88 PHILOSOPHICAL STUDIES

lems and the results of scientific research, and an indication of the points
at which the latter are actually useful in the solution of the former.
Received August 15, 1957
NOTES
1 "Science, Perception, and Some Dubious Epistemological Motives," Philosophical
Studies, 8:55-61 (June 1957).
~P. 55.
8For example, at one point (cf. pp. 57-58) his criticism relies upon a particular view
of the analytic-synthetic distinction, and throughout much of his discussion he pre-
supposes a somewhat special brand of empiricism.
' P . 56.
6 Ibid.
6 p. 60, note 9.
A. J. Ayer, The Problem of Knowledge (London: Penquin Books, 1956), pp. 87-88.

Mr. Weitz and the Definition of Art

by JOSEPH MARGOLIS
UNIVERSITY OF SOUTH CAROLINA

M~. MorRis WEixz has recently written an extremely misleading essay on

the problem of defining fine art. 1 He pleads, as he says, "for the rejection
of" the problem of providing "a true definition or set of necessary and
sufficient properties of art." 2 W e must begin, he argues, not with "What
is art?" but "What sort of concept is 'art'?" 8 And the ammunition for his
present view (for he himself acknowledges that his earlier volume, Phi-
losophy of the Arts,4 was premised on the orientation he now considers a
false one 5) is provided by applying a distinction that Ludwig Wittgenstein
proposed in his Philosophical Investigations2 Weitz summarizes the rele-
vance of Wittgenstein's remarks for his own issue in the following way:
"The problem of the nature of art is like that of the nature of games, at
least in these respects: If we actually look and see what it is that we call
'art,' we will also find no common properties--only strands of similarities.
Knowing what alt is is not apprehending some manifest or latent essence
but being able to recognize, describe, and explain those things we call 'art'
in virtue of these similarities.
"But the basic resemblance between these concepts is their open texture.
In elucidating them, certain (paradigm) cases can be given, about which
there can be no question as to their being correctly described as 'art' or
MR. WEITZ AND THE DEFINITION OF ART 89
'game,' but no exhaustive set of eases can be given. I can list some cases
and some conditions under which I can apply correctly the concept of art
but I cannot list all of them, for the all-important reason that unforeseeable
or novel conditions are always forthcoming or envisageable.
"A concept is open if its conditions of application are emendable and
corrigible; i.e., if a situation or case can be imagined or secured which would
call for some sort of decision on our part to extend the use of the concept
to cover this, or to close the concept and invent a new one to deal with
the new case and its new property. If necessary and sufficient conditions
for the application of a concept can be stated, the concept is a closed one.
But this can happen only in logic or mathematics where concepts are con-
structed and completely defined. It cannot occur with empirically-descrip-
tive and normative concepts unless we arbitrarily close them by stipulating
the range of their uses." 7
I should like to make some systematic observations about Weitz's charge
which, I trust, will show without requiring additional comment the logical
suitability of attempting to define art.
1. On Weitz's view, the error involved in defining art (cf. the third
paragraph quoted above) applies to all "empirically-descriptive" concepts
and so is not peculiarly to be found in the theory of art. On this basis, the
definition of "man" and "tree" and "stone" suffers from the same error.
But this is surely a curious view. I suggest that what Weitz wishes to say
is that the error, when it is found, is found exclusively in the "empirically-
descriptive" and "normative" domains, though it need not occur in every
case in those domains, that it never occurs in logic and mathematics where
"concepts are constructed and completely defined."
2. I agree with Weitz's view of the "open character" of "art"; Weitz
does show persuasively that an old-fashioned definition of the novel may
exclude, contrary to our wishes, Joyce's Finnegans Wake or Dos Passos'
U.S.A. or Virginia Woolf's To the Lighthouse 8 and that we therefore de-
cide to adjust the definition to incorporate these?
3. It may be true that the objects that we wish to call "novels" (any
other subclass of fine art will serve as well, and even the generic class, art)
do not have any enumerable properties that we should wish to call "neces-
sary and sufficient properties" of the entire class but have only (following
Wittgenstein) "strands of similarity." What I must insist on, however, is
that to determine whether this is so is an empirical question and not a
Iogical one. That it is an empirical question seems to be the intent of the
first paragraph quoted above (it is also the intent of Wittgenstein's advice
to "look and see"); it is part of the equivocal intent of the second para-
graph quoted above where Weitz says, "I cannot list all of them"; but it
90 PHILOSOPHICAL STUDIES
is not the intent of the third paragraph, where Weitz uses the word "only,"
nor is it the intent of the rest of the article. He states his most extreme
view unambiguously: " W h a t I am arguing, then, is that the very expansive,
adventurous character of art, its ever-present changes and novel creations,
makes it logically impossible to ensure any set of defining properties." lo
This remark of course tends to reinforce the point made in (1), namely,
that the problem of definition Weitz is interested in appears in the em-
pirical domain though it does not necessarily apply to every empirical con-
cept; it is the peculiar properties of art that affect our effort at definition.
