3 7 8 D O C U M E N T 4 1 4 F E B R U A R Y 1 9 2 9 414. To Chaim Herman Müntz [Berlin?,] 25 February 1929 Dear Mr. Müntz, Your discussion doesn’t convince me, since the equations cannot be brought into the canonical form. In the case of covariant equations, the assertion of setting to zero of a tensor with as many components as field variables cannot be maintained. For one could choose the coordinates in such a way that, e.g., four of the can be fixed everywhere in space. Then for the remaining 12 h, there are no fewer than 16 differential equa- tions. One cannot claim in general that such a system has nontrivial solutions that even seems to be rather implausible.— I am, however, quite convinced that my system of equations contains no imper- missible limitation of the solutions, although I have no real proof. So I say to myself: The system [1] alone can be postulated, owing to the identity . If arbitrary electromag- netic equations were adjoined, then it would naturally not work. The equations , however, have the property that they follow from when . It cannot be assumed that for they contain too many conditions, although they are not derivable from . As a result of this circumstance, it appears plausible to me that there is a fourfold identity not for alone, but rather for the and the together. (However, I don’t know how to search for it.) Indeed, the electromagnetic equa- tions are adapted to the others, and this adaptation must be expressed in the exis- tence of a connecting identity. Kind regards, your A. Einstein P.S. In the centrally symmetric case, the adaptation of the electromagnetic equa- tions must express itself in the fact that the gravitational mass and the electrical mass can be chosen independently of each other (apart from the fact that solutions with electrical charge must exist). h s  G  0 = G  / 0  W //  0 = G  0 =  0   0 = G  0 = G  G  W // 