Geology is to a large extent the study of unrepeatable events which happened in the very dim and very distant past. As Kim rather nicely puts it:
That 1.4 billion-year-old granite whose contact aureole I’m studying? It’s done intruding, it’s cold, it’s eroding. It’s not going back to the mid-crust any time soon.
Superficially, this conflicts with the idea that rigorous science is grounded in testing and replication of your results (and, taken to extremes, culminates in the ‘where you there?’ bleating of the more virulent creationists). After all, we can’t run the Earth again to see if we’ve got it right; so how can you ever truly test a geological hypothesis?
It is certainly true that most geological research does not fit easily into the standard picture of experimental science, where you perform laboratory experiments designed to allow you to play around with the controlling variables, and fully explore the behaviour of the system being studied. Geologists don’t study past events directly, but rather work backwards from the traces that these events leave in the rock record. Nature has already run the experiments that we are collecting the data for, but she is usually rather coy about the details: you don’t precisely know the initial conditions (in the case of Kim’s granite, things like the temperature of the melt when it was injected into the crust, and how long the intrusion took to cool), and you certainly don’t have the luxury of playing around with them to explore how different values would affect what you see.
This has led some people to single out ‘historical sciences’ like geology and archaeology from ‘observational sciences’ like particle physics and chemistry. Personally, I find the distinction rather misleading: although the intellectual toolkit required to puzzle out the stories told by billion year-old granites is obviously not exactly the same as the one needed when mixing chemicals in a test tube or firing high-energy particles into each other, when it comes to hypothesis testing, there is not as much difference as you might think. In all branches of science, testing is less about straight repetition and more about the exploration of consequences. Your hypothesis or model is usually based on only a small subset of the possible observations you could make; what predictions does it make about the behaviour of the things that you haven’t yet measured? Real testing involves moving beyond the original experiment into previously unconsidered realms; if physicists just spent their days exactly replicating old experiments, we’d still all be sitting under trees waiting for apples to fall on our heads, rather than flinging space probes at Pluto via Jupiter.
Geology fits quite comfortably into this picture of science: you can’t make a melt crystallise again, but you can look to see if all the minerals tell the same story about starting temperatures and cooling rates – not just the ones you looked at originally. If you’re sampling desert sandstones for palaeomagnetic analysis, you’d expect your results to show that they formed at low latitudes and not at the North Pole. If you’re excavating dinosaur skeletons in Montana, you wouldn’t expect to find signs that humans have chewed on the bones.
There is also another sense in which geology is testable. When studying a particular granite intrusion, or mountain range, or ancient river delta, as well as trying to understand that feature’s specific geological history, we are also attempting to construct a more general model of pluton, mountain belt, or delta formation, one which can be applied to all plutons, all mountain belts, all deltas. We can’t test the predictive power of an orogenic (mountain-building) model by resetting the Himalayas with a slightly different convergence rate or a different thickness of crust, but we can achieve the same thing by taking advantage of the fact that the Earth is a very big and very old place, and has seen many mountain building episodes which are all unique in their own way, thus providing new and different experimental data to test against. For geologists, the Earth is our laboratory, and a repository of data from over 4 billion years’ worth of experimentation in various aspects of tectonics, volcanism, geochemistry, and climatology (to name a few), all in long-term storage in the rock record. By seeking out the results of these ancient experiments, comparing different runs from different locations and times, and using them to predict what else we should find, we are undertaking science in just the same way as any physicist or chemist. Anyone claiming otherwise is just bitter that they don’t get to run around in the mountains and call it work.
That said, I certainly wouldn’t mind receiving a grant to employ the services of the Magratheans…