Turns out I actually did something ****** stupid. When I was dead bugging this, there was a solder spike across the series capacitor between the tank and the oscillator. So what's basically been happening is the coil has been happily sitting there warming up due to the DC bias across it. Tapped it with the iron and it's nowhere near as drifty as it was. Collecting data for this cock up now.
Well there goes another 3 hours of my life and another bald patch scratching at my head