One of the interests of studying scientific practice from the vantage point of literature is that it adds another series of mediations to the many that historians, sociologists, and philosophers are unfolding when they analyze laboratories, instruments, controversies, disciplines or institutions. 1 Scientific texts, to be sure, have no privilege,but neither are they inferior to the many sources we have forunderstanding science. Indeed, when properly studied, they offer a convenient model to show how many mediations can be retrieved from the scientist’s own practice. A scientific text is not only a more or less transparent medium to convey information to the author’s scientific colleagues, nor is it only a document to help historians, psychologists, or sociologists retrieve the state of mind of its author or the context in which it has been written. As many decades of literary theory have helped us to see, texts are a little bit less and a good deal more than information and document. They build a world of their own that can be studied as such in relative and provisional isolation from the other aspects. They are localized events, with their own matter and their own practice.
Although this point is taken for granted in many studies of fictional literature, 2 it has not been rendered as acceptable in the [End Page 129] study of scientific literature—in spite of Francoise Bastide’s work 3 and the many studies of scientific texts coming either from the humanities 4 or from the social sciences. 5 The same people who would not hesitate to say that laboratory practices, or citations, or instruments, or institutions, or controversies, or conversations, or rhetoric can be studied in relative and provisional isolation from the rest, would balk at allowing a full-blown study of a single text. They would insist that a text is not an island, and that it cannot be understood without its context of use, the reader’s response, and the wider social picture.
I understand these worries. They stem largely from the fact that, as Thomas Pavel has so clearly shown, many literary theories are ontologically weak, which makes them of little use for texts that insist on being referential and not fictional. 6 But as I have shown elsewhere in the case of Einstein, even this question of the referent may be tackled by semiotic theory. 7 The main advantage of practicing some sort of semiotics on scientific texts lies in the very limitation of the theory. By bracketing out the question of the referent (there exist only internal referents generated by the text itself) and by bracketing out the question of the locutor (authors and readers are built into the texts and may not relate to any authors and readers in the flesh), we let the texts deploy their own categories. Their world-making activity is no longer squeezed in between a referent that it has to grasp and a locutor or a social context from which it emerges. It becomes an event, which has the same activity, the same materiality, the same complexity, the same historicity as any other event.
This provisional independence, both from the referent and from the context, gives a valuable bonus to literary theory in the treatment of nonhumans. In the study of texts there is no a priori distinction to be made between an anthropomorphic actor and a [End Page 130] “physimorphic” or “zoomorphic” one: the same amount of work is required to attribute a role to a human or to a nonhuman character. In a fairy tale identical functions may be fulfilled by a prince, a dwarf, a magic rod, or a fox. This freedom in selecting actors and redistributing properties among them is crucial to understanding scientific practice, and, to my knowledge, no other discipline possesses that freedom. All the others have to start from a “natural” division between human and nonhuman properties. So, eventhough the ontology embedded in literary theory may be flawed, its ability to deal with nonhumans is without a par, and it allows us to go much further in the study of scientific work than do discourse...