i
i
Representations are made in a world of cooperating makers and users. p The work of making them is divided among several kinds of makers, and between makers and users. If a representation requires the four kinds of work specified earlier, who does each kind? What the makers don't do must be done by the users, if a representation is to be created and communicated more or less to the satisfaction of everyone in- t volved. Once they establish a division of labor, how do the cooperat- * ing parties coordinate the different things they do? i Sometimes the maker does most of the work, leaving only a limited i margin of autonomy to the user. When we see a film, the filmmaker has chosen and arranged everything, and our activity is restricted to seeing what we can make of what's been done and having an opinion about it and the matters it takes up. (Of course, changes in technology make it possible for us to see films in an order other than the one the maker intended, though not if we see it in a theater.) But even while allowing us an apparent freedom to interpret and judge what we see as we will, filmmakers use all the devices of their craft to channel our responses in the direction they want us to go. The authors of scientific articles, in Latour's description (1987, 21-62) of their activities, intend to keep readers under even stricter control. They anticipate questions and criticisms of their work and build answers and defenses into what they write so that the reader finds it impossible to counter their arguments. At least, they aim for that kind of control, though they often fail to achieve it and so become the target of criticism and, worse yet, see their results appropriated for uses they never intended and may not approve.
WHO DOES WHAT? : 31
In other worlds of representation making, makers leave much of the work of arranging and interpreting to users. Some artists who make representations of social reality do that deliberately. Refusing to make the generalizations that seem to leap out of the material they present, they resolutely leave that job to users. Here too the freedom is sometimes more apparent than real, because the makers use the technical and conceptual tools of their trades to channel user activity and response.
Suppose that you have made the hard choices of what to include in the report (the story, the film, whatever a report is called in the medium you are working in) you want to make on the social phenomena you've investigated. You've got the "data," the raw material. You have swallowed a bitter pill and accepted that you can't incorporate everything you have collected and think and still achieve anything useful to you or the people you mean your report for. You accept that some, perhaps a great deal, of your hard-won knowledge and material will end up, as film people used to say, on the cutting-room floor. Now you have what remains after this winnowing, a pile of fragments: strips of film, pages of numbers, files full of field notes.
How can you arrange all this stuff, put it together so that it communicates what you want to communicate to the people you want to communicate it to (and, of course, communicate what they want you to communicate to them)? Writers of social science (and other scholarly) texts typically experience this as the problem of constructing an argument, saying what needs to be said in an order that presents your ideas so efficiently and clearly that readers or viewers will not mistake them for something you didn't mean and so that all criticisms and questions will be forestalled. Dissertation advisers and journal editors repeatedly tell authors to "get your argument straight." And that advice applies, beyond the logical arrangement of propositions and conclusions and ideas, to the presentation of your evidence, the material you have selected from your investigation's data. How do you arrange that material, whatever its form, so that it says what your formal argument says and makes your conclusions manifest, unmistakable, and inescapable to any reasonable reader or viewer?
The answers to questions like these take us directly to the question of the variety of ways that makers and users can divide the representa-
32 : CHAPTER THREE
WHO DOES WHAT? : 33
tional labor between them. I'll focus on two quite different examples: the conventional social science problem of presenting statistical data—numbers—in tabular form and the problem of arranging what are usually called documentary photographs in some sort of order for presentation on the wall of a gallery, in a slide show, or in a book.
The Statistical Problem
Let's begin with the statistical problem. I have done a census or a survey or an experiment and have counted a number of things. In a census, we count people and find out a number of things about each person we count: their age, sex, race, last grade of school completed, income during the previous year, and so on, depending on the design of the particular census. In an experiment, we create two or more groups, do things (the "experimental treatment") to one group and nothing to the other (the "control group"), and measure a variety of things we think will result from that "treatment." Surveys mimic the experiment, though the researcher cannot control who gets what experimental treatment applied to them, since what is taken to be a causal variable is something like age or sex or some aspect of previous experience that the researcher cannot manipulate but will "control for" statistically.
