Sceptical analysis using the unscaled result of Ref. [8]

Once the $\times 1.52$ scaling factor on the lowest value is removed, the weighted average is shifted down and falls right in the middle of the two most measurements, as shown by the dashed line of Fig. 13,

Figure: Same as Fig. 10 removing the $\times \sqrt {2.31}\ (=1.52)$ scaling applied by the authors [9] to the `error' of the result nr. 5 of Tab. 1.

yielding $m_{K^\pm} = 493.666 \pm 0.005\,$MeV. But a sceptical analysis yields a broader distribution, overlapping the two precise measurements on the sides, thus taking into serious account also the results in between. Having to report the result as average and standard deviation of the distribution we get then $493.662 \pm 0.017\,$MeV (but remember that the complete result is provided by the posterior pdf evaluated by MCMC sampling).

Looking into the details of the inference, we see that, as imaginable, high values for $r_5$ and $r_6$ are preferred ($2.8\pm 2.4$ and $3.4\pm 2.7$, respectively), while the others remain more or less around the prior values of $\approx 1.0$.