Routledge Philosophy Guidebook to

Kripke and Naming and Necessity

Saul Kripke is one of the most important philosophers of the twentieth

century. His most celebrated work, Naming and Necessity, makes arguably
the most important contribution to the philosophy of language and
metaphysics in recent years. Asking fundamental questions – how do names
refer to things in the world? Do objects have essential properties? What are
natural kind terms and to what do they refer? – he challenges prevailing
theories of language and conceptions of metaphysics, especially the
descriptivist account of reference which Kripke argues is found in Frege,
Wittgenstein and Russell, and the anti-essentialist metaphysics of Quine.
In this invaluable guidebook to Kripke’s classic work, Harold Noonan
introduces and assesses:

Kripke’s life and the background to his philosophy

the ideas and text of Naming and Necessity
the continuing importance of Kripke’s work to the philosophy of
language and metaphysics.

The Routledge Philosophy Guidebook to Kripke and Naming and Necessity

is an ideal starting point for anyone coming to Kripke’s work for the first
time. It is essential reading for philosophy students studying philosophy of
language, metaphysics, logic or the history of analytic philosophy.
Harold Noonan is Professor of Philosophy at the University of
Nottingham, UK. He is the author of Hume on Knowledge (1998) and
Personal Identity (second edition 2003), both available from Routledge.
Routledge Philosophy Guidebooks

Edited by Tim Crane and Jonathan Wolff

University of Cambridge and University College London

Plato and the Trial of Socrates Thomas C. Brickhouse and Nicholas D.

Smith
Aristotle and the Metaphysics Vasilis Politis
Rousseau and the Social Contract Christopher Bertram
Plato and the Republic, second edition Nickolas Pappas
Husserl and the Cartesian Meditations A. D. Smith
Kierkegaard and Fear and Trembling John Lippitt
Descartes and the Meditations Gary Hatfield
Hegel and the Philosophy of Right Dudley Knowles
Nietzsche on Morality Brian Leiter
Hegel and the Phenomenology of Spirit Robert Stern
Berkeley and the Principles of Human Knowledge Robert Fogelin
Aristotle on Ethics Gerard Hughes
Hume on Religion David O’Connor
Leibniz and the Monadology Anthony Savile
The Later Heidegger George Pattison
Hegel on History Joseph McCarney
Hume on Morality James Baillie
Hume on Knowledge Harold Noonan
Kant and the Critique of Pure Reason Sebastian Gardner
Mill on Liberty Jonathan Riley
Mill on Utilitarianism Roger Crisp
Wittgenstein and the Philosophical Investigations Marie McGinn
Spinoza and the Ethics Genevieve Lloyd
Heidegger on Being and Time, second edition Stephen Mulhall
Locke on Government D. A. Lloyd Thomas
Locke on Human Understanding E. J. Lowe
Derrida on Deconstruction Barry Stocker
Kant on Judgement Robert Wicks
Nietzsche on Art Aaron Ridley
Rorty and the Mirror of Nature James Tartaglia
Hobbes and Leviathan Glen Newey
Wittgenstein and the Tractatus Michael Morris
Aristotle and the Politics Jean Roberts
Merleau-Ponty and Phenomenology of Perception Komarine Romdenh-
Romluc
Frege on Sense and Reference Mark Textor
Routledge Philosophy Guidebook
to

Kripke and Naming and Necessity

Harold Noonan
First published 2013
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN

Simultaneously published in the USA and Canada

by Routledge
711 Third Avenue, New York, NY 10017

Routledge is an imprint of the Taylor & Francis Group, an informa business

© 2013 Harold Noonan

The right of Harold Noonan to be identified as the author of this work has been asserted by him in
accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by
any electronic, mechanical, or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or retrieval system, without permission in
writing from the publishers.

Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Cataloging in Publication Data

Noonan, Harold W.
The Routledge philosophy guidebook to Kripke and Naming and necessity /
by Harold Noonan.
p. cm. – (Routledge philosophy guidebooks)
Includes bibliographical references (p.) and index.
1. Kripke, Saul A., 1940– Naming and necessity. 2. Necessity (Philosophy)
3. Reference (Philosophy) 4. Identity (Philosophical concept) I. Title.
BD417.N66 2012
121’.68–dc23
2012026860

ISBN: 978-0-415-43621-2 (hbk)

ISBN: 978-0-415-43622-9 (pbk)
ISBN: 978-0-203-07380-3 (ebk)

Typeset in Garamond
by Taylor & Francis Books
Contents

ACKNOWLEDGEMENTS

1 Introductory overview

The problem situation

Kripke’s philosophical development
The main contentions

2 The background

Frege on sense
Russell on descriptions and names
The Frege–Russell synthesizers
Quine

3 Naming

The target
Giving the meaning, fixing the reference and rigid designation
The modal argument
The debate over the modal argument
The arguments against the cluster theory qua theory of reference-fixing
The historical chain picture
4 Necessity

The intelligibility of essentialism

The rejection of the problem of transworld identity and the critique of
counterpart theory
The essential properties of individuals
The necessary a posteriori and the contingent a priori

5 Extensions

Natural kind terms as proper names of kinds

The necessity of theoretical identifications
The illusion of contingency and mind–brain identity

GLOSSARY

BIBLIOGRAPHY

INDEX
Acknowledgements

I am grateful to colleagues in Nottingham and ex-colleagues in Birmingham

for comments.
1
Introductory Overview
The Problem Situation

Saul Kripke’s three lectures Naming and Necessity were delivered at

Princeton University in 1970. Subsequently a transcript of the recording
was made and the lectures were published in 1972 (Kripke 1972). A
corrected and expanded version with a new preface was published in 1980
(Kripke 1980).
The effect of Kripke’s work was immediate and enduring. Naming and
Necessity instantly established a new orthodoxy in Anglo-American
philosophy. It is still recognized as a masterpiece. One often reads of the
Kripkean ‘revolution’ in philosophy.
The two features of Naming and Necessity which caused this were, first,
its vigorous defence and unapologetic deployment of modal concepts (the
concepts of necessity and possibility, of what must be and what can be), and
second, its brilliantly argued rejection of the conception of reference – that
relation between language and the world which enables us to use the former
to make true or false statements about reality – as dependent on, and
determined solely by, the individual language user’s knowledge, intentions
and dispositions.
In connection with the first feature, Kripke argued for a sharp division
between the metaphysical concept of necessity and the epistemological
concept of a prioricity, or knowability independent of experience, and
asserted the existence of necessary a posteriori truths, truths which could
not have been otherwise but could, on the basis of experience, though only
on the basis of experience, be known to be both true and necessary. In
connection with the second feature, Naming and Necessity sets forth a set of
theses about the meaning and reference of proper names and what Kripke
calls ‘natural kind terms’, terms for the kinds of things studied in the natural
and biological sciences – water (studied by chemists), heat (studied by
physicists) and tigers (studied by zoologists).
 To understand why this caused such uproar we need to step back and
look at what were the received philosophical views at the time Kripke was
developing his ideas.
Analytic philosophy was the dominant school in Anglo-American
philosophy in the twentieth century before Kripke came on the scene. It
originated in the work of Russell, Frege and Wittgenstein. Associated
figures were the members of the Vienna Circle (the logical positivists)
including Carnap and W. V. O. Quine, a younger colleague of Carnap.
At the time Kripke was working on the ideas published in Naming and
Necessity Quine was the dominant influence on American philosophy and
Frege and Russell were recognized as the giants of early analytic
philosophy.
These three are Kripke’s principal targets in Naming and Necessity. He
targets Frege and Russell in his arguments about naming and Quine in his
arguments about necessity.
Early and mid-twentieth-century analytic philosophy had a variety of
features. First, there was an emphasis on language as a topic of
philosophical importance in its own right. Second, analytic philosophy was
associated with a respect for natural science as the paradigm of knowledge.
The logical positivists, for example, embraced the verifiability principle as
demarcating the line between sense and nonsense, or, as they thought,
science and non-science. A third, closely associated, feature of analytic
philosophy was its commitment to empiricism. Russell’s method of ‘logical
construction’ of knowledge of the external world and the self combined the
empiricism of Hume with the new logic (the now familiar predicate logic
taught to first-year undergraduates) developed by him and Frege. The
members of the Vienna Circle also thought of themselves as the heirs of
British Empiricism. Quine viewed his philosophy as empiricism without the
dogmas of logical positivism. A fourth feature of analytic philosophy was,
of course, its emphasis on analysis as the method of philosophy. There were
two paradigms: Russell’s theory of descriptions set out in ‘On Denoting’
(Russell 1905, reprinted in Russell 1956: 41–56) and second, the Frege–
Russell logicist programme – the attempt to reduce all of mathematics
(Russell) or at least arithmetic (Frege) to the new logic. The seminal
documents of this programme were Principia Mathematica, co-authored by
Russell and Whitehead and Russell (1910–13), and Frege’s Grundlagen
(1884 translated as Frege 1968) and Grundgesetze (1893, reprinted in
1962).
A fifth feature of analytic philosophy, linked to these others and deriving
from them, was the rejection of traditional metaphysics. Traditional
metaphysicians thought of themselves as concerned with establishing
fundamental and necessary truths about the world and as delineating,
following Aristotle, the division between the essential and the accidental
features of things. Under the influence of empiricism, the earlier and later
Wittgenstein, the logical positivists and Quine’s post-positivist insistence on
the continuity of philosophy with natural science, such metaphysical
inquiries were rejected as proper work for philosophers (or indeed for
anyone).
A specific aspect of the twentieth-century rejection of metaphysics by
analytic philosophy was the rejection of any notion of necessity distinct
from the linguistic notion of analyticity or truth in virtue of meaning,
knowable a priori without appeal to experience. From an empiricist
viewpoint our convictions about possibility and necessity can seem
extraordinary. Given that we are just creatures with experiences at particular
times and places how can we have any access to what is necessarily the
case – is the case at all times and places to put it in crudely Kantian terms –
or to what is merely possible, which we seem just as a matter of definition
to have no chance of actually experiencing at all? Shrinking necessity to
analyticity was the empiricist’s way of responding to this puzzlement and
rendering modality more ordinary and, it was thought, knowledge of it more
comprehensible. Building on his assumption that this notion of necessity
was the only one of which we could make objective sense (and bracketing
his belief that even this notion was not truly intelligible), Quine argued, still
more specifically, that no non-trivial sense could be made of the distinction,
central to traditional metaphysics, between essential (or necessary) and
accidental (or contingent) features of things and hence that no non-trivial
sense could be made of what the mediaevals called ‘de re’ (of the thing) as
opposed to ‘de dicto’ (of the saying) modal ascriptions. Giving significance
to this distinction, which he saw as both the temptation and sole raison
d’être of the developing field of quantified modal logic, he argued, required
an invidious distinction between different ways of designating the same
individual, which treated only some of them as indicating what was
necessarily true of that individual, and thus departed from the conception of
necessity as merely linguistic – in short, it required ‘Aristotelian
essentialism’.
It is this particular Quinean attack on the division between the essential
and accidental properties of individuals which is the starting point of
Kripke’s onslaught in Naming and Necessity on the previous orthodox
demotion of metaphysics. Kripke embraces ‘Aristotelian essentialism’.
With his rejection of Quine’s attack goes rejection of the interdefinability of
necessary and analytic truth, and the equation or interchangeability (1971:
177) of the latter notion with that of a priori knowable truth. For once
nontrivial necessary properties of individuals – ones whose possession does
not follow analytically from every way of specifying them – are allowed,
necessity can no longer be equated with necessary truth, and Kripke argues,
necessary truth itself can no longer be equated with analytic, hence a priori,
truth. Thus, reversing the pattern of two centuries of philosophizing about
the necessary and the contingent, he argues that there are necessary a
posteriori truths and even contingent a priori truths.
A final feature of analytic philosophy taken for granted by everyone
before Kripke was the assumption that reference, and, specifically, the
relation between proper names and their bearers, is dependent on and
determined solely by something in the mind of the speaker – something that
makes it the case that, given the way the world is, the user of the proper
name is speaking about that thing rather than some other. In the British
empiricist tradition beginning with Locke this determination was thought to
be done by an idea in the speaker’s mind – a sort of image or quasi-sensory
episode. By the time Kripke wrote things had moved on and this Lockean
conception had long been rejected. Russell had developed a distinction
between knowledge by acquaintance and knowledge by description, the
latter ultimately dependent on the former, and argued that all our talk must
be about things known in one of these ways, whilst Frege had developed a
distinction between sense and reference and argued that reference to any
entity was determined by grasp of a sense, a way of thinking of, or mode of
presentation of, an entity which belonged to that entity alone (the sense
itself is a mind-independent abstract object, like a number, belonging to a
‘third realm’ of non-actual causally inefficacious objects, but the speaker’s
reference is determined by his grasp of it, which is a mental act). What
remained was the belief that reference is only achievable via a piece of
identifying knowledge, or at least identifying true belief, of the object of
reference. In the case of ordinary proper names this belief was encapsulated
in what came to be known, despite the differences that existed between the
two philosophers as the ‘Frege–Russell’ description theory of names
(notwithstanding the fact, frequently noted, but also generally
acknowledged (Dummett 1973: 110–11) not to diminish the effectiveness of
Kripke’s attack, that there is no commitment in Frege to the thesis that the
sense of every proper name must be identical with that of some
description). According to this position, a proper name (as used by a
particular speaker at a particular time) must be understood (by that speaker)
as equivalent to some description which picks out the thing named (so the
description gives the Fregean sense he then associates with the name, and is
that by which the speaker has Russellian knowledge by description of the
thing). So I speak about the philosopher Saul Kripke when I use the name
(though there are doubtless many other people called ‘Saul Kripke’)
because I associate the name with the identifying description ‘the author of
Naming and Necessity’. The principle in play here is what Donnellan
(1972) has called ‘the principle of identifying descriptions’, and illustrates
by a passage from Strawson’s Individuals (1959): ‘it is no good using a
name for a particular unless one knows who or what is referred to by the
use of the name. A name is worthless without a backing of descriptions
which can be produced on demand to explain the application’ (Strawson
1959: 20). Donnellan goes on to say that it seems
at first sight almost indisputable that some such principle governs the referential function of proper
names. Must not a user of a proper name know to whom or what he is referring? And what can this
knowledge consist in if not the ability to describe the referent uniquely?
(Donnellan 1972: 357)

Nevertheless, Donnellan disputes the principle of identifying descriptions –

as does Kripke in Naming and Necessity (the original published version of
which occurs immediately before Donnellan’s paper in the volume
Semantics of Natural Languages (1972)).
Donnellan’s and Kripke’s arguments, and particularly Kripke’s, are
widely held to have devastated the Frege–Russell description theory, along
with the more fundamental thought that reference is determined by
identifying true belief. These arguments and their extension beyond proper
names to general names (of what Kripke calls ‘natural kinds’), together
with the related work of other philosophers he mentions, particularly Hilary
Putnam, have inspired an alternative conception of how reference in
thought and talk is achieved – so-called externalism. According to the
externalist, the mind necessarily involves the world; what one means and
thinks is partly constituted by what there is in the world. Putnam’s famous
slogan is ‘cut the pie any way you like, “meanings” just ain’tin the head’
(1975: 227).
Putnam explains his position by first clarifying the notion of a
psychological state. In a wide sense, a psychological state is simply one
studied by psychology. So my being jealous of Mary is a psychological
state (this is Putnam’s own example). But it entails the existence of Mary.
Hence it is not a psychological state in the narrow sense allowed by what
Putnam calls ‘the assumption of methodological solipsism’, the assumption
that ‘no psychological state, properly so-called, presupposes the existence
of any individual other than the subject to whom that state is ascribed’
(1975: 220). A narrow psychological state is thus one an individual could
be in even if he were alone in the universe. Putnam now contends that the
traditional conception of meaning (up to and beyond Frege and Russell) is
an internalist one, in accord with the assumption of methodological
solipsism, comprising two unchallenged assumptions: (1) that knowing the
meaning of a term is a matter of being in a narrow psychological state and
(2) that the meaning of a term (together with how things are) determines its
extension – in the case of a general term, the class of things to which it
applies, and in the case of a singular term, i.e., a proper name or definite
description, the single thing it refers to – in the sense that sameness of
meaning entails, given how things are, sameness of extension. These two
assumptions entail that the extension of a term is determined by its user’s
narrow psychology. But Putnam argues that this is not so, for two speakers
can be in the same total narrow psychological state even though the
extension of the term in the idiolect of one of them is not the same as its
extension in the idiolect of the other.
He argues this point by appeal to his famous Twin Earth argument, a
version of which is also used by Kripke. He supposes that there is a distant
planet, i.e., not a planet in another ‘possible world’, not a counterfactual
state of the Earth, but simply a planet in another galaxy, far, far away, which
he calls ‘Twin Earth’, on which everything is very much like Earth. In fact,
people on Twin Earth (some of them) even speak what sounds like English.
But there are a few differences between Earth and Twin Earth, one being
that the stuff in the seas and rivers of Twin Earth has the chemical formula
‘XYZ’, not the chemical formula ‘H2O’, though there are no differences
evident except to the expert chemist. Oscar and his molecule-for-molecule
identical twin on Twin Earth (who is not in fact identical, of course, because
he is mostly XYZ) are ignorant of the relevant chemistry, Putnam supposes
(or we can roll the time back to a pre-1750 chemically uninformed era and
consider Oscar’s n-times removed great-grandfather and his twin). Yet the
extension of ‘water’ in Oscar’s mouth is the class of H2O molecules, whilst
the extension of ‘water’ in his twin’s mouth is the class of XYZ molecules.
However, their narrow psychologies in the sense Putnam de fines are the
same. So narrow psychology, ‘what is in the head’, does not determine
extension. Internalism is refuted. We shall come back to (and question)
Putnam’s argument in Chapter 5. For present purposes, though, the
important point is that if Kripke’s arguments against the Frege–Russell
theory of proper names are correct then all by themselves they establish
externalism as Putnam defines it. For what Kripke argues is that ignorance
and error are no bars to reference, which is determined by a set of
reference-preserving causal links to an initial baptism, of which links and
baptism one may be wholly ignorant. If so, two users of a proper name may,
like Oscar and his twin, be referring to different things, merely as a
consequence of their uses tracing back to initial baptisms of different
objects, even though they are, in Putnam’s sense, in exactly the same
narrow psychological state. We will return to Kripke’s arguments shortly.
First we must look briefly at their place in his own philosophical evolution.
Kripke’s Philosophical Development

Naming and Necessity emerged from Kripke’s early technical work in

modal logic. This work (summarized in Kripke 1963) provided a semantics
for propositional and quantified modal logics and established various
methodological results. Kripke introduces the idea of a model structure, a
triple < G, K, R >, where G is an element of K, K is a set and R is a
reflexive relation on K. Intuitively, he says, we may think of G as the actual
world, K as a set of possible worlds, and R as a relation of relative
possibility. This brought back into the philosophical mainstream the
Leibnizian language according to which necessity is truth in all possible
worlds and possibility is truth in some. In his technical work Kripke does
not say more about what possible worlds are (he does not need to, since he
is doing mathematics, not philosophy), but he employs the concept
extensively in Naming and Necessity, where, however, he is emphatic that
our grasp of modal language is not derivative from our grasp of the concept
of a possible world, and is particularly critical of the work of David Lewis,
who employs a notion of a possible world (subsequently set out in detail in
1986), which can be explained in wholly non-modal terms, and explains
modal notions in terms of it. Kripke emphasizes that as he thinks of them
possible worlds are simply total ways the world might have been, or
possible states or histories of the entire world (1980: 18). The actual world
– better referred to as the actual state or history of the world – should
therefore not be confused with the enormous scattered object that surrounds
us (1980: 19–20). He suggests that ‘counterfactual situation’ might be better
terminology and that one should even remind oneself that the ‘worlds’
terminology can be replaced by modal talk –‘it is possible that’ (1980: 15).
At one point Kripke even refers to the ‘metaphor’ of possible worlds.
After taking his undergraduate degree Kripke turned from the
technicalities of modal logic to more purely philosophical questions about
naming and necessity. He presented his ideas in ‘Identity and Necessity’
(1971), which focuses on the arguments for the necessary, but a posteriori,
truth of identity statements containing proper names, whilst making it very
clear that identity statements containing descriptions may be contingent,
and contains a particularly clear presentation of Kripke’s argument that a
priori philosophical reflection, combined with straightforward empirical
discovery of what are, in fact, essential properties of things, can yield
knowledge that these properties are essential, and hence knowledge of
necessary a posteriori truths (not confined to identity statements).
This [lectern] looks like wood. It does not feel cold and it probably would if it were made of ice.
Here my entire judgement is a posteriori … but one knows by a priori philosophical analysis … [if]
the table is not made of ice it is necessarily not made of ice … we know by empirical investigation
that … this table is not made of ice. We can conclude by modus ponens that it is necessary that the
table not be made of ice, and this conclusion is known a posteriori, since one of the premisses on
which it is based is a posteriori.
(Kripke 1971: 180)

‘Identity and Necessity’ nicely complements Naming and Necessity,

providing a fuller exposition of one strand in the argument of the latter.
After the publication of Naming and Necessity in 1972 two
supplementary publications followed. In his ‘Speaker’s Reference and
Semantic Reference’ (1977) Kripke turns to the assessment of Russell’s
theory of descriptions and, inspired by and in partial criticism of, Keith
Donnellan’s ‘Reference and Definite Descriptions’ (1966) and the
distinction therein between referential and attributive uses of definite
descriptions, introduces the distinction between speaker’s reference and
semantic reference. This was important to Kripke partly because of a
famous objection due to Gareth Evans to his causal-historical theory of
reference – that it could not accommodate historical cases of reference
change (as when ‘Madagascar’, originally a native word for part of the
African mainland, became a name of an island). Kripke replied to this
objection partly by arguing that in such a case what starts out as merely a
case of speaker’s reference (the island) can over time become the semantic
reference. This paper is also significant for the extensive defence of
Russell’s theory of descriptions and the careful exposition of the scope
distinctions that can be drawn when definite descriptions are made use of in
modal contexts (Kripke points to the threefold ambiguity of ‘the number of
planets might have been necessarily even’) and in propositional attitude
contexts. Here Kripke reinforces the response in Naming and Necessity to
Quine’s attack on quantified modal logic and the intelligibility of the
distinction between essential and accidental properties. In Naming and
Necessity itself Kripke attends briefly to Donnellan’s distinction to argue
that it cannot be appealed to in order to contest his arguments against the
Frege–Russell view and he elaborates this point in ‘Speaker’s Reference
and Semantic Reference’.Those suggesting otherwise, of course, never
included Donnellan himself.
In ‘A Puzzle about Belief’ (1979) Kripke responds to the worry that the
phenomena the Frege–Russell theory was designed to accommodate cannot
be explained in the alternative framework of his causal-historical ‘picture’.
He argues that the problems he faces are problems for anyone, including
Frege and Russell, and that there is a ‘puzzle about belief’, which has no
solution, but indicates that in certain circumstances our everyday
mechanisms of belief ascription ‘break down’, i.e., lead to inconsistent
descriptions.
I will conclude this introductory chapter with a brief survey of Kripke’s
most important arguments and claims in Naming and Necessity, giving their
primary locations in the text.
The Main Contentions

As emphasized, Kripke, like Putnam, is an externalist. His first target is the

internalist conception of reference as determined by the word-user’s beliefs
and intentions, thought of as narrow psychological states. According to this
view one cannot refer to a thing unless one can focus on it in thought. The
paradigm case of reference is thus:
I want to name an object. I think of some way of describing it uniquely and then I go through, so to
speak, a sort of mental ceremony: By ‘Cicero’ I shall mean the man who denounced Catiline; and
that’s what the reference of ‘Cicero’ will be … my intentions are given by first, some condition
which uniquely determines an object, then using a certain word as a name for the object determined
by this condition.
(Kripke 1980: 79)

This conception can be expressed (following Donnellan 1972) as follows:

to be able to talk about a thing one must know what (which thing), if any,
one is talking about.
Kripke’s primary contention is that this is untrue. Ignorance and error do
not prevent reference (the arguments he gives for this (1980: 78ff.) are, in
fact, sometimes called ‘the arguments from ignorance and error’–
sometimes subdivided, as we shall see into ‘semantic’ and
‘epistemological’ arguments). Kripke claims that this is so in two types of
case: reference using proper names of individuals (Lecture II) and reference
by means of natural kind terms (Lecture III). In these cases he claims that
‘what is in the head’, the narrowly defined word-user’s individual beliefs
and intentions, does not determine what his reference is. It is a great tribute
to the education of philosophers, he jokingly says, that the contrary has
been maintained for so long (1980: 81). What does determine reference is a
sort of causal-historical chain of which the speaker may have no knowledge
(1980: 91).
What is going on … a baby is born; his parents call him by a certain name. They talk about him …
others meet him. … the name is spread from link to link as if by a chain. A speaker … on the far end
of the chain, who has heard about, say, Richard Feynman, … may be referring to Feynman … even
though he cannot identify him uniquely … instead, a chain of communication going back to Feynman
himself has been established, … not by a ceremony that he makes in private in his study: ‘By
“Feynman” I shall mean the man who did such and and such’.

We will look at Kripke’s arguments later. Some consequences can be

mentioned now. The first is Kripke’s most famous thesis: that proper names
(and natural kind terms) are rigid designators (1980: 48ff.). A rigid
designator is an expression which has the same reference with respect to
every possible world. This is constituted by its behaving in a certain way in
modal statements (statements about what might have been or must have
been), in subjunctive conditional statements about what would have been
the case if such-and-such had been the case and in suppositions and
statements of their consequences. Kripke says at the beginning of the
lectures that he hopes that some people will see some connection between
the two topics in the title (1980: 23). The connection is made by the thesis
that names are rigid designators. Kripke gives the following intuitive
conjunctive test for rigid designation: a singular term ‘X’ is a rigid
designator if and only if ‘X might not have been X’ and ‘someone/thing
other than X might have been X’ are unambiguously false. By this test
typical proper names are plainly rigid designators but most definite
descriptions are not. So are natural kind terms (like ‘gold’, ‘water’, and ‘the
tiger’, which can replace ‘X’ grammatically in the test sentences) but not
(most) descriptions of natural kinds (like ‘the world’s most precious metal’,
‘the most dangerous kind of creature on the planet’).
This gives Kripke his first argument against the Frege–Russell
description theory (1980: 53, 61–3). Even if e.g., ‘Aristotle’ is a name I
associate with the description ‘the founder of formal logic’ and I know
nothing else of Aristotle I have to accept that ‘Aristotle might have not have
been the founder of formal logic’ has a true reading, whilst ‘Aristotle might
not have been Aristotle’ does not. So the name is not interchangeable with
the description without change of truth-value, salva veritate, in modal
contexts. So the name is not fully synonymous with the description.
Generalized, qualified and elaborated, this is Kripke’s ‘modal argument’
against the Frege–Russell theory, which is set out in the first lecture of
Naming and Necessity and is usually contrasted by commentators with the
subsequent semantic and epistemological arguments. It is clear that in
Kripke’s thought the modal argument came first and the conception of
proper names as rigid designators is not dependent on his subsequent anti-
internalist arguments (By contrast, it is clear that in Putnam’s thought the
rejection of externalism precedes his acceptance of the rigidity of proper
names and natural kind terms, which is something he accepts because he
equates it with his own thesis, which he regards as a consequence of his
Twin Earth arguments, that such expressions are ‘indexical’.)
However, if Kripke is right to reject the internalist conception of the
reference of proper names and natural kind terms as determined by
identifying true belief it is a plausible consequence that proper names and
natural kind terms must be rigid designators. If their reference is externally
determined it is hard to see what could prevent this being so. Consequently
many commentators think that the most important arguments in Naming
and Necessity are the arguments from ignorance and error. However,
Kripke’s observation that proper names, but not most descriptions, are rigid
designators also provides him with the independent argument given above
against the Frege–Russell description theory, or, more accurately, against
that ambitious, ‘meaning-giving’ form of the description theory according
to which names abbreviate descriptions, where this entails their
interchangeability with these salva veritate in modal contexts.
It is uncontroversial that ordinary proper names typically behave as rigid
designators, as characterized above, and it is uncontroversial that some
descriptions, including the ones mentioned in Naming and Necessity, do
not. But what is controversial is that this behaviour on the part of proper
names must be explained in a way that is incompatible with the Frege–
Russell paradigm, as Kripke’s ‘modal argument’ requires. Whether this is
so will be a main topic of Chapter 3.
However, if this is accepted other consequences follow.
 First, it follows that there are necessary a posteriori truths. Before Kripke
wrote it was generally accepted that no necessary truths are a posteriori, all
are a priori if knowable at all, at least in the broad if obscure sense in which
all knowable mathematical truths as well as all logical truths and
definitional consequences of logical truths are a priori. But the notions of
necessary and a priori truth come from different areas of philosophy
(Kripke 1980: 34ff.) and this enables Kripke to put forward
counterexamples to the traditional position. These counterexamples are of
three types. First, there are affirmations and denials of identity containing
names of individuals, e.g., ‘Hesperus is Phosphorus’, ‘Mars is not Venus’
(Kripke 1980: 99ff., Kripke actually only explicitly considers affirmative
identity statements). Given that these names are rigid designators it follows
that since ‘Hesperus’ and ‘Phosphorus’ have the same reference
‘Necessarily, Hesperus is Phosphorus’ is true (though the truth of the latter
does not entail that the names are rigid designators), and since ‘Mars’ and
‘Venus’ have distinct references ‘Necessarily, Mars is not Venus’ is true.
Given that the behaviour of these names as rigid designators cannot be
accounted for in accordance with the Frege–Russell description theory, it
follows that ‘Hesperus is Phosphorus’ and ‘Mars is not Venus’ express
necessary truths. But it seems obvious that these statements cannot be
known to be true except a posteriori. Direct reference theorists (Salmon
1982, Soames 2002), who endorse a purely ‘Millian’ picture according to
which only the reference of a name contributes to what is expressed (Kripke
1980: 20), accept that this is true of the identity statement ‘Hesperus is
Phosphorus’, but only in the sense that the English sentence ‘Hesperus is
Phosphorus’ could sometimes be used to raise an empirical issue. However,
they deny that the proposition expressed by that sentence is only knowable
a posteriori. Rather, they say, since it is the proposition that Hesperus is
Hesperus it is knowable a priori. They take this to be the logical outcome of
Kripke’s onslaught on descriptivism and his endorsement of the Millian
thesis that proper names have denotation but no connotation (1980: 26ff.).
Thus they take it that Kripke’s first example of a necessary a posteriori truth
is actually no such thing if a necessary a posteriori truth has to be a
necessary a posteriori proposition (though they accept that other examples,
and in particular, statements of non-identity, like ‘Mars is not Venus’, are
genuine examples of necessary a posteriori propositions). Kripke makes it
clear in the Preface to Naming and Necessity that he is not willing to take
‘Millianism’ so far. His claim that ‘Hesperus is Phosphorus’ is an example
of a necessary a posteriori truth is the claim that the sentence both expresses
a necessary truth and can be used to raise an empirical issue. He is neither
willing to assert nor, like the direct reference theorists, willing to deny, that
what it expresses is knowable only a posteriori, though he says that the fact
that ‘Hesperus is Phosphorus’, unlike ‘Hesperus is Hesperus’, can be used
to raise an empirical issue indicates that the mode of reference-fixing is
relevant to ‘our epistemic attitude to the sentence [sic] expressed’ (1980:
21). The possibility of combining Kripke’s thesis that proper names are
rigid designators with the affirmation (pace direct reference theorists) that
what a true identity statement between names expresses is sometimes only a
posteriori knowable will be explored in Chapter 4.
Kripke’s second class of counterexamples to the traditional position
consists still of identity statements, but ones containing names of natural
kinds, e.g., ‘water is H2O’ (1980: 100, 128ff.). These depend on the
assimilation of such names to proper names of individuals. Given this
assimilation a further class of necessary a posteriori truths is that of
subject–predicate sentences whose subject terms are natural kind names and
whose predicate terms are scientific classifications, such as ‘water is a
compound’ or ‘light is a stream of photons’ (1980: 129). Given that ‘water
is H2O’ is a necessary truth and ‘H2O is a compound’ is a necessary truth
(presumably an analytic one) it follows that ‘water is a compound’ is a
necessary truth. Kripke’s third class of necessary a posteriori truths contains
generalizations like ‘All samples of water are samples of a compound’, ‘All
tigers are mammals’ or ‘All cats are animals’ (1980: 119). These statements
are neither statements of identity nor subject–predicate statements whose
subject terms are singular names, i.e., names of individuals or kinds. But the
notion of rigidity is defined in the first place only for singular terms, so
Kripke’s claim that these are necessary a posteriori truths is possibly his
most controversial. It is also possibly his most interesting because it appears
that knowledge of such truths is a paradigm of the outcome of scientific
discovery. We will explore this thesis in Chapter 5.
Kripke also appeals to the concept of rigidity to argue for the existence of
contingent a priori truths (the flip side of the coin, as it were). His famous
example is ‘Stick S is one metre long at t’, where ‘one metre’ is introduced
as a rigid designator of the length of stick S (the standard metre rod in
Paris) at time t (1980: 54ff.). Stick S might have had a different length at t,
so it might not have been one metre. But it is apparently knowable just by
reflecting on the definition that stick S is one metre long at t (if stick S
exists at t at all). We will look more closely at this contention in Chapter 4.
Kripke also derives another startling consequence from his doctrine of
rigidity, one to which he draws attention right at the beginning of Naming
and Necessity (1980: 123–4), that it is false that there might have been
unicorns. As he explains in the Addenda, the argument for this position
depends on the assimilation of names of natural kinds to proper names of
individuals and also yields the conclusion that someone cannot use a proper
name of a fictional character, like ‘Sherlock Holmes’,to truly assert
‘Sherlock Holmes might have existed’. Kripke expands his thoughts on
fictional names in the unpublished John Locke lectures for 1973 ‘Reference
and Existence’ and in ‘Vacuous Names and Fictional Entities’ in his
Philosophical Troubles (2011).
In the Addenda the argument is that ‘unicorn’ is intended as a name of a
particular species, but different species could have satisfied all the defining
features of the myth, hence, since we cannot say which of these species
would have been the unicorns, it is not true that unicorns might have
existed. Similarly, since there is no man who answers to the description of
Sherlock Holmes in the stories, one cannot say of any possible person that
he would have been Sherlock Holmes if he had existed, since different
people might have performed the exploits of Holmes (1980: 157–8).
Although the arguments do not explicitly appeal to Kripke’s view that
‘unicorn’ and ‘Sherlock Holmes’ are rigid designators, it is implicit in them
since they appeal to the ideas that different people or species could have
satisfied the relevant description and therefore no one of them could have
been Sherlock Holmes or the unicorns since rigid designators have the same
reference across all possible worlds. Of course, although Kripke does not
explicitly argue in this way, if ‘Sherlock Holmes’ is a rigid designator ‘It
might have been that Sherlock Holmes existed’ must have the same truth-
value as ‘Sherlock Holmes might have existed’, and as ‘the detective living
at 22B Baker Street, etc., might have existed’. But the last of these cannot
be true, since there is no actual individual designated and there are no
merely possible individuals, so the description is empty.
The concept of rigid designation also enables Kripke to respond to
Quine’s objection that essentialism makes an invidious distinction between
different ways of specifying an individual (1980: 39ff.). Kripke argues that
the distinction between accidental and contingent properties is an intuitively
sound one. Intuitively, I could not have been a poached egg or a number;
but I could have spent my childhood in Scotland. Quine is correct, Kripke
says, that to make this distinction requires distinguishing different ways of
specifying an individual, but the needed distinction is just that between
rigid and non-rigid designators. Kripke also argues specifically that certain
properties of concrete individuals are in fact essential to them. One of these
contentions is the thesis of the necessity of material origin, that is, that a
material object like a table, originally constituted of a certain hunk of
matter, could not have existed and been originally constituted of a
completely different hunk of matter (1980: 114). Kripke gives an extended
argument for this thesis in a long footnote. He also defends as intuitively
plausible the thesis that an animal or human being could not have had a
different biological origin – could not have been the product of the fusion of
a different biological sperm and ovum (1980: 111ff.), so that the Queen, for
example, could not have been the daughter of Mr and Mrs Harry Truman.
So Kripke both defends the intuitive acceptability of the distinction between
essential and accidental properties and makes various claims about
particular examples of the distinction, some argued and some unargued.
The final application Kripke makes of his concept of rigidity is to the
mind–body problem (1980: 98ff., 144ff.). According to mind–brain identity
theorists mental events are identical with physical events in the brain, e.g.,
pain is the firing of C-fibres. This is a scientific discovery, they say, on a par
with the discovery that water is H2O, and an example of a contingent
identity. Kripke argues that since ‘pain’ and ‘C-fibre stimulation’ are rigid
designators, this identity statement, like the statement that water is H2O,
must be, if true, necessarily true. However, it is knowable only a posteriori,
so there is an appearance of contingency, as also in the case of ‘water is
H2O’, but in this case, unlike that one, it cannot be explained away as
illusory. So the mind–brain identity statement is not true at all.
Looking back over the uses to which Kripke puts the concept of rigid
designation we see that the most emphasis is put on the arguments for
necessary a posteriori truths. Extending the concept of rigid designation
from proper names to general terms for natural kinds, much, or maybe all,
scientific discovery becomes, when seen through Kripke’s eyes, the
discovery of such truths. As he states in the final sentence in the Addenda to
Naming and Necessity (1980: 164): ‘The third lecture suggests that a good
deal of what contemporary philosophy regards as mere physical necessity is
actually necessity tout court’.
The notion of rigidity is introduced by Kripke in Lecture One of Naming
and Necessity in the context of his presentation of the modal argument
against the ambitious, meaning-giving, version of the Frege–Russell
description theory. In the second lecture Kripke turns his attention to a less
ambitious, reference-fixing, version of the Frege–Russell description
theory, which is still consistent with the contention that users of proper
names refer in virtue of the identifying true belief they possess. According
to this position names need not be synonymous with flexible descriptions,
and so can be allowed to be rigid, but every user of a proper name must
associate some identifying true belief with it. Kripke’s argument against this
position is simply that in fact most speakers do not have the identifying
beliefs in question.
His famous arguments ‘from ignorance and error’ have proved to be
controversial, perhaps more so than his modal argument. His most quoted
examples are those of Feynman and Gödel/Schmidt. He argues that in fact
many people use the name ‘Feynman’ to refer to the famous physicist
without possessing any identifying information about him – knowing only
that he is a famous physicist – so ignorance does not prevent reference
(1980: 81). The Gödel/Schmidt fantasy is used to argue that error need
likewise be no bar to reference. If we suppose that Gödel did not in fact
prove the incompleteness of arithmetic but, unbeknownst to everyone but
Gödel himself, the long-deceased Schmidt did, the reference of the name
‘Gödel’ in the mouth of any member of the deceived general public who
would answer the question ‘Who is Gödel?’ by saying ‘the man who proved
the incompleteness of arithmetic’, would not be Schmidt but the famous
public figure, even though the description standardly associated with that
name would in these circumstances be one that was satisfied by Schmidt.
After putting forward these arguments against the Frege–Russell
description theory Kripke recommends a return to an earlier paradigm put
forward by John Stuart Mill and described previously in the text (1980: 26–
8), according to which names have denotation but no connotation. But as
we have seen, unlike subsequent philosophers influenced by him, Kripke is
unwilling to endorse the purely ‘Millian’ view that only the referent of a
name contributes to what is expressed (1980: 20). Kripke’s argument can
also be thought of as providing a reason to think of ordinary proper names
as more akin to what Russell called proper names in the logical sense and
distinguished from everyday proper names (1980: 27).
Although Kripke does not endorse the purely Millian view about proper
names of individuals, in another sense he is more Millian than Mill, since
he extends both of his conclusions about proper names – that they are rigid
designators and that their references are not fixed by associated descriptions
– to general names of natural kinds. Mill held that such expressions had
connotation as well as denotation, a general name such as ‘human being’
being defined as the conjunction of certain properties which give necessary
and sufficient conditions for being human. Kripke says that he ‘regards Mill
as more-or-less right about “singular” names, but wrong about “general”
names’ (1980: 127) (and in this he is in agreement with Putnam’s
previously sketched ideas) and that we do not use such terms, like ‘tiger’, to
pick out the individuals with a certain set of identifying marks, but rather
use them as names of natural kinds of thing. When the name is originally
introduced it is introduced as the name of that kind of thing, a kind perhaps
identified by certain ostensively demonstrated instances or by a description
like ‘the kind of thing responsible for these killings’. The original
identification thus involves ostension of particulars and an implicit appeal
to the notion of sameness of kind –‘the kind of creature whose instances are
of the same kind as these/those responsible for these killings’. But the
description may be forgotten by subsequent users who may have no
identifying description of the kind at all. And this, Kripke argues, is
analogous to the situation with proper names of individuals. Hence, just as
we must reject a description theory of proper names, we must reject a
(Millian) ‘cluster of identifying marks’ theory of natural kind terms.
With this summary of Kripke’s main contentions and lines of argument in
Naming and Necessity we can now return to a more detailed look at the
views of the great philosophers he is attacking, Frege, Russell and Quine.
2
The Background
Frege on Sense

To understand Frege’s notion of sense (Sinn) we need first to understand the

argument he gives for the distinction between sense and reference
(Bedeutung).
The reference of an expression is its semantic value, i.e. what it
contributes to the determination of the truth-values of sentences in which it
occurs. ‘John runs’ is true or false depending on what the bearer of the
name ‘John’ is doing. So the bearer of the name, John, is its reference.
Frege applies the notion of reference to expressions of the other
grammatical categories he recognizes: sentences, predicates and functional
expressions. He argues that, except in ‘indirect contexts’, which we will
come to later, the reference of a sentence, i.e. its contribution to the
determination of the truth-values of other sentences containing it, is its
truth-value. A fundamental principle to which he appeals in arguing this is
the functionality principle or compositionality principle for reference: the
reference of a whole is a function of, i.e. is determined by, the references of
its parts.
This has two implications. First, if a part of an expression lacks
reference, the whole lacks reference. Second, if a part is replaced by another
part having the same reference, the reference of the whole is unchanged. So
if a sentence contains a name lacking reference it lacks reference, i.e. truth-
value. And if a name in a sentence is replaced by another with the same
reference the truth-value of the sentence must remain unchanged. If there is
no such person as John, ‘John runs’ must be neither true nor false. If John is
Jack, ‘Jack runs’ must have the same truth-value as ‘John runs’.
Frege groups definite descriptions with proper names, ordinarily so-
called, as singular terms designating objects. So he holds that just as
‘Elizabeth II’ has as its reference its bearer, Elizabeth II, so ‘the Queen of
England’ has as its reference what it denotes, Elizabeth II. So ‘the Queen of
England is wise’ is true just in case ‘the mother of Prince Charles is wise’ is
true. ‘The Queen of England’ has reference but ‘the King of France’ does
not. So, according to Frege, ‘the King of France is wise’ is neither true nor
false.
These consequences follow because according to Frege the reference of a
singular term – a proper name or definite description – is the object it
designates, i.e. the bearer of the name or the denotation of the description,
what it describes. Now a simple view, which may be John Stuart Mill’s, is
that the only semantic feature of a singular term, the only significant feature
it can share with other terms which must be known by its users if they are to
count as understanding it, is its reference. But Frege denies this. He insists
that there is another such feature which does not depend on reference alone,
and so does not matter for truth-conditions, i.e. what determines truth-value.
This is cognitive value.
In ‘On Sense and Reference’ (1892, included in 1969) Frege compares
sentences that differ only by the substitution of coreferential singular terms,
e.g., ‘The PM is the leader of the Labour Party’ and ‘The PM is the PM’.
Since the PM is the leader of the Labour Party (as I type), ‘the PM’ and ‘the
leader of the Labour Party’ corefer. Thus, by Frege’s compositionality
principle for reference, these sentences have the same truth-value. So if its
reference was the only semantic feature of a singular term, the only feature
it has which must be known by users who count as understanding it, these
sentences would be on a par. They would not differ semantically. But they
do differ. They differ in cognitive value, i.e. one is uninformative, the other
is not.
This is the starting point of Frege’s fundamental argument for the
necessity of sense in addition to reference. But what is sense? Frege speaks
of sense as a mode of determination or mode of presentation of reference.
He also speaks of sense as ‘illuminating’ the reference.
There is an enlightening, albeit controversial, way of dispensing with
these metaphors (Evans 1982: 14ff.). Someone who understands an
utterance of ‘the PM is at least 50 years old’ must at least think of the
politician in having precisely this thought: that what the speaker is saying is
true if and only if the PM is at least 50 years old. But to think of the
politician one must think of him in a certain way. This is the principle,
Evans suggests, underlying Frege’s remarks about sense. If someone, T,
thinks of an object, O, there must be an answer to the question: In virtue of
what is T thinking of O? The answer will take the form ‘Tis thinking of O
in virtue of the fact that … T … O. …’. Filling the blank will specify the
way in which T thinks of O. A second person will think of O in the same
way if the blank has to be filled in the same way in answering the question
about him. When we give an answer to such a question we identify what
Frege calls a sense. If you give the way in which a thinker thinks of the PM
when he hears and understands an utterance of a sentence containing ‘the
PM’ then you identify the sense he then associates with ‘the PM’.
Frege’s idea is that a singular term is an element of a public language and
that to understand it one must think of the reference in a particular way:
‘The sense of a proper name [singular term] is grasped by everybody who is
sufficiently familiar with the class of designations to which is belongs’
(‘The Thought’ 1918, included in 1984: 360). So sense is public and
objective, and to the extent that it varies from speaker to speaker they do
not speak the same language.
The argument for the distinction between sense and reference appears
first in ‘On Sense and Reference’:
EQUALITY gives rise to challenging questions which are not altogether easy to answer … a = a and
a = b are obviously statements of different cognitive value. a = a holds a priori and according to Kant,
is to be labelled analytic, while statements of the form a = b often contain very valuable extensions of
our knowledge, and cannot always be established a priori. … If the sign ‘a’ is distinguished from the
sign ‘b’ only as an object (here, by means of its shape), not as a sign (i.e. not by the manner in which
it designates something), the cognitive value of a = a becomes essentially equal to that of a = b,
provided a = b is true. A difference can arise only if the difference between the signs corresponds to a
difference in the mode of presentation of that which is designated. … It is natural, now, to think of
there being connected with the sign … besides that to which the sign refers, which may be called the
reference of the sign, also what I should like to call the sense of the sign, wherein the mode of
presentation is contained. … The reference of ‘Evening Star’ would be the same as that of ‘Morning
Star’, but not the sense.
(1969: 57)
But the real problem is not just with identity statements. ‘The PM is at least
50 years old’ differs in informative value from ‘the leader of the Labour
Party is at least 50 years old’ just as ‘the PM is the leader’ differs from ‘the
PM is the PM’. A fortiori, the problem is not only about the difference
between informative and uninformative identity statements: ‘the PM is the
husband of Sarah Brown’ and ‘the leader of the Labour Party is the husband
of Sarah Brown’ are both informative identity statements but differ in
information value.
Frege thinks that the argument for sense applies to ordinary proper names
as well as to definite descriptions. In the latter case the sense is transparent;
the description carries on its face a way of thinking of an object. In a
footnote Frege comments on the difference. He says that the sense of an
ordinary proper name, like ‘Aristotle’, will differ from speaker to speaker.
For one speaker ‘Aristotle’ may mean ‘the teacher of Alexander’, for
another ‘the author of Metaphysics’. But the reference remains the same so
at the level of truth-value this variation in sense makes no difference: ‘so
long as the reference remains the same variation in sense may be tolerated’
(1969: 58). However, Frege thinks that such variation ought not to occur in
a perfect language.
According to Frege’s picture then: to each sense there corresponds at
most one reference, but many senses can determine the same reference. An
expression has reference because it expresses a sense. Reference is thus a
mediated relation between language and the world. ‘The PM’ refers to
Brown because it has a certain sense, having to do with the satisfaction of a
certain condition, and Brown is the one and only person satisfying that
condition.
An equivalent formulation of the fundamental argument for the
distinction between sense and reference can now be given.
Consider the substitution of coreferential singular terms within
propositional attitude constructions, grammatical constructions like ‘X
knows that …’ or ‘X believes that …’ ascribing an attitude (of belief or
knowledge, for example) to a proposition, that so-and-so. Given Frege’s
compositionality principle for reference he is committed to such
substitution preserving truth-value. But substitution within propositional
attitude contexts appears to provide counterexamples. The argument ‘John
knows that the Evening Star is the Evening Star. The Evening Star is the
Morning Star. So John knows that the Evening Star is the Morning Star’ is
invalid. This is a consequence of Frege’s contention that the two
expressions differ in cognitive value. In fact, this is what it means for them
to differ in cognitive value. But given just the resources of his theory of
reference Frege cannot explain this invalidity. Given that ‘the Evening Star’
and ‘the Morning Star’ are names of the same thing, they are coreferential,
so by Frege’s compositionality principle for reference, substitution of one
for the other cannot change the reference of the whole, i.e. the truth-value
of the sentence.
So there must be a feature over and above reference which accounts for
the invalidity of the argument ‘John knows that the Evening Star is the
Evening Star. The Evening Star is the Morning Star. So John knows that the
Evening Star is the Morning Star.’ This feature is sense, since to explain the
invalidity of this argument form is precisely what is required to resolve
Frege’s puzzle of the informativeness of true identity statements.
The question is now: how exactly can an appeal to sense resolve the
puzzle?
The inference from ‘John knows that the Evening Star is the Evening
Star’ and ‘the Evening Star is the Morning Star’ to ‘John knows that the
Evening Star is the Morning Star’ is valid if (a) ‘the Evening Star’ has the
same reference as ‘the Morning Star’ and (b) Frege’s compositionality
principle is correct. Frege appeals to sense to explain how (a) is false
(relevantly understood). The truth of the identity statement ‘the Evening
Star is the Morning Star’ shows that the two names have the same reference
in that context. But, Frege maintains, they do not have the same reference in
the context of the propositional attitude ascription, because in that context
they refer to their (ordinary) senses. But these are different, so the failure of
substitutivity is not a counterexample to the compositionality principle for
reference.The key point is the thesis that reference is inconstant or shifty; it
depends on context. This is how the appeal to sense enables Frege to
resolve the puzzle of identity. It provides him with the shifted reference.
The idea that reference is dependent on context is one Frege puts forward
as a general thesis: ‘If words are used in the ordinary way what one intends
to speak of is their reference. It can also appear that one wishes to talk
about the words themselves or their senses’ (1969: 58–9). In direct
quotation, Frege thinks, one talks about the words themselves. Entirely in
parallel, he thinks, in the context ‘John knows that. …’ the reference of an
expression is what would ordinarily be its sense. It is quite clear, he thinks,
that when we report in indirect quotation what someone has said, or what he
knows or believes, words do not have their ordinary reference in the ‘that’-
clause specifying what was said, or what is known or believed. When I say
‘John knows that the Evening Star is the Evening Star’ the reference of the
first occurrence of ‘the Evening Star’ is not the planet Venus, otherwise I
could substitute ‘the Morning Star’ salva veritate. But nor is the expression
itself the reference since I can substitute other expressions, e.g., ‘Hesperus’
without fear of changing the truth-value, so long as they have the same
sense. So the sense is here the only plausible reference.
This is quite intuitive. When we say what John knows we ascribe
knowledge of a thought to John, the thought that the Evening Star is the
Evening Star. Thoughts are determined by senses – they are for Frege the
senses of sentences and composed of the senses of their parts. If we
substitute an expression with a different sense we alter the thought.
So Frege’s explanation of the invalidity of the inference from ‘John
knows that the Evening Star is the Evening Star’ and ‘the Evening Star is
the Morning Star’ to ‘John knows that the Evening Star is the Morning
Star’ is that there is a change in the subject from the sense of ‘the Evening
Star’ in the first premiss to the sense of ‘the Morning Star’ in the
conclusion.
As well as the argument from the informativeness of true identity
statements Frege has a subsidiary argument for sense. There are meaningful
(complex) singular terms without reference: for example, ‘the least rapidly
convergent series’, ‘the King of England’ and ‘the largest prime number’.
These can occur in meaningful sentences (ones expressing thoughts). They
can even occur in true sentences. For example, ‘Misinformed John believes
that the King of England is called “Charles”’. Frege thinks that the
existence of such expressions is a defective feature of ordinary language,
because they produce truth-value gaps. But on his account they must have
senses. Hence sense must be something other than reference since here
there is no (direct) reference.
Frege’s theory of sense thus comprises (i) an account of cognitive value,
(ii) an account of propositional attitude reports, (iii) a theory about empty
singular terms and (iv) an account of the relation between language and the
world.
A competing account of all these matters is provided by Russell, the
second of Kripke’s targets.
Russell on Descriptions and Names

Russell made two contributions to the topic Kripke is concerned with: his
quantificational theory of descriptions and his description theory of
(ordinary) proper names. The two are logically independent: Kripke argues
against the second but raises no objection to the first; Frege, on the other
hand, although he is with Russell in assimilating ordinary proper names to
descriptions, does not group descriptions with quantificational phrases as
Russell does – in fact, Frege’s own account of descriptions is one of the
primary targets of Russell’s attack in ‘On Denoting’.
I begin with Russell’s theory of descriptions and then sketch his theory of
ordinary proper names. As we shall see, despite its centrality in Kripke’s
discussion, the latter is not much more than an appendix in Russell’s
treatment.
Russell’s theory of descriptions is most famously presented in ‘On
Denoting’. Here he regards himself as simultaneously arguing against Frege
as well as his own earlier self.
The easiest approach to Russell’s criticism of Frege and the development
of his theory of descriptions is via the problem of empty singular terms.
These pose a problem for Frege because he is led to conclude that sentences
containing such terms must be truth-valueless (except in cases where the
terms have their indirect referents); a conclusion he thought tenable
because, given the sense/reference distinction, he was able nonetheless to
ascribe such terms a sense and regard sentences containing them as
expressing thoughts.
However, in the case of empty definite descriptions like ‘the least rapidly
convergent series’ this position is deeply implausible. Russell points this
out in his first criticism of Frege (in which for ‘meaning’ read ‘sense’ and
for ‘denotation’ read ‘reference’):
… consider ‘the King of France is bald’. … this phrase, though it has a meaning … has certainly no
denotation, at least in any obvious sense. Hence one would suppose that ‘the King of France is bald’
to be nonsense, but it is not nonsense, since it is plainly false. Again, consider such a proposition as
the following: ‘If u is a class with only one member, that one member is a member of u’… This
proposition ought to be always true. But ‘the u’ or ‘that one member’ is a denoting phrase, and it is
the denotation, not the meaning, that is said to be a u. Now if u is not a unit class, ‘the u’ seems to
denote nothing, hence our proposition would seem to become mere nonsense as soon as u is not a
unit class.
(Russell 1956: 46)

This is not as clear as one might like (it is not, of course, a consequence of
Frege’s position that ‘the King of France is bald’ is nonsense), but the
general point is clear: Frege is committed to saying, both of subject-
predicate sentences like ‘The King of France is bald’ and of truth-
functionally complex sentences like ‘If u is a class with only one member,
then that one member is a member of u’, containing non-denoting definite
descriptions, that they are neither true nor false. However, that is
unacceptable, for such sentences can be false (the former case) and even
true (the latter).
The problem arises for Frege because he groups definite descriptions
with proper names, ordinarily so-called, and regards both types of
expression as having objects as their references. By contrast, he treats
quantifying expressions, that is, expressions which can be used in answers
to the questions of the form ‘How many Xs?’–‘at least one X’, ‘some X’,
‘at most one X’, ‘exactly three Xs’, ‘any X’, ‘all Xs’– as having as their
references properties of properties (in his terminology second-level
functions from first-level concepts to truth-values). Thus for Frege
quantificational statements, i.e. ‘How many?’ statements, are statements
about properties of objects, not statements about objects themselves. Hence
he analyses ‘Some King of France is bald’ as stating that a relation of
coinstantiation holds between the properties being a King of France and
being bald and not as making any statement about any object which is a
King of France.
Russell’s solution to the problem of empty definite descriptions is
essentially that definite descriptions should be treated, like ‘some King of
France’ on Frege’s theory, as quantifying expressions. We can see what ‘the
King of France is bald’ actually says, and what its true logical form is, if we
split it up into three components:

(1) There is a least one King of France

(2) There is at most one King of France
(3) Anything which is a King of France is bald.

