It's quite complicated to explain and I will probably do a bad job of trying.
Unless you are pressure coring (which they weren't) you will lose gas as the core comes out of the hole. How you determine this normally is, when the core comes out you put it in the canister and measure the rate at which the gas comes off. You plot this rate against the square root of time and if the coal is behaving itself the result is a straight line.
You know what period of time has elapsed since you starting lifting the core, which is when the core would start giving off gas. So you can extrapolate back to "time zero" and calculate how much gas has been lost between time zero and the moment you got it into the canister and starting recording how much gas was coming off.
This is called the lost gas estimate, or Q1. But it relies on the assumption that the rate of gas desorption while you were bringing the core up from the bottom of the hole obeys the same mathematical relationship that you observer once it's in the canister.
If there has been a change to the core, such as it started out one solid piece of core but then sudden fractured in an "explosive gas release" then that relationship won't hold true. The initial gas coming off will be lower because the core is one solid piece. When it fractures, you now have lots of little pieces of coal core and the surface area is much greater, so the gas coming off that core will be much higher.
When you put it in the canister you will measure this higher rate, and when you trace it back to time zero you will get an answer that assumes it's been at that higher rate the whole time since time zero. But that won't be true because until the coal fractured it was coming out at a lower rate. So you will calculate that you have lost more gas than you actually have, so your Q1 will be higher than it should be, which means your calculated gas content will be higher than it really is, not lower.
That's how it seems to me anyway. Like I say, it doesn't really matter if their reasoning is right or wrong.
To explain how the gas content measurements are underestimating the true gas content by half, you need to come up with a theory that explains how the gas might have been coming off the core much faster in the lost gas phase than it came off the core once it got into the desorption canister. Higher temperatures, higher permeabilities and higher CO2 content of the lost gas are all possibilities.
As an example, it may be that in the reservoir the coal has 25mD permeability. This creates lots of channels for the gas to come off. But as the core comes out of the hole, the coal expands as the pressure is reduced, and all the cleats close up. The permeability of the core may drop basically to zero. Now you have the reverse of the situation mentioned by mir911, where the core that goes into the canister is making gas more slowly than it was in the reservoir. Then when you do your lost gas calculation and trace it back to time zero, you underestimate your gas content because you're not accounting for the much faster rate of gas that happened in the lost gas phase.
Gas comes off faster in higher temperatures too, and CO2 comes off faster than methane because it's a smaller molecule (has a higher diffusivity).
All in all I have no idea what is causing it, but the evidence from the production testing does suggest that the lost gas estimate is underestimating the gas content of the coals in the reservoir.
(20min delay)