First, apologies to Shari, I only did a so-so job of keeping my mouth shut in class today. I will try to do better at keeping mum in future. Despite that, I do think we had a good discussion which raised a lot of issues that are well worth thinking about. A few further thoughts, in no particular order:
1. Darwin begins his introduction by emphasizing that he's not going off half-cocked--he's thought long and hard about the species problem and isn't being hasty. Clearly, Darwin was concerned with the reception he was going to get.
2. I find it both impressive and amusing that Darwin considered this book an abstract, far too short to include the many facts he'd collected to support his hypothesis. One can easily understand his reasons (see #1), but one can also be glad that he had an editor who recognized that at some point additional facts stop adding weight to the argument and instead end up obscuring it.
3. Interesting, in light of modern arguments about whether some version of creationism can/should be taught in science classes, to see "special creation" treated as a testable scientific hypothesis (and rejected not so much because the facts disprove it, but because of its failure to explain phenomena that Darwin's theory explains). Many scientists today would argue that creationism is not a scientific hypothesis at all, so that it doesn't make sense to talk about evaluating it as such. My own view is that creationism could be (and has been) pitched in both ways. For instance, if a creationist says that the world was created by God to appear as if evolution by natural selection had occurred (e.g., God created fossils in order to fool paleontologists), then that's not really scientific since the "hypothesis" is defined as "whatever's consistent with any conceivable evidence."
3a. There's a connection between #3 and long-standing philosophical issues of evidence and confirmation. Is it always bad for a theory to be consistent with the evidence "by definition"? What if a theory is developed in order to explain certain known facts, as Darwin's theory was? To what extent (if any) do those same facts count as evidence in support of the theory? After all, if the theory was developed in order to explain known facts, it couldn't possibly be inconsistent with those facts, so isn't the theory guaranteed to be "supported" by what's currently known? Put this way, it sounds like previously-known facts shouldn't count as evidence in favor of a theory. But there are many, many cases--The Origin being one example--in which known facts have been interpreted as supporting a theory that was developed to explain them. What justifies this? Is it that Darwin's theory can explain a wide variety of apparently-disparate facts, thereby "unifying" them? Does "unification" count as evidence in favor of a theory? Think of the common phrase "That explains a lot"...
4. Darwin is clearly worried that he can't offer a mechanistic theory of variation and inheritance. But does he really need one? This gets back to the issue of how to establish the domain of a scientific theory--how does one decide exactly what phenomena a given theory should explain/predict/unify? Why couldn't Darwin have said, without apology, "Organisms vary, and offspring tend to (but don't necessarily) exhibit the same variations as their parents. My theory shows how natural selection, acting on heritable variation, can give rise to new species. My theory is silent on where variation comes from in the first place, and why it is inherited. For my purposes, I will simply assume the brute empirical fact of heritable variation, which will hopefully be explained in future by some obscure monk." After all, every scientific theory takes for granted something that is explained by some other theory. For instance, the Modern Synthesis of Darwinian selection with Mendelian genetics takes for granted the underlying molecular and developmental mechanisms that drive transmission and development, and simply deals with the consequences of those mechanisms. At most, it seems to me that Darwin could've just said something about the features that any theory of variation would have to have in order to allow evolution by natural selection to occur (e.g., blending inheritance won't work, because all variation would be quickly eliminated, leaving none for selection to work on).