Each of these three sentences is quantificational in form (the third says

how many things which are Kings of France are bald). (Russell phrases this
by saying that the sentences make assertions about propositional functions.)
But their conjunction is equivalent to ‘the King of France is bald’. Putting
the pieces together, introducing modern logical symbolism and the
abbreviations ‘is F’ for ‘is a King of France’ and ‘is G’ for ‘is bald’, we
arrive at:

as the perspicuous representation of ‘The King of France is bald’. And this

comes to represent for Russell the logical form of any sentence of the
grammatical form ‘The F is G’.
Thus it is unproblematic, on Russell’s analysis, to regard ‘the King of
France is bald’ as false.
To see the parallelism with ‘some King of France is bald’ it suffices to
know that the latter has, according to both Frege and Russell, the logical
form ‘9x (Fx &Gx)’ which differs from that of ‘the King of France is bald’
by the omission merely of the final conjunct asserting uniqueness. The
difference between ‘the King of France is bald’ and ‘some King of France
is bald’ is thus merely that the former makes a more complicated assertion.
In essence, then, this is Russell’s theory of descriptions; it is simply an
extension of Frege’s treatment of quantifying expressions to expressions to
which Frege had not thought of applying it.
At a stroke then, with respect to definite descriptions at least (we shall
come on to proper names shortly), Russell appears to have resolved the
puzzle of empty singular terms, which causes so much trouble within the
Fregean framework, and consequently has shown that at least as regards
this puzzle, no appeal to the notion of sense is necessary.
Moreover, as Russell goes on to explain, his approach has other
advantages. Within the Fregean framework quantifying expressions have
what logicians call scope (about which more in the next chapter). Just as we
can distinguish between ‘(7 3) + 5’ and ‘7 (3 + 5)’ which differ in the order
in which the mathematical operations are to be applied, so we can
distinguish between ‘some King of France is not bald’ and ‘it is not the case
that some King of France is bald’, representing the former as ‘for some x,
Fx and it is not the case that Gx’ and the latter as ‘It is not the case that (for
some x, Fx and Gx)’. The difference between the two is that in the former
the quantifier has wide scope relative to the negation operator (as ‘+’ has
wide scope relative to ‘×’ in ‘(7 3) + 5’) and in the latter it has narrow scope
relative to the negation operator.
Now treating descriptions as reducible to quantifiers allows us to extend
the notion of scope to descriptions, and so we can discern an ambiguity in
‘The King of France is not bald’, which can be read either as having the
logical form ‘for some x, Fx and not: Gx and for some y, if Fy then x = y’
and so as being false, or as having the logical form ‘Not: for some x, Fx and
Gx and for all y, if Fy then x = y’, and so as being true. (Russell speaks of
‘the King of France’ as having primary occurrence when the sentence is
read in the first way and as having secondary occurrence when it is read in
the second way.) Thus Russell is able to claim for his theory that it can
account for the logical truth of the Law of Excluded Middle. By that law
either ‘The King of France is bald’ or ‘The King of France is not bald’ must
be true. According to the theory of descriptions the former is false, but the
latter is ambiguous, and if we read the description as having secondary
occurrence, it is true, so ‘either the King of France is bald or the King of
France is not bald’ has a true reading despite the non-existence of the King
of France.
The possibility of treating descriptions as having scope is also appealed
to by Russell to resolve the Fregean puzzle of identity.
 Russell uses his own example to illustrate:
If a is identical with b whatever is true of the one is true of the other, and either may be substituted
for the other in any proposition without altering the truth or falsehood of the proposition. Now
George IV wished to know whether Scott was the author of Waverley, and in fact Scott was the
author of Waverley. Hence we may substitute Scott for the author of Waverley, and thereby prove that
George IV wished to know whether Scott was Scott. Yet interest in the law of identity can hardly be
attributed to the first gentleman of Europe.
(1956: 49)

The problem is to explain the invalidity of the inference from ‘George IV

wished to know whether Scott was the author of Waverley’ and ‘Scott was
the author of Waverley’ to ‘George IV wished to know whether Scott was
Scott’. Russell’s response is to appeal to the distinction between primary
and secondary occurrences of a description and to discern ambiguity in the
argument. If we read the description ‘the author of Waverley’ in ‘George IV
wished to know whether Scott was the author of Waverley’ as having a
primary occurrence, its logical form (using ‘Wx’ to mean ‘x wrote
Waverley’) is: ‘for some x, Wx and for all y, if Wy then y = x and George
IV wished to know whether Scott = x’. This says that one and only one man
wrote Waverley and George IV wished to know whether Scott was that man,
that is, that George IV wished to know of the man who in fact wrote
Waverley whether Scott was he. This would have been true, for example, as
Russell says, if George IV had seen Scott at a distance and asked ‘Is that
Scott?’
From ‘George IV wished to know whether Scott was the author of
Waverley’ so interpreted, it does follow, given that Scott was the author of
Waverley that, in a sense, George IV wished to know whether Scott was
Scott. That is, it follows that he wished to know of Scott whether Scott was
he, i.e. ‘for some x, x = Scott and George IV wished to know whether Scott
= x’.
However, ‘George IV wished to know whether Scott was the author of
Waverley’ is naturally read in a way that gives the description ‘the author of
Waverley’ a secondary occurrence, as having the logical form: ‘George IV
wished to know whether for some x, Wx and for all y, if Wy then y = x, and
Scott = x’. But from ‘George IV wish to know whether Scott was the author
of Waverley’ so interpreted and ‘Scott was the author of Waverley’, Russell
says, we cannot infer that George IV wished to know whether Scott was
Scott, where this has the logical form ‘George IV wished to know whether
Scott = Scott’. This is so because there is no constituent, ‘the author of
Waverley’,in ‘George IV wish to know whether Scott was the author of
Waverley’ understood in this way, for which we could substitute ‘Scott’
(1956: 52). (This is obscure. In fact, Russell’s explanation only works on
the assumption that Russell’s propositional functions, like Frege’s senses,
are intensional entities, i.e. that distinct propositional functions can be
satisfied by exactly the same objects.)
The theory of descriptions just sketched is accepted by Kripke and is
always in the background in Naming and Necessity. What Kripke is critical
of is Russell’s treatment of names as descriptions, to which we now turn.
Russell proposes that whenever someone utters a sentence containing an
ordinary proper name he must have in mind some description by which he
would identify its bearer. This description might vary from speaker to
speaker, but there will always be some description which a competent
utterer of the sentence could provide, and this description will be what he
means by the name. Thus, as Russell puts it:
the names we commonly use, like ‘Socrates’, are really abbreviations for descriptions. … When we
use the word ‘Socrates’ we are really using a description. Our thought may be rendered by some such
phrase as ‘the master of Plato’,or ‘the philosopher who drank the hemlock’ or ‘the person whom
logicians assert to be mortal’.
(1956: 201)

But the most important point to appreciate for our purposes is that, despite
his putting it this way himself, it is in fact misleading, without qualification,
to describe Russell as saying that names abbreviate descriptions. As
indicated above, he thinks that although every time a speaker uses a proper
name he must have some identifying description in mind, this may vary
from speaker to speaker and time to time. He does not think that for every
name there is a description which for every speaker on every occasion
‘gives the meaning’ or ‘fixes the reference’ of the name. He does not, in
short, think of names as having a shared public meaning at all. In fact, he
thinks that from speaker to speaker and time to time the only thing constant
so long as the name is rightly used is the object to which the name applies.
Public language is of no interest to Russell except as an overlapping of
idiolects. A striking passage from Russell’s Lectures on Logical Atomism
makes this clear:
When one person uses a word, he does not mean by it the same thing as another means by it. … It
would be absolutely fatal if people meant the same thing by their words. It would make all
intercourse impossible … Because the meaning you attach to your words must depend on the nature
of the objects you are acquainted with, and … different people are acquainted with different objects.
(Russell 1918, also in 1956: 195)

This is a glimpse into the epistemology and metaphysics behind Russell’s

rejection of public language. Propositions are composed wholly of
constituents with which we are acquainted, and acquaintance is a relation
each of us has only to a very narrow set of epistemically privileged items
varying from thinker to thinker. I am acquainted with myself and items
immediately given to me in sense, but not to you and the items immediately
given to you in sense, and conversely. The sense in which idiolects overlap
is therefore a very thin one indeed: as Russell says, what will be constant
between two speakers is simply the object to which the name applies.
He illustrates how, even so, we can communicate with names in the
passage about Bismarck:
we often intend to make a statement not in the form involving the description, but about the actual
thing described. That is to say, when we say anything about Bismarck, we should like, if we could, to
make the judgement that Bismarck alone can make, namely the judgement of which he himself is a
constituent. In this we are necessarily defeated, since the actual Bismarck is unknown to us: but we
know that there is an object B called Bismarck, and B was an astute diplomatist. We can thus
describe the proposition we should like to affirm, namely ‘B was an astute diplomatist’… What
enables us to communicate in spite of the vague descriptions we employ is that we know that there is
a true proposition concerning … Bismarck, and that, however we may vary the description (so long
as the description is correct), the proposition described is still the same. This proposition … is what
interests us.
(Russell 1912: 31; 1963: 158)
According to this account Russell’s view is that the meaning of an ordinary
proper name is its bearer, but this is a fact merely about its shared meaning
in a public language. Consequently, the puzzles about proper names which
are analogous to those about definite descriptions by appeal to which
Russell motivates his theory of descriptions cannot be solved at the level of
public language, but they can be solved at the level of thought. Russell
cannot explain the possibility of a true negative existential statement, such
as ‘Vulcan does not exist’, by saying that it is to be analysed as meaning
‘The intra-Mercurial planet does not exist’. But he can say that whenever an
utterer of such a sentence is saying anything the thought in his mind will be
one that has descriptive content and is thus, by the theory of descriptions,
something that can be, and will be if he is not mistaken, clearly true. In fact
Russell’s formulations, when he is considering the problem of the
significance of affirmative and negative existential sentences containing
proper names, always emphasize the role of the individual speaker’s
association of a particular description with a proper name. For example,
‘When I say, e.g. “Homer existed”, I am meaning by “Homer” some
description, say “the author of the Homeric poems”, and am asserting that
those poems were written by one man’ (Russell 1918 also in 1956: 252).
Russell thus held that everyday proper names are merely, with the
qualifications now made clear, stand-ins for descriptions, and hence what he
called ‘incomplete symbols’. But he also held that there were proper names
in the narrow logical sense, simple symbols with no parts which were
symbols, which could name only objects with which their users were
acquainted, objects which he called ‘particulars’. Ordinary proper names do
not satisfy this paradigm and so are not genuine proper names:
what passes for names in language, like ‘Socrates’, ‘Plato’ and so forth, were originally intended to
fulfil this function of standing for particulars and we do accept, in ordinary life, as particulars all
sorts of things that really are not so. … A name, in the narrow logical sense of the word whose
meaning is a particular, can only be applied to a particular with which the speaker is acquainted,
because you cannot name anything you are not acquainted with.
(Russell 1956: 200)
When we use ordinary proper names the same puzzles arise as for
descriptions. For the use of genuine proper names of objects of
acquaintance, on the other hand, the puzzles cannot arise. If ‘a’ and ‘b’ are
two logically proper names of the same thing ‘a = b’ must be self-evidently
true if true at all; if a genuine proper name names nothing it is meaningless;
and no genuine proper name can occur in a significant existential sentence.
It follows, Russell thinks, that we are not related to any ordinary material
objects in a way that allows us to name them with logically proper names;
the objects we name using logically proper names, the objects with which
we are acquainted, must be extremely short-lived, without parts, and
incapable of appearing other than they are.
There is an important contrast here with Frege. Russell maintains the
simple view that the sole semantic function of a genuine name is to refer.
He deals with apparent counterexamples by restricting the class of genuine
names, eliminating ordinary proper names from the class. Frege just rejects
the simple view. For him there are no logically proper names. All reference
is mediated by sense and thus indirect in a way that is inconsistent with the
concept of a logically proper name. Thus their agreement on ordinary
proper names rests on a fundamental disagreement.
One way of thinking of Kripke’s arguments against the Frege– Russell
description theory is as arguments designed to show that Russell is wrong to
think that genuine proper names can only be given to the narrowly defined
class of particulars. Ordinary proper names are logically proper names in
the sense intended by Russell and there is a corresponding non-cognitive
notion of acquaintance which can be explicated via the causal-historical
account of the transmission of a name which Kripke sketches. But that does
not require that a user of a name stand in any epistemic or causal relation
whatsoever to its bearer; the causal-historical link is to a baptism, not the
object baptized.
The Frege–Russell Synthesizers

In the late 1950s and early 1960s various writers put forward accounts of
naming inspired by Frege and/or Russell. Two of the most important,
discussed by Kripke in Naming and Necessity, were Searle and Strawson.
Searle’s article ‘Proper Names’ (1958) is the locus classicus of the cluster
theory of descriptions which Kripke sets up as his most serious target. It is
intended to bring out what is shared by a community using a proper name
despite variation in descriptive associations. Searle’s contention is that
names are not descriptions, but ‘pegs’ on to which we can hang
descriptions. Referring uses of proper names do not specify any
characteristics of the object referred to, but Searle answers the crucial
question, how a name is connected to an object, by saying that the
descriptive force of ‘This is Aristotle’ is that a sufficient but so far
unspecified number of the statements regarded by users of the name as
stating essential and established facts about him are true. This is something,
he thinks, that reflects the essential function of names. For names would be
superfluous if they were logically equivalent to descriptions:
the uniqueness and universal pragmatic convenience of proper names in our language lies precisely
in the fact that they enable us to refer to objects without being forced to raise issues and come to
agreement on what descriptive characteristics exactly constitute the identity of the object. They
function not as descriptions, but as pegs on which we hang descriptions.
(Searle 1958: 167)

So, Searle says, it is not a necessary truth that Aristotle ever went into
pedagogy, as it is a necessary truth that the teacher of Alexander did, but ‘it
is a necessary truth that Aristotle has the logical sum, inclusive disjunction,
of the properties commonly attributed to him: any individual not having at
least some of these properties would not be Aristotle’ (1958: 168).
Searle thinks the answer to the question whether ordinary proper names
have sense is thus ‘yes and no’. ‘No’ because they are not used to specify or
describe characteristics of objects; ‘yes’ because they are associated ‘in a
loose sort of way’ with characteristics of the objects to which they refer.
The central contention of Searle’s article, which Kripke captures by
talking about names being associated with a weighted cluster of
descriptions, is that there is a vagueness or fluidity in our use of proper
names, which does not exist in the case of descriptions. Kripke’s arguments
are intended to show that this is not a far enough departure from the views
of Frege and Russell, if it is a departure at all, since we allow for the
possibility of far greater ignorance and error than Searle can accommodate.
The second significant discussion of naming after Frege and Russell to
which Kripke pays attention is Strawson’sin Individuals (1959). What
Strawson adds is the idea that speakers can use names deferentially to
others and so refer successfully even though individually they are woefully
ignorant or substantially erroneous in their views.
Strawson insists on the principle already noted from Donnellan that a
user of a name must be prepared to substitute a description for the name and
that ‘there must be some description [the speaker] could given which
applies uniquely to [the object] he has in mind’ (1959: 181–2). But he
explicitly notes, first, that this description need not be in purely general
terms (a point which is already implicit in Frege, Russell and Searle) and,
second, that the description may include a reference to another’s reference,
so ‘one reference may borrow its credentials from another and that from
another’ (1959: 182).
With these qualifications and additions to the Frege–Russell position we
have reached the description theory Kripke sets up as his target in Naming
and Necessity: names are associated by speakers with vague clusters of
descriptions, which may include demonstrative or indexical elements, may
differ from speaker to speaker and may incorporate deference by some
speakers to others.
Even with the qualifications introduced, Kripke argues, this theory can be
seen to be fundamentally flawed, and its fundamental error is precisely
captured in the principle of identifying descriptions expressed in Strawson’s
statement: ‘one cannot significantly use a name to refer to someone or
something unless one knows who or what it is that one is referring to by
that name’ (1959: 181–2).
Quine

As noted, Kripke’s arguments in Naming and Necessity for the existence of

necessary truths that are a posteriori, and hence not analytic, and for the
existence of a class of non-trivial essential properties, were revolutionary.
The notion of necessity, by the time Kripke wrote, had become identified
with that of a priori truth, the latter with that of analyticity, and finally
analyticity with a product of linguistic convention. Consequently, the idea
of necessary a posteriori truth was thought unintelligible – and still more
unintelligible, as it were, the idea of essential or necessary properties of
things. Something, it was thought, might be necessarily true because
analytic, established by linguistic convention, but then this would be
knowable a priori, by reflection on the convention; and of course, the only
thing that could be necessarily true in this sense would be a sentence, a
piece of language, so that the notion of an object’s possessing a property
necessarily (that is, analytically) was from this perspective quite
unintelligible (unless one were arbitrarily to stipulate a meaning for ‘x is
analytically F’, e.g., ‘it is analytically true that everything is F’).
This is the perspective of Quine’s discussion of modality, which comes
under Kripke’s fire in Naming and Necessity. Rigidity is Kripke’s answer to
Quine’s critique of essentialism and the basis of his rejection of this
dominant twentieth-century perspective on modality.
Of course, Quine has little time for modal notions generally, given his
holistic position that ‘no statement is immune to revision’ and ‘everything
can be held true come what may’ (1953: 43). But crucial to his discussion
of de re modal statements, modal statements ascribing necessary or
contingent properties to objects, and the explanation of his additional
scepticism about them over and above his scepticism about modality in
general, is his distinction between three grades of modal involvement and
his appeal to the notion of referential opacity. Quine thinks that there are
two problems with de re modality, a logical problem, to which a sufficiently
hard-nosed defender of the notion can respond, and a remaining
metaphysical problem. The latter is precisely the requirement of a reversion
to ‘Aristotelian essentialism’ (1953: 22, 155), that is, the rejection of the
linguistic conception of necessity which he regards as the only one worth
considering. Quine’s discussion of quantification into modal contexts, by
which we express de re modality, is accompanied by a parallel discussion of
quantifying into propositional attitude contexts. But Quine’s attitude to the
latter is significantly different. This is because he thinks that there is no
remaining irresoluble metaphysical problem here. And this, in turn, is
because he thinks that there is no need for quantification into propositional
attitude constructions to make objective, context-independent, sense. Not
so, he thinks, for quantification into modal constructions. He perceives the
defenders of de re modality as requiring an objective, context-independent
distinction between the necessary and accidental properties of things which
is independent of how they are described. This is the focus of his
scepticism; and it is in fact what Kripke defends.
An important thesis of Quine’s is extensionalism, which can be explained
most easily by recalling Frege. If we say that two expressions which have
the same reference are coextensive we can define a linguistic context within
a larger expression as extensional if and only if replacing an expression in
that context with a coextensive one results in a new expression coextensive
with the original whole expression. Since we are chiefly concerned with
sentential contexts we can say that a context is extensional if and only if
replacement of an expression within it by a coextensive one leaves the
truth-value of the whole sentence in which it occurs unchanged. We can
now say that a language is extensional if and only if it contains only
extensional contexts.
Quine was a lifelong advocate of extensional languages; he calls himself
a ‘confirmed extensionalist’ and says ‘extensionalism is a policy I have
clung to through thick, thin and nearly seventy years of logicizing and
philosophizing’ (2001: 215). His extensionalism is the advocacy of the
elimination of any linguistic construction which allows expressions to
contribute to the truth-values of sentences containing them in virtue of
features other than their extensions.
Quine was an extensionalist because he saw himself as constructing a
language suitable for science in which all unclear constructions are banned.
Thus any construction violating extensionalism had to be avoided or
translated. Furthermore, Quine combined this austere approach to language
with an austere empiricist, specifically Humean, approach to (our
knowledge of) the world. His view was that necessity could not be
discovered in the world because it was not a feature of it (an outlook he
shared with all his opponents in the debate about de re modality until
Kripke came on the scene). So on two accounts modal constructions were
an offence to Quine: they apparently allow us to speak of features of the
world which cannot be discovered and indeed do not exist and, unless in
some way eliminated, they make language non-extensional.
In various papers and, in particular, in ‘On Three Grades of Modal
Involvement’ (in Quine 1966) Quine thus takes up the challenge that
modality presents. The position represented in these papers is the one
Kripke attacks.
Quine distinguishes three degrees of modal involvement: the first is
entirely safe, the second is harmless but unnecessary and a temptation to the
third, the third is metaphysically repugnant to his empiricism.
Consider ‘9 > 7’. Quine’s first grade of modal involvement comes in
when one claims (in the notation of the paper): Nec ‘9>7’. This says that the
sentence ‘9> 7’ is a necessary truth. In the sentence ‘Nec “9> 7”’ ‘Nec’ is a
predicate attached to the name of a sentence ‘9>7’. Quine’s second grade of
modal involvement comes in when one claims (again in the notation of the
paper – note that ‘nec’ is here written in lower case): nec (9 > 7). Here there
are no quotation marks. We are not talking about a sentence; lower-case
‘nec’ is not a predicate, it is a statement operator. The third grade of modal
involvement comes in when one says: for some x, nec (x > 5). This says
that something is necessarily greater than 5. Again no quotation marks are
present. Hence no sentence is spoken about.
 The difference between the first two grades thus depends on the
difference between predicates and statement operators. A predicate attaches
to a singular term to form a sentence which is about that for which the
singular term stands. A statement operator attaches to a sentence to form a
larger one. So ‘it is not the case that’ is a statement operator, as is ‘John
believes that’ and ‘it is necessary that’.
So the crucial difference between Quine’s first two grades is that (upper-
case) ‘Nec’‘is a semantical predicate attributable to statements as notational
forms – hence attachable to names of statements’ (1966: 156), whereas
(lower-case) ‘nec’ operates on statements to form new statements. Thus the
presence of the semantical predicate ‘Nec’ in a language is in conformity
with Quine’s view that necessity is a feature of language, if anything, and is
consistent with extensionality.
We now turn to the difference between the second and third grades of
modal involvement. ‘For some x, nec x > 7’ employs the statement operator
‘nec’. But it occurs as part of the open sentence ‘nec x > 7’, which the
whole statement says is true of something. We here quantify into the scope
of the operator. The whole statement is formed by attaching a quantifier to
an open sentence which contains the operator ‘nec’. Quine therefore
describes the operator here as functioning not merely as a statement
operator but as a sentence operator. A sentence in which ‘nec’ appears
merely as a statement operator, i.e. attached to a closed sentence or
statement, is a de dicto modal statement. One in which it appears as a
sentence operator is a de re modal statement. Quine’s rejection of de re
modality is his rejection of the use of ‘nec’ as a sentence operator.
To see the basis of his rejection we need to turn to his concept of
referential opacity. Consider the argument: ‘The number of planets is 9. 9 >
5. Therefore, the number of planets > 5’. This is an unproblematic
application of Leibniz’s Law (the principle of the substitutivity of
identicals). Now consider the argument: ‘The number of planets is 9, nec (9
> 5). Therefore, nec (the number of planets > 5)’. There is a reading of the
conclusion on which it is false, since it is not a necessary truth, i.e. not an
analytic truth, that there are more than five planets.
 This is reminiscent of the puzzle of identity in Frege and Russell’s
writings. But Quine’s perspective is different. He concludes that in ‘nec (9
> 5)’, since ‘9’ does not stand for 9 (since if it did the argument would be
valid) it does not refer at all, it is irreferential. So he calls the context in
which ‘9’ occurs within the scope of ‘nec’ referentially opaque. This, Quine
thinks, explains why, as shown by the invalidity of the argument, the
context following ‘nec’ offends against the principle of extensionality.
It is important that ‘referentially opaque’ does not merely mean non-
extensional for singular terms. Rather, for Quine, referential opacity is the
explanation of the non-extensionality the ‘number of planets’ example
illustrates. A referentially opaque context is one in which a term does not
refer at all. Quine writes: ‘If “nec (… > 5)” can turn out to be true or false
of the number 9 depending on how that number is referred to evidently … it
expresses no genuine condition on objects of any kind’ (1966: 158). Quine
does not consider an account of the non-extensionality created by the ‘nec’
operator which involves, as in Frege, the idea of a shift in reference. For
Quine there is a simple contrast between a term’s occurring purely
referentially and its occurring irreferentially, with no intermediate
possibility.
Recall now Quine’s overall position: (i) the use of the semantical
predicate ‘Nec’ is unproblematic, (ii) the use of ‘nec’ as a statement
operator in ‘nec (9 > 5)’ is harmless but unnecessary, but (iii)
theuseofthesentenceoperator ‘nec’ in ‘for some x, nec x > 5’ is
metaphysically repugnant. Why is the use of ‘nec’ as a statement operator
harmless given that it creates non-extensional contexts?
Consider the argument: ‘The number of planets is 9. The sentence “9>5”
expresses a necessary truth. Therefore the sentence “the number of planets
> 5” expresses a necessary truth’. This is fallacious in the same way as the
corresponding argument using the statement operator ‘nec’. But the reason
here is apparent, and there is not even an apparent flouting of extensionality.
‘9’ does not occur referentially in the second premiss. In fact, it does not
occur there at all. The occurrence of the mark ‘9’ in the second premiss is
just an orthographic accident due to the fact that we form names of
expressions using quotation marks. The numeral ‘9’ no more occurs in the
second premiss than the general terms ‘cat’ and ‘car’ appear in the sentence
‘Cattle are carnivorous’.So the subject matter of the second premiss is the
sentence ‘9> 5’ of which the mark ‘9’ is merely a physical part. And the
subject matter of the conclusion is the different sentence ‘the number of
planets > 5’. So, of course, the argument is invalid and there is no
counterexample to extensionality.
Quine thinks that the flouting of extensionality that appears with the use
of ‘nec’ and the second grade of modal involvement, in fact, is no more
genuine. Strictly speaking, he thinks, ‘nec (9 > 5)’ involves a use of ‘nec’ as
a predicate rather than an operator. It means that 9 > 5 is a necessary truth.
This ascribes to the object picked out by the singular term ‘that 9 > 5’ the
property expressed by the predicate ‘is a necessary truth’. So there is no real
failure of extensionality in the invalidity of the inference to ‘nec (the
number of planets > 5)’ from the identity ‘9 is the number of planets’. The
inference requires instead the identity: that 9 > 5 = that the number of
planets > 5. And this identity is false. Moreover, the occurrences of ‘9’ and
‘the number of planets’ in the names ‘that 9> 5’ and ‘that the number of
planets > 5’ are no more problematic than their occurrences in the quotation
names ‘9>5’ and ‘the number of planets > 5’. We should be no more misled
in this case than in the latter to think that they occur here as more than
orthographic accidents or think that the truth of the identity ‘9= the number
of planets’ suffices for the validity of the inference. In fact, Quine thinks,
the second grade of modal involvement is even more similar to the first than
appears so far. For we can simply regard ‘that 9 > 5’ as another name of the
sentence ‘9> 5’. Thus we can regard ‘nec (9 > 5)’ as merely a different way
of writing ‘Nec “9>5”’. Quine writes: ‘the use of “nec” as a statement
operator is easily converted into the use of “Nec” as a semantical predicate.
We have merely to supply quotation marks’. He concludes: ‘we may, of
course, just note the conversion in principle and leave it undone in practice’
(1966: 166).
But he thinks that it is important to note it in principle. For the adoption
of ‘nec’ as a statement operator tempts one to go further and use it as a
sentence operator subject to quantification. ‘The momentousness of this
step,’ Quine writes, ‘tends to be overlooked save as one expressly conceives
of “nec”, in its use as a statement operator, as shorthand for its semantical
usage’ (1966: 166) – in which role Quine is prepared to count a sentence
containing the statement operator ‘nec’ as true only if it remains a truth
after conversion into one in which a sentence is explicitly mentioned and
said to be a mathematical or logical truth.
We now come to the final grade of modal involvement – the use of ‘nec’
as a sentence operator. This cannot be replaced by a semantical predicate of
sentences. ‘For some x, nec x > 5’ cannot be replaced by ‘for some x, Nec
“x>5”’; in the latter the quantifier is a meaningless prefix.
Hence to understand the use of ‘nec’ as a sentence operator we must have
an understanding of necessity as something that applies to the possession of
properties by objects which are not sentences and independently of how
they are described. This is why it involves accepting that some properties of
objects are essential to them, and others accidental, inconsistently with the
linguistic conception of necessity. Quine finds ‘no semblance of sense’ in
this distinction.
He gives two reasons: one logical, the other metaphysical. The first is an
argument that it is impossible to make non-trivial sense of de re modality
whilst acknowledging that modal contexts are non-extensional. The
argument is this:
(1) If something (which is not merely trivially analytically true of everything) is necessarily true of
an object then it is necessarily true of it in virtue of its being what it is. (2) If so, how we refer to the
object can make no difference to what is necessarily true of it. (3) But if it makes no difference to
something’s being necessarily true of an object how we refer to it, any two terms for the object must
be substitutable salva veritate in modal contexts. (4) However, one cannot always substitute different
terms for the same object salva veritate in modal contexts. So nothing (not analytically true of
everything) is necessarily true of any object.

The fourth premiss here is supported by the difference in truth-value

between ‘Necessarily, 9 > 5’ and ‘Necessarily, the number of planets > 5’.
But we can deny the third premiss. It depends on the assumption that two
terms which refer to one and the same object in some context must both
refer to that same object in any context in which either refers to it, so it is
indifferent which term is used. This assumption underpins Quine’s belief
that all terms occurring in referentially opaque contexts occur
irreferentially, i.e. do not refer at all.
In fact, though, Quine acknowledges that a logical system can be devised
which does not accord with this assumption (1953: 155). We can see why,
and hence we can understand why Quine can have no decisive logical
argument against the coherence of de re modality by thinking back to Frege.
A term which shifts in reference in the way Frege suggests never occurs
irreferentially, though such reference shifts will obstruct the application of
the substitutivity of identity. But if some terms (say, definite descriptions)
designating an object in non-modal contexts undergo a change of reference
in modal contexts and others (say, proper names) are constant in reference
between non-modal and modal contexts (this distinction between types of
term, of course, is something Frege himself never suggested) then the
argument against modality de re given above is blocked. That is, if in
‘Necessarily, 9 > 5’‘9’ stands for the number 9, as it does in ‘9is the number
of planets’, but in ‘Necessarily, the number of planets > 5’‘the number of
planets’ stands not for the number of planets, as it does in ‘9 is the number
of planets’, but for the sense of ‘the number of planets’ then the move from
the former to the latter on the basis of the identity claim involves a change
in subject matter. Hence it cannot be expected to preserve truth-value. We
then have a distinction between two types of term (the constant and the
shifty), the third premiss is false and Quine’s argument fails to show that
quantification into modal contexts cannot be given a non-trivial sense. Thus
the existence of the possibility of exploiting Frege’s system in this way
shows that Quine has not established the logical incoherence of non-trivial
ascriptions of de re modality.
However, he does not think that he has. Distinguishing between
coreferential terms, in this way or some other, is exactly what he thinks a
defender of de re modality must do. His deepest concern depends on his
Humean conception of necessity, not disputed by any of his opponents
before Kripke, as something belonging not to the world but only to our
conception of it. Even if we do draw a distinction between essence-
revealing singular terms and others Quine thinks that the distinction can
only be an interest-relative matter of convention and so not something fit
for purpose from the essentialist’s point of view. It cannot correspond to any
objective fact about the nature of things, as its defenders require.
That this is the fundamental source of Quine’s repudiation of de re
necessity emerges in his paper ‘Quantifiers and Propositional Attitudes’
(1956, reprinted in 1966: 183–94) and in subsequent papers on the same
topic. ‘Quantifiers and Propositional Attitudes’ deals with a problem very
like the problem of de re modality. And Quine adopts a very similar
approach. But the differences reveal a difference in his attitude to modal
and propositional attitude constructions.
Consider ‘Ralph believes that someone is a spy’. This is ambiguous
between ‘There is someone whom Ralph believes to be a spy’ and ‘Ralph
believes that there are spies’. Quine finds the first reading problematic in
exactly the way he finds de re ascriptions of necessity problematic. Belief
constructions are apparently not extensional. If the man at the beach is the
man in a brown hat and Ralph believes that the man in a brown hat is a spy
but does not believe (explicitly denies) that the man at the beach is a spy, is
there or is there not someone whom he believes to be a spy? Should we say
that he believes the man in the brown hat to be a spy? But the man in the
brown hat is the man at the beach, whom he explicitly denies to be a spy. In
the modal case Quine dismisses de re claims, but he thinks he cannot so
easily dismiss de re propositional attitude ascriptions (because as he says, it
will be of great interest to the FBI if there is someone Ralph believes to be a
spy). He accepts the distinction between de re and de dicto propositional
attitude ascriptions and attempts to explain it within the confines of his
extensionalist philosophy.
His solution is to say that propositional attitude verbs are multigrade
predicates. He begins by appealing to what he calls ‘intensions’. The
intension of a sentence is what it says, the proposition it expresses. The
intension of a predicate is the property it ascribes. Quine now says that on
its de dicto reading, ‘Ralph believes that there are spies’, our example
sentence asserts a relation between Ralph and the proposition that there are
spies. In the sentence so interpreted ‘believes’ expresses a two-place
relation between a person and a proposition. The de re claim ‘There is
someone whom Ralph believes to be a spy’ involves a three-place relation
between Ralph, another unnamed person and a property. It says that there is
someone x such that Ralph believes of x the intension being a spy. If
Ortcutt is the man in question this is so because ‘Ralph believes being a spy
of Ortcutt’ is true. All the terms in this sentence, ‘Ralph’, ‘being a spy’ and
‘Orcutt’ have referential occurrence and can be replaced salva veritate by
any coreferential term.
Now consider:
Ralph believes that the man in a brown hat is a spy

and
Ralph believes that the man at the beach is a spy.

Quine says that these sentences are ambiguous. We can read them as
expressing the two-place relation of believing between a person and a
proposition. Then the first can be true and the second false consistently with
the identity of the man in a brown hat and the man at the beach. Or we can
read them as expressing the three-place believing of relation. Then they
have the same truth-value given the identity. But this is not a problem,
Quine thinks, because when we read ‘Ralph believes the man at the beach is
a spy’ in the second way, it does not entail what it says read in the first way.
So the principle of extensionality is adhered to: every singular term can be
replaced salva veritate with any other singular term for the same object
wherever it refers.
But Quine is not content with this treatment, since it requires an appeal to
intensions, which can only be defined in terms of the obscure notion of
synonymy. So he suggests that, just as we can replace the (lower-case)
statement operator ‘nec’, in the modal case, with the (upper-case)
semantical predicate ‘Nec’, so we can regard belief as a relation between
people and linguistic items: statements and predicates (or closed and open
sentences). So ‘Ralph believes the proposition that the man in a brown hat
is a spy’ is translated as ‘Ralph believes-true “the man in a brown hat is a
spy”’; ‘Ralph believes of the man in a brown hat the intension being a spy’
is translated as ‘Ralph believes-true “x is a spy” of the man in a brown hat’.
Being explicit about the language we get as a translation of the first
sentence: ‘Ralph believes-true “the man in a brown hat” as that is
understood in English’, and as a translation of the second ‘Ralph believes-
true “x is a spy” as that is understood in English of the man in a brown hat’.
Thus it seems that Quine has been able to establish the difference
between de re and de dicto ascriptions of propositional attitude without
offending against his policy of extensionality. De re ascriptions of
propositional attitude turn out to be no worse off than de dicto ascriptions.
But, in fact, matters are more complex. Suppose a de dicto ascription is
true: Ralph believes the proposition that the man in a brown hat is a spy.
Does a de re ascription follow (given the existence of the man in the brown
hat)? Does Ralph believe of the man in a brown hat, i.e. the man at the
beach, the intension that he is a spy? In ‘Quantifiers and Propositional
Attitudes’ Quine says ‘Yes’. He puts this by saying that exportation from de
dicto propositional attitude claims to de re propositional attitude claims
must be regarded as implicative. This, for Quine, is crucial, because it gives
us an account of when an ascription of a de re propositional attitude claim is
true. It follows that if Ralph believes de dicto that the man in a brown hat is
a spy and also believes de dicto that the man at the beach is a law-abiding
citizen then he believes de re of the man in a brown hat, i.e. the man at the
beach, that he is a spy and also believes de re of the man at the beach, i.e.
the man in a brown hat, that he is a law-abiding citizen.
A comparable account in the modal case would involve that if it is
necessarily true that 9 > 5 and necessarily true that the number of planets
numbers the planets, then it is necessarily true of 9, i.e. of the number of
planets, that it is greater than 5 and also necessarily true of the number of
planets, i.e. of 9, that it numbers the planets, and, in fact, that any concept 9
numbers is one it necessarily numbers. No defender of a non-trivial
distinction between essential and accidental properties could say this, of
course, so Quine’s account of when a de re propositional attitude ascription
can be considered true does not provide his opponents in the debate over de
re modality with a rejoinder to his arguments.
However, a problem about exportation (due to Sleigh 1968; see also
Kaplan 1969) causes Quine to reject the position that de re propositional
attitude ascriptions can be objectively true or false. Suppose Ralph believes
correctly that there are spies. He also believes correctly that no two are
exactly the same height. So he believes that there is a shortest spy. Hence he
believes that the shortest spy is a spy. If we export we must say that there is
someone, the shortest spy, call him ‘Shorty’, of whom Ralph believes that
he is a spy. But this seems absurd. Surely, if there is someone of whom it is
true that Ralph believes that he is a spy, Ralph should be of interest to the
FBI. But he is of no interest to the FBI. Without exportation, however,
Quine has no general account of when de re ascriptions of propositional
attitude are true.
Others have suggested a way out. The idea is that we accept exportation,
but restrict the singular terms on which it is allowed. We distinguish, for
example, ‘the man in the brown hat’ from ‘the shortest spy’ and allow that
if Ralph believes that the man in a brown hat is a spy it follows that there is
someone of whom Ralph believes that he is a spy, but deny that it follows
from the fact that Ralph believes that the shortest spy is a spy that there is
someone he believes to be a spy. Kaplan (1969) suggests this idea and
introduces the terminology of a ‘vivid name’. So to have a de re belief
becomes to have a de dicto belief that can be expressed with a vivid name.
From Quine’s point of view, however, the notion of a vivid name is
wholly context-relative. The ability to use a vivid name for someone goes
with knowing who that person is. But, Quine argues, this ‘is utterly
dependent on context. Sometimes … we see the face and want the name,
sometimes the reverse. Sometimes we want to know his role in the
community. Of itself the notion is empty’ (1981: 121). So Quine’s final
position is that we should ‘omit propositional attitudes de re from our
overall scientific language couched in the extensional grammar of predicate
logic’ (1995: 97).
 This might seem a great loss, but Quine says, ‘it grows on one. I now
think the distinction [between de dicto and de re ascriptions of propositional
attitudes] every bit as empty, apart from context, as … that of knowing who
someone is. In context it can be important’ (1981: 121).
Quine’s view is thus not that no distinction can be drawn between those
de dicto propositional attitude ascriptions which are relevant to a de re
propositional attitude ascription and those which are not, but that we cannot
make the demarcation once and for all, outside the conversational context.
De re propositional attitude ascriptions have no place in a context-
independent scientific language, but we can still sensibly make them in
everyday life. However, in doing so the truth of our ascriptions may depend
on our purposes and the context of our utterance.
Let us now return to the difference between Quine’s final position on de
re propositional attitude ascriptions and his position on de re modal claims.
Quine is prepared to allow that in a specific conversational context a de re
propositional attitude ascription makes sense. He similarly thinks that
ascriptions of essential properties can make sense in context. He continues
the passage quoted above by making the comparison explicitly: ‘It [i.e. the
notion of who someone is] and the notion of essence are on a par. …
Relative to a particular enquiry some predicates … may be treated as
essential’ (1981: 121). Elsewhere he writes:
the requirement that distinguishes de re from de dicto, namely knowing who or what, is a function of
the contextual situation and not a general distinction. This classifies it with the indexicals; also with
necessity and possibility, according to my view of them. The indexicals and necessity and possibility
are convenient in daily discourse … ; and we can say the same of propositional attitudes de re …
while admitting none of these idioms to absolute or nonindexical scientific discourse.
(Leonardi and Santambrogio 1995: 358)

Nevertheless, he insists that the defenders of essence are wrong. He is not

prepared to allow that the legitimacy of de re modal claims can be defended
by saying that they have a context-dependent sense. Rightly, he perceives
that his opponents need more. The only possible defence of their position,
he insists, must involve an objective, context-independent, distinction
between essence-revealing terms and others (1953: 155), analogous to
Kaplan’s distinction between vivid and non-vivid names. That is what he
denies, hardly surprisingly after all, since he takes exactly the same view of
de dicto ascriptions of necessity. In Word and Object he distinguishes what
he regards as the philosophical fiction of logical modality from what
ordinary usage allows:
In ordinary non-philosophical usage ‘possibly’ usually serves merely as a modestly impersonal
rewording of …‘I am not sure but what’. Ordinarily the construction ‘necessarily’ connotes … a
propositional attitude of purpose or resolve. Sometimes, also, ‘necessarily’ and ‘possibly’ provide a
condensed way of saying that a sentence follows from or is compatible with some fixed premisses
understood as background. And sometimes they provide little more than a variant style for ‘all’ and
‘some’.
But what is called logical modality is none of these things. Used as a logical modality,
‘necessarily’ imputes necessity unconditionally and impersonally, as an absolute mode of truth, and
‘possibly’ denies necessity, in that sense, of the negation.
(Quine 1960: 195)

In From Stimulus to Science (1995: 99) he remarks, ‘the adverb

“necessarily” is useful indeed, but only as an expository guide’. And in
Perspectives on Quine (Barrett and Gibson 1990: 244) he has this to say
about logical necessity, the supposed ‘absolute mode of truth’:
Can I make sense of logical necessity? Extensionally, yes. I can make sense of logical truth. Given
truth an inventory I can demarcate the set of all logical truths. Similarly, given the chemical
vocabulary, the chemical truths. They are the truths that remain under all substitutions on the
component lexicon that leave the chemical vocabulary undisturbed, and correspondingly for the
logical truths. Calling them necessary adds nothing for me.

Kripke’s rigid designators are supposed to play the role in relation to de re

ascriptions of necessity that Kaplan’s vivid names play in relation to de re
propositional attitude ascriptions. The distinction between them and non-
rigid designators is precisely intended to be an objective, context-
independent one, providing the basis for exportation in modal contexts and
the ground of unconditional, impersonal ascriptions of de re necessity as an
absolute mode of truth. Kripke’s key claim in his discussion of de re
modality in Naming and Necessity is that we have firm intuitions,
independent of context of utterance, that certain ascriptions of essential
properties to objects are true and others are false, so that both Quine’s
critique of the distinction and his insistence that no notion of necessity
beyond the linguistic be allowed fly in the face of common-sense and
everyday linguistic practice. Whether this is so is the crucial question a
reader of Naming and Necessity must decide to determine whether to
endorse the Kripkean essentialist revolution.
3
Naming
The Target

The first target of Kripke’s attack in Naming and Necessity is the Frege–
Russell description theory of names, conceived as a Searlean cluster theory.
Kripke gives the theory a very careful, six-thesis, formulation. As he
formulates the theory it is indexical and idiolectic. This makes his claim to
have refuted it very ambitious indeed. If he is right, the principle that
reference requires an identifying description, however weakly formulated,
is refuted and externalism, as Putnam defines it, established. Nevertheless,
he is right to formulate it in this way since, as the previous discussion
makes plain, only an indexical, idiolectic theory would have been endorsed
by either Frege or Russell, or, of course, Searle or Strawson.
So Kripke’s opponent is no straw man. He breaks down the cluster theory
into the following six theses (the speaker is A) (1980: 64–5):

(1) To every name or designating expression ‘X’, there corresponds a

cluster of properties, namely the family of properties j such that A
believes ‘Xis j’

This thesis, Kripke says, is just a definition. The substantive theses are the
ones that follow:

(2) One of the properties, or some conjointly, are believed by A to pick

out some individual uniquely

This does not say that the properties pick out someone uniquely. It just says
that A thinks that they do. He thinks he knows an answer to the question
‘Which thing (who) is X?’.

(3) If most, or a weighted most, of the j’s are satisfied by one unique
object, y, then y is the referent of ‘X’.
This thesis captures the Frege–Russell intuition that the link between
language and the world must be made by the speaker’s beliefs and
intentions. The notion of a ‘weighting’ is to capture the cluster theorist’s
intuition that some properties are more important than others in the ‘vote’.
Who decides on the weighting? A. This does not mean that we have to
imagine that A is capable of saying explicitly which properties he regards
as constitutive of the meaning of the name. The weighting will be manifest
in his reaction to circumstances in which some but not all the properties in
the set are possessed by one individual. The reference to a ‘weighting’ is a
reference to A’s dispositions: his dispositions to respond to the presentation
in various circumstances of information relevant, as he sees it, to the
existence of X. We can elicit these dispositions by asking him to suppose
that these circumstances obtain, or by creating a situation in which he
actually believes that they do. We will in general be limited to doing the
former (it is by appeal to such suppositions that Kripke argues against the
Frege–Russell theory using his famous Einstein, Gödel–Schmidt and
Feynman examples).
Kripke adds three theses to complete the theory:

(4) If the vote yields no unique object, ‘X’ does not refer.
(5) The statement ‘If X exists, then X has most of the j’s’ is known a
priori by the speaker.
(6) The statement ‘If X exists, X has most of the j’s’ expresses a
necessary truth (in the idiolect of the speaker).

Kripke adds two subsidiary theses (1980: 66). Subsidiary to thesis (5) is the
thesis that it is a priori true for the speaker that if nothing has most of the j’s
‘X’ does not refer. Subsidiary to thesis (6) is the thesis that ‘If nothing has
most of the js, X does not exist’ expresses a necessary truth in the idiolect
of the speaker.
Theses (5) and (6) also have converses (Kripke 1980: 73). The converse
of thesis (5) is that it is known a priori by A that if any unique thing has the
j’s it is X. The converse of thesis (6) is that ‘if anything has most of the
j’sitisX’ is a necessary truth in the idiolect of the speaker.
This is the theory Kripke aims to refute in Naming and Necessity. But
before considering his criticisms we need to note a constraint he imposes
without which it would be circular. This is important because one of the
main responses to Kripke is that the constraint against circularity required
by description theories is much weaker than he suggests, and the
description theorist is not as restricted as Kripke thinks in the descriptions
to which he can appeal.
Kripke motivates the introduction of his non-circularity constraint by
noting what he calls an ‘incidental defect’ in the way the defenders of the
cluster theory often state it. He takes the ‘Moses’ discussion from
Wittgenstein to illustrate the point. Wittgenstein writes: ‘when one says
“Moses did not exist” one may mean: the Israelites did not have a single
leader when they withdrew from Egypt. Or: their leader was not called
“Moses”. Or: there cannot have been someone who accomplished all that
the Bible relates of Moses’. The gist of Wittgenstein’s discussion is that we
know a priori that if all the Bible relates of Moses is false, Moses did not
exist. But, Kripke points out, there is a distinction between a complete
fiction and a substantially false account of a real person. Such an account
may come to be associated with a name. Legends grow up about its bearer
and the less interesting truth gets distorted or forgotten. Eventually the tales
passed down through the generations contain a substantially false account.
But they are still tales about the real person, and not about a fictional
character of which they are true by definition (contrast here, perhaps, our
talk of King Arthur and our talk of Merlin; it is plausible that the second
character is wholly fictional). This distinction as such is no problem for the
description theory. But it indicates that an adequate account must appeal to
a Strawsonian deferential use of names. A natural response is to say that in
our talk of Moses we intend our reference to be to that person to whom the
biblical story relates, even if it gives a wholly false account of him.
It is at this point that Kripke imposes his circularity constraint.
Deferential reference is allowed, but circularity is not. We can use ‘Moses’
deferentially to the authors of the Bible only because we know that they are
not deferring to us. If they were, it is obvious that no reference would be
secured by the deferential intention. Hence Kripke imposes constraint (C)
on the properties in the cluster of properties (1980: 68):
(C) For any successful theory the account must not be circular. The properties which are used in the
vote must not themselves involve the notion of reference in a way that is ultimately impossible to
eliminate.

This is very reasonable. Suppose the only thing A can say when asked who
X is is: the person B calls ‘X’. Then if the cluster of descriptions theory is
correct, whether A refers at all when he uses the name depends on whether
B does. But, likewise, if all B can say in answer to the same question is: the
person A calls ‘X’,heis in the same boat. Neither is referring to anyone if
the cluster theory is correct.
Constraint (C) is reasonable because it is very weak. It does not exclude
the possibility that one of the properties in the reference-determining set is
one in whose specification the concept of reference must be used. Whether
such a property can play a reference-determining role depends on empirical
facts. I can con-fidently ‘pass the buck’ to the authors of the Bible because I
know they did not pass the buck to me. ‘Whatever Jones means by “X”’ can
perfectly well serve as a reference-fixing description for me if Jones does
not pass the buck back to me. The only way of specifying a reference which
constraint (C) rules out a priori is ‘the thing I here and now mean by “X”’ –
where ‘X’ is the very name whose reference I am specifying. It does not
even rule out someone referring to someone by an utterance of a name
when his only identifying description is one that makes reference to that
very utterance. The crucial question about Kripke’s attack on the
description theory in Lecture II is whether, in the cases he considers
counterexamples to it, there are any reference-fixing clusters satisfying
constraint (C), as he himself formulates it, available to the user of the name.
Giving the Meaning, Fixing the Reference and
Rigid Designation

Before getting further into Kripke’s arguments we first need to note the
crucial distinction between thesis (6) and the others. Thesis (6) is the thesis
in which modality is introduced (necessity is not mentioned in any of the
preceding theses). It must belong to the cluster theory if that is considered
as a theory of meaning. The other theses belong to it even considered, less
ambitiously, merely as a theory of reference-fixing. Interpreting the cluster
theory as a theory of meaning, it entails that any name is fully equivalent to
a description and hence interchangeable with it salva veritate in all contexts
(except where the expressions are not used but only mentioned). The cluster
theory, considered merely as a theory of reference-fixing, does not entail
this. In particular, the cluster theorist qua theorist of reference-fixation need
not require the full substitutability salva veritate of names and descriptions
in modal contexts. Kripke’s criticism of the cluster theory in Lecture I is
directed only at the cluster theory as a theory of meaning. So his target is
solely thesis (6).
Much of Lecture I is given over to explaining the difference between the
cluster theory as a theory of meaning, i.e. including thesis (6), and as a mere
theory of reference-fixing, i.e. minus thesis (6). Making the distinction also
leads Kripke into a discussion of the concept of a possible world, in terms
of which he first explains the notion of rigidity.
An expression is a rigid designator if it designates the same object with
respect to every possible world in which it exists. The fact that language
users in some other possible worlds use ‘Nixon’ to refer to someone else
does not show that ‘Nixon’ as we use it in the actual world is not a rigid
designator (it is a trivial fact that any word could have been used to mean
something different). Nor, of course, as Kripke is at pains to point out in the
Preface to Naming and Necessity (1980: 7ff.), does the fact that many actual
people are called ‘Nixon’. It is only of a particular use of the name, e.g. the
conventional use of the name as a name for a disgraced President, that it
makes sense to ask whether it is a rigid designator. Just so, there is no sense
in asking whether the description ‘the odd one’ is rigid unless one specifies
that the context is one in which one is talking of numbers and ‘odd’ means
‘not numerically even’ or one in which one is talking of philosophers and
‘odd’ means ‘eccentric’ (see Fitch 2004: 44).
But to see what the notion of rigid designation actually amounts to we
need to get rid of what Kripke calls the ‘misleading metaphor’ of identity
across possible worlds (1980: 47–8).
Consider Kripke’s example of Benjamin Franklin, who was the first
Postmaster General of the United States and the inventor of bifocals.
The sentences:

(1) The inventor of bifocals might not have been the inventor of
bifocals.

and:

(2) Someone other than the inventor of bifocals might have been the
inventor of bifocals.

are ambiguous. (1) can mean either:

(1*) Concerning the man who in fact invented bifocals: he might not
have been the inventor of bifocals.

or:

(1**) The following might have been the case: the inventor of bifocals
was not the inventor of bifocals.

This is a scope ambiguity, comparable to the scope ambiguity of Russell’s

example ‘George IV wondered whether Scott was the author of Waverley’,
or Quine’s example ‘the number of planets is necessarily greater than five’.
(1*) seems true because the inventor of bifocals, Benjamin Franklin, might
not have got round to inventing bifocals. But (1**), since it says that the
proposition that the inventor of bifocals was not the inventor of bifocals is a
possible truth, is false.
Similarly, (2) is ambiguous between the true:

(2*) Concerning the inventor of bifocals: someone other than he might

have been the inventor of bifocals.

and the absurd:

(2**) The following might have been the case: someone other than the
inventor of bifocals was the inventor of bifocals.

