Gustave Caillebotte, French Impressionism, and mere exposure


Psychonomic Bulletin & Review
2003, 10 (2), 319-343

Gustave Caillebotte was Impressionism’s anomaly, in his
life as well as his art. . . . He was 14 years younger than
his initial stylistic model Degas and when his eventual
close friends Monet and Renoir first began plotting an in-
dependent exhibition . . . he was still a teenager. . . . He
was also rich.

Varnedoe (1987, p. 1)

Inheriting his father’s wealth, acquired as a military
supplier, Gustave Caillebotte was a rich painter at age 28.
Too young and freshly started for the first Impressionist
exhibition in 1874, he was invited by Renoir to join the
second in 1876, and he participated in four others. More-
over, he organized the 1876 exhibit and subsidized it and
those in 1877, 1879, and 1882. Most important in this con-
text, he also began to buy paintings in 1876, supporting
his friends when little other money was coming in (see
Bérhaut, 1994, Distel, 1990, Marrinan, 2002, Nord, 2000,
and Varnedoe, 1987, for histories). Seven of his purchases
are shown in the left-hand panels of Figures 1 and 2;

Caillebotte painted Figure 1A himself. Caillebotte’s major
phase of acquisitions continued during the height of Im-
pressionism.1 Aside from a few drawings, he collected only
the works of Paul Cézanne, Edgar Degas, Edouard Manet,
Claude Monet, Camille Pissarro, Pierre-Auguste Renoir,
and Alfred Sisley. These painters are widely regarded as
the “major” artists of Impressionism, and their works form
the core of the Impressionist canon.2

L’AFFAIRE CAILLEBOTTE

Caillebotte died suddenly in 1894 at age 46. His will
left his entire art collection to the state of France, on the
condition that the works be hung together for the public.
Such a bequest was unprecedented, and the off icial salon
culture of Paris was still ill disposed toward Impression-
ism. Jean-Léon Gérôme, an important salon painter who
was influential in the late 19th century Paris art scene, is
reported to have said, “I do not know these gentlemen
and of the donation I know only the title—Are there not
some paintings of Monsieur Monet in it? Of Monsieur
Pissarro and others? For the state to accept such f ilth
would be a blot on morality” (cited in Mead, 1974, un-
paginated; see also Rewald, 1946, p. 422).3 But perhaps
as important as the politics was the unavailability of
space; there seemed to be no suitable place in Paris large
enough to hang the 73 or more paintings.

The haggling went on for years, taxing the patience of
Renoir, Caillebotte’s executor, and Martial Caillebotte,
the older brother. Eventually, the will was broken, and
the collection was split, 38 images going to the state of
France and the rest rejected. Two of Caillebotte’s own
paintings (Images 1c and 2c; see the Appendix) were in-
cluded in the bequest by his family, and one Sisley
(Image 60c) was later deaccessioned and sent to a re-
gional museum. Thus, we can consider Caillebotte’s
Parisian legacy to be 39 works. These were f irst hung in

319 Copyright 2003 Psychonomic Society, Inc.

This research began as a set of interests accrued in 1993, when I was
in Paris, supported by a John Simon Guggenheim Memorial Fellow-
ship, and later, in the summer of 1994, during the Caillebotte centen-
nial. I thank Claudia Lazzaro for being the final, if unwitting, impetus
for this study in the summer of 2000—again, in Paris—and for many
wonderful hours of discussion about art and canons; Justine Zee Kwok
for beginning aspects of this project as a feasibility study, for many
hours of library work, and for help with initial assembly of the data-
bases; Nicholas Epley for initial guidance through the mere exposure
literature; John Bargh and Arthur Reber for encouragement and com-
ments; Andrea Coby Riddle, the teachers, and students at the Elizabeth
Ann Clune Montessori School of Ithaca for their help with Study 3;
David Dunning and Dennis Regan for suggestions leading to Studies 4
and 5; and many colleagues for their friendly cajoling, forcing me to de-
sign and carry out Study 6. A version of Study 1 was presented at the
42nd Annual Meeting of the Psychonomic Society, November 2001, in
Orlando, FL. Correspondence concerning this article should be sent to
J. E. Cutting, Department of Psychology, Uris Hall, Cornell University,
Ithaca, NY 14853-7601 (e-mail: jec7@cornell.edu).

Gustave Caillebotte, French Impressionism, 
and mere exposure

JAMES E. CUTTING
Cornell University, Ithaca, New York

Gustave Caillebotte was a painter, a collector of some of his colleagues’ most renowned works, and a
major force in the creation of the late 19th century French Impressionist canon. Six studies are presented
as a naturalistic investigation of the effects of mere exposure to images in his collection and to those
matched to them. The probabilities of cultural exposure to the 132 stimulus images were indexed by the
frequencies of their separate appearances in Cornell University library books—a total of 4,232 times in
980 different books. Across the studies, adult preferences were correlated with differences in image
frequencies, but not with recognition, complexity, or prototypicality judgments; children’s preferences
were not correlated with frequency. Prior cultural exposure also interacted with experimental expo-
sure in predictable ways. The results suggest that mere exposure helps to maintain an artistic canon.



320 CUTTING

Figure 1. Eight Impressionist images. The four in the left column were part of the Caille-
botte legacy to the state of France; the four on the right were m atched to them  for these stud-
ies. All four pairs were used as part of a corpus of 132 images. Panels A and B are Gustave
Caillebotte’s Les Raboteurs de parquet (“Floor scrapers,” Im age 1a; Paris, Musée d’Orsay)
and his Les Raboteurs de parquet, petit version (Image 1n; private collection); panels C and
D are Paul Cézanne’s Cour de ferme à Auvers (“Auvers farmyard,” Image 6c; Paris, M usée
d’Orsay) and his La M aison du pendu (“Hanged man’s house,” Im age 6n; Paris, Musée d’Or-
say); panels E and F are Edgar Degas’ Femmes à la terrase d’un café (“Women at the café,”
Image 8c; Paris, Musée du Louvre, Départment des Arts Graphiques, Fonds du M usée d’Or-
say) and his L’Absinthe (“The absinthe drinker,” Image 8b; Paris, Musée d’Orsay); and pan-
els G and H are Edouard M anet’s Le Balcon (“The balcony,” Im age 18c; Paris, Musée d’Or-
say) and his Le Déjeuner à l’atelier (“Studio luncheon,” Image 18n; Munich, Neue Pinakothek).
See the Appendix for further information. Panels A, C, D, E, F, and G are reprinted with per-
m ission of the Musée d’Orsay; panel B is reprinted with permission of Bridgeman Art; and
panel H is reprinted with perm ission of the Neue Pinakothek.

A B

C D

E
F

G
H



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 321

the Musée du Luxembourg in Paris in 1897, moved to the
Musée du Louvre in 1927, then to the Musée du Jeu de
Paume in 1947, and f inally to the Musée d’Orsay when
it opened in 1986.

Given its division of the collection, did the French gov-
ernment select well? According to some, very well: “With
the glaring exception of Cézanne, it is arguable that [the
French state] wound up with the cream of the collection”
(Varnedoe, 1987, p. 202). Perhaps, but the splitting of
the Caillebotte collection also offers a superb natural ex-
periment in canon formation and fosters other questions.
Nonetheless, the major focus of this article is on another
issue.

CANONS, CULTURE, AND 
MERE EXPOSUR E

An artistic canon consists of a culture’s esteemed
works of painting, sculpture, architecture, music, theater,
poetry, literature, or film. The membership of works within
a canon is graded—some are central, some less central
but f irmly within, some on the margins, and some clearly
outside. Thus, canons have the structure of any natural
category (Rosch, 1973). I assume that artworks within a
canon deserve their position but also that many works on
the fringes and even well outside are equally worthy and
equally deserving of cultural reverence. Why are these
not within the canon? I would claim that a major force in
canon formation is historical accident, but I haven’t the
space here to defend such an idea; canon formation has
been discussed at length in the humanities (e.g., Guillory,
1990, 1993; Sassoon, 2001; Hallberg, 1984). Instead, I
focus here on a different issue: canon maintenance over
time through its broad cultural reception.

A force that helps maintain an artistic canon, I believe,
is the cultural generalization of the laboratory phenom-
enon called mere exposure (Zajonc, 1968, 1980), the
nonconscious acquisition of information about, and atti-
tudes toward, objects and events through their repeated
presence in our lives. These occurrences help shape our
preferences. Mere exposure is a phenomenon related to
learning without awareness, or implicit learning (e.g.,
Roediger, 1990; Schacter, 1987; Seamon et al., 1995;
Squire, 1992), but with a focus on the affective compo-
nent. In particular, from childhood through college and
throughout adulthood, we are exposed to hundreds of
thousands of images. Some are representations of art;
others, as during a museum visit, are the artworks them-
selves. We do not typically remember each occurrence of
each image or where we saw it. We often will not even
recognize it if we see it again, but its trace can influence
our future assessments. Such assessments are not overt
cognitive responses on our part. They are not directly re-
lated to the formal part of our education, but they are
very much a part of our general and higher education.

The effects of mere exposure are quite automatic and
independent of what we pay attention to in our day-to-
day activities. They accrue simply as the result of being
a member of a culture, experiencing cultural artifacts

(see Zajonc, 1970). Laboratory evidence suggests that
what we are exposed to, and then prefer, can be quite
meaningless in a larger context (Bornstein, 1989)—for
example, line drawings, polygons, ideographs, nonsense
words or syllables, or sounds. Nonetheless, they can also
be meaningful—for example, photographs of objects or
people. So why not paintings? Laboratory results here
are mixed. Berlyne (1970) found that there were mere
exposure effects for abstract paintings but that complex
ones were preferred over simpler ones. Zajonc, Shaver,
Tavris, and van Kreveld (1972) found the reverse effect
for abstract paintings and their parts. And Brickman,
Redfield, Harrison, and Crandall (1972) found both ef-
fects, depending on initial responses, favorable or unfa-
vorable. But as studies relying on laboratory exposure,
these research efforts were not concerned with everyday
exposure to artwork. The major purpose of this article,
then, is to circumvent this problem—to assess effects of
mere exposure as measured in a more ecological way.4

GOALS

These six studies had three goals. The first two are in-
tertwined and were spun from the arguments above. The
f irst concerns Caillebotte. It is sometimes suggested that
he had extraordinary taste in amassing his collection
(e.g., Bernac, 1895/1966). Unequivocally, some of these
works are among the centerpieces of the Impressionist
canon. To be sure, in many cases, he was on the scene as
the f irst possible buyer of particular artworks, and he
seems to have exercised his preferences in an optimal sit-
uation to help each friend. However, it should also be
noted that many of the paintings he purchased were
thought unsellable at the time he acquired them (Bazin,
1958; Rewald, 1946). Can one determine whether or not
Caillebotte demonstrated exquisite taste? Through the
use of paired comparisons—a major method of art his-
torical analysis—one might have the opportunity to ad-
dress such a difficult question. The second goal was to
assess the role of the Musée d’Orsay in the maintenance
of the Impressionist canon. Thus, we can ask with re-
spect to contemporary viewers: Do the paintings in its
collections have a special place within the Impressionist
canon? If so, why?

The third goal is more pertinent to experimental and
social psychology and quite incidental to both Caille-
botte and the Orsay. Its focus is mere exposure as in-
dexed by the frequency with which the images appear in
print. That is, the relative frequencies of these images in
books will serve as a cultural proxy; their differential
number should mimic the differential likelihood that in-
dividuals will have seen these images before. The more
often the images appear, the more likely it is that indi-
viduals may have seen them at least once, perhaps more.
This idea is essentially the same as that proposed by Fer-
guson (1999), a museum curator:

Every act of writing or curatorial practice, whenever it
gets to the point of naming a name, is participating in a
certain level of canon formation, no matter what the intent



322 CUTTING

of its author, no matter whether it represents a challenge
to the status quo or a confirmation of it. (p. 4)

This might be modified only by adding that every act of
publishing or broadcasting an image is a participation in
the maintenance of, or in the change in, a canon. So I set
out to enumerate particular “acts of writing or curatorial
practice,” needing only to choose which images to look
for and a reasonable place to count them.

THE IMAGES, THEIR FREQUENCIE S, AND
THEIR REPRESENTATIVENESS

Image Choice and Stimulus Preparation
For the centennial of Caillebotte’s death, Anne Distel,

later head of the Musée d’Orsay, oversaw a celebration of
his work. In Distel (1994), there are small, black-and-
white images of 65 works from the Caillebotte collec-
tion; 8 others are mentioned in brief detail without im-
ages and, sometimes, without names. Of these 73, 4 were
not considered for this study—one drawing and one wa-
tercolor by Jean-François Millet, one drawing by Paul
Gavarni, and one decorative fan by Camille Pissarro.
From the remaining 69, at least one reasonable-quality
version of 64 of the images was located—62 in books
and 2 on the Internet.5 In addition, two paintings by
Caillebotte himself that were eventually included in the
legacy were also used.

Thus, 66 images from the Caillebotte collection were
used as stimuli: 2 Caillebottes, 5 Cézannes, 8 Degas, 4
Manets, 16 Monets, 14 Pissarros, 9 Renoirs, and 8 Sis-
leys. These images are listed in the left column of the
Appendix with their French titles and the most descrip-
tive of their English titles. Also included are the dates
they were painted (if known), their current locations (if
known), and their catalogue raisonné reference numbers
or, when not available, another reference citation. Among
images for each artist, works are listed in their catalogue
raisonné order, which is generally taken to be chrono-
logically accurate.6 In all, 39 of these images are from
the Orsay, 10 in other museums, and 17 in private col-
lections or in locations unknown.

To begin, I sought many high-quality color reproduc-
tions of each image; 51 were found in color, 44 in at least
three different printings. These were inspected for appar-
ent consistency of reproduction, and what appeared to be
the best image (or in the case of the other 7, sometimes
the only image) was digitized on a LaCie Scanner III
flatbed scanner. The resulting Photoshop 3.0 f iles were
about 2 MB in size. Such resolution approximates that of
35-mm f ilm.7 These f iles were stored as JPEG files at
medium resolution. The 15 remaining images were digi-
tized in black and white (grayscale) at the same resolution.

Each image in Caillebotte’s collection was then matched
to another image, selected with several constraints: The
paired image must have been by the same artist, in the
same general style, from roughly the same period (me-
dian difference = 2 years, SD = 3.7 years), and with the
same type of subject matter. The idea is that these matched

images are ones that Caillebotte might have collected
had he had the opportunity, and in many cases he proba-
bly did. The 66 comparison images were selected from
the same general sources as those for the Caillebotte im-
ages and were screened for reproduction quality in the
same way. The right panels of Figures 1 and 2 show eight
examples. Comparison images were chosen without 
regard for their location—currently in the Musée d’Or-
say, another museum, or a private collection. These are
listed in the right column of the Appendix. These images
were digitized at the same resolution as their Caillebotte
counterparts. If matched to a black-and-white image,
they too were digitized in grayscale. In all, 33 of the
comparison images were from Orsay, 27 from other mu-
seums, and 6 from private collections or in locations 
unknown. In addition, among the 66 pairs, there were 16
direct comparisons of images from the Orsay, 5 of im-
ages from other museums, and 1 of images in private col-
lections.

When not referred to by title, the paintings will be dis-
cussed by image numbers, 1–66, with the additional no-
tation of c (for the Caillebotte images) and n (for their
matched pairs, the non-Caillebotte images). Thus,
Image 26c is Monet’s La Gare Saint-Lazare (“The Saint-
Lazare train station,” Figure 2A), and Image 8n is Degas’
L’Absinthe (“The absinthe drinker,” Figure 1F). Table 1
presents these image pair numbers and an abbreviated or
descriptive form of their English titles for data presenta-
tion and discussion.

Determination of Image Frequency
After I selected the images, a research assistant and I

became suff iciently familiar with all 132 to recognize
them spontaneously. Carrying copies of the images for
reference, we then began to consult all the relevant books
we could f ind in the Cornell Fine Arts Library and, later,
12 other Cornell campus libraries. Our intent was to
record every occurrence of each of the 132 images in
Cornell’s more than 6 million volumes. Carrying an elec-
tronic notebook with us, we created and then continually
updated separate databases for each of the 132 images.
These were Microsoft Excel files that registered each oc-
currence with its source’s call number, author, title, date
of publication, page number, and occasionally, other in-
formation. We found target images in books along shelves
by call number according to (1) each of the eight artists,
which included monographs, exhibition catalogs, cata-
logues raisonnés, and works in sections on painting, sculp-
ture, pastels, drawings, watercolors, prints, and com-
bined media; (2) all other artists closely or even loosely
associated with Impressionism, including those gener-
ally painting earlier, such as Corot, Courbet, and Turner,
those painting at the same time, such as Fantin-Latour
and Forain, and those painting later, such as Gauguin,
Seurat, Toulouse-Lautrec, and Van Gogh; (3) artistic terms,
such as Impressionism, Neoimpressionism, and Postim-
pressionism, but also the Nabis, Naturalism, Japonisme,
Cubism, Symbolism, and more generally, 19th century



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 323

Figure 2. A second group of eight Impressionist images. Again Caillebotte owned the four
on the left, and again, these 4 pairs were among the 66 pairs used in these studies. Panels A
a nd B  are Cla ude M onet’s La Gare S aint-La zare (“T he S aint-Lazare tra in sta tion,”
Image 26c; Paris, Musée d’Orsay) and his La Gare Saint-Lazare, l’arrivée d’un train (“The
arrival of a train at the Saint-Lazare train station,” Im age 26n; Cambridge, MA, Fogg Mu-
seum of Art); panels C and D are Camille Pissarro’s Les Toits rouges, coin du village, effet
d’hiver (“Red roofs,” Im age 43c; Paris, M usée d’Orsay) and his La Côte des Boeufs à Pon-
toise (Im age 26n; London, National Gallery); panels E and F are Pierre-Auguste Renoir’s Bal
du M oulin de la Galette (Im age 54c; Paris, Musée d’Orsay) and his Le Déjeuner des canotiers
(“Luncheon at the boating party,” Image 54n; Washington, DC, Phillips Collection); and
p anels G and H are Alfred  S isley ’s Les Régates à M olesey (“ The regatta at M olesey,”
Image 59c; Paris, M usée d’Orsay) and his Les Régates à Hampton Court (Image 59n; Zurich,
Stiftung Sammlung Emil G. Bührle). Again, see the Appendix for m ore information. Panels
A, C, E, and G are reprinted with permission of the Musée d’Orsay, Paris, France; panel B
is reprinted with permission of the Harvard University Art Museums; panel D is reprinted
with permission of the National Gallery, London; panel F is reprinted with perm ission of the
Phillips Collection, Washington, DC; and panel G is reprinted with permission of the Samm-
lung E. G. Bührle, Zurich, Switzerland.

