Author Correction: Past rapid warmings as a constraint on greenhouse-gas climate feedbacks

In the version of this article originally published, three estimates of equilibrium climate sensitivity (ECS) derived from different sources were used to convert feedback strength into the unitless measuregainon the assumption that these were independent.In fact, these were not independent, and so combining them yields a too-narrow uncertainty range.The authors decided to only use the "very likely" (instead of "likely") range from IPCC WG1 AR6 and treat it as a 90% confidence interval.Additionally, the gain is not normally distributed but is highly asymmetric, as it is the negative of the ratio of two approximately normally distributed variables, feedback strength (c) and the net feedback parameter ( net ), with a non-zero centre.There is no standard way to derive confidence intervals from standard error for such a variable.Therefore, in the correct version, only the standard error of the gain is provided, instead of giving confidence intervals.Besides, since calculating standard error by the error propagation rule requires the input variables to be at least approximately normally distributed, the gain was calculated directly from the net feedback parameter ( net , which is assumed to be normally distributed) corresponding to ECS (which is not normally distributed).The changes implemented have no impact on the calculated feedback strengths, but they do have an impact on the estimated gains.Since confidence intervals are no longer provided for the gains, the comparison is focused on the feedback strengths.The authors would like to thank Dr. B. B. Cael from the National Oceanography Centre for bringing this issue to their attention with advice about the choice of ECS and how the very likely range should be interpreted into confidence interval.