Journal of Philosophy, Inc.

Meaning and Reference

Author(s): Hilary Putnam
Source: The Journal of Philosophy, Vol. 70, No. 19, Seventieth Annual Meeting of the
American Philosophical Association Eastern Division (Nov. 8, 1973), pp. 699-711
Published by: Journal of Philosophy, Inc.
Stable URL: http://www.jstor.org/stable/2025079 .
Accessed: 30/01/2011 06:05

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=jphil. .

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Journal of Philosophy, Inc. is collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Philosophy.

http://www.jstor.org
 REFERENCE 699
He wantsto say that it's not a thing.As a matterof fact,it's malt
Scotch, one of the thingsI particularlydislike. And it's the only
thingI got for Chris-tmas (not the bottle,the Scotch: it came in a
different bottle). It is not a matterof wieldingthe OED or theDaily
News here to establish anythingmomentous;I'm only asking why
Chappell wantsus to talk his way.He thinksthatif we ask,What is
the Scotchin thatbottle identicalwith?we won't be able to answer
ourselves,and so will give up thinkingit's a thing.But the Scotch
in that bottle is (the same as) the whiskeyin thatbottle,the liquid
in that bottle,the stuffI'm about to give you because I can't stand
it. (I don't see that the question,What is this copy of the JOURNAL
OF PHILOSOPHY identical with? gets answers that are any better
than those.)
And he wants to say that it constitutesa thing,even when you
mix it with soda. He is not contentwith the idea that the thing
of which it is a part is Scotch and soda, or a glass of Scotch and
soda; I don't know why. He wants to say that the Scotch in that
Scotch and soda constitutesor composessomethingall its own, and
so he adopts the device of prefixing'heap of' or 'aggregate of'
to words for stuffs,to guarantee that we'll always have (a count-
noun phrase for?)a thingany stuffcomposes.And I don't see that:
if we are so sure that therealwayswill be a thingcomposedof any
stuffwe encounter,we mightas well just prefix'thingcomposedof'
and be done withit. And, anyway,whyshould we be so sure?
RUSSELL M. DANCY
CornellUniversity

U
MEANING AND REFERENCE *

NCLEAR as it is, thetraditional

doctrinethatthenotion
"meaning" possesses the extension/intensionambiguity
has certain typical consequences. The doctrine that the
meaning of a termis a concept carried the implicationthat mean-
ings are mental entities.Frege,however,rebelled against this "psy-
chologism." Feeling that meanings are public property-that the
same meaning can be "grasped" by more than one person and
* To
be presented in an APA symposium on Reference, December 28, 1973.
Commentatorswill be Charles Chastain and Keith S. Donnellan; for Donnellan's
paper, see this JOURNAL, this issue, 711-712; ProfessorChastain's comments are
not available at this time.
A very much expanded version of this paper will appear in volume 7 or 8
of Minnesota Studies in the Philosophy of Science (edited by Keith Gunderson),
under the title"The Meaning of 'Meaning'."
700 THE JOURNAL OF PHILOSOPHY