Doing any of these things produces a lot of numbers. The individual numbers don't mean or matter much. I don't care, and neither does anyone but the person's family and friends, what age this particular person is or how much money she made last year. If I add up all the incomes of people of a certain kind and find their mean, that may seem on the surface more interesting, but it actually isn't. The average income reported by people who live on this particular block in Chicago is $19,615. Twenty-seven percent of the people living on a particular block tell the census they are black (that's how the U.S. Census measures race), or thirty-six percent say they are over 65. So what? Those numbers, standing alone, still aren't interesting.
Why? Because we haven't yet asked the crucial further question: compared to what? The readers of census tables make sense of the numbers in them by comparing them with one another. They look at
I two numbers and ask: Are they the same, or is one bigger than the
I other? And if one is bigger, is the difference big enough to take seri-
I ously? To make that figure of $19,615 as the average income of a block's
I inhabitants meaningful, we have to compare it to another number. To
I what? To, perhaps, the $29,500 (or 50 percent more) that people
f who live on another block make. Armed with that comparison, we
I might conclude that the city is characterized by a geographical segre-
Igation of income groups. Or perhaps blacks or people who are over 65 make 25 percent less than people of other races or ages, so that we can conclude that there is racial or age discrimination in income. Now we i think we know something. The difference between the two numbers,
I revealed by the comparison, conveys the important information.
It isn't only the difference between two coordinate groups (black vs. white, greater than 65 vs. less than 65). We might compare the group we studied to the larger group that contains it—the people on I this block compared to the entire city—or with some external stan-
I dard, the people of this racial group compared to the "poverty line."
>, The problem of arranging my statistical results, my numbers, is a
problem of making the relevant comparisons visible. That's why the volumes of the U.S. Census don't provide any conclusions. Being files rather than arguments, they don't explicitly compare anything; they just provide the raw materials for comparisons, which is why so many people can make a living rearranging what is available free to us all in those census publications.
In fact, the census usually prints data in a tabular form that makes some comparisons easy, as in this cross-tabulation of income by age I've invented to illustrate the point. The rows of the table are labeled with age groups (0-15,15-25, 25-35, etc.), an^ the columns with income groups ($10,000-15,000, $15,000-25,000, etc.). The cells of this grid of rows and columns contain numbers, the number of people characterized by that combination of age and income. This makes it easy to compare adjacent cells and learn that there are more people age 25-35 m the $i5,ooo-$25,ooo income bracket than people 35-50 (if that's the case) but that the income difference between the two age groups lessens as income increases. All we have to do is look from one cell to the cell next to it to see that, above $40,000, the numbers
34 : CHAPTER THREE
are the same in the adjacent cells. But we might want to compare non-adjacent cells—the income differences between people aged 15-25 and people over 65—and then we would have to copy the numbers we want to another piece of paper in order to place them side by side for comparison.
Hypothetical census table
Age
Income (in $)	0-15	15-25	25-35	35-50	50-65	65+
0-15,000						
15,000-25,000	400	300	200	100	75	60
25,000-40,000	350	275	225	125	70	55
40,000-60,000	250	250	250	150	50	50
60,000-90,000	50	125	200	200	40	30
90,000 +	25	100	175	175	25	35
In statistical comparisons like these, what we are comparing appears in the labels of the rows and columns of a table. If we're interested in the relation between average income and age, we head the columns with the names of the age categories and the rows with categories of income. The reader takes on the analytic job of seeing that people over 65 make less, if that's the case, than people in other age categories.
Census tables are made by highly trained professionals for a large and varied audience of potential users. Those users don't have to create the categories of comparison: age and income, or gender, race, years of education, and all the other variables so easily available from the U.S. Census Web site or in its publications. The makers of the table have done that analytic work—creating the categories—already, just by labeling the headings of the rows and columns with those dimensions (typical headings for many census tables). Making these the headings of the rows and columns—the dimensions of the table — tells users to make comparisons like these: Do people aged 35-50 earn more than people 25-35? Or> with other variables represented in the rows and columns, do blacks get less schooling than whites? Do women earn less than men? The professionals who design tables
WHO DOES WHAT? : -15
worry about arranging the dimensions and numbers so that readers can easily make the important comparisons. (See the discussions in Tukey 1972 and Tufte 1983, 1990, and the historical discussion in Desrosieres 1993.)
