The variation theory of Ference Marton and his collaborators has widely been used as a framework for explaining what can possibly be learned in a particular classroom and what cannot. This paper reports on an experiment that put this theory to test in the context of students' learning of the orthographic structures of Chinese characters. The experiment was carried out in the classrooms of two primary schools in Hong Kong. In each of the schools, two classes of students were taught differently, as informed by the theory, about the significance of the location of a component in the orthographic structure of a character in relation to whether the component provided a clue to the meaning of the character (called the part-part relations). The results of the experiment are consistent with the prediction of the theory that those students who were given the possibility to experience variation in the locations of components in the orthographic structures significantly outperformed those who were not. The results of the experiment demonstrate the power of the theory in guiding the design of teaching that affords students' learning to happen.