by persons at differenttimes-he identifiedconcepts (and hence

"intensions"or meanings)with abstractentitiesratherthan mental
entities. However, "grasping" these abstract entities was still an
individual psychologicalact. None of these philosophersdoubted
thatunderstandinga word (knowingits intension)was just a matter
of being in a certainpsychologicalstate (somewhatin the way in
which knowinghow to factornumbersin one's head is just a mat-
ter of being in a certain verycomplex psychologicalstate).
Secondly,the timewornexample of the two terms'creaturewith
a kidney'and 'creaturewith a heart' does show that two termscanl
have the same extension and yet differin intension. But it was
taken to be obvious that the reverseis impossible: two termscan-
not differin extensionand have the same intension.Interestingly,
no argumentfor this impossibilitywas ever offered.Probably it
reflectsthe traditionof the ancientand medieval philosophers,who
assumed that the concept correspondingto a termwas just a con-
junction of predicates,and hence that the conceptcorrespondingto
a termmust always provide a necessaryand sufficient conditionfor
fallinginto the extensionof the term.For philosopherslike Carnap,
who accepted the verifiability theoryof meaning,the concept cor-
responding to a term provided (in the ideal case, where the
term had "complete meaning") a criterionfor belonging to the
extension(not just in the sense of "necessaryand sufficient condi-
tion," but in the strong sense of way of recognizingwhether a
given thingfalls into the extensionor not). So theoryof meaning
came to reston twounchallengedassumptions:
(1) That knowingthe meaningof a termis just a matterof being
in a certain psychological state (in the sense of "psychological
state," in which states of memoryand belief are "psychological
states"; no one thoughtthat knowingthe meaning of a word was
a continuousstateof consciousness,of course).
(2) That the meaning of a termdeterminesits extension(in the
sense that samenessof intension entails samenessof extension).
I shall argue thatthesetwo assumptionsare not jointlysatisfiedby
any notion,let alone any notion of meaning.The traditionalcon-
cept of meaning is a conceptwhich restson a false theory.
ARE MEANINGS IN THE HEAD?
For the purpose of the followingscience-fiction
examples,we shall
suppose that somewherethereis a planet we shall call Twin Earth.
Twin Earth is verymuch like Earth: in fact,people on Twin Earth
even speak English. In fact, apart from the differenceswe shall
specifyin our science-fictionexamples, the reader may suppose
 REFERENCE 70I

that Twin Earth is exactlylike Earth. He may even suppose that

he has a Doppelganger-an identical copy-on Twin Earth, if
he wishes,althoughmy storieswill not depend on this.
Although some of the people on Twin Earth (say, those who
call themselves"Americans"and thosewho call themselves"Canadi-
ans" and those who call themselves"Englishmen," etc.) speak
English, thereare, not surprisingly, a few tinydifferences between
the dialects of English spoken on Twin Earth and standard
English.
One of the peculiaritiesof Twin Earth is that the liquid called
"water" is not H20 but a different liquid whose chemical formula
is very long and complicated. I shall abbreviate this chemical
formulasimplyas XYZ. I shall suppose that XYZ is indistinguish-
able fromwater at normal temperaturesand pressures.Also, I shall
suppose that the oceans and lakes and seas of Twin Earth contain
XYZ and not water, that it rains XYZ on Twin Earth and not
water,etc.
If a space ship from Earth ever visits Twin Earth, then the
suppositionat firstwill be that 'water' has the same meaning on
Earth and on Twin Earth. This suppositionwill be correctedwhen
it is discoveredthat "water" on Twin Earth is XYZ, and the Earth-
ian space ship will reportsomewhatas follows.
"On Twin Earth the word 'water' means XYZ."
Symmetrically,if a space ship from Twin Earth ever visits
Earth,then the suppositionat firstwill be that the word 'water'has
the same meaning on Twin Earth and on Earth. This supposition
will be correctedwhen it is discoveredthat "water" on Earth is
H20, and the Twin Earthian space ship will report:
"On Earth the word 'water'means H20."
Note that there is no problem about the extensionof the term
'water': the word simplyhas two different meanings(as we say); in
the sense in which it is used on Twin Earth, the sense of waterTE,
what we call "water"simplyisn'twater,while in the sense in which
it is used on Earth, the sense of waterE,what the Twin Earthians
call "water" simple isn't water. The extension of 'water' in the
sense of waterE is the set of all wholes consistingof 1120 mole-
cules,or somethinglike that; the extensionof waterin the sense of
waterTEis the set of all wholes consistingof XYZ molecules, or
somethinglike that.
Now let us roll the timeback to about 1750.The typicalEarthian
speakerof English did not know that water consistedof hydrogen
and oxygen,and the typicalTwin-Earthianspeakerof English did
702 THE JOURNAL OF PHILOSOPHY

