Large, expensive, multidisciplined, multicenter studies present special problems to journals. The study by Kinsey et al. in this issue (p. 655) is a good example. This study was designed by a distinguished peer group. It was carried out in five university centers by 27 investigators, written up for publication by ten authors, evaluated by statisticians, and was ready for publication nearly nine years after the research was first begun. The goal of the study was a very good one. The authors hoped to define the level of PaO2 and the duration of exposure which could result in retrolental fibroplasia. Nearly 10,000 blood gas studies were carried out on 589 low-birth-weight infants.