In a partial replication of Samuels' (1972) experiments, 60 first grade pupils were assigned to a letter name group, a letter discrimination group, or a control group to investigate the effect of letter name knowledge on learning to read words. Artificial letters formulated by Gibson, et al. (1962) were used instead of those of Samuels, and different results were obtained. A two-way analysis of variance, followed by Tukey comparisons, revealed that the letter name group completed the transfer task (Learning words) in significantly fewer trials than either the control or letter discrimination groups, which did not differ. Sex differences and interactions were not significant. (Author)