Report on a Book Project Titled How Pictures Die: Six Stories from the End of Representation

James Elkins

This paper is an abstract of the talk I have at the conference "Das Bild" in Berlin in November 2000. It is truncated, for two reasons: first, the book is evolving, and I have already deleted some material on the sublime that formed the original subject of the talk; and second, some of the remaining material is appearing elsewhere. May I direct readers to the following two sources for additional information:

- "Logic and Images in Art History," Perspectives on Science 7 no. 2 (1999): 151-80.

- "On Some Useless Images," Visual Resources, forthcoming (2000).

Nevertheless, the material in this text is free-standing and self-explanatory. Please send all comments to jelkins@artic.edu; and for further information see www.jameselkins.com.

A. Purposes of the book

The book project, How Pictures Die, is about images in six different fields (recent painting, recent photography, astrophysics, electron microscopy, quantum mechanics, and cosmology). The common theme is that the images are made by people whose primary concern is how and when the normal methods of representation break down. In contrast to images made in previous decades and centuries, these are primarily engaged in the limits of their different representational techniques: that is, there is minimal interest in those areas of representation that are adequately served by whatever techniques (artistic, technological) are at hand.

The book is intended to chronicle those images, and propose that some of the most interesting images in several fields can be described as being at the limits of naturalistic representation.

The book also has two further purposes: it’s my contribution to the "science wars" (the ongoing misunderstanding between the sciences and the humanities); and it is intended as a critique of the normal fashion in which art history intersects science. A few words, then, about those two ancillary purposes.

I am concerned about a tendency in art history to encounter science in terms of its points of intersection with visual iconography--in other words, art historians who look at science tend to look mainly at times and places when artists had namable scientific sources.

For example, they may study what Duchamp and Picasso knew about Einstein’s relativity, or how Galileo’s interests in the arts may have influenced his science. (Here I’m thinking of Linda Henderson at Austin, and Eileen Reeves’s recent book on Galileo.)

In the last few years, I have become increasingly skeptical about that kind of work. Basically, it seems to me that very little important art—art that’s central to ongoing conversations about modernism or postmodernism—is influenced by science.

It is true that many minor artists have used mathematics and science (I’m thinking for example of Escher), and it is easy to take ideas from popularized science or technology and put them in artworks. It’s also true that some major artists, like Duchamp, have flirted with science (or rather pseudoscience, and popular science): but that doesn’t help explain what makes their work important; and in general looking for art-science links doesn’t help interpret the overwhelming majority of post-Renaissance art.

Serious art and serious science remain very different, and people who study their connections end up looking at special cases that can’t, I think, add to a plausible account of wider movements and ideas.

(As usual, Tim Clark has the best phrase for this—it’s somewhere in Farewell to an Idea—he says "art usually dines very poorly on the leavings of science.")

So my idea in this book is to stop finding causal connections between art and sciences. Instead I’m trying to find ways of talking about scientific and non-scientific images together, without special pleading or popularization.

In terms of the "science wars" (the book’s second ancillary purpose): it seems essential to me to try not to simplify either the relevant science or the art.

There has been a bit too much low-level squabbling recently between scientists and humanists, especially since Alan Sokal’s hoax and his subsequent book, and Bruno Latour’s playful, and, I think, somewhat irresponsible rewriting of special relativity.

[For Sokal’s work the best source is his own web page, www.physics.nyu.edu/faculty/sokal.]

In the journal Physics Today, which is the official publication of the major association of American physicists, the letters column is filled with attacks and defenses of Sokal’s book and of Latour’s essay. People like Irigaray, Derrida, and Lacan are still being "exposed" as ignoramuses, and scientists are still claiming that postmodernists know that what they do is nonsense and they only do it to get "cushy university jobs" (I’m quoting a physicist in the "Letters" section of the October 1999 Physics Today).

It seems to me that if the dialogue between scientists and humanists is to go forward, an essential first step is that neither side popularize unduly, so that the descriptions of works are acceptable, prima facie, to readers from various disciplines.

So the book has the actual physics—I mean the relevant equations—as well as philosophy and art criticism. I don’t mean to say that the full detail of the science can meet the full detail of the art history or criticism and result in a sensible discourse. But to me the challenge is to see what happens when the levels are kept as high as possible.

Today I can only evoke the book’s central argument by looking at three examples from different chapters: painting, astrophysics, and "transgenic art." (See the published sources listed at the top for other examples.)