J        The Photographic Problem
That's how things work in a world of representation making in which professionals do a lot of the work for a large and heterogeneous group of users. Now consider similar problems as they arise in the world of documentary photography, which seems, on the surface, very different. And it is, but there are similarities that let us specify the actual differences more precisely. Doing that shows us another way of dividing the labor of arranging between makers and users.
Suppose that 1 have made a large number of photographs—a serious documentary photographer pursuing a big topic would make many thousands of exposures—and have selected the images I think \ best convey the ideas I have arrived at about that topic as I went about
J making them. And let's take a classic example of the genre, one of the
I most discussed and admired works of its kind, often held up as a model
it for aspiring documentary photographers: Walker Evans's American
Photographs ([1938] 1975). Evans created this book from photographs he made over a period of several years, all over the eastern United States, south and north (the farthest west he got was Baton Rouge): New York, Pennsylvania, [ Mississippi, Alabama, and so on. And not all in the United States—
you have to interpret the title generously, since he made three of the pictures in Havana). He wasn't completely clear about what he was after when he made all these images. According to Alan Trachtenberg, I a profound student of his work, Evans was trying to answer the ques-
I tions the Great Depression had raised for a lot of American intellec-
tuals: "What is special about the American people? What are their characteristic beliefs, their folk history, their heroes, their work pat-I terns, and their leisure? . . . Evans's concept of America cannot easily
1 be defined by enlisting him in any particular camp, but it can be said
I that his work belongs within the general pattern of . . . the search for
36 : CHAPTER THREE
an authentic American culture and one's own Americanness" (Trach-tenberg 1989, 247).
We can find further evidence about Evans's intentions in a letter he wrote to a friend when he was at work making these pictures, listing what he was after:
People, all classes, surrounded by bunches of the new down-and-out. Automobiles and the automobile landscape.
Architecture, American urban taste, commerce, small scale, large scale, the city street atmosphere, the street smell, the hateful smell, women's clubs, fake culture, bad education, religion in decay.
The movies.
Evidence of what people of the city read, eat, see for amusement, do for relaxation and not get it. Sex.
Advertising.
A lot else, you see what I mean. (244)
His intuition, led by such concerns, produced the archive of images he drew from for the book. He finally chose one hundred pictures from that archive for his exhibit at the Museum of Modern Art. From those he took eighty-seven to be included in American Photographs.
Having made these choices, he now had to deal with what seems an apparently simple problem: in what order should the images appear in the book?
There's a preliminary, practical question. Not what order to put the images in to generate an effect you want, but what order of viewing you can get viewers to respect. You can't force people who come to an exhibit to see the photographs in a particular order, and you can easily observe that some viewers come through the entrance and immediately start working their way around to the right, while others, with equal conviction, turn left. And, maddeningly for the photographer, readers as often leaf through a photo book from the end as from the beginning. Does the order of the images in a photographic sequence matter? Photographers regard this apparently simple question as crucial and difficult.
Whatever the problems, photographers, along with exhibit design-
WHO DOBS WHAT? : 37
ers and museum curators, want to make viewers see things in a specific arrangement that they hope will push viewers to make certain comparisons along certain dimensions, generating particular moods. They understand that a single image is ambiguous and does not easily and unequivocally reveal "what it is about." When photographers make pictures for such other purposes as news and advertising, they usually compose them so as to exclude all "extraneous" detail, everything except the "point" of the news story or the product feature to which they want to call attention. They choose the details that surround that point carefully, to emphasize the story's main ideas or enhance the product's appeal (Hagaman 1996.11). Pictures made for scientific purposes similarly restrict their content to what the maker (usually the author of the scientific article) wants users to know and rigorously exclude anything extraneous to that purpose.
