Natural properties and atomicity in modal realism

The paper pinpoints certain unrecognized difficulties that surface for recombination and duplication in modal realism when we ask whether the following inter-world fixity claims hold true: 1) A property is perfectly natural in a world iff it is perfectly natural in every world where it is instantiated; 2) Something is mereologically atomic in a world iff all of its duplicates in every world are atomic. In connection to 1), the hypothesis of idlers prompts four variants of Lewis's doctrine of perfectly natural properties, all deemed unsatisfactory for the purposes of duplication and recombination. By means of 2), instead, we show that the principle of recombination does not countenance the atomicity or non-atomicity of duplicates; but it should, because it is genuinely possible that: a) something, which is atomic, is non-atomic; and b) something, which is non-atomic, is atomic. In discussing 1) and 2), the paper substantiates a tension in Lewis's metaphysics between modal intuitions and the reliance on the natural sciences