by Umberto Eco
In other words, with the exception of judgments that depend on my direct experience (such as “it’s raining”), all the judgments I can make on the basis of my cultural knowledge depend on information recorded in an encyclopedia, from which I learn both the distance from Earth to the Sun and the fact that Hitler died in a bunker in Berlin. Since I was not there to see if that was true, I trust this information because I have delegated both the information on the Sun and that regarding Hitler to expert scholars.
Moreover, every truth in the encyclopedia is open to revision. If we have a scientifically open mind, we must be prepared to discover new documents one day telling us that Hitler did not die in the bunker but escaped to Argentina, that the body burned in the bunker was not his, that his suicide was invented for propaganda reasons by the Russians or even that the bunker never existed; and in fact even though there is a picture of Churchill sitting where the bunker was, others claim that its location is doubtful. But the fact that Anna Karenina committed suicide by throwing herself under a train cannot and never will be questioned.
Fictional characters have another advantage over historical ones. In history, we are always uncertain about the identity of the Iron Mask, or Kaspar Hauser, and we are not sure if Anastasia Nikolaevna Romanova was murdered with the Russian royal family or whether she survived and was the attractive claimant to the throne later played by Ingrid Bergman. Conversely, when we read Arthur Conan Doyle, we are sure that when Sherlock Holmes refers to Watson, he is always referring to the same person, that there are not two people in London with the same name and the same characteristics, and that the person mentioned in every story will always be the one in A Study in Scarlet who is introduced as Watson for the first time by a fellow named Stamford. It is possible that in some unpublished story Conan Doyle might tell us that Watson had lied about being wounded at the battle of Maiwand during the Afghan War, or about holding a degree in medicine, but in this case, too, the man who would be unmasked as a charlatan would still be the person called Watson by Stamford in A Study in Scarlet.
The problem of the strong identity of fictional characters is an extremely important one. In 2007, a novel by Philippe Doumenc (Contre-enquete sur la morte d’Emma Bovary) was published, in which a police investigation reveals that Madame Bovary did not commit suicide with arsenic but was murdered. It is an amusing little game, but one that becomes interesting only because the readers know that Emma Bovary really did poison herself. If they did not know this irrefutable truth they would not enjoy the counter-story, just as in so-called uchronic novels if we are to appreciate a story in which Napoleon won at Waterloo we must know that it is an encyclopedically accepted truth that he lost.
And so, even though there is an undoubted ontological difference between Hitler and Anna Karenina, I am nonetheless able to underline how novelistic statements, considering the way in which we give credence to them, cite them, and refer to them in our everyday life, are indispensable for the clarification of what we mean by irrefutable truth.
If someone asked us what it means for a statement to be true, we could follow Tarski and reply that the statement “the snow is white” (between quote marks, as a verbal signifier or corresponding proposition) is true if and only if the snow is white. In other words, if the snow is like this or that regardless of the way in which we define it. Nonetheless, while this definition may satisfy logicians, it does not satisfy ordinary people. I would prefer to say that a statement is indubitably true when it is as indubitable as the statement Superman is Clark Kent (and vice-versa).
The Pope and the Dalai Lama can debate for years the truth of a statement such as Jesus Christ is truly the son of God, but if they are sensible (and informed of the facts), they cannot fail to agree that Superman is the same person as Clark Kent. And so, to know if Hitler died in a bunker in Berlin is undoubtedly true, we must check to see if it is as undoubtedly true as Superman is Clark Kent.
Thus the epistemological function of novelistic statements is that they can be used as a litmus test for the irrefutability of all other statements.
But what does it mean when we say it is true that Anna Karenina commits suicide instead of saying only that it is true that in the novel by Tolstoy it says that Anna Karenina commits suicide? Because it is obvious that if people are moved by the fact that Anna Karenina commits suicide, this does not mean at all that they are moved by the fact that Tolstoy wrote that Anna Karenina committed suicide!
