A study was conducted to replicate partially John Gould's study using more advanced machines and editors, and to test several of Colette Daiute's hypotheses about writing with the computer. Gould's study indicated that expert writers using text editors required 50% more time to compose on text editors than on hard copy, and the extra time did not improve the writing. Colette Daiute hypothesizes that the computer's speed and ease in editing will help writers to produce more text, that the blinking cursor will stimulate an audience and prompt the writer to consider the reader's needs and to keep writing, and that the machine's speed and ease of editing help to destroy writing blocks. The 15 subjects--faculty, administrators, and system designers from Carnegie-Mellon University--were experienced writers, and had several years of computer experience. Each wrote one letter on each of four topics, in a hard copy condition and three computer conditions. In two of the computer conditions, writers used "Andrew," an advanced computing system and related software being developed at the Information Technology Center as a joint venture of Carnegie-Mellon University and IBM, while the third used the IBM PC and the text editor MINCE. Three quantitative measures were collected and analyzed: time to compose the letter, total words produced, and words produced per minute. The results indicated that writers at the advanced work station produced more texts than in the pen and paper condition, in contrast to the machine Gould tested, and produced better text than in the PC condition. The results suggest that some machines can help people write more and better, but some machines may also make them write more poorly. (HTH)