I've been using CCI stingers running at 1640fps with adequate accuracy and precision.
Statistically it could be argued that the second data set you referred to Muir are the the more accurate and precise round as the one shot/10 could easily be classed as an outliar and discluded from the data set (using dixons outliar test). If you use xbar +/- the square root of ts over n taking 2 as an approximation of t for a 95% confidence interval (5% significance) then you can see the one shot outside of the 5% significance level is indeed an outliar.
One of the easiest ways to see the effect of barrel warming on shots is to plot the cusum vs shot number on a cusum chart and observe the trend of change in impact point relative to barrel temperature. (now if that didnt send you lot asleep, nowt will
)