However, if we replace ‘the inventor of bifocals’ in (1) and (2) by

‘Benjamin Franklin’ these ambiguities disappear.

(3) Benjamin Franklin might not have been Benjamin Franklin.

can only be heard one way, as saying of the man Benjamin Franklin that he
might not have been the man Benjamin Franklin. And this is absurd.
Benjamin Franklin might not have been called ‘Benjamin Franklin’, he
might have had a wholly different career and done none of the things we in
fact know him by, but he would still have been Benjamin Franklin.
Similarly:

(4) Someone other than Benjamin Franklin might have been Benjamin
Franklin.

can only be heard as saying of the man, Benjamin Franklin, that someone
other than he might have been the man Benjamin Franklin, and this is
absurd. For although someone other than Franklin might have been called
‘Benjamin Franklin’ and might have done the things which made Franklin
famous, no one other than Franklin could have been Franklin.
 That (3) and (4), unlike (1) and (2), are not ambiguous, Kripke says,
shows that the name ‘Benjamin Franklin’ is a rigid designator, whose
designation in any possible world is its actual designation, whereas ‘the
inventor of bifocals’ is a non-rigid designator, whose designation in any
possible world is whoever, in that world, satisfies the condition of being the
inventor of bifocals.
We thus see the rationale for the two intuitive tests for rigid designation
given in Chapter 1: that there are negative answers to the two questions (1)
Does ‘X might not have been X’ have a true reading? (2) Does ‘someone
other than X might have been X’ have a true reading? An account which
accords better with Kripke’s intentions than the admittedly misleading
explanation in terms of identity across worlds is the following: a singular
term ‘X’ is used as a rigid designator if it is so used that it refers, when it
figures inside counterfactual or other modal linguistic contexts, to what it
refers to, if at all, outside such contexts (this formulation is given by Hale
2004: 367). This makes no commitment to possible worlds and a fortiori
makes no use of any notion of identity across possible worlds. It explicitly
makes the rigidity of a proper name a matter of our intentions in using it,
but this is surely what Kripke thinks. Note that both tests are required
because ‘my father’ satisfies the second test, but it is not a rigid designator,
it does not designate my father with respect to every possible world in
which he exists. A stronger notion is that of an obstinately rigid designator
(Salmon 2005: 34): ‘X’ is an obstinately rigid designator if it designates the
same thing with respect to every possible world, regardless of whether that
thing exists in that world. So ‘Benjamin Franklin’ is an obstinately rigid
designator, since ‘It might have been that Benjamin Franklin failed to exist’
is unambiguously true, but ‘the man identical to Benjamin Franklin’ is not,
since ‘it might have been that the man identical to Benjamin Franklin failed
to exist’ is not unambiguously true.
An important distinction to note (not made in these terms by Kripke) is
between constant and inconstant rigid designators (Mackie 2002). An
example of a constant rigid designator is ‘the square of three’. This is what
Kripke calls a strongly rigid designator, since it designates nine, a necessary
existent. But in addition, given what it means, its conventional dictionary
significance, it could not have designated anything else (of course, it could
have meant something else and so been used by language users in some
other possible world to designate something else). Contrast the description
‘the actual number of planets’. This too is a strongly rigid designator by the
intuitive tests. But holding its conventional significance constant it could
have designated some other number (it would have done if there had been
only six planets). Hence although the necessitation of ‘the actual number of
planets is the number of planets’ (i.e. ‘it is a necessary truth that the actual
number of planets is the number of planets’) is false, ‘the actual number of
planets is the number of planets’ could not have been false, given its actual
conventional significance.
Kripke notes this distinction (1980: 59–60, n. 22) but says nothing
explicitly against the suggestion that names abbreviate such ‘rigidified’
descriptions, at one place (1980: 6, n. 8) remarking that his principal thesis
‘contrasts names with non-rigid descriptions as advocated by Russell’. But
it is highly relevant to his modal argument against the description theory, as
many have noted at least from Searle (1983) on, since the description
theorist can maintain his position, and, in particular, thesis (6), even whilst
acknowledging that proper names are rigid, if he says that such names are
inconstant rigid designators like ‘the actual number of planets’.
A distinction Kripke does emphasize, not in the original text of Naming
and Necessity, but in the Preface to the second edition, is one between de
facto and de jure rigidity.A de jure rigid designator is one whose reference
is stipulated to be a single object whether we are speaking of the actual or a
counterfactual situation, whilst a de facto rigid designator is a description
which happens to be constructed using a predicate true of one and the same
object in each possible world. A description like ‘the square of three’ or
‘the smallest prime’ is a de facto rigid designator. Kripke maintains that
names are de jure rigid designators. Given that they are rigid designators it
is trivial that this is so if it follows from the fact that they are simple
symbols and so have to have their meanings stipulated. But this is
consistent with a name’s being a mere abbreviation of an actualized
description like ‘the actual number of planets’ and hence an inconstant rigid
designator. How to classify such inconstant rigidly designating descriptions
with respect to the de jure/de facto distinction is not discussed in Naming
and Necessity. Kripke does say, however, that although he in fact believes
that names are de jure rigid, he is content with the weaker assertion of
rigidity.
We can now turn to Kripke’s account of the distinction between the
cluster theory qua theory of reference-fixing and qua theory of meaning. As
illustration he uses the example of the standard metre rod, taken from
Wittgenstein.
The puzzle he starts from is that it might seem that if ‘one metre’ is
defined to be the length of a stick S at a certain time t it must follow that it
is a necessary truth that S is one metre long at t. But, Kripke argues, this
consequence does not follow if the definition is understood merely as
giving the reference rather than giving the meaning. Someone who uses the
definition to give the reference can still say correctly ‘If heat had been
applied to stick S at t then at t stick S would not have been one metre long’.
The reason is that ‘the length of stick S at t’ is a non-rigid designator,
whereas ‘one metre’ is a rigid designator. A merely reference-fixing
definition of a name (‘one metre’ is a name of an abstract object, a length)
is then one in which the intention is to introduce a rigid designator via a
non-rigid designator. Whether it is a constant or inconstant rigid designator
is something Kripke does not discuss. Prima facie there are two possibilities
here. The introducer of the term could intend to introduce a rigid designator
which is synonymous with ‘the actual length of stick S at t’ and so
inconstant, or he could intend to introduce a constant rigid designator
whose meaning requires it to have a particular length as its designation. So
there is the question: is the meaning of ‘one metre’ in the idiolect of the
original introducer of the term such that the statement ‘one metre is the
length of stick S at t’, meaning what it actually means, has to be true? In
fact, although the notion of a constant rigid designator is unproblematic, as
shown by the example of ‘the square of three’,itis problematic to suppose
that by fixing its reference by a non-rigid designator one can introduce a
name by stipulation as a constant, rather than inconstant, rigid designator,
since to do so would be to stipulate that one’s very stipulation would have
determined a different meaning if things had been different. What would
show, so long as the name was used only by the original introducer, that it
was a constant rather than an inconstant rigid designator?
What Kripke does argue is that it is not a necessary truth that stick S is
one metre long at t since under other possible circumstances S would not
have been one metre long at t. However, he argues that for the introducer of
the term ‘one metre’, ‘S is one metre long’ is something he knows a priori
(just as, for someone who stipulates that the newly coined term
‘squatchelor’ is to apply to all and only unmarried men, it is a priori that
squatchelors are unmarried men). In this sense ‘S is one metre long’ is a
contingent a priori truth, for it is a contingent truth that S is one metre long,
but the introducer of the term knows a priori that S is one metre long.
Thus, given the distinction between rigid and non-rigid designation and
the possibility of a reference-fixing definition in which a rigid designator
has its reference fixed by a non-rigid description, the possibility in this
sense of contingent a priori truths appears to follow.
Kripke goes on to point out (1980: 57) that the distinction between
reference-fixing and synonym-giving definitions can be applied generally.
There is a difference between using the stipulation ‘Aristotle is the greatest
man who studied under Plato’ as a synonym-giving definition of the name
‘Aristotle’ and using it merely to fix the reference of the name. If we do the
latter once again there will apparently be a contingent a priori truth
forthcoming since for the introducer of the name it will be something he
knows a priori that, if he existed, Aristotle studied under Plato; but it is not
a necessary truth that Aristotle, if he existed, studied under Plato.
Kripke’s picture of naming, in fact, is that in every case a name is
introduced by a procedure that can be thought of as reference-fixing – either
by description or, what can be thought of as a special case of the former,
ostension – and then the name is passed on from speaker to speaker without
the original reference-fixing description necessarily being passed on.
 There is a final point to be made now, crucial to the assessment of
Kripke’s modal argument. If a name (‘Benjamin Franklin’) is synonymous
with a description (‘the inventor of bifocals’) thesis (6) must be true
(‘Benjamin Franklin, if he existed, was the inventor of bifocals’, must
express a necessary truth). But the converse does not follow. Thesis (6) can
be correct even if the name and the associated description differ in meaning
so long as this difference is consistent with the necessary truth in the
idiolect of the speaker of the identity statement ‘X is the j’. But
thoroughgoing synonymy is not needed for the necessary truth of this
identity statement. So defenders of Kripke’s target thesis (6) can still say
that names are not synonymous with the descriptions they are associated
with, and some, as we shall see, do.
The Modal Argument

Kripke states his modal argument against the description theory at a number
of places (1980: 53, 57, 61 and 74). Its full statement in the original text of
Naming and Necessity on page 61 is as follows:
The [description theorist] would say that the name is simply defined synonymously, as the cluster of
descriptions. It will then be necessary, not that Moses had any particular property in this cluster, but
that he had the disjunction of them. There couldn’t be any counterfactual situation in which he didn’t
do any of these things. … Such a suggestion … must clearly be false. … Most of the things
commonly attributed to Aristotle are things Aristotle might not have done at all.

In a nutshell the intended argument is this. All proper names are rigid
designators. Any descriptions (at least of contingently existent objects like
Aristotle) which can be at all plausibly thought of as fixing their references,
given what the users of the names believe and intend (in the case of
Aristotle, descriptions referring to achievements commonly associated with
him), are accidental designators. So such descriptions are not substitutable
salva veritate for proper names in modal contexts and hence a fortiori are
never synonymous with them.
To assess this argument we need to see how exactly it is supposed to
refute thesis (6).
Let us go back to the example of Benjamin Franklin and make the
supposition that the only description which is a candidate for synonymy
with ‘Benjamin Franklin’ is ‘the inventor of bifocals’. But as we have seen,
‘Benjamin Franklin’ is a rigid designator, and so it makes no sense to say
that Benjamin Franklin might not have been Benjamin Franklin or that
someone other than Benjamin Franklin might have been Benjamin Franklin.
On the other hand, ‘the inventor of bifocals might not have been the
inventor of bifocals’ has a true reading as well as a false one, as does
‘someone other than the inventor of bifocals might have been the inventor
of bifocals’. This shows that ‘the inventor of bifocals’ is a non-rigid or
accidental designator, and differs in meaning from ‘Benjamin Franklin’.
Moreover, ‘Benjamin Franklin might have existed and not invented
bifocals’ is unambiguously true. So the modal statement ‘it is necessarily
true that if Benjamin Franklin existed he invented bifocals’ is
unambiguously false. The next step in the argument is to the conclusion that
the non-modal statement ‘if Benjamin Franklin existed, then Benjamin
Franklin invented bifocals’ does not express a necessary truth (in the
idiolect of the speaker who fixes the reference of the name by the
description). Kripke concludes that thesis (6) is false since the same
reasoning can be gone through for any name and any description put
forward as the description which gives its meaning, including a description
equivalent to a cluster of descriptions. Of course, the description theorist
would not in fact offer the description ‘the inventor of bifocals’ as fixing
the reference of ‘Benjamin Franklin’. So the supposition Kripke is making
is unrealistic. But other examples make it easier to focus on the crucial
point. Kripke himself gives the example of ‘Jack the Ripper’. Dummett
(1973: 112) suggests ‘St Anne’ as a name which even educated users
associate only with the description ‘the mother of the Virgin Mary’ and
following Gareth Evans (1979) philosophers have come to use ‘Julius’ as
the name of the inventor of the zip, whoever he was.
The argument as stated is what appears to be the most obvious
interpretation of Kripke’s text, but in the Preface to the 1980 book he
refines it, in reaction to Michael Dummett’s statement (in his 1973) of the
first response we will look at below (that rigidity may be explained by
regarding proper names as synonymous with widescope descriptions).
Kripke’s Preface refinement will be explained after Dummett’s original
widescopist response has been set out, and we will then be in a position to
evaluate in turn the widescopist response to this which Dummett gives in
his subsequent book (1981). Kripke himself does not further respond, but
the debate has continued and we will look at one of the most prominent
recent attacks on widescopism.
The Debate Over the Modal Argument

What the defender of thesis (6) must do is to show how it can be true even
though ordinary proper names are typically used as rigid designators. He
cannot deny that ordinary proper names typically do function as rigid
designators so the only matter for argument can be the explanation of this
fact. He has at least four options. The first is the widescopist response,
originally put forward by Dummett (1973: 127ff.), though with
qualifications, together with the supplementary response, involving an
appeal to the distinction between ‘assertoric content’ and ‘ingredient sense’,
presented by Dummett (1981) in his reply to Kripke’s Preface refinement.
The second way of responding to the modal argument, within the
framework of Frege’s philosophy, is to appeal to the idea that proper names
differ from descriptions by maintaining their customary reference in modal
and counterfactual contexts, and do not refer therein to their ordinary
senses. This is a suggestion originally put forward by Burge (1979). A third
way of responding to Kripke’s modal argument is to say that the
descriptions with which names are synonymous are inconstant rigidified
descriptions, like ‘the actual inventor of bifocals’.As we noted, one of the
earliest defenders of this position was Searle (1983), but it has been
defended by many others including Jackson (1998). Finally, a fourth
response available to the descriptivist is to appeal to the distinction between
indicative descriptions and subjunctive descriptions, the distinction between
‘the person who is richer than anyone else’ and ‘the person who would have
been richer than anyone else’. All these responses are ways of defending
thesis (6) from Kripke’s objections by blocking the transition from the
modal claim, intuitively obviously correct in the cases he gives, that if
things had been different X would not have been j, to the conclusion that
the non-modal statement ‘If X exists, X is j’ does not express a necessary
truth in the idiolect of the speaker.
 The fundamental idea of widescopism (which has also been endorsed by
Loar 1976, Schiffer 1977 and many others) is an appeal to a scope
convention. It cannot be denied that ordinary proper names are typically
used as rigid designators, but a possible explanation of this, Dummett
suggests (1973: 127ff.), is that names are in a sense abbreviations of
descriptions, but are used subject to the general convention that in modal
contexts they are to be read as having maximal scope relative to the modal
operators. Certainly, one can introduce an expression which is stipulated to
be synonymous with another except that, unlike the latter, it is to have a
wide scope relative to a certain class of operator. As McCulloch (1989: 110)
points out, one might introduce an expression ‘*’ by the stipulation that it is
to be read as synonymous with the multiplication symbol ‘×’ except that it
is to be read as having wide scope relative to other arithmetical operators.
Thus ‘2 3’ and ‘2* 3’ will identify the same arithmetical task, but ‘4+ 2
3=18’ will be ambiguous (in the absence of a convention specifying the
order in which the calculations are to be carried out), whilst ‘4 + 2 * 3 = 18’
will unambiguously have the same truth-value as ‘(4 + 2) 3= 18’. In a sense
then ‘*’ means the same as ‘×’ but is subject to a convention not governing
the latter, in another sense ‘*’ does not mean the same as ‘×’ since to
understand it requires knowledge of a convention knowledge of which is
not required for understanding of ‘×’. Does ‘2* 3’ means the same as ‘2 3’?
They identify the same arithmetical task. It is unimportant whether one says
it therefore has the same meaning (since here the difference between ‘×’
and ‘*’ is idle) or has a different meaning (since understanding one of its
constituents requires knowledge of a convention not required for
understanding any constituent of the other).
Similarly, it might be, for all that Kripke’s unrefined modal argument
establishes, that ‘Benjamin Franklin’ does mean the same as ‘the inventor
of bifocals’ except that it is subject to the convention that it must be read as
having maximal scope in modal contexts. If so ‘Benjamin Franklin might
not have invented bifocals’ is unambiguously true, but ‘the inventor of
bifocals might not have invented bifocals’ is not. Nevertheless, consistently
with thesis (6), the sentence ‘Benjamin Franklin, if he existed, was the
inventor of bifocals’ expresses a necessary truth. On this proposal the name
and the description do not mean the same inasmuch as understanding the
former requires knowledge that it is governed by a convention that does not
govern the latter. Do ‘Benjamin Franklin invented bifocals’ and ‘the
inventor of bifocals invented bifocals’ then mean the same? There is an
argument for saying so, since the only difference is that only one of them
contains an expression governed by the widescope convention, but that
convention is idle in the context. But there is an argument for saying that
they differ in meaning since there is a sense in which their constituents
‘Benjamin Franklin’ and ‘the inventor of bifocals’ differ in meaning.
Although Dummett is the originator of the widescopist response his own
position in his first discussion of Kripke is nuanced. He does not in fact
accept Kripke’s claim that all proper names are rigid designators, and
therefore, of course, only suggests that the widescope convention governs
those that are. His own example of ‘St Anne’ is a name that he argues is not
a rigid designator. It is no good, he asserts, arguing that ‘St Anne’ cannot
mean the same as ‘the mother of the Virgin Mary’ because St Anne might
never have become a parent. ‘The mother of the Virgin Mary might never
have become a parent’ is ambiguous, as Kripke insists, between a true
reading and a false one, as is ‘The mother of the Virgin Mary could not but
have been a parent’. But ‘St Anne could not but have been a parent’ is
equally ambiguous. ‘Even though there is an intuitive sense in which it is
correct to say “St Anne might never have become a parent”, there is an
equally clear sense in which we may rightly say, “St Anne cannot but have
been a parent” provided always that this is understood as meaning that if
there was such a woman as St Anne, then she can only have been a parent’
(Dummett 1973: 113).
Kripke acknowledges that there is an ambiguity in ‘St Anne cannot but
have been a parent’, Dummett points out, in the sense that he acknowledges
that if the reference of ‘St Anne’ is fixed by the description, ‘the mother of
the Virgin Mary’, the first users of the name know a priori that St Anne, if
she existed, was a parent. In this sense we can say that this statement is
epistemically necessary. But Kripke would insist that it is not
metaphysically necessary that St Anne was a parent; like every mother, she
could have remained childless. Hence the ambiguity is in the modal
operator. The situation is quite different, Kripke thinks, with modal
statements containing descriptions, like ‘the mother of Mary cannot but
have been a parent’. The ambiguity here is a scope ambiguity, which does
not have to be explained by appeal to two notions of necessity. Of course,
Kripke has to make this distinction between the cases if he is to maintain
that ‘St Anne’, unlike ‘the mother of Mary’, is a rigid designator. So
Dummett’s claim that the same ambiguity is present in both cases is a
challenge to a fundamental Kripkean claim.
That proper names are rigid designators seems obvious when we consider
examples like ‘Benjamin Franklin’ or ‘Aristotle’,or other proper names of
well-known historical figures. And so Dummett’s challenge might seem
heroic. However, as he emphasizes, the Kripkean modal argument is
supposed to show that even in the case of a proper name whose users all
associate with it just a single non-rigid description, and which is not used
deferentially by anyone, the name and the description are not synonymous,
since the name, unlike the description, is rigid. That is why Dummett
selects the name ‘St Anne’ as a name for which this assumption is plausibly
true. Another example, from Kripke himself, of which the same is true, is
plausibly ‘Jack the Ripper’, which we may suppose to have its reference
fixed by some such description as ‘the notorious nineteenth-century London
murderer’, which is known to everyone who uses it. Now Kripke’s position
has got to be that whilst ‘the notorious nineteenth-century London murderer
might not have been a murderer’ has two readings because of the scope
ambiguity, this is not so in the case of ‘Jack the Ripper might not have been
a murderer’. In this case the ambiguity, if one is allowed, must be due to a
difference between epistemic and metaphysical modality. But, Dummett
argues, to appeal to different explanations of the ambiguity in the two cases
is implausible. There appears to be the same phenomenon, which therefore,
other things being equal, should be explained in the same way. And this
point is reinforced, he argues, when we look more carefully at the
distinction between epistemic and metaphysical modality, to which Kripke
has to appeal to explain the ambiguity in ‘Jack the Ripper cannot but have
been a murderer’ or ‘St Anne cannot but have been a parent’. The sense in
which these are true, he says, is the sense in which it is true that the
standard metre rod could not have failed to be a metre long. This is an
epistemic sense. Our knowledge that Jack the Ripper cannot but have been
a murderer is knowledge derived from the way the words are used, i.e. from
the fact that ‘the notorious nineteenth-century London murderer’ is used to
fix the reference of ‘Jack the Ripper’. By contrast, the kind of necessity we
are concerned to deny when we say that ‘Jack the Ripper might never have
been a murderer’– for example, if he had died as infant – does not depend
on our use of words, it is a metaphysical necessity. But it appears that we
can draw exactly the same contrast between the two readings of ‘the
notorious nineteenth-century London murderer cannot but have been a
murderer’. When we say that the murderer cannot but have been a
murderer, in the sense in which it is true to say this, we are expressing a
priori knowledge based solely on our understanding of the words, precisely
similar to that expressed when we say that Jack the Ripper cannot but have
been a murderer, or that the standard metre cannot but have been one metre
long. And when we say that the murderer might not have been a murderer
we are concerned with the very same kind of metaphysical necessity
involved when we say that Jack the Ripper might not have been a murderer.
But if the ambiguity in this case is to be explained in terms of scope
ambiguity we cannot regard the modal operators as ambiguous between
epistemic and metaphysical readings.
Dummett’s suggestion that in cases like that of ‘St Anne’, where a single
description is associated with the name by every user and there is no
deferential use, the same ambiguity can be discerned in modal sentences
containing the name as in modal sentences containing the description, does
not require him to say that this is also the case for more typical names, like
‘Benjamin Franklin’ or ‘Aristotle’. His widescopist thesis is that there is a
convention whereby proper names are to be read as having wide-scope in
modal contexts. But if it is only by convention that something is not done
then it is possible to do it. So where the rationale for the convention fails to
apply, the convention will sometimes be flouted. In the case of unusual
names like ‘St Anne’ and ‘Jack the Ripper’, associated with the same single
description by all who use it, it is reasonable to suppose that whatever the
rationale for the convention that names are to be read as having wide scope
relative to modal operators, it does not apply. That is why the narrow scope
reading comes into view as a possibility. And, indeed, a reading of ‘Jack the
Ripper might not have been Jack the Ripper’ as saying something true (on
which reading the first occurrence of the name must have wide scope and
the second narrow) does seem to be available.
This account is consistent with the view Donnellan (1977) puts forward
in his response to Dummett in his paper ‘The Contingent A Priori and Rigid
Designators’. Donnellan explains that the uses of proper names Dummett is
discussing, like Kripke’s example of the use of ‘Neptune’ by Leverrier
subsequent to his introduction of it via the description ‘the cause of the
perturbations in the orbit of Uranus’, are quite unusual ones, in which a
name is pegged during the relevant time to the definite description via
which it was introduced. But in the case of such uses, he suggests, it is
indeterminate whether the name is a rigid designator or an abbreviation for
a description, for nothing in the actual use of the name or its original user’s
intentions can show which it is. The two proposals entail different
consequences. For example, the first entails that ‘If Neptune exists, Neptune
is the cause of the perturbations in the orbit of Uranus’ is a contingent truth,
the second that it is a necessary truth. But all the intuitions we actually have
about the case can be explained on either proposal, so the differences are
not ones that can be used to decide between them. Historically names have
not been introduced by an explicit stipulation that they are to be taken as
rigid designators and so long as a name not introduced by such an explicit
formula, like ‘Neptune’, ‘St Anne’ or ‘Jack the Ripper’, continues to be
associated with the original description by which its reference is fixed,
Donnellan argues, there is no fact of the matter about whether it is a rigid
designator or an abbreviation for a description – entirely unworryingly
since though, for example, it will be indeterminate whether ‘Neptune is a
large planet’ can be paraphrased as ‘the cause of the perturbations in the
orbit of Uranus is a large planet’, anyone who knows how the name was
introduced will know the sentences have the same truth-value.
The description of such situations as ones in which it is simply
indeterminate whether the name is a rigid designator is entirely in accord
with Dummett’s thought. For, as noted, these are plausibly cases in which
the rationale for the general widescopist convention fails to apply, so the
possibility of reading, e.g. ‘It might have been that Neptune existed but was
not the cause of the perturbations in the orbit of Uranus’, as saying
something plainly false (by giving the description the name abbreviates
narrow scope), though excluded by the convention, becomes evident. But,
given the general convention, to which users of the name are subject, it is
excluded. So a user of the name, aware of its peculiar features, yet party to
the general convention, might well feel undecided in the way Donnellan
describes.
Of course, this story depends on a rationale being available for the
widescope convention. But we shall see below that a plausible suggestion
(from Lewis 1984) fits well with the thought that names like ‘St Anne’ and
‘Jack the Ripper’ are ones to which it would be pointless to apply the
widescope convention.
In his Preface Kripke has in fact two objections to wide-scopism. The
first objection is that the proposed convention is ‘unaccountable’ (1980:
13), i.e. has no rationale. As just noted, Lewis’s proposed rationale will be
given shortly. The second objection (1980: 6ff.), which constitutes what I
have called Kripke’s Preface refinement of the modal argument, is that the
distinction between between rigid and non-rigid designators can be made
even when we consider simple sentences containing no modal operators. He
considers the two sentences ‘Aristotle was fond of dogs’ and ‘the last great
philosopher of antiquity was fond of dogs’. These are non-modal sentences.
Nevertheless he argues, they differ in their truth-conditions as they describe
counterfactual situations. There is a certain man of whom it is true that
‘Aristotle was fond of dogs’ truly describes a counterfactual situation if and
only if he would have been fond of dogs if that situation had obtained. The
truth-condition of ‘the last great philosopher of antiquity was fond of dogs’,
as it describes a counterfactual situation, cannot be so stated. The latter is
true as describing a counter-factual situation just in case exactly one person
would have been last among the great philosophers of antiquity and would
have been fond of dogs if that situation had obtained. This (Russellian)
condition for the truth of ‘the last great philosopher of antiquity was fond of
dogs’ can differ wildly from the specified condition for the truth of
‘Aristotle was fond of dogs’. With respect to a counterfactual situation
where someone other than Aristotle would have been the last great
philosopher of antiquity identifying the counterfactual truth-condition for
‘Aristotle was fond of dogs’ with the (Russellian) counterfactual truth-
condition for ‘the last great philosopher of antiquity was fond of dogs’
would (incorrectly!) make that other person’s fondness for dogs the relevant
issuefor thecorrectness of ‘Aristotle was fond of dogs’.
Kripke’s second objection to the widescopist can be restated as follows
(cf. Kripke 1980: 13–14). Consider the simple sentences ‘Benjamin
Franklin was the inventor of bifocals’ and ‘Benjamin Franklin was
Benjamin Franklin’. The assertion:
Benjamin Franklin was Benjamin Franklin. That might not have been the case.

is not true. The assertion:

Benjamin Franklin was the inventor of bifocals. That might not have been the case.

is true. There can be no scope issues here since in each case the first
sentence is a non-modal sentence. Yet the two assertions differ in the way
specified. Hence the two initial non-modal sentences differ in their modal
status, in that one expresses a necessary truth and the other a contingency.
So ‘Benjamin Franklin was the inventor of bifocals (if he existed)’ cannot
express a necessary truth. But since any proper name is rigid the same
argument can be run for any proper name and any non-rigid description a
description theorist could plausibly put forward as fixing its reference. So
once again, thesis (6) of the description theory is refuted.
In his second discussion of Kripke, Dummett (1981) responds to this
refinement of the modal argument. David Sosa (2001) responds similarly
and quotes in a note a comment by David Lewis which nicely summarizes
the underlying thought:
If a description is eager enough for wide scope, its scope may cross sentence boundaries and also
boundaries of quotation and disquotation. … it’s fair to treat the following alike; and in each case to
say that ‘Aristotle’ may take wide enough scope to cover all the rest. (a) With respect to
[counterfactual situation] S, Aristotle was fond of dogs. (b) This is the case with respect to S:
Aristotle was fond of dogs. (c) Consider the simple sentence ‘Aristotle was fond of dogs’. That
sentence is true with respect to S.
… wide-scopers [have been accused] of solving the problem of (a) while ignoring (c). Our reply is
that (c) functions as a mere stylistic variant of (a)
(Sosa 2001, 34–7, n. 7)

Dummett frames his reply by appealing to the distinction between assertoric

and ingredient sense. We can illustrate this non-controversially using the
example of ‘×’ and ‘*’. ‘2 3=18’ and ‘2*3=18’ make exactly the same
assertion, the difference between them cannot affect what is put forward for
belief when the sentences are asserted for one cannot create a new object of
belief by a stroke of the pen. Nevertheless, they differ in ‘ingredient sense’
in that ‘4+2*3=18’ and ‘4+ 2 3=18’ differ in meaning according to the
convention for the use of ‘*’. According to that convention the former is
unambiguous (it states truly that 18 is 18) but the latter is ambiguous in the
absence of a convention for inserting brackets (so there is no sense in
asking for the ingredient sense of ‘2 3=18’ as it occurs in ‘4 + 2 3 = 18’).
Dummett’s suggestion is that ‘Benjamin Franklin’ and ‘the inventor of
bifocals’ differ in the same way as ‘×’ and ‘*’ so that ‘Benjamin Franklin
was the inventor of bifocals’ and ‘the inventor of bifocals was the inventor
of bifocals’ do not differ in assertoric content for someone who associates
the name with the description but do differ in ingredient sense. To assert
that Benjamin Franklin was the inventor of bifocals is just to assert that the
inventor of bifocals was the inventor of bifocals, but ‘it is a necessary truth
that Benjamin Franklin was the inventor of bifocals’ differs from ‘it is a
necessary truth that the inventor of bifocals was the inventor of bifocals’
because of the convention that a proper name be taken as having wide scope
in modal contexts. So ‘Benjamin Franklin was the inventor of bifocals’
differs in ingredient sense from ‘the inventor of bifocals was the inventor of
bifocals’.
Another example illustrating the distinction is that between ‘here’ and
‘where I am’. They have the same meaning for a speaker at a time (setting
aside demonstrative uses of ‘here’ as when I point at a map and say,
pointing to some distant region, ‘it is always raining here’, or Post-it notes
of the type ‘I am not here – back in 5’, the message in which could be more
long-windedly conveyed by ‘I am here, but will not be when you are
reading this – I will be back 5 minutes later than then’). But substitution for
one by the other in the context of temporal operators changes meaning and
truth-value. The two sentences ‘it is always raining where I am’ and ‘it is
always raining here’ differ. The first is ambiguous. On one reading it is true
if wherever I go, it rains. On a second it is true if I am in a place where it
perpetually rains (Manchester). The second sentence has only the second
meaning. We should not say that different things are asserted when I say ‘it
is raining here’ and ‘it is raining where I am’, however. These sentences do
not differ in assertoric content. But they do differ in ingredient sense (if we
take ‘it is always raining here’ to be constructed by attaching the operator
‘it is always the case that’ to the sentence ‘it is raining here’). And in this
case it is plausible to appeal to a scope convention to explain what is going
on. ‘Where I am’ may have wide or narrow scope relative to a temporal
operator; ‘here’ must have wide scope. So ‘here’ functions as a temporally
rigid designator, whereas ‘where I am’ functions as a temporally flexible
designator.
To bring out the analogy with the modal case Dummett suggests that we
imagine a language which contains the expression ‘Thatcherabouts’:
‘Thatcherabouts’ is stipulated to stand to ‘where Mrs Thatcher is’ exactly as
temporally rigid to temporally flexible designator (just as ‘*’ is stipulated to
have the meaning specified above). So ‘it is always raining
Thatcherabouts’, said when Mrs Thatcher is in Manchester, is
unambiguously true. By contrast ‘it is always noisy Thatcherabouts’, said
when Mrs Thatcher is in Manchester, is false (it is sometimes quiet in
Manchester, but always raining). On the other hand, ‘it is always noisy
where Mrs Thatcher is’, said when Mrs Thatcher is in Manchester, may still
be true on one reading (if Mrs Thatcher is always surrounded by a hubbub
wherever she goes). Thus ‘it is raining Thatcherabouts’ and ‘it is raining
where Mrs Thatcher is’ differ in ingredient sense but agree in assertoric
content, and Dummett suggests that there is no reason to think, on the basis
of the evidence adduced by Kripke in his presentation of the modal
argument, that ‘Benjamin Franklin’ and ‘the inventor of bifocals’ differ in
any way that is disanalogous to the difference between ‘Thatcherabouts’
and ‘where Mrs Thatcher is’.
Of course, the distinction between assertoric content and ingredient sense
can always be made whenever there is a distinction between a temporally
rigid description, like ‘the man currently standing next to Mrs Thatcher’,
and a temporally flexible description, like ‘the man standing next to Mrs
Thatcher’. ‘The man currently standing next to Mrs Thatcher is carrying a
gun’ and ‘the man standing next to Mrs Thatcher is carrying a gun’ have the
same assertoric content but differ in ingredient sense, since ‘It is always the
case that the man currently standing next to Mrs Thatcher is carrying a gun’
is unambiguous but ‘It is always the case that the man standing next to Mrs
Thatcher is carrying a gun’ is ambiguous.
With this distinction before us we can now return to the Preface
refinement of the modal argument.
The challenge for Dummett, recall, is to explain the apparent difference
between, for example:
Benjamin Franklin invented bifocals. That might not have been true.

and:
The inventor of bifocals invented bifocals. That might not have been true.

The former assertion seems to be unambiguously true (since Benjamin

Franklin might not have invented bifocals). The latter is still, on one
legitimate reading, made true by the fact that Franklin might not have
invented bifocals (something Kripke himself has to explain without appeal
to Dummett’s apparatus and without acknowledging a false reading of the
former), but on another reading it is false. Again,
The inventor of bifocals was rich. That might not have been true.

unlike:
Benjamin Franklin was rich. That might not have been true.

is ambiguous. No appeal to scope can help explain the difference since in

each discourse the initial sentence is a non-modal sentence.
Dummett’s implicit reply is that the difference here shown between
‘Benjamin Franklin’ and ‘the inventor of bifocals’ is no more significant
than, or other than, that between ‘Thatcherabouts’ and ‘where Mrs Thatcher
is’. The sentences ‘Benjamin Franklin invented bifocals/was rich’ and ‘the
inventor of bifocals invented bifocals/was rich’ have the same assertoric
content, but differ in ingredient sense in modal contexts, and it is these
ingredient senses which are under discussion in the displayed discourses.
That is, the reference of ‘That’ in the first displayed discourse is the
ingredient sense of the sentence ‘Benjamin Franklin invented bifocals/was
rich’ as it occurs preceded by a modal operator, and the reference of the
occurrence of ‘That’ in the second displayed discourse is one of the
ingredient senses the sentence ‘the inventor of bifocals invented
bifocals/was rich’ has when preceded by a modal operator. These are
different just because of the convention that names take wide scope in
modal contexts, and so the two assertions can differ in truth-value.
The temporal case is parallel. We can see this by considering the
difference made by replacing ‘Thatcherabouts’ by ‘where Mrs Thatcher is’
in:
It is raining Thatcherabouts. That is not always true.

to get:
It is raining where Mrs Thatcher is. That is not always true.
The first of these (if we suppose the assertion made in Manchester) is
unambiguously false. The second is ambiguous (supposed still to be made
in Manchester). On one reading its truth requires that it not always be
raining in Manchester. On the second reading it requires that it not always
be raining where Mrs Thatcher is then located. (There is a parallel
ambiguity in ‘The man standing next to Mrs Thatcher is carrying a gun.
That is always true’ or ‘Mary’s boyfriend is very rich. That used to be
true’.) We can account for the difference by appealing to the different
ingredient senses of ‘it is raining Thatcherabouts’ and ‘it is raining where
Mrs Thatcher is’ in the context ‘it is always the case that …’ and
hypothesizing that the two occurrences of ‘That’ refer to these ingredient
senses.
Again consider:
Thatcherabouts is Thatcherabouts. That is always true.

and:
Thatcherabouts is where Mrs Thatcher is. That is not always true.

Given that ‘Thatcherabouts’ has the meaning explained, an utterance of ‘It

is always the case that Thatcherabouts is Thatcherabouts’ must be true. But
an utterance of ‘It is always that case that Thatcherabouts is where Mrs
Thatcher is’ may be false – will be false unless Mrs Thatcher is a permanent
resident of wherever she happens to be at the time of utterance. So if we
suppose that in the language containing ‘Thatcherabouts’ it can occur in
non-temporal sentences, like ‘Thatcherabouts is Thatch-erabouts’, as well
as in temporal sentences, then a difference precisely analogous to that
between proper names and descriptions to which Kripke points in his
challenge to the widescopist is bound to emerge so long as speakers
conform to the principle that what is said by an utterance of ‘…
Thatcherabouts …’ is always true if and only if an utterance of ‘It is always
the case that … Thatcherabouts …’ in the same context is true. In the modal
case Dummett calls the analogous principle (that what is said by, say, an
utterance of ‘… Benjamin Franklin …’ is necessarily true if and only if an
utterance of ‘It is necessarily true that … Benjamin Franklin …’ in the same
context is true) the ‘modal involvement’ principle (1981: 584). He claims
that it is the fact that names and descriptions are governed by different
scope conventions in modal contexts, so that the non-modal sentences in
which they occur differ in ingredient sense, together with the fact that we
are disposed to conform to the modal involvement principle in ascribing
modal values (‘necessarily true’ and ‘possibly true’) to what is said, which
explains why we are willing to say that what is expressed by ‘Benjamin
Franklin is Benjamin Franklin’ is a necessary truth, but that what is
expressed by ‘Benjamin Franklin is the inventor of bifocals’ is not. And
hence, he argues, this willingness is entirely consistent with a
fundamentally descriptivist, Russellian, position, since it is consistent with
‘Benjamin Franklin is Benjamin Franklin’ and ‘Benjamin Franklin is the
inventor of bifocals’ having the same assertoric content and differing only
as ‘Thatcherabouts is Thatcherabouts’ and ‘Thatcherabouts is where Mrs
Thatcher is’ and ‘2*3 =18’ and ‘2 3= 18’.
Kripke has not himself replied to Dummett’s 1981 response to his
argument against widescopism in the Preface to Naming and Necessity, but
the proposal has been a topic of much discussion in recent literature
(Stanley 1997, Soames 1998 and 2002, Sosa 2001, Nelson 2002). Perhaps
the most-discussed argument against widescopism is one given by Soames
(1998), which goes as follows.
The following is a valid form of argument:
The proposition that P is the proposition that Q.
The proposition that Q is a necessary truth.
Hence:
The proposition that P is a necessary truth.

But now consider an instance of this in which ‘N’ is a name the reference of
which is fixed by the contingently satisfied description ‘the F’:
The proposition that if N is G something is both F and G is the proposition that if the F is G
something is both F and G.
The proposition that if the F is G something is both F and G is a necessary truth.
Hence:
The proposition that if N is G something is both F and G is a necessary truth.

The widescopist will note that, of course, this argument is multiply

ambiguous, because of the multiple occurrences of descriptions which may
be read with different scopes. To eliminate these ambiguities read all
descriptions explicitly occurring with narrow scope. Then Soames argues
that the widescopist must say that this argument is invalid, since he must
accept the premisses but must read the conclusion as equivalent to the ‘the
F is such that the proposition that if it is G something is both F and G is a
necessary truth’– which, on the supposition that being F is an accidental
property of the F, is not true. Hence widescopism is refuted since the
argument has a plainly valid form.
To see how the widescopist can respond consider first the two temporal
analogues:
The proposition that it is raining here is the proposition that it is raining where I am
The proposition that it is raining where I am is always true.
Hence:
The proposition that it is raining here is always true.

and
The proposition that it is raining Thatcherabouts is the proposition that it is raining where Mrs
Thatcher is.
The proposition that it is raining where Mrs Thatcher is is always true.
Hence:
The proposition that it is raining Thatcherabouts is always true.

Again, the arguments are ambiguous because of the different scopes

available to the descriptions ‘where I am’ and ‘where Mrs Thatcher is’. But
we can read these as having narrow scope. And we can suppose that Mrs
Thatcher and I are, for our sins, followed by rain wherever we go, so that
the second premiss in each argument is true, but that the places we currently
occupy are not always rainy, so that the conclusions are false (read as they
must be read, given the meanings of ‘here’ and ‘Thatcherabouts’). Must we
say that the arguments are invalid? No. We may instead say that they are
unsound. In each case the first premiss, the identity premiss, is false. ‘The
proposition that it is raining here is identical with the proposition that it is
raining where I am’ is equivalent to ‘Concerning the place I am in, the
proposition that it is raining in that place is identical with the proposition
that it is raining where I am’, which is false (recall that the description
‘where I am’ is to be read as having narrow scope). We can say the same of
‘The proposition that it is raining Thatcherabouts is identical with the
proposition that it is raining where Mrs Thatcher is’. Since ‘Thatcherabouts’
is not an English word it means only what we stipulate that it means. So far
it has just been stipulated that in temporal contexts it is equivalent to a
widescope reading of ‘where Mrs Thatcher is’. No meaning has been given
to its use in non-temporal contexts. So we can stipulate that it is equivalent
to a widescope reading of ‘where Mrs Thatcher is’ as it occurs in ‘the
proposition that it is raining Thatcherabouts is the proposition that …’.
Consequently, we can say that the first premiss of the second argument is
likewise not true, and that the argument is not invalid, but merely unsound.
Returning now to the original argument by Soames against widescopism
we can see what the widescopist can say. He need not deny the validity of
the argument. He can just say that its first premiss is false. His position is
that there is a convention whereby proper names in modal contexts are read
as widescope definite descriptions, but he is not required to say that names
cannot be read as widescope definite descriptions in non-modal sentences
like the first premiss of Soames’s argument. In fact, he can say that as it
occurs in ‘the proposition that N is G is the proposition that …’ the name
should be read as a widescope description for just the same reason that it
should be read as a widescope description in modal contexts, for the same
rationale for the widescope convention applies.
It remains, then, for the widescopist to explain what is the rationale for
the widescope convention. Now, according to the Russellian story the
meanings of proper names differ from speaker to speaker, though their
denotations remain the same. But then, if the widescope convention is not
in force, different speakers who associate different but co-denoting
descriptions with a name may fail to communicate using the name – much
as two men fail to communicate if one says ‘The bank is nearby’, meaning
the riverbank, and the other replies ‘No, it isn’t’, meaning the financial
bank. For even though the descriptions they associate with the name denote
the same object, the same modal sentence uttered by one of them may have
a different truth-value from that which it has uttered by the other if the two
speakers assign narrow scope to the (distinct) descriptions they use the
name to abbreviate. If I mean by ‘Aristotle’ the founder of formal logic, and
you mean the last great philosopher of antiquity, then if I say ‘It is
necessarily true that Aristotle, if he existed, invented formal logic’ and you
repeat this, then if we each assign ‘Aristotle’ in the sentence a narrow
scope, what I say will be true and what you say will be false. Thus,
knowing that we probably associate different descriptions with the name,
assigning it narrow scope in modal contexts is deliberately to make what we
say irrelevant to what our conversational partners say, as we would if in a
conversation about ‘the bank’ I persisted in using ‘bank’ to mean ‘river
bank’ and you persisted in using it to mean ‘financial bank’, each of us
knowing full well that the other meant something different. Again, if I
assign ‘Aristotle’ narrow scope in asserting ‘the proposition that Aristotle is
a logician is the proposition that the founder of formal logic is a logician’,
and you do the same in repeating what I say, but I mean by ‘Aristotle’, ‘the
founder of formal logic’, and you mean, ‘the greatest philosopher of
antiquity’, then you say something false whilst I say something true. From
the position of a Russellian descriptivist theorist the rationale for a
widescope convention is thus clear. As Lewis puts it in ‘Putnam’s Paradox’,
as the first of seven points a sophisticated descriptivist must take on board:
there may or may not be rigidification. If there is, that will avoid confusion between people who have
attached the same name to the same referent by different descriptions. For nothing will be true as one
person means it, but false as the other means it, not even when the name appears in modal contexts.
(1984: 223)

Of course the widescope descriptivist has to say more if he appeals to the

rationale just explained for the convention he proposes. For there are
apparent facts, which it is tempting to explain by the possibility of using a
name as equivalent to a narrow-scope description, where the rationale
would forbid it. These are the very familiar apparent facts which can be
cited by the descriptivist as evidence for his account. ‘Ralph believes that
Tully denounced Catiline’ would seem to be equivalent to ‘Ralph believes
that Cicero denounced Catiline’ with all the proper names understood as
having wider scope than the propositional attitude verb, but on an ordinary
understanding these sentences are not equivalent. Again, ‘Ralph believes of
(about or concerning) Tully that he denounced Catiline’ might be true when
‘Ralph believes that Tully denounced Catiline’ is false. Finally, on an
ordinary understanding it might be true that young Ralph believes that
Santa Claus brings presents, though there is no Santa Claus. The use of
proper names in the specification of the content of propositional attitude
ascriptions is therefore something a widescopist needs to account for.
Additionally, a widescopist has got to say something about the use of names
within the ‘that’-clauses of attitude ascriptions which are in turn embedded
within modal contexts, as in ‘it might have been the case that Ralph
believed that Cicero was an orator’. These are matters to be returned to, but
they are best examined within the context of a Fregean response to Kripke.
The widescopist response to the modal argument is presented within a
Russellian framework; its central notion, scope ambiguity, is a core notion
of Russell’s theory of descriptions. In a Fregean response the central notion
must be that of indirect reference and the central contention that proper
names preserve their direct references in modal and counterfactual contexts.
Modal contexts block substitutivity salva veritate of coreferential
singular terms. Within Frege’s framework this can be accommodated only
by regarding such coreferential descriptions as referring in modal contexts
to their customary senses.
Now the ambiguities present in such modal sentences as:

(1) The inventor of bifocals might not have been the inventor of
bifocals
are similarly present (as noted by Quine) in ascriptions of propositional
attitudes such as:

(2) George IV wondered whether the author of Waverley was Scottish.

From a Fregean viewpoint the ambiguity in the latter has to be explained in

something like the following way. On one reading of(2) it is asserted that a
relation of wondering whether holds between George IV and the thought
identified in the that-clause, the thought that the author of Waverley was
Scottish. On this reading ‘the author of Waverley’ has indirect reference in
(2). On the other reading ‘the author of Waverley’ retains its direct reference
in (2). The sentence asserts of the author of Waverley, Scott, that George IV
wondered whether he was Scottish, that is, that thinking of the author of
Waverley under some mode of presentation, George IV wondered whether
the person so presented was Scottish.
Peter Geach (1976) (see also Kaplan 1969) brings out the ambiguity as
follows. He defines the term ‘aspect’ to mean the Fregean sense of a proper
name, and speaks of an aspect α as an aspect of an object x when α is a
mode of presentation of x. Next he stipulates that ‘[α is F]’ is to stand for
the thought composed of the aspect α and the sense of the predicate ‘is F’.In
‘[α is F]’ the Greek letter does not belong to the intentional context: the
thought that [α is F] is the thought that you would express in language by
attaching the predicate ‘is F’ to a subject term whose sense is the aspect α.
The two readings of (2) are now:

(2*) George IV wondered whether [the author of Waverley is Scottish].

and:

(2**) for some α, α is an aspect of the author of Waverley and George

IV wondered whether [α is Scottish].

The relevant ambiguity in (1) can now be brought out by the two
readings:
 (1*) It might have been the case that [the inventor of bifocals was not
the inventor of bifocals].

and:

(1**) for some α, α is an aspect of the inventor of bifocals and it might

have been that [α was not the inventor of bifocals].