A B

C
D

E F

G H



324 CUTTING

T
ab

le
 1

D
es

cr
ip

ti
ve

 I
m

ag
e 

T
it

le
s 

fo
r 

M
em

b
er

s 
o

f 
E

ac
h

 P
ai

r 
E

n
u

m
er

at
ed

 A
lp

h
ab

et
ic

a
ll

y
 b

y 
P

ai
n

te
r,

 I
m

ag
e 

F
re

q
u

en
ci

es
, R

ec
og

n
it

io
n

 R
at

es
 in

 S
tu

d
ie

s 
1 

a
n

d
 2

 (
S

1
 a

n
d

 S
2)

, C
om

p
a

ra
ti

ve
C

o
m

p
le

xi
ty

 J
u

d
gm

en
ts

 in
 S

tu
d

y 
4

 (
S

4)
, P

ro
to

ty
p

ic
a

li
ty

 R
a

ti
n

g
s 

in
 S

tu
d

y 
5

 (
S

5)
, a

n
d

 P
re

fe
re

n
ce

s 
in

 S
tu

d
ie

s 
1,

 2
, 3

, 4
, a

n
d

 6
 (

S
1–

S
4,

 S
6)

Ju
d

g
m

en
ts

 f
o

r 
th

e 
M

o
re

P
ro

to
-

P
ro

to
-

F
re

q
ue

n
t 

Im
ag

e
C

ai
ll

eb
o

tt
e

ty
pi

-
ty

p
i-

C
o

m
-

C
ol

le
ct

io
n

F
re

q
u

en
cy

R
ec

og
n

it
io

n
ca

li
ty

C
o

m
p

ar
is

o
n

F
re

q
u

en
cy

R
ec

og
n

it
io

n
ca

li
ty

p
le

x
it

y
P

re
fe

re
n

ce

Im
ag

es
A

ll
G

en
er

al
S

1
S

2
(S

5
)

Im
ag

es
A

ll
G

en
er

al
S

1
S

2
(S

5
)

(S
4

)
S

1
S

2
S

3
S

4
S

6

C
ai

ll
eb

o
tt

e
1

.
T

h
re

e 
fl

o
or

 s
tr

ip
p

er
s*

6
1

3
.0

2
.1

6
2

.2
T

w
o 

fl
o

o
r 

st
ri

p
pe

rs
†

10
0

.0
1

.3
7

2
.0

.8
1

7
7

5
7

–
67

41
2

.
S

n
ow

-c
ov

er
ed

 r
o

o
fs

*
2

6
0

.0
1

–
4

.0
R

ue
 H

al
év

y,
 7

th
 f

lo
o

r†
11

0
.0

4
–

4
.0

–
2

1
–

–
–

15

C
éz

an
n

e
3

.
C

o
up

le
s 

by
 p

on
d

1
3

0
.0

4
–

4
.6

T
h

e 
pi

cn
ic

12
0

.0
3

–
4

.2
–

7
3

–
–

–
59

4
.

F
o

u
r 

b
at

he
rs

3
4

0
.0

2
.0

5
3

.2
F

iv
e 

b
at

h
er

s*
17

0
.0

1
.0

5
4

.5
.6

8
7

6
6

2
–

89
73

5
.

F
lo

w
er

s,
 r

o
co

co
 v

as
e

1
5

0
.0

1
–

2
.1

F
lo

w
er

s,
 D

el
ft

 v
as

e*
25

0
.0

4
–

3
.9

–
6

2
–

39
–

59
6

.
A

uv
er

s 
fa

rm
y

ar
d

*
3

0
1

.0
1

.2
1

4
.5

H
an

g
ed

 m
an

’s
 h

o
u

se
*

1
65

1
6

.0
1

.1
6

5
.0

.2
1

6
5

7
6

28
88

63
7

.
B

ay
 o

f 
E

st
aq

ue
*

8
6

3
.0

3
.1

6
5

.4
E

st
aq

u
e 

lo
o

k
in

g
 e

as
t

27
0

.0
2

.1
1

5
.0

.5
0

8
7

5
8

–
82

75

D
eg

as
8

.
W

om
en

 a
t 

a 
ca

fé
*

8
9

6
.0

2
.4

7
4

.0
A

b
si

n
th

e 
d

ri
n

k
er

*
1

73
3

4
.0

7
.1

1
4

.5
.1

9
6

6
7

4
52

62
55

9
.

T
h

ea
te

r 
ch

o
ru

s*
1

8
0

.0
4

–
4

.9
D

an
ce

r 
w

it
h 

b
o

u
q

u
et

*
37

4
.0

8
–

4
.9

–
4

6
–

–
–

40
10

.
W

om
an

 l
ea

vi
n

g
 b

at
h

*
2

9
0

.0
4

–
4

.7
W

o
m

an
 d

ry
in

g
 f

ee
t*

25
1

.0
5

–
4

.1
–

2
0

–
–

–
23

11
.

D
an

ce
 l

es
so

n
, v

io
li

n
2

7
0

.0
6

.2
6

4
.8

T
w

o 
d

an
ce

rs
 a

t 
b

ar
44

2
.0

9
.1

6
5

.0
.3

6
5

6
6

3
–

69
53

12
.

B
al

le
t 

st
ar

*
8

7
13

.1
4

.4
2

5
.1

A
ra

be
sq

u
e,

 b
ou

q
u

et
*

23
3

.0
9

.5
3

4
.9

.1
1

7
3

4
1

–
63

48
13

.
S

q
u

at
ti

n
g

 w
om

an
*

1
4

0
.0

1
.1

1
5

.0
W

o
m

an
 i

n
 t

u
b

*
96

1
0

.0
7

.4
0

4
.0

.8
4

7
0

8
4

–
76

67
14

.
S

p
an

is
h

 d
an

ce
r 

(b
/w

)*
8

0
.0

1
–

3
.9

A
m

b
as

sa
d

o
rs

 c
af

é 
(b

/w
)*

12
0

.0
1

–
4

.2
–

6
9

–
–

–
–

15
.

D
an

ce
r 

ru
b

b
in

g
 a

n
k

le
*

2
6

0
.1

3
–

4
.9

D
an

ce
r,

 u
m

br
el

la
32

1
.0

7
–

4
.5

–
6

0
–

66
–

58

M
an

et
16

.
A

ng
el

in
a*

1
8

0
.0

1
.0

5
2

.5
G

eo
rg

es
 C

le
m

en
ce

au
47

0
.0

2
.0

5
2

.3
.1

7
6

1
6

3
–

41
43

17
.

T
h

e 
ra

ce
s 

(b
/w

)†
6

0
.0

1
–

2
.5

L
on

gc
ha

m
ps

 h
or

se
ra

ce
 (

b/
w

)
11

0
.0

0
–

3
.6

–
5

6
–

–
–

–
18

.
T

h
re

e 
o

n
 a

 b
al

co
ny

*
2

0
1

21
.0

2
.1

5
3

.0
L

u
nc

h 
in

 t
he

 s
tu

di
o

1
04

4
.0

0
.0

5
1

.9
.0

9
5

6
4

7
68

48
43

19
.

C
ro

q
u

et
 m

at
ch

†
9

0
.0

6
.1

6
3

.7
O

n
 th

e 
b

ea
ch

31
0

.0
5

.0
5

3
.0

.8
6

7
2

6
8

65
69

66

M
o

n
et

 
20

.
M

t.
 R

ib
o

u
de

t 
(b

/w
)†

3
0

.0
1

–
4

.2
V

ie
w

, A
rg

en
te

ui
l 

pl
ai

n 
(b

/w
)*

8
0

.0
0

–
4

.5
–

7
1

–
–

–
–

21
.

A
rg

en
te

u
il

 r
eg

at
ta

*
6

6
6

.0
3

.1
6

5
.4

S
ai

li
n

g
, G

en
n

ev
il

li
er

s
24

0
.0

1
.0

5
5

.1
.6

4
4

4
5

3
–

23
33

22
.

G
ar

d
en

 l
u

n
ch

*
6

4
3

.0
3

.1
6

5
.1

W
o

m
en

 i
n

 g
ar

d
en

*
1

62
2

7
.0

4
.1

1
4

.0
.3

6
6

3
5

7
58

67
65

23
.

A
pa

rt
m

en
t 

an
d

 c
h

il
d*

3
2

0
.0

1
–

3
.5

C
or

n
er

 o
f 

a 
st

u
d

io
*

14
0

.0
1

–
2

.1
–

5
6

–
57

–
26

24
.

T
u

il
er

ie
s 

at
 d

aw
n

*
1

5
0

.0
2

–
5

.7
S

ei
n

e 
ic

e 
br

ea
k-

up
11

0
.0

2
–

4
.9

–
6

2
–

6
–

28
25

.
G

en
n

ev
il

li
er

s 
p

la
in

4
0

.0
0

–
5

.6
V

ét
h

eu
il

, 
L

av
ac

o
u

rt
*

9
0

.0
6

–
6

.4
–

6
5

–
26

–
77

26
.

S
t.

 L
az

ar
e 

st
at

io
n

*
1

2
7

21
.0

5
.2

1
5

.6
S

t.
 L

az
ar

e 
tr

ai
n

62
1

2
.0

2
.3

2
5

.3
.6

4
4

9
3

2
–

35
40

27
.

P
o

nt
 d

e 
R

o
m

e,
 L

az
ar

e†
1

4
0

.0
0

–
5

.5
P

on
t 

E
u

ro
p

e,
 L

az
ar

e
25

2
.0

2
–

6
.3

–
7

5
–

–
–

66
28

.
S

t.
 L

az
ar

e 
si

g
n

al
2

0
0

.0
0

–
4

.8
S

t.
 L

az
ar

e 
tr

ac
ks

†
10

0
.0

1
–

4
.9

–
6

7
–

68
–

56
29

.
A

pp
le

 t
re

es
 (

b
/w

)†
7

0
.0

0
–

4
.4

V
ét

h
eu

il
 l

an
ds

ca
p

e 
(b

/w
)*

5
0

.0
2

–
4

.3
–

4
9

–
–

–
–

30
.

V
ét

h
eu

il
 c

hu
rc

h
*

3
0

0
.0

1
–

4
.3

V
ét

h
eu

il
 w

in
te

r
7

0
.0

0
–

5
.1

–
4

4
–

–
–

63
31

.
P

lu
m

 t
re

es
 i

n
 b

lo
o

m
 (

b
/w

)†
3

0
.0

0
–

5
.1

U
n

d
er

 l
il

ac
s 

(b
/w

)*
14

0
.0

0
–

4
.3

–
7

4
–

–
–

–
32

.
H

oa
rf

ro
st

*
1

7
0

.0
5

.5
2

6
.0

V
ét

h
eu

il
 i

n
 m

is
t

23
0

.0
8

.2
2

6
.1

.3
3

7
0

7
8

65
70

59
33

.
R

ed
 m

u
m

s 
(b

/w
)†

4
0

.0
0

–
4

.7
M

u
m

s 
(b

/w
)*

17
0

.0
4

–
4

.3
–

7
4

–
–

–
–

34
.

S
ei

n
e 

at
 R

o
ch

e 
G

u
yo

n
 (

b
/w

)†
3

0
.0

1
–

5
.5

S
ei

n
e 

w
it

h
 ic

e-
fl

o
es

 (
b

/w
)*

32
0

.0
1

–
5

.4
–

7
5

–
–

–
–

35
.

B
el

le
-I

sl
e 

ro
ck

s*
2

7
0

.0
1

–
5

.4
B

el
le

-I
sl

e 
st

o
rm

*
24

0
.0

3
–

4
.6

–
6

3
–

69
–

61



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 325

T
ab

le
 1

 (
C

o
n

ti
n

u
ed

)
Ju

d
g

m
en

ts
 f

o
r 

th
e 

M
o

re
P

ro
to

-
P

ro
to

-
F

re
q

ue
n

t 
Im

ag
e

C
ai

ll
eb

o
tt

e
ty

pi
-

ty
p

i-
C

o
m

-
C

ol
le

ct
io

n
F

re
q

u
en

cy
R

ec
og

n
it

io
n

ca
li

ty
C

o
m

p
ar

is
o

n
F

re
q

u
en

cy
R

ec
og

n
it

io
n

ca
li

ty
p

le
x

it
y

P
re

fe
re

n
ce

Im
ag

es
A

ll
G

en
er

al
S

1
S

2
(S

5
)

Im
ag

es
A

ll
G

en
er

al
S

1
S

2
(S

5
)

(S
4

)
S

1
S

2
S

3
S

4
S

6

P
is

sa
rr

o
36

.
L

o
u

ve
ci

en
n

es
 r

o
ad

†
8

0
.0

1
.0

0
3

.1
E

nt
ra

n
ce

 t
o

 V
oi

si
n

s*
65

1
1

.0
0

.0
5

3
.8

.4
1

7
3

6
8

61
65

73
37

.
B

o
ug

iv
al

 w
as

h
ho

u
se

*
2

1
1

.0
0

–
3

.9
B

an
k

s 
o

f 
th

e 
O

is
e

7
0

.0
0

–
5

.1
–

3
7

–
–

–
39

38
.

M
o

n
tf

ou
ca

u
lt

 r
o

ck
s†

7
0

.0
0

–
5

.8
H

er
m

it
ag

e 
h

il
l*

12
0

.0
0

–
4

.9
–

6
0

–
26

–
39

39
.

P
lo

w
m

an
†

5
0

.0
1

–
4

.5
C

ha
p

o
nv

al
 l

an
d

sc
ap

e*
21

1
.0

1
–

5
.5

–
4

4
–

–
–

49
40

.
P

o
nt

o
is

e 
g

ar
d

en
 (

b
/w

)†
3

0
.0

1
–

4
.5

H
er

m
it

ag
e,

 P
o

n
to

is
e 

(b
/w

)†
2

0
.0

0
–

4
.6

–
5

6
–

–
–

–
41

.
M

o
n

tf
ou

ca
u

lt
 h

ar
ve

st
*

3
3

0
.0

4
.3

2
4

.9
H

ay
st

ac
k

, 
P

o
n

to
is

e†
13

0
.0

7
.1

1
4

.5
.5

3
3

6
2

1
–

38
27

42
.

A
ut

u
m

n
 w

oo
d

s‡
4

0
.0

0
.0

5
5

.9
L

it
tl

e 
b

ri
d

g
e,

 P
on

to
is

e
13

1
.0

2
.0

5
4

.6
.1

2
6

9
3

2
–

37
28

43
.

R
ed

 r
o

o
fs

, 
H

er
m

it
ag

e*
9

9
23

.0
1

.0
0

4
.7

C
ôt

e 
d

es
 B

o
eu

fs
38

4
.0

0
.0

0
4

.5
.1

3
4

2
4

2
–

38
46

44
.

P
o

nt
o

is
e 

o
rc

h
ar

d
*

5
6

4
.0

1
–

5
.6

L
o

uv
ec

ie
n

n
es

 o
rc

h
ar

d
23

1
.0

1
–

4
.5

–
5

7
–

–
–

35
45

.
R

ye
 f

ie
ld

s,
 P

o
n

to
is

e†
4

0
.0

6
–

5
.7

C
ow

he
rd

, 
P

o
n

to
is

e
9

1
.0

2
–

5
.6

–
6

0
–

33
–

48
46

.
P

at
h

 t
h

ro
u

g
h

 w
o

od
s*

1
3

0
.0

1
–

4
.9

R
es

ti
n

g
 i

n 
w

o
o

d
s

8
0

.0
1

–
5

.6
–

6
2

–
28

–
45

47
.

E
d

g
e 

of
 w

o
od

s 
(b

/w
)

3
0

.0
0

–
4

.8
P

in
k

 h
o

u
se

 (
b

/w
)*

6
0

.0
0

–
5

.4
–

7
2

–
–

–
–

48
.