not know that "water" consistedof XYZ. Let Oscar1 be such a

typicalEarthian English speaker,and let Oscar2be his counterpart
on Twin Earth.You may suppose that thereis no belief that Oscar,
had about water that Oscar2 did not have about "water." If you
like,you may even suppose thatOscar1and Oscar2wereexact dupli-
cates in appearance, feelings,thoughts,interior monologue, etc.
Yet the extension of the term 'water' was just as much H20 on
Earth in 1750 as in 1950; and the extensionof the term'water'was
just as much XYZ on Twin Earth in 1750 as in 1950. Oscar1 and
Oscar2understoodthe term'water'differently in 1750 althoughthey
werein thesame psychologicalstate,and although,giventhestateof
scienceat the time,it would have taken theirscientificcommunities
about fiftyyears to discoverthat theyunderstoodthe term 'water'
differently.
Thus the extensionof the term'water' (and, in fact,its
"meaning" in the intuitivepreanalyticalusage of that term)is not
a functionof the psychologicalstate of the speakerby itself.'
But, it might be objected, why should we accept it that the
term'water' had the same extensionin 1750 and in 1950 (on both
Earths)?Suppose I point to a glass of water and say "this liquid is
called water." My "ostensivedefinition"of water has the following
empiricalpresupposition:that the body of liquid I am pointingto
bears a certain samenessrelation (say, x is the same liquid as y,
or x is the sameL as y) to most of the stuffI and otherspeakersin
my linguisticcommunityhave on other occasions called "water."
If this presuppositionis false because, say, I am-unknown to me
-pointing to a glass of gin and not a glass of water,then I do not
intend my ostensivedefinitionto be accepted. Thus the ostensive
definitionconveyswhat might be called a "defeasible" necessary
and sufficient condition: the necessaryand sufficient condition for
being water is bearing the relationsameL to the stuffin the glass;
but this is the necessaryand sufficient condition only if the em-
pirical presuppositionis satisfied.If it is not satisfied,then one of a
seriesof,so to speak, "fallback" conditionsbecomesactivated.
The key point is that the relationsameL is a theoreticalrelation:
whethersomethingis or is not the same liquid as thismay take an
indeterminate amount of scientificinvestigation to determine.
Thus, the fact that an English speaker in 1750 mighthave called
XYZ "water," whereas he or his successorswould not have called
XYZ water in 1800 or 1850 does not mean that the "meaning" of
'water' changed for the average speaker in the interval. In 1750
1 See fn 2, p. 710 below, and the correspondingtext.
 REFERENCE 703

or in 1850 or in 1950 one mighthave pointed to, say, the liquid in

Lake Michigan as an example of "water." What changed was that
in 1750 we would have mistakenlythoughtthat XYZ bore the rela-
tion sameL to the liquid in Lake Michigan, whereas in 1800 or
1850we would have knownthatit did not.
Let us now modifyour science-fiction story.I shall suppose that
molybdenumpots and pans can't be distinguishedfromaluminum
pots and pans save by an expert.(This could be trueforall I know,
and, a fortiori,it could be trueforall I knowby virtueof "knowing
the meaning" of the words aluminum and molybdenum.)We will
now suppose that molybdenumis as common on Twin Earth as
aluminum is on Earth, and that aluminum is as rare on Twin
Earth as molybdenumis on Earth. In particular,we shall assume
that "aluminum" pots and pans are made of molybdenumon
Twin Earth. Finally, we shall assume that the words 'aluminum'
and 'molybdenum'are switchedon Twin Earth: 'aluminum' is the
name of molybdenum,and 'molybdenum'is the name of aluminum.
If a space ship fromEarth visited Twin Earth, the visitorsfrom
Earth probably would not suspect that the "aluminum" pots and
pans on Twin Earth were not made of aluminum,especiallywhen
the Twin Earthianssaid theywere. But thereis one importantdif-
ference between the two cases. An Earthian metallurgistcould
tell very easily that "aluminum" was molybdenum,and a Twin
Earthian metallurgistcould tell equally easily that aluminum
was "molybdenum." (The shudder quotes in the preceding sen-
tence indicate Twin Earthian usages.) Whereas in 1750 no one
on eitherEarth or Twin Earth could have distinguishedwaterfrom
"water," the confusion of aluminum with "aluminum" involves
only a part of the linguisticcommunitiesinvolved.
This example makes the same point as the precedingexample.
If Oscar, and Oscar2are standardspeakersof Earthian English and
Twin Earthian English, respectively,and neitheris chemicallyor
metallurgicallysophisticated,then there may be no differenceat
all in theirpsychologicalstateswhen theyuse the word 'aluminum';
nevertheless,we have to say that 'aluminum' has the extension
aluminum in the idiolect of Oscar, and the extensionmolybdenum
in the idiolectof Oscar2.(Also we have to say thatOscar1and Oscar2
mean differentthingsby 'aluminum'; that 'aluminum' has a dif-
ferentmeaning on Earth than it does on Twin Earth, etc.) Again
we see that the psychologicalstate of the speaker does not deter-
mine the extension (or the "meaning," speaking preanalytically)
of theword.
704 THE JOURNAL OF PHILOSOPHY