In the case of both (1) and (2) some restriction on type of aspect is
required if the second reading is to capture the intuitive English meaning.
We have already seen what is at issue in our discussion of Quine on
exportation and the ‘shortest spy’. In the case of propositional attitude
ascriptions we do not invariably allow exportation of a singular term from
an opaque position to a transparent position. We do not allow the inference
from ‘Ralph believes that the shortest spy is a spy’ to ‘Ralph believes of the
shortest spy that he is a spy’ and thence to ‘There is someone Ralph
believes to be a spy’. Similarly, from the fact that George IV wondered
whether the author of Waverley was Scottish it is not immediately evident
that we should infer that there was someone of whom he wondered whether
he was Scottish.
The Fregean must say that the existential quantifier form (2**) captures
the intuitive English meaning with the description read as having wide
scope only if the quantification is restricted to what we might call ‘identity-
revealing’ aspects – where an aspect is identity-revealing only if knowing
that [α is X] suffices for knowing who X is.
An identity-revealing aspect may be what Kaplan (1969) calls a vivid
name. Alternatively, if Quine is right that ‘knowing who’ is a highly
context-dependent notion, the notion of an identity-revealing aspect will
also be highly context-dependent. But in most contexts of discussion of
George IV’s interest in the author of Waverley it will be correct to say that
he did not know who the author of Waverley was, although he knew that the
author of Waverley was the author of Waverley, and hence in most contexts
of discussion it would be incorrect to say that there was an identity-
revealing aspect under which George IV thought of the author of Waverley
as the author of Waverley.
The Fregean must proceed similarly to account for the intuitive English
meaning of (1), according to the reading on which the first occurrence of
the description is taken as lying outside the modal operator. So read the
statement says of the man, Benjamin Franklin, that he might never have
become the inventor of bifocals. To capture this meaning we have to regard
the quantification in (1**) as restricted to what we might call ‘essence-
revealing’ aspects. Suppose that Benjamin Franklin was essentially a
human being, so that ‘the inventor of bifocals might not have been human’,
on the intended wide scope reading, is false. Still, bifocals might have been
invented by extraterrestrials. So:
For some α, α is an aspect of the inventor of bifocals and it might have been that [α was not human].

is true if the quantification ranges over all aspects, since ‘the inventor of
bifocals’ expresses an aspect of Benjamin Franklin and it might have been
that the inventor of bifocals was not human.
So the Fregean needs the notion of an essence-revealing aspect if he is to
capture the intuitive thought that the inventor of bifocals had to be human,
but might not have been the inventor of bifocals, just as he needs the notion
of an identity-revealing aspect in order to capture the intuitive thought that,
in some contexts at least, knowing that the shortest spy is the shortest spy
does not count as knowing who the shortest spy is. An aspect α is an
essence-revealing aspect of an object x if and only if it is necessary that [if
anything is α it is F] for any ‘F’ which denotes a necessary property of x,
and it is not necessary that [if anything is α it is F] for any ‘F’ which
denotes an accidental property of x. An essence-revealing aspect of an
object is one that presents it as the possessor of all and only its essential
properties.
There are complications. One possibility is that even if objects have
essential properties, they do not have individual essence, but only general
essences. It may be that I am essentially human but have no other essential
property (so that I could have been anything any human being could be). To
deal with this possibility we need to extend the notion of an ‘aspect’ of an
object x. Say that aspects of x are the senses of predicates satisfied by x.
Aspects as originally defined, senses of Fregean proper names, can be
thought of as senses of predicates of the form ‘is identical with X’, where
‘X’ is the proper name; the aspect expressed by ‘Socrates’ can be thought of
as the sense of the predicate ‘is Socrates’. In the case of quantification into
propositional attitude contexts the aspects quantified over must be
understood as aspects as originally defined.
Of course, this notion of an essence-revealing aspect, like that of an
identity-revealing aspect, may have different applications in different
contexts of use. Maybe, as Quine thinks, we have no context-independent
notion of an essential property. Maybe in some contexts it is correct to say
that Benjamin Franklin might not have been a human being. And, of course,
I have not explained the – possibly highly context-dependent – difference
between essential and accidental properties by reference to the notion of an
essence-revealing aspect; I have just defined the latter in terms of the
former. Just so, the difference between knowing and not knowing who
someone is has not been explained in terms of the notion of an identity-
revealing aspect; rather the latter has been defined in terms of the former.
We know, in most contexts anyway, that knowing that the shortest spy is the
shortest spy does not count as knowing who the shortest spy is, and we
know that the fact that the inventor of bifocals might not have been a
human being does not establish that the actual inventor of bifocals might
not have been a human being. The Fregean must be able to accommodate
the differences between the two readings of ‘the inventor of bifocals might
not have been the inventor of bifocals’, just as he must be able to
accommodate the differences between the two readings of ‘Ralph believes
that the shortest spy is a spy’. The Russellian does so by appeal to scope
distinctions. The Fregean must do so by appeal to the possibility of
quantifying over senses.
Now a Fregean response to Kripke’s modal argument can be given, i.e. a
Fregean account of what the rigidity of proper names amounts to, entirely
parallel to the Dummettian ‘widescopist’ Russellian response. Namely,
although every name has on any occasion of use a sense which might well
be that of a description, there is a convention in force whereby a proper
name, as opposed to a description, must not be used in a modal context to
refer to its indirect reference, i.e. its customary sense, and, in general, must
not be used by a speaker to refer to the sense he associates with it in
contexts where replacement by a proper name with the same direct
reference but a different customary sense is not guaranteed to preserve
truth-value. This proposal, originally put forward by Burge (1979), is
further defended by Hale (2004). Burge expresses the point by saying that
proper names always maintain their customary reference in counterfactual
contexts. Hale similarly writes: ‘We could say: d is used as a rigid
designator if and only if it is used in such a way that it refers when it figures
inside counterfactual and other contexts to what it refers to, if to anything at
all, outside such contexts’ (2004: 367). (Of course this is not, in itself, a
distinctively Fregean account of rigidity; it becomes so only if the
possibility being excluded for rigid designators is viewed as reference,
inside counterfactual or modal contexts, to sense, which is Hale’s intention.)
With this convention in force, there is no reading of ‘Benjamin Franklin
might not have been Benjamin Franklin’ corresponding to the reading of
‘The inventor of bifocals might not have been the inventor of bifocals’ as ‘It
might have been the case that [the inventor of bifocals was not the inventor
of bifocals]’.
Obviously, as noted, this Fregean response to Kripke’s modal argument is
strictly parallel to Dummett’s widescopist response. In fact, it is Dummett’s
response in Fregean guise. (Dummett would agree, since on his view
quantification into modal contexts, like quantification into any other non-
extensional context, is to be understood as quantification over senses
(Dummett 1973: ch. 9).)
Hence its rationale is the Lewisean one already provided for the
widescopist. Since the senses of proper names are idiolectic we will lack a
common subject matter if we use them to refer to the senses we individually
associate with them. Burge puts this point by saying that the (near) rigidity
of names and other indexical devices is ‘the offspring of a marriage of
convenience between cognitively promiscuous linguistic devices and
contexts where Sinn [sense] does not matter’ (2005: 225).
If the convention has this rationale other linguistic behaviours, not
involving the use of proper names to refer to senses, will be ruled out for
similar reasons. If I say ‘Cicero is Tully. That is necessarily true’, what I say
will be true if the thought I express with the sentence ‘Cicero is Tully’ is a
necessarily true thought because I associate the same sense with the two
names, and I refer to that thought when I utter ‘That’ in my comment. But if
you associate different senses with the names, what you say, if you use
‘That’ to refer to the thought you express with ‘Cicero is Tully’, may be
false. So if you repeat what I say but use ‘That’ in your comment to refer to
the thought you expressed with your previous sentence and I used ‘That’ in
my comment to refer to the thought I expressed with my previous sentence,
there is again no guarantee that our utterances will have the same truth-
value, since they may have different subject matters. This indicates how a
Fregean can respond to the refined Kripkean modal argument of the
Preface.
We can now return to the puzzles relating to the failure of substitutivity
salva veritate of codesignating proper names within propositional attitude
contexts, which we noted as puzzles to be addressed by the widescope
descriptivist given the Lewisean rationale for his proposed convention.
Within the present Fregean framework the problems are: to account for the
non-equivalence of ‘Ralph believes that Cicero was an orator’ and ‘Ralph
believes that Tully was an orator’ (and hence to resolve Frege’s puzzle of
identity) if ‘Cicero’ and ‘Tully’ are not here understood as standing for their
ordinary senses; to explain on the same basis how ‘Ralph believes of Tully
that he denounced Catiline’ might be true whilst ‘Ralph believes that Tully
denounced Catiline’ is false; and to account for the possible truth of
ascriptions of propositional attitudes using empty names in the specification
of the content clauses.
There is intuitively a difference between George IV’s believing of Scott
that he is Scottish and his believing that Scott is Scottish, just as there is
between his believing of the author of Waverley that he is Scottish and his
believing that the author of Waverley is Scottish. ‘George IV believed that
Scott was Scottish’ (like ‘George IV believed that the author of Waverley
was Scottish’) is ambiguous. It can mean (a) ‘for some α, α is an (identity-
revealing) aspect of Scott and George IV believed that [α was Scottish]’ (in
which sense it is equivalent to ‘George IV believed of Scott that he was
Scottish’) or (b) ‘for some α, α corresponds to the sense of “Scott” in my
idiolect and George IV believed that [α was Scottish]’. Here
correspondence is that relation between senses which is required if an
ascriber is to be correct in his ascription, via the use of a proper name, of a
propositional attitude to an ascribee. Correspondence is not identity. If two
people use a name to refer to the same object, one deferentially to the other,
the senses they associate with the name are different, but correspondent.
Kripke’s Pierre (1979), before his kidnap, uses ‘Londres’ in a sense which
is different from, but correspondent with, the sense I associate with
‘London’ (as does any Frenchman), which is why we must say, on the basis
of his repeated assent to the French sentence, ‘Londres est jolie’, that he
believes that London is pretty. Correspondence does, however, require
identity of reference (though both senses may have no reference as in the
case of ‘Santa Claus’ and ‘Père Nöel’). But identity of reference is not
sufficient. It may be that nothing more general can be said and that what
correspondence requires is contextually determined (see for further
discussion Noonan 1979 and 1981; there are other similar discussions in the
literature, see in particular Chalmers 2011).
The solution to the puzzle of failure of substitutivity salva veritate within
propositional attitude ascriptions of codesignating proper names
(equivalently, of Frege’s puzzle of informative identity statements) is now
evident.
… believes that Phosphorus …

and
… believes that Hesperus …

differ because the first has the truth-condition:

for some α, α corresponds to the sense I express with ‘Phosphorus’ and … believes that [α …]

whilst the second has the truth-condition:

for some α, α corresponds to the sense I express with ‘Hesperus’ and … believes that [α …]

Evidently, also, on the account, ‘George IV believes of Scott that he is

Scottish’ might be true whilst ‘George IV believes that Scott is Scottish’ is
false, and ascriptions of propositional attitudes using empty proper names
may also be true.
What of sentences containing propositional attitude verbs within modal
constructions? ‘George IV might have believed that Scott was Polish’ is
ambiguous between ‘it might have been that for some α, α is an (identity-
revealing) aspect of Scott and George IV believed that [α was Polish]’
(which says of Scott and George IV that the latter might have believed of
the former that he was Polish), and ‘for some β, β is the sense of “Scott” in
my idiolect and it might have been that for some α, α is an aspect of Scott
corresponding to β and George IV believed that [α was Polish]’. Hence a
possible world in which George IV believes that Scott is Polish is one in
which he at least has a belief concerning Scott himself, i.e. ‘our’ Scott, that
he is Polish. That this must be so is one of the lessons of Burge’s work
(1979b). Burge imagines a counterfactual circumstance in which Oscar, a
medically ignorant patient, who does not realize that arthritis is a disease of
the joints, and uses the term deferentially to the medical community, asserts
‘I have arthritis in my thigh’ but nonetheless is not correctly described as
believing that he has arthritis in his thigh since the medical community to
whom he defers in fact uses ‘arthritis’ (unlike the actual medical
community) as a name of another disease (which can, perhaps, affect the
thigh). By contrast counterfactual Oscar’s actual twin, actual Oscar, who is
equally medically uninformed, is correctly described as believing that he
has arthritis in his thigh, because in the actual community to which he
defers ‘arthritis’ is used by the medical experts as a name of arthritis.
It seems to me perfectly plausible for someone working within a Fregean
framework who wishes to accommodate Kripkean intuitions of rigidity in
the way described to respond in this way to the puzzles indicated
concerning the role of proper names within the context of verbs of
propositional attitude. The basic thought is that the difference between
someone who only believes of a person N that he is so-and-so and someone
who believes that N is so-and-so is a difference between someone who
could not, and someone who could, express his belief using the name ‘N’ or
some translation, and this is a thought that is independently tempting. The
same is true of the thought that ‘Ralph believes that Tully …’ and ‘Ralph
believes that Cicero …’ differ by mentioning different names and ascribing
different linguistic dispositions to Ralph.
Finally, it is perhaps a point in favour of the Fregean viewpoint that from
it we can attempt to explain Kripke’s puzzling intuition (1980: 60) that the
definition of pi as the ratio of the circumference of a circle to its diameter is
not an abbreviative, but a merely reference-fixing, definition, even though
both definiens and definiendum are strongly rigid designators. Since pi is a
proper name it conventionally preserves its customary reference where
descriptions and abbreviations of descriptions have their indirect referents.
It is this which distinguishes pi from the description. So, as Kripke says,
there are two different types of rigidity. The rigidity of proper names, de
jure rigidity, is a conventional matter, whilst the de facto rigidity of rigid
definite descriptions is explained by their picking out their referents by
essential properties. What Kripke calls a reference-fixing definition is
merely one in which the term introduced is a rigid designator of the first
kind, while the term by means of which it is introduced is at most a rigid
designator of the second kind.
The third popular response to Kripke’s modal argument is an appeal to
inconstant rigidified descriptions. As noted, one of the earliest proponents
of this response was John Searle (1983), in his vigorous defence of
descriptivism, understood, as he emphasized, as an internalist account (in
the sense defined by Putnam) of reference. Searle writes (1983: 258):
any definite description at all can be treated as a rigid designator by indexing it to the actual world. I
can, by simple fiat, decide to use the expression ‘the inventor of bifocals’ in such a way that it refers
to the actual person who invented bifocals and continues to refer to that very person in any possible
world, even in a possible world in which he did not invent bifocals. Such a use of the definite
description will always take wide scope or will be in a sense scopeless in a way that is characteristic
of proper names.

Searle refers to earlier occurrences of the idea (though not in writers

wishing to defend descriptivism) in Kaplan (1970) and Plantinga (1978).
Subsequently the proposal has been extensively defended by Jackson
(1998) and is associated with the two-dimensional programme of which
Jackson, along with Chalmers (1996) are distinguished defenders (others
endorsing it include Stanley (1997) and Pettitt (2004)). The two-
dimensional programme derives partly from the work of Gareth Evans
(1979) via Davies and Humberstone (1980), but these authors are not
proponents of rigidified descriptivism for proper names, though Davies and
Humberstone tentatively suggest that it might be correct for natural kind
terms. Searle thus deserves recognition as the first unambiguous proponent
of rigidified descriptivism as a response to Kripke’s modal argument. The
proposal is very obvious. ‘The inventor of bifocals’ is an accidental
designator, but ‘the actual inventor of bifocals’ is not. Hence the thesis that
proper names abbreviate such descriptions is invulnerable to the modal
argument and thesis (6) of Kripke’s statement of the cluster theory is not
refuted since the non-modal statement ‘the actual inventor of bifocals is the
inventor of bifocals (if just one person invented bifocals)’ expresses a
necessary truth even though it is unambiguously true that it might have
been that the actual inventor of bifocals (existed but) did not invent
bifocals. Nevertheless, Kripke never considers an appeal to rigidified
descriptions as a way of responding to his modal argument (though he notes
their possibility). So what reply is there to Searle and Jackson?
‘The actual inventor of bifocals’ is an inconstant rigid designator, so it
could have been used with the same conventional significance to speak of
someone other than Benjamin Franklin. Now suppose Benjamin Franklin
had died moments after his birth and never been christened, but the midwife
had substituted, unbeknownst to anyone, another baby, subsequently
christened ‘Benjamin’, who had gone on to have exactly the life Benjamin
Franklin actually had. If this had been so, we would now have been using
‘Benjamin Franklin’ to refer to someone other than Benjamin Franklin.
Would we therefore have been using the name with a different meaning? It
would be mere stipulation to say so. It is thus no argument against the
rigidified descriptivist that his account has the consequence that if this had
been the case the name would have had the same meaning in our mouths
that it actually has.
A recent much-discussed argument that names cannot abbreviate
rigidified descriptions, put forward by Scott Soames (1998: 16), goes as
follows. Consider someone who believes that Aristotle was a philosopher. If
the world had been slightly different he might still have believed that
Aristotle was a philosopher. But if he believes that the man who was
actually the last great philosopher of antiquity was a philosopher he could
not have believed that if the world had been in any way different in any
past, present or future respects, since to believe that is to have a belief about
the actual world, the actual world state, and no one could have had a belief
about the actual world if things had been in any way different. Hence
‘Aristotle’ cannot abbreviate ‘the man who was actually the last great
philosopher of antiquity’ since otherwise the two beliefs would be the same.
In short, the idea is that the beliefs we express with sentences containing
proper names, such as ‘Aristotle was a philosopher’, are modally robust. I
could have had the same beliefs if the world been slightly different; for
example, if the Second World War had finished a second earlier than it in
fact did, or if the next female Prime Minister of Great Britain were going to
be someone other than the woman it will in fact be. But the beliefs I express
with rigidified descriptions are modally fragile. The belief I have that the
man who was actually the last great philosopher of antiquity was a
philosopher is a belief about the actual world. If things had been, were now
or were going to be, even slightly different I would not have had that belief.
To gain perspective on this argument it is useful to compare it with its
temporal analogue. Consider the description ‘the place where Mrs Thatcher
is now’. Corresponding to Soames’s premiss that the belief actually
expressed by ‘the man who was actually the last great philosopher of
antiquity was a philosopher’ is modally fragile, is the proposition that it is,
always has been and always will be that someone who believes that it is
raining where Mrs Thatcher is now must have a belief about the present
time; hence that it is not the case that yesterday someone believed that it
was raining where Mrs Thatcher is now without having any thoughts about
the present time. But this is implausible. Suppose Mrs Thatcher is now in
Manchester, and suppose that yesterday someone believed that (as usual) it
was raining in Manchester. Then he believed that it was raining where Mrs
Thatcher is now. Of course, his belief was a belief about the place where
Mrs Thatcher is now, but it was not a belief about the present time.
Analogously, just as, even if things had been slightly different, someone
might have believed that Aristotle was a philosopher, so, even if things had
been slightly different, someone might have believed that the man who was
actually the last great philosopher of antiquity was a philosopher. In order to
do so he would have had to have a belief about the man who was actually
the last great philosopher of antiquity. But to do that he would not have had
to have any beliefs about the actual world; it would have sufficed that he
had beliefs about Aristotle. The modal fragility of the belief that the man
who actually was the last great philosopher of antiquity was a philosopher
need not be accepted. And, in general, it does not seem that the beliefs we
express using ‘actually’ as a rigidifying device are as modally fragile as
Soames thinks. ‘It could have been that everyone who actually died
survived’ is, in consequence of the rigidifying effect of ‘actually’,
unambiguous. It means that those who died are such that it could have been
that every one of them survived. Equivalently, it means that the class of
those who died is such that it could have been that every member of it
survived. (This meaning cannot be expressed in the language of quantified
first-order modal logic without a rigidifying ‘Actually’ operator, because
this language can only express the non-equivalent thoughts that everyone
who died could have survived [perhaps by killing and eating the others] – in
symbols 8x (Dx ! à S x), and that it might have been the case that the
following situation obtained: everyone, if he died, survived – à x (Dx
disaster,8 ! Sx) (Humberstone 1982).) If, referring to some I say ‘It could
have been that everyone who actually died survived’ (if only, say, the rescue
helicopter had arrived earlier), it does not seem plausible to say that the
belief I thus express is one that I would have failed to have had if things had
been or were going to be slightly different in wholly irrelevant respects, e.g.
if the Second World War had finished half a second earlier. But, if so, the
same seems to be true if we consider instead a rigidified singular
description. ‘It could have been that the man who actually died survived’ is
unambiguously equivalent to: ‘the class of those who alone died is such that
it could have been that every member of it survived’ (even though we can
this time express the thought in the language of quantified first-order modal
logic using a widescope quantifier). Again, then, it seems implausible to say
that the belief I thus express is modally fragile in the way Soames’s
argument against rigidified descriptivism requires. This implies that the
actual world should not be thought of as an aspect of context, like speaker,
place or time, for difference in context makes a difference to what is said
and believed (Davies 2004: 90–1).
The final response to Kripke’s modal argument listed appeals to the
distinction between indicative descriptions, i.e. descriptions formed using
indicative verb forms, and what I shall call subjunctive descriptions,
descriptions formed using non-indicative verb forms used in subjunctive
conditionals. The distinction is between ‘the person who is richer than
anyone else in the world’ or ‘the book Mary is reading’ and ‘the person
who would have been richer than anyone else in the world’ or ‘the book
Mary would have been reading’. The response to Kripke’s modal argument
is that indicative descriptions are indeed always rigid designators and so do
not contrast with proper names.
The distinction is noted by John Burgess (2006: 175). He comments:
even when discussing a counterfactual situation in which Bill Gates has given all his wealth to Ivana
Trump, it is not unambiguously the case that when we use the description ‘the richest person in the
world’ we must be referring to her and not to him. For the description contains an implicit verb, made
explicit in ‘the one person to be richer than anyone else in the world’. And this verb is subject to
inflection for grammatical mood –‘to be’ may become the indicative ‘is’ or the conditional ‘would
have been’– in a way that creates a flexibility of reference. Thus ‘If Bill Gates had given all his
wealth to Ivana, the richest person in the world would have been female’ is ambiguous between the
truth ‘If Bill had given all his wealth to Ivana, the one person who would have been richer than
anyone else in the world … would have been female’ and the falsehood ‘If Bill had given all his
wealth to Ivana, the one person who is richer than anyone else in the world would have been female’.

Because the latter is unambiguously false ‘the one person who is richer than
anyone else in the world’ designates rigidly. The distinction between the
two types of description can be confirmed by noting the difference between
the questions ‘If Bill had given half his money to Ivana, who would have
been the person who would have had more money than anyone else in the
world?’ and ‘If Bill had given half his money to Ivana, who would have
been the person who has more money than anyone else in the world?’ The
answer to the first question may be ‘J. K. Rowling’ (suppose Bill has 100
billion dollars, J. K. 80, Ivana 20, and no one else is close). The answer to
the second question is, trivially, ‘Bill’.
So it appears that present-tense indicative descriptions in English, like the
ones given, are rigid designators by Kripke’s tests. The same is true of all
future-tense indicative descriptions, e.g. ‘The person who will give the John
Locke Lectures in 2150’, ‘the book Mary will be reading next month’. Past-
tense descriptions, e.g. ‘the book Mary was reading last month’, are
ambiguous: ‘If Mary had joined a different book club, the book Mary was
reading last month would not have been the book Mary was reading last
month’, has a true interpretation because the past tense is ambiguous
between an indicative and subjunctive reading. This also accounts for the
ambiguity of ‘If Mary had been sicker on Monday than Jane was on
Tuesday …’ (contrast the unambiguous ‘If Mary had been sicker on
Monday than Jane is today …’ which unambiguously makes a cross-
comparison between how things would have been in counterfactual
circumstances and how things are). Past-tense descriptions are rigid on the
indicative reading. In the case of the rigid present-tense description ‘the
book Mary is reading now’ change to the passive voice and omission of the
tensed verb introduce flexibility: ‘If she had changed book clubs the book
now being read by Mary would not have been the book now being read by
Mary’ has a true interpretation (note, then, that the temporal indexical does
not impose modal rigidity and can also occur in the subjunctive description,
‘the book Mary would have been reading now’). Insertion of the rigidifying
adverb, ‘actually’, to yield ‘the book actually being read by Mary now’
brings back rigidity. The only non-rigid descriptions, then, are ones, like
‘Mary’s current husband’, ‘the richest person in the world’ and ‘the
inventor of bifocals’, not containing tensed indicative verbs.
If this is accepted what is to be the response to the refined Preface
version of the modal argument in Naming and Necessity already
encountered? Here the modal argument goes as follows, adapted to
Burgess’s example. Suppose I fix the reference of a name ‘Richie’ by the
description ‘The person who has more money than anyone else in the
world’. Then the simple, non-modal, sentence ‘Richie is a female’
unambiguously expresses something that could not have been true. On the
other hand the simple, non-modal sentence ‘the person who has more
money than anyone else in the world is a female’ does not. Hence the name
and description differ in a way that the description theorist cannot
accommodate.
But why think that ‘Richie is female’ unambiguously expresses
something that could not have been true? Presumably because Richie is Bill
Gates and so it couldn’t have been that Richie was female (assuming one’s
gender is an essential property). But then it couldn’t have been that the
person who has more money than anyone else in the world was female
(though it could have been that the richest person in the world/the person
who had more money than anyone else was female, as it would have been if
Bill had given all his money to Ivana). And both ‘It could have been that
Richie is a female’ and ‘It could have been that the person who has more
money than anyone else in the world is a female’ are ungrammatical
(contrast the grammatical ‘It could be that Richie/the person who has more
money than anyone else in the world is a female’). But, a defender of
Kripke’s argument may rejoin, is it not the case that with respect to certain
counterfactual circumstances, e.g. if Bill Gates had given all his money to
Ivana, what ‘Richie is a pauper’ says would have been true, whereas with
respect to those circumstances, what ‘the person who has more money than
anyone else in the world is a pauper’ says would not have been true? But
why say that with respect to those circumstances what ‘Richie is a pauper’
says would have been true? Because it is unambiguously true that if those
circumstances had obtained Richie would have been a pauper (recall here
Dummett’s modal involvement principle). But equally, it is unambiguously
true that if those circumstances had obtained the person who has more
money than anyone else in the world would have been a pauper, though, of
course, it is not the case that if those circumstances had obtained the person
who would have had more money than anyone else in the world would have
been a pauper. When asked whether what ‘Richie is a pauper’ says might
have been true I answer the question by asking whether it might have been
true that Richie was a pauper, i.e. whether there is a way things could have
been such that if things had been that way it would have been that Richie
was a pauper. When asked whether what ‘the person who has more money
than anyone else in the world is pauper’ says might have been true I answer
the question by asking whether it might have been true that the person who
has more money than anyone else in the world was a pauper, i.e. whether
there is a way things could have been such that if things had been that way
it would have been that the person who has more money than anyone else in
the world was a pauper. In both cases I must answer affirmatively, though,
of course, I acknowledge that there are no ways things could have been
such that if things had been those ways it would have been the case that the
person who had more money than anyone else in the world was a pauper.
Of course, this final response only speaks to Kripke’s challenge to the
descriptivist to come up with descriptions in English which may plausibly
be thought of as equivalent to proper names, given that such names function
as rigid designators. Languages can be imagined, and, in fact, actually exist,
in which there is no distinction between indicative and subjunctive forms
and no descriptions function as required by this response. But it is no part of
the descriptivist position to deny the possibility of such languages. This is
obviously so since the descriptivist can acknowledge the possibility of a
language in which proper names occur but there are no definite descriptions
at all (what Kripke (1977) calls a ‘strong Russell language’); indeed, a
descriptivist who follows Russell, for whom definite descriptions are a mere
abbreviatory convenience, must insist on the possibility of such languages,
expressively equivalent to languages in which definite descriptions can be
formed. Thus the possibility of such languages is irrelevant to the issue
between Kripke and the descriptivists. If definite descriptions formed in
English using indicative verb forms are rigid designators it suffices to refute
the modal argument since the second premiss of that argument is that no
descriptions (at least of contingently existent objects) that can be at all
plausibly thought of as synonyms of proper names are rigid designators.
But indicative descriptions, like ‘the book Mary is reading’, form an
extremely large class of commonplace English expressions and are
plausibly available to users of proper names to define them.
I now conclude this section. The simple and straightforward modal
argument that dominates the early part of Naming and Necessity is hotly
contested. At least four ways of opposing it are in the literature. There is
much more to be said about each of these. However, Kripke has, of course,
a second, more ambitious line of argument against the descriptivist’s cluster
theory, beginning in the second lecture, which is designed to show that even
as a theory of reference-fixing it fails (and the debate over this argument
survives translation into a strong Russell language). If this succeeds the
modal argument is redundant (or at best necessary to sweep up ‘stragglers’,
atypical proper names like ‘Jack the Ripper’ and ‘St Anne’).
The Arguments Against the Cluster Theory Qua
Theory of Reference-Fixing

It is important to appreciate how ambitious the argument of the second

lecture is, how modest the form of the cluster theory it targets. The target is
constituted solely by theses (1)–(5) of Kripke’s formulation, assumed to be
satisfied subject to the circularity constraint (C). This encapsulates the basic
thought that we cannot refer to something unless we have the means,
however indirect, of picking it out and distinguishing it from other
candidates. The means need not be shared by others; it need not be in purely
qualitative or general terms; it need not identify the thing by any essential
properties; it need not be by a specific description as opposed to a vague
cluster; but it must exist if one is able to refer to a particular thing. Rightly,
Kripke perceived that if his attack on this conception of reference was
successful it was revolutionary. We need to assess how successful it really
was.
The general form Kripke’s attack takes is the presentation of examples
intended to illustrate that when the circularity constraint (C) is taken into
account, successful reference can be made even in the absence of
identifying descriptions and even in the presence of wholly mistaken
beliefs, which identify no one or an object distinct from the actual
reference.
Often philosophical counterexamples to a theory are convoluted and
complicated cases that the proponent of the theory has not thought of.
Kripke’s attack on the cluster theory is quite different. He argues that in a
whole collection of central cases of reference the theses of the cluster
theory are mistaken. The counterexamples lie right under our noses. His
arguments against it are often subdivided by commentators into semantic
arguments and epistemological arguments. The semantic arguments are
aimed at theses (2), (3) and (4); the epistemological arguments at thesis (5)
and the theses subsidiary to theses (3) and (4) Kripke mentions. The
semantic arguments are designed to refute the description theory as a theory
of reference-fixing; the epistemological arguments are designed to refute
the idea that the truth of the description theory is known a priori by users of
proper names.
Theses (2), (3) and (4), he claims, turn out to have a large number of
counterexamples. Moreover, even when they are true, thesis (5) is false; the
truth of theses (2)–(4) is an empirical accident that the speaker hardly
knows a priori. What determines reference in these cases is something other
than what the cluster theorist thinks and it is just coincidence that this is
what theses (2)–(4) would determine as the reference. Only in a rare class
of cases are theses (2)–(5) all true. The examples he singles out as ones in
which this is so are cases of initial baptism.
The point that Kripke is not claiming that theses (1)–(5) are never true is
important. The picture of meaning theses (1)–(5) give, he says, is this. I
want to name an object. I think of some way of describing it uniquely. I go
through a mental baptism ceremony: ‘Cicero’ is to denote the man who
denounced Catiline. In such a case all theses (1)–(5) are correct, and Kripke
allows that such cases are possible. Indeed, it is implicit in his developed
account that all cases of reference with proper names start out in such a
situation. So he thinks the description theory necessarily has application,
though primarily to cases of initial baptism. He does not, therefore, require
an initial causal link to the reference (the initial baptism may be in purely
qualitative terms and, even if indexicals are used, Kripke says that we may
think of this as a limiting case of baptism by description; he nowhere
suggests even in such a case that a causal link to the reference is needed). In
fact, the role of a causal link in his account is even more limited; it is not
said even to be essential to the transmission of a name. Nevertheless, he
argues that in most cases the description theory is incorrect.
He starts by arguing against thesis (2), which he emphasizes might seem
in some a priori way to be something that has to be true ‘because if you
don’t think that the properties you have in mind pick out anyone uniquely
… then how can you say which one them you are talking about’ (1980: 80).
Nevertheless, he argues, thesis (2) is false.
His argument against it hangs on the fact that most people use names of
famous individuals, like Cicero, on the basis of really scanty information.
Defenders of the cluster theory, when asked what descriptions fix the
reference of a name, typically offer descriptions of famous deeds. But, in
fact, most people do not have such detailed information. They think of
Cicero just as a famous orator. Or to take Kripke’s other example, they
think of Feynman just as a famous physicist, without being able to give any
details that distinguish him from Gellman.
Moreover, Kripke argues, often, when thesis (2) of the cluster theory
seems to be satisfied, it emerges that this is not so when the circularity
constraint (C) is recalled. Any student of philosophy, having read Quine,
will know that Cicero denounced Catiline – so it seems at first sight that
such a student has an identifying description of Cicero available, namely,
the man who denounced Catiline. But if the student has got his classical
knowledge from Quine it is probable that all he will know about Catiline is
that he was denounced by Cicero. So constraint (C) is not satisfied.
Consequently Kripke concludes that there are many cases in which
reference by way of proper names is achieved in a way incompatible with
thesis (2) and many cases in which this is so that at first sight might appear
to be in conformity with thesis (2) because of failure to attend to the need to
avoid circularity.
Kripke illustrates these points with another example: Albert Einstein.
Everyone knows who Einstein is: the man who discovered the theory of
relativity. And everyone knows what the theory of relativity is: Einstein’s
theory. Clearly the knowledge stated here does not satisfy the constraints of
the cluster theory.
These arguments are directed against thesis (2) of the cluster theory. The
claim is that many actual mundane situations are counterexamples to thesis
(2). Since the cluster theory is a philosophical analysis of what it is for a
name to have reference, and hence if true is necessarily true,
counterexamples need merely be possible, but Kripke’s claim is that they
are not merely possible, but actual and everyday.
Kripke brings arguments of exactly the same character against thesis (3),
the thesis that if most of the js, suitably weighted, are satisfied by a unique
object y, y is the reference of ‘X’ (for the speaker A). In fact, he argues that
there are cases in which a speaker associates a uniquely identifying
description with a name but does not refer to what the description denotes.
Such cases are uncommon because the more common cases are ones in
which speakers do not associate any identifying descriptions with names,
but they occur. One hypothetical example Kripke gives is that of Gödel.
Suppose someone says that Gödel proved the incompleteness of arithmetic
and is able to specify exactly what this means (he is not in the position of
someone who knows only that the theory of relativity is Einstein’s theory).
Then he has a definite description capable of fixing the reference of ‘Gödel’
in accordance with thesis (3). But, Kripke argues, it need not be that his
reference is determined by this description. For suppose, counter-factually,
that Gödel was not the author of the incompleteness theorems
(unbeknownst to anyone). Then the reference of the name ‘Gödel’, as used
by the person who can offer no other description, but is still a member of
our linguistic community, will still be Gödel, the famous public figure,
hence it will not be the unknown person (Schmidt, Kripke calls him) who,
according to the supposition, did discover incompleteness of arithmetic, as
thesis (3) requires. The basic point Kripke is insisting on is that we often
use names on the basis of considerable misinformation about their bearers,
but such misinformation does not make it impossible for us to use these
names to refer to their bearers. The Gödel counterexample to thesis (3) is
hypothetical, but others are actual. Kripke suggests the case of the name
‘Christopher Columbus’, as used by a child who has been told, or has
retained from what he has been told, only that Christopher Columbus
discovered America. According to thesis (3) of the cluster theory such a
child will, of necessity, be referring to the discoverer of America, that is,
some Norseman or other, whenever he uses the name. But, Kripke says, this
simply need not be so. Another real-life example he gives is that of Peano,
who did not in fact discover the axioms to which he gave his name, the
Peano axioms. Their discoverer was Dedekind. But many people do not
know this and if they do not know much about the history of mathematics
they may identify Peano solely by the description ‘the author of the Peano
axioms’. If thesis (3) of the cluster theory is correct, when they use the
name ‘Peano’ they refer to Dedekind. But Kripke says they simply do not.
Once again, Kripke’s claim is that there are in fact a host of familiar
everyday counterexamples to thesis (3) of the cluster theory. The cases are
not merely hypothetical (the Gödel case is atypical in this respect and
inessential to Kripke’s argument, although it is the most celebrated of his
cases), so they do not show merely that the cluster of descriptions theory of
reference for proper names fails as a philosophical theory because it is not
necessarily true; they show that it is actually false.
In the context of his discussion of his counterexamples to thesis (3)
Kripke briefly discusses Donnellan’s celebrated distinction between
referential and attributive uses of descriptions, which he returns to later in
his paper on speaker’s reference and semantic reference (1977). One of
Donnellan’s examples of the distinction is the following. We see Jones in
the dock, accused of the murder of Smith. We are both convinced that he is
the murderer. Case one: on the basis of the bizarre nature of the murder I
say, ‘the murderer of Smith is insane’. Case two: on the basis of the way
Jones is acting in the dock I say, ‘the murderer of Smith is insane’. The first
is an attributive use of the description – I intend to speak of the murderer of
Smith whoever he is. The second is a referential use. I intend to speak of the
man in the dock and believe that I can do so using the description ‘the
murderer of Smith’ since I believe that he is the murderer. A second famous
example of a referential use of a description is the following. We are at a
party. I say, gesturing at a man in the corner, ‘The man over there drinking
champagne is very happy’. In fact, the man in question is drinking sparkling
water, but I am using the description merely as a way of identifying the man
in the corner for my audience. When used referentially in modal contexts
descriptions cannot be heard as having narrow scope. Hence the proposal
that proper names be equated with referentially used descriptions is a
possible response to Kripke’s modal argument which can be added to our
list. But Kripke does not consider this proposal in the context of his
discussion of the modal argument, but only as a way of defending thesis (3)
of the cluster theory. The proposal he considers (1980: 87) is that the use of
‘Gödel’ can be equated with a referential use of a description. So although
‘Gödel’ abbreviates ‘the author of the incompleteness theorem’ we are
talking about the public figure even if he turns out not to have proved the
theorem (just as, when, in the case of Smith’s murderer, I say on the basis of
Jones’s bizarre behaviour in the dock, ‘The murderer of Smith is insane’, I
am talking about Jones even if I am wrong and he is innocent). He makes a
conclusive reply:
a referential definite description is typically withdrawn when the speaker realizes it does not apply to
its object. If a Gödelian fraud were exposed, Gödel would no longer be called ‘the author of the
incompleteness theorem’, but he would still be called ‘Gödel’. The name, therefore, does not
abbreviate the description.

In ‘Speaker’s Reference and Semantic Reference’ Kripke adds: ‘[i]f the

hypothetical fraud were discovered [w]e would withdraw any previous
assertions using the description to refer to Gödel (unless they were also true
of Schmidt). We would not similarly withdraw the name “Gödel”’ (1977:
12).
Kripke explains the phenomenon to which Donnellan has drawn attention
by appeal to a distinction between speaker’s reference and semantic
reference (1977: 14–15):
the semantic referent of a designator … is given by a general intention … to refer to a certain object
… . The speaker’s referent is given by a specific intention, on a given occasion, to refer to a certain
object. If the speaker believes that the object he wants to talk about … fulfils the conditions for being
the semantic referent, then he believes that there is no clash between his general intentions and his
specific intentions.

Attributive uses of singular terms, Kripke suggests, are ones in which,

simply, the speaker’s specific intention is to refer to the semantic referent.
Referential uses are ones in which the speaker has a complex intention: ‘he
has a specific intention, which is distinct from his general intention, but
which he believes, as a matter of fact, to determine the same object as the
one determined by his general intention’ (1977: 15). Given this account, he
thinks, proper names cannot be regarded as abbreviating referentially used
definite descriptions since, like descriptions, they can be used in both ways.
A possible response by the descriptivist is that the use of a proper name is
always a complex use. When a speaker uses a name ‘N’ his use of it is
equivalent to a complex use of the description ‘the thing called “N”’, since
he has both the general intention to refer to the (salient) thing called ‘N’ and
another specific, reference-determining, intention.
After discussing thesis (3) Kripke turns more briefly to thesis (4), the
thesis that if the cluster identifies no unique object, the name does not refer.
We can easily see, he says, that counterexamples are available. To begin
with, the Cicero and Feynman examples establish that no unique object may
be located. Furthermore, it is clear that if someone can use a name to refer
to someone on the basis of a completely false set of beliefs about that
object, as the Gödel, Columbus, and Peano examples are intended to show,
then he can do so even when the things he falsely believes to be true of the
object are true of nothing. It is, perhaps, rather doubtful whether anyone
deserves to be called ‘the discoverer of America’, but this cannot affect the
question whether the young child who believes that Columbus discovered
America, and nothing else about Columbus, refers to anyone when he uses
the name ‘Columbus’. Kripke refers to the example of Jonah. Biblical
scholars believe that Jonah really existed, but everything reported about him
is false. The name ‘Jonah’ has a reference, inconsistently with thesis (4).
Finally, we come to thesis (5): that the speaker knows a priori that in his
idiolect the statement ‘if X exists then X has most of the js’ is true. But, of
course, if there are actual counterexamples to theses (3) and (4) this cannot
be so; and in fact even the possibility of such counterexamples is enough to
show that it cannot be so, since if counterexamples are possible, the speaker
can hardly know a priori that his case is not one of them.
These are Kripke’s arguments against the cluster of descriptions theory
considered in its modest role as a theory of reference. There are
immediately plausible responses, to which Kripke replies.
 The first response is that the speakers Kripke refers to have more
knowledge than he credits them with. Speakers who really do lack an
identifying description of an individual, as he supposes in the cases he
describes, do in fact fail to refer or refer to the item, if any, which their
identifying description in fact picks out. For example, someone who really
does know nothing more about Cicero than that he was a Roman orator is
not a counterexample to thesis (2) since he does not refer to anyone.
Similarly, someone who really does associate no description with the name
‘Richard Feynman’ other than ‘a famous physicist’ refers to no one. Again,
if someone really does associate only the description ‘the author of the
incompleteness theorem’ with the name ‘Gödel’ then, on the supposition
that Kripke’s fiction is true, when he uses the name he must be referring to
the unknown author, rather than the famous public figure. And someone
who really does associate with the name ‘Peano’ merely the description ‘the
author of the Peano axioms’ must, when he uses the name, be referring to
Dedekind.
Kripke cannot deny that it is possible for someone just to use the name
‘Cicero’ with the overriding intention of referring to ‘the famous Roman
orator’ (under the mistaken apprehension that there was only one) or ‘the
man who denounced Catiline’, since he explicitly admits the possibility of
cases in which the theses of the cluster theory are not only true but even
give a correct picture of how reference is determined. He suggests that
examples may be the use of ‘Hesperus’ by someone who announces ‘I shall
call the heavenly body over there “Hesperus”’, the use of ‘Neptune’ by
Leverrier and the use of ‘Jack the Ripper’ by the police investigating the
Whitechapel murders. But his claim is that the actual cases he suggests as
counterexamples to the cluster theory are not such cases. They are cases in
which reference is achieved despite the identifying descriptions required by
the cluster theory being absent.
But the reply now is that this is not so. In these cases the speaker is
ignorant of, or badly misinformed about, the history of the bearer of the
name. But such cases are not ones in which he is capable of associating no
definite description, or only an incorrect definite description, with the name.
For the speaker can and does in such cases use the name deferentially. The
classically uneducated philosopher defers to Quine in his use of ‘Cicero’
and so associates the description ‘the man Quine refers to by the name
“Cicero”’ with the name. In almost all real cases no one uses names with
the limited knowledge Kripke suggests, and when they do, the situations are
the unusual ones mentioned by Kripke himself, like ‘Neptune’ and ‘Jack the
Ripper’. So the challenge to Kripke is twofold. First, none of the actual
cases he gives are counterexamples to the cluster theory. Second, in the
fictitious case of Gödel, the story is interpretable in two ways: if we
suppose that ‘Gödel’ is used by someone who intends to refer to the author
of the incompleteness theorem, whoever he is, then the reference is
Schmidt; if, on the other hand, we suppose that the user of the name is
aware that there is a famous public figure who has that name and intends to
refer to him, then the case is again no counterexample to the cluster theory
since he has a description which denotes Gödel.
Of course, Kripke is vividly aware of the possibility of a deferential use
of names and quotes Strawson precisely on the point. But he thinks that
appeal to deference is an insufficient reply to his arguments. He makes
three main points. First, he points out that such a description as ‘the man to
whom the incompleteness of arithmetic is commonly attributed’ will
succeed in securing a reference only if it is not the only description used for
that purpose in the community. No one will be referring to anyone by the
name ‘Gödel’ if everyone declares that he will mean by the name ‘the man
to whom the incompleteness of arithmetic is commonly attributed’. Second,
Kripke points out that if I defer to Jones, in the sense of associating such a
description as ‘the man Jones thinks proved the incompleteness of
arithmetic’ with the name ‘Gödel’, I cannot be sure that there is no circle
involved. I therefore cannot use such a putatively identifying description
with any confidence. Kripke’s third point is that if you use such descriptions
as ‘the man Jones called “Gödel”’ in order to determine the reference of a
name, you need to remember from whom you got the name, but often you
do not. This shows, he says, that the view he advocates has consequences
which can actually diverge from those suggested by Strawson’s footnote.
Suppose that the speaker has heard the name ‘Cicero’ from Smith and
others, who use the name to refer to a famous Roman orator. He later
thinks, however, that he picked it up from Jones who, unbeknownst to him,
uses ‘Cicero’ as a name of a notorious German spy and has never heard of
any orators in the ancient world. Then, according to Strawson’s paradigm,
the speaker must determine his reference by the resolution: ‘I shall use
“Cicero” to refer to the man whom Jones calls Cicero’, while on the view
that Kripke advocates the reference will be the orator despite the speaker’s
false impression about where he picked up the name. Kripke takes this to
show that his own view is preferable, because it conforms to the principle
that it is the actual chain of communication, and not what the speaker thinks
is the chain of communication, which is relevant.
This, in essence, is Kripke’s response to the proposal that the speaker
may use a name deferentially to others, referring only because he himself
has identifying knowledge of his reference in the form of descriptions
which relate to it via the knowledge and activities of others. It is unclear
whether it succeeds.
While there is a great deal more to be said, the key point is that the
constraint against circularity to which a description theory of reference
must conform, Kripke’s constraint (C), is much weaker than he presents it
as being in the course of his arguments. Other kinds of descriptions than
those Kripke discusses are available to the description theorist. What
Kripke has to show is that one can refer to a thing by name even when one
knows of no description which determines the reference of the name. But
none of his examples show this, as argued above. And he never considers a
description like ‘the person referred to as “Feynman” by the person from
whom I got this use of the name’ (cf. Searle 183: 253 ‘Feynman’
substituted for ‘Thales’) or ‘the person called “Feynman” such that my
familiarity with the name as a name of a physicist derives from having
encountered references made to him by that name’ (cf. Schiffer 1977: 68,
where ‘Feynman’ is substituted for ‘Schoenberg’). (The closest he comes to
doing so is in the Addenda where he considers the possibility of
determining reference by the stipulation ‘let “Glumph” denote the man
called “Glumph” by the people from whom I got it [whoever they are],
provided that my present determination of reference satisfies the conditions
sketched in Naming and Necessity and whatever other conditions need to be
satisfied’.) But his criticism focuses on the material following the comma in
the stipulation (e.g. ‘the resulting theory … would only occur to those
speakers who have mastered a complex theory of reference’ (1980: 162)
and he never considers a stipulation of the simple kind suggested by Searle
‘let “Glumph” denote the man called “Glumph” by the people from whom I
got it’.)
Now suppose Jones uses ‘Feynman’ to refer to Feynman, knowing little
more than that he is a famous physicist. This will not be all Jones knows.
He will also know that the description just given of Feynman in the last
paragraph is satisfied, that is, that there is just one person named ‘Feynman’
such that his familiarity with the name ‘Feynman’ as a name of a physicist
derives from having encountered references to that person by that name.
Suppose Jones believes this to be so, but it is not: there are two physicists
named ‘Feynman’ and Jones has encountered references to both of them
under that name, without realizing that there are two. Then, evidently
enough, Jones is not referring to anyone when he uses the name ‘Feynman’;
he is merely in a muddle. But if Jones does know that there is just one such
person, he can simply identify Feynman by the description given. (Note that
this description may denote even if the description ‘the famous physicist
called “Feynman”’ does not; for example, if there are two famous physicists
called ‘Feynman’ but by chance Jones has only encountered references to
one of them. If Jones is not a physicist or member of a related profession,
however, this situation is an unlikely one; most probably if the description
‘the famous physicist called “Feynman”’ does not denote, neither will the
longer description, as used by Jones. This explains why Jones would be
embarrassed by the information that there were two physicists called
‘Feynman’, both famous. If there were only one famous physicist called
‘Feynman’, however, though several physicists who were less well-known
were also called ‘Feynman’, then more than likely Jones, as a non-scientist,
would only have encountered references to the famous one, so the longer
description, as used by him, would still denote. This explains why this piece
of information would not put Jones out of countenance.)
Again, while, as Kripke says, one does not use the name ‘Gödel’ to refer
to the unknown Schmidt just because Schmidt proved the incompleteness of
arithmetic and the only biographical detail one associates with the name
‘Gödel’ is ‘prover of the incompleteness of arithmetic’, this proves nothing
as strong as he claims. For to use ‘Gödel’ to refer to Gödel in this
circumstance, one must know that there is just one person such that (a) he is
named ‘Gödel’ and (b) one’s familiarity with the name, as the name of a
logician, derives from having encountered references to him by that name.
But if one knows this one can identify Gödel by the description ‘the person
called “Gödel” such that my familiarity with the name “Gödel”, as the name
of a logician, derived from having encountered references to him under that
name’. (Cf. Searle (1983: 251): ‘When Jones uses “Gödel” he has “the man
called ‘Gödel’ in my linguistic community or at least by those from whom I
got the name”. The reason he does not give this as an identifying
description is that he assumes that something more is wanted’.)
In general, it seems that some such egocentric, metalinguistic, causal
description is bound to be available, if nothing else is, whenever one is
capable of referring to an object using a proper name, and, of course, such
descriptions are not circular in the manner of ‘the person I (here and now)
mean by “Jones”’ – they are genuinely reference-determining, in
conformity with Kripke’s own formulation of the circularity constraint (C).
Thus contrary to the suggestion of Kripke’s discussion, the requirement that
the description theory of reference not involve the notion of reference in an
ineliminable way does not entail that no description theory can be true. In
his 1997 paper ‘Naming the Colours’ this conclusion is famously drawn by
David Lewis (though many others, in addition to those mentioned already,
for example Kroon 1987, think the same):
did not Kripke and his allies refute the description theory of reference, at least for the names of
people and places? … I disagree. What was well and truly refuted was a version of descriptivism in
which the descriptive senses were supposed to be a matter of famous deeds and other distinctive
peculiarities. A better version survives attack: causal descriptivism. The descriptive sense associated
with that name might for instance be ‘the place I have heard of under that name “Taromeo”’ or
maybe ‘the causal source of this token: Taromeo’, and for an account of the relation being invoked
here, just consult the writings of causal theorists of reference.
(1997: 339)

I think a stronger conclusion can be drawn. It is not just that Kripke’s

examples do not refute the description theory. No such examples could
refute it. There is a difficulty in principle with Kripke’s contention that the
reference of a proper name might be determined independently of the
speaker’s intentions in a way that is inconsistent with the constraint on
reference embodied in the description theory. For if a person intends the
reference of a name to be the so-and-so it must be the so-and-so, or nothing.
(Of course, he can believe the reference of a name to be the so-and-so when
it is not.) If a person really intends the reference of ‘Gödel’ when he uses it
to be the person who proved the incompleteness of arithmetic, and has no
other competing reference-determining intention, that is who it will be –
even if that is Schmidt (cf. Searle 1983: 251). As we have seen, Kripke
could not deny this, for he allows the possibility of reference being
determined in some cases by description, and such cases must be ones
where the speaker’s intentions are allowed to be dominant in determining
reference. So, when a speaker has an intention capable of determining a
specific object of which he has identifying knowledge as the reference of a
name, his reference will be determined by that intention. To find
counterexamples to the thesis that the reference of a proper name is
determined by the speaker’s intention one must, therefore, apparently
consider cases where no such unambiguous reference-determining intention
exists; where the speaker has no intention capable of determining reference
or several intentions which do not determine the same reference. But,
however the reference is determined in such a case, it must be in a manner
which, if it were described to them, the speaker and other members of his
community could be brought to acknowledge as correct, that is, as bound to
determine the correct reference on any relevant occasion of utterance. For,
as Dummett says (1974: 527):
Suppose that the causal theory of reference is correct in that it gives an accurate account of the way
in which, in problematic cases, it is generally agreed that the reference of a name is to be determined.
… Then the causal theory … merely gives an account of what senses [names] have. The alternative is
to suppose that the causal theory gives a correct account of the conditions for the name to have a
particular object as its reference, even though, in critical cases, most speakers would repudiate that
means of determining the truth-values of sentences containing the name. This … would mean that a
certain means of determining the truth-values of a sentence might be the right one although it was not
acknowledged as such by any speaker of the language. Such an idea would seem to involve the same
fallacy as ‘They are all out of step but our Willie’.

