[brucepick]

What I'm wondering is, how accurate and precise are those scales at a track?

I calculated that you need to detect a weight difference of about 4 oz or 1/4 lb, to detect a difference of about 1 mpg over a 100 mile run at approx 50 mpg.

Unfortunately that spreadsheet isn't here where I can get at it. I may be slightly mis-remembering some of the particulars - but I did end up with a delta of approx 4 oz. as being what you'd need to be able to detect.