P
at

h
 c

li
m

b
in

g
 f

ie
ld

s*
7

0
.0

1
–

5
.5

W
o

m
an

, 
E

ra
g

ny
 f

ie
ld

*
29

2
.0

4
–

5
.6

–
5

2
–

–
–

47
49

.
W

h
ee

lb
ar

ro
w

*
1

7
1

.0
0

–
5

.4
A

u
tu

m
n 

p
o

n
d

†
3

0
.0

1
–

4
.5

–
6

0
–

22
–

27

R
en

o
ir

50
.

G
ir

l 
re

ad
in

g
*

4
5

2
.0

5
.1

1
5

.7
M

ad
am

e 
D

ar
ra

s*
8

0
.0

2
.2

2
3

.2
.7

9
8

1
7

9
–

77
59

51
.

S
t.

 G
eo

rg
es

 P
la

ce
 (

b
/w

)†
4

0
.0

0
–

4
.9

A
ra

b 
fe

st
iv

al
, A

lg
ie

rs
 (

b
/w

)*
21

0
.0

1
–

4
.7

–
6

7
–

–
–

–
52

.
T

o
rs

o
 i

n
 s

u
n

li
g

h
t*

9
0

6
.0

3
.1

1
5

.6
A

n
n

a,
 n

u
d

e
29

2
.0

3
.1

6
4

.2
.6

3
5

1
5

8
–

68
44

53
.

T
h

e 
sw

in
g

*
9

4
7

.0
5

.4
2

5
.4

T
h

e 
bo

w
er

, G
al

et
te

18
0

.0
7

.2
1

5
.8

.4
4

7
2

5
3

–
79

66
54

.
M

o
u

li
n

 G
al

et
te

 p
ar

ty
*

2
8

2
51

.2
5

.5
8

5
.8

B
oa

ti
n

g
 l

u
n

ch
eo

n
1

92
3

5
.2

1
.5

8
5

.3
.8

5
5

4
4

7
34

52
40

55
.

S
ei

n
e 

at
 C

h
am

p
ro

sa
y

*
2

1
0

.0
0

–
6

.1
S

n
ow

y
 l

an
ds

ca
p

e
7

0
.0

1
–

6
.0

–
4

4
–

92
–

25
56

.
R

ai
lr

o
ad

 b
ri

d
g

e*
1

5
0

.0
1

–
5

.6
W

o
m

an
 i

n
 g

ar
d

en
7

0
.0

8
–

6
.3

–
3

0
–

–
–

46
57

.
C

h
at

ea
u

 o
f 

m
is

t 
(b

/w
)

2
0

.0
1

–
4

.6
W

in
d

in
g

 p
at

h
 (

b
/w

)*
44

0
.0

1
–

4
.4

–
6

6
–

–
–

–
58

.
G

ir
ls

 a
t 

p
ia

no
†

‡
(*

 +
4

1
)

6
0

.0
4

–
2

.7
G

ir
ls

 a
t 

p
ia

n
o

 i
n 

re
d

19
0

.0
2

–
2

.9
–

5
1

–
33

–
38

S
is

le
y

59
.

B
o

at
s 

at
 M

o
le

se
y

*
4

1
1

.0
0

.0
5

5
.1

H
am

p
to

n 
C

o
ur

t 
B

o
at

s
11

0
.0

4
.1

6
5

.6
.8

6
5

0
3

2
–

40
41

60
.

L
o

u
ve

ci
en

n
es

 s
tr

ee
t

7
0

.0
1

–
5

.7
S

ab
lo

n
s 

w
o

o
d

s*
10

4
.0

1
–

5
.2

–
4

6
–

88
–

57
61

.
S

ei
n

e 
at

 S
ur

es
n

es
*

1
3

0
.0

1
–

5
.5

B
ri

d
g

e 
at

 S
èv

re
s

2
0

.0
5

–
6

.2
–

4
5

–
71

–
41

62
.

S
t.

 M
am

m
ès

 f
ar

m
*

1
5

0
.0

5
.0

5
5

.4
V

o
is

in
s 

v
il

la
g

e*
12

1
.0

5
.2

1
5

.6
.4

7
7

0
5

3
74

75
61

63
.

E
d

g
e 

of
 f

o
re

st
 (

b
/w

)*
9

1
.0

0
–

5
.3

R
es

ti
n

g
 b

y
 a

 b
ro

o
k

 (
b

/w
)*

9
0

.0
1

–
4

.4
–

§
4

5
§

–
–

–
–

64
.

S
ei

n
e 

at
 S

t.
 M

am
m

ès
 (

b
/w

)*
9

0
.0

1
–

5
.7

L
o

in
g

 c
an

al
 (

b
/w

)*
4

0
.0

3
–

5
.3

–
8

1
–

–
–

–
65

.
S

ei
n

e 
at

 B
il

la
n

co
u

rt
 (

b/
w

)†
2

0
.0

1
–

4
.5

S
ei

n
e 

at
 B

o
u

gi
va

l 
(b

/w
)*

13
0

.0
0

–
5

.1
–

9
2

–
–

–
–

66
.

S
ei

n
e 

su
n

se
t 

(b
/w

)†
2

0
.0

1
–

5
.3

B
ou

g
iv

al
 l

oc
k

 (
b

/w
)*

14
0

.0
1

–
4

.5
–

2
9

–
–

–
–

N
o

te
—

F
re

q
u

en
cy

, f
re

qu
en

cy
 c

o
u

nt
s 

o
f 

ea
ch

 i
m

ag
e 

as
 s

ea
rc

h
ed

 f
o

r 
in

 t
h

e 
C

o
rn

el
l 

L
ib

ra
ri

es
; 

re
co

g
n

it
io

n
, t

h
e 

p
ro

b
ab

il
it

y 
o

f 
se

lf
-r

ep
o

rt
ed

 r
ec

og
n

it
io

n
 o

f 
th

e 
im

ag
e;

 p
ro

to
ty

p
ic

al
it

y,
 t

h
e 

ju
d

g
m

en
t 

o
f 

ho
w

 t
y

p
-

ic
al

 e
ac

h
 i

m
ag

e 
is

 o
f 

Im
p

re
ss

io
ni

st
 a

rt
 (

1
–

7
 s

ca
le

, 
7

=
 m

o
st

);
 a

ll
, 

nu
m

er
ic

al
 f

re
q

u
en

cy
 o

f 
im

ag
e 

ac
ro

ss
 a

ll
 l

ib
ra

ry
 b

o
o

k
s;

 g
en

er
al

, 
n

u
m

er
ic

al
 f

re
qu

en
cy

 i
n

 i
n

tr
o

d
u

ct
o

ry
 a

rt
 h

is
to

ry
 t

ex
tb

o
o

k
s,

 w
or

ld
 i

m
ag

e
so

ur
ce

bo
o

k
s,

 e
n

cy
cl

o
p

ed
ia

s,
 a

n
d

 d
ic

ti
o

n
ar

ie
s;

 c
o

m
p

le
x

it
y,

 p
ro

ba
b

il
it

y
 t

ha
t 

th
e 

m
o

re
 f

re
q

ue
n

t 
im

ag
e 

of
 t

h
e 

pa
ir

 w
as

 a
ls

o
 j

u
d

g
ed

 a
s 

m
or

e 
co

m
p

le
x;

 p
re

fe
re

n
ce

, 
th

e 
p

er
ce

n
ta

g
e 

th
at

 t
h

e 
m

o
re

 f
re

q
ue

n
t 

im
ag

e
w

as
 l

ik
ed

 b
et

te
r;

 b
/w

, 
bl

ac
k

 a
n

d
 w

hi
te

 c
om

pa
ri

so
n

s,
 c

o
lo

r 
im

ag
es

 u
n

av
ai

la
bl

e 
fo

r 
C

ai
ll

eb
o

tt
e 

im
ag

es
.

*
Im

ag
es

 i
n

 t
h

e 
M

u
sé

e 
d

’O
rs

ay
 o

r 
M

u
sé

e 
d

u 
L

o
u

v
re

, 
fo

n
d

s 
d

u
 M

u
sé

e 
d

’O
rs

ay
.

†
Im

ag
es

 i
n

 p
ri

va
te

co
ll

ec
ti

o
n

s,
 o

r 
in

 lo
ca

ti
o

n
s 

u
nk

n
ow

n
.

‡
R

en
o

ir
 p

ai
n

te
d

 s
ev

er
al

 J
eu

n
es

 f
il

le
s 

a
u

 p
ia

n
o,

 a
n

d
 th

ey
 a

re
 v

ir
tu

al
ly

 in
d

is
ti

n
g

u
is

h
ab

le
. C

ai
ll

eb
ot

te
’s

 i
m

ag
e 

is
 s

li
g

h
tl

y
 d

ar
ke

r 
in

 c
o

lo
r 

an
d

 h
as

 a
 s

li
g

h
tl

y 
d

if
fe

re
nt

 h
an

d
po

si
ti

o
n 

fo
r 

th
e 

g
ir

l 
n

o
t 

p
la

y
in

g
 t

h
e 

p
ia

no
. B

ec
au

se
 o

f 
th

ei
r 

st
ri

k
in

g 
si

m
il

ar
it

y,
 t

h
e 

fr
eq

u
en

ci
es

 f
o

r 
th

e 
O

rs
ay

 v
er

si
o

n
 w

er
e 

ad
d

ed
 t

o
 t

h
o

se
 o

f 
th

e 
C

ai
ll

eb
o

tt
e 

v
er

si
o

n
 i

n
 f

re
q

u
en

cy
 a

n
al

y
se

s.
§
W

h
en

 f
re

qu
en

-
ci

es
 a

re
 t

ie
d

, 
th

e 
p

er
ce

n
ta

g
e 

p
re

fe
re

nc
e 

o
f 

th
e 

C
ai

ll
eb

o
tt

e 
im

ag
e 

is
 g

iv
en

.



326 CUTTING

art, French art, European art, and modern art; (4) paint-
ing and drawing techniques and the use of color; (5)
landscapes, seascapes, portraits, still lifes, and flowers;
(6) feminism, nudes, bodies, fashion, visual culture, and
modern life; (7) aesthetics and form; (8) art appreciation
and the psychology and philosophy of art; (9) pictorial
art as it relates to music, poetry, and the other arts; (10)
guidebooks to Paris and the French patrimoine, a reflec-
tion of French culture, geography, and history; (11) books
on art collectors and collecting; (12) books on museums
and museum design; (13) source books of museum hold-
ings around the world and their guide books; (14) books
of images for sale as posters; (15) auction house catalogs;
(16) introductory art history textbooks; (17) world art
textbooks and sourcebooks; (18) encyclopedias and dic-
tionaries specific to art; and (19) general encyclopedias.

Later, it will be useful to distinguish among image ap-
pearances in three types of books: (1) the most focused
texts featuring one of the eight artists; (2–15) the broader,
more topical texts featuring many artists or those featur-
ing one artist but using works of another (e.g., a book on
Cézanne with an image by Pissarro); and (16–19) gen-
eral textbooks, sourcebooks, encyclopedias, and dictio-
naries, whose purpose it is to cover all of art across all of
time. In this process, we attended to shelf lists and the
on-line catalog, recalled books when necessary, obtained
missing books on interlibrary loan, and perhaps most im-
portant—and certainly most delightful—simply roamed
the aisles of the Cornell Fine Arts and other libraries in
search of candidate volumes. We thumbed through as many
as were deemed even remotely likely to have any of these
images.

Several constraints governed our tallies, following the
spirit of counting the different “acts of writing or cura-
torial practice” (Ferguson, 1999): Multiple copies of the
same book were not considered (these were intended for
the same audience, had the same call number, and were
published at the same time), although a foreign language
book and its English language translation were counted
separately, as well as different editions of the same book
(these were intended for different audiences, had differ-
ent call numbers, and were published at different times).
Occasionally, in a given book, there would be both a full
image and a detail of it. Both of these were counted, with
the idea being that if the author wished to show the
image twice (or more), it was important to count it twice
(or more). In the databases, we recorded the number of
details for each image and the number of times the image
appeared in color (vs. black and white), on covers, as
frontispieces, or spread out on two pages versus one.
None of these factors had any statistical leverage in the
results that follow, so they will not be considered again.

We searched books—sometimes intensively, some-
times more leisurely—over the course of 20 months in at
least 200 library visits. After several months, totals were
accumulated each month or so and were compared, and
correlations were computed. These were always ex-
tremely high (rs > .99), and increasingly so as time pro-

gressed. Thus, however many books we might have
missed in assembling our databases, image counts that
included them would not have changed the shape of the
relationships among the frequencies reported here.

In all, we located 4,232 reproductions of the 132 im-
ages in 980 different books. Multiple volume sets were
counted as a single book. Publication dates were between
1901 and 2002. In this effort, possibly 6,000 books were
examined. Frequencies of image occurrence ranged from
2 (1 Pissarro, 1 Renoir, and 3 Sisleys) to 282 (Renoir’s Bal
du Moulin de la Galette, Image 54c and Figure 2E).8 Mean
frequency for all the images was 32.1; median frequency
was 16. Distributions varied widely: Eight images oc-
curred more than 100 times (including Figures 1D, 1F,
1G, 1H, 2A, 2E, and 2F), 13 from 50 to 100 times (in-
cluding Figures 1A, 1E, 2B, and 2C), 27 from 25 to 49
times (including Figures 1C, 2D, and 2G), 44 from 10 to
24 times (including Figures 1B and 2H), and 40 fewer
than 10 times. Thus, among the 132 images, there is a
clear gradation from the high canon of Impressionism to
its base corpus. Raw frequencies for the images in all the
texts and in general texts are shown in Table 1, but a dis-
cussion of them will be interlaced with the results of
Study 1.

An important question arises about the utility of these
frequencies: How representative are the relative occur-
rences of these 132 images to what would be more
broadly available in our culture? This is not an easy
question to resolve. Nonetheless, one can approximate
an answer in two ways. The f irst related question is: How
representative are the numbers of books on each artist in
the Cornell collections to all the books and other sources
on these painters. This was assessed in three ways.

On the Representativeness of the 
Cornell Collections

First, in April 2001, the on-line catalog of the Cornell
libraries was searched for all occurrences of the last
names of eight artists, used as keywords. Individual
records were inspected for relevancy, and the number
was recorded.9 The same was then done on line for the
Bibliography of the History of Art (BHA), a professional
database of books and articles since 1973 that can be
searched by the same keyword technique. The correla-
tion between the frequencies of the relevant titles in the
Cornell collections and those in the BHA was very high
(r = .97, p < .001). Relative counts are shown in Table 2.
This suggests that the holdings at Cornell on these artists
and the scholarly work done on them in the last third of the
20th century are quite tightly related. Given the assump-
tion that the artworks of each artist should be distributed
roughly in the same way in both corpora (Cornell’s and
the BHA’s), Cornell’s collection seems representative of
the professional literature.

Second, an on-line search was conducted of current
and out-of-print books on Amazon.com, screened in the
same manner as before. The results are also shown in
Table 2, and the correlation between the Cornell hold-



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 327

ings and the Amazon listings is reasonably high (r = .80,
p < .02). Differences are not many. The Amazon offer-
ings overrepresented Monet and underrepresented
Cézanne, as compared with the Cornell holdings and the
BHA listings.

Third, Internet explorations were conducted in April
2002, with the search engine Google. Search was done
three ways: f irst, using the last name of the painter only,
then the name in conjunction with “art”; and then the
name plus “Impressionism.” The results are shown in
Table 2. None of these was reliably correlated with the
frequencies in the Cornell collections or in the BHA
(rs < .36, n.s.). Monet and Renoir were overrepresented,
and Cézanne and Manet were underrepresented, as com-
pared with the Cornell collections and the BHA.

On the Representativeness of the Images
The second methodological question concerns the re-

lation of the stimuli used here to the Impressionist
canon. To this end, an index of obtained versus expected
(O/E) values was created for the images of each painter.
Its numerator for each artist is the total number of Cor-
nell books in which any of his stimulus images appeared.
This is the obtained value, O (column 8 of Table 2). The
denominator is the expected value, E. This is the total
number of Cornell books accessed using the ar tist’s
name as a keyword and screened for content (column 2
of Table 2), plus the number of books accessed using the
open-ended keyword “Impression$.” In the latter count,
books on Postimpressionism were excluded, and the
residual was checked for relevant content and for the
presence of many images.10 These totaled 91. The logic
of the index is as follows: If the artist’s works among
these stimuli were central to his oeuvre, one might rea-
sonably expect that at least one of them would appear in
every book on the painter and in every book on Impres-
sionism generally. In this manner, if the stimuli were rep-
resentative, index values (O/E ) should be near 1.0.