Thus, whatever the mode of reference-determination is, a speaker must be

disposed to regard his reference as determined in that way in the absence of
any specific reference-determining intention on his part, and so, if he lacks
such a specific reference-determining intention, he must intend to let his
reference be determined in that way. And so, once more, the speaker’s
intentions will be determinative of his reference. (Kripke acknowledges this
in footnote 38 of Naming and Necessity where he allows that if any ‘causal’
theory of reference is correct a causal version of the description theory must
be trivially true, but insists that no theory is true and he can only give a
picture.)
This sort of consideration has caused many writers to reject Kripke’s
arguments against the cluster theory. It can be developed further when we
turn to Kripke’s own alternative picture of what naming involves. There are
at least two other difficulties for Kripke. The first, considered at the end of
this chapter, is to explain the possibility of reference change within his own
account. The second, considered in the next chapter, is to explain those
failures of substitutivity of coreferential expressions salva veritate in the
context of propositional attitude ascriptions which the Frege–Russell view
is designed to accommodate. Of course, Kripke holds that in modal contexts
rigid designators are interchangeable salva veritate, but he does not hold
that this is so in propositional attitude contexts. Otherwise he would be
unable to recognize the possibility of necessary a posteriori truths. This
difficulty is the focus of his later paper ‘A Puzzle about Belief’.
The Historical Chain Picture

When he presents his alternative to the cluster theory, Kripke stresses the
word ‘picture’. He says that he is not attempting a theory (a set of necessary
and sufficient conditions, which he thinks is impossible), but only an
alternative way of looking at reference. Recall that the descriptivist story is
motivated in large part by the need to provide an account of how we
manage to connect symbols with the things to which they refer. The
descriptivist’s answer is: by association with properties (specified by
description). Since Kripke rejects this story he recognizes that he needs to
say something else. He does so by sketching his alternative picture:
Someone … is born; his parents call him by certain name. They talk about him to their friends. Other
people meet him. Through various sorts of talk the name is spread from link to link as if by a chain.
A speaker who is on the far end of the chain, who has heard about Richard Feynman, may be
referring to Feynman even though he can’t remember from whom he first heard of Feynman or from
whom he ever heard of Feynman. … He is then referring to Feynman even though he can’t identify
him uniquely. … a chain of communication going back to Feynman himself has been established, by
virtue of his membership in a community which passes the name from link to link, not by a ceremony
that he makes in private in his study: ‘by “Feynman” I shall mean the man who did such and such’.
(1980: 91–2)

He adds some details:

A rough statement of a theory might be the following. An initial ‘baptism’ takes place. Here the
object may be named by ostension, or the reference … may be fixed by description. When a name is
‘passed from link to link’, the receiver of the name must … intend when he learns it to use it with the
same reference as the man from whom he heard it. If I hear the name ‘Napoleon’ and decide it would
be a nice name for my pet aardvark, I do not satisfy this condition.
(1980: 96)

This divides the account of reference into two parts: the initial fixation of
reference and the maintenance and transmission of reference. The former
concerns the first use of the name by the first user of the name, the latter,
later uses of the name by the first user and the first and later uses of the
name by later users who acquire it from an earlier one. The first part of the
theory is in accord with the description theory, except that the name is
introduced as rigid.
This division between the initial fixation of reference and the
maintenance of reference is important in relation to Kripke’s claims about
the contingent a priori and the necessary a posteriori. His examples of the
contingent a priori are cases in which something is a priori for the initial
baptizer who has used a description to fix the reference. These examples
depend on the contrast between the rigidity of the name and the non-rigidity
of the reference-fixing description. His explanation of the ‘illusion of
contingency’ associated with necessary a posteriori truths, i.e. the fact that
they are knowable only a posteriori, also depends on the association of the
name with a non-rigid description in the initial baptism. In both cases the
description linked initially with the name may cease to be associated with it,
so that what is initially contingent a priori for the first user may, for
subsequent users or the same user at a later time, be merely a posteriori, and
the explanation of the a posteriori status of the necessary a posteriori truth
known initially by the baptizer and retained by him or transmitted to others
will not be available for him later or for subsequent users.
At first sight it appears that there is a difference in kind between Kripke’s
explanation of reference and that of the descriptivist. One is given in terms
of facts about linguistic communities and the other in terms of facts about
the psychology of individual language users. But the contrast is overdrawn.
Individual language users’ intentions are still crucial to the Kripkean story:
(a) at the time of the initial introduction of the name and (b) when the name
is first passed on from one user to another. On the other hand, as we have
seen, the Frege– Russell descriptivist account allows for deference and the
thought that it is essential to my reference that I am a member of a
community.
This brings us back to the thought that no counterexample of the kind
Kripke envisages to the description theory can possibly succeed. As just
noted, speaker intentions play two crucial roles in Kripke’s picture. There is
an initial baptism in which the baptizer intends the reference of a name to
be fixed by description. The name is then passed on to a second user who
must intend when he picks it up to refer to the object to which the person
from whom he is picking it up referred – he cannot decide to initiate a new
use of the name as a name for his pet aardvark, say. And the same must be
true of all subsequent users. Now when the second user picks up the name
he will not count as intending to use the name with the same reference as
those from whom he acquired it just because he does in fact intend to use it
as a name for his pet aardvark and, in fact, quite unbeknownst to him, the
people he heard using the name were already using it as a name for his pet
aardvark. Rather, it must be that the content of his intention is: to use the
name to refer to whatever was the referent in the mouths of those from
whom he acquired it. Similarly, when the baptizer introduces the name he
must have an intention whose content is: to use the name to speak of the so-
and-so (or that object which … if he fixes the reference of the name
ostensively). But then neither the second user’s first use, governed by his
deferential intention, nor the baptizer’s first use, governed by his baptismal
intention, can possibly be a counterexample to the Frege–Russell
description theory as Kripke defines it. Counterexamples can only emerge
once the second user (to focus on him for the moment) ceases to have the
intention to refer to the object to which the people from whom he acquired
the name referred with it and acquires no other reference-determining
intention in its place (he cannot acquire the intention to refer to the object to
which he most recently referred with the name, for example, or to what
others in his community now refer to by the name, or to his own or Kripke’s
pet aardvark). Similarly, uses of the name by the baptizer can only emerge
as counterexamples to the description theory once the baptizer ceases to
have the intention to refer to the thing satisfying the description which
originally fixed the reference of the name for him and acquires no other
reference-determining intention in its place, not even the intention to refer
to the object to which he previously referred using the name. Such uses can
be counterexamples only if their reference is the object standing in a certain
causal-historical relation R to them, to be described by Kripke – which
requires that an object X would be acknowledged by the name user to be
the reference if evidence were provided to him that X did indeed stand in
relation R to his utterances – but the name user has no intention to refer to
the object standing in relation R to his utterances, nor any other intention
capable of determining that object as his reference. But Kripke never
describes a case satisfying these requirements. In fact, the intelligibility of
the stories he tells depends crucially on the fact that the user of ‘Gödel’,or
whatever, is a member of a community and intends to use the name in the
same way as others do – it is only this that makes it plausible that, despite
his ignorance and mistaken beliefs, he does not refer to Schmidt but does
succeed in referring to someone when he uses the name.
One thing that Kripke explicitly allows is the possibility of reference
change. In the ‘Napoleon’ example the man who decides to use ‘Napoleon’
as a name for his pet aardvark consciously opts out of a previous referential
practice and institutes a new use. But there is also the possibility of
referential shift without any such conscious decision. Kripke notes this
himself, acknowledging the contribution of Gareth Evans:
Gareth Evans has pointed out that … cases of reference shift arise … from one real entity to another
of the same kind. …‘Madagascar’ was a native name of part of Africa; Marco Polo, erroneously
thinking that he was following native usage, applied the name to an island. Today the usage of the
name as a name for an island has become so widespread that it surely overrides any historical
connection with the native name.
(Kripke 1980: 163)

Neither Marco Polo nor any subsequent European users of the name
consciously initiated a new practice. Each intended to use the name when
he received it with the same reference as those from whom he got it.
Nevertheless, reference shift occurred. Kripke offers the following
explanation:
… a present intention to refer to a given entity … overrides the original intention to preserve
reference in the historical chain of transmission. The matter deserves extended discussion. But the
phenomenon is perhaps roughly explicable in terms of the predominantly social character of the use
of proper names … We use names to communicate with other speakers in a common language. This
character dictates that ordinarily a speaker intends to use the name the same way as it was transmitted
to him, but in the ‘Madagascar’ case this social character dictates that the present intention to refer to
an island overrides this distant link to native usage.
 (1980: 163)

Kripke here indicates his awareness of the crucial role of the present
intention to refer to the island in determining the present reference of the
name. Surely this is right. Initially Marco Polo was referring to part of the
African mainland and merely had mistaken beliefs about it – at this stage he
would have withdrawn his statement ‘Madagascar is an island’ if informed
of his error by a native speaker. But Kripke also indicates that he does not
think that such examples of reference change, and the role in explaining
them of speakers’ intentions, are problems for him or evidence in favour of
the description theory. He suggests later that the case may be understood as
one in which what is a speaker’s reference by an earlier user is
unintentionally transmitted as a semantic reference to later users (Kripke et
al. 1974: 512).
But this is problematic. Dummett argues as follows:
Kripke expressly wishes to allow that the association with a name of a description that does not apply
to the person … for which the name was originally introduced does not deprive the name of reference
to that person … It merely reveals a false belief about the reference. There is therefore no room in
Kripke’s account for a shift of reference in a chain of communication … Intuitively … there is no
such guarantee [of preservation of reference]. … Once this is conceded, the account crumbles away
altogether. We are left with this: that a name refers to an object if there exists a chain of
communication … at each stage of which there was a successful intention to preserve its reference.
This proposition is indisputably true; but hardly illuminating.
(1973: 150–1)

The challenge of Evans and Dummett to Kripke is that he explain reference

shift without transforming his positive ‘picture’ into a descriptivist one.
Cases of reference shift are indeed no counterexamples to Kripke’s picture
since he is not giving a theory. He is not claiming that a name refers to a
thing if and only if it is the thing the introducer of the name baptized with
the name. The ‘Napoleon’ case would be a counterexample to such a claim.
Nor is he claiming even that a name refers to a thing if and only if it is the
thing that the introducer of the name baptized with the name, and there is an
unbroken chain of intentions to preserve reference linking the introduction
of the name with the current use the name. The ‘Madagascar’ case would be
a counterexample to this claim. He is claiming at most that a name typically
refers to the thing its introducer baptized as long as, as in the ‘Napoleon’
case, there is no conscious opting-out of the initial referential practice and,
as in the Madagascar case, no widespread communal intention to which the
user intends to conform, which is in fact an intention to refer to something
other than the thing initially baptized. But this claim is consistent with the
description theory. Dummett’s criticism is that once Kripke’s story is
elaborated in a way that explains the possibility of reference change, it just
becomes a ‘social’ form of the description theory. What Kripke needs to do,
to answer this criticism, is to provide, without appealing to the speaker’s
intentions to give certain descriptions a determinative role in fixing the
reference of the name, an account of why Schmidt is not the reference of
the name in the Gödel scenario, while Madagascar is the reference of the
name in the situation Evans describes. But he does not do so.
4
Necessity
The Intelligibility of Essentialism

Much of the first lecture of Naming and Necessity is devoted to rebutting

Quine’s attack on quantified modal logic and essentialism. Kripke first
argues that the distinction between essential and accidental properties is an
intuitively intelligible one.
He next goes on to discuss and attack the notion of a possible world and
the associated notion of transworld (or crossworld) identity that he thinks
may lie behind the reluctance of some philosophers, perhaps Quine, to
accept the intelligibility of essentialist claims, and then to criticize David
Lewis’s counterpart-theoretic interpretation of de re modality. He finally
argues specifically for some non-trivial essentialist theses, including, in a
long footnote (footnote 56), the thesis of the essentiality of material origin.
First, we will consider Kripke’s response to Quine and a possible
rejoinder. As we shall see, ideas of Lewis not critiqued in Kripke’s
discussion (and not even developed by Lewis at the time) provide a
Quinean response to Kripke.
Quine’s critique, recall, begins from the conviction that the only notion of
necessity of which we can have any hope of making sense, as an ‘absolute
mode of truth’, is that of analyticity, a property of statements, not of things
in the extra-linguistic world. So it makes no sense to ask of a thing,
considered absolutely, whether it is necessarily the possessor of a certain
property. It is an analytic truth that 9 is greater than 7, and it is not an
analytic truth that the number of planets is greater than 7. So if ‘necessarily’
abbreviates ‘it is an analytic truth that’, ‘necessarily 9 is greater than 7’ is
true and necessarily ‘the number of planets is greater than 7’ is false, but it
is simply meaningless to ask whether 9, i.e. the number of planets, is such
that it is necessarily greater than 7. However, if it is meaningless to ask of
an individual whether it is necessarily greater than 7, the predicate ‘is
necessarily greater than 7’ can make no sense either, thought of as denoting
a property of individuals, and so the existential claim, ‘something is
necessarily greater than 7’, is also meaningless.
The moral Quine draws is that if we explain necessity in terms of
analyticity we cannot quantify across modal operators, i.e. make general
claims like ‘something is necessarily greater than seven’. And if we cannot
do that we might just as well restrict ourselves to the first grade of modal
involvement and the use of the semantic predicate ‘is analytic’ attachable to
quotation names of sentences.
The alternative is to reject the explanation of necessity in terms of
analyticity and embrace a conception of necessity as some feature of the
world and being necessarily true of as a relation between things and
properties. But this for Quine involves ‘Aristotelian essentialism’ and a
distinction between different ways of specifying the same object. However,
while Quine allows that in a particular linguistic context we may distinguish
between essential and accidental properties, treating some predicates as
denoting essential traits and others as denoting accidental traits, just as in a
particular linguistic context we may distinguish between knowing who
someone is and not doing so, treating possession of some pieces of de dicto
knowledge of a person rather than others as knowing who that person is, he
thinks that in both cases the belief in a context-independent notion is
mistaken (Quine 1981: 121) – that is why he labels the distinction
Aristotelian essentialism requires ‘invidious’.
Quine is here, in 1977, reiterating his bafflement with the idea that
something may, independently of how it is described, have some of its
properties essentially and others accidentally, already expressed in his
discussion of the cycling mathematician in Word and Object:
Mathematicians may conceivably be said to be necessarily rational and not necessarily two-legged;
and cyclists two-legged and not necessarily rational. But what of an individual who [is both]? Is this
concrete individual necessarily rational and contingently two-legged or vice versa? Just insofar as we
are talking referentially of the object, with no special bias towards a background grouping of
mathematicians as against cyclists or vice versa, there is no semblance of sense in rating some of his
attributes as necessary and others as contingent … Curiously, a philosophical tradition does exist for
such a distinction. … But however venerable … it is surely indefensible.
(Quine 1960: 199)
Quine is not here just saying that neither of these attributes is an essential
attribute of the mathematical cyclist. He is denying that any such attributes
are, in the absence of special bias, to be rated as either essential or
accidental. He is insisting that such ratings only make sense in a context
where a certain ‘background grouping’ is presupposed.
Kripke’s response is a bluff appeal to common sense:
it is very far from being true that this idea [that a property can be meaningfully be held to be essential
or accidental to an object independently of its description] is a notion which has no intuitive content,
which means nothing to the ordinary man. Suppose that someone said, pointing to Nixon, ‘That’s the
guy who might have lost’. Someone else says ‘Oh no, if you describe him as “Nixon”, then he might
have lost; but, of course, describing him as the winner, then it is not true that he might have lost’.
Now which one is being the philosopher, here, the unintuitive man? It seems to me obviously to be
the second. The second man has a philosophical theory. The first man would say, and with great
conviction, ‘Well, of course, the winner of the election might have been someone else. … On the
other hand the term “Nixon” is just a name of this man’. When you ask whether it is necessary or
contingent that Nixon won the election, you are asking the intuitive question whether in some
counterfactual situation, this man would in fact have lost the election. If someone thinks the notion of
a necessary or contingent property (forget whether there are any nontrivial necessary properties [and
consider] just the meaningfulness of the notion) is a philosopher’s notion with no intuitive content, he
is wrong.
(Kripke 1980: 41–2)

So Kripke claims against Quine that it is intuitively meaningful to ask of

Nixon, considered independently of how he is described, whether he has
essentially or only accidentally a property, being the winner, analytically
implied by some, but not all, descriptions of him. The property of being the
winner is one intuition rates as an accidental property of Nixon, but since
what Quine is rejecting is only non-trivial essentialism, Kripke must also
maintain that it is intelligible to suppose that some such properties are
essential properties of Nixon. Intuitively this is right. It is as firmly a part of
common-sense that Nixon could not have been a prime number or a month
of the year as it is that he could have lost the election. So it appears that the
position common-sense supports is a moderate, but non-trivial essentialism,
according to which things have some, but not all, their non-trivial properties
essentially. This is, in fact, Kripke’s position, but he cautiously endorses a
greater range of non-trivial essential properties than the negative ones just
indicated, including, for material objects, their origin and material
constitution. His response to Quine does not require him to do so, however.
The most important point to make about Kripke’s appeal to intuition is
that the mere firmness of our common-sense modal intuitions does not by
itself establish that ascriptions of truth-value to modal judgements are
independent of a background grouping. When we say that Nixon might
have lost, or pointing to Nixon, say ‘That man might have lost’, what we
say would be intuitively acceptable in almost all imaginable contexts. The
same is true of the statement of the cycling mathematician ‘He might have
had only one leg’ or ‘He might not have been capable of rational thought’.
But it does not follow that a background grouping is not presupposed in any
context in which one knowingly speaks of persons, against which all of
these attributes must be rated as accidental.
Quine’s crucial claim is that when there is no special bias towards one
way of describing a thing rather than another, it makes no sense to speak of
some of its non-trivial attributes as essential and others as accidental.
Kripke denies this. But Quine thinks the evidence does not support him.
Just as different tokens of the predicate ‘is believed by Ralph to be a spy’
stand for different properties, so that it can be true in one conversational
context to say of Ortcutt that he is believed by Ralph to be a spy and false in
another, in the same way different tokens of a modal predicate may stand
for different properties in different contexts. Consequently, one and the
same thing may correctly be truly said to be essentially so-and-so in one
context and truly denied to be essentially so-and-so in another. Such
variation in token predicate reference is entailed by the context-relativity of
essence Quine endorses. As Quine (2000: 427–8) puts it, with necessity
thought of as ‘a single idiom but an indexical or token-reflexive one like the
personal and demonstrative pronouns … each occurrence of it depends for
its interpretation on its textual or circumstantial context’. That is to say, it is
a consequence of Quine’s position that de re modal predication is, in the
terminology of David Lewis (1986), inconstant. A special case of such
inconstancy is that two tokens of the same modal predicate may be forced
to vary in reference when attached to two coreferential tokens of certain
different singular terms. The possibility is analogous to the one Quine
illustrates with his ‘Giorgione’/‘Barbarelli’ example (1953: 139).
‘Giorgione is so-called because of his size’ is true. ‘Barbarelli is so-called
because of his size’ is false. This does not count against the identity of
Giorgione and Barbarelli. The explanation is that the reference of the
predicate ‘is so-called because of his size’ depends on the singular term to
which it is attached. Attached to ‘Giorgione’ it stands for the property of
being called ‘Giorgione’ because of his size, attached to ‘Barbarelli’ it
stands for the property of being called ‘Barbarelli’ because of his size.
Consequently the existential claim ‘someone is so-called because of his
size’ is meaningless. Similarly, tokens of the same modal predicate attached
to coreferential tokens of different singular terms may, in the different
contexts thus created by those very attachments, have different references.
Hence, according to the thesis of the inconstancy of modal predication, the
truth of a singular modal claim does not guarantee the meaningfulness of
the corresponding existential claim.
Recognizing Quine’s thesis of the context-relativity of essence as
equivalent to Lewis’s thesis of the inconstancy of modal predication a range
of familiar examples of apparent de re opacity become relevant to the
debate between Quine and Kripke. Lewis famously argues against the
dualist that, consistently with the identity of me and my body, ‘I am
necessarily a thinking thing’ is true, whilst ‘my body is necessarily a
thinking thing’ is false. Many speakers of English will be inclined to
endorse Lewis’s modal judgements whilst being reluctant to draw the
conclusion that I am distinct from my body. This intuitively acceptable
position is consistent if the modal predicate ‘is necessarily a thinking thing’
is inconstant in reference and tokens of it attached to tokens of ‘I’ have a
different reference from tokens attached to tokens of ‘my body’. But Kripke
must reject it. Gibbard (1975) gives the famous example of Lumpl, the
lump of clay, and Goliath, the all-time coincident statue. Many are tempted
to say that there are not two purely material things where Lumpl/Goliath is,
although Lumpl could have survived being rolled into a ball and Goliath
could not, whilst Goliath could have undergone, through repair and
replacement of parts, a complete material change whilst Lumpl could not.
Hence they are disposed to regard the inferences: ‘Lumpl could have been
rolled into a ball and not destroyed, Goliath could not have been rolled into
a balled and not destroyed, so Lumpl is not Goliath’ and ‘Goliath could
have undergone complete material change, Lumpl could not, so Goliath is
not Lumpl’ as invalid. But again Kripke must reject this intuition. The
Lewisean can say that the modal predicates ‘could have been rolled into a
ball and not destroyed’ and ‘could have undergone complete material
change’ are inconstant in reference, and that attachment to tokens of
‘Goliath’ determines a different reference from that determined by
attachment to tokens of ‘Lumpl’. Similarly, he can say that the intuitive
difference in truth-value of ‘Any statue can be rolled into a ball and not
destroyed’ and ‘Any lump of clay can be rolled into a ball and not
destroyed’ can be explained, consistently with some statues being lumps of
clay, by saying that attachment to the restricted quantifier ‘any lump of
clay’ determines a different reference for the predicate from that determined
by attachment to the restricted quantifier ‘no statue’. Again, Kripke must
say that our dispositions to say the things Lewis can explain are mistaken.
Lewis gives the famous example of the dish pan and the piece of plastic.
The plastic is synthesized right in the mould, so that it no sooner exists than
it constitutes a dishpan; and the dishpan is destroyed just when the plastic is
incinerated. But suppose the factory had received its order for plastic
dishpans a day later. The same bit of plastic would have been made; for the
raw materials were already divided into portions just right to fit one mould.
But it would have been made in another mould, so that it would have
constituted a wastebasket. The dishpan would have been made the next day
of different plastic. The piece of plastic could have been a wastebasket, the
dishpan could not, the dishpan could have been composed of wholly
different atoms, the piece of plastic could not. But it reeks of double
counting, Lewis says (1986: 252), to say that here we have a dishpan, and
we also have a dishpan-shaped piece of plastic that is just where the dishpan
is. Again, this is an intuition many share, but Kripke must reject.
 Lewis also gives the example of GWR and GWR-. GWR is the Great
Western Railway. It came into being in 1923 after the grouping of railways:
the pre-grouping Great Western ought to have absorbed two other railways early on: the Bristol and
Gloucester and the Birmingham and Gloucester. But it tried to drive too hard a bargain. The line from
Bristol to Gloucester to Birmingham fell into rival hands. Therefore, after the grouping of railways in
1923, the post-grouping Great Western Railway lacked a part it might have had. Let GWR- be the
Great Western Railway as it actually was, without the missing line. Let GWR be the Great Western
Railway …‘they’ are one thing. What might have happened to it? It is GWR; so it would have been
greater. … It is GWR- so it would only have been a part of GWR.
(Lewis 1986: 249)

Kripke himself gives an example, in his unpublished lectures on identity,

which the Lewisean can cite as evidence for his position. It involves a plant
that never flowers so the plant and the stem coincide at all times (forget
about the roots as Kripke does). It is intuitively correct to say that the plant
but not the stem might have had a flower as a part, but to deny that there are
two things present.
We can easily cook up similar cases. Russell illustrates his distinction
between primary and secondary occurrences of descriptions by imagining
the retort of an irate yacht owner to the insensitive remark ‘I thought your
yacht was larger than it is’. Suppose the yacht owner had had the yacht built
to his own specification and had originally planned a three-decker, but
having incurred heavy gambling debts had to stop the construction before
the top deck was added. The yacht might have been larger, it might have
had the missing deck as a part, but the yacht as it actually is, without the
missing deck, could not have had the missing deck as a part. If the yacht
owner had had better luck the yacht as it actually is would merely have
been a part of a larger three-decker. Yet the yacht is the yacht as it actually
is.
In all these cases, the Lewisean will say, de re modal predications are
referentially opaque because the modal predicates are inconstant, different
tokens differing in reference. As Lewis puts it:
Attend to the variety of what we say about modality and counter-factuals de re, and I think you will
find abundant evidence that we do not have settled answers, fixed once and for all, about what is true
concerning a certain individual according to a certain … world. … Different answers are often right
in different contexts, as witness the comfort with which we adhere to, or presuppose, opposed
answers. It can very well happen that no answer is determinately right, for lack of the contextual
guidance that normally does the determining.
(Lewis 1986: 252)

If Quine and Lewis are right in emphasizing the role of context in

determining the reference of modal predicates and the truth-values of modal
predications it follows that it is not sufficient to answer worries about the
objective and context-independent meaningfulness of essentialism to point
to the distinction between rigid and flexible designators. ‘Goliath’ is a rigid
designator: small-scope and wide-scope readings of a statement of the form
‘Goliath might have been F’ necessarily coincide in truth-value, unlike
small-scope and wide-scope readings of ‘the statue owned by Jones might
have been F’. The same is true of ‘Lumpl’. But if Gibbard’s intuitions about
the example are correct, although Goliath is Lumpl, ‘Goliath might have
been rolled into a ball and not destroyed’ is false, whilst ‘Lumpl might have
been rolled into a ball and not destroyed’ is true, whilst ‘Lumpl might have
undergone complete material change’ is false, and ‘Goliath might have
undergone complete material change’ is true. So to preserve ‘Aristotelian
essentialism’ a distinction is needed within the class of rigid designators.
Lewis develops his account of the inconstancy of de re modal predication
in the framework of genuine modal realism, which Kripke rejects, as
explained in the next section. However, as he himself emphasizes, it is not
necessary to do so. But it will be useful to get a better sense of one way of
understanding inconstancy to look at some of the details of Lewis’s genuine
modal realism.
Lewis is a reductionist about modality. He provides a translation of
modal talk into non-modal talk. The language of his genuine modal realism
is extensional in Quine’s sense. The modal operators are replaced by
quantifiers over possible worlds, which are just big concrete objects,
maximal summations of objects related pairwise by spatio-temporal
connectedness (or an analogous relation). Two things are worldmates, or
belong to the same world, just in case they are (analogously) spatio-
temporally connected. Something is possibly true if and only if it is true in a
possible world, where this just means that it is true if we restrict the
quantifiers in the statement in question to parts of some possible world.
Hence, it is possibly true that there are talking donkeys on Lewis’s account
just in case there is a possible world in which there is a talking donkey, i.e.
just in case there is some maximal summation of spatio-temporally
connected things, a part of which is a talking donkey. Since Lewis accepts
that it is possible that there is a talking donkey he accepts that there is such
a massive summation, and hence that there is, really is, a talking donkey
(though not around here, i.e. spatio-temporally related to us). This is why,
as Lewis puts it, his genuine modal realism attracts ‘the incredulous stare’.
So far I have only sketched Lewis’s story about de dicto modality, which,
as noted, is entirely Quinean in both its ideology and ontology. The same is
true of Lewis’s development of his account to cover de re modality. This
appeals to the notion of a counterpart and develops a counterpart-theoretic
semantics for quantified modal statements. The key idea is that a
counterpart relation is a relation of similarity. The truth-conditions of de re
modal claims about individuals are a matter of those individuals standing in
similarity relations to others. I am necessarily a thinking thing because all
of my counterparts are thinking things; I am possibly an immaterial
thinking thing because some counterpart of me is an immaterial thing. Of
course, there are no immaterial things here, in the actual world, that is, the
maximal concrete object consisting of me and everything spatio-temporally
connected to me (there are no ghosts, Angels or whatever). But there are
other concrete universes not spatio-temporally connected to this one, Lewis
thinks (which is why the de dicto modal claim that there might have been
talking donkeys is true) and one contains an immaterial thing similar
enough to me to count as my counterpart. So I might have been an
immaterial thing.
But, crucially for our purposes, Lewis thinks that which counterpart
relation, which similarity relation, is relevant to the assessment of a de re
modal statement is dependent on context. In this respect de re modal
statements are no different from explicit statements of similarity, e.g. ‘the
United States is similar to the Roman Empire’, ‘human beings are like other
primates’.So there is no question of saying truly once and for all which
things are my counterparts. Hence there is no question of it being true of me
once and for all that I am possibly an immaterial thing. The occurrence of
the first person pronoun ‘I’ in the statement ‘I might have been an
immaterial thing’ will tend to ‘evoke’,as Lewis puts it, to bring to greater
salience, a counterpart relation, the ‘personal’ counterpart relation we might
say, which relates material things to immaterial things, so that if this effect
is not overridden by other features of the context the statement will be true.
But if ‘I’ is replaced by the coreferential ‘my body’ the effect will be to
bring to salience the ‘bodily’ counterpart relation, a relation which can only
hold between two material things, so that unless this effect is overridden by
other features of the context the resulting statement will be false. As noted,
though, other features of the context may override the effect of the meaning
of the subject term. Indeed, the mere fact that a de re modal assertion has
been made has some tendency, Lewis thinks, to evoke a counterpart relation
which allows it to be interpreted as a truth. Thus he responds to Kripke:
… those philosophers who preach that origins are essential are absolutely right – in the context of
their own preaching. They make themselves right: their preaching constitutes a context in which de re
modality is governed by a way of representing (as I think, by a counterpart relation) that requires
match of origin. But if I ask how things would be if Saul Kripke had come from no sperm and egg
but had been brought by a stork, that makes equally good sense. I create a context that makes my
question make sense, and to do so it has to be a context that makes origin not to be essential.
(Lewis 1986: 252)

It goes without saying, of course, that it is consistent with this that the facts
reported in context by assertions of de re modality are entirely objective
non-modal facts of similarity. Which facts of similarity are relevant is
determined by context, but these facts, once selected, are in no sense
subjective.
It may be objected to this Lewisean story that it allows too much
inconstancy. De re modal claims about one and the same thing do not
exhibit wide context-dependent variation in truth-value in the way Lewis
suggests. It is not the case that whatever de re modal statement is in
question there is some context in which a counterpart relation is selected
which makes it literally true and another in which a counterpart relation is
selected which makes it literally false. However, the examples given
already suggest that that it will be hard for an opponent of Lewis to defend
the claim that any non-trivial modal predicates have context-independent
referents. But it is important to see also that where there is constancy it can
be accommodated within the Lewisean framework. It may be, for example,
that our intentions and conventions are so entrenched that certain proper
names are associated with connotations that determine certain counterpart
relations irrespective of any other features of the context. It may be that the
use of any name of a person brings to such salience the personal counterpart
relation that it is impossible for other features of the context to override it,
and it may be that the personal counterpart relation is such that nothing can
count as a personal counterpart of Saul Kripke unless it is a human being
with a father who is a personal counterpart of Rabbi Kripke. If so it will be
true in any context to say, ‘Kripke could not have had a different father’,
and the statement resulting from replacement of ‘Kripke’ by any other
proper name of Kripke, thought of as a name of a person, will also be true
in any context. But this will simply be a matter of what is rightly called ‘a
personal counterpart of Saul Kripke’. Nothing in this account is inconsistent
with the thesis of the context-relativity of essence. So, whilst the Lewisean
story is tailor-made to account for the inconstancy of de re modal
predication, it can also account for constancy if it occurs.
Turning back to Quine: he does not, of course, use the sort of examples in
his critique of quantified modal logic that have since become familiar and
have been given here since his focus is merely on establishing that when ‘is
necessarily so-and-so’ is understood in terms of ‘is necessarily true’,
interpreted as ‘is an analytic truth’, quantified modal logic is either trivial or
nonsense; once this has been established, he takes the context-relativity of
essence, in any other sense of essence, as evident and needing no
illustration. And, in fact, from the Lewisean viewpoint sketched Quine’s
example of 9/the number of planets – which, like his other examples, is
intended merely to illustrate that anything true of a thing follows
analytically from some way of specifying it – would not be best-chosen to
illustrate the context-relativity of essence. Inconstancy of modal
predication, as underpinned by a multiplicity of counterpart relations must
be irrelevant in this case since 9 is its own only counterpart under the
counterpart relation evoked by either ‘9’ or ‘the number of planets’ (though
that this is so, for a Lewisean, reduces to the mere fact that we do not treat
anything else, however similar, as a counterpart of 9, i.e. as relevant to the
truth-value of de re modal assertions about 9 (Lewis 1984: 235)).
Nonetheless, appeal to the sort of examples sketched here is entirely in
conformity with Quine’s position. For Quine material objects are simply the
material contents of spatio-temporal regions. So Goliath is Lumpl. The
piece of plastic is the dishpan. So the difference between our intuitive
ascriptions of truth-value to the claims ‘Goliath might have been rolled into
a ball and not destroyed/undergone complete material change’ and ‘Lumpl
might have been rolled into a ball and not destroyed/ undergone complete
material change’ is grist to his mill. Quine does not accept Lewis’s plurality
of (concrete) possible worlds, but as emphasized, the thesis that modal
predication is inconstant does not need development within the Lewisean
modal realist framework. What the thesis does require is that modal
predication is reducible to non-modal predication in the sense that what is
ascribed in context by the use of a modal predicate can be ascribed
independently of context by the use of a non-modal predicate. But there is
nothing here in tension with Quine’s thinking.
To sum up this section: Kripke insists against Quine that ascriptions of
essential and accidental properties to individuals make objective sense and
argues that this is the intuitive, common-sense, position. But there are
familiar intuitions Kripke cannot accommodate, and so it is doubtful
whether de re modal ascriptions have the sort of intelligibility he thinks
they have. The alternative view of essence as context-relative can explain
the recalcitrant intuitions, and even if there is more constancy than Lewis
thinks, it can be accommodated within his framework.
The Rejection of the Problem of Transworld
Identity and the Critique of Counterpart Theory

After defending essentialism Kripke goes on to criticize the idea that there
is a genuine problem of ‘transworld identity’ or ‘transworld identification’.
He suggests that this idea is one motivation people have for thinking that
the distinction between essential and accidental properties is unintuitive.
Given the context of these remarks it is tempting to read Kripke as thinking
of Quine as one of the people so motivated, though he does not say so
explicitly. Quine does indeed raise some worries about transworld identity
in ‘Propositional Objects’:
How is Catiline to be identified in … various possible worlds? Must he have been named ‘Catiline’
in each, in order to qualify? How much can his life differ from the real life of Catiline without his
ceasing to be our Catiline and having to be seen as another man of that name? Or again, how much
can the [Great] Pyramid differ from the real one? … Is it sufficient, for its identification in other
worlds that it have been built by Cheops? How much then can his life differ from the life of the real
Cheops without his ceasing to be our Cheops?
(Quine 1969: 153)

But it is clear that the primary focus of Quine’s worry is that, unlike Kripke,
he does not have any strong convictions about the truth-values of de re
modal statements. He does not have any answer to the question ‘How
different from the way he is might Catiline have been?’ Quine’s worries
about transworld identity are not the basis of his doubts about de re modal
predication. He is merely using the language of possible worlds to express
them. The basis of his doubts is his long-standing rejection of context-
independent essence.
Other philosophers who were waxing sceptically at the time about
transworld identity were similarly primarily sceptical about the
intelligibility and truth-values of de re modal ascriptions rather than about
statements of transworld identity expressed in the language of possible
worlds. Thus Chisholm discusses the possibility of a world in which Adam
and Noah ‘swap roles’ and the consequent apparent possibility that there
could be two worlds (the actual one and this one) discernible only inasmuch
as in one Adam plays the Adam role and Noah the Noah role and in the
other things are reversed. The existence of this pair of worlds can be easily
ruled out if it is accepted that Adam and Noah differ in their essential
properties. However, this is precisely what worries Chisholm. He does not
see how we can tell which of Adam’s properties are essential and which
accidental:
What … properties … are essential to Adam? Being the first man? Having a name which, in English,
begins with the first letter of the alphabet? But why these properties? … It seems to me that even if
Adam does have such essential properties, there is no procedure at all for finding out what they are.
And it also seems to me that there is no way of finding out whether he does have any essential
properties.
(Chisholm 1967: 6)

Nevertheless, Kripke thinks, resistance to the idea that there is a meaningful

distinction between the essential and the accidental properties of individuals
is motivated for some philosophers by the (correct) thought that the
question of essential properties is equivalent to the question of identity
across possible worlds, together with the (incorrect) thought that we must
possess a criterion of identity in order to ascertain which individual in
another possible world (in which no one is exactly the same as he is in the
actual world) is, say, Nixon (Kripke 1980: 42).
This second thought is incorrect, Kripke argues, because it depends on a
mistaken picture of what possible worlds are:
One thinks … of a possible world as if it were like a foreign country. One looks upon it as an
observer. Maybe Nixon has moved to the other country and maybe he hasn’t, but one is given only
qualities. One can observe all his properties, but, of course, one doesn’t observe that someone is
Nixon. So we better have a way of telling in terms of properties when … we come across one of
these possible worlds, who was Nixon.

But:
A possible world isn’t a distant country … a possible world is given by the descriptive conditions we
associate with it. …‘Possible worlds’ are stipulated, not discovered by powerful telescopes. There is
no reason why we cannot stipulate that, in thinking about what happened to Nixon in a certain
counterfactual situation, we are talking about what would have happened to him.
 (Kripke 1980: 44)

Of course, if someone makes the demand that every possible world has to be described in a certain
qualitative way, we can’t say ‘Suppose Nixon had lost the election’. Why … make this demand? That
is not the ordinary way we think about counterfactual situations. We just say ‘Suppose this man had
lost’. … even if there were a purely qualitative set of necessary and sufficient conditions for being
Nixon, … I … would not demand that we find these conditions before we can ask whether Nixon
might have lost the election, nor does it demand that we restate the question in terms of such
conditions. We can consider Nixon and ask what might have happened to him had various
circumstances been different.
(Kripke 1980: 46–7)

Again:
Don’t ask: how can I identify this table in another possible world except by its properties? I have the
table in my hand, I can point to it, and when I ask whether it might have been in another room, I am
talking, by definition, about it. I don’t have to identify it after seeing through a telescope.
(Kripke 1980: 52)

Two points are being emphasized in these passages. First, possible worlds
are ‘stipulated not discovered’. Second, I can identify an individual about
which I can go on to make a supposition without giving a purely qualitative
necessary and sufficient condition – just as ‘this man’ or ‘Nixon’.
When Kripke says that possible worlds are stipulated not discovered he is
not claiming that I can stipulate that something is possible, as I can stipulate
that I will use a new word, say ‘squelschy’, as synonymous with, say,
‘wise’. Whether there is a possible world at which so-and-so is the case,
whether it is possible that so-and-so is the case, is, in Kripke’s view, an
objective matter of fact, which is so or not independently of any stipulation.
But I can stipulate that my counterfactual supposition is about a particular
individual, say Nixon, that is, that I am to be regarded as speaking correctly
when I say ‘Nixon might have lost’ if and only if Nixon might have lost. I
might say ‘The PM might have been a member of the Conservative Party’
and to ward off misunderstanding say, ‘I mean that man, who is in fact PM,
might have been a member of the Conservative Party’. Here I am
stipulating that my ambiguous utterance is to be read with the description
‘the PM’ given wide scope, and so that I am entertaining the possibility of a
situation in which a particular actual individual has a certain property.
But it does not follow from the possibility of such a stipulation that I can
guarantee, by stipulating that I am considering a possible world in which
Nixon, that very man, lost, that there is such a world. If I stipulate that I am
considering a possible world in which Nixon had different parents I do not
thereby guarantee that there is a possible world in which he had different
parents. In fact, if Kripke’s own ‘necessity of origin’ thesis is correct, there
is no such possible world. This point may be put (Mackie 2008) by saying
that if Kripke’s thesis of the necessity of origin is correct then in stipulating
that I am talking about Nixon in saying that Nixon might have lost, I am
implicitly, whether I know it or not, stipulating that I am talking about
someone with Nixon’s actual parents, and that my supposition is not merely
that Nixon might have lost, but that someone with Nixon’s actual parents
might have lost.
Kripke’s second point is that I can identify someone using a name or
demonstrative expression, as ‘Nixon’ or ‘this man’, and go on to make a
statement about what might have happened to him, without having any
knowledge of necessary or sufficient qualitative conditions for being him.
There is no more difficulty in understanding the statement that there is a
possible world in which Nixon lost the election than there is in
understanding the statement that there is a possible world in which a man of
such and such a qualitative description lost the election. And this is so, at
least, because the former statement means no more than the familiar modal
claim ‘Nixon might have lost the election’ (‘It is better … to avoid
confusion, not to say “In some possible world, Humphrey would have
won”, but rather, simply, “Humphrey might have won”’ (Kripke 1980: 48)).
It is hard to grasp the point of Kripke’s remarks about trans-world
identity because the target is a position no one has ever held. It has two
components. The first is the Lewisean reductionist notion of a possible
world as a maximal sum of spatio-temporally related parts, according to
which other possible worlds are indeed to be thought of as like other
countries. We do not need modal vocabulary to explain this notion of a
possible world. This is the ‘foreign country’ conception Kripke repudiates.
The second component of Kripke’s target is the idea that some individuals
(the same individuals) ‘live simultaneously’ in these different countries.
This is no part of Lewis’s conception – as Kripke emphasizes. Lewis thinks
that no individual is part of several worlds, he merely has counterparts in
several. Kripke’s arguments that the problem of transworld identity is a
pseudo-problem are aimed at showing that if, as he thinks, the first
component of this picture is mistaken, i.e. that the notion of a possible
world is rather the notion of a possible state or history of the one actual
world, there is no genuine question to be asked about the necessary and
sufficient condition for an individual to be an ‘inhabitant’ of two possible
worlds. His remarks seem bafflingly platitudinous because they are
platitudinous – in the same sort of way a mathematical proof can seem
obvious once it is given.
However, there are contentious issues on which Kripke takes a stand
which can be, but need not be, phrased in the language of transworld
identity. In the Preface to Naming and Necessity Kripke indicates one such
issue and his stance on it in a footnote (1980: 18): ‘if there is a
philosophical argument excluding qualitatively identical but distinct worlds,
it cannot be based exclusively on the supposition that worlds must be
stipulated purely qualitatively’. The world in which Adam and Noah have
switched roles and the actual world would be such a pair of indiscernible
but distinct worlds, but the existence of this world can be ruled out given
Kripke’s thesis of the essentiality of biological origin, because such a world
would be one in which Adam and Noah had origins other than their actual
ones. However, Kripke’s thesis of the essentiality of origin does not rule out
another possible world in which someone other than Noah, an actual or
possible individual with no essential property incompatible with Adam’s
origin, plays the Adam role, for, as he emphasizes, his thesis is one about
the necessary conditions of being an individual, not the sufficient
conditions. To rule out such a world it seems that we must suppose that
every individual has a non-trivial individual essence, a property essential to
that individual and necessarily not possessed by any other. Kripke does not
take a stand on whether (all) individuals have individual essences. He does
not assert or deny that some possible or actual individual other than Adam
could have played the Adam role. But he does assert that there is no
argument for denying this based simply on the supposition that worlds, i.e.
world states, must be stipulated purely qualitatively, since they need not be.
As noted, Kripke recognizes that although Lewis adheres to the first
component of his ‘foreign country’ target picture of possible worlds, he
does not think that there are individuals living simultaneously in distinct
worlds, but rather thinks that individuals are world-bound, having only
counterparts in other worlds. In a footnote, bracketing his scepticism about
Lewisean reductionism, Lewis presents his famous ‘Humphrey’ objection
to counterpart theory.
[For Lewis] if we say ‘Humphrey might have won the election’, we are not talking about something
that might have happened to Humphrey, but to someone else, a counterpart. Probably, however,
Humphrey could not have cared less whether someone else, no matter how much resembling him,
would have been victorious in another world. Thus Lewis’s view seems to me even more bizarre than
the usual notion of transworld identification that it replaces.
(Kripke 1980: 45 n. 13)

The objection also occurs in ‘Identity and Necessity’ (1971: 148):

Even more objectionable is the view of David Lewis. According to Lewis, when we say ‘Under
certain circumstances Nixon would have gotten Carswell through’, we really mean ‘Some man, other
than Nixon but closely resembling him, would have gotten some judge, other than Carswell, but
closely resembling him, through’. Maybe that is so, that some man closely resembling Nixon would
have gotten some man closely resembling Carswell through. But that would not comfort either Nixon
or Carswell, nor would it make Nixon kick himself and say ‘I should have done such and such to get
Carswell through’. The question is whether under certain circumstances Nixon himself would have
gotten Carswell through.

This argument has generally been found unconvincing. The starting point
is that when we say ‘Humphrey might have won’ we mean that Humphrey
might have won, i.e. that Humphrey himself might have won. We are
talking about Humphrey. But on the counterpart theory we are talking about
someone else, a counterpart to Humphrey. So counterpart theory gets the
truth-conditions wrong, it assigns truth-conditions to our modal statements
which render them irrelevant to our concerns.
 The objection to this argument is that according to counterpart theory
‘Humphrey might have won’ is about Humphrey, as well as being about
someone else; it says that Humphrey, the man himself, has a winning
counterpart. Lewis expresses this response to Kripke himself:
Counterpart theory does say that someone else … enters into the story. … Insofar as the intuitive
complaint is that someone else gets into the act, the point is rightly taken. But I do not see that is an
objection, any more than it would be an objection against ersatzism that some abstract whatnot get
into the act. What matters is that someone else … should not crowd Humphrey out. And there all is
well.
(1986: 196)

But perhaps Kripke has another point. According to Lewis ‘Humphrey

might have won (though he didn’t)’ does not just say something about
someone else, as well as Humphrey; it entails that there is a winner (a
counterpart of Humphrey). But intuitively it does not seem that the claim
that I might have been so-and-so ever entails that someone or something (to
whom I am similar) is so-and-so. That I might have had green hair (because
someone might have poured a pot of paint over my head), does not, it
seems, entail that there is someone with green hair. But according to Lewis,
it does. Of course, this is of a piece with the fact that for Lewis the de dicto
modal claim that there might have been talking donkeys entails that there is
something which is a talking donkey. But at the time of Naming and
Necessity Lewis had not made this clear. Perhaps, then, we can regard
Kripke as anticipating in his ‘Humphrey’ objection the objection to Lewis’s
extravagant ontology which later writers made explicit. But if so the
objection is not just to the extravagant ontology. It is rather to the Lewisean
position that this ontology is entailed by the truth of familiar modal claims
– e.g., that I might have had green hair entails that there is someone who
has green hair. And, indeed, it does seem what is most objectionable about
Lewis’s position is not his extravagant ontology (about which agnosticism
seems the only sensible position), but his claim that we can know, on the
basis of our knowledge of such familiar modal facts as those just indicated,
that there are people with green hair and talking donkeys.
The Essential Properties of Individuals

Having argued against Quine that the distinction between essential and
accidental properties makes objective, context-independent, sense, Kripke
goes on in the third lecture to defend specific non-trivial essentialist claims.
In particular, he defends the necessity of biological origin, the necessity of
material origin for artefacts, the necessity of constitution for material
objects and the necessity of identity and distinctness for everything.
In a famous passage, prompted by an article by Timothy Sprigge, Kripke
affirms that Queen Elizabeth II could not have had different parents, and, in
particular, could not have been the biological daughter of Mr and Mrs Harry
Truman:
They might have had a child resembling her in many respects. Perhaps … even … a child who
actually became the Queen of England. … This still would not be a situation in which this very
woman whom we call ‘Elizabeth II’ was the child of Mr and Mrs Truman. … How could a person
originating from different parents, from a totally different sperm and egg, be this very woman? One
can imagine, given the woman, that various things in her life would have changed. … One is given,
let’s say, a previous history of the world up to a certain time, and from that time it diverges from the
actual course. … And so it is possible that even though she were born of those parents she never
became Queen. … But what is harder to imagine is her being born of different parents. It seems to
me that anything coming from a different origin would not be this object.
(1980: 112–13)

Kripke’s appeal to intuition here is generally acknowledged to have some

weight, but how great has been contested. We have seen that the Lewisean
story can accommodate the existence of such intuitions and is consistent
with Kripke’s claim that it is harder to imagine Elizabeth’s being born of
different parents than to imagine her not becoming Queen. Other
philosophers have similarly expressed opposing intuitions about biological
origin. Dummett suggests: ‘we may quite intelligibly wonder what
difference it would have made to Franz Kafka’s outlook if he had not been
of Jewish descent and upbringing’ (1973: 132). Wiggins expresses doubts
about the necessity of origin in these terms:
Perhaps the speculator has to be able to rebut the charge that he has lost his subject of discourse
[Julius Caesar] if he changes his parents or origin. But now I ask can he not rebut the charge by
claiming to speculate about how the man whom Brutus murdered in 44 BC would have fared if, say,
Marius had been his father?
(Wiggins 1980: 116, n. 22)

It is the most natural thing in the world to ponder on how things would have
been if one’s parents had been different. ‘If I had been the son of the King
I’d have had more opportunities in life’.
Thinking about origin more generally the same point holds. It is natural
to say in an appropriate context that the same building might have been
built of a different lot of bricks or the same ship of a different batch of steel
(Strawson 1979). We understand Lewis very well when he says, describing
his dishpan/plastic case ‘… suppose the factory had received its order for
plastic dishpans a day later. The same piece of plastic would have been
made. But … it would have constituted a wastebasket. The dishpan would
have been made the next day out of different plastic’ (1986: 252). There is a
clash of intuitions, and such a clash cannot itself constitute a reason to
endorse Kripke’s intuitions rather than those of Dummett, Wiggins,
Strawson and Lewis.
But Kripke also gives or suggests several arguments in support of the
necessity of origin.
The first argument, suggested by Kripke’s remark that when considering
alternative possibilities ‘one is given … a previous history of the world, and
from that time it diverges from the actual course’ is spelled out by
Dummett:
It may truly be said of President Nixon, for instance, that he might never have been a politician,
because there was a time in his life at which it would have been true to say that he might never
become a politician. … But we cannot push back the moment in respect of which a property is to be
characterised as [something it might not come to possess] beyond the point at which the object came
into existence.
(Dummett 1973: 131)
If all possibilities for Elizabeth are limited in this way to divergences from
actuality after she came into existence she could not have had different
parents. But the principle is absurd since it makes all features she possessed
at her origin (however intuitively accidental or relational) essential. Kripke
does not in fact endorse the principle. However, he does tentatively endorse
a slightly weaker one which requires that possibilities for Elizabeth must
involve divergences from the actual course of world history but allows that
such divergences may take place before Elizabeth came into existence:
Ordinarily when we ask intuitively whether something might have happened to a given object, we
ask whether the universe could have gone on as it actually did up to a certain time, but diverged in its
history from that point forward so that the vicissitudes of that object would have been different from
that time forth. Perhaps this feature should be erected into a general feature about essence. Note that
the time at which the divergence from actual history occurs may be some time before the object itself
is actually created.
(1980: 115)

Of course, this principle, though indeed intuitive, cannot explain why we

find biological origin, as opposed say, to original location, hard to think
away.
Kripke’s second argument for the necessity of origin is given in support
of his thesis that (some of) an artefact’s original material is a necessary
feature of it: a table could not have been made from a completely different
block of wood from that from which it was originally made. This argument
has been much debated. Kripke offers it as ‘something like a proof’ in his
footnote 56:
Let ‘B’ be a name (rigid designator) of a table, let ‘A’ name the piece of wood from which it actually
came. Let ‘C’ name another portion of wood. Then suppose B were made from A, as in the actual
world, but another table D were simultaneously made from C. (We assume there is no relation
between A and C which makes the possibility of making a table from one dependent on the
possibility of making a table from the other.) Now in this situation B 6¼ D; hence even if D were
made by itself and no table were made from A, D would not be B.
(1980: 114, n. 56)

Most commentators interpret this argument for the necessity of origin as

having as a premiss a ‘sufficiency of origin’ principle for table identity, a
principle to the effect that sameness of origin (in some sense of ‘origin’) is
sufficient for table identity across worlds, and there has been considerable
debate about whether any ‘sufficiency of origin’ principle which can do the
work required has any plausibility.
The most obvious way to interpret the argument is as relying on the
simple principle that if a table could have been made entirely of a certain
hunk of matter, no other table could have been made entirely from that hunk
of matter (the following draws on Noonan 1983).
The argument can then be understood as follows, with reference to the
following specification of possible worlds (‘x/y’ means ‘x is made from y’):
W0: Bt/Am; W1: T/Cm; W2: Bt/Am, Dt/Cm. Suppose a table Bt made
from a hunk of matter Am in the actual world W0, and another hunk of
matter Cm – completely different from hunk of matter Am – from which a
table exactly like table Bt could be made. Now consider a possible world
W1, in which matter Cm is made into a table (T), no table in this world
being made from matter Am. Is the table made from matter Cm in W1, the
table Bt, which in the actual world W0 is made from matter Am? No,
Kripke answers, for there is another possible world, W2, in which both
matter Am and matter Cm are made into tables. Clearly the table made from
Am in W2 is table Bt, the table made from matter Am in the actual world,
W0; so the table made from matter Cm in W2, call it Dt, cannot be table Bt
in W2, and hence (by the necessity of distinctness, which Kripke has argued
for in his earlier paper ‘Identity and Necessity’) cannot be Bt anywhere at
all. So the table made from matter Cm in W1, which is identical with table
Dt, the table made from Cm in W2, cannot be table Bt either. Thus table Bt
could not have been made from a hunk of matter completely distinct from
that from which it was actually made, for the supposition of this being so is
the supposition of just such a hunk of matter Cm and possible world W1 as
the argument has shown to be impossible. Since the role of the actual world
in the argument could be filled by an arbitrary possible world, the argument,
if good, establishes also that table Bt could not have been made from a hunk
of matter completely different from any from which it could have been
made.
 This argument assumes the simple ‘sufficiency of origin’ principle for
table identity – that sameness of original matter is sufficient for table
identity across possible worlds, or that any hunk of matter has only one
possible table ‘in’ it. This is what justifies the claim that the table made
from Cm in W1 is the table Dt made from Cm in W2. Evidently any version
of the argument will have to assume some sufficient condition of table
identity across worlds. Any such assumption, of course, requires
essentialism. The stated assumption entails, for example, that the matter
from which table Bt is made is necessarily such that no other table is made
from it, and that table Bt is necessarily such that it alone is ever made from
the matter from which it is actually made. But Kripke is not trying at this
point to respond to a sceptic about essentialism. He thinks he has already
done enough to refute such scepticism already; he is aiming rather to give a
proof of a particular essentialist claim appealing to a less controversial one
as a premiss.
The assumption of Kripke’s argument, as reconstructed, that sameness of
original matter is a sufficient condition for transworld identity, plausibly
holds for many other material objects if it holds for tables. So the argument
may have a wide range of application. But there are nonetheless kinds of
material object for which it is not a plausible sufficient condition of identity.
Consider rivers. It is not the first quantity of water which constitutes it
which determines the identity of a river, but its source and its bed. So
different rivers in different possible worlds might be originally constituted
of just the same water and the same river of different water (the same is true
of artificial lakes). In fact, the assumption that sameness of original matter
is sufficient for identity is not even plausible for tables. The matter which
was used to make table Bt could have been used to make a table of a
completely different type, original structure and design. Intuitively that
table would, or might, not have been table Bt. (Or if you think otherwise,
consider a completely different piece of furniture, say a chair. Intuitively if
a chair had been made out of matter Am it would not have been table Bt,
constituted of the same matter but in a different form, but a table of a
completely different type may be as unlike Bt as the chair.) So it is wrong to
say that any tables identical in their original matter must be identical tout
court. The obvious response is to modify the sufficiency principle so that
the identity of the plan according to which the table is made also figures in
it (Salmon 1982: 211). With this modification we do not have to allow that
any table, no matter how different, originally made from the same matter as
table Bt is table Bt. It is, of course, an implication of this modified
sufficiency principle that table Bt could not have been made from matter
Am according to a radically different plan (according to which another table
could have been made from Am). It is not clear that this is intuitively
acceptable.
Whatever is said about this, replacing the original sufficiency principle
by this weakened one still allows us to deduce Kripke’s conclusion that
table Bt (in fact made in accordance with plan P1) could not have been
originally made (in accordance with some possible distinct plan Pn) from
matter Cm, completely distinct from that from which it was actually made.
(The relevant specification of world is: W0: Bt/AmP1; W1: T/CmPn; W2:
Bt/AmP1, Dt/CmPn.)
However, reflection on the problem of the ship of Theseus suggests that
the plausibility of sameness of original matter plus plan as a sufficient
condition for table identity is quite severely restricted (assuming that tables
are not relevantly different from ships, of course). Thomas Hobbes
imagines a world W2 where the ship of Theseus is gradually repaired and
its planks replaced over time and in which the replaced planks are hoarded,
carefully preserved and afterwards put together so as once again to
constitute a ship of exactly the same design as the original ship of Theseus.
Call the ship of Theseus in this world Bs, its original planks Am, the ship
subsequently made from its original planks Ds. Ships Bs and Ds are distinct
ships with the same origin (they may be floating side by side). So far this
does not refute the principle that sameness of original matter and plan is a
sufficient condition of identity across distinct possible worlds. But now
consider the world W1 in which the ship of Theseus survives miraculously
intact and unrepaired until annihilated in a twenty-first-century nuclear
holocaust. Consider the sole ship made from matter Am in this world. (The
relevant specification of worlds is: W1: T/Am; W2: Bs/Am; Ds/Am.) Is it
Bs or Ds or neither? In accordance with the assumption that sameness of
original matter and plan is sufficient for ship identity across worlds it must
be both.
The problem can be avoided, of course, by accepting that sameness of
original matter plus plan is not sufficient for identity (of ship or table), but
maintaining the sufficiency of original matter, together with both plan and
moment of origin. With this further weakening of the original sufficiency
principle we do not have to identify the two ships in the world Hobbes
describes, nor to identify the single ship in the nuclear holocaust world with
the two ships in Hobbes’ world. Nevertheless, since any artefact must be
made at some time, Kripke’s conclusion that table Bt could not have been
made from a completely different hunk of matter Cm from the matter Am
from which it was actually made is still secured. (Imagine a world W1 in
which a table is made from Cm. Since Am is completely distinct from Cm
there is a world W2 in which two tables are made, one from Am according
to the plan in accordance with which Bt is actually made and at the time Bt
is actually made and one from Cm according to the plan in accordance with
which the table in W1 is made from Cm and at the time the table in W1 is
made from Cm. One is Bt, the other, Dt, is the table made from Cm in
world W1, so Bt is not the table made from Cm in world W1.)
However, endorsing this weakened sufficiency of origin principle, in
which time of origin is a component, and rejecting the stronger principles
we have seen to run into problems with the example of the ship of Theseus,
requires accepting that table Bt could not have been originally made at any
other time than its actual time of origin from its actual original matter in
accordance with its actual plan. Otherwise we cannot allow that another
table (a plank hoarder’s table) could have been made both from that matter
and in accordance with that plan at that other time. This seems implausible.
We can easily imagine that the making of the table was delayed for a few
days (or weeks, or years …) and then it, that very same table, was made as
originally planned from the hunk of matter set aside.
 It may also seem that we run into problems with the ship of Theseus. For
are we not required to identify the ships Bs and Ds in Hobbes’ world W2?
For consider a world W3 where the planks taken from the ship of Theseus
are not replaced but are simply stored and then reassembled into a ship at
the time at which in W2 Ds is constructed. This is a world in which the ship
of Theseus undergoes disassembly and later reconstruction. Now given the
revised sufficiency principle for transworld identity are we not forced to
identify the single ship in this world with both Bs and Ds – and hence to
identify Bs and Ds with each other? No. If there is indeed a single ship in
W3 which has undergone disassembly and reconstruction its moment of
origin is the same as the moment of origin of Bs in W2. So there is no ship
in W3 which has the same moment of origin as Ds in W2. What is true is
only that the reconstruction of the single ship in W3 takes place during the
same period of time as the original construction of Ds in W2.
Nevertheless, we have to allow that artefacts made at a particular time
could, as a result of delay, have been made at other times (from the same
material according to the same plan). So we still need to find some further
weakening of the sufficiency of origin principle if we are to rescue Kripke’s
footnote argument for the necessity of origin.
One suggestion is that we replace the original sufficiency of origin
principle in the argument with the premiss that sameness of original matter
plus plan is sufficient for transworld identity when the matter in question
constitutes two distinct artefacts so planned in neither of the worlds
between which the transworld identity obtains. But this does not get us past
the problem with the ship of Theseus. Imagine a world W4 where the ship
of Theseus is repaired and patched up exactly as ship Bs is in W2, and has
exactly the history Bs has in W2, but its replaced planks are never brought
together again to make another ship, and another possible world W5 where
there is no original ship of Theseus but where a ship is constructed from the
same planks as ship Ds is constructed from in W2, in exactly the same way,
beginning at exactly the same time, and that this ship subsequently has the
history Ds has in W2. Unless sameness of original matter, moment of
origin, plan and history is denied to be sufficient for transworld identity, the
ship in W4 is Bs and the ship in W5 is Ds. But Bs is not Ds so the ship in
W4 is not the ship in W5 despite the fact that they are related as the latest
proposed sufficiency of origin principle requires.
A second suggestion is to replace the original sufficiency of origin
principle by the principle that sameness of original matter plus plan is
sufficient for transworld identity when the matter in question is in no
possible world the original matter of two distinct artefacts made according
to that plan. The ship of Theseus creates no problems for this suggestion,
but it is useless nevertheless. If Kripke’s argument is revised in this way it
goes through only if it is assumed that neither matter Am nor matter Cm is
in any possible world the original matter of two distinct tables (made
according to the same plan), which must be false.
A third suggestion is to replace the original principle by the principle that
sameness of original matter plus plan together with sameness of position in
a chronological series of things made from that matter according to that
plan is sufficient for transworld identity. This suggestion avoids the
implausibility of saying that a table or ship’s time of origin is essential to it.
But it entails that in the example of worlds W4 and W5 just given the ship
in W4 is the ship in W5 (each is the only ship in its world constructed from
the matter Am); but the former is Bs and the latter is Ds (unless the latter is
Bs too, as the principle in fact entails, in which case it must, absurdly, be a
necessary property of ship Ds that it is preceded in existence by ship Bs, so
that sameness of original matter plus plan, plus time of origin, plus total
subsequent history is not sufficient for transworld identity).
A fourth suggestion is to replace the original sufficiency of origin
principle with the principle that sameness of original matter, plan and causal
origin is sufficient for transworld identity. It seems clear that we can
interpret ‘causal origin’ in such a way as to ensure that the ship of Theseus
presents no problem for this principle, but if we interpret it, as seems
plausible, so that in the case of an artefact made by a single craftsman its
‘causal origin’ includes the activities of the craftsman and the specific
intentions he was guided by in manufacturing his product, the revision of
Kripke’s argument got by this repair allows that, if table Bt was in fact
made by a single craftsman, it could have been made from matter Cm
entirely if it had been made by the same craftsman and for the same specific
purposes as in the actual world.
This review suggests that there is no evident weakening of the original
sufficiency of origin principle which will both yield Kripke’s conclusion
and not issue in counter-intuitive consequences. Some of the arguments
given here, however, depend upon a controversial description of the ship of
Theseus case. But as we shall see now, describing the case differently will
provide no support for Kripke’s footnote 56 argument for the necessity of
origin. We have assumed that in the world described by Hobbes two ships
come into existence at different times, constituted from the same matter and
in accordance with the same plan: the original ship, which undergoes repair
and replacement of parts and the plank hoarder’s ship, which is constructed
later from the original planks of the continuously repaired ship.
But this description is controversial since it conflicts with an intuitively
plausible principle, governing identity over time, ‘the only x and y
principle’ (Noonan 2003: 129ff.). The formulation of this is delicate, but
intuitively what it says is that whether a later individual x is identical with
an earlier individual y can depend only on facts about x and y and the
relationships between them; it cannot depend upon facts about individuals
other than x or y. Otherwise put, what the principle asserts is that whether x
is identical with y can depend only upon the intrinsic relationships between
them; it cannot be determined extrinsically. The intent of the principle is to
rule out ‘best candidate’ theories of identity over time, according to which
whether a later individual x is identical with an earlier individual y can
depend upon whether there is any better candidate than x around at the later
time for identity with y. It implies that if two events are parts of the history
of a single entity of a kind in one situation, then they must be parts of the
history of a single entity of the kind in any second situation in which both
they and all the events which are parts of the history of the entity in the first
situation remain present – no matter what additional events which have no
causal effect on the history of the entity are also present in the second
situation. To see the significance of this principle for the description of the
case of the ship of Theseus, we need to compare the situation Hobbes
describes (world W2) with two others. The first situation (called W4 in the
preceding discussion) is that in which the ship of Theseus undergoes repair
and replacement of parts, but the replaced planks are simply discarded or
destroyed, and the other situation (called world W3 in the preceding
discussion) is that in which no repair or replacement work takes place, but
the ship of Theseus is dismantled plank by plank and later reconstructed.
Common sense tells us that there is just one ship in the first of these
situations, which undergoes repair and replacement of parts, and that in the
second situation there is just one ship, which undergoes dismantling and
reassembly. Given the only x and y principle we must say similarly that in
the situation Hobbes describes there is a ship which undergoes repair and
replacement of parts and a ship which undergoes dismantling and
reassembly. These two ships must share their origin and an initial part of
their history, i.e. they must for a time be multiple occupants of a space–time
region but distinct because they subsequently diverge.
If this is so, sameness of original matter together with plan and time of
origin cannot be a sufficient condition of artefact identity within a world.
But then it follows that it cannot be a sufficient condition of transworld
identity either. For the single ship in the first situation cannot be identical
with the two ships in W2, i.e. Hobbes’ situation; and the single ship in the
first situation cannot be the single ship in the second situation (since the
single ship in the first situation is the repaired ship in Hobbes’ situation and
the single ship in the second situation is the distinct disassembled and
reassembled ship in Hobbes’ situation).
However we describe the ship of Theseus example, then, if we accept a
sufficiency of origin principle for identity which allows us to infer Kripke’s
‘necessity of origin’ conclusion in his footnote 56 we come into conflict
with fairly strong common-sense intuitions.
The argument for the necessity of origin is Kripke’s most explicit
argument for an essentialist thesis. But he also argues in the following
footnote for the necessity of constitution or original substantial ‘make up’.
The question he asks is whether the table originally made from wood could
originally have been made of anything other than wood. He argues:
Obviously this question is related to the necessity of the origin of the table from a given block of
wood and whether that block, too, is essentially wood (even wood of a particular kind). Thus it is
ordinarily impossible to imagine the table made from any other substance other than the one of which
it is actually made without going back through the entire history of the universe, a mind-boggling
feat.
(1980: 115, n. 57)

The argument here is brief but appears to rely on an appeal to the necessity
of origin. Cases in which it is plausible to say that an artefact is originally
made of completely different matter in two circumstances also suggest that
it is plausible that an artefact might be originally made of different kinds of
matter in different circumstances. Even if the necessity of origin is allowed
the argument is not convincing unless it is assumed that if a wooden table is
originally made of a block of wood that block is both essentially originally
wooden and could not have changed over time from being wood to being
some other kind of substance. Otherwise, to imagine the table in fact made
of the block originally made of a different kind of substance does not
require going back over the entire history of the universe but only
imagining a change in the block in the course of its existence.
At any rate, it is clear that the argument for the necessity of constitution
Kripke suggests requires a substantial essentialist commitment as a premiss.
It cannot, nor of course does Kripke ever suggest it can, itself convince
someone with no prior commitment to essentialism.
The necessity of identity and distinctness figure prominently (as
premisses) in Kripke’s last lecture (and implicitly, as we have seen, in his
arguments for the necessity of origin). The argument for the necessity of
identity is given briefly in the Preface to Naming and Necessity and more
carefully in the much earlier ‘Identity and Necessity’. In the Preface he
writes (p. 3):
Waiving fussy considerations derived from the fact that x need not have necessary existence it was
clear from (x)&(x = x) and Leibniz’s Law that identity is an ‘internal’ relation: (x)(y)(x = y ! &x = y).
(What pairs < x, y > could be counterexamples? Not pairs of distinct objects, for then the antecedent
is false; nor any pair of an object and itself, for then the consequent is true.)

