>>Fill the tank, zero the counter, drive for a while (the longer the
>>better) refill the tank, and enter the fuel quantity used to refill.
>
>>During this process, we've been counting injector pulses, so we now
>>know that there are X pulses per gallon. (liter or whatever)
>
>How does that work, when the injector pulse widths vary with engine
>load etc.


Worked pretty well on several car computers I've owned. Obviously it's
averaging.. Mind, they were not trying to control the engine, just to
monitor consumption and performance.

I suppose one could collect min and max widths and average, or even
get more sophisticated about it.