Recently, Jiayi Hu and co-workers reported that they did not resolve any thermal Hall signal in La
2CuO
4 by `high resolution' measurements, setting an upper bound of
|\kappa_{xy}| <2\times 10^{-3}~Wm
−1K
−1 at 20 K. Two points have apparently escaped their attention. First, thermal Hall signals with an amplitude well below this resolution bound have been detected in disordered perovskites. Second, the longitudinal thermal conductivity of their sample is significantly lower than the La
2CuO
4 sample displaying a thermal Hall signal. We find that a moderate reduction of
κxx in SrTiO
3 is concomitant with a drastic attenuation of
κxy. A trend emerges across several families of insulators: the amplitude of
κxy anti-correlates with disorder.