As in Fig. 6, but using richness at the galaxy-derived r₂₀₀. This richness adopts a radius uncorrelated with the weak-lensing mass estimate and a different cluster center estimate. A tight scaling is also observed here. O26 is not fitted due to its possibly contaminated weak-lensing mass. Including it in the sample would not alter the scaling relation. Note that in this case we fit mass as a function of richness but plot it with mass on the abscissa to facilitate comparison with Fig. 6. The scaling relation seems to differ from the one derived using caustic masses and an X-ray-selected sample (Andreon 2015, dashed line), but interpretation of the cause is premature owing to the degeneracy between sample selection and mass bias (see text for details).