On Thu, 20 Jul 2017 04:29:26 -0000 (UTC), Mad Roger
wrote:

On Wed, 19 Jul 2017 22:59:45 -0400,
wrote:

So 20mpg is really from 19mpg to 21mpg, which makes calculation difficult
(when I was looking up the wheel size stuff earlier today, Consumer Reports
said as much so that's why I'm curious how close you think you can get to a
repeatably precise figure (yes, both repeatable, and precise).
With the fuel mileage rig I had (for carbureted vehicles) you could
see the differnce made by changing tire pressure by a few PSI if the
wind didn.t change. You could see the difference from winding down a
window.

Take everything I say with a grain of salt because I'd have to dig up the
references of where I got this idea that even the EPA can't get better than
4% (as I recall), and they don't measure volume (since that's inaccurate).
They measure weight. And they used, as I recall, plastic bags filled with a
known *weight* of fuel.

In addition, they ran on test course, and they repeated the tests.
And even then, they couldn't get better than 4% as I recall.

Unless I dig up the reference, that's all hearsay, but, as I recall, nobody
is going to get accuracies and repeatabilities anywhere near decimal places
at home if they're not repeating the tests under extremely well controlled
circumstances and weighing the fuel consumed.
Weighing the fuel is NOT required if the tests are done at the same
temperature - which would be a requirement for accuracy in fuel
mileage measurementin any case due to the difference in air density.