The results are shown in the column 9 of Table 2. As
one can see, Renoir’s index (2.04) was more than twice
what might be expected. This is in good measure due to
his Bal du Moulin de la Galette, part of the Caillebotte
legacy. His index would be nearer unity (1.43) without it.
Three other indices were reasonably above 1.0: those for
Pissarro (1.46), Degas (1.34), and Monet (1.28). Those
of Manet and Cézanne were less, partly because the
numbers of their paintings used here were fewer than
those for the other artists. In addition, Manet’s was less
(0.93) also because of extensive literatures on his Déjeu-
ner sur l’herbe (1863), his Olympia (1863; both in the
Orsay), his Le Bar aux Folies-Bergère (1882; Courtauld
Institute, London), and works on Spanish influences and
topics in his art. Except for the Folies-Bergère, Caille-
botte’s collection of Manet’s work, and images here
matched to them, was subsequent to these. In a reverse
manner, Cézanne’s index (0.69) was considerably lower,
because Caillebotte’s collection was prior to many more
important works, such as the near-Cubist series of La
Montagne Sainte-Victoire. Indeed, these are subsequent
to Caillebotte’s death. Finally, Sisley’s index was lower
still (0.56) for at least two reasons. First, his works are
increasingly excluded from presentations of Impression-
ism, and second, his images from the more celebrated se-
ries on the floods at Marly (e.g., L’Inondation à Port-
Marly in the Orsay) were not included here. The fact that
indices for these seven artists are generally greater than
1.0 (mean index = 1.19), coupled with the high correla-
tion of works in the Cornell collections with those in the
BHA database, suggests that these stimuli form a suit-
able and representative sample of the Impressionist
canon.

Table 2 also shows two related indices in the last two
columns: that for the representativeness of the Caille-
botte images for each painter to the Impressionist canon
and that for those images matched to them. Across
artists, the means of these indices were 0.77 for both the

Table 2
For Each Painter, the Number of Sources Found in Electronic Searches, the Number of Books in the Cornell

University (CU) Libraries With Their Im ages, and the Indices of the Representativeness of their Images in the 
Entire Stimulus Set, Those in the Caillebotte Collection, and Those of the Im ages Matched to Them

Number
of BooksOn-Line Searches

in CU Indices of
Academic and Google (31,000) Libraries Obtained/Expected

Popular Sources Painter With Frequencies

CU Painter +“Impres- These Caille- Comparison
Artist Libraries BHA* Amazon Painter +“Art” sionism” Images All† botte Images

Caillebotte 16 , 90 8 8.8 5.0 1.6 61 – – –
Cézanne 243 1,200 83 116.0 47.2 8.7 230 0.69 0.39 0.53
Degas 156 ,853 119 131.0 62.0 14.3 330 1.34 0.70 1.09
Manet 168 ,942 84 155.0 48.2 9.3 240 0.93 0.68 0.52
Monet 156 ,800 130 846.0 230.0 25.2 316 1.28 0.79 0.89
Pissarro 60 ,414 27 41.6 20.3 5.8 221 1.46 0.95 0.92
Renoir 115 ,372 73 298.0 105.0 13.0 420 2.04 1.46 1.11
Sisley 33 , 77 8 153.0 14.2 3.5 70 0.56 0.38 0.32

*BHA is the Bibliography of the History of Art. †The obtained variable is the number of books in the Cornell libraries with any
of the selected images by each painter (column 8). The expected variable is the number of books found in the on-line catalog, using
the painter’s name (column 2), plus the 91 books in the catalog on Impressionism, screened for content.



328 CUTTING

Caillebotte images and their comparisons, suggesting
that the two groups are well matched in representing
each artist’s work.

STUDY 1
Frequency Counts, Recognition, and 

Preferences for Images From the 
Impressionist Canon and Corpus

Using the frequency counts of the images as an esti-
mate of the likelihood of prior exposure, I set out to de-
termine whether there was a relationship between the
counts and viewers’ preferences for the images and be-
tween the counts and viewer recognition. Also of inter-
est were the relative preferences for, and recognition of,
the Caillebotte and Orsay images.

M ethod
All JPEG files were commercially made into 5 3 5 cm slides.

Two carousel trays were prepared, with members of each pair in
corresponding slots— one image from the Caillebotte collection in
one tray and one matched to it in the other. Their order in sequence
was random, with several constraints: No images by the same artist
could follow one another; black-and-white pairs could not follow
each other; half of the Caillebotte images appeared on the left, and
half appeared on the right; and 31 images from the Musée d’Orsay
appeared on the left, and 41 on the right.

Stimulus pairs were presented for about 8 sec each. They were
projected onto a large screen by two projectors. The smaller di-
mension of each image (vertical if landscape, horizontal if portrait)
projected a size of about 8º from the middle of the auditorium.
Viewers were 166 Cornell University undergraduates enrolled in a
perception course. They were asked to look at each pair and judge
which image they liked best. In addition, should they recognize any
image, they were to mark which of the two, or both, they recognized
of each pair. No record was kept of where the viewers sat. Previous
research using a wide variety of stimuli seen from a wide variety of
locations in such situations has shown no effect (Cutting, Wang,
Flückiger, & Baumberger, 1999; Gibson, 1947; Kozlowski & Cut-
ting, 1977).

The viewers also filled out a brief questionnaire, indicating how
many times they had visited the Johnson Art Museum on the Cor-
nell University campus each year (54% said at least once a year),
the number of times they had visited any other museum (52% said
at least once a year), the number of times they had ever visited the
Musée d’Orsay (8% had visited at least once), and how many art
history courses they had taken (only 16% had taken any). The stu-
dents also reported spending a mean of 10.8 hours per week on the
Internet (median = 10, SD = 9.3).

Results and Discussion
The results divide several ways. In increasing order of

psychological interest, f irst the results pertinent to the
Caillebotte collection and to that of the Musée d’Orsay
and then those interrelating recognition, preference, and
frequency will be considered.

Caillebotte and the Orsay. Across the database, im-
ages from the Caillebotte collection did not appear with
any different frequency than their matched pairs. Means
were 33.5 (SEm = 5.8) and 30.6 (SEm = 5.1, n.s.), respec-
tively. In addition, the Caillebotte images were not claimed
to be recognized with any reliably different frequency—

2.5% versus 2.9% (n.s.), respectively. And f inally, the
viewers expressed no preference for the Caillebotte im-
ages, choosing them 47.5% of the time (vs. 52.5%, n.s.).
Thus, there is nothing unusual here about the paintings
and pastels in the Caillebotte collection. He cannot be
said to have had extraordinary taste in the selection of
his images over and above the selection of other images
by groups of other collectors—at least for these com-
parisons, as judged by a contemporary, relatively naive,
but appreciative artistic audience.

Next, consider the Musée d’Orsay. Images in its col-
lection appeared significantly more often than those
elsewhere. Means were 43.1 (SEm = 6.0) versus 20.9
(SEm = 3.2, p < .0001), respectively. This difference oc-
curred even when the comparison was restricted only to
those images in all the other museums (43.1 vs. 27.5, p <
.005). This is not a surprise, since the French govern-
ment and the Réunion des Musées Nationaux of France
have been thorough in promoting their art for a long
time. However, the viewers did not claim to recognize
the images in the Orsay more often than others—2.8%
versus 2.6% (n.s.), respectively. As might be expected,
viewers did prefer them somewhat more often—54%
versus 46% [r = .32; t(38) = 2.14, p < .04]—but when
frequency differences were factored out, there was no ef-
fect of images being in the Orsay (t < 1, n.s.). Thus, what
distinguishes this selection of the Orsay holdings is only
that its images appear more often.

What about Varnedoe’s (1987, p. 202) claim that the
state of France got the “cream” of the Caillebotte col-
lection? Of course, an empirical analysis of the kind here
carries a different force than an academic or professional
assessment. Yet one must be wary of how this retrospec-
tive prophecy may have been fulfilled; certain of Caille-
botte’s images went to the state of France, which made
them available and promoted them, which we now re-
vere. It is unequivocal that the Caillebotte images in the
Orsay appear more often than those that are not (means
of 49.7 vs. 9.6, p < .0001).11 And indeed, the Caillebotte
Orsay images were somewhat preferred over their matched
pairs (53% to 47%, n.s.), whereas those not in the Orsay
were not (42% vs. 58%, p < .01). But once the difference
in relative frequency is factored out [r = .50; t (63) =
4.08, p < .0002], there was no residual effect of images
being in the Orsay (r = .24; t < 1, n.s.). Thus, being in the
Orsay does not make a painting part of the canon inde-
pendently of how often it appears. Instead, an image may
appear more often because it hangs in the Orsay (and
hung in its predecessor museums), and appearing there
often goes some distance toward maintaining an artwork
in the canon as acknowledged by professionals and the
public.

Finally, paintings and pastels that reside in any museum—
the Orsay or elsewhere—appeared reliably more often
than those in private collections (37.4% vs. 5.6%; p <
.0001), they were recognized more often (3.3% vs. 1.5%;
p < .0001), and when paired directly they were preferred
more often (61% vs. 39%; p < .001). None of this, of



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 329

course, is a surprise. Essentially, by definition, artworks
in private collections cannot be in a canon. What drives
all of this, at least statistically, would appear to be fre-
quency of appearance. The other relationships among
frequency, recognition, and preference are the center-
piece of the f indings in this study and are a bit complex.

Recognition and frequency. Viewers claimed to rec-
ognize only 2.7% of all the images, but this varied ac-
cording to observer experience, as is shown in the top
panel of Figure 3. Those having never taken an art his-
tory course, or having taken only one, recognized 1.6%
of all images, and those having had at least two courses
recognized 11.9% ( p < .001). Those claiming not to go
to an art museum each year recognized 1.1% of the im-
ages, whereas those claiming to go at least once a year
recognized 4.1% ( p < .001). Finally, those never having
been to the Musée d’Orsay recognized 1.9%, whereas
those having been at least once recognized 13.4% ( p <
.001). Multiple regression showed that the number of im-
ages viewers claimed to recognize was correlated with
how often they went to any museum (r = .49, p < .0001)
and how often they had been to the Musée d’Orsay (r =
.49, p < .0001), but once these were factored out, not
with how many art history courses they had taken (r =

.35, n.s.). The f irst two factors accounted for 39% of the
variance in recognition judgments across viewers (R =
.62, p < .0001). Recognition of individual images ranged
from 0% (2 Manets, 9 Monets, 10 Pissarros, 2 Renoirs,
and 3 Sisleys) to 25.4% (Renoir’s Bal du Moulin de la
Galette, Image 54c and Figure 2E). Rates for each of the
132 images are given in Table 1. The general patterns
will be discussed here.

Analyses were then conducted predicting viewers’
recognition of all the images from three different sets of
frequencies: (1) those from general texts, (2) those from
more topical texts with narrower scope but not featuring
a single artist, and (3) those from more focused texts and
monographs featuring a single artist. The counts from
the first group seem the most f irmly in the Impression-
ist canon, those from topical books a bit less so, and
those from more focused texts less firmly still, leaving
those in none of the books except the catalogues raison-
nés in the broad base of the corpus. Thus, in moving
from high canon to corpus, it is best to compare the fre-
quencies of images in (1) the most general texts with
(1+2) those in the general group plus the topical texts,
with (1+2+3) those in all texts. These can be conceived
as different slices through likely cultural exposure.

Figure 3. The contrasts between recognition and preference results as a function of
observer sophistication and experience in Study 1. Standard errors are shown.

.16

.12

.08

.04

.60

.55

.50

N Y Y N Y N Y1 2+

N YN YN Y
Took An Art

History Course?
Visited a Museum
in the Past Year?

Ever Visited the
Musée d’Orsay?

P
ro

b
ab

il
it

y 
of

 P
re

fe
rr

in
g

th
e 

M
or

e 
F

re
q

u
en

t 
Im

ag
e

P
ro

b
ab

il
it

y 
of

 R
ec

og
n

it
io

n

n = 14
n = 137 n = 13 n = 77 n = 87 n = 151 n = 13



330 CUTTING

Consider a hypothetical example: If an image ap-
peared in five general texts, 15 topical texts, and 25 mono-
graphs, the numerical values for the three classes would
be 5, 20, and 45. Psychophysical practice suggests that
these should be logarithmically transformed. Since zeros
occur (most images never appear in general texts) and
since the log of zero is undefined, the natural log (n + 1)
was used, where n is the value under consideration (More-
land & Zajonc, 1977). Thus, the values above would be
log (5 + 1), log (20 + 1), and log (45 + 1), or 1.79, 3.04,
and 3.83, respectively. Since such measures are always
correlated, multiple regression is the best statistical tool
for titrating effects of the appearances of images in the
different types of academic sources.

Claims of image recognition were most correlated
with their log frequency of occurrence in general art
texts [r = .46; F(1,128) = 7.5, p < .01], but not with the
sum of the general and topical texts [r = .41; F(1,128) <
1, n.s.] or with occurrences across all texts [r = .43;
F(1,128) = 1.8, n.s.]. The multiple correlation accounted
for 23% of the variance in the data [R = .48; F(3,128) =
12.0, p < .0001], but clearly, almost all of the effect came
from the frequencies of images in the most general
sources. This result is not a surprise. Those images
should be the most recognizable. However, in this con-
text, it is important to note that there is currently no in-
troductory art history course at Cornell and that only 27
of 166 (16%) undergraduates had taken any art history
courses. Thus, I claim that recognition reflects general
knowledge of particular viewers, not their perusal of in-
troductory art history textbooks or encyclopedias in the
Cornell libraries or elsewhere.

Most of the image pairs were in color, but 15 were
black and white. Did this effect recognition? Yes. That is,
when the differential frequencies were factored out,
there remained an effect of color: Color images were
claimed to be recognized 3.7% of the time, but black-
and-white images only 0.7% [r = .31; F(1,127) = 9.52,
p < .003].12 The most frequently occurring black-and-
white stimulus was Image 57n. If comparison color im-
ages are restricted to its value (n = 44) and less, the dif-
ference is still reliable (2.8% vs. 0.7%, p < .01). Color
also improves recognition of objects and scenes in labo-
ratory studies (Joseph & Proff itt, 1996; Oliva & Schyns,
2000; Wurm, Legge, Isenberg, & Luebker, 1993).

Note that no assumption is made that the observers’
responses necessarily represent the true recognition of a
particular painting or pastel. There is no way to verify
them. Nonetheless, there are additional interesting trends.
For example, against a backdrop recognition of 2.7%, the
16 images by Degas were recognized at a rate of 6.1%
( p < .001), and the 7 of his images that were clearly
dancers were recognized at a mean rate of 9.4%. Recog-
nition of the dancer images, it would seem, is an exam-
ple of generic recognition—recognition by that individ-
ual only that he or she had seen images of Degas-like
dancers before (for a laboratory analogue, see Monahan,
Murphy, & Zajonc, 2000). Given that there are 600 pas-
tels and paintings of dancers in the Degas catalogue

raisonné (Lemoisne, 1946), this is perhaps not entirely
surprising.

Preference and frequency. Herein lie the key results
of the study. Over all pairs, the viewers preferred the
more frequently occurring image of each pair on 59% of
all the trials. This highly reliable effect [t(165) = 9.09,
p < .0001] is about the size of many mere exposure ef-
fects in the literature (e.g., Seamon & Delgado, 1999).
Indeed, 48 of the more frequent images in 64 pairs were
preferred (with one tie in preference and one in fre-
quency; z = 3.9, p < .001). Unlike the recognition results,
this effect was uniform across all types of observers, as
is shown in the bottom panel of Figure 3. It occurred
equally for those who never had had an art history course
[59%; t (130) = 8.14, p < .0001] and those who had taken
at least one such course [59%; t(27) = 4.01, p < .0001].
It occurred for those not visiting a museum in the past
year [59%; t(79) = 5.1, p < .0001], those visiting once
[59%; t(46) = 6.8, p < .0001], and those visiting at least
twice [58%; t (38) = 5.4, p < .0001], and it occurred
equally for those not visiting and those visiting the
Musée d’Orsay (59% each, ps < .001). Importantly,
when differences in recognition rates were compared
with preferences for each of the 66 images pairs, there
are no reliable correlation (r = .18, n.s.).

Bornstein (1989), among others, noted that although
mere exposure effects are a function of number of expo-
sures, the effect asymptotes with increasing numbers of
presentations. Preferences were then compared against
the difference in log values of the images for the three
levels of texts in which they appeared. Again, if one
image appeared in f ive encyclopedias, 15 topical texts,
and 25 monographs, the natural log transformation of the
frequencies of 6, 21, and 46 would be 1.79, 3.04, and
3.83, respectively. If its comparison image appeared in 0,
25, and 40 such texts, the log of 1, 26, and 66 would be
0.0, 3.25, and 4.19. Differences in log values would then
be 1.79, 20.21, and 20.36. These values would be used
as inputs to the regression.

Multiple regression analysis showed that viewer pref-
erence was not correlated with the difference in log fre-
quency of image appearance in the most general texts
(r = .36, F < 1, n.s.) or with the differences in the sum of
frequencies in the f irst group and the topical texts (r =
.37, F < 1, n.s.). However, it was reliably correlated with
the difference in frequencies in all occurrences [r = .54;
F(1,62) = 11.5, p < .001]. The multiple correlation on
preference results accounted for 30% of the variance in
the data [R = .55; F(3,62) = 9.1, p < .001], again with
most of the effect coming from a single source—this
time, differences in overall frequency. Overall frequen-
cies for each image, I claim, act as a proxy for the like-
lihood and frequency with which an individual may have
been exposed to that image in his or her broader cultural
experience. Interestingly, the correlation between pref-
erences and the differences in occurrences in all books
(r = .54) is the same regardless of whether all the occur-
rences of the images since 1901 are considered, or only
those since 1989. The latter, of course, are likely the oc-



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 331

currences correlated with what is most relevant to these
viewers. This result also speaks strongly to the stability
of the canon of images across the course of the 20th cen-
tury, a topic to which I will return in the conclusion.

