This is a text-only version of Chapter Nine of William J. Mitchell's The Reconfigured Eye: Visual Truth in the Post-Photographic Era (Cambridge: MIT Press, 1994): 190-223. It is used here with the permission of the author and MIT Press.
The central metaphor of André Gide's Les Faux-Monnayeurs suggests that, just as economies are destabilized by counterfeit coinage, so the practices of textual production and exchange through which subjects construct their understandings of the world are subverted by the surreptitious interjection of signifying tokens of uncertain value. Gide's character Strouvilhou remarks of "those promissory notes which go by the name of words":
I must own that of all nauseating human emanations, literature is one of those which disgust me most.... We live upon nothing but feelings which have been taken for granted once for all and which the reader imagines he experiences, because he believes everything he sees in print.... And as everyone knows that "bad money drives out good," a man who should offer the public real coins would seem to be defrauding us. In a world in which everyone cheats, it's the honest man who passes for a charlatan. 
Similarly, the photographic falsifier holds up not a mirror to the world but a looking glass through which the observing subject is slyly invited to step, like Alice, into a place where things are different--where facts seem indistinguishable from falsehoods and fictions and where immanent paradox continually threatens to undermine established certainties (figure 9.1). To grasp precisely how this can be so, we must consider not only how photographs and pseudo-photographs are made, but also how they are used--how their potential uses are established, how they are appropriated and exchanged, how they are combined with words and other pictures and made to play roles in narratives, and how they may have the effect of creating beliefs and desires. This strategy finds some precedent among the high modernists in Clement Greenberg's well-known insistence that photographs must be considered essentially in terms of their narrative uses. In 1964 Greenberg argued:
The art in photography is literary art before it is anything else; its triumphs and monuments are historical, anecdotal, reportorial, observational before they are purely pictorial. The photograph has to tell a story if it is to work as art. And it is in choosing and accosting his story, or subject, that the artist-photographer makes the decisions crucial to his art. 
It also draws from the legacy of structuralism--on Roland Barthes's
influential suggestion that the press photograph should not be regarded
as "an isolated structure." It is always, Barthes noted, "in
communication with at least one other structure, namely the text--title,
caption or article--accompanying every press photograph." 
This is essentially an expansion and reworking of some earlier remarks
by Walter Benjamin, who had suggested that, without captions "all
photographic construction must remain bound in coincidences" and who
had gone on to ask, "Will not captions become the essential component
of pictures?" 
Philosophers of a more analytical disposition have sometimes made the similar point that pictures, when combined with labels, can be used like declarative sentences to make assertions which are either true or false. (Ludwig Wittgenstein no doubt encouraged this by insisting, in Tractatus Logico-Philosophicus, that "A picture is a fact." ) John G. Bennett, for example, has written:
Consider a picture postcard. It has on one side a picture of a sunny beach with a large modern hotel in the background. On the face of the postcard there is the written phrase, 'Diddle Beach.' I get the impression that Diddle Beach is a sunny place with a fancy modern hotel and a pleasant beach. Later, when I go out of my way to visit it, I find nothing of the sort; Diddle Beach consists entirely of sharp rocks, the largest building within twenty miles is a rundown gas station, and the sun hasn't shown there in the memory of anyone living. I have been misled....
The picture by itself would not have misled me.... The label 'Diddle Beach' was necessary. On the other hand, the label without the picture could not have given me the false belief.... It was the combination of the label and the picture which led me to have false beliefs. 
This example leads Bennett to conclude that the label is analogous to
a name, that the picture is analogous to a predicate, and that combining
the label and the predicate gives something which can be true or false,
like a sentence. According to this account, it seems, you can perjure yourself
by proffering a picture with a false label (the place pictured is not really
called Diddle Beach) or a label with a false picture (the actual Diddle
Beach does not really have the attributes shown in the picture). Photographs
with false labels are nothing new: they have been used to mislead since
the earliest days of photography--as, for example, when Hippolyte Bayard
photographically portrayed himself as a drowned man in 1839 (figure 9.2).
But labels with pseudo-photographic false pictures have become much easier
to produce in the era of digital imaging.
The most useful place to start, however, is with the traditional distinction between facts and evidence.  A piece of evidence is a fact with significance in some context, a fact that has been pressed into service, used to support some claim or argument. It serves to tell us about something that happened in the past or is happening somewhere else or will happen in the future or is just too small to see or otherwise takes place in a setting to which we have no direct access. Photographs, then, present facts but are frequently used as evidence. Any photograph might be used as evidence of many things, but it only becomes evidence when somebody finds a way to put it to work. In The High Window Raymond Chandler shows us Philip Marlowe as a not-too-bright would-be exegete performing this task--picking out a relationship as significant, finding a way to use it as evidence, then embedding it in a reconstructed chain of events:
There I was holding the photograph and looking at it. And so far as I could see it didn't mean a thing. I knew it had to. I just didn't know why. But I kept on looking at it. And in a little while something was wrong. It was a very small thing, but it was vital. The position of the man's hands, lined against the corner of the wall where it was cut out to make the window frame. The hands were not holding anything, they were not touching anything. It was the inside of the wrists that lined against the angle of the bricks. The hands were in air.
The man was not leaning. He was falling.
This narrative turned the mute facts of the photograph into the telling
evidence that Marlowe needed to crack the case.
In much the same fashion, a famous snapshot of Lee Harvey Oswald holding the rifle used at Dealey Plaza has often been seized upon as evidence demonstrating that Oswald was, indeed, John Kennedy's assassin. But Oliver Stone's film JFK shows this photograph being doctored--suggesting that production of this picture was part of a conspiracy. Here, one aspect of what the picture shows--Oswald's possession of the rifle--is picked out and tellingly embedded in alternative narratives.
