跨年級研究有助於課程的縱向發展，但卻耗時、費力且難以效化。而藉由電腦建立學生於不同科學主題之最適概念演化樹的「支序分類之概念演化取向」卻可克服上述限制。據此，研究者採用此取向，並以概念改變研究中已被不同學者確認心智模式類型的「地球形狀」為主題，建立了地球形狀概念演化樹之最適假說。本研究則以一、三、五年級共336 位學生為研究對象，設計地球形狀概念測驗，藉此獲得學生於各認知特徵比例消長情形，以檢驗地球形狀概念演化樹之最適假說。研究結果顯示，此假說與跨年級調查結果大致相符，且藉由假說與跨年級資料之整合，本研究清楚呈現兒童地球形狀概念發展之全貌，並對支序分類之概念演化取向之可行性提供進一步的支持。

Although a cross-grade study is helpful for designing vertical curricula, such studies require substantial time and effort and present validation concerns regarding the research results. However, a cladistic approach in conceptual evolution could be used to construct a best-fit conceptual evolution tree regarding various science topics by using a computer to overcome these limitations. Therefore, we adopted this approach and used Earth as the topic to build a best-fit conceptual evolution tree of Earth. The purpose of this study is to verify a prediction of children’s conceptual evolution trees of Earth in a cross-grade study and to validate this novel approach. We adopted a diagnostic test to investigate 336 students who attended Grades one, three, and five, to obtain the percentages of various cognitive characteristics and statuses of students in different grades. The results indicated that the prediction of the best-fit conceptual evolution tree of Earth closely fit the empirical data of the cross-grade study. The results showed that children’s conceptual evolution trees represent the development of their concepts of Earth clearly and that the feasibility of using a cladistic approach in conceptual evolution is supported. The results can aid researchers of science education in understanding the development of children’s conceptions.