A variety of other frequency transforms were correlated
with preference results, but none provided as strong a
correlation as log (n + 1). Given this result, the greater
the difference in the logarithms, the stronger the prefer-
ence should be. Indeed, this is what occurred. Figure 4
shows the relative preferences for the more frequent image
across three groups of image pairs. The increasing trend
seen in the f igure is what drives statistically reliability
[r = .54; t(64)= 5.1, p < .0001].

Unlike recognition responses, preferences were not af-
fected by color versus black-and-white presentation.
That is, once the differences in relative frequencies of
the various types of books are factored out, there is no re-
maining contribution of color [F(1,61) < 1, n.s.]. Nonethe-
less, without considerations of frequency, the more fre-
quent image in a black-and-white pair was actually
preferred more often (67%) than was the more frequent
image of a color pair [57%; t(62) = 2.37, p < .01]. This
would appear to be due to variance in small numbers:
The generally lower frequencies of occurrence for black-
and-white pairs generated greater mean difference ratios
(4.9:1) than for color pairs (2.8:1).

In addition, all digital f iles were inspected for mean lu-
minance and RGB values. Preference within pairs was not
related to differences in luminance (r = .11, p > .35) or to
distributions of red (r = 2.04), green (r = 2.05), or blue
(r = .03) colors. Indeed, the multiple correlation of all
these variables was also not significant (R = .09, p > .20).

STUDY 2
Preference and Recognition in an Older Group

The goals of this second study were to replicate the
primary results of Study 1 with an older group. In par-

ticular, the f irst study showed preferences related to dif-
ferences in frequencies of occurrence for pairs of images
and no relation between preference and recognition. This
joint result is important for demonstrating mere expo-
sure. Nonetheless, the lack of relation could be due, in
part, to a floor effect, because so few of the images were
claimed to be recognized. An older group would likely
recognize more.

Method
Twenty-five pairs of images were selected from the 66 pairs used

in Study 1. These are indicated in Table 1—1 Caillebotte,  3
Cézannes, 4 Degas, 3 Manets, 4 Monets, 4 Pissarros, 4 Renoirs, and
2 Sisleys. These were among the more frequent in the database of
images from all books (mean frequency of 61.0 vs. overall mean of
32.1 in Study 1). They were also among the more recognized (mean
of 4.8% vs. 2.7% in Study 1). No black-and-white pairs were used
in this study. To vary presentation format, digital images were
mounted in pairs on the same slide as part of a PowerPoint presen-
tation. The left–right arrangement of half the pairs was reversed
from that of Study 1, and the presentation order was also changed,
but otherwise with the same constraints. In all, 13 Caillebotte im-
ages appeared in the left, 12 on the right; 14 Orsay images were on
the left, 16 on the right. For most of the viewers, the largest dimen-
sion of each image subtended about 8º. As a group, 9 faculty mem-
bers (mean age = 46 years) and 10 graduate students (mean age =
25 years) viewed the 25-pair sequence; 12 had visited the Musée
d’Orsay at least once. Each made preference judgments and marked
which images they remembered having seen before.

Results and Discussion
This group reported recognizing 18.6% of the images,

many more than in Study 1 [t(47) = 7.8, p < .0001]. Indi-
vidual image rates are given in Table 1. Claims of recog-
nition were unrelated to frequencies of appearance in all
books (t < 1, n.s.) and in general tests [t(47) = 1.2, n.s.] and
were unrelated to their preferences [r = 2.31; t(23) = 1.6,
n.s.]. There was no difference in recognition of Caillebotte
versus comparison images and, independent of frequency,
no superior recognition of Orsay images (ts < 1, n.s.).

Nonetheless, as in Study 1, the viewers preferred the
more frequent image of each pair—here, 57% of the time
[t(23) = 3.6, p < .002]; 17 of 25 pairs showed this effect
(z = 2.2, p < .02). Rates for each comparison are given in
Table 1. This set of images had about the same prefer-
ence margin in Study 1 (58% vs. 42%). In addition, there
was a high correlation of preferences among the 25
image pairs used in the two studies (r = .84, p < .0001).
Again, preferences were related to the log difference in
frequencies across all books [r = .57; F(1,22) = 4.7, p <
.05], but less so to the log difference in frequencies in
general texts (r = .43, F < 1). Independent of frequency,
however, there was no preference for Caillebotte images
or Orsay images (ts < 1, n.s.). Thus, the major features
of the results of Study 1 were replicated.

STUDY 3
Preferences of Children

The goal of this study was to determine whether the
pattern of preference results found in Studies 1 and 2

Figure 4. Difference in the ratio of overall frequency of the im-
ages in the pairs and the mean preference of the more frequent of
the pair in Study 1. In general, images were preferred over their
m ates as a function of how much more often they occurred.



332 CUTTING

might be found in a group of children. Should children’s
preferences match those of previous studies, something
other than mere exposure must be at work; children sim-
ply lack broad exposure to art. Here, I also wanted to
focus on a group of images that would be least likely to
have been seen before.

M ethod
Twenty-four different pairs of images were selected from the 66

pairs. These are indicated in Table 1: 2 Cézannes, 2 Degas, 2
Manets, 7 Monets, 5 Pissarros, 3 Renoirs, and 3 Sisleys. Nineteen
pairs contained images that were among the least frequent from all
the books— mean image frequency of 15.0 versus 32.1 in Study 1.
These will be called rare images. For comparison purposes, f ive
pairs containing the most frequent images were also included—
Image Pairs 6, 8, 18, 22, and 54, with a mean frequency of 146 per
image. These will be called common images. Mean Study 1 prefer-
ence rates for the more frequent images were 59% for both the rare
and the common pairs. Nine of the 24 pairs had been used in
Study 2; 5 were common-image pairs. Again, no black-and-white
pairs were used in this study. Again, digital images were mounted
in pairs on the same slide as part of a PowerPoint presentation. The
left–right arrangement was balanced and randomized for the Caille-
botte images and for the more frequent images, and the presentation
order was randomized anew. Otherwise, the same constraints were
followed as those in Study 1.

Sixty-three students in a Montessori school participated in three
groups as part of an art class. There were 13, 28, 19, and 10 stu-
dents, whose ages were 6, 7, 8, and 9 years old, respectively. In ad-
dition 5 teachers participated. Each made preference judgments on
an answer sheet. Particular care was taken that all the students un-
derstood the instructions. Pairs were presented for a minimum of
10 sec each. Questions were allowed during the tests, so presenta-
tion rates varied.

Results and Discussion
These children showed no preference for the Caille-

botte (49.5% vs. 50.5%) or Orsay (45.2% vs. 54.8%; ts <
1, n.s.) images. More important, they also showed no
preference for the more frequent image of each pair—
51.2% versus 48.8% [t (22) = .26, p > .75]. Common
(49.4% vs. 51.6%) and rare (52.0% vs. 48.0%) pairs did
not differ in this regard. Interestingly, mean teacher pref-
erence was 60% for the more frequent image of each pair
[t(22) = 4.1, p < .06]. The children’s responses, however,
were far from random; they seemed to like paintings with
brighter colors. Responses of each age group correlated re-
liably with those of all other age groups [rs > .63; ts(22) >
3.9, ps < .001], but they did not correlate with the pref-
erences of the adults of Study 1 for the same images (r =
2.16, p > .40). Overall preferences for more frequent
images were 48.7%, 47.1%, 53.0%, and 55.8%, respec-
tively, for the 6-, 7-, 8-, and 9-year-old groups. Nonethe-
less, the variance was large, and the apparent increasing
trend was not reliable, either for all the pairs [t(61) =
1.56, p > .12] or for the rare pairs [t(61) = 1.6, p > .10].

Thus, whatever governed the adult preferences in
Studies 1 and 2 is not operative in the preferences of chil-
dren. This is an important null result. Although elemen-
tary school children may have seen a few Impressionist
paintings before, they lack the broad cultural exposure to
Impressionist art that adults have experienced.

STUDY 4
Complexity Is Not a Mediator of Preference

The goals of this fourth study were twofold: to replicate
the preference results of Studies 1 and 2 and to explore
the possible contributions of what is called stimulus
complexity. Bornstein (1989) reviewed the literature on
the relation between complexity and preference in labo-
ratory experiments on mere exposure. He found that six
of nine published studies found stronger mere exposure
effects for complex stimuli than for simple stimuli, but
more relevant are those using art works as stimuli. Here,
the results are mixed. As was mentioned earlier, Berlyne
(1970) and Zajonc et al. (1972) found opposing effects:
one favoring a preference for complex stimuli, the other
for those simpler. Nonetheless, the results of neither of
these studies are generally applicable to representational
art. Both confined themselves to abstract art (where bits
and pieces could be counted in the context of complex-
ity), and the latter carved up the artworks (under the as-
sumption that a detail is less complex than the whole).
Since complexity cannot be defined with rigor in many
domains (Goodman, 1972), I let the observers define it
for themselves.

Method
The viewers were 112 students from a classroom population, dif-

ferent from those in Study 1, but taking the same class a year later.
Only 17% had taken an art history course, and 7% had been to the
Musée d’Orsay. The same 25 pairs were used as those in Study 2,
again in a PowerPoint presentation, but shown in a different order
and with some counterbalanced changes in left– right positions.
Again, pairs were presented for about 8 sec each. Viewers were
asked to make a judgment about which image in each pair they pre-
ferred. In addition, after the first run through of the stimulus set, the
pairs were presented a second time, and the viewers were asked to
make judgments about which image was more complex.

Results and Discussion
Again, from the database tallies of all the books, the

more frequent image of each pair was preferred over its
counterpart [58% vs. 42%; t(24) = 16.7, p < .0001], the
same preference margin for these images as in Study 2.
Again, 17/25 pairs followed this pattern (z = 2.2, p <
.02), but 4 pairs changed polarity of preference. Overall,
preferences here were well correlated with those for the
same images in Study 1 (r = .80, p < .0001). Individual
pair results are again shown in Table 1. The correlation
of preference with frequency differences across all the
books was reliable and about the same as that in Study 1
[r = .47; F(1,22) = 6.3, p < .02], and again, that for gen-
eral texts was not [r = .21; F(1,22) = 1.0, n.s.]. Finally,
the Caillebotte images were not preferred over their
matched pairs (48% vs. 52%, n.s.), Orsay images were
not preferred either (49% vs. 51%, n.s.), and preferences
were unrelated to museum trips or art history courses
taken.

Complexity judgments were not correlated with pref-
erences (r = .18; t < 1, n.s.). Individual means for each
image are shown in Table 1. Inspecting the relative judg-



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 333

ments for each pair suggests that viewers were simply
counting things, usually people, in the pictures. Consider
four image pairs exhibiting strongly judged complexity
differences. Degas’s Femmes à la terrase d’un café, le
soir (“Women outside a café in the evening”) was judged
as more complex than his L’Absinthe 81% of the time
(Image Pair 8, Figures 1E and 1F). This may be because
there are f ive people in the foreground of the former
(with many in the background) and only two in the lat-
ter. Renoir’s Bal du Moulin de la Galette was judged
more complex than his Le Déjeuner des canotiers (“Lun-
cheon of the boating party;” Image Pair 54, Figures 2E
and 2F) by 85% of the viewers. This is likely because the
former has uncountably many people and the latter but a
dozen. Pissar ro’s La Côte des Boeufs à Pontoise was
judged as more complex than his Les Toits rouges (“The
red roofs”) 87% of the time (Image Pair 43, Figures 1D
and 1C). Both are views of the Hermitage in Pontoise in
winter, but the former has more trunks of tall trees in the
foreground. Finally, Manet’s Le Déjeuner à l’atelier
(“Lunch in the studio;” with three people) was judged
more complex than his Le Balcon (with three people
clearly visible and a fourth in shadow; Image Pair 18,
Figures 1H and 1G) 91% of the time. This is probably
because the table in the former is cluttered with food,
and the scene is cluttered with other objects. There is no
such clutter in the latter.

STUDY 5
Prototypicality is Not a Mediator of Preference

Another account for the preference results of the Stud-
ies 1, 2, and 4 might be that viewers, when faced with
making preference judgments, were comparing images
on the basis of what they thought were the most repre-
sentative (prototypical) Impressionist paintings. This
study was designed to address this issue.

M ethod
Twenty-one undergraduate students in an advanced visual percep-

tion seminar viewed a PowerPoint sequence of 138 images—all 132
used in Study 1, plus 6 more by Gustave Caillebotte. 13 These images
were presented singly, and viewers rated them on a scale of 1 to 7 as
to how representative each was of Impressionist paintings, with 7
being the most prototypical . Presentation was haphazard, with the
constraints that images by the same artist could not follow one an-
other, nor could black-and-whit e images. Six students had participated
in either Study 1 or 4, and 6 had taken at least one art history course.

Results and Discussion
Prototypicality judgments were not correlated with

the experimental variables of previous interest—the fre-
quencies of the images in all the texts or in the introduc-
tory texts (rs = 2.05 and .06, respectively, n.s.) or the
recognition rates in Studies 1 and 2 (rs < .20, n.s.). The
differences in prototypicality judgments within a pair
were also not correlated with preferences in Studies 1, 2,
or 3 (rs < .14, n.s.) or complexity judgments in Study 4
(r = 2.05, n.s.). Prototypicality judgments for individual
images are shown in Table 1. These null results may

seem odd. After all, the most frequently occurring im-
ages are the prototypes of the canon. Nonetheless, it
seems that the viewers were making judgments in a dif-
ferent way.

There were some striking effects of prototypicality
judgments by painter. Most prototypic were the 16 im-
ages of Sisley (mean rating = 5.28). This is interesting,
because he is clearly the least major of the seven “major”
Impressionists, as suggested by the book counts in
Table 2. Clustered next and together were the 32 works
of Monet (4.96), the 28 of Pissarro (4.93), and the 18 of
Renoir (4.96), with the f irst two reliably different from
Sisley ( ps < .05). Clustered next, and reliably below
these four, were the 16 works of Degas (4.59) and the 10
of Cézanne (4.24). Finally, well below these were the 8
of Manet (2.81) and the 10 of Caillebotte (2.77). Inter-
estingly, Cézanne and Manet are often described as not
really being Impressionist painters (Cézanne’s most im-
portant works are later than the period of the 1870s and
1880s, and Manet’s earlier), and Caillebotte as well. In
addition, Degas never painted outdoors, which may have
influenced judgments.

Other classifications of images also show some inter-
esting differences. Of this set of 138 images, 90 can be
classified as landscapes, 44 as portraits (often of groups
and often outside), and 4 as still lifes. Mean ratings for
landscapes (4.96) were reliably higher than those for
portraits [4.14; t(132) = 4.71, p < .0001] and still lifes
[3.75; t(92) = 5.2, p < .0001]. Portraits and still lifes did
not differ. It should be noted that, among these images
and throughout their oeuvres, Sisley painted only land-
scapes, Pissarro mostly landscapes, Renoir mostly por-
traits, and Degas almost exclusively portraits.

OVERVIEW

Preference
The adult viewers of Studies 1, 2, and 4 generally

liked the more frequent images of each pair, but the chil-
dren of Study 3 did not. The effect in the adults was
salient for differences measured across all the books in
the Cornell databases; preference was not independently
related to frequency differences in general texts. Prefer-
ence strength was not a function of whether or not im-
ages were in color (Study 1; in Studies 2 and 4, all were
in color), nor were they a function of whether or not the
observers took trips to museums or attended art history
courses.

Recognition
In Study 1, viewers recognized few of these Impres-

sionist images—less than 3%. Low recognition rates are
requisite for laboratory demonstrations of mere expo-
sure (Kunst-Wilson & Zajonc, 1980). In that study, recog-
nition rates were related to their frequency in general
texts in the Cornell collections, but not to their frequency
in all the books. Recognition was more frequent for color
images, and it was also related to viewers’ visits to the
Musée d’Orsay and other art museums. In Study 2, a



334 CUTTING

more seasoned set of viewers recognized 18% of a smaller
set of images, but their recognition rates were not related
to any frequency counts. In Studies 1 and 2, the viewers’
recognition rates and preferences were not related, an-
other requisite for laboratory demonstrations of mere ex-
posure. This pattern has been recognized for quite some
time (Zajonc, 1980), and there is neurophysiological ev-
idence in its support (Elliott & Dolan, 1998).

Art Collections
Several conclusions can be drawn. First, at least with

respect to this type of analysis and experiment, images
from the Caillebotte collection were neither preferred
more often nor recognized more often than those matched
to them, and the Caillebotte images were not more fre-
quently occurring. Second and in contrast, the Musée
d’Orsay holdings did occur more often in this sample.
This is not a surprise. A systematic culling of images
from the Orsay, which owns one tenth of all Impression-
ist paintings publicly available, would form the bulk of
the core of the canon.14 However, with the same caveats
as above, the Orsay’s holdings were neither preferred nor
recognized more often than other images matched to
their frequencies of occurrence. Third, art in private col-
lections is not in the Impressionist canon. These images
occurred less often in the literature; they were less fre-
quently recognized; and, lacking exposure, they were
preferred less.

On the Availability of Art
Together, all of these trends support the idea that it is

not where an image is, or who bought it, but how often it
appears that affects public appreciation. Any artwork in
a prized location—such as in the Musée d’Orsay—has a
great advantage over other artworks, but systematic pro-
motion by other museums and authors can overcome this
advantage.

