A study determined the effect of three types of text structures (attribution, comparison, and collection) on fifth graders' reading comprehension. Subjects, 71 students from 3 classrooms in 1 public elementary school in southern California, were randomly assigned to 1 of those 3 conditions. There was no instruction given. Students read oceanography essays for approximately 20 minutes per day over the course of a week and a half. At the end of the reading sessions, students took a comprehension posttest. Two weeks later the same posttest was administered to assess retention effects. Results indicate a significant interaction effect for time of measurement by treatment. Univariate follow-up tests indicated that the collection group significantly outscored the comparison group on the immediate posttest. There were no significant differences among the groups at the delayed posttest. A follow-up study done with sixth graders (to assess the impact of the effect of a crude drawing in the text read by one group) attained similar results. (One table and one figure of data are included.) (Author/RS)