Before discussingthis example further,let me introducea non-

science-fictionexample. Suppose you are like me and cannot
tell an elm froma beech tree.We still say thattheextensionof 'elm'
in my idiolect is the same as the extensionof 'elm' in anyone else's,
viz., the set of all elm trees,and that the set of all beech treesis
the extensionof 'beech' in both of our idiolects.Thus 'elm' in my
idiolecthas a different extensionfrom'beech' in your idiolect (as it
should). Is it really credible that this differencein extension is
brought about by some differencein our concepts? My concept
of an elm tree is exactly the same as my concept of a beech tree
(I blush to confess).If someoneheroicallyattemptsto maintainthat
the difference between the extensionof 'elm' and the extensionof
'beech' in my idiolect is explained by a differencein my psycho-
logical state, then we can always refute him by constructinga
"Twin Earth" example-just let the words 'elm' and 'beech' be
switchedon Twin Earth (the way 'aluminum' and "molybdenum'
were in the previous example). Moreover, suppose I have a
Doppelganger on Twin Earth who is molecule formolecule "iden-
tical" with me. If you are a dualist, then also suppose my Doppel-
gangerthinksthe same verbalizedthoughtsI do, has the same sense
data, the same dispositions,etc. It is absurd to think his psycho-
logical state is one bit differentfrommine: yet he "means" beech
when he says "elm," and I "mean" elm when I say "elm." Cut the
pie any way you like, "meanings"just ain't in the head!
A SOCIOLINGUISTIC HYPOTHESIS
The last two examples depend upon a fact about language that
seems,surprisingly, never to have been pointed out: that there is
divisionof linguisticlabor. We could hardlyuse such wordsas 'elm'
and 'aluminum' if no one possesseda way of recognizingelm trees
and aluminum metal; but not everyoneto whom the distinction
is importanthas to be able to make the distinction.Let us shiftthe
example; considergold. Gold is importantformanyreasons: it is a
precious metal; it is a monetarymetal; it has symbolicvalue (it
is importantto mostpeople that the "gold" weddingring theywear
really consist of gold and not just look gold); etc. Consider our
communityas a "factory":in this "factory"some people have the
"job" of wearing gold wedding rings; other people have the
"job" of selling gold wedding rings; still other people have the
job of telling whetheror not somethingis reallygold. It is not at
all necessaryor efficient that everyone who wears a gold ring (or a
gold cufflink, etc.), or discusses the "gold standard," etc., engage
in buyingand sellinggold. Nor is it necessaryor efficient that every
 REFERENCE 705