Preference Judgments and Artistic Quality
Surely the most interesting result of these studies is

the relation of viewer preferences to how often images
occurred in the Cornell libraries. Mere exposure aside,
how else might this effect be accounted for? In dis-
cussing these results with colleagues, quite a few have
suggested that perhaps viewers can simply judge quality,
choosing the “better” picture. I don’t believe this for a
moment, but this is not the place to deal at length with
this important and thorny issue. Here, let me simply ac-
knowledge that there are many statements outside the art
historical literature (e.g., Kant, 1794/1952; Pirsig, 1974)
about people’s culturally independent ability to judge
quality, as well as many within it (e.g., Rosenberg, 1967;
Woodford, 1983). Nevertheless, there are also more re-
cent and culturally sophisticated counters to this claim
(e.g., Bal & Bryson, 1991; Cheetham, 2002; Moxey,
1994). Indeed, even Kenneth Clark, director of the Na-
tional Gallery London in his 20s and, years later, a BBC

icon of the visual arts, wrote in his memoirs (Clark,
1974):

At the age of nine or ten I said with perfect confidence
“this is a good picture, that is a bad one” . . . This almost
insane self-confidence lasted till a few years ago, and the
odd thing is how many people have accepted my judge-
ments. My whole life might be described as a long, harm-
less confidence trick. (p. 47)

Nonetheless, rather than trying here to rule out quality as
a mediator of these results, let me try one last time to
rule in mere exposure in a different way. This also pro-
vides an experimental effect that the notion of quality
cannot explain.

STUDY 6
Preferences From the Combination of Cultural

and Classroom Exposure

If mere exposure mediates preferences for artworks, it
ought to be possible to combine the effects of exposures
to art across two situations—the viewers’ personal his-
tories with these images outside the classroom and class-
room exposure to them.

Method
All 51 color pairs from Study 1 were used as stimuli. During 21

class periods in an introductory perception course (a year after
Study 4 and 2 years after Study 1) and at the beginning of each lec-
ture, students were presented 12 (on 18 days) or 13 (on 3 days) im-
ages, for a total of 255 presentations, 5 per pair. Across the 21 ses-
sions and for each pair, the less frequent image from the Cornell
Library tallies (see Table 1) was presented four times; the more fre-
quent image was presented only once. Each image was presented
singly for about 2 sec as a PowerPoint slide, without comment. Pre-
viously, Harrison and Zajonc (1970), Marcus and Hakmiller (1975),
and Vanbeseleare (1983) found no effects of presentation rates in
the range from 2 to 50 sec; it is the presentation that matters, not its
duration. Several constraints governed the composition of a day’s
set of slides: Images that would later be a test pair were not both
presented, and no more than two images by the same artist could ap-
pear successively. More important, the same image did not appear
twice on the same day or even on successive days.

On a 22nd class period, the 51 pairs of images were presented
side by side in a PowerPoint presentation for about 6 sec per pair in
a new random sequence. Insofar as possible, three presentation fac-
tors were counterbalanced, left and right— more frequent images
(25 left, 26 right), Caillebotte images (26/25), and Orsay images
(28/30). In addition, no pairs of images by the same painter fol-
lowed one another. The delays between exposure(s) and test ranged
between 12 and 84 days. Although no studies in Bornstein’s (1989)
meta-analysis used delays as long as these, the general result across
the literature is that short delays produce weak effects, those up to
two weeks relatively strong effects, and naturalistic studies (such
as those presented here) with indeterminate delays often produce
the strongest effects. Indeed, Harrison (1977) showed that both
heterogeneous display sequences (no successively repeating items)
and longer delays enhance exposure– affect relationships.

The 116 viewers in attendance were asked to indicate on a re-
sponse sheet which image of each pair they liked best. They also
filled out a brief questionnaire, as in Study 1. Mean self-reported
attendance over the 21 exposure sessions was 18 days (median =
19, SD = 3.2); 70% said that they visited the Johnson Art Museum



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 335

on the Cornell University campus at least once a year, 47% said
they visited at least one other art museum each year, 14% said they
had visited the Musée d’Orsay, and 22% said they had taken at least
one art history course. The students also reported spending a mean
of 12.5 h per week on the Internet (median = 10, SD = 9.6).

For the critical comparisons, the preferences of viewers in this
study were compared with those of Study 1, a similar group but
without the systematic classroom exposure. Thus, major analyses
here are based on a between-group design, looking at differences in
preferences across the 51 image pairs.

Results and Discussion
More frequently published images were no longer pre-

ferred, accruing only 48% of all the judgments. This was
reliably lower than the 57% preference in Study 1 for
these 51 colored-image pairs [F(1,50) = 26.8, p < .001].
Indeed, in 41 of 50 image pairs (with one tie), the more
frequent image received a smaller proportion of prefer-
ence judgments (z = 4.2, p < .0001). Preference rates for
the more frequent image are found in the last column of
Table 1. No other results proved statistically reliable:
Image pairs that were published more frequently (sum >
50, n = 26) changed, on average, as much as those that
occurred less often (sum < 51, n = 25; 9.0% vs. 9.2%, re-
spectively), and image pairs whose frequency ratios were
relatively large (>2.5, n = 26) changed as much as those
whose ratios were nearer unity (< 2.5, n = 25; 10.1% vs.
8.1%, respectively).

If observers were able to judge quality alone in the
image pairs, their judgments should not have been con-
taminated by appearance differences in the classroom.
To be sure, quality could still play a role, but such an ac-
count must then rely on two processes—mere exposure
and quality assessment (however that might be done).
My proposal is that these are one-process results and
done on the basis of mere exposure inside and outside
the classroom.

CONCLUSION: M ERE EXPOSUR E IS A
MEDIATOR OF CANON MAINTENA NCE

How might mere exposure affect an artistic canon, its
reception, and its maintenance? All of us, as members of
a culture, absorb what is around us. As visual beings, we
digest images voraciously, even without noticing. A very
small proportion of these images are from the Impres-
sionist corpus and canon. Nonetheless, we respond to
their occurrences in our future interactions with Impres-
sionism. We like the ones we have seen before and, par-
ticularly, those we may have seen many times.15

We f ind these images everywhere. Impressionist
paintings are not only in galleries, but also in books and
on textbook covers, calendars, posters, coasters, tee
shirts, and towels, and one can find them readily on the
Internet. Mere exposure dictates that every occurrence
can matter, particularly when an image is otherwise rare.
Museum curators ought to note the full implications of
this f inding. Museums already do a reasonably good job
at promoting their collections, but placing images every-
where and without cost to the public will go a long way

toward ratifying the importance of their collections as re-
ceived by the broader public. But the competition is stiff;
everyone seems to be doing it. Currently, the correlation
of what’s in the literature and what’s on the Internet, as
shown in Table 2, is not high. If this difference is main-
tained, the canons of the future may change in directions
independent of goals and interests of art professionals.

Thus, I claim that artistic canons are promoted and
maintained, in part, by a diffuse but continual broadcast
of their images to the public by museums, authors, and
publishers. The repeated presentation of images to an au-
dience without its necessarily focused awareness or re-
membrance makes mere exposure a prime vehicle for
canon maintenance. Tacitly and incrementally over time,
this broadcast teaches the public to like the images, to
prefer them, eventually to recognize them as part of the
canon, and to want to see them again. In turn, it seems
likely that this implicit education also reinforces the
choices made by professionals in what they present to
that public. The public’s appreciation rewards museums,
scholars, and the publishing industry by demonstrating
an interested and responsive audience.

And so it goes, with mere exposure cyclically rein-
forcing the canons through generations of authors and
curators, on the one hand, and of museumgoers and book
buyers, on the other. Although it may be tacit, I do not
claim that this is necessarily a subversive trend or one to
be denigrated. I claim it is part of the same force that
binds a culture. It is part of our human nature, built on an
evolutionary substrate that makes very good sense. It
helps ensure steadiness in culture more generally and
relative constancy in artistic canons more particularly.

REFERENCE S

Bal, M., &  Bryson, N. (1991). Semiotics and art history. Art Bulletin,
73, 174-208.

Bazin, G. (1958). Impressionist paintings in the Louvre (S. Cunliffe-
Owen, Trans.). London: Thames & Hudson.

Bérhaut, M. (1994). Gustave Caillebotte: Catalogue raisonné des
peintures et pastels. Paris: La Bibliothèque des Arts.

Berlyne, D. E. (1970). Novelty, complexity, and hedonic value. Per-
ception & Psychophysics, 8, 279-286.

Bernac, J. (1966). The Caillebotte bequest to the Luxembourg. Reprinted
and translated in D. Sutton, Gustave Caillebotte, 1848–1894: A loan
exhibition in aid of the Hertford British Hospital in Paris, 15th
June–16th July, 1966. London: Wildenstein. (Original published in
1895)

Bonnano, G. A., &  Stilling, N. A. (1986). Preference, familiarity,
and recognition after repeated brief exposure to random geometric
shapes. American Journal of Psychology, 99, 403-415.

Bornstein, R. (1989). Exposure and affect: Overview and meta-analysis
of research, 1968–1987. Psychological Bulletin, 106, 265-289.

Brickman, P., Redfield, J., Harrison, A. A., &  Crandall, R. (1972).
Drive and predisposition as factors in the attitudinal effects of mere
exposure. Journal of Experimental Social Psychology, 8, 31-44.

Chastel, A. (1971). Paris. New York: Holt, Rinehart & Winston.
Cheetham, M. A. (2002). Kant, art, and art history. Cambridge: Cam-

bridge University Press.
Clark, K. (1974). Another part of the wood. London: John Murray.
Cutting, J., Wang, R. F., Flückiger, M., &  Baumberger, B. (1999).

Human heading judgments and object-based motion information. Vi-
sion Research, 39, 1079-1105.

Daulte, F. (1959). Alfred Sisley: Catalogue raisonné de l’oeuvre peint.
Paris: Durand-Ruel.

http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0031-5117^28^298L.279[aid=348035]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0002-9556^28^2999L.403[aid=293644]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0033-2909^28^29106L.265[aid=188552]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0042-6989^28^2939L.1079[aid=2385745]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0031-5117^28^298L.279[aid=348035]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0042-6989^28^2939L.1079[aid=2385745]


336 CUTTING

Daulte, F. (1971). Auguste Renoir: Catalogue raisonné de l’oeuvre
peint: Vol 1. Figures 1860-1890. Lausanne: Durand-Ruel.

Distel, A. (1990). Impressionism: The first collectors. (B. Perroud-
Benson, Trans.). New York: Harry N. Abrams.

Distel, A. (Director) (1994). Gustave Caillebotte: 1848–1894 [Art
exhibition]. Paris: Réunion des Musées Nationaux.

Elliott, R., &  Dolan, R. J. (1998). Neural response during preference
and memory judgments for subliminally presented stimuli: A func-
tional neuroimaging study. Journal of Neuroscience, 18, 4697-4704.

Ferguson, R. (1999). Can we still use the canon? Art Journal, 58, 4.
Fezzi, E. (1972). L’opera completa di Renoir [The complete works of

Renoir]. Milan: Rizzoli Editore.
Fox, B. (1983, October 20). Video thrills the Hollywood star. New Sci-

entist, 100, 203-208.
Gibson, J. J. (1947). Motion picture testing and research (Army Air

Force Aviation Psychology Research Rep. 7). Washington, DC: U.S.
Government Printing Office.

Goodman, N. (1972). Problems and projects. Indianapolis, IN: Bobbs-
Merrill.

Guillory, J. (1990). Canon. In F. Lentrichhia & T. McLaughlin (Eds.),
Critical terms for literary study (pp. 233-262). Chicago: University
of Chicago Press.

Guillory, J. (1993). Cultural capital: The problem with literary canon
formation. Chicago: University of Chicago Press.

Hallberg, R. v on (Ed.) (1984). Canons. Chicago: University of
Chicago Press.

Harrison, A. A. (1977). Mere exposure. In L. Berkowitz (Ed.), Ad-
vances in experimental social psychology (Vol. 10, pp. 39-83). New
York: Academic Press.

Harrison, A. A., &  Zajonc, R. B. (1970). Effects of frequency and du-
ration of exposure on response competition and affect ratings. Jour-
nal of Psychology, 75, 163-169.

Hayward Gallery (1985). Renoir. Catalog of the exhibit at the Hay-
ward Gallery, London, 30 June–21 August 1985, Galérie Nationale
de Grand Palais, Paris, 14 May–2 September 1985, Museum of Fine
Arts, Boston, 9 October–5 January, 1986.

Jamot, P., &  Wildenstein, G. (1932). Edouard Manet (Vols. 1 and 2).
Paris: Les Beaux-arts.

Joseph, J. E., &  Proffitt, D. R. (1996). Semantic versus perceptual in-
fluences of color on object recognition. Journal of Experimental Psy-
chology: Learning, Memory, & Cognition, 22, 407-429.

Kant, I. (1952). Critique of aesthetic judgment. (J. Meredith, Trans.)
Oxford: Oxford University Press, Clarendon Press. (Original work
published 1794)

Klinger, M. R., &  Greenwald, A. (1994). Preferences need no infer-
ences? The cognitive basis of unconscious mere exposure effects. In
P. M. Niedenthal & S. Kitayama (Eds.), The heart’s eye: Emotional
influences in perception and attention (pp. 67-85). San Diego: Aca-
demic Press.

Kozlowski, L. T., &  Cutting, J. E. (1977). Recognizing the sex of a
walker from a dynamic point-light display. Perception & Psycho-
physics, 21, 575-580.

Kunst-Wilson, W. R., &  Zajonc, R. B. (1980). Affective discrimina-
tion of stimuli that cannot be recognized. Science, 207, 557-558.

Lemoisne, P. A. (1946). Degas et son oeuvre (Vols. 1– 4). Paris: Arts et
Métiers Graphiques.

Marcus, M. G., &  Hakmiller, K. L. (1975). Effects of frequency, du-
ration of study trials and total duration of exposure on affective judg-
ments. Psychological Reports, 37, 195-200.

Marrinan, M. (2002). Caillebotte as a professional painter: From Stu-
dio to the public eye. In N. Broude (Ed.), Gustave Caillebotte and the
fashioning of identity in Impressionist Paris (pp. 21-65). New
Brunswick, NJ: Rutgers University Press.

Mead, K. H. (Director) (1974). The Impressionists and the salon
(1874–1886) [Art exhibition]. Riverside, CA: Regents of the Uni-
versity of California.

Monahan, J. L., Murphy, S. T., &  Zajonc, R. B. (2000). Subliminal
mere exposure: Specific, general, and diffuse effects. Psychological
Science, 11, 462-466.

Moreland, R. L., &  Zajonc, R. B. (1977). Is stimulus recognition a
necessary condition for the occurrence of exposure effects? Journal
of Social & Personality Psychology, 35, 191-199.

Moxey, K. (1994). The practice of theory: Poststructuralism, cultural
politics, and art history. Ithaca, NY: Cornell University Press.

Musée d’Orsay (1990). Catalogue sommaire illustré des peintures
(Vols. 1 and 2). Paris: Réunion des Musées Nationaux.

Nord, P. (2000). Impressionists and politics. London: Routledge.
Oliva, A., &  Schyns, P. G. (2000). Diagnostic colors mediate scene

recognition. Cognitive Psychology, 41, 176-210.
Pirsig, R. (1974). Zen and the art of motorcycle maintenance. New

York: Morrow.
Pissarro, L.-R., &  Venturi, L. (1939). Camille Pissarro – son art –

son oeuvre (Vols. 1 and 2). Paris: Paul Rosenberg.
Rewald, J. (1946). The history of Impressionism. New York: Museum

of Modern Art.
Rewald, J. (1996). The paintings of Paul Cézanne: A catalogue

raisonné (Vols. 1 and 2). New York: Harry N. Abrams.
Roediger, H. L., III (1990). Implicit memory: Retention without re-

membering. American Psychologist, 45, 231-246.
Rosch, E. H. (1973). Natural categories. Cognitive Psychology, 4, 328-

350.
Rosenberg, J. (1967). On quality in art: Criteria of excellence, past

and present. Princeton, NJ: Princeton University Press.
Saltzman, C. (1998). Portrait of Dr. Gachet. New York: Penguin.
Sassoon, D. (2001). Becoming Mona Lisa. New York: Harcourt.
Schacter, D. L. (1987). Implicit memory: History and current status.

Journal of Experimental Psychology: Learning, Memory, & Cogni-
tion, 13, 501-518.

Seamon, J. G. , Brody, N., &  Kauff, D. M. (1983). Affective discrim-
ination of stimuli that are not recognized: II. Effect of delay between
study and test. Bulletin of the Psychonomic Society, 21, 187-189.

Seamon, J. G., &  Delgado, M. R. (1999). Recognition memory and af-
fective preference for depth-rotated solid objects: Part-based struc-
tural descriptions may underlie the mere exposure effect. Visual Cog-
nition, 6, 145-164.

Seamon, J. G., Williams, P. C., Crowley, M. J., Kim, I. J., Langer, S. A.,
Orne, P. J., &  Wishengrad, D. L. (1995). The mere exposure effect
is based on implicit memory: Effects of stimulus type, encoding con-
ditions, and number of exposures on recognition and affects judg-
ments. Journal of Experimental Psychology: Learning, Memory, &
Cognition, 21, 711-721.

