I read you can use gas (e.g. Butane) to find a leak. I'm thinking of using an unlit blowtorch, and moving it around the intake area.
Would the instability suggest the leak was isolated to a particular cylinder or so? I.e., if the leak was around the carb mounting plate, it would be low, but static, but if it's one cylinder, then vacuum would be lower when that cylinder should be drawing in air, so the reading would fluctuate?
I guess swapping out the inlet manifold gasket is dead easy, and I could add some sealant when I next fit the carb, and see how that goes. After that, I guess it gets more serious, with valve seals, low cylinder compression, etc.