A proposal that the normal reader does not pay attention to individual letters but to sets of features which are characteristic of a word as a whole is presented and tested in this investigation. It was hypothesized that the reader learns (1) to extract distinctive feature information simultaneously from several parts of the configuration and (2) to integrate this information for identification of the whole. Two hundred and sixteen college students were given booklets with six 150-word passages of text. Each passage was printed in one of six typographic styles making use of upper and lower case letters. The subjects were asked to locate 20 words in each passage. The results of the experiments supported the hypothesis that disruption in the total look of the word does not interfere with the reader's ability to identify that word unless discriminability of elements is disrupted, as in the passages where the size of the individual letters was mixed. Even then, the readers coped well with the disruptions. References are included. (NH)