B. First example from the book: painting

The first example concerns an attempt made by the art historian Rosalind Krauss to rescue Agnes Martin’s paintings from the criticism of the sublime. In the book this analysis proceeds slowly, in steps: first an account of the Kantian mathematical sublime; then a discussion of why the Kantian sublime has a bad name in art history; an introduction to Martin’s paintings; and a summary of Krauss’s phenomenological account, whose purpose is partly to show how an unbounded Kantian sublime is not an adequate description of the experience of seeing Martin’s paintings.

Here I abbreviate that sequence.

First, to introduce the sublime, I will review Kant’s original description. Kant divides sublime experience into two types: the mathematical, and the dynamic. The latter happens when an experience is overwhelmingly, irresistibly powerful, like a stormy ocean. The first kind of sublime, the mathematical sublime, has often been taken to be a perfect description, avant la lettre, of central developments in twentieth-century image making.

Kant defines it as an experience of something unencompassable, so large that it exceeds our capacity for comprehension. His example, the starry sky, is also one of my subjects in the book. A four-step sequence—which is derived from a common reading of Kant—can serve to introduce the salient points.

At first, confronted with the starry sky, you are dazzled, confused, and humbled.

Then you try out some concept to help comprehend the incomprehensible heavens: say for example the concept of infinity. You say to yourself, "This is infinite," and for a moment you are comforted.

But there follows a third moment when it becomes clear that the object is too expansive to be understood. It cannot be gathered under a single concept, a single intuition. And further: the concept itself, in this case "infinity," is not directly experienced--it can’t be encompassed by human understanding.

There follows a fourth and final moment when you realize that your innate capacity to reason is what drives the desire to encompass an unencompassable object with an inadequate concept.

Even though the attempt to understand the object fails, you become aware of a mysterious, inbuilt capacity to try to match your imagination to objects: a capacity that includes the very idea of a fully adequate concept even though no such concept can be brought into the intuition.

Epistemologically, it is also possible to imagine that Kant’s description divides experience into three parts: the part that is known, the borderline, and the part beyond that is not known.

[In lectures introducing this idea, I draw a picture on the chalkboard: the area of cognition and concepts is on the left, the region of the nonconceptual on the right, and a barrier in the middle.]

The word "sublime" means "up to the threshold" (limen means "lintel" or "threshold"): so the sublime experience is one that takes you up to the edge of what you can understand.

Limen is a post-Kantian term (his term is Erhabene, which has entirely different connotations and etymology): but it is a good representation of the barrier that exists, in different terms, in his own account. In Kant, the limen, the threshold, also functions as a limes, a path: the experience can move beyond what is given to the intuition by means of the faculty of suprasensible reason.

So for the immediate purpose the four-step sequence and three-step schemata comprise an adequate description of the Kantian account.

Especially since the Second World War, the literature on the sublime has been growing exponentially. My approach in the book is to focus on what I take to be a common feature in a number of theories. For the purpose of argument I call the theories the postmodern sublime, though in practice there are several—maybe dozens—of definitions of postmodern sublimes.

For me the salient feature of a postmodern sublime is the focus on the fourth stage—the recuperation of pleasure through the awareness of a supersensible faculty of reason—or, using the other schemata, the rejection of the region beyond the limen. Essentially the plurality of postmodern accounts deny the fourth step, turning the sublime into an experience of epistemological defeat. It is in that context that I present Krauss’s reading of Agnes Martin’s paintings.

[For images of Martin’s paintings I recommend exhibition catalogues and monographs. For a quick idea try

http://www.artcyclopedia.com/artists/martin_agnes.html or

http://contemporaryart.about.com/arts/contemporaryart/cs/agnesmartin/index.htm and click on "Artnet."]

Martin paints pictures that are in a single color, and covered with a grid, usually done with a pencil and ruler. She calls her paintings Starlight, Hill, Blue Flower, Islands, Night Harbor, Mountain, or Lemon Tree-- naturally, her titles have made people think that these are "secretly" landscape paintings; but she always said they were purely abstract.

Krauss uses a phenomenological reading by the critic Kasha Linville, which is still one of the best accounts of what it is like to stand up close to Martin’s paintings. Linville says a viewer will first notice how the thin pencil lines skim over the canvas surface, becoming "dotted or broken."

At that distance, perhaps a foot or two, the paintings are just materials: irregular canvas weave, skipping and stuttering lines. After considering the picture’s fine structure, Linville says a viewer will step back; and that is when something like illusion takes place.

Linville doesn’t see atmosphere "in the spatially illusionistic sense I associate with color field painting"; instead she senses "a non-radiating, impermeable mist."

