ABSTRACT
W e see that a series of testing sessions are p rogrammed at specified intervals of t ime. A t the first of these sessions, the author tests out the first vers ion of the first chapter. T h i s is then revised, i n the l ight of the data collected, and a further chapter is drafted. A t the second session, the same student (or smal l group) that was used to test the first draft of Chap te r 1, now studies and comments on the first draft of Chap te r 2. T h e new version of Chap te r 1 is tested again , but us ing a different student or small group. W i t h luck, the second draft of Chap te r 1 w i l l now teach m u c h better and may indeed be fit to publ ish and use. The re may appear some new problems, however, that were not encountered i n the first testing. It is a good pol icy to take each chapter through as m a n y stages of testing as is necessary to el iminate al l weaknesses and obtain consistent l ea rn ing results wi th consecutive groups of students. T h e m i n i m u m recommended n u m b e r of testings is three, even i f no problems are encountered on the first testing. T h e feedback that the author gathers on Chap te r 1 w i l l also help to make the second and subsequent chapters better adapted to the target popula t ion right from the start, so the later chapters may need less repeats than the earlier ones. H o w e v e r , one should remember the need for cont inui ty - i n order to study a later chapter, the student usual ly needs to have knowledge of earlier parts of the materials. It is best, therefore, to p lan a developmental testing schedule for a m i n i m u m of, say, four or five repeats of each chapter, thus us ing the same four or five groups of students throughout the project.
