Philosophers of science tell us that there is no such thing as a naked observation. How we design an experiment and how we interpret the results are embedded in a constantly expanding web of beliefs and assumptions. As Einstein said, it is the theory that allows us to see the facts. Still, when we see two foil leaves in a jar on a laboratory bench push each other apart as though repelled by like charges, or a dot on a plate in a vacuum tube glowing as though struck by a particle beam, it seems perfectly reasonable to believe in electrons. This phenomenon called electricity arises in so many contexts and can be so simply manipulated -- all you need are batteries and wires -- that it seems very much a part of our world.
But as we go from electrons to positrons to neutrinos to quarks, and from gas-filled discharge tubes to cloud chambers to accelerators shooting beams at detectors so elaborate that they are among the most complex, delicate devices ever made -- it becomes harder to be sure that experimenters are simply observing what the theorists predict. There is a wide gulf between the beholder and the beheld, consisting not only of millions of dollars worth of detecting equipment but the complex of theories with which the equipment is designed, its results interpreted.
In even the simplest experiment, there are the random disturbances we call noise, and one must always make a judgment of when enough pains have been taken to reduce it. To register the outcome of rare, vanishingly tiny events that last fractions of fractions of seconds, a detector must be as sensitive as possible. But the more sensitive the equipment, the more subject it is to noise. There is a constant trade off between capturing the feeble signal that you seek and drowning it out. When the results are finally in, a judgment must be made. Which is data, which is noise? Experimenters trust in their ability to distinguish signal from noise, but there is always the danger of seeing pictures in the clouds.
Even if we can separate foreground from background, we are still left to wonder: Did nature cause the reading, or was it somehow hidden in the design of this very complicated machine? A theory requires a particle and there is a race to find it. The detector is built and then tuned and retuned until, lo and behold, the predicted effect is observed -- the effect, not the particle itself, which does not live long enough to leave a track. The best we can say is that the hypothetical particle, acting according to theory, interacted with other hypothetical particles, whose existence is also built from a long chain of inferences, and at the end of this series of hypothesized reactions, photons or electrons were produced -- the two particles we understand the best. Photons cloud photographic plates or collide with photoelectric cells, producing electrons. And it is electrons that drive our gauges, whose readings we take by bouncing photons from the dial into our retinas, where they generate electrons again, sending signals to the brain. Everything, it seems, comes down to a dance between these particles of electricity and these particles of light.
Looking back, we assume that the newfound phenomenon was there all along and the experimenters cleverly ferreted it out. Or, if the quarry is never found, the theory that predicted it still might be saved, for awhile at least: perhaps the particle was just too massive to produce in existing accelerators -- or in any machine that could be conceivably built. In this domain the connection between map and territory is very subtle. Some philosophers worry how easy it is for so delicate a science to become infected by what they call retrospective realism. The experimental design that produced the right result is retrospectively taken to be correct; the ones that failed to find the phenomenon are judged to be mistaken. The particle exists because it was verified by experiment; the experiment is deemed to have been designed correctly because it found the particle.
In an accelerator experiment, more data is produced than we could ever hope to interpret; it is sifted with computers programmed to look for the patterns the theorists have decided are important. Theory restricts the search space. But maybe more important truths lie in what we thought was noise. To take the most skeptical stance, there is always the possibility that we are simply building big machines, more complex than we can completely understand. We are studying their behavior under different conditions. When an experiment is replicated at another accelerator, it is simply a matter of building or tuning another machine until it behaves in a similar manner. Perhaps we are interpreting machines, not nature.
And perhaps, in our inevitably imperfect ways, we are inching steadily toward a better and better picture of the subatomic world, a realm so alien that it is amazing our minds can enter it at all. Physicists are quite aware that their instruments are not transparent windows on nature. As we move further and further from everyday phenomena, which are mysterious enough, to studying events that last for microseconds -- and then only under the most assiduously controlled conditions -- it takes faith as well as ingenuity to unearth these hidden orders. The faith is driven by the universal passion to find symmetry in the world.