one who buys and sells gold be able to tell whetheror not some-
thing is really gold in a societywhere this formof dishonestyis
uncommon(selling fake gold) and in which one can easily consult
an expertin case of doubt. And it is certainlynot necessaryor effi-
cient thateveryone who has occasion to buy or wear gold be able to
tell with any reliabilitywhetheror not somethingis really gold.
The foregoingfacts are just examples of mundane division of
labor (in a wide sense). But theyengendera division of linguistic
labor: everyone to whom gold is importantfor any reason has to
acquire the word 'gold'; but he does not have to acquire the method
of recognizingwhethersomethingis or is not gold. He can relyon a
special subclassof speakers.The featuresthat are generallythought
to be present in connectionwith a general name-necessary and
sufficientconditions for membership in the extension, ways of
recognizingwhethersomethingis in the extension,etc.-are all
presentin the linguisticcommunityconsideredas a collectivebody;
but that collectivebody divides the "labor" of knowing and em-
ploying these various parts of the "meaning" of 'gold'.
This division of linguisticlabor restsupon and presupposesthe
division of nonlinguisticlabor, of course. If only the people who
know how to tell whethersome metal is really gold or not have
any reason to have the word 'gold' in their vocabulary,then the
word 'gold' will be as the word 'water' was in 1750 with respect
to that subclass of speakers, and the other speakers just won't
acquire it at all. And some words do not exhibit any division of
linguistic labor: 'chair', for example. But with the increase of
division of labor in the societyand the rise of science,more and
more words begin to exhibit this kind of division of labor.
'Water', for example, did not exhibit it at all before the rise of
chemistry.Today it is obviouslynecessaryfor everyspeaker to be
able to recognize water (reliably under normal conditions), and
probably most adult speakers even know the necessaryand suffi-
cientcondition"wateris H20," but only a fewadult speakerscould
distinguishwater fromliquids that superficiallyresembledwater.
In case of doubt, other speakers would rely on the judgment of
these "expert" speakers.Thus the way of recognizingpossessedby
these "expert" speakers is also, through them, possessed by the
collectivelinguisticbody, even though it is not possessedby each
individual memberof the body,and in thisway the mostrecherchd
factabout watermaybecomepart of thesocial meaningof the word
although unknown to almost all speakers who acquire the word.
It seems to me that this phenomenon of division of linguistic
7o6 THE JOURNAL OF PHILOSOPHY

to
labor is one that it will be very importantfor sociolingu-istics
investigate.In connection with it, I should like to propose the
followinghypothesis:
HYPOTHESIS OF THE UNIVERSALITY OF THE DIVISION OF LINGUISTIC LABOR:
Everylinguisticcommunity exemplifies the sortof divisionof lin-
guisticlabor just described;thatis, it possessesat least someterms
whoseassociated"criteria"are knownonlyto a subsetof thespeakers
who acquiretheterms, and whoseuse by theotherspeakersdepends
upon a structured cooperationbetweenthemand thespeakersin the
relevantsubsets.
It is easy to see how this phenomenonaccounts for some of the
examples given above of the failure of the assumptions(1 and 2).
When a termis subject to the divisionof linguisticlabor, the "aver-
age" speaker who acquires it does not acquire anythingthat fixes
its extension.In particular,his individual psychologicalstate cer-
tainlydoes not fix its extension;it is only the sociolinguisticstate
of the collectivelinguisticbody to which the speaker belongs that
fixestheextension.
We may summarizethisdiscussionby pointingout that thereare
two sortsof tools in the world: thereare tools like a hammeror a
screwdriverwhich can be used by one person; and thereare tools
like a steamshipwhich require the cooperativeactivityof a number
of persons to use. Words have been thoughtof too much on the
model of the firstsort of tool.
INDEXICALITY AND RIGIDITY
The firstof our science-fictionexamples-'water' on Earth and on
Twin Earth in 1750-does not involve division of linguisticlabor,
or at least does not involve it in the same way the examples of
'aluminum' and 'elm' do. There were not (in our story,anyway)
any "experts" on water on Earth in 1750, nor any experts on
"water" on Twin Earth. The example does involve thingswhich
are of fundamentalimportanceto the theoryof referenceand also
to the theoryof necessarytruth,which we shall now discuss.
Let W1 and W2 be two possible worlds in which I exist and
in which this glass exists and in which I am giving a meaning
explanation by pointing to this glass and saying "This is water."
Let us suppose that in W1 the glass is full of H20 and in W2 the
glass is full of XYZ. We shall also suppose that W1 is the actual
world, and that XYZ is the stufftypicallycalled "water" in the
world W2 (so that the relationbetweenEnglish speakersin W1 and
English speakersin W2 is exactlythe same as the relationbetween
English speakerson Earth and English speakers on Twin Earth).
 REFERENCE 707

