l x x x v i i i I N T R O D U C T I O N T O V O L U M E 1 5 a second requirement, motivated by the first. He demands that if a system is indeed composed of two subsystems that do not couple, with each subsystem having a cer- tain set of possible energy eigenvalues, then the possible motions of the total sys- tem must be combinations of the possible motions of the subsystems. Einstein then argues that his modified wave mechanics does not fulfill this requirement: the the- ory does not in general allow decomposition of the motions of a composite system into the motions of its subsystems. He justifies this claim by appeal to his tensor of ψ-curvature, . He argues that the latter does not vanish for the kind of situation discussed and that this implies that the principal directions of the composite system do not coincide with the principal directions of the subsystems if the latter are regarded as isolated systems. Einstein evidently considered this to be a major problem. The addendum fin- ishes with an expression of hope that it might yet be possible to overcome the issue by following through with an idea of Jakob Grommer, namely, to use lgψ instead of ψ to define the principal directions. But the idea did not pan out and, only two and a half weeks after presenting the paper to the Academy, Einstein called the journal and withdrew the paper from publication. X. Refrigerators and Patents Refrigerators had already aroused Einstein’s interest in December 1919 (see Vol. 9, Doc. 207) when, together with Walther Nernst, he had worked on a “funny techni- cal thing, an ice machine.” By March 1922, they were considering a patent appli- cation (Vol. 13, Doc. 67). Einstein must have been an excellent discussion partner, but it was Nernst who led negotiations with a company and worried about techni- calities (Vol. 13, Abs. 63). Leo Szilard claimed that in 1926 it was Einstein who initiated a project for developing safe household coolers after having read in a newspaper that an entire family was killed in bed by poisonous gas leaking from the pump of their refriger- ator. Einstein therefore planned to develop coolers that had no moving parts and were hermetically sealed. How and why he chose Szilard as a partner is not known perhaps he wanted to help Szilard, a man always in need of money.[56] The first reference to their collaboration in developing refrigerators can be found in Szilard’s letter of 10 September 1926 (Abs. 579). Because they had been in con- tact since the early twenties and were becoming closer at least since the end of 1923 (Vol. 14, Doc. 198), there was no need of communicating by letters, since they both lived in Berlin and could meet in person (Doc. 221). Szilard informed Einstein that he had submitted an application for a patent on a refrigerator (Abs. 579). An appli- cation known only by its title, “on a cooler with capillary pump,” by Szilard and ψμν