This 1971 letter to Mondadori contains an early version of Lewis’s reply to the famous problem now known as the ‘Humphrey objection’. In the first paragraph, Lewis agrees to publish ‘Languages and Language’ in Italian. It duly came out in 1973 as ‘Lingue e lingua’ in Versus, 4: 2–21. The paper wasn’t to appear in English until 1975, slightly revised.
The Humphrey Objection
In the second paragraph, Lewis turns to discuss the ’Humphrey objection’ to counterpart theory. We now know it as Kripke’s objection from Naming and Necessity (1972). But Mondadori and Koether had already raised it in correspondence with Lewis. The objection is, roughly speaking, that counterfactual statements about actual things, like Humphrey, or in Mondadori’s letter, Cicero, are about Humphrey or Cicero. They are not about their counterparts in other possible worlds. Lewis’s response in 1971 was that those statements are about both. The character of actual things settles whether things that don’t belong to our world are counterparts of the things in our world. This suffices to establish that counterfactuals under a counterpart theoretic construal are still about things that belong to the actual world. Lewis’s reponse here is different from the longer response that Lewis subsequently made in Plurality of Worlds (1986), though compatible.
In fact Kripke had originally raised this objection to counterpart theory in a lecture at NYU, sometime in 1969-70. He later published it in ‘Identity and Necessity’ (1971). There is no evidence here that Lewis know that Kripke had raised this objection. What Lewis’s letter also reveals is that the objection was clearly in the air. He remarks that it had already had been raised when he spoke at Bedford College in the University of London.
Necessary vs. A Priori
In the next paragraph Lewis states very clearly that he distinguishes necessary truths from a posteriori ones. He cite the Continuum Hypothesis as an example of a necessary truth not known a priori. But he resists apparent cases of necessary a posteriori which are currently popular. Lewis claims, for example, that the identification of electricity with a stream of protons is not impossible, only false a posteriori.