Notice she does not say Martin has painted a mist: she says "It feels like, rather than looks like, atmosphere," and it "somehow… dematerialize the canvas, making it hazy, velvety."

After taking in the "feeling" of "impermeable mist," Linville says a viewer will naturally want to step back even farther, to try to see the picture all at once, but then "the painting closes down entirely, becoming completely opaque."

In the end, Martin’s paintings are "impermeable, immovable as stone."

[In lectures I draw this as a sequence of three position: up close, where the gesso and pencil lines look flat and hard; a middle distance, where there is a strange sensation of "impermeable mist"; and a distant vantage, where the painting once again seems flat.]

Krauss recounts Linville’s analysis and says it "brackets" the middle-distance view in between two flat hard "walls"—an excellent metaphor for the refusal of the Kantian sublime, and the confinement of the imagination which counts, in my discussion, as one of the form of the limits of representation.

C. Second example from the book: astrophysics

The analysis of Martin’s paintings is an especially clear example of a theory (a variant of the postmodern sublime) being brought to bear to explain how a given pictorial practice (Martin’s paintings) work at the limits of representation. In the book that analysis serves as an introduction and model for others in which the theories and images are very different, but the interests of those making and interpreting the images are structurally similar.

An example from astrophysics:

If the history of twentieth-century images were not biased toward fine art, one of the central works of the century would be the photographs collectively known as the Hubble Deep Field.

It is a picture--or rather, an collection of pictures--of a tiny patch of sky near the Big Dipper. The idea was to point the Hubble Space telescope at a part of the sky far from any bright star, and see what it would record if its cameras were left open for a maximal amount of time.

For over a hundred hours the telescope peered into space, and brought back an astonishing image, full of galaxies fainter than any that had ever been seen. The Hubble Deep Field was made in December 1995. In October 1998, the same was done for a small area of sky in the Southern hemisphere, producing the Deep Field South. Again an apparently blank stretch of sky yielded thousands of faint galaxies.

The deep fields are extremely small parts of the sky, almost at the limit of what I can imagine.

[A web animation does this for the Hubble Deep Field North: http://oposite.stsci.edu/pubinfo/pr/96/01/HDF.html.]

The inhabitants of this distant realm are decidedly strange. Brighter objects, like nearby galaxies with magnitudes of 19 or 20, tend to be elliptical, spiral, or barred: that is, they have predictable, orderly shapes. After magnitude 25, almost one-third of the objects are "peculiar." There are "head-tail galaxies resembling tadpoles," and all kinds of asymmetric objects.

[An example is available in: Sidney van den Bergh, Roberto Abraham, Richard Ellis, Nial Tanvir, Basilio Santiago, and Karl Glazebrook, "A Morphological Catalogue of Galaxies in the Hubble Deep Field," available on the internet at http://xxx.lanl.gov/abs/astro-ph/9604161.]

Easily the most wonderful act of seeing in regard to the HDF-N was the attempt to see beyond the faintest galaxies on the plate--to see something in the black regions between the faint bright spots, at the very end of the visible universe.

It seems the Deep Field images reach up to a region where visibility drops off, named the wall of "dust extinction." What if it were possible to discern something there, among the dark pixels? The telescope was at its limit, but if a better telescope were to be built (and one is already being planned), would it see further? Would the black voids resolve into little lights, and fill with even more distant galaxies? Did the visible universe continue on, or was the HDF-N a view of the very end of the visible universe?

It was immediately clear that the space between the faintest lights in the HDF-N is not a perfect void, because it is not a uniform black. Rather it is a set of pixels with slightly different colors and luminosities. Color images of the HDF-N have an uncanny resemblance to a deep summer sky, with scudding clouds.

There have been several attempts to see into that hazy darkness. Among them is a paper written in late 1997 by two astronomers in Seoul.

[The paper is Changbom Park and Juhan Kim, "Diffuse Dark and Bright Objects in the Hubble Deep Field," available at http://xxx.lanl.gov/abs/astro-ph/9712039.]

Park and Kim found "tiny spots" of light, "embedded in extended backgrounds," "galaxies in the process of active star formation and merging," "primordial galaxies," and "proto-galactic objects." In addition to such objects they also find smooth, dark shapes, where the "sky" is darker than in surrounding areas; they interpret those as "dark clouds" between the "proto-galactic objects." These dark clouds and proto-galaxies are so faint and so entangled in the background light that they actually cannot be seen by simply looking at the images.