Documentary photographers like Evans don't reduce the contents of a photograph in so ruthlessly comprehensive a way. Looking for photographic truth, they let what's there be there. As a result, most pictures made as "documentary" purposely contain a large amount of detail, all sorts of stuff that was in the area when the image was made, even when that stuff does not support any simple interpretation of what's going on. The crucial work of interpretation is left to the user, with whatever control the maker attempts left implicit. Though the pictures are carefully composed, so that the detail is not just random noise, viewers can interpret them in many ways, depending on which details they emphasize and what they make of them.
An image that contains so much detail will always support more than one interpretation, and certainly more than the simple scripts that inform newspaper stories or advertisements. Which raises this question: since this division of labor leaves the interpretation to users, how will those users know what's important, what the idea is, what the photographer had in mind, what they are "supposed to get out of this picture'"? How can photographers arrange the pictures so that what they had in mind will shape the interpretations of the people who see their work?
Ordinarily, a caption tells us what's important, points out what we should attend to, tells us what we can ignore, indicates how the con-
38 : CHAPTER THREE
WHO DOES WHAT? : 39
nections that link the objects and people in the picture. Some documentary photographers help viewers along with extended captions. Dorothea Lange sometimes attached a lengthy explanation, as when she captioned the image (sometimes called "Tractored Out" and reproduced in many places, for instance, in Stryker and Wood 1973,100) of a small farmhouse marooned in a plowed-up field—the result of the buying up of small Dust Bowl farms by large agricultural conglomerates, which didn't even bother to demolish the small owner's home—"Abandoned farmhouse on a large mechanized cotton farm" (see figure 3.1). Sometimes photographers embed their images in a text.
Danny Lyon's book about a motorcycle gang (Lyon 1968) mixes photographs of the gang in action with long interviews with its members. Other photographers— Evans was one—leave their images verbally unattended, except for the place and date of the image's making, and this has the result Trachtenberg describes:
An uncaptioned sequence of pictures suggests a hidden author, one who keeps out of the reader's way—like Flaubert or Henry James— but maintains a consistent point of view, a physical and moral perspective. The analogy cannot be exact, for what choice does the editor of photographs really have? Except for its denotations, what it is a picture of, a photograph can arouse widely varying interpretations, and thus, unless an editor anchors the image in an unambiguous caption, its meaning is too open and indeterminate to provide a reliably secure point of view. (251)
The maker can, however, indicate the image's meaning, using what the film director Sergei Eisenstein called montage. Again, Trachtenberg:
Any grouping of images within the book can be taken as an example of Evans' adaptation of the montage device, which can be restated as a dialectical process of thesis giving rise to counter-thesis, together producing as feeling and/or idea an unseen, unstated synthesis. Each picture discloses a link to the next, a hint or germ of an antithetical image to follow. The reader is expected to remember each image fully, in all its details and nuances, for the most inconspicuous details become significant in echoes and allusions further on. What the pictures say they say in and through the texture of relations which unfold—continuities, doublings, reversals, climaxes, and resolutions. (259)
3.1 Dorothea Lange: Tractored Out: Abandoned farmhouse on a large mechanized cotton farm.
That is, the image an image follows, the image it precedes, and those even farther away in the sequence of pictures the viewer sees—all those pictures condition our understanding of the picture we are looking at right now. In fact, every image influences our understanding of every other image. Nathan Lyons distinguishes a series, in which the order of the photographs is important, from a sequence, where it isn't. If what eventually matters are the resonances and echoes between the photographs, which attentive readers, as Trachtenberg says, have stored up in their heads, then the initial order in which we encounter them may not, after all, be so important to our ultimate understanding of the work. Whatever the order, on this view, all the images we have seen affect our understanding of any single image.
Comparison
How does that happen? How do we use the materials of a sequence of images to create our understanding of what they "mean," the ideas they convey beyond a mere listing of what's there?