This brings us to the reason I began to get interested in these problems. Some time ago, a colleague of mine suggested that I should organize a seminar on why we cry (or, at any rate, feel emotions) about things that happen to fictional characters. At first I told him that it is a matter for psychologists, who have studied the mechanisms of projection and identification. After all, I said, don’t we occasionally dream or even fantasize that a loved one dies, and are then moved to tears? So is there any reason why we cannot be moved by what happens to the female lead in Love Story?
Then I told myself that, no matter how people may be moved by imagining that their beloved has died, after a while they realize that it is not true, and stop crying; in fact, they are relieved, whereas swarms of young romantics killed themselves after weeping over Werther’s suicide, even though they knew both before and after that he was a fictional character. Which means that these readers kept thinking that in some world Werther really killed himself.
It is probable that none of my readers has wept over the misfortunes of Scarlett O’Hara, but no one can tell me that he was unmoved by those of Medea. I have seen sophisticated intellectuals wiping away stealthy tears during the last act of Cyrano de Bergerac, even though they had seen it several times, even though they knew very well how it was going to end and had gone to the theatre only to compare Dépardieu’s Cyrano with Belmondo’s. As a sensitive lady friend of mine once said: “Whenever I see a flag fluttering on the screen, I cry, and I don’t care which country’s flag it is.”
So there is a difference between pretending that the person we love has died, and pretending that Anna Karenina or Madame Bovary dies. In the first case the delusive state lasts almost no time at all, in the second we continue to talk seriously about the misfortunes of the two women, and we write books about them.
In any case, if we look at two versions of Madame Bovary—Umberto Brunelleschi’s 1953 illustrations and the one portrayed by Isabelle Huppert in the 1991 film—at least one of them has nothing directly to do with the novel. It is as if she had left it and gone to live in some other means of expression—the cinema, in this case. Then there is the petit bourgeois Bovary and the risqué Bovary, all the way down to the Bovary used in advertisements for recipes.
Why talk about these versions of the lady? The fact that there are so many types of Bovary who act in ways different from Flaubert’s text means that we are no longer dealing with a character from Flaubert’s world, but a fluctuating character.
Many fictional characters can live outside their own text of origin and act in an area of the universe that is difficult to identify and delimit. Sometimes they even migrate from one text to another, as in novels or films about the sons of the musketeers or Pinocchio the aviator. They no longer belong to the source text. To become a fluctuating character it is not necessary to come from a great work of art, and we should study elsewhere why both Hamlet and Robin Hood, Gargantua and Tintin, Heathcliff and Milady, Leopold Bloom and Superman, Faust and Popeye have all become fluctuating characters, while those who have not include Baron de Charlus, Le grand Meaulnes, Stelio Effrena, and Andrea Sperelli.
A survey showed that 25 percent of Britons believed that Churchill, Gandhi, and Dickens were fictional characters while some percentage (I don’t remember how great) believed that Sherlock Holmes and Eleanor Rigby really did exist. So it is possible to become a fluctuating character for a great many reasons. Disraeli does not fluctuate but Churchill does, Scarlett O’Hara does but the Princesse de Clèves does not. (Nicolas Sarkozy has repeate
dly stated that he could never read that novel by Madame de La Fayette, and my French friends tell me that this fact has proved a shot in the arm for the unfortunate Princesse de Clèves because, out of spite toward Sarkozy, many people have now started to read it.)
Many characters have become so fluctuating that most people know them better through their extratextual avatars than through the text that introduced them in the first place. This is the case with Little Red Riding Hood, for example. Perrault’s version differs from that of the Grimms (in Perrault, the hunter does not come to rescue Riding Hood and her grandmother), but the fluctuating story that mothers tell their children, even if they stick to the ending in the Grimms version, merges the two versions and sometimes deviates from both.
Even the three musketeers are no longer those of Dumas.
