The full story of how we know what we know about the existence and structure of atoms is fascinating, complex and, perhaps fortunately for you, too long to go into complete detail here. This is partially because much of it is considered physics - although in fact where physics ends and chemistry begins is rather arbitrary. What we do want to do is to consider a number of key points that illustrate how our ideas of atoms arose and have changed over time. We will present the evidence that has made accepting the atomic theory unavoidable if you want to explain, and manipulate, chemical reactions and the behavior of matter. Atomic theory is an example of a scientific theory that began as speculation and, through the constraints provided by careful observation, experimentation, and logical consistency, evolved over time into a detailed set of ideas that made accurate predictions and was able to explain an increasing number of diverse, and often previously unknown, phenomena. As new observations were made and became well established, atomic theory was adapted to accommodate and organize them. A key feature of scientific, as opposed to other types of ideas, is not whether they are right or wrong, but whether they are logically coherent and make unambiguous, observable, and generally quantitative predictions. They tell us what to look for and what we will find if we measure it. When we look, we may find the world to be as predicted (confirmation of the ideas) or something different. If the world is different from what our scientific ideas suggest, then we assume we are missing something important - either our ideas need altering, or perhaps we are not looking at the world in the right way. As we will see, the types of observations and experimental evidence about matter have become increasingly accurate, complex, and often abstract (that is, not part of our immediate experience). Some of it can be quite difficult to understand, since matter behaves quite differently on the atomic and sub-atomic (microscopic) scale than it does in the normal (macroscopic) world. It is the macroscopic world that evolutionary processes have adapted us to understand (or at least cope with) and with which we are familiar. Yet, if we are to be scientific, we have to go where the data leads us. If we obtain results that are not consistent with our intuitions and current theories, we have to revise those theories rather than ignore the data.

However, scientists tend to be conservative in revising well established theories because new data can sometimes be misleading; this is one reason there is so much emphasis placed on reproducibility. A single report, no matter how careful it appears, can be wrong; the ability of other scientists to reproduce the observation/experiment is key to its acceptance (this is why there are no “miracles” in science.) Even then, the meaning of an observation is not always obvious or unambiguous; more often than not a "revolutionary observation" has a prosaic, that is simple, unrevolutionary, and often boring explanation. Truly revolutionary observations are few and far between (and one might argue, getting rarer as we approach an increasingly complete scientific understanding of the world in which we live. [optional link]) This is one reason that the oft quoted statement by Carl Sagan is oft quoted: "Extraordinary claims require extraordinary evidence." In many cases where revolutionary data is reported, subsequent studies often find that results are due to poor experimental design, sloppiness, or irrelevant factors - the fact that we all do not have cold fusion energy plants driving perpetual motion refrigerators in our homes is evidence of that.

A common misconception about scientific theories is that they are simply “ideas” that someone came up with on the spur of the moment. In everyday use, the word “theory” may well mean an idea or even a hypothesis or working assumption, but in science the word theory should be (but often isn't) reserved for explanations that encompass and explain a broad range of observations. More than just an explanation, however, a theory must be well tested and make clear predictions relating to new observations or experiments. For example, the theory of evolution predicts that when one looks at the fossil record one will find evidence for animals that share many of the features of modern humans. This is a prediction made before any such fossils were found; in fact, many fossils of human-like organisms have, and continue to be discovered. Based on these discoveries, and comparative analyses of the structure of organisms (a field known as cladistics), it is possible to propose plausible "family trees" (phylogenies) connecting different types of organisms - modern molecular genetics methods, particularly genome (DNA) sequencing, have confirmed these predictions, and produced strong experimental support for the current view that all organisms currently living on the earth are part of the same family, that is they share a common ancestor, an ancestor that lived billions of years ago. The theory of evolution also predicts the older the rocks, the more different will the fossilized organisms found be from modern organisms. In rocks dated to be 410 million years old, we find fossils of various types of fish, but not fish that exist today; we do not find evidence of humans, there are, in fact, no mammals, no reptiles, no insects, and no birds.

A scientific theory is also said to be falsifiable, which doesn’t mean that it is false, but rather that it may be proven false by experimentation or observation. For example, it would be difficult to reconcile the current theory of evolution with the discovery of fossil rabbits from rocks older than 300 million years ago. Similarly, the atomic theory would require some serious revision if an element were discovered that did not "fit" into the periodic table; the laws of thermodynamics would have to be reconsidered if a successful perpetual motion machine were to be developed. A "theory" that can be easily adapted to any new evidence has no real scientific value.

A second foundational premise of science is that all theories are restricted to natural phenomena, that is phenomena that can be observed and measured, either directly or indirectly. Explanations that invoke the supernatural, or the totally subjective, are by definition not scientific, since there is no imaginable experiment that could be done that might provide evidence one way or another for their validity. In an important sense, it does not matter whether these supernatural explanations are true (in the deepest sense) or not, they remain unscientific. Imagine an instrument that could detect the presence of angels – if such an instrument could be built, angels could be studied scientifically; their numbers and movements could be tracked and their structure and behaviors analyzed, it might even be possible to predict or control their behavior - they would cease to be supernatural and would become just another part of the natural world. Given these admitted arbitrary limitations on science as a discipline and an enterprise, it is rather surprising how well science works in explaining the world around us. At the same time, science has essentially nothing to say about the meaning of the world around us - although it is often difficult not to speculate on meaning based on current scientific theories. Given that all theories are tentative, and may be revised or abandoned, perhaps it is wise not to use scientific ideas to decide what is good or bad, right or wrong (morally).