Horizontal readability, the analysis of essentially similar passages through classification of words and phrases according to their cognitive similarity, is discussed in relation to its usefulness in differentiating among materials designed for the same reading level. Three studies of horizontal readability in which passages were rated for high and low comprehension are described. Research findings indicated five variables of reading difficulty. (1) Vagueness, which resulted from indeterminate qualifiers and probability words, lowered comprehension. (2) Explaining links such as prepositions and conjunctions, which indicated that the cause, result, or means of an event or idea was being presented, raised comprehension. (3) The frequent use of examples produced greater understanding. (4) A rule-example-rule pattern of explanation was more effective than either inductive or deductive explanation. (5) The elimination of irrelevant sentences facilitated comprehension. It is noted that experimental research will be necessary to further clarify these findings, but it is recommended that the findings be implemented now for textbook evaluation. References are listed. (BS)