Shikes, R. E., &  Harper, P. (1980). Pissarro: His life and work. New
York: Horizon.

Smith, E. R. (1998). Mental representation and memory. In D. T.
Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psy-
chology (4th ed., pp. 391-445). Boston: McGraw-Hill.

Squire, L. R. (1992). Memory and the hippocampus: A synthesis of
f indings with rats, monkeys, and humans. Psychological Review, 99,
195-231.

Vanbeseleare, N. (1983). Mere exposure: A search for an explanation.
In W. Doise & S. Moscovici (Eds.), Current issues in European social
psychology (Vol. 1, pp. 239-278). London: Cambridge University
Press.

Varnedoe, K. (1987). Gustave Caillebotte. New Haven, CT: Yale Uni-
versity Press.

Venturi, L. (1936). Cézanne, son art, son oeuvre (Vols. 1 and 2). Paris:
Paul Rosenberg.

Wadley, N. (Ed.) (1987). Renoir: A retrospective. New York: Macmil-
lan.

White, B. E. (1984). Renoir: His life, art, and letters. New York:
Harry N. Abrams.

Wildenstein, D. (1974–1985). Claude Monet: Biographie et catalogue
raisonné (Vols. 1–5). Paris: La Bibliothèque des Arts.

Winkielman, P., Zajonc, R. B., &  Schwarz, N. (1997). Subliminal
affective priming resists attributional interventions. Cognition &
Emotion, 11, 433-465.

Woodford, S. (1983). Looking at pictures. Cambridge: Cambridge
University Press.

Wurm, L. H., Legge, G. E., Isenberg, L. M., &  Luebker, A. (1993).
Color improves object recognition in normal and low vision. Journal
of Experimental Psychology: Human Perception & Performance, 19,
899-911.

Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of
Personality & Social Psychology, 9, 1-27.

http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0270-6474^28^2918L.4697[aid=211328]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0022-3980^28^2975L.163[aid=665028]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0278-7393^28^2922L.407[aid=847164]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0031-5117^28^2921L.575[aid=296578]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0036-8075^28^29207L.557[aid=213661]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0956-7976^28^2911L.462[aid=2203542]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0010-0285^28^2941L.176[aid=5185419]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0278-7393^28^2913L.501[aid=57313]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0278-7393^28^2921L.711[aid=304092]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0033-295X^28^2999L.195[aid=215396]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0269-9931^28^2911L.433[aid=259293]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0096-1523^28^2919L.899[aid=311706]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0022-3514^28^299L.1[aid=73201]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0022-3980^28^2975L.163[aid=665028]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0278-7393^28^2922L.407[aid=847164]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0031-5117^28^2921L.575[aid=296578]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0956-7976^28^2911L.462[aid=2203542]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0278-7393^28^2913L.501[aid=57313]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0278-7393^28^2921L.711[aid=304092]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0033-295X^28^2999L.195[aid=215396]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0269-9931^28^2911L.433[aid=259293]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0096-1523^28^2919L.899[aid=311706]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0022-3514^28^299L.1[aid=73201]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0096-1523^28^2919L.899[aid=311706]


MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 337

Zajonc, R. B. (1970, February). Brainwash: Familiarity breeds com-
fort. Psychology Today, 4, 33-35, 60-62.

Zajonc, R. B. (1980). Feeling and thinking: Preferences need no infer-
ences. American Psychologist, 35, 151-175.

Zajonc, R. B. (2001). Mere exposure: A gateway to the subliminal.
Current Directions in Psychological Science, 10, 224-228.

Zajonc, R. B., Crandall, R., Kail, R. V., &  Swap, W. (1974). Effects
of extreme exposure frequencies on different affective ratings of
stimuli. Perceptual & Motor Skills, 38, 667-678.

Zajonc, R. B., Shaver, P., Tavris, C., &  van Kreveld, D. (1972). Ex-
posure, satiation, and stimulus discriminability. Journal of Personal-
ity & Social Psychology, 21, 270-280.

NOTES

1. Caillebotte left no records concerning the accumulation of his col-
lection. He seems to have acquired 57 of the 66 artworks used here in
the 7 years up to and including 1882, and probably only 9 in the 12 years
thereafter. Among the latter, 3 were gifts (1 from Monet and 2 from
Renoir), and 6 were purchased (3 Sisleys and 3 Manets). The Manets
were purchased in 1884 at the studio sale to support his widow.
Nonetheless, in 1882, Caillebotte retired from Paris and removed him-
self from new developments in art. In that year, Georges Seurat had yet
to paint his f irst divisionist ( pointilliste) painting; Paul Gauguin was
still a stockbroker; Vincent Van Gogh, although a painter for 2 years
(having given up teaching and mission work), was still 4 years away
from moving to Paris; and Henri Toulouse-Lautrec was a teenage art
student.

2. As a part of a larger project, I inspected 30 books on Impression-
ism published over the 20th century and found that these seven artists
had works appearing in at least 29 of them. The next most frequent
artists were Berthe Morisot (25), Mary Cassatt (22), Georges Seurat
(21), Frédéric Bazille (19), Paul Gauguin (18), Henri Toulouse-Lautrec
(16), Gustave Caillebotte (14), and Vincent Van Gogh (13).

3. Much has been made of this quote, and it takes many different
forms. Shikes and Harper (1980), for example, cited it as including
Manet rather than Monet.

4. Most recent research on mere exposure methodologically allies it-
self with subliminal perception. That is, stimuli are presented briefly
and then masked so that observers cannot report what they have seen but
can be shown to have processed it through results of priming or prefer-
ence (e.g., Kunst-Wilson & Zajonc, 1980; Monahan, Murphy, & Za-
jonc, 2000; Moreland & Zajonc, 1977; Seamon, Brody, & Kauff, 1983).
These results are interesting and important, but from the perspective of
this article, subliminal perception is a laboratory phenomenon used to
mimic the processes in real life perception—inattention and forgetting
over the long haul. Otherwise, it can have no general interest. Thus, in
this context, I am less interested in alternative theories that may explain
mere subliminal exposure (e.g., Bonnano & Stilling, 1986; Klinger &
Greenwald, 1994; Smith, 1998; Winkielman, Zajonc, & Schwarz, 1997;
Zajonc, 2001) than in the more general phenomenon itself.

5. Once found, the images not appearing in Distel (1994) were
checked for their provenance (ownership history), to be sure they had
been Caillebotte’s. In addition, despite 20 months search in libraries and
on the Internet, I was unable to find five of these works. Mention of
these appeared without images in Distel (1994) or in images in the cat-
alogue raisonné of each painter. They include one Monet, three Pissar-
ros, and one Sisley—all listed at the end of the Appendix.

6. A catalogue raisonné assembles and illustrates, typically chrono-
logically, the entire corpus of an artist’s work, often with additional ma-
terial, such as correspondence. Inevitably, it has inconsistencies and er-
rors revealed by later research. In this context, dating works is a

particular problem with Cézanne. Thus, Rewald’s (1996) compendium
was a necessary revision of Venturi (1936). Similarly, there are dating
problems with Degas. For example, Distel (1994) noted that Image 14c
(Danseuse espagnol, “Spanish dancer”) was purchased by Caillebotte
before May 1879, yet Lemoisne (1946) listed it as having been painted
in 1880. Some catalogues raisonnés, such as that of Manet (Jamot &
Wildenstein, 1932), occasionally list images thematically (e.g., Image
Pair 17), rather than chronologically.

7. Fox (1983), from Kodak data and in the context of predictions
about high-definition TV, reported that 35-mm film has a digital reso-
lution of about 2,500 3 1,800 pixels, or 4.5 MB. The f iles used in
Study 1 were about 65% of that resolution along one dimension.

8. Renoir’s Bal du Moulin de la Galette is almost surely the most re-
produced of all Impressionist paintings. For example, in one Paris
guidebook (Chastel, 1971), I found only two paintings representing
what might be found in the city—Mona Lisa and Bal du Moulin de la
Galette. In turn, the Mona Lisa (also La Joconde) is often regarded as
the world’s most famous painting (Sassoon, 2001) or, if one counts
Michelangelo’s ceiling of the Sistine Chapel, the second most famous
(September 24, 1995, London Sunday Times, The Culture, Section 10,
p. 29). However, the latter assessment was made not long after the Sis-
tine ceiling was cleaned, and with understandable effects of its accom-
panied media coverage.

9. Inspecting each entry is necessary: “Monet” is also one of the mid-
dle names of both Lamarck and Lavoisier, as well as the Russian word
for money; Pierre-Auguste Renoir must be distinguished from his
younger son the filmmaker, Jean Renoir (except in the latter’s book
about the former); Camille Pissarro must be separated from his sons,
particularly Lucien who was also an artist (but who also wrote of his fa-
ther); and there are several authors named Sisley, as well as an Italian
fashion line.

10. To exclude Postimpressionism is to exclude some descriptions of
the work of Cézanne and, to a lesser degree, Degas. However, these
works were composed after Caillebotte stopped collecting in 1882. Pis-
sarro also had a Neoimpressionist ( pointilliste) period, but this too oc-
curred after Caillebotte actively collected.

11. The Orsay comparison images appeared a mean of 39.5 times and
the non-Orsay comparisons 25.1 times. Note, however, than many more
of the comparison images (27) were from other museums than the
Caillebotte images (10).

12. There was no reliable effect of color versus black and white in the
counts of images in the Cornell books that effected either recognition
or preference, but the effect reported above concerns the presence of
color versus black and white in the stimuli as presented to the viewers.

13. The six additional Caillebotte paintings were: Déjeuner (“Lun-
cheon,” 1876, private collection), Portraits à la campagne (“Country
portraits,” 1876, Musée Baron Gérard, Bayeux, France), Le pont de
l’Europe (“The Europe bridge,” Paris, 1876, Musée du Petit Palais,
Genève, Switzerland), Peintres en bâtiment (“House painters,” 1877,
private collection), Rue de Paris; temps de pluie (“Paris street, rainy
weather,” 1877, Art Institute of Chicago), and Boulevard vu d’en haut
(“Boulevard viewed from above,” 1880, private collection). Reference
citations for these images are Bérhaut (1994) #37, 40, 49, 53, 57, and
154, respectively.

14. This figure was determined through analysis of the catalogues
raisonnés of the seven “major” Impressionists and the holdings catalog
of the Musée d’Orsay (Musée d’Orsay, 1990).

15. It is often noted that the effects of mere exposure may even de-
cline with experience repeated very many times (e.g., Bornstein, 1989;
Zajonc, Crandall, Kail, & Swap, 1974). This may well be true in labo-
ratory situations or in the real world with massed practice, but it is not
clearly the case with exposures distributed over years, even decades.

(Continued on next page)

http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0003-066X^28^2935L.151[aid=23836]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0022-3514^28^2921L.270[aid=2203544]
http://gessler.ingentaselect.com/nw=1/rpsv/cgi-bin/linker?ext=a&reqidx=/0022-3514^28^2921L.270[aid=2203544]


338 CUTTING

APPENDIX
The Caillebotte and Comparison Images,

Their Dates (If Known), Their Locations (If Known), and Their Reference Citations

From the Caillebotte Legacy and Collection Comparison Images

Caillebotte

1. Les Raboteurs de parquet Les Raboteurs de parquet (petit version)
“The floor strippers” “The floor strippers (small version)”
1875, Musée d’Orsay (RF 2718) 1876, private collection, Paris
Bérhaut (1994) #34; Figure 1A Bérhaut (1994) #35; Figure 1B

2. Vue de toits, effet de neige Rue Halévy, vue d’un sixième étage
“Snow-covered roofs in Paris” “Rue Halévy, view from the 7th floor”
1878, Musée d’Orsay (RF 876) 1878, private collection
Bérhaut (1994) #96 Bérhaut (1994) #100

Cézanne

3. Au bord de l’étang or Scène champêtre Le Déjeuner sur l’herbe
“Couples relaxing by a pond” or “The pond” “The picnic” or “Luncheon on the grass”
1876–1877, Museum of Fine Arts, Boston 1876–1877, Musée de l’Orangerie
Venturi (1936) #232, Rewald (1996) #244 Venturi (1936) #238, Rewald (1996) #287

4. Baigneurs au repos, III Les cinq baigneurs
“Bathers at rest” “The five bathers”
1876–1877, Barnes Foundation 1875–1877, Musée d’Orsay (RF 1982-42)
Venturi (1936) # 276, Rewald (1996) #261 Venturi (1936) #268, Rewald (1996) #254

5. Fleurs dans un vase rococo Bouquet au petit Delft
or Vase de fleurs “Flowers in a Delft vase”

“Flowers in a rococo vase” 1873, Musée d’Orsay (RF 1951-33)
1876, National Gallery of Art, DC Venturi (1936) #183, Rewald (1996) #227
Venturi (1936) #222, Rewald (1996) #265

6. Cour de ferme à Auvers La Maison du pendu, Auvers-sur-Oise
“Farmyard at Auvers” “House of the hanged man”
1879–1880, Musée d’Orsay (RF 2760) 1873, Musée d’Orsay (RF 1970)
Venturi (1936) #326, Venturi (1936) #133, 

Rewald (1996) #389; Figure 1C Rewald (1996) #202; Figure 1D

7. Le Golfe de Marseille, vu de l’Estaque* La Baie de l’Estaque vue de l’est
“Estaque” “The bay of Estaque looking east”
1878–1880, Musée d’Orsay (RF 2761) 1878–1879, Memorial Art Gallery,
Venturi (1936) #428, Rewald (1996) #390 University of Rochester, NY

Venturi (1936) #408, Rewald (1996) #394

Degas

8. Femmes à la terrase d’un café, le soir Dans un café or L’Absinthe
or Un Café, boulevard Hausmann “The absinthe drinker”

“Women outside a cafe in the evening” 1876, Musée d’Orsay (RF 1984)
1877, Musée d’Orsay (RF 12257) Lemoisne (1946) #393; Figure 1F
Lemoisne (1946) #419; Figure 1E

9. Choristes or Les Figurants Danseuse au bouquet saluant
“The chorus” or “The supernumeraries” “Dancer with bouquet, bowing”
1876–1877, Musée d’Orsay (RF 12259) ~1877, Musée d’Orsay (RF 4039)
Lemoisne (1946) #420 Lemoisne (1946) #474

10. Femme sortant du bain Après le bain
“Woman leaving the bath” “Nude woman drying her feet”
1876–1877, Musée d’Orsay (RF 12255) 1885–1886, Musée d’Orsay (RF 4045)
Lemoisne (1946) #422 Lemoisne (1946) #874

11. La Leçon de danse Danseuses à la barre
“Portrait of a dancer at her lesson” “Dancers practicing at the bar”
~1879,Metropolitan Museum of Art, New York 1876–1877, Metropolitan Museum of Art, New York
Lemoisne (1946) #450 Lemoisne (1946) #408

12. L’Étoile (Danseuse sur la scène) Danseuse tenant un bouquet à la main
“Ballet (The star)” “Arabesque”
1876–1878, Musée d’Orsay (RF 12258) ~1877, Musée d’Orsay (RF 4039)
Lemoisne (1946) #491 Lemoisne (1946) #418



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 339

APPENDIX (Continued)

13. Femme nue, accroupie, de dos Le Tub
“Squatting woman seen from the back” “Woman bathing in a shallow tub”
~1879, Musée d’Orsay (RF 12254) 1886, Musée d’Orsay (RF 4046)
Lemoisne (1946) #547 Lemoisne (1946) #872

14. Danseuse espagnol Le Café-concert
or Chanteuse de café-concert “At the Café des Ambassadeurs”

“Study for the bust of a ballet dancer” 1885, Musée d’Orsay (RF 4041)
1880, Musée d’Orsay (RF 12260) Lemoisne (1946) #814
Lemoisne (1946) #608

15. Danseuse nouant son brodequin L’Attente
“Seated dancer kneading her ankle” “Dancer and woman with a black umbrella, waiting”
1881–1883, Musée d’Orsay (RF 12271) ~1882, Getty Museum, Malibu, CA
Lemoisne (1946) #658 Lemoisne (1946) #698

Manet

16. Angelina Georges Clemençeau
“Angelina” “Georges Clemençeau”
1865, Musée d’Orsay (RF 3664) 1879–1880, Musée d’Orsay (RF 2641)
Jamot and Wildenstein (1932) #118 Jamot and Wildenstein (1932) #372

17. Les Courses† Courses à Longchamps†
“The races” “At the races”
1865, location unknown 1875, National Gallery, Washington
Jamot and Wildenstein (1932) #204 Jamot and Wildenstein (1932) #205

18. Le Balcon Le Déjeuner à l’atelier
“The balcony” “Luncheon in the studio”
1868–1869, Musée d’Orsay (RF 2772) 1868, Neue Pinakothek, Munich
Jamot and Wildenstein (1932) #150; Figure 1G Jamot and Wildenstein (1932) #149; Figure 1H

19. Croquet à Boulogne Plage avec personnages
or La Partie de croquet “On the beach at Boulogne”

“The croquet game” 1869, Virginia Museum of Art, Richmond
1868–1871, Woll family, private collection Jamot and Wildenstein (1932) #166
Jamot and Wildenstein (1932) #197