40 : CHAPTER THREE
We do it by comparison, just as the readers of statistical tables make sense of numbers by comparing them with one another. To be explicit, we look at two pictures together and see what they have in common, and we take that common feature to be maybe not everything the picture is about but, at least provisionally, one of the things it is about. Using the language Leonard Meyer (1956) and Barbara Herrnstein Smith (1968), respectively, used about music and poetry, we might say that we hypothesize that that common feature is what these pictures are about. We go on, of course, to test the hypothesis with succeeding pictures, as Meyer and Smith suggest we do in listening to music or reading poetry. We look at a third picture, seeing if it has the features our hypothesis about similarities suggests. When it doesn't have them exactly, but only partly, we revise our hypothesis, our idea of what the sequence is about. And so on, comparing each picture that follows, again and again, to the images that have come before, using the accumulated understanding of similarities to arrive at our understanding of what the whole sequence is about.
We don't, of course, just find similarities, any more than the statistician finds that all the numbers in a table are the same. The statistician sees which number is bigger. But photographs contain more detail than an unadorned number, so we have more comparisons to make and more complicated hypotheses to consider than whether two items are the same. We find differences as well as similarities, and we note those differences and see what we can make of them. Do they suggest a second theme? A variation on the first theme? Do we see a connection between the two themes?
Trachtenberg does just that with the first six images in American Photographs, explaining how the successive references to cameras and photographs and situations of photographing lead viewers to conclude, if their reading of similarities and differences coincides with Trachtenberg's, that the sequence is about photography and image making (it is useful to read what follows with Evans's book in hand, looking for the features and relationships Trachtenberg describes):
The movement from the opening picture through the second to the third encapsulates the method of the book: from a conception of the
WHO DOES WHAT? : 41
photograph as mere identification to a subversion of that idea in the second image (where "Studio" cues our response to the wit in the event: a single picture made of, and commenting upon, many small pictures), to a picture free of writing and full of ambiguity, of the two boys looking elsewhere. Their glances beyond the frame of the image tell us that the world is wider and more full of circumstance than any photograph can show, that photographs cannot properly "identify" because they leave out too much, that reading has its limits and must take the arbitrariness of the picture's frame into account: an admission of contingency absent from the "studio" images implied or shown in the preceding pictures. (264)
The subtlety of Trachtenberg's analysis shows what a sophisticated reader can make of a carefully arranged sequence of photographs. But note two things about a reading like this. One is that the reader must really be sophisticated, must know how to "read" photographs in a sophisticated way. The other makes itself evident in a comparison with the reading of statistical tables.
A sophisticated reader of photographs does consciously and carefully what any ordinary reader of photographs does unreflecti vely and carelessly. A conscious and careful reading differs from an "ordinary" reading, first of all, in its deliberate thoroughness. We can guess that all viewers of a photograph respond, whether they know it or not, to everything in the frame. They are all affected by the tonalities and composition, they register all the small details, but they don't know they are doing it. They take a quick look, add it all up, and say, "Oh, yeah, that's striking," or "That's sad," or "It really captures the essence of that thing." But they don't know what went into their summing up of what the photograph captured or just how they conducted their interpretive operations. How you conduct these summarizing operations makes a difference, just as how you calculate a statistical measure of central tendency makes a difference; a mean is not a median is not a mode.
A conscious and careful reading, on the other hand, takes time. The sophisticated viewer goes over every part of the picture, registering explicitly what's there, what point of view it represents (where the photographer put the camera in order to get that particular view,
4^ : CHAPTER THREE
among the many that might have been chosen), the time of day, the things that were left out but perhaps hinted at by the framing of the image, and so on. The sophisticated viewer knows the photographer could have made, and perhaps did make, many other versions of the same material, in which all those choices were made differently, and so reads what's in the frame as the result of the photographer's deliberate choices, which combine to produce the final effect. A careful reader of photographs spends a long time on each image.
As a result, a sequence of photographs has the kind of meaning Trachtenberg teaches us to look for only when the reader puts that kind of time into the consideration of every photograph and of the relations of each of the photographs to all the others. A book like American Photographs thus requires as careful a reading as a complex poem of similar length {Trachtenberg compares Evans's book to T. S. Eliot's The Waste Land).
