As I said in class, Strawson pretty much picks up Wittgenstein's very nonRussellian conception of linguistic meaning or significance and clubs Russell over the head with it.  For Russell, linguistic meaning was word-world correspondence; therefore Russell had a prima facie problem about singular terms that correspond to nothing but are perfectly meaningful, and the Theory of Descriptions is the result of Russell's effort to solve that problem.  Strawson maintains that no such drastic solution is necessary once we give up "'Fido'-Fido" or referential theories of meaning in the first place.  Linguistic items have the meanings that they do in virtue of the social conventions that govern their appropriate uses.  There are linguistic conventions that apply to such English phrases as "The present King of France" and "Sherlock Holmes"; we know the circumstances under which those phrases would appropriately be used, even though we also know as a matter of extralinguistic fact that those circumstances are never going to be realized.  So there is no mystery about nonreferring singular terms, and hence no reason either to posit eerie Meinongian Objects or to subject singular terms to a rigid procedure of "analysis" that "reveals" their hitherto secret "logical forms."
    According to Strawson, Russell's characteristically preWittgensteinian failure to see that language is a sweaty, grittily real human activity rather than a crystalline abstract Platonic structure caused him to make more specific errors, leaving the Theory of Descriptions open to criticism that even a preWittgensteinian theorist must take into account.
    Strawson holds that expressions do not refer at all; people refer, using expressions for that purpose.  This is reminiscent of the National Rifle Association's slogan, "Guns don't kill people, people kill people."  Certainly there is an obvious sense in which Strawson is right.  To use an example of his, if I write down, "This is a fine red one," "This" does not refer to anything -- and no determinate statement has been made -- until I do something to make it refer.  (But note Andrea's excellent point that this ignores negative existentials.  Strawson will have to find something special to say about negative existentials.)  An expression will come to refer only if I use it in a suitably well-engineered context, so that it does refer to a particular person.  But that is a matter of the expression being used, and when I do use it, it is I that am doing the work, not the expression.  Strawson tries to bring out the general antiRussellian point by offering three more specific arguments against the Theory of Descriptions.

    Objection 2.  According to Russell, sentence S, "The present king of France is bald," is false owing to the lack of any such King. Strawson points out that that verdict is implausible.  Suppose someone comes out and asserts S.  Would that person's hearers react by saying "That's false" or "I disagree"?  Surely not.  Rather, Strawson maintains, the speaker has produced an only ostensibly referring expression that has misfired; the speaker has simply failed to refer to anything and so has failed to make a complete statement.  The speaker's utterance is certainly defective, but not in the same way that "The present Queen of England has no children" is defective.  It is not incorrect but abortive; it does not even get a chance to be false.  Since no proper statement has been made in the first place, a fortiori nothing either true or false has been said.  A hearer would either just not comprehend or would say "Back up" and question the utterance's presupposition ("I'm not following you; France doesn't have a king").  Strawson therefore solves the Problem of Apparent Reference to Nonexistents by denying proposition 3:  S is meaningful, in that it has a legitimate use in the language and could be used to say true or false things if the world (or the French) were more coöperative, but not because it succeeds in picking out any individual thing.

    Objection 3.  Strawson further criticizes the claim, which he attributes to Russell, that "part of what [a speaker] would be asserting [in uttering S] would be that there at present existed one and only one king of France."  That claim too is implausible, for although the speaker presupposes that there is one and only one king, that is certainly not part of what the speaker asserts.
    (But that is a misunderstanding: Russell had made no such claim.  He said nothing at all about acts of asserting.  Perhaps Strawson is assuming on Russell's behalf that whatever is logically implied by a sentence is necessarily asserted by a speaker who utters that sentence.  But the latter principle is false: If I say "Fat Tommy can't run or climb a tree," I do not assert that Tommy is fat, even though my sentence logically implies that he is; if I say "Tommy is five feet seven inches tall" I do not assert that Tommy is less than eighteen miles tall.)

    Objection 4.  Strawson points out that many (probably most) descriptions are context-bound.  He offers the example of

            (T) The table is covered with books.

Presumably the subject term is a definite description, used in a standard way rather than in any different or unusual way.  But if we apply Russell's analysis, we get "At least one thing is a table and at most one thing is a table and any thing that is a table is covered with books" -- which by way of its second conjunct entails that there is at most one table, in the entire universe!
    That cannot be shrugged off.  However unwillingly, Russell is going to have to take some notice of the context of utterance.
    (He has several options.  After all, Strawson has no monopoly on the fact that when someone says "The table," we hearers generally know which table is meant, because something in the context has made it salient.  It may be the only table in sight, or the only one in the room, or the one we have just been talking about.  Russell may say that there is ellipsis here, that in the context, "The table" is short for a more elaborate description that is uniquely satisfied, though the ellipsis view has some disturbing implications.  There is also the nasty example I mentioned due to Marga Reimer.  I have not been able to find any version of the example online, or I would have supplied a link; it first appeared in her article "Incomplete Descriptions," Erkenntnis 37 (1992).)