave us a theory of gravity that is even more accurate, and today we can make measurements that show that Einstein’s theory is indeed better than Newton’s. There is no doubt, though, that someday Einstein’s theory will be replaced by something even more accurate, and the precision of our measurements will advance to the point where we can tell the difference.

The Role of Math in Physics

Math is the language we use to describe the logical relationships between measurable physical quantities. Putting numbers on things is more accurate than describing them in words (at least this is true for the kinds of things we measure in physics—there is still a place in the world for poets). Math is a precise and unambiguous language. By using mathematics to describe the behavior of physical quantities, we can describe them in a way that everyone understands the same way (those who speak the language, anyway).

How Research is Done

Stripping a Problem Down to its Fundamentals: Reduce or control things unrelated to what you want to investigate

Events in real life are affected by many factors, and we cannot measure or control all of them at once. Instead, we try to focus on what is fundamental. When measuring falling objects, we will find that air resistance plays a role. But this is a complication not related to the fundamental force of gravity. By measuring relatively dense objects at speeds that are not too high, we can reduce the impact of air resistance to the point where it will not interfere too much with our measurement of gravity. Of course, if air resistance were the thing we wanted to measure, we would strive to make it the dominant factor.

You can never completely eliminate undesired factors. All you can do is try to minimize them. They will still be there, though, and they will mess up your experiment to some degree. But with careful procedure and statistical analysis it is possible to separate the crucial results from the background noise.

Understand the limitations of measurements

Just as no theory is 100% accurate, no measurement can be either. Any tool that we use has its limitations, and so we  do experiments with the understanding that our results are only meaningful within the limitations of our precision. This motivates us to constantly strive to increase the precision of our experiments, and that in turn stimulates the theoreticians to refine their theories.

Document everything: Make it repeatable

An experimental result is of no value unless it can be communicated to other scientists, and in a way that makes them trust the result. This means that the experimenter must not only follow a carefully planned procedure, but he or she must also carefully document that procedure. An experiment is generally not considered valid until other scientists have independently reproduced the same result. This will not be possible without a detailed description of the experiment.