Let us begin, then, by considering how photographs and pseudo-photographs may be pressed into service as evidence--how they may be employed within larger signifying structures to report the significant facts (or "facts") about states of affairs that are claimed to have existed and events that are claimed to have taken place, and how such reports are used to convince, to create belief, and to command assent. Let us note, in particular, how the potential uses and misuses of images are modified by various types of interventions in image-production processes. This by no means exhausts the possibilities: the issue might usefully be examined from any of the alternative perspectives opened up by the recent fecundity of literary theory--from before Bakhtin to beyond Barthes. But it's a start.
The photographer is more of a pointer than a painter. Just as the pointing
finger indicates something real out there, so does the pointing camera.
Above all else, a photograph denotes objects, persons, or scenes about
which something may then be said. (To "denote"--literally from
the Latin root--is to "mark out.") But we should not be misled
into assuming, therefore, that a photograph works just like the referring
expression of a sentence: we can easily write, "The present king of
France is bald," but it is beyond the powers of the most intrepid
paparazzi to frame such an individual in their viewfinders. As we have
seen, a photograph can denote only something that exists at definite
spatial and temporal coordinates in the physical world, so a photograph
always tells us--as the footprint on the beach told Robinson Crusoe--that
something was actually out there. Vicki Goldberg has noted that "bearing
witness is what photographs do best; the fact that what is represented
on paper undeniably existed, if only for a moment, is the ultimate source
of the medium's extraordinary powers of persuasion." 
Conversely, if a photograph of a bald individual is captioned "The
present king of France," the label of a nonexistent thing is being
falsely applied to a real person.
This presumptive anchorage of the photograph to the real provides an opportunity for photographic fakers to take advantage of us. They can subvert ontology by assembling available image fragments into pseudo-photographs that convincingly match accepted conceptions of what some nonexistent thing would be like--much as sixteenth-century entrepreneurs manufactured specimens of mermen and mermaids, furry fish, sea bishops, unicorn horns, and griffin claws and their nineteenth-century counterparts produced grotesque "medieval" torture devices and sinister-looking chastity belts to satisfy expectations aroused by gothic tales.  (Conversely, when a specimen of the unlikely looking platypus was first carried back to Europe, it was widely suspected of being an assembled fake. Nobody thought that anything would look like that!) The assemblage--depicting, say, a UFO, a Loch Ness monster, or a street in Atlantis--is proffered as evidence that the thing denoted by the label was witnessed by the photographer. If we can be fooled into thinking that the assemblage is a photograph, then the presumption that photographs can show only things that exist will do the rest. These sorts of images function as pseudo-acheiropoietoi--ersatz relics used to create belief that something existed on earth. 
A variant of this game is use of photographic evidence--claimed to be of recent date--to create the belief that something still exists. In the long and bitter aftermath of America's defeat in the Vietnam War, for example, fake photographs were used by those with a propaganda or blackmail interest in the situation to perpetuate the pathetic myth of the continued existence of American prisoners in the jungles of Southeast Asia--prisoners waiting to be rescued and returned to their loved ones by the likes of Rambo.  There was a sadly symmetrical trade in doubtful relics: photographs with fake provenances were produced to demonstrate that prisoners still remained, and bones--sometimes of animals if sufficient human ones were not available--were proffered by the Vietnamese to show that there were indeed no prisoners left.
It matters little whether the image fragments are deceptively assembled before or after the moment of exposure. A particularly famous example of assembly before exposure was used to hoodwink sir Arthur Conan Doyle--the creator of Sherlock Holmes, masterly interpreter of physical evidence--in 1917.  Conan Doyle was a convinced spiritualist, had faith in the existence of fairies, and thought he knew what they would be like. Two girls, Elsie Wright and Frances Griffiths, produced a doctored photograph showing a group of fairies buzzing like big blowflies around the head of a child (figure 9.3). Even though the fake was quite crude, Conan Doyle welcomed it as confirming evidence. He wanted to believe, and the Wright girls had the right stuff to convince him. In fact, the girls used cutout fairy figures from a children's book, held in place by hatpins.
There is, then, a symmetry of uses. Straightforward photographs can be used legitimately and effectively to show that things exist. But mislabeled or manipulated photographs can be used to create illusions of existence.
Walt Whitman, it seems, perpetrated a deception similar to that of the
Wright girls.  A studio portrait
of 1883, which was later used as a frontispiece to Leaves of Grass,
shows a butterfly dramatically perched on his index finger (figure 9.4).
He was not, of course, attempting to demonstrate that butterflies exist,
but making a self-serving point about his own nature. "I've always
had the knack of attracting birds and butterflies and other wild critters,"
Whitman told the historian William Roscoe Thayer. Skeptically, Thayer later
commented: "How it happened that a butterfly should have been waiting
in the studio on the chance that Walt might drop in to be photographed,
or why Walt should be clad in a thick cardigan jacket on any day when butterflies
would have been disporting themselves in the fields, I have never been
able to explain." A later biographer, Esther Shephard, reported that
Whitman's memorabilia contained "a small cardboard butterfly with
a loop of fine wire attached, by means of which it could be fastened to
a finger." This famous photograph is certainly not worthless as evidence:
it told the truth--at least about some things--but it did not restrict
itself to nothing but the truth.
