Russell's On Denoting: The Theory of Definite Descriptions
Bertrand Russell's 1905 paradigmatic analytic philosophy achievement introducing the theory of definite descriptions and resolving puzzles about meaning and reference.
PhilosophyAll chapters Reading tools Notes & comments
Descriptions and logical form › A Sentence Without a KingFull chapter · Sections in this chapter1 / 8 ↓
A Sentence Without a King
Imagine someone telling you that the current king of France is bald. You don't need to find the king and look at his head to see the problem. France has no current king. Yet you have understood the sentence well enough to object. Where, in your understanding of it, did the missing king enter?
Bertrand Russell's “On Denoting”, published in Mind in 1905, explains how a statement with a definite description can make sense when nothing answers to it. The explanation also shows how the same description can behave differently when it falls under a negation or occurs inside a report of what someone wishes to know. The king is a useful example because we can fix the relevant fact in advance. Nothing in the situation satisfies the description. We are free to attend to the work that the sentence does with it.
Phrases That Denote and Phrases That Fail
Russell starts with a family of denoting phrases. Some pick out a particular individual, some pick out none, and some leave open which individual is at issue. The definite description the author of Waverley identifies an author if exactly one person wrote the novel. The present king of France has the same sort of grammatical form, but fails to denote in the situation Russell considers. By contrast, a writer can figure in a claim that some writer did something even when the speaker has nobody in particular in mind. The different cases give him a reason to distinguish what a phrase contributes to a statement from the thing, if any, it denotes. Opening of “On Denoting,” pp. 479–480
Russell also uses this distinction to talk about what we can think about when we lack acquaintance with it. Direct acquaintance and descriptive access are different ways of being in a position to think about something. Russell illustrates descriptive access with the centre of mass of the Solar System at a particular time. You can understand a question about it without having encountered the point directly. The description gives a condition whose satisfaction is a separate question. This reasoning explains how one can understand what is being asked without already knowing which thing satisfies the condition. “On Denoting,” pp. 479–480
An empty description is a different case from a successful one whose referent we have never met. Someone can understand the words of an empty description and the condition it sets without thereby knowing an object by description. The condition might fail altogether. Conversely, grasping a condition is compatible with more than one object meeting it. This distinction is easy to lose if we start with a picture in which every grammatically singular subject must already have picked out a thing.
Existence, Uniqueness, and Predication
Russell analyzes the proposition expressed by a sentence containing the definite description. Take the form The F is G. For example, in The keeper of the archive is out, let F stand for being keeper of the archive under discussion and G for being out at the relevant time. The analysis gives three demands:
- At least one individual is F.
- At most one individual is F.
- The individual who is F is G.
Combine the first two and you have exactly one F. The third then gives a further condition about that unique individual. In a contemporary rendering, there is some x such that x is F, anything that is F is identical with x, and x is G. This is an explanatory rendering of Russell's conditions rather than a quotation of his notation. “On Denoting,” exposition and application, pp. 481–482 and 492
The three demands can fail independently. In a building with a vacant keeper position, the existence condition fails. In a building with two equal keepers and no further restriction, uniqueness fails even if both keepers are out. If there is one keeper and she is in, existence and uniqueness hold and predication fails. The analysis separates failures that the original sentence could leave you unsure how to diagnose. When the job is defined more narrowly, such as keeper of the manuscript archive on duty this evening, the domain of the description changes. You then test the same conditions using the narrower description.
On Russell's analysis, The F is G is true only if there is exactly one F and that individual is G. Identifying these conditions does not require supplying an F in advance. If there is no F, the analysis gives a false proposition with intelligible conditions.
In the king example, it is enough for the first condition to fail. We never reach a separate question about a particular person's hair. This makes it possible to explain the failure without treating a nonexistent monarch as the subject of a further statement. Russell replies to a different proposed treatment of descriptions by asking how we could accept a round square as the object designated by the round square. He can instead analyze the whole sentence in a way that asks for a unique thing satisfying contradictory conditions. On that analysis there is no need to supply such a thing. “On Denoting,” pp. 482–483 and 490–491
What Does the Negation Deny?
Now put the sentence in the negative. The king of France is not bald can be understood in two ways. It can say that there is exactly one present king of France and that he is nonbald. That reading fails because the existence condition still has to hold. It can also deny the full claim that there is a unique present king of France and that this individual is bald. On the second reading it is true when there is no king. Russell's distinction between primary and secondary occurrences allows him to separate these readings. “On Denoting,” discussion of negation, pp. 489–490
Return to the archive example. The first reading of The keeper of the archive is not out requires one keeper who is in, taking in and out here as exhaustive. The second says that it is false that there is a unique keeper and that the keeper is out. When the position is vacant, the first reading is false and the second is true. When there is exactly one keeper and she is in, both readings are true. The agreement in the second case can disguise the difference. A case in which the description fails is therefore a useful test of where the negation applies.