4. Weitz appears to confuse in his argument logical and merely practical
reasons. For when he explains why it is that "art" is an "open" concept, he
says: "We can, of course, choose to close the concept. But to do this with
'art' or 'tragedy' or 'portraiture,' etc., is ludicrous since it forecloses on the
very conditions of creativity in the arts." 11 He may be right, though I
suspect he is not; after all, we do not anticipate that the definition of
"living organism" will "foreclose" on biological evolution, and yet we have
reasonable hopes of defining the term (in Weitz's sense) so that it will
comprehend "forthcoming and envisageable" types of organisms. But more
important even than this, the reason l~e supplies for demurring to defini-
tions is very clearly a practical one; he is simply concerned that the effort
to define an enormously complicated and creative domain like art is prob-
ably beyond the capacities of anyone, that every effort will probably
(though he has no logical basis for saying "necessarily") fail because the
theorists are ordinarily myopically committed to the special features of
narrow art traditions they have studied, that commitment to a definition
may have unfortunate consequences for the very practice of art.
5. That Weitz's objections are merely practical is obvious once again
from his admission that "there are legitimate and serviceable closed con-
cepts in art" for such as "have been drawn for a special purpose." 12 lie
goes on to show, and persuasively, that "a theory or real definition . . . at
least of the extant Greek tragedies" can be given, that in fact Aristotle's
definition is false? 8 But if this is admitted (and it is difficult to see how it
could be reasonably denied), it suddenly becomes obscure what the sense
is in which such definitions cannot be given for "tragedy, . . . . comedy,"
"art." W e see now the implicit circle Weitz has unwittingly entered when
he makes the extreme remark quoted above. For he speaks of the "ad-
venturous character of art" when he wishes to say that a previous definition
of art would probably (though again, he has no basis for saying "neces-
sarily") not apply to the objects he n o w wishes to be admitted as works
of art. The confusion in short rests here: there is a difference between not-
ing the inadequacy of any formulated definition of art, if we wish to include
MR. WEITZ AND THE DEFINITION OF ART 91
as art certain obiects that do not share the necessary and sufficient proper-
ties listed in that definition, and (on the other hand) proving logically
impossible the enumeration of necessary and sufficient properties for any
set of objects already agreed upon. It is our practical dissatisfaction with
any empirical definition of this sort that urges us to revise it, to make a
"decision" (as Weitz would put it).
6. Weitz must be wrong in denying that the same problem of definition
occurs in mathematics and logic (as he argues in the third paragraph quoted
above). For even though the concepts there are "constructed," it is con-
ceivable that, examining the empirical use of these constructions, we
"decide" (again for practical reasons) to change the definitions of certain
concepts. And this is all that is required to fulfill Weitz's criterion of an
"open" concept (see the third paragraph quoted above). But we surely
can imagine a definition of number, for example, made in a relatively early
phase of the history of mathematics, that could not be applied to other
invented entities of a later stage of development. W e could, conceivably,
"decide" that the manipulation of these latter entities constituted play
with something other than "number" or we could even insist that
"number" refers to "family resemblances" only. Or we could "decide" to
revise our definition to accommodate the new development; and this would
surely be the most reasonable thing to do. W e must be clear, however, that
the change is for a practical reason and that changes made on any given
occasion do not argue that similar changes must be made for every innova-
tion in a given domain.
7. There is something odd about Weitz's statement (third paragraph
quoted above) that we cannot supply closed concepts in the "empirically-
descriptive" domain "unless we arbitrarily close them by stipulating the
ranges of their uses." This seems to suggest that there is a conceivable
alternate procedure for securing closed concepts. But, on inspection, we
see that there is none. For one thing, even the privileged domain of mathe-
matics and logic, on Weitz's view, employs such stipulative definitions; in
this respect, the logic of the empirical domain resembles that of mathe-
matics and logic. For another, Weitz actually, as we have seen in (5), does
not always treat such stipulative definitions in the empirical domain in a
pejorative fashion; he merely insists that there must be an assignable prac-
tical reason for the use of such a definition. W e are forced to a rather sur-
prising conclusion: Weitz's entire argument presupposes in a subterranean
way that we are, in some sense, able to grasp the eternal forms of things.
