3 1 8 D O C U M E N T S 3 3 2 , 3 3 3 D E C E M B E R 1 9 2 8 332. To Julius Mueller-Alberti[1] [Berlin/Gatow?, after 12 December 1928][2] Your plan seems reasonable and welcome, like every effort seeking to counter through positive work the narrow nationalist attitude still prevailing in the school.[3] 333. To Chaim Herman Müntz [Berlin?,] 13 December 1928 Dear Mr. Müntz, The identity for the  can be found as follows: One starts with the identity for [the case of] :[1] . By cyclic permutations of k, l, m, one obtains two other equations. Then all three are added. The first term gives, together with the second, once cyclically permuted term: . We combine the third term together with the fourth, twice cyclically permuted: or + Thus, one initially obtains + or + (I) This is still not a genuinely covariant form of the identity, since differentiating with regard to m does not give a “covariant” derivative (we would write those as ). However, we have . (1) If we permute cyclically and add, then on the right, owing to (1),[2] the terms orig- inating with the first two terms on the right in (1) cancel out. If we combine the third with the once cyclically permuted fourth, then we obtain or . curvature 0  i l m – i m l  i l  m i m  l – + + l m  –i i l  m  i l  m – l i  m i l m – . –. –   i   ml . . + +   0  i l m  + . . +     i  lm . . + +   0  i kl m i l m i l m  l   m i i l  m –    i  lm – +  i l  m i l  m –   – l  m –i