Monet

20. Le Mont Riboudet à Rouen au printemps Vue de plaine à Argenteuil
“Mt. Riboudet at Rouen in spring” “View of the plain at Argenteuil”
1872, private collection, U.S. 1872, Musée d’Orsay (MNR 855)
Wildenstein (1974–1985) #216 Wildenstein (1974–1985) #220

21. Régates à Argenteuil Voilier au Petit-Gennevilliers
“Regattas at Argenteuil” “Sailboat at Petit-Gennevilliers”
~1872, Musée d’Orsay (RF 2778) 1874, Lucille Ellis Simon Collection, USA
Wildenstein (1974–1985) #233 Wildenstein (1974–1985) #336

22. Le Déjeuner Femmes au jardin
“Luncheon in the garden” “Women in the garden”
1873, Musée d’Orsay (RF 2774) 1866, Musée d’Orsay (RF 2773)
Wildenstein (1974–1985) #285 Wildenstein (1974–1985) #67

23. Un Coin d’appartement Coin d’atelier
“Apartment interior” “Corner of a studio”
1875, Musée d’Orsay (RF 2776) 1861, Musée d’Orsay (MNR 136)
Wildenstein (1974–1985) #365 Wildenstein (1974–1985) #6

24. Les Tuileries (esquisse) La Débâcle
“The Tuileries” “Ice break up”
1875, Musée d’Orsay (RF 2705) 1880, Museum of Art, University of Michigan
Wildenstein (1974–1985) #403 Wildenstein (1974–1985) #565

25. La Plaine de Gennevilliers Vétheuil, vu de Lavacourt
or La Plaine d’Argenteuil “Vétheuil, view from Lavacourt”

“The plain near Gennevilliers” 1879, Musée d’Orsay (RF 1998)
1877, Fogg Art Museum, Harvard Wildenstein (1974–1985) #528
Wildenstein (1974–1985) #437



340 CUTTING

APPENDIX (Continued)

26. La Gare Saint-Lazare La Gare Saint-Lazare, l‘arrivée d’un train
“St. Lazare train station, Paris” “Gare Saint-Lazare, the arrival of a train”
1877, Musée d’Orsay (RF 2775) 1877, Fogg Art Museum, Harvard
Wildenstein (1974–1985) #438; Figure 2A Wildenstein (1974–1985) #439; Figure 2B

27. La Gare Saint-Lazare: sous le pont Le Pont de l’Europe, gare Saint-Lazare
de l’Europe “The Pont de l’Europe”

“Gare Saint-Lazare: exterior view” 1877, Musée Marmottan, Paris
or “Pont de Rome” Wildenstein (1974–1985) #442

1877, private collection
Wildenstein (1974–1985) #447

28. La Gare Saint-Lazare à l’extérieur, le signal Les Voies à la sortie de la gare Saint-Lazare
“The Gare Saint-Lazare, the signal” “The tracks in front of the Gare Saint-Lazare”
1877, Landesmuseum, Hannover 1877, private collection, Japan
Wildenstein (1974–1985) #448 Wildenstein (1974–1985) #445

29. Pommiers, Vétheuil Paysage, Vétheuil
“Apple trees, Vétheuil” “Landscape, Vétheuil”
1878, Wesley M. Dixon Collection, U.S. 1879, Musée d’Orsay (RF 2623)
Wildenstein (1974–1985) #490 Wildenstein (1974–1985) #1986

30. L’Église de Vétheuil, neige Vétheuil l’hiver
“The church at Vétheuil, snow” “Vétheuil, winter”
1878–1879, Musée d’Orsay (RF 3755) 1879, Frick Gallery, New York
Wildenstein (1974–1985) #506 Wildenstein (1974–1985) #507

31. Pruniers en fleurs or Poiriers en fleurs Lilas, temps gris
“Blossoming plum trees” “Resting under the lilacs”

or “Blossoming pear trees” or “Lilacs, gray weather”
1879, location unknown 1972–1973, Musée d’Orsay (RF 1680)
Wildenstein (1974–1985) #519 Wildenstein (1974–1985) #203

32. Le Givre Vétheuil dans le brouillard
“Hoarfrost (Vétheuil)” “Vétheuil in the mist”
1880, Musée d’Orsay (RF 2706) 1879, Musée Marmottan, Paris
Wildenstein (1974–1985) #555 Wildenstein (1974–1985) #518

33. Chrysanthèmes rouges Chrysanthèmes
“Red chrysanthemums” “Chrysanthemums”
1880, location unknown 1878, Musée d’Orsay (RF 1951-36)
Wildenstein (1974–1985) #635 Wildenstein (1974–1985) #492

34. La Seine entre Vétheuil et La Roche Guyon Les Glaçons or Débâcle sur la Seine
“The Seine between Vétheuil and “The ice-floes”

La Roche Guyon” 1880, Musée d’Orsay (RF 1965-10)
1881, private collection, France Wildenstein (1974–1985) #567
Wildenstein (1974–1985) #674

35. Les rochers de Belle-Île Tempêtes, Côtes de Belle-Île
“The rocks of Belle-Isle” “Storm at Belle-Isle”
1886, Musée d’Orsay (RF 2777) 1886, Musée d’Orsay (RF 3163)
Wildenstein (1974–1985) #1100 Wildenstein (1974–1985) #1116

Pissarro

36. Louveciennes Entrée du village de Voisins
“Louveciennes” “Entrance to the village of Voisins”
1871, private collection, Paris 1872, Musée d’Orsay (RF 2456)
Pissarro and Venturi (1939) #123 Pissarro and Venturi (1939) #141

37. Le Lavoir, Pontoise Pontoise, Côte de l’Oise
“The washhouse at Bougival” “Pontoise, banks of the Oise”
1872, Musée d’Orsay (RF 2732) 1872, Getty Museum, Malibu
Pissarro and Venturi (1939) #175 Pissarro and Venturi (1939) #182

38. Paysage avec rochers, Montfoucault Côteau de l’Hermitage, Pontoise
“Landscape with rocks, Montfoucault” “Hill at the Hermitage, Pontoise”
1874, private collection, Paris 1873, Musée d’Orsay (RF 1983-8)
Pissarro and Venturi (1939) #282 Pissarro and Venturi (1939) #209



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 341

APPENDIX (Continued)

39. Le Labourer Paysage à Chaponval
“The plowman” “Chaponval landscape”
1876, private collection, Paris 1880, Musée d’Orsay (RF 1937–51)
Pissarro and Venturi (1939) #340 Pissarro and Venturi (1939) #509

40. Jardin en fleurs, Pontoise L’Hermitage à Pontoise
“Garden in bloom, Pontoise” “The Hermitage, Pontoise”
1876, private collection, Paris 1872, private collection, Japan
Pissarro and Venturi (1939) #350 Pissarro and Venturi (1939) #529

41. La Moisson à Montfoucault La Meule, Pontoise
“Harvest at Montfoucault” “The haystack, Pontoise”
1876, Musée d’Orsay (RF 3756) 1873, private collection, Paris
Pissarro and Venturi (1939) #364 Pissarro and Venturi (1939) #223

42. Sous bois avec une femme assise Le Petit pont, Pontoise
“In the woods” “Little bridge, Pontoise”
1876, private collection, Paris 1875, Kunsthalle, Mannheim
Pissarro and Venturi (1939) #371 Pissarro and Venturi (1939) #300

43. Les Toits rouges, coin du village, effet d’hiver La Côte des boeufs à Pontoise
“The red roofs” “La Côte des Boeufs, the Hermitage”
1877, Musée d’Orsay (RF 2735) 1877, National Gallery, London
Pissarro and Venturi (1939) #384; Figure 2C Pissarro and Venturi (1939) #380; Figure 2D

44. Printemps à Pontoises, potager et Verger en fleurs, Louveciennes
arbres en fleurs “Orchard in blossom, Louveciennes”

“Orchard with flowering fruit trees, Pontoise” 1872, National Gallery, Washington
1877, Musée d’Orsay (RF 2733) Pissarro and Venturi (1939) #153
Pissarro and Venturi (1939) #387

45. Les Seigles, Pontoise, côte des Mathurins Gardeuse de vache sur la route du Chou, Pontoise
or Les Orges “A cowherd at Pontoise”

“Rye fields, Pontoise, seen from the Mathurins” 1874, Metropolitan Museum of Art, New York
1877, private collection, Japan Pissarro and Venturi (1939) #260
Pissarro and Venturi (1939) #406

46. Chemin sous-bois, en été Au repos sous-bois, Pontoise
“Path through the woods” “Resting in the woods, Pontoise”
1877, Musée d’Orsay (RF 2731) 1878, Kunsthalle, Hamburg
Pissarro and Venturi (1939) #416 Pissarro and Venturi (1939) #466

47. Lisière de bois or Clairière Le Châlet, la maison rose
“Edge of the woods” “The chalet, the pink house”
1878, private collection, Paris 1870, Musée d’Orsay (RF 1937-58)
Pissarro and Venturi (1939) #455 Pissarro and Venturi (1939) #82

48. Chemin montant à travers champs. Femme dans un clos
Côte des Grouettes, Pontoise “Woman in an enclosure, spring

“Path across the fields” sunshine in an Eragny field”
1879, Musée d’Orsay (RF 2736) 1887, Musée d’Orsay (RF 1937-47)
Pissarro and Venturi (1939) #493 Pissarro and Venturi (1939) #709

49. La Brouette, verger Automne, l’étang de Monfoucault
“The wheelbarrow” “Autumn, Monfoucault pond”
~1881, Musée d’Orsay (RF 2734) 1875, private collection, Paris
Pissarro and Venturi (1939) #537 Pissarro and Venturi (1939) #329

Renoir

50. La Liseuse L’Amazone
“Girl reading” “Madame Darras”
1874–1876, Musée d’Orsay (RF 3757) 1873, Musée d’Orsay (RF 1965-11)
Daulte (1971) #106, Fezzi (1972) #202, Daulte (1971) #93, Fezzi (1972) #103

Wadley (1987) #36

51. La Place Saint-Georges La Mosquée, fête arabe
“St. George Place” “The mosque” or “Arab festival in Algiers”
~1875, private collection 1881, Musée d’Orsay (RF 1957-8)
Distel (1994) p.65, fig. 55 Fezzi (1972) #462, Hayward Gallery (1985) #56



342 CUTTING

APPENDIX (Continued)

52. Étude, torse, effet de soleil Nue or Torse d’Anna
“Torso of a woman in sunlight” “Female nude (Anna)”
1875–1876, Musée d’Orsay (RF 2740) 1979, Pushkin Art Museum, Moscow
Daulte (1971) #201, Fezzi (1972) #204, Daulte (1971) #213, Fezzi (1972) #250

Hayward Gallery (1985) #36,
Wadley (1987) #44

53. La Balançoire La Tonnelle du Moulin de la Galette
“The swing” “The arbor” or “The bower”
1876, Musée d’Orsay (RF 2738) 1875–1876, Pushkin Art Museum, Moscow
Daulte (1971) #202, Fezzi (1972) #242, Daulte (1971) #197, Fezzi (1972) # 240

Hayward Gallery (1985) #39

54. Bal du Moulin de la Galette, Montmartre‡ Le Déjeuner des canotiers
“Ball at the Moulin de la Galette” “Luncheon of the boating party”
1876, Musée d’Orsay (RF 2740) 1881, Phillips Collection, Washington
Daulte (1971) #209, Fezzi (1972) #249, Daulte (1971) #379, Fezzi (1972) #468,

Hayward Gallery (1985) #40; Figure 2E Hayward Gallery (1985) #52; Figure 2F

55. Bords de Seine à Champrosay Paysage de neige
“Banks of the Seine at Champrosay” “Snowy landscape”
1876, Musée d’Orsay (RF 2737) 1875, Musée de l’Orangerie, Paris
Fezzi (1972) #256, Wadley (1987) #43 Hayward Gallery (1985) #33, Wadley (1987) #42

56. Le Pont du chemin de fer à Châtou Femme avec parasol dans le jardin
“The railroad bridge at Châtou” “Woman with a parasol in the garden”
1881, Musée d’Orsay (RF 3758) 1873, Thyssen-Bornemisza Gallery, Madrid
Fezzi (1972) #470 Fezzi (1972) #199, Wadley (1987) #22

57. Le Château des brouillards Chemin montant dans les hautes herbes
or Soleil couchant à Montmartre “Path winding through tall grass”
“The chateau of the mists” 1876–1877, Musée d’Orsay (RF 2581)

(Renoir family home) Fezzi (1972) #141, Wadley (1987) #67
~1890, private collection
Distel (1994) p. 65, Figure 56

58. Jeunes filles au piano§ La Leçon de piano
“Young girls at the piano” “Piano lesson”
1892, private collection, Paris ~1889, Joslyn Art Museum, Omaha
Hayward Gallery (1985) #90 Daulte (1971) #561, Fezzi (1982) #661,

Hayward Gallery (1985) #84

Sisley

59. Les Régates à Molesey Les régates à Hampton Court
“Boat races at Molesey, near Hampton Court” “Boat races at Hampton Court”
1874, Musée d’Orsay (RF 2787) 1874, Sammlung E.G. Bührle, Zurich
Daulte (1959) #126; Figure 2G Daulte (1959) #125; Figure 2H

60. Une Rue à Louveciennes Un Coin de bois aux Sablons
“A street in Louveciennes” “A corner of the woods, the Sablons”
~1876, Musée de Beaux-Arts, Nice 1883, Musée d’Orsay (RF 2079)
Daulte (1959) #221 (deaccessioned Daulte (1959) #502

from Musée d’Orsay, RF 2783)

61. La Seine à Suresnes Le Pont à Sèvres
The Seine at Suresnes” or “The bridge at Sèvres”

“The banks of the Seine” 1877, National Gallery, Prague
1877, Musée d’Orsay (RF 2786) Daulte (1959) #262
Daulte (1959) #267

62. Cour de ferme à Saint-Mammès Village de Voisins
“Farmyard at Saint-Mammès” “Village of Voisins”
1884, Musée d’Orsay (RF 2700) 1874, Musée d’Orsay (RF 2019)
Daulte (1959) #544 Daulte (1959) #142

63. Lisière de forêt au printemps À repos au bord du ruisseau
“Edge of the forest near Fontainbleau” “Resting by a brook”
1885, Musée d’Orsay (RF 2784) 1872, Musée d’Orsay (RF 1693)
Daulte (1959) #350 Daulte (1959) #42



MERE EXPOSURE HELPS MAINTAIN A CANON OF ART 343

APPENDIX (Continued)

64. Saint-Mammès Canal du Loing
“Saint-Mammès” “Loing canal”
1885, Musée d’Orsay (RF 2785) 1884, Musée d’Orsay (RF 1972-33)
Daulte (1959) #629 Daulte (1959) #522

65. Seine à Billancourt La Seine à Bougival
“The Seine at Billancourt” “The Seine at Bougival”
undated, private collection 1881, Musée d’Orsay (MNR 208)
Distel (1994) p. 67, Figure 63 Daulte (1959) #90

66. Bords de la Seine, effet du soleil couchant Bateaux à l’écluse de Bougival
“Banks of the Seine, sunset” “Boats at the Bougival lock”
undated, private collection 1873, Musée d’Orsay (RF 1690)
Distel (1994), p.67, Figure 59 Daulte (1959) #90

Caillebotte Images Not Found

Monet Pissarro
Une colline rose (Vues de Vétheuil) Coin de village – les choux
“A pink hill in Vétheuil” “Village corner – cabbages”
undated, location unknown 1875, location unknown
not in Wildenstein (1974–1985) Pissarro and Venturi (1939) #312,

image not presented

Sisley La vallée en été, Pontoise
“The valley in summer, Pontoise”

Station de bateaux à Auteuil 1877, location unknown
“Boat dock at Auteuil” Pissarro and Venturi (1939) #407,
undated, location unknown image not presented
not in Daulte (1959)

Sous-bois en automne, Pontoise
“In the autumn woods, Pontoise”
1879, location unknown
Pissarro and Venturi (1939) #505,

image not presented

*Cézanne’s Image 7c is very similar to two others perhaps equally well known: One is in the New York Met-
ropolitan Museum of Art (Venturi, 1936, #625), and the other is in the Art Institute of Chicago (Venturi, 1936,
#626), both with the same name, “The gulf of Marseille, seen from Estaque.” †Manet’s Image 17c was a study
for Courses à Longchamps, 1867, Art Institute of Chicago (Jamot & Wildenstein, 1932, #202). Image 17n is
a similar study, but done several years later. ‡Renoir painted two versions of 54c. Caillebotte owned the
original. For many years the smaller copy was on loan to the Metropolitan Museum of Art in New York (see
Distel, 1990; White, 1984), but was sold at Christie’s to Ryoei Saito for $78 million in 1990. The week be-
fore, Mr. Saito had purchased Van Gogh’s “Portrait of Dr. Gachet” for $82.5 million. Mr. Saito soon went
bankrupt, and these two most expensive paintings ever sold (through mid 2003) are now probably in a Tokyo
bank vault (Saltzman, 1998). §Under commission from the French state, Renoir painted at least five versions
of 58c. That chosen by the government is now in the Musée d’Orsay (RF 755), another in the Musée de 
l’Orangerie, a third in the Metropolitan Museum, and at least a fourth in a private collection (Hayward Gallery,
1985, pp. 261–263; Wadley, 1987).

(Manuscript received December 10, 2001;
revision accepted for publication May 17, 2002.)