Every reader of the Nero Wolfe and Archie Goodwin stories knows that Wolfe lived in Manhattan, in a brownstone at a certain number on West 35th Street. In reality, over the years Rex Stout gave us at least ten different numbers on 35th Street—in which, among other discrepancies, there are no brownstones. But at some point a sort of unspoken pact among Wolfe fans convinced everyone that the right number was 454. On June 22, 1996, the city of New York and The Wolfe Pack put up a bronze plaque at 454 West 35th Street to commemorate the fact that the famous brownstone once stood there.
So Medea, Dido, Don Quixote, Monte Cristo, and Gatsby have become individuals who live outside their original scores and even those who have never read those scores hold that they know them and can make correct statements about them. Some characters, in their wanderings outside the original texts, have become muddled amalgams of one another, such as Philip Marlowe, Sam Spade, and the Rick Blaine of Casablanca fame. (I should point out that Casablanca was originally a play called Everybody Comes to Rick’s.) These characters, having become independent of the texts of origin, live among us in a certain sense; often they inspire our behavior, and sometimes we elect them as criteria of judgment, saying that someone has an Oedipus complex, for example, or a gargantuan appetite, or the jealousy of an Othello, or Hamletic doubts.
So when we assert that it is true that Anna Karenina committed suicide or that Holmes lived in Baker Street, we make statements not on a given score (that is, what a given author wrote) but on a fluctuating creature, whose ontological status appears fairly bizarre, because it should not exist and yet somehow it moves among us and can occupy our thoughts.
Is it possible to fluctuate without existing in a physical form? Are there any objects that do not necessarily exist in physical form? Of course, it would be enough to define as an object every entity we can conceive of and some of whose properties can be predicated. For example, let’s consider the case of the husband and wife, one a history teacher and the other a math teacher, who often talk about both Julius Caesar and the right-angled triangle, but would also like to have a little girl.
So they begin to talk every day not only about Julius Caesar and the right-angled triangle but also about the child they would like to call, as one does, Gessica (with a G of course): how to bring her up, which sports she should play, and how nice it would be if she became a showgirl. So husband and wife are talking (i) about someone who has existed physically but physically no longer exists (Caesar), (ii) about something that some call an ideal object although it is not clear where it exists, unless we assume platonically that there is a world of ideas, and (iii) about someone who hopefully will exist physically but does not yet exist (Gessica). But what happens if, apart from these things, the couple also start talking about freedom and justice?
Freedom and justice are certainly objects of thought, but different from Caesar and Gessica—first and foremost, because they are not so well-defined as Caesar and Gessica, because depending on the various cultures, places, historical periods, and religious beliefs peoples have had contrasting ideas about them, and second, because they are not individuals but concepts. And yet, there are concepts, such as the right-angled triangle, that are better defined than the concept of justice.
Are fictional characters entities like Caesar and Gessica, or the right-angled triangle, or freedom?
They have something in common with Caesar, Gessica, the triangle, and freedom because they are semiotic objects—that is to say, sets of properties expressed by a given term, which a culture recognizes by mutual consent and records in its encyclopedia. Examples of semiotic objects are the right-angled triangle, woman, cat, chair, Milan, Everest, Article Seven of the Constitution, the quality of being a horse. Semiotic objects also include those expressed by proper names, and in this sense semiotic objects include not only Julius Caesar (who now exists for us only as a set of properties) but also, assuming they exist somewhere or other, a John Smith or Joe White who, regardless of the fact they are physical entities, are also, when we mention them with a name, sets of properties (and even without ever having met him we could identify Joe White as the son of Tom, born in Slough, currently a cashier with the Bank of such-and-such, living in such-and-such street, and so on). And since the properties expressed by a proper name include those of having existed in the past or existing at present, or those, recorded in any good encyclopedia, of being a mythological entity or a character in a story, so fictional characters are semiotic objects, too.
The boundaries of many semiotic objects are defined so to speak ab aeterno (for example, the properties whereby a square is recognized as such are not subject to variation or negotiation in approximate terms); others have defined boundaries (for example, the frontiers between two states) but can take losses or additions of property (Italy has remained identifiable as such even though it has been deprived of Zara or Nice); and a great many others are variously fuzzy.