The second major difference between the statistical table and the photographic sequence, and the more important one here, is that the division of labor between users and makers differs in the two cases. The maker of a table does a lot of interpretive work for users that the maker of a photographic sequence requires users to do for themselves. In a table, remember, the rows and columns are labeled with the names of the categories, and their subdivisions, which we are to take account of. The statistician who prepared the table has done that analytic work for users, telling them, in those row and column headings, that age, sex, race, income, education, and other variables are what matter and that they are divisible into just the divisions recognized in the labels (25-35 years old, $i5,ooo-$25,ooo, male or female). The grid constructed by putting two or more of these divided categories together (as I put age and income together earlier, creating what statisticians call a cross tabulation) lays out all the possible combinations. The entries in the resulting cells tell us how many cases of each combination there are: how many people who are between 25 and 35 years old make between $15,000 and $25,000 a year, and how many make between $60,000 and $90,000, and so on for every combination of age and income.
We can think of the sequence of photographs in American Photo-
WHO DOES WHAT? : 43
graphs as something like the entries in such a table or grid, each image a piece of "data," a fact given to users to work with. When they compare the images in a photographic sequence, however, they don't have the kind of help given by the headings of the table's rows and columns- No one made a table or labeled the rows and columns for them. Ho one told them what the important dimensions for comparison are, at least not explicitly. And consequently, no one has described the range of possible combinations for them. The photographer leaves that work to the viewer, the first step of whose analytic job is to find out what the dimensions of comparison are, or might be, or could be. The next step is to work out from that what kinds of combinations of people and situations and their interaction the segment of society the photographer is telling us about contains. The result of their work is not the items to be found in the cells of the resulting table but the la-| bels on the rows and columns themselves, the dimensions that the I comparisons between the images tell us are important. I What kind of dimensions can we find in American Photographs, and
i what would the resulting table look like? What follows is one possibility, a sketchy, merely illustrative analysis that starts with two images Evans made of the experience of women in the streets of New York. Other interpretations than the one I've made are possible, which is one of the results that follows from this exercise.
In "A Girl on Fulton Street, New York, 1929" (39), we see a slim young white woman, turned away from us so that we see only her left side in profile (figure 3.2). She's wearing a dark coat with a large fur collar and holding a muff of the same fur, and she wears a black cloche over her bobbed hair. She has what you might want to call a "hard," even "angry" look; you might want to say she looks "wary." Or you might not. We can agree that she doesn't look relaxed or at ease. She's the only figure in the frame in sharp focus. Three men behind her, all wearing fedoras, are a little blurred, and the figures beyond them even more so. They're in a crowded downtown street lined with stores, some advertising signs, and a construction crane.
"42nd Street" (43), separated from the first image by "Interior of Negro Preacher's House, Florida, 1933," shows a black woman, older and heavier, well dressed in a coat with a fur collar, a string of pearls
44 : CHAPTER THREE
3.2 Walker Evans: A Girl on Fulton Street, New York, 1929. Image © The Metropolitan Museum of Art.
around her neck, and a hat perhaps a little less stylish than the white woman's (figure 3.3). She stands near the stairs to the elevated train, down which a man is coming, and the street behind her is filled with traffic, signs, and the supports for the train tracks. The tones are darker than those in the Fulton Street picture. The woman's look is harder to describe: heavy-lidded, a little suspicious of the man who is making the photograph, a little wary too.
Considering these two pictures, you might provisionally conclude something about women's experience on these New York streets, and
WHO DOES WHAT? : 45
3.3 Walker Evans: 42nd Street. Image © The Metropolitan Museum of Art.
perhaps something more general about the lives of women, as those are embodied in just such moments on the streets as Evans gives us. When we compare the two images, our intuitive grasp of how they seem alike tells us some of the dimensions of comparison. We might say that women in New York are ill at ease and wary when they are on the streets. And our next thought is that these two women are alike in that way, the likeness emphasized by the similarity of their hats and furs, even though they differ in age and race, but they are both very different from the country woman who appeared earlier in the book ("Alabama Cotton Tenant Farmer's Wife, 1936," 33) in her plain dress and hairdo, standing against the weathered boards of her house (figure 3.4). She doesn't look wary, but you couldn't say she looks at ease either; we might think she's a little bashful and embarrassed to have the photographer from New York making her picture with that big camera, that she might be wondering what he wants with a picture of her. That tells us that "wary" does not exhaust the possibilities; there are still more things to include in our thinking about women's lives.
