l x x v i I N T R O D U C T I O N T O V O L U M E 1 3 assumption as well, and briefly discussed a generalized action in which a term depending on the antisymmetrized Ricci tensor is added to the purely gravitational Riemann curvature scalar. The manuscript ends with the derivation of a generalized compatibility condition that would hold in the case that electromagnetic fields are present. It is unclear whether Einstein ever sent this manuscript off for publication. Ac- cording to the travel diary, on the afternoon of 13 January he had “found a fat fly in my electricity soup. A pity.” He therefore spent the next week in intense work. On 15 January, he already had “new ideas for the electric problem.” The following three days were an uphill march: “Disciplined work on the problem despite the heat. Advancing with many setbacks” (Doc. 379, [p. 30]). On 22 January, some two weeks after composing the first version, he noted that he had completed the final version of the paper on gravitation and electricity. In all probability, this is the version that was eventually published in the Sit- zungsberichte and is included in the present Volume as Doc. 425. The article bears the same title as the earlier manuscript, is dated “Haruna Maru, Januar 1923,” and was presented to the Prussian Academy for publication by Max Planck in its ses- sion of 15 February 1923. The published version differs substantially from the manuscript begun on 9 Jan- uary. Einstein now no longer assumed the independent existence of a metric along- side the connection. Rather, he assumed that the theory would only depend on the affine connection. He now identified the symmetrized part of the Ricci tensor with the “natural” metric of the theory, and, as before, he linked the antisymmetrized Ricci tensor to the electromagnetic field. As a consequence of this change in basic assumptions, the published version no longer presented a contribution to the vari- ational treatment of standard general relativity, and it also no longer provided an explicit variational derivation of the Levi-Civita connection. Einstein also altered his choice of the Lagrangian for the variational integral. His criticism of Eddington’s approach now focused on Eddington’s failure to provide field equa- tions that determine all forty connection coefficients, and the derivation of such field equations became the focal point of the published version. Lastly, Einstein ex- plicitly introduced a scale factor λ that mediates between the scale of the “natural” metric defined by the Ricci tensor and that of the physical metric. The time elapsed between the first (Doc. 417) and final (Doc. 425) versions of the mansucript was less than a fortnight. During these two weeks Einstein greatly expanded his initial understanding of the subject matter covered in the article. He also appears to have identified and remedied the source of the problems that he had encountered with his early manuscript. It is therefore fortunate that we have anoth- er document that provides additional evidence of Einstein’s thinking during this