The spirit photographers who flourished in the early decades of the twentieth century employed a different insertion technique.  Their strategy was to preexpose parts of a negative plate with images of a client's deceased loved ones, then to use that same plate in a portrait sitting. When the plate was developed, the picture showed ghostly spirit images (figure 9.5). The spirits had been invisibly there, it seemed, but the eye of the camera had revealed them.
Digital dissemblers, of course, need have no recourse to cardboard fairies or butterflies or to double exposures: they can scan the images that they want to insert and fix these apocrypha in place without the use of hatpins or wire loops. On October 27, 1989, for example, Newsday's cover photo showed eighteen Grumman F-14 fighter jets taking off in formation; but it never occurred: the photograph was manufactured by copying one of a single jet landing, rotating it to point the nose upward, and repeatedly pasting it into the scene (figure 9.6).  A similar technique can be applied to "complete" unfinished or ruined buildings--to produce electronic Potemkin villages by cutting and pasting repetitive facade elements.
Accompanying text can put these sorts of images to use in several different ways. It can falsely assert, or encourage the natural assumption, that the depicted event took place or that the depicted state of affairs did exist. It can enframe the content as that construction dear to philosophers, a counterfactual conditional, such as "If a butterfly had landed on my finger, then this is how I would have looked."  Or it can make a prediction, such as "When the building is completed it will look like this." In other words, it can present a possible world rather than the actual one. Photographs can present only the actual world, but constructed pseudo-photographs can--like naturalistic novels or carefully staged film scenes--present possible worlds as if they were actual.
The converse way to fake the photographic evidence is to efface something.
The result is an image that tells the truth up to a point, but not the
whole truth. For example, one of the most notorious photographs of the
twentieth century exists in two versions (figure 9.7). The first, taken
on May 5, 1920, shows a dramatically posed Lenin addressing a meeting with
the conspicuous figure of Trotsky at his side. In the second, Trotsky is
absent--deleted from the image as he was in general from Stalinist history.
 Those who want to rewrite political
narratives know such strategies well--hence also the removal of Alexander
Dubcek from a 1968 photograph of Czech leaders outside Saint Vitus Church
in Prague, and the removal of the Gang of Four from a photograph of Chinese
leaders at Mao's funeral in Tiananmen Square in 1976. 
(Similar effects can be achieved by selectively excising portions of audio
tapes, as in the notorious Nixon Watergate tapes.) The photograph's metonymic
power-- its capacity to stand for a larger world outside the frame and
to suggest a larger narrative that embraces the moment of exposure--is
being exploited here. Modifying what the photograph explicitly shows has
the more important effect of changing what it implicitly constructs.
On another famous occasion, the wife of a British newspaper proprietor was offended by a photograph of a prize bull and caused its testicles to be erased prior to publication. The owner of the bull, furious at this misrepresentation of his animal's capacities, sued. (Presumably, if the photograph had been labeled with the name of some other bull, he would not have found reason to complain.) For not dissimilar reasons (and perhaps to similar indignation), Rolling Stone magazine electronically deleted a shoulder holster and pistol from the portrait of a macho television actor for its March 28, 1985, cover. The holster could just as easily been removed before the moment of exposure to produce an image with exactly the same content, so the difference between directorial and electronic manipulation seems theoretically insignificant in this case. (Of course it is significant to the bull.)
Aesthetically inconvenient elements have frequently been excised from architectural photographs. Some of the most influential plates in Le Corbusier's 1923 modernist tract Vers une architecture, for example, portrayed the dramatically pure, unadorned, geometric forms of North American grain elevators: they compellingly illustrated Corbusier's definition of architecture as "the masterly, correct and magnificent play of volumes brought together in light."  But earlier versions of these photographs, published by Walter Gropius in the Jahrbuch des Deutschen Werkbundes of 1913, had shown something quite different. Corbusier removed prominent classical pediments from atop some Canadian silos to leave simple cylinders and in another instance dropped out a splendid classical dome so that what remained was a pure composition of rectangular boxes and square pyramids (figure 9.8). "So much the worse," he insouciantly wrote, "for those who lack imagination!"
Traditionally, photographers have tendentiously effaced and elided, when they wished to do so, through carefully selective framing and cropping and through use of camera angles in which foreground objects occlude unwanted background objects. (For example, Time magazine was accused of digitally altering its striking December 16, 1988, cover image of Reagan, Gorbachev, and Bush to remove the surrounding crowd.  In fact, the photographer had succeeded in getting an unusual angle that excluded the crowd.) This is a game of layering and sightlines, much like that played by stage designers. Digital image processing simply extends the game by allowing easy insertion of new foreground layers after exposure to obscure objects that do not fit the narrative purposes to which the image is being put and that were not or could not be obscured by more conventional means.
Logically, any photograph can be used to show the absence of indefinitely many things-- that Elvis Presley did not accompany Lenin in May 1920, that there was no UFO fly-by at Mao's funeral, and that the posturing actor did not sport a sequined handbag. Absence only becomes interesting as evidence when it conflicts with our presuppositions--that Trotsky was an important Bolshevik who would naturally have accompanied Lenin, that buildings would normally have classical details, or that a stud bull would typically have testicles. When such absences can be made to seem sufficiently plausible, they can change our beliefs.
Just as a declarative sentence has a referring expression and a predicate,
so many photographs that we regard as informative have components which
uniquely identify individuals and other components that assign certain
attributes to those individuals. Conversely, as in traditional pictures
of saints, certain attributes--uniforms, tools of trade, characteristic
haircuts, and so on--may be displayed so that we can identify an individual
who might otherwise be anybody. (The problem with saints is that we usually
have no idea of what their faces were actually like, so we have to rely
on stereotyped attributes for visual identification.) In either of these
cases, a photographic manipulator can change the meaning and potential
uses of an image as evidence by adding, deleting, or interchanging identifying
elements. Thus, for example, a nineteenth-century scandal was created by
a photograph manipulated to show the pope improbably wearing the insignia
of a freemason. 
