The clearest memory I have of Donald M. Murray is watching him writing at a long white wooden table in his study, which looks out on the New Hampshire woods. Beside his desk is a large framed poster of a small boy sitting on a bed staring at a huge dragon leaning over the railing glowering at him. The poster is captioned, "Donald imagined things." And so he did, as he addressed the problems writers face each time they confront a new assignment. During the summer of 1981, as I listened to him daily recording his thoughts aloud as he worked on two articles, a short story, and an editorial, I came to understand in what ways each writer's processes are unique and why it is important that we pay close attention to the setting in which the writer composes, the kind of task the writer confronts, and what the writer can tell us of his own processes. If we are to understand how writers revise, we must pay close attention to the context in which revision occurs. Janet Emig, citing Eliot Mishler, has recently described the tendency of writing research toward "context stripping."'1 When researchers remove writers from their natural settings (the study, the classroom, the office, the dormitory room, the library) to examine their thinking processes in the laboratory, they create "a context of a powerful sort, often deeply affecting what is being observed and assessed."2 Emig's essay points to the need to examine critically the effects of these practices. The subject of the present study is not anonymous, as are most subjects, nor will he remain silent. I began the investigation with a critical eye regarding what he has said about revision, he with an equally critical attitude toward methods of research on cognitive processes. To some extent our original positions have been confirmed-yet I think each of us, researcher and writer, has been forced to question our assumptions and examine our dogmas. More important, this project stirs the dust a bit and suggests a new direction for research on composing processes.