Then thereare two theoriesone mighthave concerningthe mean-

ing of 'water':
(1) One mighthold that 'water' was world-relativebut constant
in meaning (i.e., the word has a constantrelative meaning). On
this theory,'water' means the same in W1 and W2; it's just that
wateris H20 in W1,and water is XYZ in W2.
(2) One might hold that water is H20 in all worlds (the stuff
called "water" in W2 isn't water),but 'water'doesn't have the same
meaningin W1 and W2.
If what was said beforeabout the Twin Earth case was correct,
then (2) is clearly the correcttheory.When I say "this (liquid) is
water,"the "this" is, so to speak,a de re "this"-i.e., the forceof my
explanation is that "water" is whateverbears a certainequivalence
relation (the relation we called "sameL" above) to the piece of
liquid referredto as "this" in theactual world.
We might symbolizethe differencebetween the two theoriesas
a "scope" difference in the followingway. On theory(1), the follow-
ing is true:
(1') (ForeveryworldW) (For everyx in W) (x is water_ x bearssameL
to theentityreferredto as "this"in W)
while on theory(2):
(2') (ForeveryworldW) (Foreveryx in W) (x is water-x bearssameL
referred
to theentity to as "this"in theactualworldW1)
I call this a "scope" differencebecause in (1') 'the entityreferred
to as "this"' is within the scope of 'For everyworld W'-as the
qualifyingphrase 'in W' makes explicit-whereas in (2') 'the entity
referredto as "this"' means "the entityreferredto as 'this' in
the actual world," and has thus a referenceindependent of the
bound variable 'W'.
Kripke calls a designator"rigid" (in a given sentence)if (in that
sentence)it refersto the same individual in everypossible world in
which the designator designates. If we extend this notion of
rigidityto substancenames, then we may express Kripke's theory
and mine by sayingthat the term'water' is rigid.
The rigidityof the term'water' followsfromthe fact that when
I give the "ostensivedefinition":"this (liquid) is water," I intend
(2') and not (1').
We may also say, followingKripke, that when I give the osten-
sive definition"this (liquid) is water," the demonstrative'this' is
rigid.
 708 THE JOURNAL OF PHILOSOPHY

What Kripke was the firstto observe is that this theoryof the
meaning (or "use," or whatever) of the word 'water' (and other
natural-kindtermsas well) has startlingconsequencesforthe theory
of necessarytruth.
To explain this, let me introduce the notion of a cross-world
relation.A two-term relationR will be called cross-world when it is
understoodin such a way thatits extensionis a set of orderedpairs
of individualsnot all in the same possible world.For example,it is
easy to understandthe relationsame heightas as a cross-world rela-
tion: just understandit so that,e.g., if x is an individual in a world
W1 who is 5 feettall (in W1) and y is an individual in W2 who is 5
feettall (in W2), then the orderedpair x,ybelongs to the extension
of same heightas. (Since an individual mayhave different heightsin
differentpossible worlds in which that same individual exists,
strictlyspeaking,it is not the ordered pair x,y that constitutesan
elementof the extensionof same heightas, but ratherthe ordered
pair x-in-world-W1, y-in-world-W2.)
Similarly,we can understandthe relationsameL (same liquid as)
as a cross-worldrelation by understandingit so that a liquid in
world W1 which has the same importantphysical properties(in
W1) that a liquid in W2 possesses(in W2) bears sameL to the latter
liquid.
Then the theorywe have been presentingmay be summarizedby
sayingthat an entityx, in an arbitrarypossible world, is water if
and only if it bears the relation sameL (construedas a cross-world
relation) to the stuffwe call "water" in the actual world.
Suppose, now, that I have not yet discoveredwhat the important
physicalpropertiesof water are (in the actual world)-i.e., I don't
yetknow that wateris H20. I may have waysof recognizingwater
that are successful(of course,I may make a small number of mis-
takes that I won't be able to detectuntil a later stage in our scien-
tificdevelopment),but not know the microstructure of water. If I
agree that a liquid with the superficialpropertiesof "water" but a
different microstructure isn't really water,then my ways of recog-
nizing water cannot be regarded as an analytical specificationof
what it is to be water. Rather, the operational definition,like the
ostensiveone, is simplya way of pointingout a standard-pointing
out the stuffin the actual world such that, for x to be water,in
any world,is for x to bear the relationsameL to the normal mem-
bers of the class of local entitiesthat satisfythe operational defini-
tion. "Water" on Twin Earth is not water, even if it satisfiesthe
operational definition,because it doesn't bear sameL to the local
 REFERENCE 709