Since people are most readily identified by their facial features, such shifts can effectively be accomplished by substituting well-known heads onto whatever available bodies have the appropriate attributes--a practice which reduces those bodies to signifying commodities. (It is not too much different from inserting a portrait photograph into a false passport.) There is a well-known precedent for this from Roman antiquity: the emperor Hadrian removed the head of Nero from an imperially garbed statue and replaced it with another more to his liking. Similarly, Pierre Lombart's famous Headless Horseman prints of an equestrian figure exist in three states--one with no head (which we may read as a kind of visual aposiopesis), one with the head of Oliver Cromwell, and one with the head of Charles I (figure 9.9).  Political power is a passing thing, but statues and printing plates endure.
In the prewar and Civil War era in the United States, which coincided with the first decades of photography, images of various politicians were recycled with the highly recognizable head of Abraham Lincoln. At different times, Lincoln's likeness assumed the engraved bodies of Alexander Hamilton, Martin Van Buren, and the Southern politician John C. Calhoun (figure 9.10).  Henry S. Sadd's elaborate 1852 mezzotint engraving Union showed Calhoun in a prominent central position, but a later version (ca. 1861), replaced his face with that of an unbearded Lincoln and introduced the portraits of half a dozen lesser pro-Union figures as well (figure 9.11)--a shrewd marketing move under the circumstances.  After Lincoln was assassinated, new pictures of the dead president were created by pasting his head, from a famous Mathew Brady photograph (the one engraved on the five-dollar bill), onto an appropriately statesmanlike full-length portrait of Calhoun (figure 9.12). (A new text is also substituted on the papers under the figure's hand: "Strict Constitution" becomes "Constitution," "Free Trade" becomes "Union," and "The Sovereignty of the States" becomes "Proclamation of Freedom.") Lincoln's head had to be mirrored in order to make it fit; the deception was discovered when somebody noticed that the late president's highly recognizable mole was on the wrong side.
In our own era, interchangeable heads have mostly been used pornographically rather than politically--to present women as desirable boy toys rather than to incorporate men in the emblematic raiments of political power. The integral female subject is reconstructed as stereotyped sexual object. The magazine TV Guide, for example, was caught embodying the not-so-svelte talk show host Oprah Winfrey as the more lithe actress Ann-Margret to produce a cover picture (figure 9.13). Use of a body double in this way is quicker than a diet and cheaper than a health club membership, but a smitten fan would certainly be disappointed if he sought out this alluring composite. (This deception was discovered when Ann-Margret's husband noticed a familiar ring on one of the reallocated fingers.) The distributors of the film Pretty Woman put together a similarly rebuilt fantasy figure for a publicity poster by supplanting the body of the film's star, Julia Roberts, with that of an anonymous, seductively clad and posed model. This game can, of course, be recast as one of ambiguity, androgyny, and scrambled sexual difference; in 1991, in response to a Vanity Fair cover photograph of the unmistakably pregnant actress Demi Moore, Spy magazine sniggeringly substituted the face of the father, actor Bruce Willis. The composite became a rhetorical figure--an oxymoron.
There is an old but still-robust tradition of maliciously recapitating photographed bodies to show public personages compromised, naked, in bad company, or just having too good a time. In 1861 an attempt was made to discredit the exiled Queen of Naples by producing a composite showing her cavorting naked with the pope and cardinals.  More recently, the head of the photogenic Princess Di has been grafted onto nude bodies to produce the pictures that tabloid magazines wanted to get but couldn't. At the time of the Third French Republic, a photographer named Eugene Appert fanned anti-Communard feeling by posing models in dramatic tableaux and then substituting heads of Communards and their victims. 
A more subtle and increasingly common form of transfiguration by attribute substitution is alteration of eye, hair, or skin color. The color of a model's eyes can be changed by fitting tinted contact lenses before a shot is taken or by color correction afterward. Similarly, hair color and style can be changed by a prephotographic visit to the hairdresser or by postexposure manipulation of the digital image: in 1991 the Japanese weekly Shukan Bunshun published manipulated photographs of the unmarried Crown Prince Naruhito with ten alternative hairstyles, together with an article entitled "Hairstyle Remodeling Plan," in which a hundred young women were asked which style they preferred (figure 9.14).  (It is reported that the Imperial Household Agency was not amused.) And a pale-faced model can get a tan on the beach or in the digital darkroom. When images substitute for bodily presence, digital makeovers may serve just as well as visits to hairdressers and makeup artists.
Props and scenery can also be recolored for cosmetic effect: the Orange County Register once color corrected an implausibly colored swimming pool to a more normal shade of blue: unfortunately, the accompanying story was about how vandals had dyed the pool red. Rather less trivially, in a 1970 budget request to Congress, NASA presented a colorized version of black-and-white film footage from the Apollo moon mission: it showed not the recorded colors of the scene of this historic event, but somebody's reconstruction of those colors. 
Sometimes digital disguises are put on to feign the fit of a photograph to a story. In 1989 the Washington producers of ABC News photographed a staged scene of one man passing a briefcase to another, then electronically manipulated the image so that one of the anonymous protagonists appeared to be Felix Bloch, a diplomat who had been accused of espionage.  An ABC News spokesperson embarrassed by the revelation of this deception--later said that the image had been intended as a "simulation" and that failure to identify it as such had been an oversight.