We can go on to compare these women to the men we see—for instance, the dapper older black man wearing a white suit and a white
46 : CHAPTER THREE
3.4 Walker Evans: Alabama Cotton Tenant Farmer's Wife, 1936. Image © The Metropolitan Museum of Art.
straw hat with a black band standing in front of a newsstand filled with Spanish-language magazines and newspapers, topped by a Coca-Cola sign ("Citizen in Downtown Havana, 1932," 45). He seems so at home, so unwary, so at ease, in another urban milieu, in another country (figure 3.5).
The first outcome of such a photographic analysis, conducted by the viewer with the materials the photographer provided, might be that "Girl on Fulton Street" tells us that this white woman, and perhaps all white women or all white women of a certain age and class, standing in the street in New York look like this, the "this" suggesting perhaps a mood or an attitude toward being in public and on display. When we see "42nd Street," we conclude, provisionally, that this black woman standing in the New York street also looks like that, her own
WHO DOES WHAT? : 47
3.5 Walker Evans: Citizen in Downtown Havana, 1932. Image © The Metropolitan Museum of Art.
version of "like that." But we also compare the entries in what now look like two adjacent cells in a grid, a table in the making. We decide that the two have this look in common and that what they have in common suggests something about the way women feel they must conduct themselves in public in New York. We might decide, looking hard, that the looks differ as well—that the black woman's gaze is per-
48 : CHAPTER THREE
haps more guarded—in ways that maybe traceable to the differing social situations of black and white women, or to the different situation of women of different ages, or to the different situations of women of different social classes. And we take those notions to other pictures in the sequence, and perhaps decide that being a woman in New York makes you hard in ways that living in Alabama don't, and vice versa. And that adds another dimension to the table of possibilities. So the viewer's work produces not just a list of possible combinations of life situations but the grid of comparisons itself, the space defined by the intersections of all these possibilities and their interconnections.
Let's be logical about what's going on. Eve 17 time we describe someone as a "woman" or "white," or describe a situation as "urban," we automatically introduce other possible labels, which might be symmetrical—"man"—but more likely will be a list of coordinate alternatives: "black," "Asian," "Native American," and so on. If one situation is "urban" that points to other degrees of population density: "suburban" and "rural," maybe "exurban," perhaps others. The term we use alerts us to the existence of a dimension along which there are other positions than the one we've pointed to.
The imaginary table I've been referring to expresses the logical analysis visually. It shows all the possible combinations of the descriptive dimensions we have used informally. Having included "woman," because the two New Yorkers are women, we created, for our analysis, the dimension of gender (so making room for the category "man"). Noting that the two women differ racially gave us race as a dimension; we don't know all the subtypes we will use under that heading yet. Seeing the women react "warily" to being observed on the street in New York, we created a dimension of "reactions to being observed in public." We have to add, remembering the Alabama farm woman, the rural-urban continuum, with whatever stops along it we think appropriate or necessary.
In this way, we do the work that census statisticians do for us when they lay out a table. We name the rows and columns. When we combine them, labeling the columns with gender terms and the rows with names of possible attitudes toward being observed in public (recognizing that we will surely add more rows as we see more kinds of re-
WHO DOES WHAT? : 49
aCtions to being in public), we see a larger conceptual space than Evans pictured, but one that is implied (if you accept this analysis) by the images he put in his book. We have some help from the photographer, who composed the images so as to suggest some possibilities rather than others and then arranged them in a way that hints, through the comparisons I've been discussing, at what the dimensions and intersections of the table are or might be.
Having done all that, which sounds like more work than it is, we can go on to inspect other images, about which it hadn't occurred to us to raise such questions, to see what they add to our understanding of the specific cases pictured but also of the general ideas and categories suggested.