W e are to recognize that Joyce's Finnegans Wake, for example, is a novel
iust as Flaubert's Madame Bovary and hence are to reject, as false, defini-
tions of the novel which fail to include Finnegans Wake. Weitz himself
92 PHILOSOPHICAL STUDIES
of course would not subscribe to this reading of his view. But it is difficult
to see how else such statements as the following may be understood:
"If we take the aesthetic theories literally, as we have seen, they all fail;
but if we reconstrue them, in terms of their function and point, as serious
and argued-for recommendations to concentrate on certain criteria of ex-
cellence in art, we shall see that aesthetic theory is far from worthless." 14
None of the criteria of recognition is a defining one, either necessary or
sufficient, because we can sometimes assert of something that it is a work
of art and go on to deny any one of these conditions . . ." 15
Either (a) the definition offered for an agreed set of obiects is empiri-
cally inadequate, in which case we must try to better it; or (b) the new
obiects now to be called by the same name as the previous set (for which
a definition was empirically adequate) are arbitrarily selected, in which
case every definition may be struck down; or (c) we now wish, for practi-
cal reasons, to enlarge the set of obiects to be grouped together, in which
case we must attempt (and we cannot antecedently decide that it is im-
possible) to provide a definition of their necessary and sufficient properties.
That is, either all definitions have some stipulative basis or we must hold
to some version of the theory of forms (even if recognized only darkly
through "family resemblances").16 I hasten to remark that, even if Weitz's
empirical findings on the definition of art were defensible, even if only
"family resemblances" could be enumerated, it would not follow that em-
pirical definition is logically impossible; C. L. Stevenson, it may be ob-
served, has suggested a procedure for defining art based on this very notion
(which appears to be logically sound).17
8. It is necessary to observe that both Weitz and Wittgenstein, in speak-
ing of "family resemblances," use a categorical negative regarding the
enumeration of necessary and sufficient properties. This is simply unwar-
ranted. All Weitz has done is show that specific well-known definitions
are inadequate for a determined set of obiects. The notion of "family re-
semblances" is at best an empirical compromise; having failed to arrive at
a satisfactory definition, we are inclined to think none can be formulated
(Weitz's own disappointment on his use of the organic theory of art is to
the point). But this is to transform an empirical finding (and that a nega-
tive one) into the strongest logical obiection. The use of "family resem-
blances" is inevitably a makeshift; it is never logically impossible that we
may agree on a suitable definition at a later date, else the logical obiections
would apply as well to the use of "family resemblances." Surely, "family
resemblances" between different kinds of energy, for example, has had to
give way gradually to an empirically adequate definition of the necessary
and sufficient properties of energy.
MR. WEITZ AND THE DEFINITION OF ART 93
9. Weitz carries his argument to a disingenuous extreme when he asserts:
"None of the criteria of recognition is a defining one, either necessary or
sufficient, because we can sometimes assert of something that it is a work
of art and go on to deny any one of these conditions, even the one which
has traditionally been taken to be basic, namely, that of being an artifact:
Consider, 'This piece of driftwood is a lovely piece of sculpture.'" is
There is no question that ordinary language supplies us with remarks
such as the one Weitz has selected; but we are not required, even in ordi-
nary language, to take every remark as a literal statement of fact. If anyone
were pressed to explain such a remark, he would of course say that the
driftwood looks very much like a sculpture, that it is as if nature were a
sculptor, that we could imagine the driftwood actually fashioned by a
human sculptor into its present form. And this means that, in making such
a remark, we hardly wish to deny what is a necessary condition for an
object's actually being a work of art, "namely, that of being an artifact."
Disputes of this sort raise a question about what sort of evidence is to be
supplied to support or defeat Weitz's thesis. Whether anyone (Weitz
himself) really prefers the interpretation Weitz advances of the above re-
mark seems to me to be utterly beside the point. It seems possible to
accommodate the sense in which a piece of driftwood is called a piece of
sculpture, to get agreement on the sense in which it differs fundamentally
from the vast class of objects that ordinary usage calls sculptures or works
of art, and thence to "decide," independently of, though not without
attention to, ordinary usage, to define the term to cover the vast class rather
than the marginal case of the driftwood. I cannot see how this stipulative
feature of definition can be eliminated. Even if one were to ask, in an in-
ductive sense, how do people define art and we were to find that "family
resemblances" are probably the only traits we could enumerate, the finding
would be quite irrelevant for any systematic effort to classify our knowledge;
we would simply construct a concept which (in this instance) would
accommodate at least an important part of ordinary usage and which,
together with other distinctions, would allow us to classify empirical state-
ments about any other related phenomena without contradiction. An im-
mediately apparent analogy may be seen in considering ordinary references
to porpoises or whales as interesting fish and t h e usage of science.