The converse process, of disguising recognizable facial features for the sake of anonymity-- like a movie star putting on dark glasses--has also surfaced from time to time. A former photographer for Hearst's New York Journal, for example, has described how file photographs were adapted to serve new narrative purposes in this way. Photographers would "dig up a real photograph of, say, John L. Sullivan, remove his ferocious moustache, paint a General Grant beard across his massive chin, and send it to the engravers as a legitimate picture of an unidentified body in a foul murder."  The result is an image that passes for a photograph of somebody--but not anybody that we know.
Another informative type of photograph (one that we might regard as
newsworthy) shows an event--a meeting, a tryst, or a murder perhaps--taking
place. Such a photograph locates certain individuals at a specific location
at some particular moment. We presume that photographed events--unlike
drawn or painted ones--must really have happened, and we value photographs
showing significant events as part of the historical record. Alfred Hitchcock's
Rear Window is grounded on this premise: Jimmy Stewart, the immobilized
photographer, takes pictures that add up to evidence of a murder and is
stalked by the sinister Raymond Burr--the killer who realizes that the
testimony of photographs can put him away. And Antonioni's photographer
in Blow-Up is convinced by the evidence of his enlargements that
somebody was killed--until both the photographs and the body disappear.
News photographs showing us young women perched illicitly on middle-aged presidential candidates' knees and amateur snapshots of family gatherings and posing vacationers have the same power to demonstrate that significant events indeed transpired. Photofinish pictures provide evidence acceptable to punters and bookmakers of the order in which horses crossed the line. Videotapes of politicians taking bribes in motel rooms have been used to put these malefactors in jail. Terrorists and kidnappers provide photographs of hostages to demonstrate that they really do have them. And the effectiveness of blackmail photographs depends entirely on the victim's concession that they are undeniable. (Blackmail drawings, however, are easily deniable and would not work.)
This effect of displaced witnessing of being confronted with real events in the lives of real people--can make photographs disturbing in a way that the most horrific drawings can never be. Eddie Adams's famous 1968 photograph of General Loan shooting a Viet Cong suspect through the head created such outrage because we knew that we were not seeing allegory or agitprop but the actual slaughter of a helpless human being. Weegee's flashlit pictures of murder victims at the scenes of the crimes are numbing in their matter-of-fact inscriptions of squalid death. Brassai's photographs of prostitutes turn us into voyeurs. Execution photographs and snuff films (the privatized equivalent of this traditional state product) have the uniquely malevolent power to transform their viewers into complicitous witnesses. A 1991 videotape of white Los Angeles police officers brutally beating black motorist Rodney King provided such irrefutable evidence of what really happened that, when a jury with no blacks failed to convict the officers on criminal charges, Los Angeles exploded into days of violent rioting; most people believed their eyes, not what the legal system tried to tell them. 
But events that never occurred can also be shown--sometimes to similarly telling effect-- by bringing together, within a frame, photographic images taken at different times in different places. The early pictorialist photographers--notably Oscar G. Reijlander and Henry Peach Robinson--exploited this possibility to produce wannabe-paintings in the traditional modes of allegory, sentimental narratives, and history pieces. Reijlander's The Two Ways of Life convoked posed figures from thirty different negatives to produce an action-packed Victorian crowd scene that today makes a life of gambling, liquor, and lust seem a good more interesting than the proffered alternative of religion, charity, and industry. And in 1890 a photograph of painter Henri de Toulouse-Lautrec--somehow anticipating poststructuralist fetishization of reproducibility and reflexivity--pictured Toulouse-Lautrec portraying Toulouse-Lautrec (figure 9.15).
Visual misrepresentations of what happened have obvious political uses. During the McCarthy era a crude cut-and-paste fake photograph showing US Senator Millard Tydings in a meeting with the communist leader Earl Browder (the one with the Joe Stalin moustache) probably cost the senator his seat (figure 9.16).  (Stalin's propagandists found it useful to take a widely known communist out of a picture, and McCarthy's found that it served their purposes to put one in.) Since the collapse of the Soviet Union, of course, caricature communists have become much less effective as emblems of unreliability; male politicians are now more easily discredited through juxtaposition with stereotypical bimbo figures.
One of John Heartfield's most compelling political collages--Like Brother, Like Murderer (1933)--simply conjoins a portrait of brownshirt leader Julius Streicher with a picture of a bloodied murder victim from the Stuttgart police archives and an Italian officer holding a dagger (figure 9.17). An alternative version, without the Italian officer, was published under the title A Pan-German. Within this directorial mode, it is actually of only minor technical interest whether genuine image fragments are recombined or impersonators are posed for a real photograph. And it makes little difference whether a backdrop is painted on canvas in the photographer's studio before the exposure is made or electronically matted in afterward.
