Gasoline is made up of a mixture of different hydrocarbons (alcohols are also oxygenated). The lighter fractions, generally with the fewer carbon atoms, tend to be more volatile. Some will evaporate, but not all. It will stabilise when the pressure reaches a certain value depending on what the RVP of the mixture is and the temperature etc.
The calorific value of most saturated hydrocarbons doesn't vary very much, typically around the 43MJ/kg for the liquids used in fuels, ethanol is around 27MJ/kg.
https://www.engineeringtoolbox.com/fuels-higher-calorific-values-d_169.htmlI'm not sure what the typical density of the vapour is, but propane gas is around 1.5 times that of air. I don't think there's anything lighter than C5 or C6 in liquid gasoline, so I'd guess the vapour density is at least that of propane, somewhere between 1.5 and 2x air, so will sit on top of the fuel and stop any air getting to it in the tank. The volume ratio of vapour to liquid is of the order of 250:1 or more, you'll need at most 100 gms or so of liquid to make up 20 or 30 lts of vapour. That will not significantly detract from the characteristics of the remainder of the gasoline. Also light fraction vapour which does exit the tank to the carbon can will get taken into the engine anyway, an evaporative loss system cannot allow breakthrough of vapour to atmosphere, it has to be contained to a high level of efficiency to meet regs. and subsequently consumed by the engine.
As far as your economy readings go, I suppose it's possible that the evaporative loss recovery might be just about enough to make a difference, a full fresh tank will make more vapour and send it to the canister, and this will gradually reduce as the tank gets used up and lighter fractions are driven off. The engine will consume these fractions as the canister purges and the injector pulse width will reduce slightly to compensate, but of course the ECU only records the injector on-time for calculating economy so it will appear better with higher canister loading (that's effectively not measured). If this is what is behind your readings, it's actually a false representation because fuel is being used from the tank, it just isn't going through the injectors. You're still getting bang for your buck, so to speak. Of course this is just educated guesswork, it might be something else entirely.
