Good work. I took the liberty of making a plot with error bars:
Weight Rev/Mile Diameter Circumference
Dunlop 19# 856 24.4” 77 1/4”
Michelin 22# 786 26.5” 83 5/8”
Statistically speaking, if you were to repeat the tests over and over 67% of the results would fall in the span of the error bars and 95% would fall into 2x the span of the error bars.
Sure there's more than just tire size involved, but it's clear that the one tire is better. One other factor to keep in mind is that tread depth also plays a roll--meaning that worn tires typically have lower RR.
Another key takeaway is that 10 mph increase costs 10+ mpg.