6 1 8 D O C U M E N T 4 0 0 N O V E M B E R 1 9 2 6 400. From Gilbert N. Lewis [Berkeley, California,] 3 November 1926 My dear Professor Einstein: It was a great pleasure to me to receive your letter and to read your very inter- esting remarks about some of the problems which are troubling me. Owing to our isolation here I am afraid I spend a good deal of time puzzling out questions which have already been solved by others. I wish that I could hear more frequently the informal opinions of those who are studying the same problems that I am. For the last year or two I have been completely absorbed in the problem of radiation, look- ing at it from the standpoints of optics, thermo dynamics, probability, and quantum theory. I have gradually formed a view of the problem which has answered succes- sively the various questions which have occurred to me, but whether it will give an answer to questions which may occur to others I do not know. I start with the fundamental idea, which I owe originally to yourself, that when an atom loses energy by radiation this energy is not disseminated in all directions, but is taken up in its entirety by some other atom, so that if we have a small number of atoms radiating to one another all the changes in the state of the system which have occurred in short interval of time (with the possible exception of the radiant energy still in transit) can be stated by means of a finite and small number of data which give the several changes in the positions and the momenta of the atoms and their parts. In other words, when we are dealing with a small number of atoms over a limited period of time, the changes in the microscopic state, including changes due to ra- diation, can be completely expressed in a few words, or with a few data. I postulate further that for every process there is an inverse process identical, except for sign, to the last detail, and that in the state of thermal equilibrium every process and its inverse are occurring with equal frequency. It does not seem possible to reconcile this idea of reversibility to the last detail with the classical picture of radiation. In that picture energy leaves a radiating atom in a diverging spherical wave, but an atom does not receive energy by means of any similar spherical wave converging upon the atom. To specify the course of a radiant wave from a single atom and what eventually becomes of its energy would require an infinite number of data. While therefore with the classical theory of radiation it might still be possible to keep the idea of Gibbs and Boltzmann that entropy is the measure of probability of certain groups of microscopic states, these microscopic states, even in the case of one or two radiating atoms, would be extremely numer- ous, or infinite.