I ASK your permission to reply to the arguments brought forward by Prof. Milne (NATURE, Mar. 22) against my treatment of the problem of stellar luminosity. The outcome of my investigation is a formula which predicts the luminosity of a star of given mass and radius (or mass and effective temperature). I have followed the common procedure of first employing special assumptions to make the mathematical answer definite, and then removing the assumptions by calculating the effect of the greatest admissible variation from the conditions adopted as standard—thus obtaining what is equivalent to a probable error of the prediction. The calculations of luminosity must, of course, be considered in conjunction with these estimates of probable error. I think that the discussion in “Internal Constitution of the Stars” of all known sources of uncertainty is exhaustive; and I conclude that the error is not so great as to impair the practical value of the formula. I have shown that differentially the result agrees excellently with observation, but absolutely it makes all the stars about ten times too bright. In view of this remaining discordance, I have been as eager as my critics in searching for possible loopholes or complications. If Milne's scrutiny brought to light any new possibility I should be grateful; but his criticism is not of this type. He claims that there is a fundamental inadequacy in my method, so that not even a rough value can be computed in this way.