Now we can see some advantages the photographic method has over the tabular one used by statisticians. I was originally going to create the table that the preceding analysis of Evans's book might generate, but I gave up the idea when I realized what a mess that would make. Tabular forms are very helpful when you deal with a relatively few categories. You generate amanageable number of labels and cells. But every time you add a new dimension, you double the number of cells required. (You'll find a very clear discussion of the process in Danto 1964. He uses the example of judgments of artistic worth but also explains the logic of the analysis very clearly.) En the simplest case, two variables, each taking only two values, generates four cells. Example: age, divided into old and young, cross-tabulated with gender, divided into male and female. (As an exercise, you can draw these tables for yourself.) Each cell contains an important fact: how many people have just that combination of characteristics (in a slightly more complex version, what percentage of the people in that cell have a value of x on a third characteristic, like "rich" as opposed to "poor"). If you now add the population density variable, divided into rural and urban, you must divide each of the four age/gender cells into two, one for rural, one for urban, ending up with eight cells. Every additional subdivision—if you, for instance, added a category for suburban—increases the number of subheads in a row or column and the number of cells. (We'll return to the problems of displaying such information in a table in chapter 5.)
so :
CHAPTER THREE
WHO DOES WHAT? : 51
When you cross-tabulate four or five characteristics, the resulting table has so many cells that it's difficult—not impossible, but difficult—to find the two numbers the table is supposed to help you compare, and it has defeated its own purpose. A table with ten cross-tabulated variables, containing 1,024 cells, is so unwieldy you would have trouble publishing it, and if you managed that, users would have a hard time manipulating it physically, let alone making sense of its entries.
The documentary photograph works differently. It typically contains so much detail that an interested user can easily make a great many comparisons between any two such images, every comparison suggesting a dimension of variation and its possible subdivisions and adding to a list of questions to be asked in looking at succeeding images in the sequence. It contains, in embryo, all of these possibilities, the number mainly constrained by the user's ingenuity in exploring what's there. Not all the comparisons will produce ideas that can be sustained over the course of a long sequence, hypotheses about what the sequence is about that hold up when confronted with the succeeding images. But some, and not just a few, will do that. These ideas will not contradict each other. They will be complementary, suggesting more complex hypotheses that link the subthemes a viewer might construct.
All this work of constructing categories of comparison and their divisions, creating hypotheses and checking them, falls to the user. The maker furnishes the raw materials (in truth, not so raw as all that), artfully chosen and arranged to be sure, but after that it's up to the user to construct the analysis, witli all its paraphernalia. That's a far different division of representational labor from the one that goes into the making and use of a census table.
The multitude of details in a documentary image gives viewers material with which to construct more than one comparison of the kind I've alluded to. You can make more than one table out of a lengthy sequence of detailed photographs. There are many comparisons to make, many dimensions to explore, many stories to tell. We might, for instance, focus not on the women standing in the street but on the streets themselves, the way they look, and what they tell us about life
f
3.6 Walker Evans: Main Street, Saratoga Springs, New York, 1931. Image © The Metropolitan Museum of Art.
in America. That means that we will now include in our comparison all the images of streets in which no people appear, such as the haunting image of automobiles parked head-in to the curb in the rain ("Main Street, Saratoga Springs, New York, 1931," 59). Which leads us to comparisons with other streets seen in other pictures, in Bethlehem, Pennsylvania (117), Fredericksburg, Virginia (153), and a variety of other towns, large and small (figures 3.6, 3.7, 3.8).
WHO DOES WHAT? : 53
3.7 Walker Evans: Street and Graveyard in Bethlehem
So a well-made photographic sequence supports a large number of comparisons and thus a large number of interpretations, which is why we can continue to attribute more and more meaning to what is, after gj^ a small number of images. And why it is difficult-—in fact, impossible —to settle on a definitive interpretation of such a work, and why American Photographs repays repeated readings by giving you new interpretive possibilities. Evans did his part of the job. He made and selected the pictures, which contain the possibilities, and put them in a book. He left the rest for the user to do.
3.8 Walker Evans: Frame Houses in Virginia, 1936.