Abstract: This paper discussed the necessity of self-study and experimental methodsof self learning, and problems concerning how to coach a self-studyexperimental class. Quantified data showed four times that the fiveexperimental classes were all superior to control classes in their achievementand self-study ability. One can never affirm that self-study cannot godeep in understanding concepts and principles. The experiment showed thatstudents with relatively poor foundation, too, are able to study on their own.

Key words: NULL