Park and Kim provide an amazing illustration which purports to show the objects.

[Refer to their Fig. 3, available in the paper cited above.]

Each row here is a single object, seen in three different wavelengths. The graphs along the right side plot the surface brightness of each object across the field, measured in arcseconds. The top three objects are considered bright, and the bottom three dark (notice the negative surface brightness, -SB).

But in what sense are these objects really there?

I don’t have trouble believing that a mathematical analysis can discover objects that my eye can’t, but the authors also offer a phenomenological description. "Most of these objects," they say, "are sprinkled with many noise-like glares and show extended backgrounds. Some of them are highly elongated with emission of connections and seem to be undergoing merging."

Certainly, if these are discrete objects, then they are irregular and multicolored: but nothing else about them, I think, is visible without mathematical help.

[In lectures I picture this as a variant on the three-part schema used to visualize the Kantian mathematical sublime. Here the fuzzy galaxies are on the left, unknown objects on the right, and a line, the limit of resolution, in the middle.]

This case is partly analogous to the first example, of Martin’s paintings; but here the limit of representation is not provided by phenomenological criticism, but by the mathematics itself. Park and Kim’s mathematical analyses prove the existence of objects in the plates that cannot be seen with the unaided eye. In effect they provide a bracket (see the sketch above) which is partly in the realm of the visible, and partly beyond it.

As in Martin, these images push at the limits of seeing, and their accompanying critical (mathematical) apparatus provides the limits of what is construed as visibility. The salient feature of the parallel is that mathematics has substituted for critical and philosophical analysis. It is mathematics that provides the limit, the limen or limes.

D. Third example: transgenic art.

My final example—for the purposes of this abbreviated version of the lecture—is an artist and geneticist named Joe Davis who works at MIT.

[See for example the essay at:

http://www-tech.mit.edu/V120/N26/bioartists.26f.html; also Steve Nadis, "Science for Art’s Sake," Nature 407 (12 October 2000): 668-70, and "Microvenus," Art Journal 55 no. 1 (1996):. 70-74.]

He been working on a way to put an image of a galaxy into a mouse, where it would become part of the mouse’s genetic code and be passed on to future generations. Davis began with an infrared picture of the Milky Way galaxy.

[The image is available on the internet at http://www.gsfc.nasa.gov/astro/cobe/cobe_home.html.]

<

He displayed the picture in digital form, using hexadecimal notation, and then devised a series of codes that change the sixteen decimals of hexadecimal notation into DNA base pairs.

The challenge was to create a code that would accurately transcribe the image and still be viable as DNA in a living cell. His final answer is a "supercode" that reads amino acids from a base-20 code. The "supercode" is a biologically viable and stable form of the digitized image of the Milky Way. Once it is sequenced and inserted into a mouse, it can be passed on from generation to generation.

It could conceivably even be inherited by species descended from the mice, and barring significant random mutations Davis expects it to remain stable for "geological time." He thinks of his project as a variations on "pictures" that animals and plants already have encoded in their genetic material. He speculates that his galaxy picture may be "redundant," because "mice and other living organisms already inherently possess subtle ‘maps’ of the local cosmos," in the form of responses to diurnal and annual rhythms. (There are "pictures" hidden in many microscopic forms, and who knows?--perhaps mice even have an awareness of the Milky Way.)

Whether or not Davis is right, his "picture" of the Milky Way is one of the most marvelously distorted pictures I know: it is further in the realm of the unrepresentable than any astronomical image I know. Its "bracket"—and here I’m stretching the notion to include complex forms of correspondence—is the complex series of codes and supercodes, and the result is wholly beyond human vision, though it can be recuperated for vision at any time.

D. Summary and conclusion

These three examples are intended to evoke the book and give a sense of the kind of argument it involves.

I will end with a summary and a moral.

The summary goes like this: I argue in the book that some painters, photographers, particle physicists, electron microscopists, cosmologists, "genetic artists," and astrophysicists, can all be understood in terms of a common concern. They are each interested in images only insofar as the images can serve as pointers to forms than cannot currently be represented. In that sense the book is about a kind of interrogation of visibility that is articulated in different forms in different fields. I would claim that some of the most interesting images made in each of those fields in the past twenty-five or thirty fields are concerned primarily with the places and reasons why naturalistic representation breaks down.

And the moral is for anyone who works with science-art connections:

Often, I think, it’s not a matter of science taking from art, or vice versa: we have a common image culture that entails the speaking different languages simultaneously. Acknowledging that can only make things more interesting and more difficult.