As an estimate of probability of readability is discussed within the problem of ensuring that science tests meet the purposes of measuring skills, knowledge and ability of students. The necessity of technical terms is cited while arguing that testing for language should not interfere with tests for knowledge and skills. A review of formulas for measuring readability is presented to establish general principles, and several approaches to solutions in readability and testing are discussed. The appendices include a table and a scale for measuring reading ease by Flesch and a summary of findings by Gardner on student understanding of science connecting words. A word list shows percentages of correct usage of science terms in the context indicated. (Author/CM)