For example, we are able to recognize that German shepherds and chihuahuas are both dogs. They share only some salient properties that for now I will limit myself to define as diagnostics. But this also happens for entities such as the city of Milan—otherwise, those who saw it for the first time, like me, in 1946, half destroyed, without either the Pirelli building or the Torre Velasca, would be unable to identify the modern city as the postwar one. And the same happens, for example, with historical figures, otherwise it would not be possible to make statements such as if Cleopatra’s nose had been a little longer it would have changed the history of Rome (in other words, we can take our idea of Cleopatra and remove some properties from it without this causing us to stop recognizing it as such—and we can imagine counterfactual situations such as what would have happened if Caesar had not been killed on the Ides of March).
Which diagnostic properties must be kept in order to identify something as belonging to the same species or class is an open problem, and in any case we ought to think that a property becomes or remains diagnostic depending on the context or the universe of discourse.
Fictional characters are fluctuating semiotic objects because they can lose some of their properties without losing their identity—so much so that, in the popular imagination, D’Artagnan is a musketeer while we know that in The Three Musketeers he is merely a cadet. If Madame Bovary had lived in Italy rather than in France her story would not have been so different. What then are the truly diagnostic properties of Madame Bovary? One would say that she committed suicide for sentimental reasons. So why can we read a parody like Woody Allen’s The Kugelmass Episode, where the protagonist uses a time machine to take Madame Bovary from Yonville and carry her off to New York to live the good life she had always dreamed of? Only because the context stresses Madame Bovary’s diagnostic property as that of being a provincial petit bourgeois with kitsch passions? In fact, the parody works because the main character hastens to take her back before she commits suicide.
While on the one hand this suicide does not occur in the parody, it remains essential, radically diagnostic, for the identification of Madame Bovary. And this point should be emphasized because, as we shall see at the end, the fascinating thing about fictional characters is that their destiny can
not be changed. We can imagine what would have happened if Napoleon had won at Waterloo, and such a counterfactual exercise would certainly be very interesting, but a story in which Madame Bovary did not kill herself and lived happily ever after somewhere would be insipid to say the least …
Why can we be moved by semiotic objects such as fictional characters? We might answer: for the same reason that many people die for justice or liberty. But there is a difference between being moved by Anna Karenina and being moved by the right-angled triangle. (I believe Pythagoras was the only one to experience the latter.)
We are moved by Anna Karenina because, having signed the fictional pact, we pretend to live in her world as if it were ours, and after a while (as if seized by some mystical rapture, certainly due to some qualities of the narrative) we forget that we are pretending. Not only that, but since we are not signed up to that world—so to speak—or we are not a relevant presence in it, we instinctively try to take the place of some rightful inhabitant or inhabitants of it with whom we share the most aspects.
If we accept this definition of fictional characters, we see that the gods of all mythologies are semiotic objects, like dwarves, fairies, Santa Claus, and the entities of the various religions. Some may think that comparing religious entities with fairies is merely an expression of atheism, but I invite every believer to try a mental experiment: imagine being a Catholic and believing that Jesus really is the son of God. Fine: in this case, Shiva, the Great Spirit of the prairies, and the Exu of Brazilian cults are merely fictional characters. But imagine now being a Hindu: if Shiva really exists somewhere, then it is obvious that the Great Spirit, Exu, and the God of Israel are fictional characters. And so on, until we have to admit that, whatever our religious belief, all religious entities minus one are fictional characters. And therefore, even if we refuse to decide which one defies the general law, we can be certain that 99 percent of religious entities are fictional characters, who, like Madame Bovary or Othello, are usually born from a text. The only difference is that the number of people sharing opinions and beliefs about Shiva is greater than those who know Madame Bovary—but let’s not get involved in quantitative or statistical matters.
