I can see if one was using the fuel pump approach it can be a problem but my methodology does not rely on filling up the tank (details provided in my TDIClub link) and even so my Fuelly measurements are also independent of the pump equipment as D2 can be filled always to the neck of the fuel tank (non-flammable).
In any case I think we have taken enough bandwidth in this thread regarding MoS2. For follow ups please respond in the oil additives forum where I have posted my results/link. Thanks.
Originally Posted By: Garak
The error bars are still going to be, unfortunately, significantly higher than 3%. I understand your point, and I have "seen" differences myself by doing certain things differently (i.e. E0 versus E10), but 3% far too easily falls within the noise. Gasoline pumps aren't even required to have accuracy under 3%, so that puts the data points beneath the error bars there, all on its own.
The central limit theorem does have its place in fuel economy calculations, but I'd be looking to a much greater sample size that accurately covers off more variables; even then, there are pitfalls. Even for a large fleet with a steady driver pool, there are still variables that should be minimized, including fuel reformulation, even.
For an individual trial, you're stuck with eliminating as many variables as you can, to the point that you need a lab.