Back in late September I wrote about a controversial paper in the Proceedings of the National Academy of Sciences. It attracted comment for its way-out-there hypothesis: that caterpillars and other larvae arose through a spectacular interspecies gene transfer rather than through conventional evolutionary processes. And it may have been the last paper to make it into the journal by the now-eliminated "Track III" route, which allowed members to essentially cherry-pick their own reviewers. This paper may well have hastened the disappearance of that system, actually - it created quite an uproar.

At the time, I wrote that the paper's hypothesis seemed very likely to be wrong, but at least the author had proposed some means to test it. Now in the latest PNAS come a letter and a full article on the subject. Both mention the testability of the original paper, and go on to point out that such tests have already been done. The paper is written in a tone of exasperation:

Williamson suggested that "many corollaries of my hypothesis are testable." We agree and note that most of the tests have already been carried out, the results of which are readily available in the recent literature and online databases. Here, we set aside (i) the complete absence of evidence offered by Williamson in support of his hypothesis, (ii) his apparent determination to ignore the enormous errors in current understanding of inheritance, gene expression, cell fate specification, morphogenesis, and other phenomena that are implied by his hypothesis, and (iii) the abundant empirical evidence for the evolution and loss of larval forms by natural selection. Instead, we focus on Williamson's molecular genetic predictions concerning genome size and content in insects, velvet worms, and several marine taxa, and we point out the readily available data that show those predictions to be easily rejected.

And you know, they really should set aside those first three points. Entertaining as it is to read this sort of thing, the real way to demolish a paper like Williamson's is to rip it up scientifically, rather than hurl insults at it (however well-justified they might be). There seems to be plenty of room to work in. For example, Williamson predicts that a class of parasitic barnacle will be found to not be barnacles at all, and to have an abnormally large genome, with material from three different sorts of organisms. Actually, though, these organisms have smaller genomes than usual, and from their genes they appear to be perfectly reasonable relatives of other barnacles.

And so on. Williamson predicts that the genomes of insects with caterpillar-like larval stages will tend to be larger than those without, but the data indicate, if anything, a trend in the opposite direction. His predictions for specific insects don't pan out, nor do his predictions about the genome size of velvet worms and many other cases. If I read the paper right, not one of Williamson's many predictions actually goes his way. In some cases, he appears to cite genome size data that line up with his hypothesis, but miss citing similar organisms that contradict it.

So that would appear to be that. Indeed, as the authors of the latest PNAS paper mention, one might have thought so years ago, since these very authors have shot down some of Williamson's work before. That's the real problem here. I have a lot of sympathy for people who are willing to be spectacularly wrong, but that starts to evaporate when they don't realize that they've been spectacularly wrong. Williamson appears to have had a fair hearing for his ideas, and as far as I can tell, they've come up well short. And while we need brave renegades, cranks are already in long supply.