The argument is given at slightly greater length in ‘Identity and Necessity’

(1971: 134–5):
First, the law of substitutivity of identity says that, for any objects x and y, if x is identical to y, then
if x has a certain property F, so does y:

(1) (x)(y)(x = y → (Fx → Fy))

On the other hand, every object surely is necessarily self-identical:

(2) (x)◻(x = x)

But:

(3) (x)(y)(x = y → [◻(x = x) → ◻(x = y)])

is a substitution instance of (1), the substitutivity law. From (2) and (3), we can conclude that, for
every x and y, if x equals y, it is necessary that x equals y:

(4) (x)(y)(x = y → ◻(x = y)).

This is because the clause ◻(x = x) of the conditional drops out because it is known to be true.
(1971: 135–6)

This is an argument for the thesis that an identity pair (any pair of objects x
and y such that x = y) is necessarily an identity pair; similarly Kripke could
argue that a distinctness pair is necessarily a distinctness pair. Nothing
follows about the necessary truth of any statement, much less the necessary
a posteriori truth of any statement. Kripke emphasizes the necessary
distinctions himself:
We must distinguish three distinct theses: (i) that identical objects are necessarily identical; (ii) that
identity statements between rigid designators are necessary; (iii) that identity statements between
what we call ‘names’ in actual language are necessary.
(1980: 4)

The necessity of identity and distinctness are thus metaphysical claims on a

par with the necessity of origin. They, like it, are claims about the world,
not about language, albeit the argument for them is a logical proof.
 Is Kripke’s proof correct? It may be questioned on Quinean grounds
whether ‘(x)◻(x = x)’ expresses an objective truth unless read as equivalent
to the claim that ‘Everything is self-identical’ is an analytic truth. The proof
will also be challenged by defenders of Lewisean counterpart theory who
want to allow that a thing can have more than one counterpart in a single
possible world – making it true that an identity pair has a de re possibility of
being a non-identity pair (Lewis 1986: 263). This is consistent because the
counterpart-theoretic translation (when the box is replaced by a universal
quantifier over possible worlds and quantifiers ranging over counterparts)
turns out not to be a substitution instance of Leibniz’s Law at all. Whether it
is a mark against counterpart theory or a mark in its favour that it allows
this is debatable.
To sum up this section: Kripke asserts various essentialist claims and
offers brief proofs of some of them. Some of these proofs are disputable,
though the proofs of the necessity of identity and distinctness, if Quinean
doubts about the intelligibility of quantifying into modal contexts are put
aside, can only be challenged by a Lewisean counterpart theorist. A crucial
point to understand, however, as Kripke himself makes totally clear, is that
it does not follow from the truth of these essentialist claims, if allowed, that
there are any necessary truths, much less any necessary a posteriori truths.
Nor does it follow from the truth of any ‘anti-essentialist’ claim, i.e. any
claim to the effect that some property is a merely contingent property of
some object, that there are any contingent a priori truths. Just as we have to
distinguish the three claims (i) that identical objects are necessarily
identical, (ii) that identity statements between rigid designators are
necessary and (iii) that identity statements between what we call ‘names’ in
actual language are necessary, so also we have to distinguish, for example,
the claims (i) that biologically related objects are necessarily biologically
related, (ii) that statements of biological relation made using rigid
designators are necessary and (iii) that statements of biological relation
using what we call ‘names’ in actual language are necessary. If it is the case
that any child is necessarily the child of her actual parents then given
appropriate rigid designators ‘a’ and ‘b’ understanding rigidity as Kripke
does, ‘if a exists a is a child of b’ is a necessary truth and, if ‘Elizabeth II’
and ‘Elizabeth Bowes-Lyon’ are rigid designators so understood, ‘if
Elizabeth II exists Elizabeth II is a child of Elizabeth Bowes-Lyon’ is a
necessary truth. But an essentialist has no obligation to recognize the
category of Kripkean rigid designators, much less to classify any actual
proper name as a Kripkean rigid designator.
In short, essentialism is one thing, the admission of necessary a posteriori
truth another. The topic of necessary a posteriori truth and contingent a
priori truth and Kripke’s routes thereto are our next business.
The Necessary a Posteriori and the Contingent a
Priori

A truth is a priori if it can be known a priori, a posteriori if its truth can be

known, but not a priori. Any unknowable truth is neither a priori nor a
posteriori. Given that all mathematical truths are necessary, if there are any
unknowable mathematical truths they therefore count as necessary but
neither a priori nor a posteriori. However, given such an unknowable
mathematical truth N, its disjunction with an only a posteriori knowable
contingent truth C will count as a necessary a posteriori truth. But it will not
be knowable that it is necessary. The examples of the necessary a posteriori
that Kripke has in mind are not like this; they can be known not merely to
be true, but to be necessary, they are knowably necessary a posteriori truths.
The puzzle of the Kripkean necessary a posteriori is therefore not merely
that there can be necessary truths which are not a priori knowable, nor that
there can be necessary truths which are knowable but only a posteriori, but
that there can be necessary truths which are knowably necessary but
knowable only a posteriori.
The category of the contingent a priori is more straightforward to
characterize; it consists of truths which are contingent but knowable a
priori.
Kripke thinks that his examples of knowably necessary a posteriori truths
can be constructed by a mechanism whose principal components are rigid
designation and (instances of) general essentialist principles like the
necessity of origin. He thinks that his examples of contingent a priori truths
can be constructed by a mechanism whose principal components are rigid
designation and (instances of) general anti-essentialist principles, for
example, that the length of a bar of metal at a time is a contingent property
of it. In both cases, then, rigid designation is crucial. The arguments for the
contingent a priori are seemingly a mirror image of the arguments for the
necessary a posteriori. But Kripke is more tentative in his endorsement of
them, as have been subsequent supporters of the Kripkean necessary a
posteriori.
In the case of the necessary a posteriori sceptics of Kripke’s account of
rigid designation (for example, the Dummettian wide-scopist) will have an
alternative explanation of the phenomena that will enable them to deny the
necessity of the purported examples. For those convinced that Kripke’s
account of rigidity is correct this is not an option, but then Kripke’s
argument that the examples are instances of a posteriori truth becomes
problematic. In the case of the putative examples of the contingent a priori
sceptics about Kripke’s account of rigid designation will deny the
contingency of the examples. For those convinced by Kripke’s account this
is not an option, but then Kripke’s argument that the examples are instances
of a priori truth becomes problematic.
We will begin by looking at Kripke’s account of the mechanism by which
knowably necessary a posteriori truth can be manufactured from essential
truth, and the crucial role of rigid designation. Then we will be able to see
how scepticism about Kripke’s account of rigid designation must create
doubt about the manufacturing process. Next we will look at Kripke’s
arguments for the a posteriori status of his examples, and the worries about
this someone not sceptical about Kripke’s concept of rigid designation
might have.
Then we turn to the contingent a priori in order to see how Kripke
constructs his examples and how a sceptic about Kripkean rigidity might
resist their classification as contingent and a non-sceptic might query their
status as a priori.
Kripke claims that all his examples of the necessary a posteriori have the
characteristic of mathematical statements that we can know a priori that if
they are true they are necessarily true. In ‘Identity and Necessity’ he
suggests the schema:
illustrating with the reading of ‘P’ as ‘this table is not made of ice’. One
knows by a priori philosophical analysis, he says, that if this table is not
made of ice it is necessarily not made of ice. One knows by a posteriori
empirical investigation that this table is not made of ice. Hence the
conclusion that it is necessary that the table is not made of ice is known a
posteriori.
This makes it plausible that Kripke thinks of all his examples of the
necessary a posteriori as established by a priori reasoning from a posteriori
truths and general essentialist principles which can be known a priori to be
true and necessary. In more detail, for example, in this instance this will
presumably be: ‘Any table is such that if it is not made of ice it is
necessarily not made of ice (a universal generalization which is an a priori
knowable essentialist premiss). So if this table is not made of ice it is
necessarily not made of ice (instance of the universal claim). This table is
not made of ice (empirical observation). Therefore this table is necessarily
not made of ice (by modus ponens).’
Similarly, the conclusion that Hesperus and Phosphorus are necessarily
identical can be deduced as follows: ‘Any objects which are identical are
necessarily identical, i.e. any pair is such that if it is an identity pair it is
necessarily an identity pair (a universal generalization which is an a priori
knowable essentialist premiss). So if Hesperus is Phosphorus then
necessarily Hesperus is Phosphorus (instance of universal claim). Hesperus
is Phosphorus (empirical observation). Therefore necessarily Hesperus is
Phosphorus (via modus ponens).’
On this account Kripke’s route to the necessary a posteriori is the same
both in the case in which the destination is a purportedly necessary a
posteriori truth ascribing an essential property to an object (or an essential
relation to a pair of objects) and in the case in which the destination is a
purportedly necessary a posteriori identity statement; identity is just a
special case of an essential relation, what Kripke calls an internal relation
(what is special about the case of identity is that the general essentialist
principles to which he appeals is a logical truth of quantified modal logic,
whereas Kripke does not think that this is true of the other general
essentialist principles in question).
This explanation of Kripke’s schema brings out the crucial role of rigid
designation. The inference remains valid in both cases if accidental
designators are employed to instantiate the universal claim, but all that
follows then is a statement ascribing an essential property to an object (or
an essential relation to a pair of objects). One cannot conclude that anything
is a necessary truth.
So from the essentialist premiss that any table which is not made of ice is
necessarily not made of ice and the empirical statement that Kripke’s
favourite table is not made of ice we can validly infer that Kripke’s
favourite table is such that it is necessarily not made of ice and hence that
on one reading (on which the description is given wide scope) ‘It is
necessarily true that Kripke’s favourite table is not made of ice’ is true. But
one cannot infer that ‘It is necessarily true that Kripke’s favourite table is
not made of ice’ is unambiguously true, or that ‘Kripke’s favourite table is
not made of ice’ is or expresses a necessary truth.
However, given that ‘this table’ is a rigid designator the conclusion of the
argument ‘this table is necessarily such that it is not made of ice’ is
equivalent to the small-scope reading of ‘it is necessarily true that this table
is not made of ice’ and hence it follows also that ‘it is a necessary truth that
this table is not made of ice’ is unambiguously true and ‘this table is not
made of ice’ is or expresses a necessary truth.
Thus on Kripke’s story if we may take it that it is an essential property of
any table which is not made of ice that it is not made of ice, if this table is
in fact not made of ice it is a necessary truth, but only a posteriori
knowable, given that ‘this table’ is a rigid designator, that this table, if it
exists, is not made of ice. The same conclusion follows for any rigid
designator in place of ‘this table’.
 But this conclusion can be resisted by sceptics about Kripke’s concept of
rigid designation. I mentioned Dummettian widescopism at the beginning
of this section, but I will illustrate the point here by reference to a Fregean
scepticism about rigidity. As we saw, one way a Fregean can respond to
Kripke’s arguments is by saying that the simple designators which Kripke
classifies as rigid behave as they do because they are governed by the
convention that they must not be used in modal contexts to refer to their
indirect references, their ordinary senses. On this Fregean view a term ‘T’,
introduced as a name for this table, which functions as a rigid designator,
will be such that ‘T is not made of ice’ will express contingently true
thoughts in most people’s mouths, different ones in different mouths, just as
‘this table is not made of ice’ will, as uttered by different people
demonstrating the table. But if I prefix the operator ‘necessarily’ I cannot be
understood as ascribing a property to the thought I express with that
sentence. Rather, I have to be understood as saying of the direct reference of
the name in my mouth, i.e. this table, that it is necessarily not made of ice.
And given that not being made of ice is an essential property of the table –
which is something the Fregean need not deny – what I say will be true. But
nothing will be both necessarily true and knowably so but only a posteriori.
Consider another example. If I fix the reference of the rigid designator ‘N’
by the description ‘the number of planets’ I can find out, but only a
posteriori, that N is odd. I know a priori that if any number is odd it is
necessarily odd, let us suppose. So I can infer that if N is odd it is
necessarily odd. Whence I can infer that N is necessarily odd. Hence I can
come to know, but only a posteriori, that I will speak the unambiguous truth
if I say ‘Necessarily, N is odd’. But on the Fregean account it does not
follow that there is any a posteriori necessary truth expressible in the form
‘X is odd’. The thought that I as reference-fixer express when I say ‘N is
odd’ is just the thought that the number of planets is odd, which is not a
necessary truth.
The Fregean sceptic about Kripkean rigidity will respond in the same
way to Kripke’s putative examples of necessarily true but only a posteriori
knowable identities, such as ‘Hesperus is Phosphorus’. It follows from the
Fregean story that in most people’s mouths this will express contingently
true thoughts, different thoughts in different mouths. The reason for the
appearance of necessity is that when the operator ‘necessarily’ is prefixed
the resultant statement is one to which the convention that one must not use
a proper name to refer to the sense one associates with it applies. Hence,
‘Necessarily, Hesperus is Phosphorus’ must be understood, not as ascribing
necessity to the thought that a free-standing occurrence of ‘Hesperus is
Phosphorus’ would express, but as saying of the objects Hesperus and
Phosphorus that they are necessarily identical, and given that an identity
pair is necessarily an identity pair (as Kripke’s proof of the necessity of
identity shows so long as we set aside Quinean scruples about quantifying
into modal contexts and the possibility of a counterpart theoretic
interpretation of modality) this is true. But it does not follow, the Fregean
will say, that there is any thought which is both knowably necessarily true
and knowable only a posteriori.
From the Fregean perspective sketched, then, the transition Kripke makes
from his essentialist claims to his contentions about the necessary a
posteriori is illegitimate. The former need not be denied, nor need it be
denied that they are knowable a priori. But, the Fregean can say, what
Kripke identifies as a statement which expresses a knowably necessary a
posteriori truth is merely one expressing an only a posteriori knowable
contingent thought in which an object picked out by an expression subject
to the convention specified, which proscribes reference to sense in modal
contexts, is ascribed an essential property. Kripke asserts (1971: 180): ‘the
notion of essential properties can be maintained only by distinguishing
between the notions of a priori and necessary truth’. But this is what the
Fregean denies. According to him the distinctively Kripkean category of
metaphysical necessity, knowably necessary a posteriori truth, is an illusion.
Of course, this conclusion depends on the highly contentious Fregean
explanation of the rigidity phenomenon. The point is just that it is entirely
consistent for someone who gives such an explanation to accept
essentialism but deny the necessity of the Kripkean examples of the
necessary a posteriori.
 The other side of the coin is that for those who accept Kripkean rigidity
the a posteriori status of the examples becomes contentious. Kripke argues
that true identity statements containing proper names whose references are
fixed differently, such as ‘Hesperus is Phosphorus’, state a posteriori as well
as necessary truths in the last few pages of Lecture II of Naming and
Necessity. He then generalizes his account in Lecture III to explain how
other necessary truths can be knowable only a posteriori. But the initial
story can be questioned, and the worries about it are ones that Kripke’sown
‘A Puzzle about Belief’ brings into prominence.
Kripke argues that it is knowable only a posteriori that Hesperus is
Phosphorus as follows:
The evidence I have before I know that Hesperus is Phosphorus is that I see a certain heavenly body
in the evening and call it ‘Hesperus’, and in the morning and call it ‘Phosphorus’. … There certainly
is a possible other world in which a man should have seen a certain star at a certain point in the
evening sky and called it ‘Hesperus’ and a certain star in the morning and called it ‘Phosphorus’ and
… have found out by empirical investigation that he names two different … heavenly bodies. … and
so it’s true that given the evidence that someone has antecedently to his empirical investigation, he
can be placed in a sense in exactly the same situation, that is a qualitatively identical epistemic
situation, and call two heavenly bodies ‘Hesperus’ and ‘Phosphorus’ without their being identical. So
in that sense it might have turned out either way. Not that it might have turned out either way as to
Hesperus’s being Phosphorus … that couldn’t have turned out any other way, in a sense. But one …
could have had qualitatively identical evidence and concluded that ‘Hesperus’ and ‘Phosphorus’
named two different objects … So … we do not know a priori that Hesperus is Phosphorus …
because we could have had evidence qualitatively indistinguishable from the evidence we have and
determined the reference of the two names by the positions of two planets in the sky, without the
planets being the same.
(1980: 103–4)

Kripke returns to this issue and restates and generalizes this argument in the
third lecture:
Now in spite of the arguments I gave before for the distinction between necessary and a priori truth,
the notion of a posteriori necessary truth may still be somewhat puzzling. Someone may … argue as
follows. You have admitted … that this table might have turned out to be made from ice from water
from the Thames. I gather that Hesperus might have turned out not to be Phosphorus. What can you
mean when you say that such eventualities are impossible? If Hesperus might have turned out not to
be Phosphorus, then Hesperus might not have been Phosphorus.
(1980: 140–1)
Kripke responds to this by invoking the idea of a qualitatively identical
epistemic situation:
The objector is correct … that if I hold that this table could not have been made of ice, … I must …
hold that it could not have turned out to be made of ice; it could have turned out that P entails that P
could have been the case. What, then, does the intuition that the table might have turned out to have
been made of ice … amount to? I think it means simply that … I (or some conscious being) could
have been qualititatively in the same epistemic situation that actually obtains, I could have the same
sensory experience that I in fact have about a table, which was made of ice.
(1980: 141–2)

He sums up:
The general answer to the objector can be stated, then, as follows: any necessary truth, whether a
priori or a posteriori, could not have turned out otherwise. In the case of some necessary a posteriori
truths, however, … under appropriate qualitatively identical evidential situations an appropriate
corresponding qualitative statement might have been false. … The inaccurate statement that
Hesperus might have turned out not to be Phosphorus should be replaced by the true contingency
[that] two distinct bodies might have occupied, in the morning and the evening, respectively, the very
positions actually occupied by Hesperus–Phosphorus–Venus.
(1980: 142–43)

These passages contain Kripke’s response to the sceptic about the necessary
a posteriori who is prepared to accept that the statements in question are
necessary (unlike the Fregean sceptic mentioned earlier) but wonders why
they should be called a posteriori. Kripke’s argument is that we can only
know a posteriori that ‘Hesperus’ and ‘Phosphorus’ are coreferential, and
hence that ‘Hesperus is Phosphorus’ expresses a truth. So we can only
know a posteriori that Hesperus is Phosphorus. Or again, that we can only
know a posteriori that the heavenly body visible in the evening is the
heavenly body visible in the morning. So we can only know a posteriori
that Hesperus is Phosphorus. In the case of the table the argument is that
Kripke can only know a posteriori that the table in front of him is not made
of ice (he can only know a posteriori what he expresses by saying ‘the table
in front of me is not made of ice’) and hence he can only know a posteriori
that what he expresses by saying ‘this table is not made of ice’ (or, if he
introduces ‘T’ as a rigid designator of the table in front of him, ‘T is not
made of ice’) is the case. Thus it is necessary but only knowable a posteriori
that Hesperus is Phosphorus and that this table (T) is not made of ice.
The reason for doubting that this is an adequate explanation of the a
posteriori status of these statements is brought out by Kripke himself in his
paper ‘A Puzzle about Belief’ (1979). If we take seriously Kripke’s thought
that the description originally associated by a speaker with a name is
inessential to his understanding of it, so that once introduced by a
description a name can later be used by the original baptizer or other
speakers in association with no, or quite different, descriptions, it becomes
quite unclear why evidence for the associated contingent statement
(whether this is thought of as the metalinguistic statement that ‘Hesperus’
and ‘Phosphorus’ are coreferential or the statement that the heavenly body
seen in the night sky is the heavenly body seen in the morning sky) is
necessary for justified belief that Hesperus is Phosphorus. Before acquiring
evidence that ‘Hesperus’ is coreferential with ‘Phosphorus’ a competent
user of the names justifiably believes that Hesperus is Hesperus. The
question is why he does not also already justifiably believe that Hesperus is
Phosphorus despite not believing that ‘Hesperus’ and ‘Phosphorus’ are
coreferential. Kripke’s story of puzzling Pierre brings out the urgency of
this question. Pierre, a monolingual French speaker living in Paris believes
that London is pretty (on the basis of the pictures he has seen), a belief he
expresses by saying in French ‘Londres est jolie’. He is then kidnapped and
transported to an ugly area of London, where he has to learn English by the
direct method. Disliking his surroundings and never thinking that the city
he has been taken to is the pretty city in the pictures he does not assent to
‘London is pretty’ and repeatedly assents to ‘London is ugly’, whilst often
repeating to himself ‘Londres est jolie’. As Kripke asserts, it is undeniable
that Pierre continues to believe after his kidnap that London is pretty and is
justified in doing so, although he does not believe that the sentence ‘London
is pretty’ is true and would not be justified in doing so. Nor, of course, does
he believe, nor would he be justified in believing, that the city he is living in
is pretty. Granted that it is a necessary truth that Hesperus is Phosphorus,
therefore, it is unclear how, given his own views, Kripke’s argument in the
quoted passage can establish that it is knowable only a posteriori. What
Kripke appears to be assuming is that in order to be justified in believing
that Hesperus is Phosphorus someone must be justified in believing that the
sentence ‘Hesperus is Phosphorus’ is true, or that the heavenly body in the
evening sky is the heavenly body in the morning sky. But the unfortunate
Pierre is justified in believing that London is pretty despite not justifiably
believing either that the sentence ‘London is pretty’ is true or that the city in
which he is now living is pretty (even though the only reference-fixing
description he has for ‘London’– as opposed to ‘Londres’– is the
description ‘the city in which I am now living’). So why should a justified
belief that Hesperus is Phosphorus have to fulfil these further requirements?
To see the point from a slightly different perspective, consider the
proposition that London is Londres. Before his kidnap Pierre believes that
London is London, a proposition he expresses by saying ‘Londres est
Londres’. After his kidnap he retains this belief. After his kidnap he is not
prepared to say (in an English– French word-salad) ‘Londres is London’
(anymore than he is prepared to say ‘London is jolie’) and in fact would
insist that what this expresses is false. But the necessary truth he
nonetheless expresses with this sentence, according to Kripke, is the
necessary truth that London is London, which he believes. So he believes
that London is Londres, just as he believes that London is pretty. That he is
not prepared to say in the French–English word-salad ‘London is Londres’
is irrelevant. Just so, then, a sceptic about the a posteriori status of
‘Hesperus is Phosphorus’ will say, the fact that someone is not prepared to
say ‘Hesperus is Phosphorus’ is not evidence that he does not believe that
Hesperus is Phosphorus.
The same point can be made about Kripke’s other purported examples of
necessary a posteriori truth. Given that it is a necessary truth that this table
is not made of ice (or, if ‘T’ is a rigid designator whose reference is fixed by
Kripke by the description ‘the table now in front of me’, that it is a
necessary truth that T is not made of ice), Kripke’s argument that it is not
knowable a priori that T is not made of ice (if it exists) is dubious. Given
that the table is not made of ice a correct description of it is ‘the table here
which is not made of ice’. If a rigid designator ‘T*’ is introduced with its
reference fixed by this description it is, according to Kripke’s account,
knowable a priori that T* is not made of ice if it exists. So if it can also be
known a priori that T* is T and if the possibility of a priori knowledge is
closed under logical implication, it can be known a priori that T, if it exists,
is not made of ice. Setting aside this argument, the dubiousness of Kripke’s
claim that it can only be known a posteriori that T (this table) is not made of
ice, if it exists, is once again highlighted by direct comparison with the case
of puzzling Pierre. Although Pierre’s evidential situation after his kidnap is
such that he is not warranted in assenting to the English sentence ‘London
is pretty’, he is still justified in believing that London is pretty since he is
justified in believing to be true the sentence ‘Londres est jolie’, which he
understands and says that London is pretty. Just so, a sceptic about Kripke’s
argument can say, although my evidential situation when I confront the
table does not warrant me in assenting to the English sentence ‘T is not
made of ice if it exists’ I can still be justified without that evidence in
believing that T is not made of ice if it exists since I can be justified in
believing that T* is not made of ice if it exists. I understand the sentence
‘T* is not made of ice if it exists’ and justifiably believe that what it says is
true; since it says that T is not made of ice if it exists I am justified in
believing that T is not made of ice if it exists despite having no justification
for believing that the sentence ‘T is not made of ice if it exists’ is true or
that the table in front of me if it exists is not made of ice. (The same
considerations can be used to cast doubt on the idea that non-identities are
necessary a posteriori truths. The sentence ‘If they exist Venus and Mars are
different entities’ can only be known to be true empirically, but if I fix the
reference of the rigid designator ‘Mars*’ as ‘the planet which is Mars and is
other than Venus’ I can know a priori that if they exist Venus and Mars* are
different entities. The considerations sketched above can now be brought to
bear.)
Consider also the example of the number of planets, which can be
explained more simply since no existential assumption need be made. If I
introduce ‘N’, with its reference fixed by the description ‘the number of
planets’, in ignorance of how many planets there are, according to Kripke
what I express by ‘N is greater than 7’ is a necessary truth. But accepting
this claim the question can be pressed why it is knowable only a posteriori.
It is true that without empirical justification I cannot know that ‘N is greater
than 7’ expresses a truth or that the number of planets is greater than 7. But
I can know a priori that 9 is greater than 7. If I can also know a priori that 9
= N, I can know a priori that N is greater than 7 if possible a priori
knowledge is closed under logical implication. Independently of this
argument the dubiousness of the Kripkean claim that it can only be known a
posteriori that N is greater than 7 is once again highlighted by direct
comparison with the case of Pierre. Pierre is justified in believing that
London is pretty although he has no evidential warrant for believing to be
true the sentence ‘London is pretty’.Heisso justified since he is justified in
believing to be true the sentence ‘Londres est jolie’, which he understands
and which says that London is pretty. Just so, a sceptic about Kripke’s
argument can say, although my evidential situation when I view the heavens
does not warrant me in assenting to ‘N is greater than 7’ I can still be
justified without that evidence in believing that N is greater than 7 since I
can be justified in believing that 9 is greater than 7.
In short, there is a unresolved tension in Naming and Necessity which
Kripke himself draws attention to in ‘A Puzzle about Belief’. It is tempting
to think that if proper names are, as Kripke argues, de jure rigid designators,
grasp of which is not essentially tied to any descriptive content, they must
be substitutable salva veritate not only in modal contexts but also in
epistemic contexts. As Kripke puts it, adopting some terminology from
Geach (1962), such contexts, it appears, must be ‘Shakespearian’ (alluding
to ‘a rose by any other name’) if proper names are de jure rigid designators
without descriptive content. But if so, his putative examples of the
necessary a posteriori no longer qualify as such. On the other hand, if
proper names are not substitutable salva veritate in epistemic contexts, the
most readily available explanation is some version of the Frege–Russell
view, according to which the examples cease to be examples of the
necessary a posteriori, because they cease to be examples of necessary
truth. (The defender of the Frege–Russell view will accept that it is correct
to say, on the basis of what he says in French, that Pierre believes that
London is pretty but deny that just because I assent to ‘9>7’ and so can be
said to believe that 9 > 7 I can also be said to believe that N is greater than
7; he can explain the difference by the correspondence between the senses
of ‘London’ and ‘Londres’, which is lacking in the case of ‘9’ and ‘N’.)
Defenders of the Kripkean necessary a posteriori need to fight on two
fronts: to deny that epistemic contexts are Shakespearian, whilst
maintaining that modal contexts are.
The challenge the Kripkean faces in explaining the necessary a posteriori
can be put another way. The task for him is to provide an explanation,
without appeal to resources only available to the descriptivist, of the
informativeness of true identity statements of the form ‘a = b’ which cannot
also be deployed to ‘explain’ the informativeness of true, but
uninformative, identity statements of the form ‘a= a’. Suppose I use
‘Cicero’ to refer to Cicero, but without any associated identifying
description and under the misapprehension that Cicero was a Greek
admiral. I am in the same position with respect to ‘Tully’. I am open to the
thought that there is more than one Greek admiral called ‘Cicero’ and
equally open to the thought that there is more than one Greek admiral called
‘Tully’. Kripke needs to explain, without using descriptivist resources, how
in this situation, which according to him is absolutely typical, I can be
informed by the statement ‘Cicero is Tully’, but not by the statement
‘Cicero is Cicero’. Equivalently, he needs to explain how the information
provided to me by a statement of the form ‘Cicero is F’ can be different
from that provided to me by a statement of the form ‘Tully is F’, and to
explain this in a way that does not entail that successive token statements of
the form ‘Cicero is F’ will provide me with different pieces of information.
Kripke’s putative examples of contingent a priori truth also crucially
involve rigid designators. The general form of a candidate statement is
‘aisF’ or ‘if a exists a is F’ where ‘a’ is a rigid designator whose reference is
fixed by the description ‘the F’ and being F is an accidental property of the
thing designated by ‘a’. Kripke’s most discussed example is the length of
the standard metre rod, but he thinks that the same mechanism by which a
priori contingent knowledge can be generated in this case can be used to
generate it in other cases also: for example ‘Jack the Ripper’ (where the
reference is fixed by the description ‘the man, whoever he was, who
committed all those murders, or most of them’) and ‘Neptune’ (‘the planet
if there is one causing such and such perturbations in the orbits of these
other planets’). Other philosophers have introduced their own examples.
One of the most well-known is Gareth Evans’s example of Julius, the
inventor of the zip. Dummett suggests the example of St Anne, the
reference-fixing description being ‘the mother of the Blessed Virgin Mary’.
Kripke argues for the contingent a priori status of his examples as
follows. Suppose we fix the reference of ‘one metre’ by means of the
stipulation: one metre is to be the length of stick S at time t. Stick S is an
ordinary rod chosen as the standard of length. It is not a necessary property
of stick S that it is one metre long at t, since it is an accidental feature of
stick S that it is the length it is at t (equivalently, it is a contingent property
of the length one metre that it is the length of stick S at t). So the
description ‘the length of stick S at t’ is an accidental designator of the
length. The length of stick S at t might not have been the length of stick S at
t. So one metre might not have been the length of stick S at t. So it is not a
necessary truth that one metre is the length of stick S at t, i.e. the statement
‘It is a necessary truth that one metre is the length of stick S at t’ is false if
the description ‘the length of stick S at t’ is read as having narrow scope.
Consequently ‘one metre is the length of stick S at t’ is not a necessary
truth. However, on the basis of the reference-fixing stipulation the stipulator
can know a priori that one metre is the length of stick S at t (if stick S exists
at t), automatically, without further investigation, or at least he can do so if
someone who stipulates that ‘squatchelor’ is to apply to all and only
unmarried men can know a priori that all and only squatchelors are
unmarried men. How can someone who stipulates that ‘squatchelor’ is to
apply to all and only unmarried men come to know that all and only
squatchelors are unmarried men? He can reason as follows: ‘All and only
those things to which “squatchelor” as-I-use-it-at-the-moment applies are
squatchelors. All and only those things to which “squatchelor” as-I-use-it-
at-the-moment applies are unmarried men. So all and only squatchelors are
unmarried men.’ Similarly Kripke’s stipulator can reason: ‘All and only
those things to which “a metre long” as-I-use-it-at-the-moment applies are
things which are a metre long. All and only those things to which “a metre
long” as-I-use-it-at-the-moment applies are the same in length as stick S at
t. So all and only those things which are a metre long are the same in length
as stick S at t.’ If the former reasoning provides a priori justification for the
truth that all and only squatchelors are unmarried men the latter provides a
priori justification for the truth that things a metre long are the same length
as S at t.
The argument generalizes. It applies whenever the reference of a rigid
designator is fixed by a description giving a contingent property of its
bearer. Dummett’s example of St Anne can be used to illustrate. Very little
is known of St Anne, and the reference can be taken to be fixed by the
description ‘the mother of the Virgin Mary’. St Anne might have died in
infancy, so she might not have been the mother of Mary, or a parent at all.
Consequently it is not a necessary truth that St Anne was the mother of
Mary nor a necessary truth that St Anne was a parent. On Kripke’s account,
then, neither ‘St Anne was the mother of Mary’ nor ‘St Anne was a parent’
is a necessary truth. But it is knowable a priori that if she existed St Anne
was the mother of Mary.
In these cases the referents are contingent existents so what Kripke
claims to be knowable a priori are conditional statements. But we can
appeal to the example of the rigid designator ‘N’ of the number of planets.
Since ‘the number of planets’ must have reference it can be known, on
Kripke’s account, completely a priori that the categorical statement that N is
the number of planets is true.
What is puzzling about the contingent a priori, Kripke suggests, can be
brought out as follows:
If someone fixes a meter as ‘the length of stick S at t’, then in some sense he knows a priori that the
length of stick S at t is one meter, even though he uses the statement to express a contingent truth.
But, merely by fixing a system of measurement, has he thereby learned some (contingent)
information about the world, some new fact that he did not know before? It seems plausible that in
some sense he did not, even though it is undeniably a contingent fact that S is one meter long.
(1980: 63)

Consider the case of ‘squatchelor’. Before I make the stipulation I know

that all and only unmarried men are unmarried men. After I make the
stipulation I know that I have made a certain stipulation and so know
something new about myself and about language. But I know nothing new
about the world apart from language. I know that squatchelors are
unmarried men. But this is just the truth that unmarried men are unmarried
men, something I knew all along.
Now consider the case of ‘N’. Before I introduce ‘N’ as a rigid
designator of the number of planets I know a priori the necessary truth that
the number of planets is the number of planets and I am either ignorant of
or know only by a posteriori means the contingent truth that 9 is the number
of planets. Afterwards I know that I have made a certain stipulation and so
know something new about myself and about language. But, according to
Kripke’s account, I have learned some fact about the world apart from
language that I did not know before. I have learned that N is the number of
planets. This is not the necessary truth that the number of planets is the
number of planets, which I knew before. Nor is it the contingent and only a
posteriori knowable truth that 9 is the number of planets, which, perhaps, I
also knew before or of which I was and am still ignorant. It is a contingent,
a priori knowable truth, or fact, about the world apart from language,
different from these. But what is this fact?
Scott Soames expresses the worry here, referring to the example of one
metre:
… consider the claim that if stick S exists at t, then its length is one meter … we know that [this
claim] cannot be the claim that if stick S exists at t, its length at t is the length of stick S at t. So what
… is supposed to be both contingent and a priori? … Perhaps it is the claim that if stick S exists at t,
then its length is this length.
(2003: 400)

Soames goes on to express the obvious doubt:

Prior to going through a little verbal ceremony, we are entirely ignorant of a certain entirely non-
linguistic, non-trivial empirical fact – namely that stick S if it exists is this long. We say a few
formulaic words … and, presto, we know the fact that we were completely ignorant of. … True belief
is not that easy to come by.
(2003: 411)

The sentiment is widely shared and Kripke’s examples of contingent a

priori truth are widely regarded as dubious.
Sceptics about Kripkean rigidity, like Dummett, say that in all these cases
what is known is indeed a priori, but is also necessary. Someone who before
making the stipulation knows that the length of stick S at t, if it exists, is the
length of stick S at t, or that the number of planets is the number of planets
knows nothing new about the world apart from language afterwards, but
just has a new way of expressing this old piece of knowledge (just as in the
case of ‘squatchelor’). The Fregean sceptic about Kripkean rigidity, for
example, can respond to his putative examples of the contingent a priori in
a way that exactly mirrors his response to Kripke’s examples of the
necessary a posteriori. Consider the example of the putatively contingent a
priori truth that St Anne was a parent if she existed. In the mouth of
someone who uses ‘St Anne’ with the sense of ‘the mother of Mary’ this
expresses an a priori knowable, necessarily true thought. The reason for the
appearance of contingency is that when the operator ‘Necessarily’ is
prefixed the result is a statement to which the convention applies that one
must not use a name to refer to the sense one associates with it. Hence one
can only be understood as speaking, not of the sense one associates with ‘St
Anne’ but of the person herself. But since she was only contingently a
parent, one’s statement has then got to be understood as saying something
false. But there is no thought here, which one expresses when one says ‘St
Anne, if she existed, was a parent’, that is both contingently true and
knowable a priori. From the Fregean perspective sketched the transition
Kripke makes from his claims about the contingency of the properties of the
things his examples concern to his contentions about the contingent a priori
is illegitimate. What Kripke identifies as a statement which expresses an a
priori contingent truth is merely one expressing an a priori knowable and
necessary thought in which an object picked out by an expression subject to
the convention specified, which proscribes reference to sense in modal
contexts, is ascribed a contingent property.
For those who accept Kripkean rigidity, on the other hand, the a priori
status of the examples becomes contentious. A reaction which many have
found plausible, expounded most famously by Donnellan (1979), is that
although ‘if stick S exists at t, the length of stick S at t is one metre’ does
express a contingent truth, it is not one that is knowable a priori by the
reference-fixer. All the reference-fixer knows a priori is that the sentence ‘If
stick S exists at t, the length of stick S at t is one metre’ expresses a truth.
But not knowing which contingent truth is expressed he does not know a
priori that if stick S exists at t then the length of stick S at t is one metre.
The idea is that if one simply stipulates, in complete ignorance of the
existence of stick S or, therefore, its length, that ‘one metre’ is to designate
the length of stick S at t, one cannot thereby come to have de re knowledge
of the length in question. But the knowledge that if stick S exists at t its
length at t is one metre is a piece of de re knowledge of the length, so one
does not come to have it via the stipulation.
Donnellan actually focuses on Kripke’s example of ‘Neptune’,
supposedly introduced by Leverrier to designate the planet hypothesized as
causing certain perturbations in the orbit of Uranus. In this case he argues
that what Leverrier knows after the stipulation is that the sentence ‘If the
planet causing such and such perturbations exists, Neptune is the planet
causing such and such perturbations’ is true. He does not know the
contingent truth that this sentence expresses since to know that would
require him to have de re knowledge of Neptune, and since he has no
acquaintance with it (it is a mere hypothesis) Leverrier has no such
knowledge. Indeed, he has not even a belief that if the planet causing such
and such perturbations exists, Neptune is the planet causing such and such
perturbations, since to have this belief is to have a de re belief about
Neptune and given his lack of acquaintance with Neptune he has no such de
re belief.
 Donnellan illustrates the distinction here between knowing that a
sentence is true and knowing the truth it expresses with a variety of
examples. The first is a case in which someone is told by someone he trusts,
and is reliable, that a sentence of German, a language he does not
understand, is true. He then believes and, if it is the case, knows, that the
sentence is true, but since he does not know what it means, he does not
know the truth it expresses. A second example is one in which I read in
Scienti fic American that the oblateness of Mars is 0.003. I have no idea
what oblateness is, but trusting the magazine I believe the sentence to be
true. But I do not believe the oblateness of Mars is 0.003. A third type of
example is one in which I hear and accept to be true a sentence containing
an indexical without having the contextual knowledge to determine the
reference of the indexical. I overhear a trustworthy and honest friend saying
‘That is mine’ but have no idea what he is pointing at.
In all these cases, Donnellan plausibly suggests, there is a gap between
knowing that the relevant sentence is true and knowing the truth it
expresses.
The difficulty, however, for Donnellan, is that all these cases seem to be
disanalogous to the examples of ‘Neptune’ and ‘one metre’ in precisely the
crucial respect. In all these cases it is plausible that there is a gap between
knowing the sentence to be true and knowing the truth the sentence
expresses, because what is required to bridge the gap, i.e. knowing what
truth the sentence expresses, is lacking. In the first and second cases what is
lacking is semantical knowledge – the speaker does not know German or
does not know what ‘oblateness’ means. In the case of the sentence
containing an indexical the speaker lacks the relevant context-specific
knowledge. But when Leverrier stipulated that ‘Neptune’ was to be the
name of the planet he hypothesized as the cause of certain perturbations in
the orbit of Uranus he did not lack semantic understanding –‘Neptune’ was
not already a term in public use; nor did he lack any relevant contextual
knowledge. In fact, Donnellan’s insistence in this case that there is a gap
between knowing the sentence to be true and knowing the truth it expresses
is implausible and goes against the way we ordinarily talk. As Hughes
(2004: 100) suggests, if after fixing the reference of ‘Neptune’ Leverrier
gives a talk to a group of astronomers about his research it will be natural to
report this by saying ‘Leverrier thinks that there is an eighth planet – he
calls it “Neptune”. He believes that Neptune is not only responsible for such
and such perturbations but also …’. Donnellan must say that this ascribes to
Leverrier a belief he does not have. There seems to be no good ground to
accept this. The example of the number of planets is even clearer if
anything. Having introduced ‘N’ via the description ‘the number of planets’
I surely know exactly what I mean by the sentence ‘N is the number of
planets’ and hence know not only that that sentence is true but that N is the
number of planets. But I do not know that 9 is the number of planets (unless
I did so already), and what I do clearly know, that the number of planets is
the number of planets, is a necessary truth, so the puzzle of the contingent a
priori status of my knowledge that N is the number of planets is unresolved
by Donnellan’sproposal.
But if these examples are accepted, as Kripke (somewhat tentatively)
suggests that they should be, as examples of genuine a priori knowledge of
non-metalinguistic contingent truths, there is once again a tension in his
views. For, to revert to the example of the number of planets, if it is
knowable a priori that N is the number of planets by the stipulator then
epistemic contexts are not Shakespearian, or else it is knowable a priori that
9 is the number of planets. Thus defenders of Kripke’s position on the
contingent a priori need to deny that epistemic contexts are Shakespearian
without maintaining that modal contexts are, a position most readily
explained by some version of the Frege–Russell view.
In ‘A Puzzle about Belief’ Kripke indicates his awareness of this tension
with respect to both the contingent a priori and the necessary a posteriori.
Talking of the Millian view to which he inclines, he writes:
According to Mill, a proper name is, so to speak, simply a name. … If a strict Millian view is correct,
and the linguistic function of a proper name is completely exhausted by the fact that it names its
bearer, it would appear that proper names of the same thing are everywhere interchangeable not only
salva veritate but also salva significatione … If Mill is right not only should ‘Cicero was lazy’ have
the same truth-value as ‘Tully was lazy’, but the two sentences should express the same proposition.
… If such a consequence of Mill’s view is accepted, it would seem to have further consequences
regarding ‘intensional’ contexts. … any simple sentence should retain its ‘modal value’ (necessary,
impossible, contingently true, or contingently false) when ‘Cicero’ is replaced by ‘Tully’ in one or
more places since such a replacement leaves the content of the sentence unaltered. … The situation
would seem to be similar with respect to contexts involving knowledge, belief and other epistemic
modalities. Whether a given subject believes something is true or false of such a subject no matter
how the belief is expressed; so if a proper name substitution does not change the content of a
sentence expressing a belief, coreferential proper names should be interchangeable in belief contexts.
Similar reasoning would hold for epistemic contexts … and contexts of epistemic necessity (‘Jones
knows a priori that …’)
(Kripke 1979: 104–5)

Kripke is noncommittal about the argument here sketched for the

Shakespearian character of belief contexts. He writes: ‘Philosophers have
often, basing themselves on Jones’ and similar cases, supposed that it goes
virtually without saying that belief contexts are not “Shakespearian”. I think
that at present, such a definite conclusion is unwarranted’ (1979: 136).
Mutatis mutandis, such a conclusion must be unwarranted for contexts of
epistemic necessity, for the Shakespearian character of which on a Millian
conception of names, as Kripke notes, similar arguments can be given. But,
as we have seen, if such contexts are Shakespearian Kripke’s contentions
that there are necessary a posteriori and contingent a priori truths cannot be
maintained. To take the example of the number of planets again, Kripke’s
claim is that it is a contingent a priori truth that N is the number of planets
and a contingent, but a posteriori truth that 9 is the number of planets. But if
‘it is a priori knowable that …’ is a Shakespearian context this distinction
cannot be made. Similarly, his contention is that it is a necessary a
posteriori truth that N is greater than 7 and a necessary but a priori truth that
9 is. But given that ‘it is a priori knowable that …’ is Shakespearian this
cannot be the case.
In short, a defender of Kripke’s division of metaphysical from epistemic
modality cannot sit on the fence with regard to the Shakespearian character
of contexts of epistemic necessity and possibility.
It remains that, putting this conclusion aside, it can still be accepted by
Kripke’s Fregean and Russellian opponents, as well as by his defenders,
that (i) a division can be drawn between essential and accidental properties,
(ii) that ‘It is a necessary truth that Hesperus is Phosphorus’ is
unambiguously true although empirical enquiry is needed to ascertain that it
is and (iii) that ‘It is a merely contingent truth that Hesperus, if it exists, is
visible in the night sky’ is unambiguously true although it can be
determined a priori that if Hesperus exists it is visible in the night sky (in
whatever sense of ‘a priori’ it can be determined a priori by our imagined
stipulator that squatchelors are unmarried men). But to accept these three
things is not yet to accept the distinctive claims of Naming and Necessity,
according to which ‘Hesperus is Phosphorus’ expresses a knowably
necessary a posteriori truth and ‘Hesperus, if it exists, is visible in the night
sky’ a contingent a priori one.
5
Extensions
Natural Kind Terms as Proper Names of Kinds

After discussing proper names of individuals Kripke extends his apparatus

to a discussion of what he calls names of natural kinds. These are terms for
natural phenomena, such as heat or light, terms for natural kinds of stuff,
such as gold or water and terms for natural kinds of thing, such as the tiger.
It is in this context that Kripke’s views come together with Putnam’s (1973,
1975), and his examples can be seen as illustrating some of the same points
as Putnam’s.
Kripke extends his discussion to the associated expressions of other
grammatical categories linked with these names, the adjective ‘hot’, the
relational predicate ‘hotter than’, the predicates ‘is (constituted of) gold
(water)’, and the predicate ‘is a tiger’. Thus, pursuing the analogy with
proper names, he arrives at conclusions about the necessary a posteriori
status of ‘theoretical identifications’, which include not only statements of
apparent identity of kinds, such as ‘water is H2O’, ‘Heat is the motion of
molecules’ and ‘Gold is the element with atomic number 79’, but also
apparent subject-predicate statements like ‘Light is a stream of photons’
and generalizations like ‘Cats are animals’ and ‘tigers are mammals’.
Moreover, he takes his view that identity claims like ‘Water is H2O’ and
‘Gold is the element with atomic number 79’ are necessary a posteriori
truths to entail that generalizations about particular samples (‘If something
is a sample of water it is composed of atoms with atomic number 79’) are
also necessary a posteriori truths. If such terms as ‘water’ and ‘H2O’ are
genuine rigid designators, then when scientific investigation reveals that
water is identical to H2O, as he thinks it does, what has been discovered is
the truth of a genuine identity expressed using rigid designators flanking the
sign for identity, and hence one that holds necessarily. Moreover, since the
proposition that something is a sample of water just in case it is a sample of
H2O follows from the proposition that water is H2O (just as the proposition
that someone is an inhabitant of Hesperus if and only if he is an inhabitant
of Phosphorus follows from the proposition that Hesperus is Phosphorus), it
is also necessary, if water is H2O. So the picture of science as merely
discovering truths about the actual world has to be replaced. Rather, the
results of scientific enquiry, insofar as they are discoveries about natural
kinds of phenomena, stuff and things, are genuinely necessary truths that
hold in any metaphysically possible world. As Kripke put it, ‘In general,
science attempts by investigating basic structural traits, to find the nature,
and thus the essence (in the philosophical sense) of’ natural kinds (1980:
138).
The key question in assessing Kripke’s discussion of natural kind terms is
whether the points of analogy he sees between them and proper names of
individuals justify these claims.
Kripke sums up his position in the following passage:
terms for natural kinds are much closer to proper names than … ordinarily suggested. The old term
‘common name’ is … quite appropriate for predicates … such as ‘cow’ or ‘tiger’. My considerations
also apply … to … mass terms such as ‘gold’, ‘water’. … It is interesting to compare my views to
those of Mill. Mill counts both predicates like ‘cow’, definite descriptions and proper names as
names. He says of ‘singular’ names that they are connotative if they are definite descriptions but non-
connotative if they are proper names. On the other hand, Mill says that all general names are
connotative. Such a predicate as ‘human being’ is defined as the conjunction of certain properties
which give necessary and sufficient conditions for humanity – rationality, animality and certain
physical features. The modern logical tradition, as represented by Frege and Russell, seems to hold
that Mill was wrong about singular names but right about general names. More recent philosophy has
followed suit, except that, in the case of both proper names and natural kind terms, it often replaces
the notion of defining properties by that of a cluster of properties. … My own view … regards Mill
as more-or-less right about ‘singular’ names but wrong about ‘general’ names. … Perhaps some
general names (‘foolish’, ‘fat’, ‘yellow’) express properties. In a significant sense, such general
names as ‘cow’ and ‘tiger’ do not. … ‘cow’ and ‘tiger’ are not short for the conjunction of properties
a dictionary would take to define them. … Whether science can discover empirically that certain
properties are necessary of cows, or of tigers, is another question, which I answer affirmatively.
(1980: 128)

