I'm reading Kuhn, "The structure of scientific revolutions". I'm only up to chapter 7 so far, and not desperately impressed. Perhaps it gets better later. Anyway, by curious co-incidence, a piece of it is relevant to the "methane debate", recently energised by Keppler et al. (btw, don't miss the clarification). We had a "discussion group" about this at work: mostly ice core folks whose ultimate interest was interpreting delta-C-13 in ice cores; but most of the discussion was trying to make sense of the paper. And the thing that came up repeatedly is that the *mechanism* for the methane production is unknown. Given that, its really not clear what to measure. At some point, the mechanism will be understood, or at least there will be a testable hypothesis, and suddenly there will be a mini-paradigm-shift (K states that P-shifts are not always major revolutions; they are allowed to be little things in their own area).
To illustrate this: the paper measures methane production from both dead and living leaves. The dead stuff produces much less methane (order of magnitude). Is this because the methane in this case is the breakdown product of some molecule produced by the living plant? In which case a time series would be instructive. But the paper doesn't present this. Then again: for living plants, the paper asserts that production is bigger in sunlight. The measurements weren't terribly well controlled, as the paper notes make clear: "sunshine" is defined as what you get in Heidelburg. And then the problem is: if you expose the plant to sunlight, it will get warmer. And elsewhere they note the strong temperature sensitivity of their results. So all of this is rather typical of the early stages: something odd is found, but the measurements are all in a bit of a fog and its not clear exactly what you are trying to measure.
A bit more commentary on the results: all this turns out to come from about 10 living plants, extrapolated to the planet. So the error bars are huge. It may turn out that these plants are quite atypical. Or, they may turn out to be quite typical. In which case, the methane budget is overthrown? Well, probably not that either. The sinks in the methane budget seems to be well known (mostly atmos oxidation); the sources (wetlands, rice paddies, gas leaks) have pretty big error bars. Most of the overall budgets people construct from these various sources add up to about the same number, but that is because people know the number they are supposed to get: viz, the number that makes the current observed increase work out about right. So (especially if the new results come in at the low end of their range) they can be fitted into the existing methane budget quite easily.
[Update: when I wrote the above, I didn't put any numbers in cos I was doing it from memory. But now I have the data: this (may not be subscription). From which you see, firstly, that Nature says the current methane source-sinks is negative (-47 Tg/y; thats based on a range of sources of 500-600 and deciding, for some unknown reason, that 530 is the best value; IPCC seems to use much the same data but get an *increase*, which is correct...), from those estimates: whereas the atmos measurements clearly show an increase (14 Tg/y). So at the very least, flipping that to +47 could be done without surprising anyone: ie, there is room for an extra source of 94 Tg/y. Kepplers estimates are 62-236 Tg/y, so the lower end can be accomodated without trouble: it would even help! And there is even more scope for fiddling (or, put another way, accomodating the new source without changing much else): Nature sources the current balance to IPCC table 4.2. Is the wetlands source 115, 237, 225, 145 or 92 Tg/y? Is rice 100, 88, 25-54, 60 or 53? Confused?]