Convincing assemblages of this sort are easy to produce by digital image manipulation. For a 1989 article on the film Rain Man, for example, the picture editors of Newsweek separately photographed actors Dustin Hoffman and Tom Cruise one in New York and the other in Hawaii--then produced a composite showing them beaming together, as if sharing a joke.  Similarly, in 1990 The New York Times demonstrated this principle by showing a composite in which Rambo and Groucho Marx appear at the Yalta Conference (refer back to figure 9.1).  And in 1991 the Times illustrated the steps involved in combining a photograph of Iraqi President Saddam Hussein taken in Baghdad on July 8 with a photograph of US Secretary of State James Baker taken in Kuala Lumpur on July 23 to produce the appearance of a cordial meeting.  More prosaically, formal group pictures of boards of directors for annual reports, groups of political candidates for magazine covers, and the like are now fairly routinely put together with image-processing systems from photographs taken on different occasions. (In much the same fashion, record producers tape musicians on separate tracks and assemble musical performances that never took place. The practice is so prevalent that recordings of actual performances have to be labeled "live" or "in concert." Recorded sounds can also be combined with recorded images, as when videos of performers--such as the pop group Milli Vanilli--who look better than they sing are lip-synched with audio tracks from others who sing better than they look. Rap music goes even further, with its practice of "sampling"--freely appropriating and recombining fragments of existing recordings. )
Similar techniques can be used to produce anachronistic assemblages of filmed characters. In the 1983 Woody Allen film Zelig, for example, Allen's character convincingly interacts with historical figures from 1920s black-and-white newsreels. And in 1991 a Diet Coke television commercial seamlessly combined colorized old film footage of Humphrey Bogart, James Cagney, and Louis Armstrong with modern footage to yield a spectacular nightclub scene in which these long-dead characters mingle and converse with the living (figure 8.1). 
Of course it does not suffice merely to surround characters by the same frame: their actions or attitudes must be connected visually in a way that serves some narrative purpose. The picture must have a point. Thus, for example, Heartfield's Like Brother, Like Murderer achieves its barbed clarity by fitting the three characters into an easily understood stereotypical pattern of standing victors and fallen, bloodied victim--one that is found both in Greek vase paintings and in boxing pictures on the sports pages of newspapers. Just as Vladimir Propp  analyzed folk tales as embeddings of characters in stereotyped action sequences-- contracts, seductions, betrayals, and so on--we might analyze successful news pictures in terms of the stereotypical (and therefore recognizable) frozen-action patterns that they instantiate. They are iterated enactments of basic narrative functions, and we might well develop a systematic and quite exhaustive narratology of them.
An interpretation may be suggested or contested by proposing that an image shows enactment of some familiar action sequence. Many people who saw the notorious videotape of Rodney King being beaten by Los Angeles police officers immediately and confidently assimilated it to the ancient narrative of the helpless victim--the powerless member of an underclass being ill-used by his oppressors. But the defense in the criminal trial of those officers successfully constructed it in the minds of the jury as the well-worn racist narrative (long familiar from such instantiations as Birth of a Nation and King Kong) of the big, threatening, irrational black male who had to be brought under control by the forces of law and order.
Digital cut-and-paste rearrangements of the elements of a photograph can transform one action pattern into another, and in so doing dramatically alter the image's meaning--our understanding of what the protagonists are doing. Figure 9.18 illustrates this. In the original photograph Margaret Thatcher and George Bush have their heads inclined toward each other as they engage in what clearly appears to be an amicable chat. But, in the manipulated version, Thatcher's figure has been mirrored, so that the two now turn away from each other and seem to be quarrelling. Conversely, simply moving them closer together makes their conversation seem more intimate.
Sometimes a narrative connection can be established without even placing the protagonists within the same pictorial frame. Textual enframing of separate images may work almost as well. Consider the supermarket tabloid story "5-Ton Rhino Rams Train--and Knocks It OFF the Tracks."  To illustrate this story, it suffices to juxtapose a stock photo of an angry rhinoceros captioned "Tourist captured this head-on shot of charging rhino on film" with another one of a train wreck presented as "Rhinoceros 1, train 0! Shocking aftermath of passenger train derailment."
Thus images suitable for appropriation by narratives can be produced in several different ways. The procedure of a traditional photojournalist is to watch for the emergence of significant patterns out of the flux of continuous action and to expose precisely at the decisive moment. That of the directorial photographer-- such as Edward Curtis in producing some of his ennobled scenes of Native American life-- is to marshall characters, props, and scenery into the desired relationships.  And that of the electronic collagist is to assemble such patterns from available image fragments: disparate found objects may encounter each other on the digital dissecting table.
In all these cases of image manipulation we feel that somebody is not
quite playing by the rules, that we are somehow being cheated, that invalid
reports are being given.  But
precisely how? Speech-act theorists have thrown some light on the matter
by pointing out that the transaction of valid reporting, stating, or asserting
(like other speech acts and analogous nonverbal or partially verbal acts
of communication) is defined by constitutive rules. 
Most obviously, the maker of the report must be committed to the truth
of the expressed proposition and cannot simultaneously hold some contradictory
proposition to be true. Then there is the preparatory rule: the maker of
the report must be authorized by an ability to provide evidence or arguments
for the truth of the expressed proposition. (Thus responsible historians
and newspaper reporters check sources and do not just guess at what happened
or make things up.) Next, there is the informativeness rule: the expressed
proposition must not, in the relevant context, be obviously true to both
the speaker and the hearer. And finally, there is the sincerity rule: the
maker of the report must believe in the truth of the expressed proposition
or else be open to charges of lying, prevarication, or perjury.
A successful act of reporting conveys information that the reporter appropriately believes to be true and the addressee accepts as credible. But there are various cases of failure due to violation of the constitutive rules: makers of reports may not sincerely believe what they say or be committed to its truth, their beliefs may not be firmly grounded in evidence, or they may not be conveying anything that is not already obvious to the addressee. Invalid reporting conveys inaccuracies or untruths or does not make coherent sense. It may not be believed or it may create false beliefs or it may have no effect whatsoever on the addressee's beliefs.