Kripke’s views about natural kind terms are opposed to what we can call the
‘cluster of identifying marks’ theory, according to which such terms as
‘water’ are, in their primary usage, predicates whose meaning and extension
are fixed by the cluster of properties we regard as the criteria for their
application. This cluster of identifying marks theory of natural kind terms is
obviously similar to the cluster of identifying descriptions theory of proper
names and was held by much the same philosophers, e.g. Wittgenstein, as
Kripke emphasizes. It is because of this that Kripke thinks it appropriate to
describe his views by saying that he holds, like Mill, that proper names
have no connotation, but goes beyond Mill in extending the same
conclusion to natural kind terms. The question we have to keep carefully in
mind, then, is how far Kripke’s arguments against the cluster of identifying
marks theory of natural kind terms, construed as predicates, also qualify as
arguments against the cluster of identifying descriptions theory of natural
kind terms, construed as proper names of kinds. I shall conclude, not very
far.
Kripke proceeds against the cluster of identifying marks theory by way of
examples.
Consider the term ‘tiger’. According to the cluster of identifying marks
theory ‘x is a tiger’ means ‘x has such and such (superficially observable)
properties’, e.g. is large, striped, ferocious and cat-like in appearance. But,
Kripke argues, even if these were all the known properties of tigers, as
presumably they were for primitive man, it would not be the case that any
possible creature having these properties was a tiger, nor even that any
actual creature having these properties was a tiger. Suppose that in a
counterfactual situation there are creatures having the appearance of tigers
which are, in fact, not mammals at all, but very peculiar reptiles, then these
reptiles are not tigers because they have the wrong internal structure.
Suppose that, on another planet in the actual universe or in an unexplored
part of the globe, such creatures really do exist; still they are not tigers –
their discovery would not be the discovery that there were reptilian tigers as
well as mammalian tigers. If it is accepted that the tiger-like reptiles in the
counterfactual situation are not tigers, then the cluster of identifying marks
story does not give the meaning of ‘x is a tiger’, and if the tiger-like reptiles
would not be tigers even if they existed in the actual world, it cannot be that
this cluster even fixes the extension of ‘x is a tiger’, i.e. the cluster of
identifying marks story must be wrong both as a theory of meaning and as a
theory of reference- or extension-fixing for natural kind terms.
Moreover, Kripke argues, there might have been and might be tigers that
do not have all the properties involved in the cluster of identifying marks,
or indeed any of the properties in the cluster. A three-legged tiger is not a
contradictio in adjecto. It might be that no tigers have four legs (we are
deceived by an optical illusion). Indeed it might have been and might be
that tigers have none of the properties by which we originally identified
them.
Perhaps none are quadrupedal, none tawny yellow, none carniverous, and so on; all of these
properties turn out to be based on optical illusion or other errors … So the term ‘tiger’… does not
mark out a cluster concept in which most, but not all of the properties used to identify the kind must
be satisfied. On the contrary, possession of most of these properties need not be a necessary condition
for membership of the kind, nor need it be a sufficient condition.
(1980: 121)

Consider now the predicate ‘is water’. According to the cluster of

identifying marks theory of natural kind terms, the meaning and extension
of this predicate are determined by the properties we know to be
characteristic of samples of water. Hence, it is both necessarily true and
something we know a priori, that something is water if and only if it has
these properties. But, Kripke argues, it is not. Imagine a possible world (or
a distant but actual planet) in which the stuff in the seas, lakes and rivers is
not H2O but something with the chemical formula XYZ, which, however,
has the appearance, taste and texture of water and behaves in all respects
known to the average non-scientist like water (this is Putnam’s [1975] Twin
Earth example which elaborates Kripke’s remarks). Then, Kripke argues,
this stuff is not water because it is not H2O, just as fool’s gold is not gold.
This is so even if the planet on which XYZ occupies the seas and rivers is
not in another possible world but is a Twin Earth situated at the ‘other end’
of the actual universe.
If there were a substance, even actually, which had a completely different atomic structure from that
of water, but resembled water in these respects, would we say that some water wasn’tH2O? I think
not. We would say instead that just as there is a fool’s gold there could be a fool’s water.
 (1980: 128)

What fills the seas and rivers of this planet is not water even if its
inhabitants speak English and call it ‘water’ and despite the fact that any
non-scientist from Earth, if he were to visit Twin Earth, would be unable to
tell that the liquid in its seas and rivers was not water. Moreover, the stuff
on Twin Earth did not belong to the extension of the predicate ‘is water’ as
used on Earth even, say, 700 years ago, when the properties the average
non-scientist associates with that predicate now were the only properties
associated with it by anyone – non-scientist or scientist (alche-mist) alike –
even though, if anyone living on Earth 700 years ago had been carried off
by the Martians to Twin Earth, he would have found the stuff in the seas
and rivers to be exactly as he would have expected water to be.
Again, if Kripke is right about all this, the cluster of identifying marks
theory of natural kind terms is mistaken.
Now Kripke’s view of what is going on in these cases is as follows.
When we introduce a term like ‘water’ or ‘gold’ or ‘tiger’ into the language
we introduce it as a proper name of a kind. For example, ‘tiger’ is first
introduced by a reference-fixing definition of the following type: ‘Tigers
are the kind of fierce, cat-like, yellow and black striped creatures which live
in the jungle over there and occasionally steal and eat our goats and
infants’. This is analogous to fixing the reference of a man’s name by the
definition: ‘Tom Jones is the man presently living in the old house by the
river over there, who is always so rude and unsociable’. Kripke emphasizes
the likeness to the case of proper names of individuals:
In the case of proper names, the reference can be fixed in various ways. In an initial baptism it is
typically fixed by an ostension or description. Otherwise the reference is usually determined by a
chain, passing the name from link to link. The same observations hold for such a general term as
‘gold’. If we imagine a hypothetical baptism of the substance, we must imagine it picked out by some
such definition as ‘gold is the substance instantiated by the items over there, or at any rate, by almost
all of them’. … I believe that in general names for natural kinds … get their reference fixed in this
way, the substance is defined as the kind instantiated by (almost all) of the given sample. The
‘almost’ all qualification allows that some fool’s gold may be present in the sample. If the original
sample has a small number of deviant items, they will be rejected as not really gold. If, on the other
hand, the supposition that there is some uniform substance or kind in the initial sample proves more
radically in error, reactions can vary; sometimes we can declare that there are two kinds of gold,
sometimes we may drop the terms ‘gold’… the original samples get augmented by the discovery of
new items. … More important, the term may be passed from link to link exactly as in the case of
proper names, so that many who have seen little or no gold can still use the term. Their reference is
determined by a causal (historical) link.
(1980: 135–6)

On the Kripkean picture the term ‘gold’ acquired its reference in the way
we imagined above the name ‘Tom Jones’ to acquire its reference, or, to use
Kripke’s own example, in the way ‘Jack the Ripper’ acquired its reference.
‘Jack the Ripper’ was introduced by fixing its reference as: the man who
committed those murders, or at any rate, most of them. If evidence had
emerged that a few of the murders were committed by a different person (a
copycat killer) the reference would have remained the man who committed
most of the murders (or maybe the most gory, or those committed or
discovered first). The victims of the copycat killer would not have been
included in the history books as victims of Jack the Ripper. But reactions
might have varied if honours, as it were, had been divided. Be that as it
may, the name’s original reference was: that man, the man who committed
those murders, whoever he is – and it was understood that it was the job of
the experts (the detectives) to discover the identity. In the same way, ‘gold’,
on Kripke’s picture, was originally introduced just as a name for: that kind
of substance, the kind instantiated by those samples, whatever it is – and it
was understood that its nature was a matter for future discovery. So it was
entirely comprehensible to its first users that a substance absolutely
indistinguishable from gold by all the tests available to them for being gold
might not be gold but a different kind of stuff.
Moreover, since reference-fixing definitions do not give meaning, once
‘Tom Jones’, ‘Jack the Ripper’, ‘the tiger’ and ‘gold’ have been introduced
in this way, it is perfectly sensible to speculate about, say, a possible world
in which not Tom Jones, but someone else is the unsociable old man
presently living in the old house by the river or in which not tigers, but
another kind of creature are the fierce, cat-like, black and yellow striped
animals that steal and eat our goats and infants. And just as in each case
what is being named is not something that can only exist where and when it
is named, it is perfectly sensible to speculate about actual places and times
where it is considerably different from how it is here and now: places and
times where Tom Jones is not rude and unsociable and tigers are not fierce,
black and yellow striped carnivores. Similarly, it is possible to speculate
that there are, or might be, other rude and unsociable old men around or
other kinds of black and yellow striped carnivores.
It is with this kind of comparison of names for natural kinds with proper
names of individuals that Kripke suggests that we can make sense of the
counterexamples which refute the cluster of identifying marks theory of
natural kind terms. To be Tom Jones is to be the same man as the man so
baptized; to be a tiger to be the same kind of thing as the kind so baptized.
Being the same kind of thing is no more a matter of known superficial
appearance than being the same man is. What it is a matter of is somewhat
vague; all the same, according to Kripke it is a notion of which we all have
some intuitive grasp and which is essential to our understanding of natural
kind terms. According to the cluster of identifying marks theory someone
confronted by a superficially tiger-like creature ought to be able, if he
understands the term ‘tiger’, to determine whether it is really a tiger by
examining it. On this theory it makes no sense for him to say, after
examining it, that it might be a tiger but that he remains uncertain. But,
according to Kripke’s account, such a remark is perfectly comprehensible.
Someone for whom the reference of tiger is fixed by the description ‘the
kind of ferocious striped carnivore of which my village lives in terror’ will
know that there could be many kinds of creature of the same superficial
appearance as tigers and so, without a knowledge of the internal structure of
tigers – say, whether they are mammalian or reptilian – ought not to be
prepared to say with certainty of any particular creature of such an
appearance that it must be a tiger. Similarly, someone transported to Twin
Earth and confronted with a sample of XYZ can perfectly intelligibly, on
Kripke’s account, withhold judgement whether the stuff is water even
whilst acknowledging that it has all the properties he knows water on Earth
to possess. For he can point out that the stuff might not be of the same kind
as the samples on Earth despite sharing these properties. Another example.
Suppose a chemist comes across a previously unknown substance
somewhere on Earth which he calls ‘nemonite’; he has just begun to
explore its properties when he is abducted by the Martians and transported
to Twin Earth. He is then confronted by a substance of chemical formula
UVW and asked whether it is nemonite. He carries out the same few tests
on it that he was able to carry out on nemonite on Earth and it behaves in
the same way. Obviously, it would be absurd for him to announce on this
basis that the stuff on Twin Earth is nemonite, for he knows that it might
turn out, on further investigation, that nemonite and the stuff on Twin Earth
are quite different kinds of stuff. Equally obviously, he will no more be able
to say whether the stuff on Twin Earth is nemonite even after he has
become fully familiar with its properties, since however extensive his
knowledge of it, it cannot possibly enable him to say whether it is of the
same kind as the stuff on Earth which he was only beginning to investigate.
Kripke’s account of how natural kind terms function can make sense of
all this; the cluster of identifying marks theory cannot. But it is important to
see what feature of Kripke’s account makes the stories intelligible. It is not
that natural kind terms are construed as rigid designators of kinds. This can
play no role in explaining the intelligibility of the native’s refusal to say
whether the creature he has examined is a tiger, nor the refusal of the
chemist abducted by the Martians to say whether the stuff on distant Twin
Earth is nemonite. This is obviously so, since the distinction between rigid
and non-rigid designators is a matter of how the reference is determined in
counterfactual, not actual, situations. What seems to make these cases
intelligible is the thesis that natural kind terms are names of kinds together
with our knowledge, which is part of our grasp of the concept of a kind, that
super-ficial similarity in external appearance is not a sufficient condition of
sameness of biological kind and that substances which to the inexperienced
eye are very similar can be of distinct material kinds or be of the same kind
despite being superficially very different. (It might be, as Dummett (1973:
143) says, that in the case of animal kinds descent from a common stock is
also important.)
 Let us recap. Kripke writes as if his account of how natural kind terms
are introduced is inconsistent with Frege–Russell descriptivism and takes
the examples he gives in support of his position as more grist to his ‘more
Millian than Mill’ account of naming. But although these cases are
problematic for the cluster of identifying marks account they are not
problematic for Frege–Russell descriptivism. The commitments of the
cluster of identifying marks account go far beyond the commitments of
descriptivism. If it were applied to proper names of individuals the cluster
of identifying marks account would amount to the claim that the reference
of a name of an individual is determined solely by qualitative features of the
individual available to inspection by anyone who uses the name. But this is
not entailed by descriptivism. If I introduce a name of a man by such a
description as ‘the tall fat man over there now’, I am not committed to the
claim that there are no circumstances in which that man is short and thin or
no circumstances in which other men are tall and fat. What I intend when I
fix the reference of a name in this way is that it is henceforth used correctly
only if it is used to refer to the same man, and it is part of our understanding
of the notion of the same man that the same man may be superficially
different at different times and places and different men may be
superficially the same. Similarly, as Kripke’s work, along with Putnam’s,
has brought out, the notion of sameness of kind allows that superficially
similar things may be of different kinds and superficially different things of
the same kind. It is for this reason that the cluster of identifying marks
account is mistaken. But it does not follow that descriptivism is incorrect.
An account which is compatible with Kripke’s discussion (even if it is
not compatible with Kripke’s intention) is that when natural kind terms are
introduced they are introduced by descriptions of the form: the kind of thing
of which these (ostensively introduced) samples are instances (as Kripke
says). Subsequently, users intend to refer to what the first users referred to –
they use the term deferentially as descriptivism allows. Kripke does not, in
fact, attempt to sketch out in the case of natural kind terms, as he does in
the case of proper names of individuals, a counterexample to the
descriptivist view. Rather, he writes as if he has already done so in his
account of how natural kind terms are introduced. But this account only
shows that natural kind terms are not equivalent to descriptions free of
names, demonstratives and other indexicals, just as proper names of people
are not. Kripke’s story in fact requires, in the case both of natural kind
terms and proper names of people, that when the term is first introduced an
identifying description must be available to the baptizer, in conformity with
descriptivism. In the case of proper names of people Kripke goes on to
argue (a) that even as used by the initial baptizer whilst he retains the
identifying description in question the proper name is not equivalent to any
description because the name is a rigid designator and the description is not
and (b) that subsequently users of the name will lack any identifying
description of the sort descriptivism requires. In the case of natural kind
terms the corresponding arguments are simply not given, so it is open to the
descriptivist to accept the conclusion Kripke explicitly draws, that the
cluster of identifying marks theory is incorrect, whilst maintaining his
position.
In short, it appears that the main thrust of Kripke’s explicit arguments
about natural kind terms is that they cannot be adequately defined in terms
of indexical-free descriptions known to their users – i.e. clusters of
identifying marks – an important thesis but not one inconsistent with
descriptivism. To drive this point home recall that if Kripke’s attack on the
description theory of proper names, as defined by the theses (1)–(6) he lists,
is successful, then it establishes externalism in the sense defined by Putnam
–‘meanings just ain’t in the head’. But Kripke’s arguments against the
cluster of identifying marks theory of natural kind terms leave internalism
unscathed. Of course, the same is also true of Putnam’s own Twin Earth
arguments. As arguments for externalism they are simply unsound. Even we
accept, with Putnam, that ‘water’ on Earth denotes H2O and ‘water’ on
Twin Earth denotes XYZ, externalism is not established since it is just false
that Oscar and Twin Oscar are narrowly psychologically identical in the
sense of ‘narrow psychology’ Putnam himself defines (and needs, if his
argument is to have the revolutionary implications he perceives). Oscar
believes, say, of himself, that is, of Oscar, that he has a headache. Twin
Oscar does not. This is a narrow psychological state in Putnam’s sense since
Oscar can be in it even if nothing else exists (though no one else can). So
Oscar and Twin Oscar differ in their narrow psychological states (Noonan
1981, 1985). Similarly, Oscar intends to use ‘water’ to speak of the
substance he believes that he, that is, Oscar, bathes in. Twin Oscar does not.
Again, this intention is a narrow psychological state, so again Oscar and
Twin Oscar differ in their narrow psychological states. Hence Putnam’s
argument does not establish that meanings are not, in the sense intended, ‘in
the head’. Of course, it does establish that there is an indexical component
in our concept of water – for me ‘water’ is defined as ‘the liquid stuff I am
in contact with, in the seas and rivers near to me’ (or perhaps I need the past
tense ‘water is the liquid stuff I have been in contact with throughout my
life, in the seas and rivers near to me’ so that when I am kidnapped by the
Martians and transported to Twin Earth I speak falsely when I say ‘this
liquid I am now drinking is water’). But this is not something inconsistent
with internalism, i.e. descriptivism as Kripke understands it, but a form of
it.
This is not to say that Kripke’s alternative to the cluster of identifying
marks account, even when not viewed as a rival to descriptivism, is
unproblematic. The intuition that ‘gold’ or ‘tiger’ might have been
introduced in a primitive community to name a kind that members of the
community could not distinguish, given their limited knowledge, from a
kind instantiated elsewhere is highly plausible. In fact, there are two
plausible intuitions here. The first is that a primitive community in Bengal
might introduce the term ‘tiger’ for the kind of large, dangerous, striped
creature that comes in the night and attacks them, whilst recognizing that
there might well be other kinds of large, dangerous, striped creatures that
they could not distinguish from those they call ‘tigers’– since they have
never caught and killed a tiger or even got close to one. The second is that
the community could employ a notion of sameness of kind whilst
recognizing that their knowledge of and tests for sameness of kind are so
limited that pairs of things they could not at all tell apart even on the closest
examination might be of different kinds, i.e. might not fall in the extension
of the relational predicate they employ which we translate as ‘is of the same
kind as’. However, what seems most readily to make sense of these
intuitions is the thought that the term could have been introduced to name
the narrowest, the most uniform, kind instantiated by the samples, so that
even if the stuff (thinking of the ‘gold’ example) elsewhere (on Twin Earth)
were indistinguishable from that on Earth by all the means available to the
community, they could still make sense of the idea of differences they could
not discern which made the difference. And, indeed, it is only charitable to
assume that some such restriction must be intended by Kripke since kinds
form a hierarchy, so to talk of ‘the kind’ instantiated by the samples, unless
elliptical, is to fail to refer. If Kripke does intend such a restriction it fits
well with our intuitions about XYZ on Twin Earth. However, it does not fit
well with all our intuitions. It is, of course, particularly difficult to see how
the restriction can give the right results in the case of biological kinds, since
every individual organism is different, genetically and in its history, from
every other. Yet the members of a primitive community could presumably,
on Kripke’s story, pick out the kind Tiger by reference to a single tiger: ‘the
kind of thing (animal?) of which that is an example’, or by reference to a
female tiger and its cubs. In the case of chemical kinds the problem is still
present, if less obvious. Gold naturally occurs in one isotope, so any
baptismal sample for ‘gold’ on Earth consists entirely of one isotope, but it
seems that the term ‘gold’ could have been introduced in the way Kripke
sketches by a primitive community without its extension being composed
only of samples of the naturally occurring isotope, as a restriction to the
narrowest kind instantiated by the samples would require. Similarly, if all
our baptismal samples for ‘water’ had happened to be H2O, then, given the
restriction, the term ‘water’ could not have been introduced in the way
suggested, and at the same time have been given an extension including
samples of heavy water. The restriction does not even allow for all of
Kripke’s own intuitions. He remarks that if this substance (H2O) can take
another form – such as the polywater allegedly discovered in the Soviet
Union with very different identifying marks from that of what we now call
water – this is a form of water because it is the same substance (1980: 129).
 An alternative way of elaborating, or modifying, Kripke’s proposal,
which might give intuitively acceptable results, would be to propose that
the introducers of a natural kind term might themselves use it deferentially
– deferentially to future users in their community. This is suggested by
Putnam’s statement that ‘the key point is that the relation sameLis a
theoretical relation: whether something is or is not the same liquid as this
may take an indeterminate amount of scientific investigation to determine’
(1973: 702). Putnam’s account, in outline, is that the term ‘water’ gets its
reference fixed as follows: speakers latch on to certain actual samples and
apply the term ‘water’ to them and their use of ‘water’ accords with the
following rule: ‘for all possible x, x is a sample of water if and only if x is
of the same kind as the actual samples’– where the same kind relation
(sameL) istobe determined by scientific theory. The contrast here is with,
e.g., the definition of a ‘paediatrician’ as a doctor specializing in the
treatment of children. A tempting way to understand this proposal and
contrast is that users of ‘water’ use it deferentially to future users because
they fix its reference as ‘stuff of the same kind as these samples’, and they
use the relational expression ‘of the same kind as’ deferentially to future
users in the community. So the communities on Earth and distant Twin
Earth might refer respectively to H2O and XYZ when they use the term
‘water’ even before any chemistry is known, because the introducers of the
term on Earth intend its reference to be that which future users in their
community would pick out as ‘water’– and, by contrast, the users of the
term on Twin Earth intend exactly the same. This proposal at least provides
a response to the prima facie puzzle about how the chemically ignorant
speakers on Earth and Twin Earth, who are unable to distinguish H2O and
XYZ by any means available to them, can grasp a notion of sameness of
kind which distinguishes them, and it is a proposal in accord with Putnam’s
notion of deferential reference, or division of linguistic labour – thought of
as extending across generations. The suggestion is that the supposition that
the chemically ignorant users of ‘water’ on Earth and Twin Earth must
make if they are to be referring to different kinds, is that they have a
possibly imperfect grasp of a notion, i.e. that of sameness of kind, of which
other future users will know more.
This is vague. But the point I wish to emphasize is that it is not merely if
Kripke’s account of natural kind terms is interpreted within a descriptivist
framework that it is necessary to explain how the initial ignorant users can
secure reference to a unique kind; the same is true if Kripke’s account is
read, as he intends, as opposed to descriptivism. Kripke’s proposal that
natural kind terms be understood as analogous to proper names of
individuals requires that the initial baptism contains an identifying
reference to a kind, just as an introduction of a proper name requires an
identifying reference to an individual.
So far I have written as if the intuitions to which Kripke appeals in his
attack on the cluster of identifying marks theory cannot be contested. Of
course, this is not so. If we in fact discovered a Twin Earth on which there
was a substance super-ficially like water with a different chemical
composition, what would we say? What would we say about the extension
of ‘water’ as used by our chemically ignorant ancestors if such a discovery
were now made? Kripke and Putnam write as if it is clear what our
reactions would be. But others are not convinced. Lewis writes:
Like any up-to-date philosopher of 1955, I think ‘water’ is a cluster concept. Among the conditions
in the cluster are: it is liquid, it is colourless, it is odourless, it supports life. But, pace the philosopher
of 1955, there is a lot more to the cluster than that. Another condition in the cluster is: it is a natural
kind. Another is indexical: it is abundant hereabouts. Another is metalinguistic: many call it ‘water’.
When we hear that XYZ off on Twin Earth fits many of the conditions but not all, we are in a state of
semantic indecision about whether it deserves the name ‘water’. … When in a state of semantic
indecision, we are often glad to go either way and accommodate our own usage to the whim of our
conversational partners. … So if some philosopher, call him Schmutnam, invites us to join him in
saying that the water on Twin Earth differs in chemical composition from the water here, we will
happily follow his lead. And if another philosopher, Putnam … invites us to say that the stuff on
Twin Earth is not water … we will just as happily follow his lead. We should have followed Putnam’s
lead only for the duration of that conversation, then lapsed back into our accommodating state of
indecision. But, sad to say, we thought that instead of playing along with a whim, we were settling a
question once and for all.
(1994: 424)
If Lewis is right Twin Earth cases pose no threat to internalism or
descriptivism (though the mere fact that they are unclear suffices to refute
the cluster of identifying marks theory of the 1950s philosopher). But, as
we have seen, they pose no threat to these positions anyway. It is an
empirical question what our linguistic dispositions are, but whether they
incline us to withhold ‘water’ from XYZ, or apply it, or to dither, the
sophisticated Frege– Russell internalist descriptivist Kripke intends to
refute in Naming and Necessity need not worry.
The Necessity of Theoretical Identifications

According to Kripke just as there are necessary a posteriori truths about

individuals, there are necessary a posteriori truths about kinds. Some of
these have the form of identity statements (‘Water is H2O’), others have the
form of singular predications (‘lightning is an electrical discharge’), others
have the form of universal generalizations (‘cats are animals’).
Kripke is relatively inexplicit about how the necessary a posteriori status
of his examples is to be established. But it is clear that he thinks that it is
crucial that natural kind terms, like proper names of individuals, are rigid
designators. But how exactly does this help to explain the necessary a
posteriori status of theoretical identifications of the sorts listed above?
In his earlier ‘Identity and Necessity’ Kripke gives an argument which
straightforwardly extends his argument for the necessary a posteriori truth
of identity statements containing two rigid designators of the same
individual. Kripke has just been discussing the statement that heat is the
motion of molecules. He writes:
To state the view succinctly: we use both the terms ‘heat’ and ‘the motion of molecules’ as rigid
designators for a certain external phenomenon. Since heat is in fact the motion of molecules, and the
designators are rigid, by the argument I have given here, it is going to be necessary that heat is the
motion of molecules.
(1971: 140)

In Naming and Necessity itself there are similar passages. ‘Theoretical

identities’, Kripke says, ‘according to the conception I advocate, are
generally identities involving two rigid designators, and therefore are
examples of the necessary a posteriori’ (1980: 140) and ‘[w]e have seen
above that since “heat” and “molecular motion” are both rigid designators,
the identification of the phenomenon they name is necessary’ (1980: 148).
The proposal is that just as we can argue that ‘Hesperus is Phosphorus’
expresses a necessary truth, despite being a posteriori, so we can argue that
‘heat is molecular motion’ or ‘water is H2O’ expresses a necessary a
posteriori truth. But the matter is not so straightforward.
In the former case the reasoning appeals to the necessity of identity,
which we know to be true a priori, together with the empirical truth that
Hesperus is Phosphorus. So we argue ‘If things are identical it is necessary
that they are. Hesperus is Phosphorus. So Hesperus and Phosphorus are
necessarily identical’. We can equally well argue ‘If things are identical it is
necessary that they are. The planet visible in the night sky is the planet
visible in the morning sky. So the planet visible in the night sky and the
planet visible in the morning sky are necessarily identical’. But given that
‘Hesperus’ and ‘Phosphorus’ are rigid designators the conclusion of the
former piece of reasoning is equivalent to ‘it is necessary that Hesperus is
Phosphorus’ (the widescope reading of the modal claim is equivalent to the
narrow scope), and, if Kripke’s explanation of rigidity is correct, ‘It is
necessary that Hesperus is Phosphorus’ is true if and only if ‘Hesperus is
Phosphorus’ expresses a necessary truth. The reasoning to the conclusion
that ‘Heat is the motion of molecules’ expresses a necessary truth, Kripke
suggests, is entirely parallel. But the necessity of ‘Heat is the motion of
molecules’ yields the natural kind essentialism which Kripke advertises
only if it entails the coextensiveness in all possible worlds of the predicate
‘is hot’ and ‘contains moving molecules’, i.e. only if it entails the necessary
truth of ‘something is hot if and only if it contains moving molecules’, and
indeed of the interesting phenomenological discovery ‘when it is hotter the
molecules are moving faster’ (1980: 129). Similarly, the necessary truth of
‘water is H2O’ is of interest only if it entails the necessary coextensiveness
of the predicates ‘is a sample of water’ and ‘is composed of H2O
molecules’. So the argument for the necessity of ‘heat is molecular motion’
must be extended. However, it seems straightforward to do so. Since it is, if
Kripke is right, a necessary truth that Hesperus is Phosphorus it is a
necessary truth that any inhabitant of Hesperus is an inhabitant of
Phosphorus. Mutatis mutandis, if it is a necessary truth that heat is the
motion of molecules it is a necessary truth that any instance of heat is an
instance of molecular motion. So if it is necessary that something is hot if
and only if it is an instance of heat and necessary that something contains
molecules in motion just in case it is an instance of molecular motion we
can conclude that it is necessary that something is hot just in case it
contains molecules in motion.
But now the worry is that it is not a straightforward empirical discovery
that heat is molecular motion when this is understood as an identity claim,
as it is a straightforward empirical discovery that Hesperus is Phosphorus.
If the ground for asserting this identity is that the predicates ‘is hot’ and
‘contains molecules in motion’ are necessarily coextensive the argument for
the necessary identity of heat and molecular motion, unlike that for the
necessary identity of Hesperus and Phosphorus, is question-begging. But, if
the ground for the identity is just the actual coextensiveness of the
predicates, an opponent of Kripke can simply resist the claim. He can
accept the principle of the necessity of identity and acknowledge that ‘heat’
and ‘molecular motion’ are rigid designators, but deny that they are
designators of a single kind, rather than merely, like the kind terms ‘creature
with a heart’ and ‘creature with kidneys’, designators of two actually
coextensive kinds. Of course, given the coextensiveness of ‘creature with a
heart’ and ‘creature with kidneys’, the class of creatures with a heart is
identical with the class of creatures with kidneys. So if I introduce rigid
designators ‘H’ and ‘K’ for these classes I can infer, using the necessity of
identity, that necessarily H is K, and hence that necessarily any member of
H is a member of K. But this is no more a necessary a posteriori scientific
discovery of the kind Kripke is interested in than the fact that necessarily,
any inhabitant of Hesperus is an inhabitant of Phosphorus, or, using ‘IH’
and ‘IP’ as rigid designators of the class of Hesperus inhabitants and the
class of Phosphorus inhabitants respectively, that necessarily, IH is IP and
necessarily, any member of IH is a member of IP.
There is another concern about extending Kripke’s argument for the
necessity of identity statements involving proper names of individuals to
theoretical identity statements like ‘Heat is molecular motion’ and ‘Water is
H2O’. In one way, the latter are obviously not analogous to the former since
‘molecular motion’ and ‘H2O’ are not, like ‘Hesperus’, simple symbols.
The correct comparison is therefore not with ‘Hesperus is Phosphorus’ but
with ‘Elizabeth II is the fusion of sperm S and ovum O’, taking these to be
the actual sperm and ovum from which the Queen developed. Or we can
compare ‘heat is molecular motion’ with the example used before, with ‘N’
thought of as having its reference fixed by the description ‘the number of
planets’: ‘N is the square of 3’. Of course, examples of identification
statements involving simple names of natural kinds are perfectly
conceivable; maybe ‘furze is gorse’ is one and perhaps ‘Brontosaurus is
Apatosaurus’ is another. (Both the latter terms were introduced by O.C.
Marsh. Marsh supposed that the specimens he used to fix the references of
the two terms were not from the same genus because they differed so
greatly in size. Later Elmer Riggs discovered that he was wrong.) But
examples of this kind are of no interest to Kripke since they do not provide
instances of scientific discovery that can be thought of as discoveries of
essential properties. The appropriate comparison for Kripke’s purposes is
therefore with the identity statement about Elizabeth II. But if so, to
conclude that it is a necessary truth that water is H2O Kripke needs the
premiss that ‘H2O’ is a rigid designator, just as he needs the premiss that
‘the product of the fusion of sperm S and ovum O’ is a rigid designator to
conclude that ‘Elizabeth II is the product of the fusion of sperm S and ovum
O’ is a necessary truth. However, to know that requires knowing that H2O is
essentially H2O. So the argument from the principle of the necessity of
identity cannot get Kripke the conclusion he wants without prior knowledge
of the essential properties of natural kinds.
In fact, that the argument from the principle of the necessity of identity
cannot be the only argument Kripke needs to establish his natural kind
essentialism is obvious from the fact that many of his examples of
necessary a posteriori truths about kinds are not identity statements, or even
biconditionals, at all, but one-way conditionals like ‘cats are animals’ and
‘tigers are mammals’.
The more appropriate comparison to draw to understand Kripke’s
reasoning is thus with the more general mechanism he sketches by which a
posteriori knowledge of the necessary possession of properties can be
acquired. I can come to know that necessarily P by inference if I know that
P and that if P then necessarily P. In the cases of necessary a posteriori truth
that interest him, that P is something I can only know a posteriori and ‘if P
then necessarily P’ is knowable a priori by philosophical analysis, hence, as
Kripke says in the Appendix to Naming and Necessity:
All the cases of the necessary a posteriori advocated in the text have the special character attributed
to mathematical statements: Philosophical analysis tells us that they cannot be contingently true, so
any empirical knowledge of their truth is automatically empirical knowledge that they are necessary.
This characterization applies, in particular, to cases of identity statements and of essence.
(1980: 159)

As already noted, the general form of argument by which Kripke argues for
the necessary truth of ‘Hesperus is Phosphorus’ is a special case of this
reasoning. I discover empirically, what I can only know a posteriori, that
Hesperus is Phosphorus. I acquire by philosophical analysis the knowledge
that if Hesperus is Phosphorus it is a necessary truth that it is and I infer that
it is a necessary truth. The knowledge that, if Hesperus is Phosphorus, it is a
necessary truth that it is results from universal instantiation of the principle
of the necessity of identity. Thus knowledge of a general essentialist
principle is required for knowledge of a particular necessary a posteriori
truth of identity that it is necessary.
Entirely in parallel, Kripke thinks, if being made of molecules is an
essential property of Kripke I can come to know that it is a necessary truth
that Kripke, if he exists, is made of molecules by ascertaining a posteriori
that he is, and hence that if he exists, he is – something I can only know a
posteriori, inferring the conclusion by modus ponens using the conditional,
which I can know a priori to be true, that if Kripke is made of molecules if
he exists it is a necessary truth that he is if he exists. And I can know this
conditional about Kripke in particular to be true a priori, if it is, because I
can know a priori that, say, any material object made up of molecules is
essentially so. Thus, knowledge of a general essentialist principle about
individuals is required for knowledge of a particular necessary a posteriori
property ascription that it is necessary. The route to the conclusion involves
universal instantiation followed by modus ponens. The indispensability of
universal instantiation makes it clear that what primarily comes to be
known is the possession of an essential property by an object. The
reasoning remains valid if ‘Kripke’ is replaced by ‘the author of NN’, but
the conclusion then inferred would not then be transformable into ‘it is a
necessary truth that if he exists the author of NN is made of molecules’.
Similarly, in the case in which I fix the reference of ‘N’ by ‘the number of
planets’, I can infer that it is a necessary truth that N is odd, appealing to the
general principle that if any number is odd it is necessarily odd. Thus the
inference remains valid if ‘N’ is replaced by ‘the number of planets’, but in
this case we cannot arrive at the conclusion that it is a necessary truth that
the number of planets is odd.
In the same way, Kripke thinks that I can come to know that all tigers, i.e.
all instances of the kind Tiger, are mammals. I can know by a posteriori
investigation that all actual instances of the kind are mammals. And I can
know by philosophical analysis that if all instances of the kind are
mammals the kind is such that necessarily all instances of it are mammals.
Thus I can infer that the kind Tiger is such that necessarily all instances of it
are mammals. Given that ‘the kind Tiger’ is a rigid designator this can be
transformed into ‘It is a necessary truth that all instances of the kind Tiger
are mammals’. Given that ‘is a tiger’ means ‘is an instance of the kind
Tiger’, it follows that it is a necessary truth that all tigers are mammals. But
I can know by philosophical analysis that if all instances of the kind Tiger
are mammals then the kind Tiger is necessarily such that all instances of it
are mammals only because I can know by philosophical analysis, for
example, the general truth that any biological kind is such that if every
instance of it is a mammal that kind is necessarily such that every instance
of it is a mammal. Again, this is a general essentialist principle without
which my reasoning cannot proceed, but this time it is one about kinds.
Consequently, just as in the case of knowledge of necessary a posteriori
truths about individuals, the knowledge of necessary a posteriori truths
about kinds in which Kripke is interested can be achieved via the
mechanism he suggests only if we have a priori knowledge of other general
essentialist principles, in addition to a priori knowledge of the general
principle of the necessity of identity. But Kripke never argues in detail, or
indeed at all, except in the case of the necessity of individual origin, for any
such general essentialist principles either governing individuals or kinds.
He simply appeals to our intuitions.
Nevertheless, in the case of kinds, it is at least suggested by the text how
an argument for the general essentialist principles Kripke employs might be
constructed. At one point Kripke recalls the example of the standard metre:
The definition [‘gold is the substance instantiated by the items over there, or at any rate, by almost all
of them’] … express[es] an a priori truth, in the same sense as (and with the same qualifications as)
‘one meter = the length of S’:it fixes a reference. I believe that, in general, terms for natural kind …
get their reference fixed in this way …
(1980: 135–6)

In the case of the definition of ‘one metre’ as the length of stick S at t,

Kripke emphasizes the contingency of the property by which the length is
picked out and hence the apparently contingent, albeit a priori, status of the
statement that one metre is the length of stick S at t, if it exists. Similarly, if
we define ‘the kind Tiger’ as the kind of creature of which this individual
animal is an instance, it is contingent that the kind Tiger does have this as
an instance, but it is apparently a priori that the kind Tiger has this as an
instance if it exists. Kripke emphasizes in other cases that the reference of a
natural kind term is typically fixed by a contingent property of the kind:
although ‘heat’ is a rigid designator, the reference of that designator was determined by an accidental
property of the referent, namely the property of producing in us the sensation S.
(1980: 152)

In all these cases, then, just as in the case of the standard metre, if Kripke is
right, there are not only necessary a posteriori truths, but contingent a priori
truths, about the kind.
But what are the necessary a posteriori truths about one metre,
comparable to ‘heat is molecular motion’ or ‘the tiger is a mammal’? It
appears plausible that one such necessary a posteriori truth is ‘A metre is a
greater length than a yard’ (Kripke (1980: 76) suggests that the reference-
fixing description for ‘one yard’ is ‘the distance when the arm of King
Henry I of England was outstretched from the tip of his finger to his nose’).
Given that I have fixed the reference of ‘one metre’ as the length of stick S
at t I can certainly not know this without knowing the length of stick S at t,
which I can do only by measuring it. But once I have measured stick S, I
can infer that it is a necessary truth that one metre is a greater length than a
yard, and thus that anything, in any possible circumstance, which is one
metre is longer than anything, in that or any other circumstance, which is a
yard. It appears that this is analogous to the reasoning by which Kripke
thinks that I can come to know, but only a posteriori, that it is a necessary
truth that the Tiger is a mammal, and thus that necessarily all tigers are
mammals. I can come to know by empirical investigation that the instance
of the kind (tiger Tim, say) by which I have fixed the reference of the kind
term ‘the Tiger’ is a mammal. Consequently I can infer that it is a necessary
truth that the Tiger is a mammalian kind and thus that any creature in any
possible circumstance which is a tiger is a mammal.
My inference that it is a necessary truth that one metre is a greater length
than a yard appeals to two propositions which I know a priori. First, that
any possible or actual things that are the same in length are such that if
either is greater than a yard in length the other is also. That is, that if
something in one possible circumstance is the same in length as something
(else) in the same or a different possible circumstance then, if the first thing,
in the first circumstance, is longer than a yard, the same is true of the
second in the second circumstance. This proposition cannot be stated
without quantifying over merely possible things or making reference to
possible circumstances. It has a counter-factual content greater than the
content of the claim that in fact things that are the same in length are such
that if one is greater than a yard in length the other is also, and its content is
not captured by the modal claim that it is necessarily true that if things are
the same in length and one is longer than a yard then the second is also.
Nevertheless, it does appear to be something knowable a priori, because
analytic, by anyone who grasps the concept of sameness of length and the
Imperial system of length measurement. The second proposition my
inference rests on is that two actual or possible concrete objects are the
same in length just in case the length of the first is the length of the second.
This is an instance of an abstraction principle, comparable to the criterion of
identity for numbers Frege gives: the number of Fs is the number of Gs if
and only if there is a one–one correspondence between the Fs and the Gs, or
the principle by which Frege introduces the concept of a direction in order
to provide a simpler analogue of this: the direction of line a is the direction
of line b if and only if a is parallel to b (Grundlagen 1968: 74). It introduces
discourse about lengths conceived of as abstract objects. And, like Frege’s
criterion of identity for numbers and the direction principle, on its natural
understanding it too, despite having counterfactual content, is knowable a
priori. Putting these propositions together I can infer that the abstract object,
the length one metre, is such that if some concrete object (a fortiori, every
concrete object) the length of which it is is longer than a yard, then that
length is such that necessarily every concrete object of which it is the length
is longer than a yard. This is an instance of the general essentialist principle
about lengths that any length is such that if some concrete object (a fortiori
every concrete object) the length of which it is is longer than a yard, then
that length is such that necessarily, every concrete object of which it is the
length is longer than a yard. From this, given that ‘one metre’ is a rigid
designator of the length of stick S at t, and that stick S at t is longer than a
yard, one can infer that it is a necessary truth that any concrete object which
has a length of one metre is longer than a yard, which is a necessary a
posteriori truth about an abstract object, a length, on a par with Kripke’s
proposed necessary a posteriori truth about the natural kind, the Tiger, that
everything which belongs to it is a mammal.
The mechanism by which such necessary a posteriori truths about
abstract objects can be generated can be illustrated in another example
(Donnellan’s, given by Salmon (2005: 165)).
I understand what it is for two people to be of-the-same-marital-kind,
where this is defined as meaning that, i.e. as an abbreviation for the claim
that, they are both bachelors, both spinsters, both married men, both
married women, both widows, or both widowers. Given this definition I
know that if two people are of-the-same-marital-kind and one is unmarried
the other is also (just as I know that if two things are the same in length and
one is longer than a yard, the other is also). And I know not only that any
actual people who are so related are such that if either is unmarried, both
are, but that any actual or possible such people are such that if either is
unmarried both are, that is that if anyone in any actual or possible
circumstance were to be of-the-same-marital-kind-as anyone else in any
other actual or possible circumstance is or was then the former would be
unmarried if the latter is or was. This is trivially so, because it is analytic
that if two people are both bachelors or both spinsters or … (spelling out
the full definition of same-marital-kind), both are unmarried if either is. So
far, however, no discourse about marital kinds as objects of reference has
been justified. I have simply defined by abbreviative definition ‘x is-of-the-
same-marital-kind-as y’ to mean ‘x and y are both bachelors, both spinsters,
both unmarried men, both unmarried women, both widows or both
widowers’. However, having given this definition I can now introduce talk
of marital kinds as objects by a Fregean definition by abstraction: the
marital kind to which x belongs is the marital kind to which y belongs if
and only if x is-of-the-same-marital-kind-as y. On its natural understanding
this definition has counterfactual force. If I then fix the reference of a
proper name, ‘B’, to be the marital kind to which my (reclusive) neighbour,
Hilary, belongs, then when I make the a posteriori discovery that he is a
bachelor, I can infer, by the reasoning sketched in the example of one metre,
that it is necessary that the marital kind B is such that anyone belonging to
it is unmarried. If I now define the predicate ‘x is a squatchelor’ to mean
‘the marital kind to which x belongs is identical with the marital kind B’, I
can infer that it is a necessary truth that anyone who is a squatchelor is
unmarried; something I can know only a posteriori.
The reasoning by which Kripke suggests that one can infer that it is a
necessary truth that all tigers are mammals, or that it is a necessary truth
that all samples of water contain atoms of oxygen and hydrogen, appears to
be entirely similar. The difference is that it is transparent in the cases just
considered how I can know a priori the crucial relevant proposition linking
sameness of kind with the sharing of the relevant property. It is simply
analytic, given how sameness of marital kind is defined, that if two actual
or possible people are of-the-same-marital-kind then both are unmarried if
either is, and it is, if not analytic, an obvious conceptual truth that if two
actual or possible things are the same in length both are longer than a yard
if either is (just as, in the Fregean examples, it is an obvious conceptual
truth that if two lines are parallel and one lies at 90 degrees to a third so
does the second, and an obvious conceptual truth that if the Fs are one– one
correlated with the Gs then if there are more than three Fs there are more
than three Gs). But the proposition that, if two actual or possible things are
of the same natural kind, both are mammals if either is, is not similarly
transparently knowable a priori. Nor is this true of the proposition that if
two actual or possible samples are samples of the same natural kind of stuff
both are compounds of hydrogen and oxygen if either is. As we have seen
Kripke emphasizes that the notion of sameness of natural kind he is
operating with is a primitive one, available to biologically and chemically
uninformed communities of speakers. Thus it is left unexplained how such
propositions as these can be known a priori. Nevertheless, it does appear
likely that the best account of the mechanism by which Kripke thinks
necessary a posteriori truths about natural kinds can be generated is the one
sketched above, in which such propositions play a crucial role. Whether this
means that Kripke is presupposing knowledge of essentialist principles
other than those he explicitly relies on is moot (see Salmon 1985: ch. 6). Is
the proposition that if two actual or possible things are of the same natural
biological kind either is a mammal if the other is itself an essentialist
principle? The argument for saying so is that it entails (together with the
definition by abstraction by which reference to natural biological kinds as
abstract objects can be introduced) that any natural kind is such that if
something which is a member of that kind is a mammal it is a necessary
truth that anything which is a member of it is a mammal, which is a general
essentialist principle about natural kinds.
But on the same basis it can be argued that the proposition that if two
actual or possible things are the same in length either is longer than a yard if
the other is is itself an essentialist principle on the ground that, together
with the abstraction principle for lengths, it entails that every length is such
that if something of which it is the length is more than a yard long it is
necessarily true that everything of which it is the length is more than a yard
long. Or again, that the proposition that if two actual or possible people are
of-the-same-marital-kind either is male if the other is is itself an essentialist
principle on the ground that, together with the abstraction principle for
marital kinds, it entails that every marital kind is such that if someone of
whom it is the marital kind is male it is necessarily true that everyone of
whom it is the marital kind is male.
So perhaps Kripke cannot be charged with presupposing any general
essentialist principles additional to those to which he explicitly appeals.
Nevertheless, it is important to see that the general essentialist principles
about kinds conceived as abstract objects on which he does rely can only be
derived, in the manner sketched above, from the a priori knowable general
propositions about individuals, by appeal to the abstraction principles by
which talk of the kinds as abstract objects is introduced. And these, on the
understanding of them which allows the derivations, are not expressible in
the language of quantified modal logic. The abstraction principle for marital
kinds, for example, cannot be understood merely as saying that it is a
necessary truth that for all persons x and y, the marital kind to which x
belongs is identical with the marital kind to which y belongs if and only if x
is of-the-same-marital-kind-as y. To allow the derivation of the general
essentialist principle that any marital kind is such that if any person
belonging to that marital kind is unmarried necessarily any person
belonging to that marital kind is unmarried, its content has to be that for all
possible worlds W and W* and possible persons x and y the marital kind to
which x belongs in W is identical with the marital kind to which y belongs
in W* if and only if x in W is of-the-same-marital-kind-as y in W* –
something that cannot be expressed in the language of quantified modal
logic, but requires for its expression an explicit quantification over possible
worlds (Salmon 1985). (Alternatively, assuming Lewis’s modal realism and
the world-boundness of possibilia, its content can be captured without
quantification over possible worlds, in the statement: for all possible
persons x and y, the marital kind to which x belongs is the marital kind to
which y belongs if and only if x is-of-the-same-marital-kind-as y. The
content of the proposition that necessarily all squatchelors are unmarried is
thus that all possibilia of-the-same-marital-kind-as reclusive Hilary are
unmarried. But this, from the modal realist viewpoint, is not a de re modal
statement about an abstract object at all (it does not say anything about all
or some counterparts of such an abstract object), but a de re modal
statement about Hilary, to the effect that he is essentially unmarried (taking
being-of-the-same-marital-kind-as as the relevant counterpart relation).
Similarly, from the modal realist viewpoint, the proposition that it is
necessarily true that all tigers are mammals is equivalent to the proposition
that all possibilia of the same natural kind as paradigm tiger Tim are
mammals, which is not a de re statement about a kind, the Tiger, but a de re
modal statement about Tim, to the effect that, under a certain counterpart
relation, he is essentially a mammal. It is therefore, from this viewpoint,
entirely unsurprising that it can only be known a posteriori that all tigers are
mammals, but the fact that it can only be so known does not establish that
any proposition is a necessary a posteriori truth, since the proposition that
necessarily all tigers are mammals, i.e. that Tim is essentially a mammal, is
not equivalent to any purely de dicto modal truth.)
The Illusion of Contingency and Mind–Brain
Identity

Kripke’s discussion of theoretical identifications is preparatory to his attack

on mind–brain identity theories. At the time Kripke gave the Naming and
Necessity lectures a popular view was that mental states were contingently
identical with brain states, and that the particular identifications were a
matter for scientific discovery. Thus it was thought it might be discovered
that pain was, as a matter of fact, C-fibre stimulation, in the way it was
discovered that heat was molecular motion or water H2O. Kripke rejects the
comparisons. In all these cases, he argues, the identity if true at all must be
necessary, so it is confused to talk of contingent identity. But the mind–
brain identity theorist cannot just sit back and say that the identity of pain
with C-fibre stimulation is a necessary truth as is, on Kripke’s own view,
the identity of water with H2O or heat with molecular motion. For in all
these cases there is, in a sense, an appearance of contingency, since the
identity, if knowable at all, is knowable only a posteriori. But, for reasons to
be explained below, in the case of pain and C-fibre stimulation the
appearance cannot be diagnosed as an illusion, as it can be in the other
cases. So it has to be accepted and the identity rejected.
This argument against the mind–brain identity theorists requires that
whenever there is a genuinely necessary a posteriori truth its a posteriori
status, which engenders an illusion of contingency, must be diagnosable in
the way Kripke explains. His explanation is that in every such case there is
an associated contingent truth, which is a priori equivalent to the necessary
truth and with which we are disposed to conflate it. The passages in which
he puts forward this thesis are those already discussed, from the end of the
second lecture and the third lecture, in which Kripke considers what he
refers to as the puzzle of necessary a posteriori truth, or to put it in Kantian
form, the question ‘How is necessary a posteriori truth possible?’ But as we
have seen it is not straightforward to say what is puzzling about Kripkean
necessary a posteriori truths. It is sometimes said that they show that
conceivability is no proof of possibility. But that is not a distinctive feature
of them, for the falsehood of a necessary a priori (mathematical) truth may
be conceivable for me simply because I have not done the necessary
reasoning or, given my limited intellectual power, cannot. And if there are
unknowable necessary truths their falsehood will be conceivable by anyone.
In this respect they will be as good counterexamples to the dictum that
conceivability implies possibility as ‘Hesperus is not Phosphorus’ or ‘Water
is not H2O’. And, as we have also seen, if there are unknowable necessary
truths there are a fortiori necessary truths which are knowable but only a
posteriori – any disjunction of a knowable contingent truth with one of
these will qualify. But these truths will not be knowably necessary. What
Kripke argues for in particular is the existence of knowably necessary a
posteriori truths.
But why should we expect the truth of knowably necessary truths to be
knowable a priori?
Kripke himself says that:
[People might have thought that] … if something not only happens to be true in the actual world but
is also true in all possible world, then of course, just by running through all the possible worlds in our
heads we ought to be able with enough effort to see, if a statement is necessary, that it is necessary,
thus know it a priori. But really this is not so feasible at all.
(1980: 38)

It is not clear how seriously Kripke intends this diagnosis of what is

puzzling about his examples of the necessary a posteriori. Nevertheless, it
fails to explain why they should be any more puzzling than unknowable
necessary truths.
The question ‘How is knowably necessary a posteriori truth possible?’ is
Kantian in form, and any question of this form makes sense only against the
background of an argument which purports to show that such-and-such, in
this case knowably necessary a posteriori truth, is not possible. Kripke
never formulates any such background argument, but one can be
reconstructed, given his solution to his puzzle, that is, that any knowably
necessary a posteriori truth is paired with a contingent truth, knowledge of
which is required for knowledge of it. The reconstructed argument is the
following:
Any knowably necessary truth can be known to be true without knowledge of any contingent truth.
Any truth which can be known to be true without knowledge of any contingent truth can be known to
be true a priori.
So:
Any knowably necessary truth can be known to be true a priori.