10. If we reconsider the quotation given in (3), we can pinpoint the
fundamental weakness of Weitz's view. As Weitz puts it, the creative
nature of art "makes it logically impossible to ensure any set of defining
properties." I have italicized the clue words. In the very next sentence,
Weitz admits that we "can . . . choose to close the concept," which
shows that for him such a concept is not self-contradictory. It is not logi-
94 PHILOSOPHICAL STUDIES
cally impossible to construct such a concept but only logically (sic) im-
possible to ensure it. If Wittgenstein's corresponding entry 1~ is examined,
we see that Wittgenstein's purpose was to draw attention to the fact that
the usage of "family resemblances" is familiar and distinct from that of
the use of deliberately closed concepts (it is interesting that Wittgenstein's,
contrary to Weitz's, emphasis (as in the third paragraph quoted above)
admits that even the concepts of mathematics may be used in an "open"
sense). In other words, Wittgenstein distinguishes between two types of
usage and really examines the usage of "games" in the sense of that of
"family resemblances." He expressly says that he can use mathematical
concepts either in the "open" or "closed" sense, but he insists (and his
argument is surely inconclusive here) that we use "games" in the "open"
sense--as if to say, only in the "open" sense. Weitz consequently decides
that we cannot logically ensure any "closed" sense of "art," though he only
means to say that such a "closed" sense would be different from that of
an "open" sense and that we are not entitled somehow to prefer this
"closed" sense to the "open." What the priority of the "open" sense is,
is left unexplained.
11. It may further be argued that the innovations Weitz has introduced
(following Wittgenstein) are self-defeating. He fails (the passages cited
from Wittgenstein's Investigations also fail) to discuss the need for dis-
tinguishing, and the distinction itself, between the "closed" and "open"
senses even of concepts defined in terms of "family resemblances." Con-
sider the concept "games." Is courtship a game? Is love a game? Is life a
game? There seems to be a stipulative element required even here to give
discipline to usage; else we run the risk of linguistic anarchy. But if a
"closed" sense of "family resemblances" is allowed, why not a "closed"
sense for "necessary and sufficient properties"? In short, the notion of an
"open" concept is equivocally employed. It sometimes means a concept
defined in terms of "family resemblances" and it sometimes means the
opposite of a "closed" concept. On the argument advanced, the two are
independent notions and "open" is best employed if restricted to the latter
sense. Wittgenstein's concern, to summarize this as well, is to argue that
concepts based on "family resemblances" are usable as such; he sometimes
exceeds his argument when he appears to claim that some concepts, like
"games," are usable on/), in terms of "family resemblances." This is clearly
not so; but even with this restriction, the "openness" of the concept is not
automatically decided. And therefore, the mere substitution of "family
resemblances" for "necessary and sufficient properties" fails to eliminate
the problem Weitz originally proposed, namely, of employing the con-
cept of art in such a way as to avoid "foreclosing" on artistic innovation.
MR. W E I T Z AND THE DEFINITION OF ART 95
By the same token, the use of any formulated "necessary and sufficient
properties" in an open sense need not prejudice such innovation.
I believe that Weitz's argument founders on the objections enumerated.
And since the definitional effort in question is not self-contradictory and
resembles other such efforts that are both meaningful and actually success-
ful, I say let us simply try again.
Received September 5, 1957
NOTES
"The Role of Theory in Aesthetics," Journal of Aesthetics and Art Criticism, 15:
27-35 (September 1956); the essay is one of the Matchette Foundation prize essays
for 1955.
Ibid., p. 27.
a Ibid., p. 30.
Cambridge: Harvard University Press, 1950.
"The Role of Theory in Aesthetics," p. 29.
Translated by G. E. M. Anscombe (New York: Macmillan, 1953); cf. I, sections
65-75, cited by Weitz.
T"The Role of Theory in Aesthetics," p. 31.
8 Ibid., p. 31.
1bid., p. 32.
aoIbid.
Ibid.
= Ibid.
Ibid.
~4Ibid., p. 35.
Ibid., p. 34.
~6It is an interesting oddity that the appeal to ordinary usage may hide an incipient
theory of forms.
~7Cf. "On 'What Is a Poem?' " Philosophical Review, 66:329-62 (July 1957), esp.
pp. 340-47.
as "The Role of Theory in Aesthetics," p. 34.
Philosophical Investigations, I, section 68.

Books Received
AgcY~, MICHAEL. T h e Scientific Study of Social Behavior. New York:
Philosophical Library, 1957. $6.00.
BAIER, KURT. T h e Moral Point of View. Ithaca, N.Y.: Cornell University
Press, 1958. $4.00.
BANERJEE, NIKUNJA VmAgI. Concerning H u m a n Understanding. New
York: Macmillan, 1958. $6.75.
BARTA, ALVIN. Timetable of Civilization. New York: Vantage Press, 1958.
$2.75.