Under normal circumstances, presentation of an unmanipulated photograph to show that something was there, that some state of affairs existed, or that some event took place is playing by the rules of valid reporting. But presentation of photorealistic synthesized images or pseudo-photographic assemblages (photographs with additions, deletions, substitutions, or rearrangements) as straightforward photographs usually is not, and the resulting transaction then becomes something other than valid reporting--either falsehood or fiction. (There are some exceptions: a combination print carefully made from two negatives to retain details of a cloudy sky might reasonably be regarded as truer to nature than a conventional print made from a single negative in which the sky is completely bleached out by overexposure.)
In the simplest case, photographic forgers and manipulators violate the sincerity rule by using their productions in the context of enframing narratives to convey knowingly false information--lies: the propagandists who removed Trotsky from Lenin's podium or who juxtaposed Senator Tydings and Earl Browder surely did not believe that what they were showing really happened, but they wanted us to. Successful use of pseudo-photographs in this way has the effect of producing or confirming false beliefs--as when the Wright girls convinced Conan Doyle of the existence of fairies.
Heartfield's AIZ cover image Like Brother, Like Murderer, though, is subtly and importantly different. Julius Streicher may or may not actually have stood over a bloodied corpse in a Stuttgart street, but (we may reasonably presume) Heartfield was not committed to the belief that Streicher did, and he was not prepared to produce any evidence that such an event took place. However, we cannot accuse Heartfield of insincerity: he was not reporting anything at all, he was not (unlike the National Geographic editors who shifted the pyramids for their cover) being cavalier with the truth, and he was certainly not trying to take anybody in. This powerful picture is fiction: Heartfield pretends to show us something that took place, and we recognize that it is a pretense.  We understand that the artist is projecting a possible world, not reporting on the actual one. Like Sir Philip Sydney's poet, "he nothing affirms, and therefore never lieth." We take his work as a realization of the imaginary rather than an image of the real. It succeeds in its purpose not by creating false belief about a particular event in a Stuttgart street, but by engaging us and by compellingly suggesting a general way of understanding Streicher's character and political role.
To take a work as fiction in this way, we must somehow understand that what we are seeing is just a picture, that it is not being used to report on an actual scene or event--even though it may look as if it could be. We must, in other words, appreciate that the constitutive rules governing valid reporting are suspended. If this suspension of the rules is not signaled with sufficient clarity, or if it is deliberately fudged, then we may justifiably feel deceived-- that fiction has slipped into falsehood. Oliver Stone's 1991 film on the Kennedy assassination, JFK, for example, was vigorously attacked by many press critics because it combined actual newsreel footage with simulated footage and therefore, they claimed, did not play fair. A typically apoplectic response was that of Richard Christiansen, who complained in the Chicago Tribune that the film tries to "persuade its audience that because certain incidents are shot in grainy black-and-white newsreel style, these incidents did, in fact, happen." He concluded that JFK "tries to make its viewers believe that speculation is truth and that fiction is verity."  The basic point is that if we do not know when the rules of reporting are in effect, we cannot know what to believe.
There are many ways to intimate that the rules of reporting are not in effect. A framing narrative may simply tell us that this is the case by specifying that an image is a "simulation" or "reenactment." Or the context may signal it: according to widely accepted ethical conventions, we can expect to see visual fictions when we go to the movies and to find them on the covers of magazines or in art galleries, but not to encounter them in the pages of respectable newspapers. (Oliver Stone could reasonably claim that JFK--made with actors and shown in movie theaters--was using realistic fiction to make a valid political point in much the same fashion as Heartfield.) Even the image itself may signal that it is not actually a straightforward photographic report, for example by incorporating obviously hand-drawn lines or by displaying conspicuous inconsistencies of perspective or shading.
Sometimes just the same image can be used, in different contexts, to make valid reports, to deceive, and to project fictions. Consider the photograph of Walt Whitman and the butterfly. Historians can legitimately use this picture to report on Whitman's appearance at a particular date--to show the length of his beard and the style of his attire. Whitman himself apparently used it, on at least some occasions, to create the false belief that a live butterfly had spontaneously settled on his finger. But Whitman's readers might most reasonably take it as a posed fiction; it is, after all, the frontispiece of a book in which the poet repeatedly constructs and contemplates images of his own psyche's transmutation and resurrection. Why should not the photograph be read as a visual allegory on the same theme? Whatever the photographer's (or Whitman's) original intentions, the image can be appropriated and used for a variety of purposes.
Even completely straightforward photographs can sometimes be used to dissimulate or project fictions. Imagine a photograph of a bloody dictator smilingly dandling an infant on his knee- a work in a propaganda genre that has been popular from Stalin to Saddam Hussein. The photograph may well be unposed and unmanipulated; there is no reason to doubt that dictators are capable of sentimentality about grandchildren. But if the photograph is used as evidence of the dictator's generally benevolent nature, then it is used to create a false belief.
Thus we can begin to see that the truth, falsehood, or fictionality of an image is not simply a congenital property--one conferred at birth by a particular capture or construction process. It is, at best, only partially determined by the maker of the image. It is a matter, as well, of how that image is being used--perhaps by somebody other than the maker--in some particular context.
So far, though, we have taken a rather narrow view of the uses of pictures.
They can actually be used to do many more things than report on events
and states of affairs, tell lies, and project fictions. In a sushi restaurant
with a pictorial menu, for example, you can use pictures to order. (Point,
and say "Bring me one of these." You do not need to know the
word for it.) Architects use pictures to specify work that is to be done,
then to create contracts for execution of that work. Gruesome pictures
can be used to warn or threaten, and pornographic pictures can be used
to elicit sexual response. Photographs in mail-order catalogues are used
to promise goods in return for money. Passport photographs are used to
identify, and photographs of birds in the Audubon Society Field Guide
are used to classify. In the terminology of speech-act theorists, the use
of a picture in a particular act of communication gives that picture a
certain illocutionary force, and the illocutionary force given to a particular
picture may vary from context to context. 
