D O C . 3 6 8 O N U N I F I E D F I E L D T H E O R Y 3 4 9 368. “On Unified Field Theory: Continuation of the Work of the Same Title (These Proceedings I. 29)” [Berlin, after 10 January 1929][1] The paper[2] to which the present article is a supplement contains an error, as was pointed out to me by Mr. Lanczos and Mr. Müntz.[3] While thinking about possi- bilities for its correction, I discovered that the system of field equations was not yet complete in the following, it will be shown that the four conditions … (1) must be added. The whole system then consists of 24 algebraically independent equations, related by 12 mutually independent differential identities. This means that 24 – 12 mutually independent equations exist for the 16 field variables . This is precisely correct, since in a relativistic theory, the number of independent field equations must be 4 less than the number of independent field variables. The incorrect conclusion in the quoted article was the following: The field equa- tions (10 1.c) form a tensor system containing 16 equations for the 16 field vari- ables .[4] These equations must not completely determine the continuation of the h from a section with , since, owing to the general covariance, the continuation of the coordinate system and thus the continuation of four of the must remain undetermined.[5] This led me to conclude, incorrectly, that four differ- ential identities must hold among those equations. In fact, the search for such iden- tities was unsuccessful. If, however, no such identities exist, then the 20 field equations (10 1.c) and (11 1.c) are related only by the four identities (8 1.c), which is not sufficient. The elimination of this discrepancy has so considerably improved the complete- ness of the theory that I now regard it as fruitful to give the derivation of the field equations here once more, quite independently of their previous formulation. Of course I base this on the purely formal results of the earlier investigation.— §. 1 On the identities which are important for the theory. To begin, I recall the identity derived in §2.1.c. (3b 1.c). … (2) By twice applying the divergence-commutation relation (5 1.c) to the tensor , taking (2) into account, or else using the antisymmetric character of with respect to the indices  and , one also obtains the two identities [p. 1] S  V  V  V  + + h       + +  0 = = =  sh  sh x 4 konst. =  sh [p. 2] V / 0  V  V 