stuffthat satisfiesthe operational definition,and local stuffthat

satisfiesthe operational definitionbut has a microstructure differ-
ent from the rest of the local stuffthat satisfiesthe operational
definitionisn't water either,because it doesn't bear sameL to the
normal examples of the local "water."
Suppose, now, that I discover the microstructureof water-
that water is H20. At this point I will be able to say that the
stuffon Twin Earth that I earlier mistookfor water isn't really
water. In the same way, if you describe,not anotherplanet in the
actual universe,but another possible universe in which there is
stuffwith the chemical formulaXYZ which passes the "operational
test" for water,we shall have to say that that stuffisn't water but
merelyXYZ. You will not have describeda possible world in which
"wateris XYZ," but merelya possibleworldin whichthereare lakes
of XYZ, people drink XYZ (and not water),or whatever.In fact,
once we have discoveredthe nature of water,nothingcounts as a
possible world in which water doesn't have that nature. Once we
have discoveredthat water (in the actual world) is H20, nothing
counts as a possible world in which water isn't H20.
On the other hand, we can perfectlywell imagine having ex-
periencesthat would convinceus (and that would make it rational
to believe that) waterisn't H20. In thatsense,it is conceivablethat
waterisn'tH20. It is conceivablebut it isn'tpossible! Conceivability
is no proofof possibility.
Kripke refersto statementsthatare rationallyunrevisable(assum-
ing there are such) as epistemicallynecessary.Statementsthat are
true in all possible worldshe refersto simplyas necessary(or some-
timesas "metaphysicallynecessary").In this terminology, the point
just made can be restatedas: a statementcan be (metaphysically)
necessaryand epistemicallycontingent.Human intuition has no
privilegedaccess to metaphysicalnecessity.
In this paper, our interestis in theoryof meaning,however,and
not in theoryof necessarytruth.Words like 'now', 'this', 'here'
have long been recognizedto be indexical,or token-reflexive-i.e.,
to have an extension which varies from context to context or
token to token. For these words, no one has ever suggestedthe
traditionaltheorythat "intensiondeterminesextension."To take
our Twin Earth example: if I have a Doppelgangeron Twin Earth,
then when I think"I have a headache," he thinks"I have a head-
ache." But the extension of the particular token of 'I' in his
verbalizedthoughtis himself(or his unit class,to be precise),while
the extension of the token of 'I' in my verbalized thoughtis me
7IO THE JOURNAL OF PHILOSOPHY

