Geography is facing the danger of being elbowed out by other disciplines in the research on the relationship between human and land. Through analysis it is pointed out that the fundamental reason is the fact that geography lacks an abstractive paradigm that suits the principle of geography and is different fr om that of any of the other disciplines. For the purpose of raising the differentiation rate on the human-land relation, human and land are put in a framework of graded order of geographical system and examined in line with Fractal theory, Boundary Hull Theory and the Theory of the Dissipative Structure, thus reaching the symmetrical abstractive principle of the relationship between human and land. This principle has two advantages:1) Both contrapositive parties, human and land, obtain obvious individually characterized mark in the principle of geography by receiving a "regional birthmark". Whereas the regional graded system provides a structure for the further study of the graded systematic relation between human and land, which in fact paves the theoretic way for the geographical evidence study. The result will undoubtedly raise the differentiation rate of the human-land relation. 2)A "standard insight" is offered for the study of relation bet ween human and land which suits the principle of geography. It is neither that of the Environmental Determinists', which believes that environment influences human, nor that of the Human Ecologists', which holds that human adapts himself or herself to the environment. Rather, it studies the relationship from a human-land mutual-influence angle. Because of the existence of different dimensions of regional boundary hull, the study of the human-land mutual influence from the symmetrical abstractive angel becomes more effective.
For the great importance of this principle, the essay briefly explains its significance in theory building as well as in practical usage. It is hoped that the research of the human-land relation will be pushed back to the principle of geography itself.