D O C U M E N T 2 2 9 J U N E 1 9 2 8 2 2 7 The thus transform exactly like the matrix potentials , and one could eas- ily arrive at a false conclusion. Just as before, it namely follows that Hence, putting , one would obtain . The error is due to the fact that , since indeed, !! and that is good, since we have thus far not made explicit use of the three conditions 8.) . Instead, one has to proceed as follows: It holds that: or , with vectors in the Euclidean representation space, 9.) . Unfortunately, it is getting rather late, dear Professor, and I can’t calculate it any further but I presume that the Christoffel tensor formed with your object has precisely the properties that we are looking for. It of course has less symmetry than the usual Riemannian tensor, and thus our earlier irrational covariant directly as a Christoffel tensor will allow your parallel displacement to be reproduced.— Addendum: I used equation 1a.) for determining the automatically in the case that , one can set or something similar. . Hoping to be able to greet you, dear Professor, in the best of health next Tuesday at 11:30,[3] I send my warmest greetings to you and to your esteemed wife. Yours very gratefully, C. Schwarz P.S. A mathematical chiming is stealing softly through my mind how splendidly does physics shine down upon me, and Planck’s bald pate—behold, it is smiling at us as well! Heine-Goethe-Schiller[4] Q S Q SQ S –1 = Q Q x --------- - Q x ---------- – Q Q Q Q – + Q HP H –1 P P = Q 0 Q H----------- H –1 x - = SS' 1 = ss = r srslr l = ssl l = g 12 g 23 g 31 0 = = H g 11 0 0 h 1 0 g 22 0 h 2 0 0 g 33 h 3 0 0 0 h 4 = h g 4 g ------------ -