This also helps with statements that deny existence. If someone says there is no golden mountain, a useful rendering is that nothing satisfies the condition of being a golden mountain. That differs in general from saying there is no unique golden mountain. If there were two, the first denial would be false and the second true. In the empty case they agree, but they make different claims about other cases. Russell's method asks us to state which proposition we are denying. The surface position of a phrase does not decide that for us. “On Denoting,” discussion of empty descriptions, pp. 489–491
Why George IV's Question Is Informative
Russell's second famous example involves the authorship of Waverley. Scott wrote the novel. If the author of Waverley is just another name for Scott, it might seem harmless to interchange the two expressions in a report about what George IV wished to know. Yet the result would say that the king wished to know whether Scott was Scott. That changes the question. The king could understand identity perfectly well and remain curious about the authorship. Russell presents the example as a puzzle about what substitution is licensed inside a report of an attitude. “On Denoting,” first puzzle, p. 485
Apply the analysis to the sentence about Scott on its own. There was exactly one author of Waverley, and Scott was that individual. The analysis still contains an identity. The difference is that the description has been analyzed into conditions, and the identity appears within a claim about a unique author. There is no separate name-like constituent in that analyzed proposition that can simply be replaced by Scott. This explains why a person can know that Scott is Scott and still have something to learn from the authorship claim. “On Denoting,” discussion of Waverley, pp. 487–489
In the report about George IV, there are two places we can put those conditions. Inside the question, the report says that he wished to know whether there was a sole author of Waverley and Scott was that author. Outside the question, it says there was a sole author, and George IV wished to know of that person whether he was Scott. The two reports can focus on different questions. In the second case you might imagine someone pointing out a particular distant figure and the king asking if it was Scott, independently of any question about the book. Russell uses such differences to explain his distinction between primary and secondary occurrence. “On Denoting,” pp. 488–489
Try paraphrasing a report twice. Put the existence and uniqueness conditions first, outside the reported question. Then move them inside it. In the first version the speaker is making those claims as part of the report. In the second they are part of the proposition whose truth the person is asking about. The difference can survive even when the two versions mention the same person.
What the Analysis Leaves Open
The description can matter differently to people using the sentence on different occasions. P. F. Strawson develops this point in “On Referring” (1950). He distinguishes the meaning of an expression from what a speaker does with it on a particular occasion. Someone saying the sentence about the present king can use it to try to refer. The success of that attempt may be taken for granted rather than asserted as another part of what the speaker is saying about the monarch. In the empty case, a listener may challenge the setup instead of offering the ordinary verdict that the speaker made a false claim about a monarch's hair. Strawson takes this response to be significant for how an account of ordinary utterances should work. Stanford Encyclopedia of Philosophy, “Peter Frederick Strawson,” §2
The response and the formal analysis can address different questions. The formal analysis gives conditions for an F-and-G proposition. A further theory is needed to say what a speaker puts forward in an utterance and which background assumptions a listener is expected to share. Asking what would make the formal proposition true can still be helpful, even if it doesn't tell us everything that happens in the conversation. One dispute is whether the listener's response gives us a reason to change the analysis of the utterance or a reason to explain a difference in how the utterance is used.
Keith Donnellan distinguishes attributive and referential uses in “Reference and Definite Descriptions” (1966). In one of his examples, a person inspecting Smith's murder scene remarks that the murderer must be insane. The person could have no suspect in mind. The property is being attributed to whoever murdered Smith. The same description could be used during a trial to direct an audience's attention to a particular defendant. The speaker can succeed in drawing attention to that defendant even if he turns out to be innocent. The description can therefore play different roles as an aid to identification. The example opens a further question about the relation between what was said and what the hearer understood the speaker to mean.
One way to see the issue is to give the two speakers different follow-up questions. Ask the first whether the claim would apply if someone different were shown to be the killer. Ask the second whom she meant to draw attention to when she uttered the sentence. The answers can track different features of the episode. A successful answer about the intended person need not establish that the description fitted him. The point of the example is to keep that possibility visible when deciding what the existence and uniqueness analysis explains.
Practising the Distinction
Suppose you hear that the only gate into a garden is locked. Ask what would have to be true for the statement to hold. There must be a gate into the garden. There must be only one gate into it, within the intended restriction. That gate must be locked. Learning that it is locked doesn't settle the other questions. Conversely, finding a second gate can change the truth of the analyzed statement without telling you anything about the first lock. If the speaker was talking only about entrances on a particular street, you need to agree on that restriction before testing uniqueness.
Now consider a message that says a visitor wants to know whether the only gate is locked. It may report a question that takes for granted that there is a unique gate. It might also report an inquiry into the whole matter, including whether there is only one. Asking a follow-up question can help resolve the ambiguity. Russell shows why it arises by spelling out where the description's conditions enter. That is a method readers can try before they decide whether to accept the theory's account of every particular utterance.
A similar issue turns up when an organization publishes a rule about an office. “If there is a sole keeper on duty, the keeper has the alarm code” can lay down a policy that applies only after existence and uniqueness are established. It does not announce that someone is on duty now. If at a particular hour there are two keepers, you may still want a policy for that case, but you should not infer it from a rule with the sole-keeper condition. A paraphrase that moves the description's conditions outside the conditional changes a practical rule into a claim about current staffing.
Or ask what information you need to challenge a gate notice. If an official says the gate is unlocked, and another says the only gate is locked, the two might be disagreeing about a lock, or they might have different sets of gates in view. Bringing the two officials to the same gate subjects one part of their statements to a common test. Agreeing which paths count as entrances subjects a different part to a test. This is one benefit of stating existence, uniqueness, and predication separately. The analysis identifies several ways to respond to what sounded like a single assertion.
Further Reading
Read Russell's original paper through the three puzzles and the later applications. Try to write the two readings of the king's negative sentence before reading Russell's solution. Then compare the Waverley case with Donnellan's murder example. George IV's report turns on the scope of a description in a question. Donnellan asks how a speaker can use a description to direct an audience to a person. Keeping those questions distinct makes it easier to see what later objections owe to Russell's analysis and where they ask for something further.
Join the discussionAsk a question or share what you learned.
Keep exploring
Reference
Citation details
Cite or share this article.

Comments
Add to the conversation
A question, a correction, or an example from your own work.
No comments yet. Be the first to share your thoughts!