But there are constraints: pictures--like other types of physical artifacts--must be fit for the particular uses to which they are put. They must have properties that assure their functional adequacy. In this they differ from spoken sounds: language, as Saussurean semiotics long ago taught us, puts available speech sounds to arbitrary and conventionally maintained rather than causally constrained uses.
Firstly, image-production processes make certain representational commitments: they record certain kinds of things and not others, and they record some kinds of things more completely and accurately than others. These representational commitments determine in a very obvious way the limits of a resulting image's potential uses in acts of communication. A color photograph or drawing can be used, for example, to report on the hue of an object, to specify that an object should have a certain hue, or to promise that an object will have a certain hue; but a black-and-white drawing or photograph clearly cannot. Photographs and mathematically constructed perspectives are committed to absolute spatial consistency and can be used to report the existence of, specify, or promise precise spatial relationships; but hand sketches are not committed in this way and so cannot be used to show anything more than approximate relationships--except, perhaps, when they are appropriately annotated with dimensions and comments.  Instantaneous photographs are committed to temporal unity, but collages are not. A photofinish picture can be used to report that horses crossed the line in a certain order and at a certain spacing, but not (since it is the trace produced by the horses crossing in front of a slit aperture rather than an instantaneous snapshot) to report that that a particular configuration of horses existed at a specific moment. One-point perspective drawings or photographs of buildings are committed to representing certain elevation planes without foreshortening, but two-point and three-point views are not. Different medical-imaging techniques--CT, ultrasound, PET, MRI, and so on--are committed to acquiring different types of data about bony and soft tissue diseases and physiological activities, and so are used for different diagnostic purposes.
Secondly, a picture used in an act of communication must have the correct type of intentional relationship to its subject matter. (In other words, it must be about the right sort of thing.) A photographic or identikit portrait of Abu Nidal can be used by an immigration officer to identify that individual but not to classify somebody else as a terrorist, and a drawing of a fictional character such as Sherlock Holmes is of no use to an immigration officer at all. Conversely, an ornithologist's generic drawing of a grackle in a field guide can be used by a birdwatcher to classify a particular specimen as a member of that species, but not to identify that bird as the individual Tweety. A sushi-menu photograph is used appropriately to order fresh pieces of the depicted sushi types--certainly not the particular sushi instances that were actually photographed some time ago! A news photograph can be used effectively by a journalist or a historian to make a precise report about an event or a state of affairs at some definite moment in the past, but a handmade drawing works less well in this role. And a synthesized perspective view can be used by an architect to project a possible future world, but a photograph can only show an anterior state of the actual material world.
In general, then, an attempted pictorial act of communication--making a report, carrying out an identification, giving an order, or whatever--will fail if the image used does not have the requisite sort of graphic content or if its intentional positioning is not appropriate. This is misuse of a physical artifact--like attempting to drive nails with a marshmallow. The act will misfire or it will succeed only in some hollow or fraudulent way: your report may be greeted with disbelief, you may wrongly identify somebody (who happens to resemble a fictional illustration) as Sherlock Holmes, or your sushi order may produce an unappetizing surprise.  Just as you must understand the different uses afforded by marshmallows and hammers if you want to perform physical tasks successfully, so you must distinguish between the varying functional capabilities of paintings and drawings, photographs, and digital images produced under various different circumstances in order to use these different types of images felicitously.
Now, the process of photographic image construction is highly standardized, its representational commitments are well known, and the intentional relationships of standard photographs to their subject matter are relatively straightforward and unambiguous. Furthermore, if one accepts the Foucaultian thesis that modern science reversed the scholastic view of an assertion's authority as something derived from its author and substituted the notion that matters of fact are impersonal things, then it becomes obvious that the impersonal process of photography answered to a dominant conception of what the coinage of communication should be. Thus the rules that societies have evolved for acceptable and effective usage of photographs in acts of communication are both clear (if not always explicit) and widely understood. These rules valorize photographs as uniquely reliable and transparent conveyors of visual information and concomitantly structure familiar practices of graphic production and exchange--among them the practices of photojournalism, feature illustration, advertising photography, photo-illustrated fiction, the legal use of photographic evidence, the family snapshot, photographic portraiture, photo identification, medical imaging, and art photography. Photography has established a powerful orthodoxy of graphic communication.
But digital images--as we have seen--have much less standardized production processes than photographs. These processes are less subject to institutional policing of uniformity, offer more opportunities for human intervention, and are far more complex and varied in their range of possible representational commitments. Furthermore, digital images can stand in a wider variety of intentional relationships to the objects that they depict. And, because they are so easily distributed, copied, transformed, and recombined, they can readily be appropriated (or misappropriated) and put to uses for which they were not originally intended. Thus they can be used to yield new forms of understanding, but they can also disturb and disorient by blurring comfortable boundaries and by encouraging transgression of rules on which we have come to rely. Digital imaging technology can provide openings for principled resistance to established social and cultural practices, and at the same time it can create possibilities for cynical subversion of those practices.
The growing circulation of the new graphic currency that digital imaging technology mints is relentlessly destabilizing the old photographic orthodoxy, denaturing the established rules of graphic communication, and disrupting the familiar practices of image production and exchange. This condition demands, with increasing urgency, a fundamental critical reappraisal of the uses to which we put graphic artifacts, the values we therefore assign to them, and the ethical principles that guide our transactions with them.