IN THE RECENT MIKE LEIGH FILM Topsy Turvy, composer Arthur Sullivan tells his librettist, W. S. Gilbert, "I have such respect for your words that I have continually kept down my music in order that they can be heard." Gilbert replies, "I have always subordinated my words to your music." Their deference to each other helped keep their partnership going over a twenty-five year period, but such guarded mutual respect is not temperamentally available to all collaborators. Ford Madox Ford and Joseph Conrad collaborated for eleven years, but didn't speak for the following two. Lou Reed and John Cale's partnership lasted only from 1965 to 1968, and it's surprising it went on for so long, in view of the extent to which the band's "shared intelligence [as Cale puts it in his newly published autobiography] could not offset the antagonism of differing musical notions."

Collaboration is a touchy subject for both the artist and the audience. Despite Barthes's famous eulogy for the author and Foucault's severing of the work from its creator, our view of art is still haunted by the Romantic notion of the artist as an isolated consciousness "expressing itself" through the masterworks it sends forth. To this view, multi-author work looks suspicious. Like those notorious political capitulations that have also been called "collaborations," artistic collaborations have the taint of betrayal and compromise. They make attribution waver, frustrating our desire to know who did what. Thus, when we come across a song credited to "Lennon-McCartney," we satisfy our deep need to know the author by assuming it was written mainly by whomever is singing. Multi-author texts like The Two Noble Kinsmen (written by Shakespeare and John Fletcher) and Macao (directed by Josef von Sternberg and Nicholas Ray) have endured more as puzzles than as works of art. Unacknowledged collaborations -- like the underground labor of blacklisted screenwriters in fifties Hollywood -- carry the stigma of exploitation, or are seen as a breach of trust between author and readers. In 1999, the press reported as a scandal the revelation that mystery writer Dick Francis received significant uncredited help from his wife in writing his successful novels. The division of labor between Vladimir and Vera Nabokov, or Raymond Carver and Gordon Lish, is a subject of perennial interest in certain circles.

But the same tensions that make us uneasy with collaboration underlie artistic production in general. All texts are collaborations -- in ways that range from Barthes's insight that every text is "a tissue of quotations" to the not always trivial fact that, to reach a reader, a text relies on the efforts of several persons: writers, editors, copy editors, webmasters, and accountants. (The function of the last-named is especially crucial, though too often omitted or delayed.) The most indispensable of all collaborations is, however, that between author and reader. No text exists without a reader, and no author totally controls or anticipates how a work will be read. When a text has two authors, each becomes the other's first reader, each writes for the other, and the text becomes a dialogue in which the authors set each other off, lay traps for each other, or pay mutual tribute. To pick some texts, almost at random, is to see a startling variety in the way collaborators work: The dialogue between Gilbert and Sullivan is programmed and directed outwards, toward the abstract perfection of the work. In the film The Life and Death of Colonel Blimp, by Michael Powell and Emeric Pressburger, the creative dialogue turns progressively inward, and duality of perspective thematizes itself. The literary collaboration of Ford and Conrad becomes a dialectical process involving not two voices, but three. Most dialectical of all is the collaboration of Cale and Reed: Each represents what the other lacks and wants to be. Their creative tension escalates into a struggle in which all the ambiguous, painful aspects of collaboration come to the fore.

GILBERT AND SULLIVAN, who banished such darkness, represent a cultural ideal of collaboration: a clearly defined division of labor that produces marvels of delicacy. Sullivan's music enhances Gilbert's words -- and vice versa -- in myriad intricate ways that the listener gives up analyzing as she surrenders to the combination's unique charm. That neither collaborator's solo works are much esteemed may reflect less on the merits of those compositions than on the cultural need to think of Gilbert and Sullivan as a team. And it's the tension between the commercial pressures to keep the collaboration going and Sullivan's desire to distinguish himself as a composer in his own right that drives the first half of Mike Leigh's (historically rather accurate) portrait.

In Leigh's account, a certain "otherness" plays the role of saving Gilbert and Sullivan's relationship from the pull of individuality. When their partnership is in doubt, Gilbert visits a Japanese exhibition. He watches a Kabuki play in which a man severely harangues another, who then leaps up and bashes his companion with a sword. In Leigh's next scene, Gilbert admires a Japanese sword in his study. As the camera moves in for a low-angle close-up of his suddenly inspired face, we hear the prelude to "Behold the Lord High Executioner" from The Mikado. The transition makes the process of creation purely imaginary -- as imaginary as the impetus that drives Gilbert to identify "Japan" as the theme for the next Gilbert and Sullivan work. The Kabuki play isn't just any play, but a play about a dispute between two men, the tension between whom becomes magnetized in a sword like the one in Gilbert's study. Such externalizations are the escape routes that make collaboration possible.

Four years after the events depicted in Leigh's film, Gilbert wrote to Sullivan: "You are an adept in your profession, and I am an adept in mine. If we meet, it must be as master and master -- not as master and servant." But, in Topsy Turvy, they simply stay out of each other's way while each takes care of his own end. The work of writing and staging The Mikado carries them along, negating their differences. In a sense, they never meet: They dwell in the work. In the infrequent moments in the film in which the two are together, Leigh shows them awkwardly making way for each other in a doorway or being guardedly polite. When Gilbert reads his outline of The Mikado aloud to Sullivan, both face in the direction of the camera, as if facing the future stage on which the play will take shape. Without the play, they have no relationship; and both tacitly recognize that the demands of the play are weightier than personal pride. Deferring to these demands, Gilbert and Sullivan find a joint style that subsumes the individual, and thus the tension between them.

PERHAPS LEIGH'S SYMPATHY for Gilbert and Sullivan's predicament is made possible by his own, unusually collaborative, filmmaking process. (Leigh's films are developed through extensive improvisations, during which he and his actors discover the characters, their relationships, and their world. To some extent, even Topsy Turvy was prepared in this way, although the historical record limited the leeway for improvisation.) A sustained body of work signed by two directors constitutes a no less impressive exception to the rule of the film director as demiurge, and the work of Michael Powell and Emeric Pressburger, who together wrote, produced, and directed fifteen films during as many years, is all the more interesting since their films themselves focus on collaborative situations. These situations include artistic creation by a group (The Red Shoes), non-artistic group efforts (One of Our Aircraft Is Missing, A Canterbury Tale, Black Narcissus), and corporate decision-making (the heavenly tribunal in A Matter of Life and Death). But it's their 1943 masterpiece The Life and Death of Colonel Blimp that perhaps comes closest to capturing the spirit of their partnership.