Kripke’s response to the puzzle is the denial of the first premiss of this
argument. Another response would be to deny the second premiss. Kripke
does not consider this alternative, but as we shall see, it would undermine
his argument against mind–brain identity theories.
According to Kripke the necessary a posteriori status, and hence the
appearance of contingency, of his necessary a posteriori truths is to be
explained by their association with contingent a posteriori truths which
must be known if they are to be known. He illustrates with the case of the
table in front of him, which is known to be wooden only a posteriori but, he
thinks, is essentially (originally) wooden. In this case he suggests that we
have the intuition that the table could have turned out to be made of ice. But
what this means is just that he could have been in qualitatively the same
situation with just the same evidence even though an ice table was present.
So although the statement ‘this table is made of wood if it exists’ states a
necessary truth, there is an associated contingency, ‘the table in front of me
is made of wood’ in which the actual wooden table is picked out by a
contingent property. So, according to Kripke:
Any necessary truth could not have turned out otherwise. In the case of some necessary a posteriori
truths, however, we can say that under appropriate qualitatively identical evidential situations, an
appropriate corresponding statement may have been false. The loose and inaccurate statement that
gold might have turned out to be a compound should be replaced by the statement that it is logically
possible that there should have been a compound with all the properties originally known to hold of
gold.
(1980: 143)
In the case of identities using rigid designators he says, ‘there is a simpler
paradigm. … The references of [the rigid designators “R1” and “R2”] may
well be fixed by non-rigid designators “D1” and “D2”… Then …“D1=D2”
may well be contingent and that is often what leads to the erroneous view
that “R1=R2” might have turned out otherwise’ (1980: 143–4).
The reason we are thus led into error, into an illusion of contingency,
Kripke thinks, is that there are two senses in which ‘could have turned out
otherwise’ can be used. In one sense ‘it could have turned out that P’ entails
that it could have been the case that P, in the sense that it is metaphysically
possible that P. In another, epistemic, sense it does not. When I say, in this
second sense, that it could have turned out that Hesperus was distinct from
Phosphorus what I say could be accurately expressed by saying that it could
have turned out in the first sense that the body visible in the evening was
distinct from the body visible in the morning, which does entail that the
body visible in the evening could have been distinct from the body visible
in the morning in the sense of a metaphysical possibility. Confusing the two
senses, I am disposed, Kripke thinks, to infer that Hesperus could have been
distinct from Phosphorus, in the sense of metaphysical possibility. So I am
disposed to be led into error because I can only come to know that Hesperus
is Phosphorus, a necessary a posteriori truth, via knowing the associated
contingent truth (which is why the necessary a posteriori truth is a
posteriori), and I am inclined to confuse two senses (epistemic and
metaphysical) of ‘it could have turned out that P’.
At the end of Naming and Necessity Kripke brings these considerations
to bear against the mind–brain identity theory. He distinguishes three
different identifications: of a person with his body, of a particular token
sensation (or the event or state of having that sensation) with a particular
token brain state, and of types of mental event with types of physical event.
His aim in each case is to show at least that the only option for the identity
theorist is to embrace a necessitarian version of his position: to say that the
identity is both true and necessary since contingent identity is a confusion.
Thus he argues in the first case that someone who accepts, with
Descartes, that he could exist without his body must also accept that he is in
fact distinct from his body since, if ‘B’ is a rigid designator of my body,
because ‘I’ is also a rigid designator, ‘Iam B’ must be necessary if true at
all. He notes that a precisely parallel argument can be used to conclude,
from the premiss that a statue and the all-times-coincident hunk of matter
from which it is composed could exist apart, that they are not in fact
identical. Many do accept the parallel argument, as Kripke himself does;
others, as we have seen, follow Lewis in embracing the inconstancy of
modal predication, which allows them to accept the premiss in either case,
and reject the conclusion. At any rate Kripke does not think the argument
establishes Cartesian dualism (the identity of the self with a Cartesian
immaterial soul), with which, as he notes, his own thesis of the essentiality
of material origin is inconsistent. The anti-materialist arguments Kripke
takes seriously are the arguments against the token–token identity theory
and the type–type identity theory.
The argument against the token–token identity theory runs as follows.
Let ‘A’ name a certain particular sensation of pain and ‘B’ name a particular
brain state occurring in a man Jones. If ‘A = B’ is true it is necessarily true.
But it is possible for A to exist without B and B without A. So it is not
necessarily true, hence it is not true. Since ‘A’ and ‘B’ are rigid designators
the point of contention can only be the premiss that A could exist without B
and B without A. Kripke’s argument that B could exist without A is that B
is only contingently a pain if it is a pain at all, whereas A is necessarily a
pain: ‘it is at least logically possible that B should have existed without
Jones feeling any pain at all’ (1980: 146); on the other hand, ‘can any case
of essence be more obvious than the fact that being a pain is a necessary
property of each pain?’ (1980: 146). His argument that A could exist
without B is that A is only accidentally a brain state if at all, but B is
essentially a brain state: ‘the pain could have existed without the
corresponding brain state [but] being a brain state is evidently an essential
property of B’. Thus A and B differ in their essential properties and are
distinct.
The token–token identity theorist can respond by denying that A is
essentially a pain. He can say that ‘being a pain, as a property of a physical
state, is to be analyzed in terms of the “causal role” of the state, in terms of
the characteristic stimuli which cause it and the characteristic behaviour it
causes’ and that the causal role of a state is a contingent feature of it (1980:
147). Against this Kripke says only that it strikes him as self-evidently
absurd, since it amounts to the view that this very pain I now have could
have existed without being a mental state at all.
Whether or not this is right, it is clear that Kripke’s argument against the
token–token identity theory needs more than the supposition that ‘A’ and
‘B’ are rigid designators: it also needs the essentialist premiss that pain
tokens are necessarily pains.
In fact, it requires more, because the token–token identity theorist might
allow this but still resist the argument by denying that it is possible for B to
exist without pain. He might say that it is an illusion that the token event B
could exist without pain being felt, just as it is an illusion that the particular
wooden table Kripke uses as an example could have existed without being
made of wood. According to Kripke a table made of ice rather than wood
might have existed and looked exactly the same as the wooden one and
been exactly where the wooden one was, but it would not have been the
wooden one but merely an epistemic counterpart of it. Similarly, the token–
token identity theorist might say, a brain state token could have existed
which looked exactly like B (when Jones’s brain was examined), but was
not a pain. But it would not have been the very token brain state B, which is
essentially a pain. So the token–token identity theorist could accept that all
actual brain state tokens which are pains are essentially so, and explain the
appearance of contingency by appeal to the possibility of brain states which
are not pains but can serve as epistemic qualitative counterparts of the
actual brain states which are pains.
Against this Kripke can only appeal to an (anti-) essentialist intuition: ‘it
would seem logically possible that B could exist without any sensation with
which it might plausibly be identified’ (1980: 146).
However, even if the token–token identity theorist can defend himself
against Kripke’s argument that B might have existed without A, either by
insisting that A is accidentally a pain or by insisting that B is essentially a
pain, he also needs to respond to Kripke’s argument that A might have
existed without B, since A is only accidentally a brain state and B is
essentially a brain state. But, for the token–token identity theorist who says
that A is only accidentally a pain (since it only contingently occupies the
causal role of pain) it is not a great stretch to say that what A essentially is
is a brain state (just as on Kripke’s account a token of heat which causes a
sensation of heat in a perceiver is only accidentally a cause of a heat
sensation, but essentially an occurrence of molecular motion). Finally, to
complete the catalogue of possibilities, the token–token identity theorist
might accept that A is only accidentally a brain state, but say exactly the
same of B. B, he might say, qualifies as a brain state because of its
neurological and physiological role, but it is therefore not essentially a brain
state and is in fact essentially a pain. This is a non-materialist version of the
token–token identity theory.
Thus the token–token identity theorist can resist Kripke’s argument by
rejecting one or more of his intuitions about the (differing) essential and
accidental properties of A and B. But another possibility for him, as in the
case of mind–body identity, is to accept Kripke’s intuitions but deny the
inference to non-identity by rejecting Kripke’s assumption that modal
predication is constant in denotation. Thus, he can say that A is essentially a
pain and B is not and that A is B just as Goliath is essentially a statue and
Lumpl is not though Goliath is Lumpl.
Kripke’s most developed argument in the final pages of Naming and
Necessity is against the type–type identity theorist, who holds not only that
each mental state token is identical with some physical state token, but also
that any two token events of the same psychological type are events of the
same physical type, e.g. that pain is the firing of C-fibres. So type–type
identity theory is put forward as precisely analogous to such theoretical
identifications as that of water with H2O and heat with molecular motion.
Kripke’s argument is that the analogy does not obtain. The argument has
two parts. First, he argues that in the case of pain and C-fibre stimulation, as
in the other cases, the identity, if it holds, must hold necessarily since ‘pain’
and ‘C-fibre stimulation’ are rigid designators. But second, he argues that
the a posteriori status and thus apparent contingency of ‘pain is C-fibre
stimulation’ cannot be explained in the way that the a posteriori status and
apparent contingency of ‘heat is molecular motion’ can be explained. The
reason is that we do not fix the reference of the kind term ‘pain’ by a
flexible designator in the way in which we fix the reference of ‘heat’ by a
flexible designator and so the a posteriori status of the identity statement
cannot be explained away by appeal to an associated genuine contingency.
(Note that Kripke’s challenge to the identity theorist is not merely to
explain the epistemic possibility of the falsehood of ‘pain is C-fibre
stimulation’, but to explain its a posteriori status. Unobvious unproved
mathematical truths share with ‘pain is C-fibre stimulation’ the property
that their denials are epistemically possible, but they are not a posteriori.)
One way the type–type theorist can resist this argument is simply by
denying that ‘pain’ is a rigid designator. This is what Lewis does, defending
the standard functionalist position:
If the state of having neurons hooked up in a certain way and firing in a certain pattern is the state
properly apt for causing and being caused, as we materialists think, then that neural state is pain. But
the concept of pain is not the concept of that neural state. The concept of pain, unlike the concept of
the neural state which is in fact pain, would have applied to some different state if the relevant causal
relations had been different. Pain might not have been pain.
(Lewis 1983: 124)

However, this seems unpersuasive. The linguistic considerations that count

in favour of regarding ‘heat’ as a rigid designator of a kind are also present
in the case of ‘pain’. In both cases we can only apply the notion of rigidity
Kripke explains if we regard the terms as names of kinds, analogous to
proper names of individuals. So the inference from a claim about the modal
status of a proposition expressed using the name to a claim about the modal
status of a proposition expressed using the corresponding predicate is non-
trivial, as we have seen. But if this is a problem for the contention that
‘pain’ is a rigid designator it is equally a problem for the contention that
‘heat’ is a rigid designator. In short, no doubt we can read ‘pain might not
have been pain’ as saying something true, but we can equally well read
‘heat might not have been heat’ as saying something true.
 If we accept with Kripke that ‘pain’ is a rigid designator and that ‘C-fibre
stimulation’ is also a rigid designator, we must say that the only tenable
form of type–type identity theory is one according to which the identity is
necessary. But Kripke argues that the a posteriori status and consequent
appearance of contingency in this case cannot be explained. Heat could not
have been anything other than molecular motion, but the cause of the
sensation by which we in fact identify heat (sensation S) might not have
been heat, i.e. molecular motion, but some other physical phenomenon and
heat, i.e. molecular motion, might have been present but not caused
sensation S. ‘Heat’ is a rigid designator, but its reference is determined by
an accidental property, the property of being the cause of sensation S. So
corresponding to the a posteriori status and consequent apparent
contingency of ‘heat is molecular motion’ is the genuine contingency of
‘the phenomenon causing sensation S is heat’. Consequently, though it
seems like a contingent scientific discovery that hot things contain
molecules in motion, it is not; what is a contingent matter is that things that
cause sensation S are hot.
In the case of ‘pain’, however, Kripke argues that no flexible designator
can play the role in explaining the a posteriori status and consequent
apparent contingency of ‘pain is C-fibre stimulation’ that ‘the cause of
sensation S’ plays in explaining the apparent contingency of ‘heat is
molecular motion’:
In the case of … heat, what seemed really possible is that molecular motion might have existed
without being felt as heat … without producing sensation S. … [i]s it analogously possible that a
stimulation of C-fibres could have existed without being felt as pain? … then the stimulation of C-
fibres can exist without pain, since for it to exist without being felt as pain is for it to exist without
there being any pain. … The trouble is that the identity theorist does not merely hold that the physical
state produces the mental state, he wishes the two to be identical. In the case of … heat … the
sensation of heat … is an intermediary. … In the mental-physical case no such intermediary is
possible. … The apparent contingency of the connection between the mental state and the
corresponding brain state thus cannot be explained by some sort of qualitative analogue in the case of
heat.
(1980: 151)
Pain is a feeling, so the description by which we fix the reference of ‘pain’
cannot be ‘the cause of such a feeling’. The only possibility that remains for
the identity theorist, and the one he would clearly take, is to say that the
reference is fixed, not as Kripke says, by its immediate phenomenological
quality, the property of being pain itself, and so by an essential property, but
by its causal role as the intermediary between characteristic physical stimuli
and characteristic behaviour. Kripke regards this proposal as self-evidently
absurd (1980: 147) since it entails that the very pain (token) I now have
could have existed without being a mental state at all, which he thinks
evidently false. But, as we have seen, this essentialist contention is
debatable.
The proposal, then, is that just as the reference of ‘heat’ is fixed by the
description ‘the cause of sensation S’, the reference of ‘pain’ is fixed by the
description ‘that which is caused by physical stimuli P and causes
behaviour B’. So, just as ‘heat might not have been molecular motion’ is
false, but ‘the cause of sensation S might not have been heat’ is true, so
‘pain might not have been C-fibre stimulation’ is false, but ‘the
intermediary between physical stimulus P and behaviour B might not have
been pain’ is true.
But actually this consequence seems quite implausible. To say that the
cause of sensation S might not have been heat is to say that things that felt
hot might not have been hot because they might not have contained any
molecules in motion. Analogously, to say that the intermediary between P
and B might not have been pain is to say that creatures who suffered
physical damage actually apt to cause pain and were caused by that damage
to go into a state which produced behaviour actually apt to be caused by
pain might not have been in pain because they might not have had occurring
in their brains any C-fibre stimulation. But a point stressed by all
functionalists is that mental states might be variously realized in different
(possible) creatures, so that what plays the role of pain in one kind of
organism need not in another. Pain, the suggestion is, is like glue (see
Hughes 2004: 228); any actual or possible bit of stuff, even if it is of a quite
different kind from the glue we are familiar with (which may be the only
substance actually used as glue), will count as glue if it bonds things
together (and perhaps is manufactured or used for that purpose). And any
creature is in pain if there is an internal state mediating in it between
appropriate physical inputs and behavioural outputs.
If so, the identity theorist cannot explain the a posteriori status and
consequent apparent contingency of ‘pain is C-fibre stimulation’ in the way
suggested (any more than a type–type identity theorist about glue can
explain the apparent contingency of ‘glue is UFR’ if UFR [Urea
Formaldehyde Resin] is the only substance of which glue is actually made).
So if he accepts Kripke’s claim that ‘pain’ is a rigid designator, he must
agree that it is actually false that pain is C-fibre stimulation. Analogously,
the type–type identity theorist about glue must admit defeat and agree that it
is actually false that glue is UFR.
Or rather, the type–type identity theorist must accept defeat if he goes
along with Kripke’s two assumptions (a) that pace Lewis modal predication
de re is constant and (b) that the a posteriori status and consequent apparent
contingency of a knowably necessary a posteriori truth can only be
explained by appeal to an associated genuine contingency, knowledge of
which is required for knowledge of the necessary a posteriori truth. If the
type–type identity theorist rejects the first of these assumptions he can, like
the token–token identity theorist, accept all Kripke’s modal claims and deny
the non-identity. The remaining possibility for the type–type identity
theorist is to reject Kripke’s contention that an appeal to an associated
genuine contingency is needed to explain the a posteriori status of a
necessary a posteriori truth. An alternative way to reject the argument
against the possibility of necessary a posteriori truth to which we took
Kripke to be replying by appeal to the possibility of an associated genuine
contingency is to reject its second premiss. To do this is to say that there are
knowably necessary features of the world which simply cannot be known a
priori, although our route to a posteriori knowledge of them need not be via
knowledge of any contingent truth. Thus, the type–type identity theorist can
say, if he responds in this way, that Kripke is right that pain is not identified
by an accidental property as heat is, and that it can only be known a
posteriori that pain is C-fibre stimulation, but this is consistent with the
identity he maintains. It is knowable that pain is C-fibre stimulation and
necessarily so. This is consistent because an appeal to an associated genuine
contingency to explain the a posteriori status of a knowably necessary a
posteriori truth is unnecessary. Similarly, he can say, whereas my a
posteriori knowledge of the necessary truth that this table is wooden if it
exists is, as Kripke says, associated with my a posteriori knowledge of the
contingent truth that the table in front of me is wooden if it exists, my a
posteriori knowledge that I am a human being if I exist – which Kripke
would agree is knowledge of a knowably necessary truth – is not
necessarily associated with a posteriori knowledge of a contingent truth,
since there is no flexible description of myself which I associate with ‘I’;
rather I pick myself out directly (Wright 2002).
If this response can be plausibly developed, Kripke’s argument against
the type–type identity theory fails. But, as we have seen, even if it does not,
the argument’s assumption of the constancy of modal predication can be
challenged. And even if this assumption is left in place the most the
argument can establish is that ‘pain’ stands to ‘the firing of C-fibres’ as
‘glue’ stands to ‘UFR’, which is hardly a contention a materialist need
strenuously reject.
Glossary

a posteriori/a priori: These terms refer to kinds of knowledge. A truth is

knowable a priori if our justification for it need not rely on experience.
It is a posteriori if it can only be known on the basis of experience.
abstract object/concrete object: Abstract objects are not located in space
and time and hence are causally inefficacious. Examples are numbers
and classes.
abstraction principle: A principle specifying the condition of identity of
one type of entity as a relation between entities of another type. The
paradigm is Frege’s specification of the condition of identity for
directions: the direction of line a is the direction of line b if and only if
line a is parallel to line b.
accidental designator: see rigid designator.
accidental property: see essential property.
analytic/synthetic: An analytic truth is true merely in virtue of the
meanings of words (e.g. ‘all bachelors are unmarried’), whereas a
synthetic truth is not. The terminology derives from Kant.
analytic philosophy: A twentieth-century philosophical movement which
makes conceptual analysis the foundation of philosophy. Its historically
most important paradigms are Frege, Russell and Wittgenstein.
assertoric content/ingredient sense: Assertoric content is what is said by
an utterance of a sentence, and is also the object of belief, doubt and
other propositional attitudes. The ingredient sense of a sentence is what
the sentence contributes to the assertoric content of more complex
sentences of which it is a part. The terminology derives from Dummett.
attributive use/referential use: A speaker who uses a definite description
attributively states something about whomever or whatever is the so-
and-so. A speaker uses a definite description referentially in an assertion
when he uses the description to enable his audience to pick out whom or
what he is talking about and states something about that person or thing.
The terminology derives from Donnellan.
cluster theory: The theory that the meaning associated with a name by a
speaker is given by a family of definite descriptions, its meaning is
roughly ‘whoever or whatever satisfies most of the following
descriptions …’.
cognitive value: That feature of an expression’s meaning which contributes
to the information conveyed by an utterance of a sentence containing it.
compositionality principle: The Fregean principle that the sense/reference
of an expression is determined by the senses/references of its parts.
concrete object: see abstract object/concrete object.
connotation: A term introduced by John Stuart Mill for the attribute
implied by a connotative term, which both denotes a subject and implies
an attribute. Non-connotative terms signify a subject or an attribute
only. ‘Whiteness’ is non-connotative, it signifies an attribute only.
‘White’ is connotative, it denotes all white things and implies the
attribute whiteness. Proper names, Mill held, are non-connotative and
thus unmeaning marks, which denote the individuals called by them.
counterpart theory: David Lewis’s account of de re modality, which is
part of his genuine modal realism. According to counterpart theory to
say that someone or something might have been so-and-so is to say that
there is a possible world in which he or it has a representative, a
counterpart, which is so-and-so.
de dicto/de re: Ascriptions of properties to objects are de re; ascriptions to
what is said are de dicto. Modal ascriptions de dicto ascribe being
necessary, possible or contingent to what is said. Modal ascriptions de
re ascribe being necessarily, possibly or contingently so-and-so to a
thing. De re propositional attitude ascriptions relate the subject of the
ascription to a thing to which the attitude is said to be being held. De
 dicto propositional attitude ascriptions ascribe to the subject an attitude
to what is said in the following ‘that’-clause.
de facto rigidity/de jure rigidity: The rigidity of a designator is de jure
when its reference is stipulated to be a single object with respect to any
possible world, otherwise it is de facto. So rigid definite descriptions
like ‘the smallest prime’ are de facto rigid.
demonstrative: An indexical expression the reference of an utterance of
which is determined by an accompanying ostension.
denotation: That which is denoted by a phrase. A definite description
denotes what it describes. Russell called definite and indefinite
descriptions ‘denoting phrases’ whether or not anything was actually
denoted by them.
empiricism: The view that all our knowledge of the external world is a
posteriori. Traditionally empiricists also believed that all our ideas came
from experience.
epistemic necessity/possibility: An epistemic possibility (for me) is
something that is so for all I know; an epistemic necessity something
whose negation is not an epistemic possibility.
essence: the set of a thing’s essential properties.
essential/accidental property: An essential property of a thing is one that
it could not have lacked and still existed. An accidental property of a
thing is one that is not an essential property of it.
essentialism: Minimally the thesis that some things have essential
properties. Contemporary essentialists typically hold that things have
accidental properties as well and that the essential properties of one
thing may not be essential properties of, or properties at all of, others.
exportation: The principle of inference that licenses the move from a de
dicto propositional attitude ascription to a de re propositional attitude
ascription.
extension/intension: The extension of a predicate is the class of objects it
describes, the extension of a singular term is its reference, and the
extension of a sentence is its truth-value. The intension of a predicate is
the condition something has to satisfy to belong to its extension. The
 intension of a singular term is the condition something has to satisfy to
be its reference and the extension of a sentence is what it says.
Extensional contexts are ones in which expressions with the same
extension are always substitutable for one another without alteration in
the truth-value of the whole sentence.
externalism/internalism: In the philosophy of mind and language
externalism is the thesis that what is said and thought is dependent on
aspects of the world external to the mind of the subject: ‘Meanings ain’t
in the head’ (Putnam). Internalism is the denial of externalism.
indexical: An expression whose reference upon an occasion is dependent
upon the context: the spatial or temporal location of the speaker or some
other feature that can vary from occasion to occasion.
intension: see extension/intension.
internalism: see externalism/internalism.
Law of Excluded Middle: the principle that every sentence of the form
‘Por not: P’ is true.
metaphysical necessity/possibility: A metaphysical possibility is
something that is actually true or true in some non-actual possible
world. A metaphysical necessity is something that is true in all possible
worlds.
methodological solipsism: The principle of methodological solipsism as
stated by Putnam is that no psychological state properly so-called
presupposes the existence of any individual other than that to which it is
ascribed.
modality: The modal notions are those of necessity and possibility, of what
must be and might be and the derivative notion of contingency, what
might or might not be so, in any of the meanings of these expressions.
modal logic: the logic of necessity and possibility.
modal realism: A modal realist holds that possible worlds exist and that
talk about them is literally true. A genuine modal realist holds that
possible worlds are just like the actual world –‘ourselves and all our
surroundings’ – but are not spatio-temporally connected to the actual
world. An ersatz modal realist holds that possible worlds, while real, are
 not themselves objects like the actual world, but total representations of
how the actual world could be.
narrow psychological state: A narrow psychological state is one allowed
by the assumption of methodological solipsism, that is, one that does
not presuppose the existence of any individual other than that to which
it is ascribed.
natural kind: Natural kinds include elements and compounds, like gold
and water, vegetable and animal species, like elm trees and tigers, and
kinds of phenomena, like heat and lightning. According to Kripke
natural kinds are susceptible to necessary a posteriori theoretical
identifications.
necessity of identity: The principle that if objects are identical they are
necessarily identical.
necessity of origin: The principle that the material or biological origin of
an individual is an essential property of it.
operator: An expression for a function, mathematical or logical, including,
for example, truth-functional operators, modal operators and quantifiers.
possible world: A total way things can be. Genuine modal realists and
ersatz modal realists dispute the nature of possible worlds.
primary occurrence/secondary occurrence: A primary occurrence of a
denoting phrase is one which does not lie within the scope of any
operator; a secondary occurrence does so. Russell, to whom the
terminology is due, says that a description has a primary occurrence
when the proposition in which it occurs results from substituting the
description for ‘x’ in some propositional function jx; a description has
secondary occurrence when the result of substituting the description for
‘x’ gives only part of the proposition concerned.
proper name: A (possibly many-worded) simple symbol which is a name
for some one object. In Russell’s writing ordinary proper names are
distinguished from logically proper names. The latter are name of
objects of immediate acquaintance, directly presented in experience.
Ordinary proper names are, in a sense, abbreviated descriptions.
propositional attitude: A state such as belief, desire or fear. The general
form of ascription of a propositional attitude is: X Vs that P, which can
be said to ascribe to the subject X an attitude towards what is said by the
clause ‘that P’.
quantifying in: A sentence involves quantifying into a modal or
propositional attitude construction when a variable within the scope of
the modal operator or propositional attitude verb is bound by a
quantifier outside it.
reference: see sense/reference.
referential opacity/transparency: A context is referentially transparent if
coextensive expressions can be substituted for one another salva
veritate within it. Otherwise it is referentially opaque.
referential use: see attributive/referential use.
rigid designator/accidental designator: A rigid designator designates the
same thing with respect to every possible world in which that thing
exists. A designator which is not rigid is called accidental.
salva veritate: Two expressions can be substituted salva veritate if
replacement of one by the other will not change the truth-value of the
whole sentence in which the substitution is made.
scope: The scope of an operator is the part of an expression over which it
operates. Scope ambiguities are caused by insufficient information
about the scope of an operator.
semantic value: The semantic value of an expression is that which it
contributes to the determination of the truth-values of sentences
containing it.
sense/reference (German Sinn Bedeutung): According to Frege two
expressions may have the same reference but present it in different
ways. The modes of presentation are the senses associated with the
expressions. ‘Hesperus’ and ‘Phosphorus’ have the same reference but
different senses.
transparent: see referential opacity/transparency.
type–type identity theory/token–token identity theory: A type–type
identity theory is a form of theory which asserts that every type of
mental event is identical with some type of physical event. A token–
token identity theory asserts only that any token mental event is a
physical event.
Bibliography

Barrett, Robert and Gibson Roger, 1990, Perspectives on Quine. Oxford:

Blackwell.
Burge, T. 1979, ‘Sinning against Frege’, Philosophical Review 88: 398–
432.
——2005, Truth, Thought, Reason: Essays on Frege, Oxford: Oxford
University Press.
Burgess, J. 2006, ‘Saul Kripke: Naming and Necessity’, in J. Shand (ed.).
Central Works of Philosophy, vol. 5, Chesham: Acumen, 166–86.
Chalmers, D. 1996, The Conscious Mind: In Search of a Fundamental
Theory, Oxford:Oxford University Press.
——2011, ‘Propositions and Attitude Ascriptions: A Fregean Account’,
Noûs 4: 595–639.
Chisholm, R. 1967, ‘Identity through Possible Worlds: Some Questions’,
Noûs 1: 1–8.
Davies, M. 2004, ‘Reference, Contingency and the Two-Dimensional
Framework’, Philosophical Studies 118: 83–131.
Davies, M. and Humberstone, I. L. 1980, ‘Two Notions of Necessity’,
Philosophical Studies 38: 1–30.
Donnellan, K. 1966, ‘Reference and Definite Descriptions’, Philosophical
Review 25: 281–304.
——1972, ‘Proper Names and Identifying Descriptions’, in D. Davidson
and G. Harman (eds). Semantics of Natural Language, Dordrecht:
Reidel, 252–355.
——1977, ‘The Contingent A Priori and Rigid Designators’, Midwest
Studies in Philosophy 2: 12–27.
Dummett, M. 1973, Frege: Philosophy of Language, London: Duckworth,
2nd ed. 1981.
——1974, ‘Postscript’, Synthese 27: 523–44. Reprinted in Truth and Other
Enigmas, London: Duckworth, 1978 as ‘The Social Character of
Meaning’, 420–31.
——1981, The Interpretation of Frege’s Philosophy, London: Duckworth.
Evans, G. 1979, ‘Reference and Contingency’, Monist 62: 160–89.
——1982, The Varieties of Reference, Oxford: Oxford University Press.
Fitch, G.W. 2004, Saul Kripke, Chesham: Acumen.
Frege, G. 1884, Die Grundlagen der Arithmetik, Breslau: W Koebner; tr. as
Frege 1968.
——1892, ‘On Sense and Reference’,in Translations from the
Philosophical Writings of Gottlob Frege, Peter Geach and Max Black
(eds), Oxford: Basil Blackwell, 1969, 56–78.
——1918, ‘The Thought’,in Collected Papers on Mathematics, Logic and
Philosophy, B. McGuinness (ed.), tr. M. Black et al., Oxford: Blackwell,
1984.
——1962, Grundgesetze der Arithmetik (vol. 1 of 1893 and vol. 2 of 1903,
combined in one vol.), Hildesheim: Olms; preface, introd. and section
1–52 of vol. 1 and appendix to vol. 2; tr. in Frege 1964; parts of vol. 2
tr. in Frege 1997.
——1968, The Foundations of Arithmetic, tr. of Frege 1884 by J. L. Austin,
with German text, Oxford: Blackwell.
——1969, Translations from the Philosophical Writings of Gottlob Frege,
Peter Geach and Max Black (eds), Oxford: Basil Blackwell, 1960.
——1984, Collected Papers on Mathematics: Logic and Philosophy, B
McGuinness (ed.), tr. M. Black et al., Oxford Blackwell; includes ‘The
Thought’ (1918).
Geach, P. 1962, Reference and Generality, Ithaca: Cornell University Press.
——1976, ‘Two Kinds of Intentionality?’, Monist 59: 306–20.
Gibbard, A. 1975, ‘Contingent Identity’, Journal of Philosophical Logic 4:
187–221.
Hale, R. V. 2004, ‘Putnam’s Retreat: Some Reflections on Hilary Putnam’s
Changing Views about Metaphysical Necessity’, in French and
Wettstein (eds). Midwest Studies in Philosophy, vol. XXVIII, The
American Philosophers, 351–78.
Hughes, C. 2004, Kripke: Names, Necessity, and Identity, Oxford: Oxford
University Press.
Humberstone, I. L. 1982, ‘Scope and Subjunctivity’, Philosophia 12: 99–
126.
Jackson, F. 1998, ‘Reference and Description Revisited’, Philosophical
Perspectives 12: 201–18.
Kaplan, D. 1969, ‘Quantifying In’, in Davidson and Hintikka (eds). Word
and Objections: Essays on the Work of W.V. Quine, Dordrecht: Reidel,
178–214.
Kripke, S. 1963, ‘Semantical Considerations on Modal Logic’, Acta
Philosophica Fennica 16: 83–94.
——1971, ‘Identity and Necessity’, in M. K. Munitz (ed.). Identity and
Individuation, New York: New York University Press. Reprinted in A.
W. Moore (ed.). Meaning and Reference, Oxford: Oxford University
Press, 1993. Page references to reprint.
——1972, ‘Naming and Necessity’, in D. Davidson and G. Harman (eds).
Semantics of Natural Language. Dordrecht: Reidel, 252–355. Reprinted
in revised form as a monograph by Basil Blackwell, Oxford, 1980.
——1973, ‘Reference and Existence: The John Locke Lectures for 1973’.
——1977, ‘Speaker’s Reference and Semantic Reference’, Midwest Studies
in Philosophy 2: 255–76.
——1979, ‘A Puzzle About Belief’,in Meaning and Use, A. Margalit (ed.),
Boston: Reidel, 239–83.
——1980, Naming and Necessity, Oxford: Basil Blackwell. Expanded
monograph version of Kripke 1972.
——2011, Philosophical Troubles: Collected Papers, vol. I, Oxford:
Oxford University Press.
Kripke, S. et al. 1974 ‘Second General Discussion Session’, Synthese 27:
509–21.
Kroon, F. 1987, ‘Causal Descriptivism’, Australasian Journal of Philosophy
65: 1–17.
Leonardi, P. and Santambrogio, M. 1995, On Quine: New Essays,
Cambridge: Cambridge University Press.
Lewis, D. 1983, Philosophical Papers, vol. 1, Oxford: Oxford University
Press.
——1984, ‘Putnam’s Paradox’, Australasian Journal of Philosophy 62:
221–36.
——1986, On the Plurality of Worlds, Oxford: Basil Blackwell.
——1994, ‘Reduction of Mind’, in S. Guttenplan (ed.). A Companion to the
Philosophy of Mind, Oxford: Blackwell, 412–31.
——1997, ‘Naming the Colours’, Australasian Journal of Philosophy 75
(3): 325–42.
Loar, B. 1976, ‘The Semantics of Singular Terms’, Philosophical Studies
30: 353–77.
Mackie, P. 2002, ‘Deep Contingency and Necessary A posteriori Truth’,
Analysis 62 (3): 225–36.
——2008, ‘Transworld Identity’, in Edward N. Zalta (ed.). The Stanford
Encyclopedia of Philosophy (Fall 2008 Edition), available online at:
http://plato.stanford.edu/archives/fall2008/entries/identity-transworld/.
McCulloch, G. 1989, The Game of the Name, Oxford: Oxford University
Press.
Nelson, M. 2002, ‘Descriptivism Defended’, Noûs 34: 408–36.
Noonan, H. 1979, ‘Rigid Designation’, Analysis 39: 174–82.
——1981, ‘Methodological Solipsism’, Philosophical Studies 40: 269–74.
——1983, ‘The Necessity of Origin’, Mind 92: 1–20.
——1985, ‘Methodological Solipsism: A Reply to Morris’, Philosophical
Studies 48: 285 –90.
——2003, Personal Identity, London: Routledge.
Pettitt, P. 2004, ‘Descriptivism, Rigidified and Anchored’, Philosophical
Studies 118: 323–8.
Plantinga, A. 1978, ‘The Boethian Compromise’, American Philosophical
Quarterly 15: 129–38.
Putnam, H. 1973, ‘Meaning and Reference’, Journal of Philosophy 70:
699–711.
——1975, ‘The Meaning of “Meaning”’, in K. Gunderson (ed.). Minnesota
Studies in the Philosophy of Science VII: Language, Mind and
Knowledge, University of Minnesota Press, 215–71; reprinted 1975 in
Mind, Language, and Reality: Philosophical Papers Volume 2,
Cambridge: Cambridge University Press, 215–71.
Quine, W. 1953, From a Logical Point of View, Cambridge, Mass.: Harvard
University Press, 1953; revised edition 1980.
——1956, ‘Quantifiers and Propositional Attitudes’, Journal of Philosophy
53: 177–87; reprinted in Quine 1966, 185–96.
——1960, Word and Object, Cambridge, Mass.: MIT Press, 1960.
——1966, Ways of Paradox, New York: Random House. Second edition,
enlarged, Cambridge: Harvard University Press, 1976.
——1969, Ontological Relativity and Other Essays, New York: Columbia
University Press.
——1977, ‘Intensions Revisited’, Midwest Studies in Philosophy 2: 5–11.
Reprinted in Quine, Theories and Things, Cambridge, MA: Harvard
University Press, 1981.
——1981, Theories and Things, Cambridge, MA: Harvard University
Press.
——1995, From Stimulus to Science, Cambridge, MA: Harvard University
Press
——2000, ‘Quine’s Responses’, in A. Orenstein and P. Kotatko (eds).
Knowledge, Language and Logic: Questions for Quine, Dordrecht:
Kluwer Academic Publishers.
——2001, ‘Confessions of a Confirmed Extensionalist’, in J. Floyd and S.
Shieh (eds). Future Pasts: the Analytic Tradition in Twentieth Century
Philosophy, Oxford: Oxford University Press, 215–21.
Russell, B. 1905, ‘On Denoting’, Mind 14: 479–93.
——1910–11, ‘Knowledge by Acquaintance and Knowledge by
Description’, Proceedings of the Aristotelian Society, 108–28. Reprinted
 in Mysticism and Logic, London: George Allen and Unwin, 1963, 152–
67.
——1912, The Problems of Philosophy, Oxford: Oxford University Press.
——1918, ‘Lectures on the Philosophy of Logical Atomism’, Monist 28–9:
1918–19. Reprinted in R. Marsh, (ed.), Logic and Knowledge, London:
George Allen and Unwin, 1956, 177–281.
——1956, Logic and Knowledge, R. Marsh (ed.), London: George Allen
and Unwin.
——1963, Mysticism and Logic, London: George Allen and Unwin.
Salmon, N. 1982, Reference and Essence, Oxford: Basil Blackwell.
——2005, Reference and Essence, second edition with added appendices,
Prometheus Books.
Schiffer, S. 1977, ‘Naming and Knowing’, in P. A. French, T. E. Uehling
and H. K. Wettstein (eds). Contemporary Perspectives in the Philosophy
of Language, Minneapolis: University of Minnesota Press, 61–74.
Searle, J. 1958, ‘Proper Names’, Mind 67: 166–73.
——1983, Intentionality, Oxford: Oxford University Press.
Sleigh, R.C. 1968, ‘On a Proposed System of Epistemic Logic’, Nous 2:
391–8.
Soames, S. 1998, ‘The Modal Argument: Wide Scope and Rigidified
Descriptions’, Noûs 22: 1–22.
——2002, Beyond Rigidity: The Unfinished Semantic Agenda of Naming
and Necessity, Oxford: Oxford University Press.
——2003, Philosophical Analysis in the Twentieth Century: Volume 2, the
Age of Meaning, Princeton, NJ: Princeton University Press.
Sosa, D. 2001, ‘Rigidity in the Scope of Russell’s Theory’, Noûs 35: 1–38.
Stanley, J. 1997, ‘Names and Rigid Designation’, in R. Hale and C. Wright
(eds). A Companion to the Philosophy of Language, Oxford: Blackwell.
Strawson, P. 1959, Individuals: an Essay on Descriptive Metaphysics,
London: Methuen.
——1979, ‘Maybes and Might have Beens’, in A. Margalit, (ed.), Meaning
and Use, Dordrecht: Reidel, 229–39.
Whitehead, A. and Russell, B. 1910–13, Principia Mathematica, 3 vols.
Cambridge: Cambridge University Press. Second edition 1925–7.
Wiggins, D. 1980, Sameness and Substance, Oxford: Basil Blackwell.
Wright, C. 2002, ‘The Conceivability of Naturalism’, in T. Szabo and J.
Hawthorne (eds). Conceivability and Possibility, Oxford: Clarendon
Press.
Index

a posteriori knowledge (known on the basis of experience) 2, 39, 220, 221; contingency illusion and
mind-brain identity 210, 211; Naming and Necessity lectures, main contentions 14, 15, 16;
necessary 117, 158–80; philosophical development of Kripke 9–10; theoretical identifications,
necessity of 197, 201, 203
a priori knowledge (knowability independent of experience) 2, 3, 4, 9, 14, 201, 210, 221; contingent
117, 158–80; naming 56, 64, 103
‘A Puzzle about Belief’ (Kripke) 10–11, 166, 170, 178
abstract object/concrete object 5, 131, 221
abstraction principle 204, 221
absurdity, and truth 60
accidental designators 65–66; see also rigid designation
accidental properties of individuals 17, 81, 88, 135; versus essential, 4, 10, 50, 89, 137
actual world 9
Adam (Old Testament) 136, 137, 140, 141
ambiguity 60, 61, 68, 70, 75–76
analogous principle 80
analytic philosophy 2–5, 39, 221
analytic truth 4, 124, 221
Aristotle/Aristotle example 3, 13, 24–25, 37, 64, 83, 84, 96, 97; essentialism, 4, 40, 124, 125, 131
assertoric content 67, 75, 77, 222
attributive use 10, 222

baptism 36, 119, 193; initial 8, 103, 104, 117, 118, 186, 195
Barrett, R. 52
Bible 56, 57
British Empiricism 3, 5
Burge, T. 67, 89, 90
Burgess, J. 98

Carnap, R. 2
Cartesian dualism 213
causal-historical theory of reference 10, 11, 12, 36
C-fibre stimulation 209, 216, 217, 219, 220
Chalmers, D. 94–95
Chisholm, R. 136, 137
Cicero example 11, 84, 85, 90, 91, 93, 103, 104, 108–11, 170, 171, 178
circularity constraint 56, 57–58, 111–12, 114
cluster of identifying marks theory 20, 183–86, 188–92, 195, 196
cluster theory 37, 56, 58, 63, 222; arguments against 102–16; counterexamples 110; and reference-
fixing 102–16
co-denoting descriptions 83
cognitive value 23, 222
compositionality principle 21, 22, 222
concrete objects 5, 131, 204–5, 221
connotation 15, 222
context-dependence 26–27, 133
context-relativity, of essence 128, 134
contingency, illusion of 209–20
‘The Contingent A Priori and Rigid Designators’ (Donnellan) 72
contingent properties 17
coreferential singular terms 25
counterfactual situation 9
counterpart theory 157, 222; critique 141–43

Davies, M. 95
de dicto (of the saying) 4, 222–23; propositional attitude 49, 51
de facto rigidity 62, 63
de jure rigidity 62, 63, 94, 223
de re (of the thing) 222–23; de re modality 4, 40, 42, 46, 50, 53, 123, 222; opacity 128; predications
130; propositional attitudes 49, 50–51, 53
deferential reference 57
demonstrative elements 38, 223
denotation 15, 223
descriptions 6, 24, 29, 83; abbreviations for 72–73; cluster theory see cluster theory; Frege–Russell
theory 6, 10–11, 13, 14, 18, 19, 36, 119, 190; indicative versus subjunctive 67–68, 98–99, 101;
and names 27–36, 68; primary and secondary occurrences 32, 130, 225
descriptivism 15, 94, 190, 191, 192, 195, 196; rigidified 95, 98, 114
direct quotation 26
direct reference theorists 14–15, 15
distinctness pairs 156
Donnellan, K. 6, 10, 11, 38, 106, 175, 176, 222; ‘The Contingent A Priori and Rigid Designators,’ 72
Dummett, M. 67, 68–78, 80, 90, 100, 115, 121, 144, 145, 159, 171, 172, 189

empiricism 2, 3, 223
empty singular terms 28, 30
epistemic necessity 179, 223
epistemological arguments 11
epistemology 34
error 8, 11, 13, 19
essence 128, 134, 223
essence-revealing terms 47, 52, 87, 88, 89
essential properties of individuals 135, 143–58, 223; versus accidental 4, 10, 50, 89, 137
essentialism 17, 39, 136, 148, 155, 157, 200, 207, 223; Aristotelian 4, 40, 124, 125, 131;
intelligibility of 123–35
Evans, G. 10, 66, 95, 120, 121, 122
exportation 49, 50, 53, 86, 87, 223
extensionalism/extensionality 40–41, 43, 44
extension/intension 6, 7, 181–220, 223–24
externalism/internalism 6, 8, 11, 224

Fitch, G. W. 59
flexible (non-rigid) designators 64, 73–74, 118, 131, 189
Franklin, B/Franklin as inventor of bifocals example 60, 61, 65, 66, 67, 69, 70, 74, 78, 79, 80, 87, 88,
90, 95
Frege, G. 2, 3, 20, 85, 88, 226;
Frege–Russell description theory 6, 10–11, 13, 14, 18, 19, 28, 36, 119, 190; Frege–Russell logistic
programme (mathematics) 3; Frege–Russell synthesizers 37–39; Frege–Russell theory of proper
names 8; ‘King of France is bald’ example 29, 30, 31; and Quine 46; on sense 5, 21–27; ‘On
Sense and Reference’ 22, 24
From Stimulus to Science (Quine) 52
functionalism 216
functionality principle 21

Geach, P. 86
Gibbard, A. 128, 131 ‘
Giorgione’/‘Barbarelli’ example (Quine) 127–28
Goliath see Lumpl (lump of clay) and Goliath example
Great Western Railway (GWR) example 129–30 Grundgesetze (Frege) 3 Grundlagen (Frege) 3, 204

Hale, R. V. 89, 90
heat 196–98, 203, 209, 215–19, 220
‘here’ and ‘where I am’ distinction 76–77
‘Hesperus is Phosphorus’ example 14, 15, 160, 162–64, 166, 167, 179, 180, 182, 197–200, 210, 212
historical chain picture 116–202
Hobbes, T. 149, 150, 153, 154
Hughes, C. 177, 219
Humberstone, I. L. 95
Hume, D. 3

identity: identity-revealing aspects 87; mind-brain 18, 209–20; ‘misleading metaphor’ of 59; names
of individuals 14, 15; necessity of 225; puzzle of 26, 31, 42; theoretical identifications, necessity
of 196–209; token-token identity theory 213, 214, 215, 226; transworld see transworld identity;
type-type identity theory 213, 216, 220, 226
‘Identity and Necessity’ (Kripke) 9, 10, 141–42, 156, 160, 196–97
identity statements 9, 24, 26, 27, 196
ignorance 8, 11, 13, 19
incomplete symbols 35
indexical expressions 13, 38, 54, 176, 224
indicative descriptions 67–68, 98–99, 101
Individuals (Strawson) 6, 38
ingredient sense 67, 75, 77, 222
initial baptism 8, 103, 104, 117, 118, 186, 195
intension 47
internalism 8, 11, 192, 224
intuition 126
island names 120, 121

‘Jack the Ripper’ example 66, 70–73, 102, 110, 171, 187
Jackson, F. 94
John Locke lectures ‘Reference and Existence’ 1973, 16
Jonah (Old Testament) 109
Julius (inventor of the zip) 66, 171

Kafka, F. 144
Kaplan, D. 50, 53, 87, 94
knowledge: logical construction of 2–3; a posteriori see a posteriori knowledge (known on the basis
of experience); a priori see a priori knowledge (knowability independent of experience)
Kripke, S.: ‘Identity and Necessity’ 9, 10, 141–42, 156, 160, 196–97; philosophical development 8–
11; ‘A Puzzle about Belief’ 10–11, 166, 170, 178; ‘Speaker’s Reference and Semantic Reference’
10, 107; see also Naming and Necessity lectures, 1970
Kroon, F. 114

language, individual users 118

Law of Excluded Middle 31, 224
Lectures on Logical Atomism (Russell) 33–34
Leibniz’s Law 42, 157
Leonardi, P. 51
Lewis, D. 9, 75, 91, 114, 123, 127–34, 140, 142, 145, 157, 208, 216, 219, 222; ‘Putman’s Paradox’
84
Loar, B. 68
Locke, J. 5
logical necessity 52
logical positivists 2
London example 92, 167, 168, 169, 170
Lumpl (lump of clay) and Goliath example 128–29, 131, 135, 215

Mackie, P., 61 139

Madagascar example 120, 121, 122
‘Mars is not Venus’ example 15, 169
Marsh, O.C. 199
McCulloch, G. 68
metaphysics 3, 34, 179; metaphysical necessity/possibility 2, 212, 224
Metaphysics (Aristotle) 25
methodological solipsism 7, 224
metre rod example 63–64, 171–72, 174, 175–76, 202–3, 205
Mill, J. S. 19, 20, 22, 183
Millianism 15, 19, 20
mind–brain identity 18, 209–20
modal argument 12, 13, 19, 65–67; debate over 67–102; Preface refinement 77–78, 91, 100
modal logic 8, 123, 224
modal predication 128
modal realism 131, 224
modality 39, 132, 224; de re modality 4, 40, 42, 46, 53, 123, 222; degrees of modal involvement
(Quine) 41–42, 44, 45, 124; modal argument 65–102
‘molecular motion’ example 197, 198, 199, 203, 209, 216, 217, 218
Moses (Old Testament) 56, 57

naming: ambiguity 60, 61, 68, 70, 75–76; descriptions and names 27–36, 68; fiction versus
substantially false account of real person 56; Franklin as inventor of bifocals example 60, 61, 65,
66, 67, 69, 70, 74, 78, 79, 80, 87, 88, 90, 95; ‘here’ and ‘where I am’ distinction 76–77; modal
 argument 65–102; names as rigid designators or abbreviations for descriptions 72–73; rigid
designation 58–65, 223; target 54–58; ‘Thatcherabouts’ example 77, 78, 79–80, 82, 96; thesis 58,
68; vivid names 50, 52, 87; see also proper names
Naming and Necessity lectures 1970, 1, 2, 6, 8, 9, 164; Addenda 16, 17, 18; Appendix 200; main
contentions 11–20; Preface 15, 59, 62, 67, 73, 77–78, 80, 91, 100, 140, 155; supplementary
publications following 10; see also Kripke, S.
narrow psychological state 7, 8, 224
natural and biological sciences 2, 3
natural kind terms 2, 12–13, 224–25; as proper names of kinds 181–96; as rigid designators 13;
subject–predicate sentences 15
natural phenomena 181
‘nec,’ as sentence operator 42, 44, 45, 48
necessity 17, 39, 40, 123–35, 223, 225; and analyticity 3, 4; metaphysical 2, 224; of origin 139, 151,
225; of theoretical identifications 196–209
Nelson, M. 80
Neptune/Uranus example 72–73, 176, 177
Nixon, R./Nixon example 126, 137, 139–40, 142
Noah (Old Testament) 136–37, 140
non-connotative terms 222
non-extensionality 43, 45
non-modal sentences/statements 68, 74–75
non-trivial sense 4, 123, 126, 127, 134, 141, 143, 217; essential and accidental properties, non-trivial
distinction 49–50; and Quine 39, 45, 46, 49–50, 126
Noonan, H. 147, 153

of-the-same-marital-kind example 205–6, 207, 208

‘On Denoting’ (Russell) 3, 28
‘On Sense and Reference’ (Frege) 22, 24
‘On Three Grades of Modal Involvement’ (Quine) 41
operator 225
ordinary proper names 5, 14, 19–20, 24, 225; Frege–Russell synthesizers 37–38; modal argument
debate 67, 68; and Russell 28, 33, 35, 36; see also naming; proper names
origin, necessity of 139, 225

pain 209, 214, 216, 217, 218, 219, 220

Pettitt, P. 95
Philosophical Troubles (Kripke) 16
pi 94
Plantinga, A. 94
Plato 64
possibility 3, 223; possible worlds 9, 12, 59, 225
predicate terms 15–16; and statement operators 41–42
primary occurrences 32, 130, 225
Princeton University, Naming and Necessity lectures see Naming and Necessity lectures, 1970
Principia Mathematica (Russell and Whitehead) 3
proper names 2, 5, 8, 19, 24, 36, 170, 225; description theory 28; and descriptions 29, 35; Feynman
and Gödel examples 19, 104–8, 110–13, 119, 122; in logical sense, 19–20; natural kind terms as
proper names of kinds 181–96; non-genuine 35–36; ordinary see ordinary proper names; as rigid
designators 12, 13, 15, 20, 67, 75; see also naming; rigid designation
‘Proper Names’ (Searle) 37
propositional attitudes 10, 25, 26, 47, 225; de dicto 49, 51; de re 49, 50–51, 53
propositional functions 30, 33
‘Propositional Objects’ (Quine) 136
propositions 34, 48
psychological states 6–7, 8, 224
public language 23, 35; Russell’s rejection of 33, 34
‘Putman’s Paradox’ (Lewis) 84
Putnam, H. 6–8, 11, 20, 181, 224; Twin Earth argument 7–8, 13, 185, 186, 188, 189, 191, 192, 193,
195, 196
puzzle of identity 26, 31, 42

quantified modal statements 132

‘Quantifiers and Propositional Attitudes’ (Quine) 47
quantifying expressions 29, 30, 31
quantifying into 40, 225
Quine, W. 2, 3, 4, 17, 20, 39–53, 60, 86, 88–89,123,124, 131,134–35, 163; degrees of modal
involvement 41–42, 44, 45, 124; ‘Giorgione’/‘Barbarelli’ example 127–28; ‘Propositional
Objects’ 136; ‘Quantifiers and Propositional Attitudes’ 47; From Stimulus to Science 52; ‘On
Three Grades of Modal Involvement’ 41; Word and Object 125

reference 1, 4, 11–12; causal-historical theory 10, 11, 12, 36; dependence on context 26–27, 133;
sense distinguished 5, 21, 25; see also sense/reference
‘Reference and Definite Descriptions’ (Donnellan) 10
reference shift 120, 121
reference-determination 115
reference-fixing 58, 63, 94, 102–16, 117, 118, 186; and cluster theory 102–16; versus synonym-
giving 64; see also cluster theory
referential opacity/transparency 39–40, 43, 225
referential use 10, 222
rigid designation 17, 53, 212, 223, 226; concept of rigidity 16, 18; naming 58–65, 89–90, 93;
necessity 131, 157, 162; and non-rigid (flexible) designators 64, 73–74, 118, 131, 189; proper
names as rigid designators 12, 13, 15, 20, 67, 75; strongly rigid designators 61–62; see also de
facto rigidity; de jure rigidity
rigidified descriptivism 95, 98, 114
Russell, B. 5, 20, 27–36, 101, 130;‘On Denoting’ 3, 28; Frege–Russell description theory 6, 10–11,
13, 14, 18, 19, 28, 36, 119, 190; Frege–Russell logistic programme (mathematics) 3; Frege–
Russell synthesizers 37–39; Frege–Russell theory of proper names 8; Lectures on Logical
Atomism33–34; logical construction of knowledge 2–3; public language, rejection of 33, 34

Salmon, N. 14, 61, 148, 205, 207, 208

salva veritate (truth-value) 13, 14, 65–66, 226
Santambrogio, M. 51
sceptics 148, 159, 161, 165–66, 168, 169, 174
Schiffer, S. 68
scope 10, 30, 31, 68, 226
Scott, Walter/Scott as author of Waverley example 32, 59–60, 85–86, 87, 91–93
Searle, J. 37, 38, 54, 67, 94, 115
secondary occurrences 32, 130, 225
semantic arguments 11
semantic reference, 10; versus speaker’s reference 108
semantic value 21, 226
semantical predicates 42, 43
Semantics of Natural Languages 6
sense/reference 21–27, 226; definition of sense 23; distinction between 5, 21, 25; Elizabeth II
example 22; Evening Star/Morning Star examples 25–26, 27; non-trivial sense see non-trivial
sense
sentence operators 42, 44, 45, 48
Sherlock Holmes stories 17
singular terms 23, 25, 28, 42; empty 28, 30; essence-revealing 47, 52, 87, 88, 89
Sleigh, R.C. 50
Soames, S. 80, 81, 83, 95–96, 98, 174
Sosa, D. 75, 80
speaker’s mind/speaker’s reference 5, 10, 108
‘Speaker’s Reference and Semantic Reference’ (Kripke) 10, 107
Sprigge, T. 143
‘squatchelor’ example 173, 175
St Anne as mother of Virgin Mary example 66, 69–73, 102, 171, 172, 173, 175
Stanley, J. 80, 95
statement operators 41–42
Strawson, P. 6, 37, 38–39, 54, 57, 110, 144, 145
subject–predicate sentences 15–16, 29
subjunctive conditional statements 12
subjunctive descriptions 68, 98–99, 101
sufficiency of origin principle 147, 149, 152
synonym-giving, versus reference-fixing 64
synthetic truth 221

table of ice/wood example 211–12, 214

temporal analogues 81–82
‘Thatcherabouts’ example 77, 78, 79–80, 82, 96
Theseus, ship of 149, 150, 153, 154
thesis 58, 68
‘all tigers are mammals’ example 16, 182, 200, 202, 203, 204, 206, 208, 209
token–token identity theory 213, 214, 215, 226
transparency of sense 24
transworld identity 151–52, 154; rejection of problem 135–40
truth: and absurdity 60; analytic 4, 124, 221; necessary 4; salva veritate (truth-value) 13, 14, 65–66,
226; in sentences 29; synthetic 221; true beliefs, identifying 19
truth-values 21, 22, 40–41, 45, 83, 89; see also salva veritate (truth-value)
Twin Earth argument (Putnam) 7–8, 13, 185, 186, 188, 189, 191, 192, 193, 195, 196
type–type identity theory 213, 216, 220, 226

unicorns 17

verifiability principle 2
Vienna Circle 2, 3
vivid names 50, 52, 87

‘water is H2O’ example 8, 15, 16, 18, 181, 182, 185, 196, 197, 198, 199, 209, 210, 216
Whitehead, A.N. 3
widescopism: and naming 68, 69, 73, 74, 80–85, 90, 91; and necessity 159, 161–62
Wiggins, D. 145
Wittgenstein, L. 2, 3, 56, 63–64, 183
Word and Object (Quine) 125
worldmates 131
world-user’s beliefs and intentions 11, 12
Wright, C. 220