This looks like a valid point. That would suggest that there is slippage, stretching or shrinkage going on in the tire (or maybe a combo of 2 or all 3). I hadn't considered that - I was thinking of it more like a car tire, which has a designed contact patch that is already very flat side to side, so the contact patch primarily gets longer when under load.

In that case the question becomes - would this affect the circumference at the center of the tread? I don't have that answer.

I still think it isn't large enough to get into saying you have to do a rollout weighted down or else it won't be accurate - the error in the measurement technique most of us will be using will wash it out, so neither method should improve accuracy necessarily. My own measurements support this, even though as others point out there's simply not enough repeatability for the difference in averages to be statistically significant. I'd just like to point out that my difference in averages was very small - well under 1% (2mm), and the difference in samples was somewhat larger (+-3mm). Even adding the difference in between the largest outliers gives you a change in rolliout of less than 0.5% - still not real reliable, but probably accurate and definitely not worth worrying about when setting your computer. Yes, more samples would make this a more valid data set. FWIW, I'm the only one who bothered to post any data at all.

JB

PS and for what it's worth, comparing a deflated tire with a properly inflated tire weighted down is specious - as has already been noted, there's good reason to think simply inflating the tire stretches it, increasing the circumference.