The protein misfolding avoidance hypothesis explains the universal negative correlation between protein abundance and sequence evolutionary rate across the proteome by identifying protein folding free energy (ΔG) as the confounding variable. Abundant proteins resist toxic misfolding events by being more stable, and more stable proteins evolve slower because their mutations are more destabilizing. Direct supporting evidence consists only of computer simulations. A study taking advantage of a recent experimental breakthrough in measuring protein stability proteome-wide through melting temperature (Tm) (Leuenberger et al. 2017), found weak misfolding avoidance hypothesis support for the Escherichia coli proteome, and no support for the Saccharomyces cerevisiae, Homo sapiens, and Thermus thermophilus proteomes (Plata and Vitkup 2018). I find that the nontrivial relationship between Tm and ΔG and inaccuracy in Tm measurements by Leuenberger et al. 2017 can be responsible for not observing strong positive abundance–Tm and strong negative Tm–evolutionary rate correlations.