Are Atoms Real?

I gave a talk to a group of parents last Friday and asked whether we should believe atoms are real. The group overwhelmingly thought it was a silly question until one parent, clearly a scientist, piped up and informed everyone that atoms are not directly observed, but merely a part of our best current model.

Models, I mentioned, are not true or false, but better or worse. A model airplane may be good if it looks like an airplane, but it may not be as good as another model that bears a stronger resemblance. Scientists use models all the time, indeed many contemporary philosophers of science see scientific theories as sets of models. A theory is successful if the actual system in the world bears a significantly strong resemblance relationship to the posited model, where the resemblance is strong if it allows us to make a wide enough array of successful predictions that it suits our needs.

Of course, every theory will eventually be replaced with a better theory. What does this mean for the way we understand the relationship between theories and the world?
Realists contend that we have good reason to think that we should understand our scientific theories literally, that is, we have good reason to believe that the world actually looks the way described by our best theories. Instrumentalists hold theories to merely be useful tools in providing images by which we can work with systems and produce further results, but one should not confuse a successful tool with a description of some underlying reality.

So, should we believe that atoms are real and if so, why?