(or my unit class, to be precise). So the same word, 'I', has two
differentextensions in two differentidiolects; but it does not
followthatthe conceptI have of myselfis in any way different from
the concept my Doppelganger has of himself.
Now then,we have maintained that indexicalityextendsbeyond
the obviouslyindexical words and morphemes(e.g., the tenses of
verbs). Our theorycan be summarizedas saying that words like
'water' have an unnoticed indexical component: "water" is stuff
that bears a certain similarityrelation to the water around here.
Water at anothertime or in anotherplace or even in anotherpos-
sible world has to bear the relation sameL to our "water" in
orderto be water.Thus the theorythat (1) wordshave "intensions,"
which are somethinglike concepts associated with the words by
speakers;and (2) intensiondeterminesextension-cannot be trueof
natural-kindwordslike 'water'forthe same reasonit cannotbe true
of obviouslyindexical words like 'I'.
The theory that natural-kindwords like 'water' are indexical
leaves it open, however,whetherto say that 'water' in the Twin
Earth dialect of English has the same meaning as 'water' in the
Earth dialect and a differentextension-which is what we normally
say about 'I' in different
idiolects-therebygiving up the doctrine
that "meaning (intension) determinesextension,"or to say, as we
have chosen to do, that differencein extensionis ipso facto a dif-
ferencein meaning for natural-kindwords,therebygiving up the
doctrine that meanings are concepts,or, indeed, mental entities
of any kind.2
It should be clear, however,that Kripke's doctrinethat natural-
kind wordsare rigid designatorsand our doctrinethat theyare in-
dexical are but two waysof makingthe same point.
We have now seen that the extensionof a termis not fixedby
a concept that the individual speaker has in his head, and this is
true both because extension is, in general, determinedsocially-
thereis divisionof linguisticlabor as much as of "real" labor-and
2 Our reasons for rejecting the firstoption-to say that 'water' has the same
meaning on Earth and on Twin Earth, while giving up the doctrine that mean-
ing determinesreference-are presented in "The Meaning of 'Meaning'." They
may be illustrated thus: Suppose 'water' has the same meaning on Earth and
on Twin Earth. Now, let the word 'water' become phonemically differenton
Twin Earth-say, it becomes 'quaxel'. Presumably, this is not a change in
meaning per se, on any view. So 'water' and 'quaxel' have the same meaning
(although they refer to differentliquids). But this is highly counterintuitive.
Why not say, then, that 'elm' in my idiolect has the same meaning as 'beech' in
your idiolect, although they refer to differenttrees?
 REFERENCE 7II

because extensionis, in part,determinedindexically.The extension

of our termsdepends upon the actual natureof the particularthings
that serve as paradigms,and this actual nature is not, in general,
fullyknown to the speaker.Traditional semantictheoryleaves out
two contributionsto the determinationof reference-the contribu-
tion of society and the contributionof the real world; a better
semantic theorymust encompass both.
HILARY PUTNAM
Harvard University

SUBSTANCES AS INDIVIDUALS *

pUTNAM'S centralcontentions
seemto me bothtrueand ex-
tremelyimportant.My contributionto the symposiumwill,
therefore,not take the form of a critical reply. Instead, I
want to take up a point that is, perhaps, peripheral to Putnam's
main concerns,but one that I have found the need to get straight
about in thinkingabout thesematters.
Putnam gives us a theoryabout how names of substancesand
species, 'water', 'gold', 'tiger', etc., function,a theorythat is in-
consistentwith the conjunctionof two assumptionsabout meaning
to be found just about whereverone looks in the historyof philo-
sophical semantics.Putnam points out that this theoryis veryclose
to one developed by Saul Kripke. And, using Kripke's notion of
rigiddesignators,he says(707), "we mayexpressKripke'stheoryand
mine by sayingthat the term'water' [for example] is rigid." Now
Kripke introduced"rigid designators"in his paper, "Naming and
Necessity,"1 firstin connectionwith singular terms,in particular,
propernames. A rigid designatoris a termthat designatesthe same
individualin everypossibleworld.
The point I wish to discuss has to do with the relationships
among threetheseswhich are to be found in Kripke'spaper: (1) a
thesis about proper names concerningthe way in which they are
connectedwith what theyname; (2) a thesisabout the referentsof
*Abstract of a paper to be presented in an APA symposium on Reference,
December 28, 1973, commentingon Hilary Putnam, "Meaning and Reference,"
this JOURNAL, this issue, 699-711.
1 In Donald Davidson and Gilbert Harman, eds., Semantics of Natural Lan-
guage (Dordrecht: Reidel; New York: Humanities, 1972), pp. 253-355.