To: Alistair L.
From: Geoffrey Klempner
Subject: Proper names and 'Sherlock Holmes does not exist'
Date: 7th January 2009 11:49

Dear Alistair,

Thank you for your email of 17 December, with your essay for the University of London Logic paper, in response to the question, 'Can we give a defensible account of the proper name 'Sherlock Holmes' which allows it to be true that Sherlock Holmes does not exist?'

I am sorry that I was not able to respond to this sooner. I came back to my office on Monday, facing an avalanche of work. I hope that you had an enjoyable holiday.

This is an excellent piece of work, which fully justifies my belief that you have the ability to do very well in the BA.

Having said that, there is an issue about whether you have given an answer to the question, rather than to a different question, e.g. 'Can we give a defensible account of existence which allows it to be true that Sherlock Holmes does not exist?' In other words, to my ear what the question is intended to focus on is the analysis of proper names (cluster of descriptions theory vs Kripke vs ...).

Reading between the lines, you are clearly aware of the problem. You say at one point that treating 'Sherlock Holmes' as a Russellian proper name ('no object, no thought') has the consequence that 'Sherlock Holmes does not exist' cannot be true. The problem, as you state, is that in ordinary language we consider this a perfectly proper thing to say (e.g to the Japanese tourist looking for 221b Baker Street).

By contrast, according to the cluster of descriptions theory, when we say, 'Sherlock Holmes does not exist' there are a variety of things we could mean, depending on which descriptions are considered salient. (Consider the variety of contexts in which one might make the statement: a reply to the Japanese tourist, an expression of frustration of a London murder detective wishing that SH were on the case, an ignorant tax investigator who comes across a site dedicated to Sherlock Holmes on the internet etc.)

If we don't like the cluster of descriptions theory as a general account of proper names, and want to allow that it can be true to say that 'Sherlock Holmes does not exist' then we need to start thinking about the various alternatives that you canvass: Meinong's theory of objects as developed by Parsons/ Zalta; Gareth Evans' account of fictional discourse as part of a 'game of make believe'; as well as the idea of 'free logics' which do not make existential commitment (here, you gesture towards the idea but don't give the reader enough information to be able to assess it).

I liked your point about the difference between an existing individual, of whom a potentially infinite number of predications are true, and a fictional or mythical or hypothetical individual defined by a strictly finite set of properties/ relations, such as 'golden mountain' or 'Sherlock Holmes'. (Although, see below on my second thoughts regarding SH.)

I am sympathetic to a Meinongian approach. Although you refer to objections raised by Russell (e.g. the objection to the idea of a 'round square') you don't mention the most serious problem -- a problem which I believe can be resolved.

On a Meinongian approach, some objects have being and exist, while other objects have being but don't exist. This in effect treats 'existence' as a property which an entity either has or doesn't have. This makes it look as if an entity could acquire, or lose, the property of existence. This IS nonsensical. Consider what it would require to 'give' Sherlock Holmes, the non-existing character of fiction, the property of existence. If I chance my name by deed poll and set up a private detective agency at 221b Baker Street, you might say, as a joke, 'Now it is true that Sherlock Holmes exists!' But of course it isn't. Sherlock Holmes the character created by Arthur Conan Doyle and 'Sherlock Holmes' the former philosopher now offering services as a detective are two entirely different entities.

However, we can still treat 'exists' as a first-order predicate (the difference between 'exists' as a first-order and as a Fregean second-order predicate is one of the issues relevant to your essay) if we are prepared to accept the consequence that we can never truly state that 'X does not exist'. Sherlock Holmes in Conan Doyle's stories does exist. When we say, 'Sherlock Holmes does not exist' on this account we are saying something like, 'Sherlock Holmes refers to a a character in fiction' (as with the cluster of descriptions theory of proper names, we have to allow for a variety of responses, depending on the context).

It occurs to me also that it is actually not true that 'Sherlock Holmes' when treated as a fictional character only possesses a finite number of properties. All the things ever said by literary critics about the character of Sherlock Holmes belong to the properties of that fictional entity. In that sense, it could be argued that fictional entities have a 'real essence' from which their properties flow. Unlike cooked-up philosophical examples, fictional entities have a 'reality' which is inexhaustible, in a way similar, but also dissimilar to entities in the 'real' world. The difference is the difference between 'existing in fiction' and 'existing in the world', and of course that is a mighty big difference.

All the best,

Geoffrey