本研究計畫分為二部分進行,第一部分為探討集區設計對觀測值漏失時之穩健性,計畫中求出任一集區設計,當某一集區內漏失t個對照處理之觀測值時,該設計仍維持穩健之充要條件,一個較簡單計算之充分條件亦在報告計畫中給出。殘留設計之A效率值之上下限也在計畫報告中求出。一些特殊形式之BTBD之A效率值可準確算出,其實例亦在第一部分之第五節中列出。 計畫之二部分為探討行列設計之連接性, Raghavarao and Federer(1975)提出當行與處理以及列與處理均為連接時,行列設計亦具連接性之充分條件,一個較Raghavarao and Federer更廣義之充分條件將在計畫報告中給出。 This research project is composed of two parts. The first part concerns the robustness of block designs against missing data is investigated. Necessary and sufficient condition for the robustness of an arbitrary block design is derived when t observations of control treatment in a block are missing. A simple sufficient condition is also provided. The lower bound and upper bound to the efficiency of the residual designs are obtained also. The robustness of some special BTBD's is derived and examples are given. Second part of the project is dealing with the connectedness of row-column designs. Raghavarao and Federer (1975) gave some restrictions so that if a row-column design is row-treatment and column-treatment connected then it is also treatment connected. However, their restriction seem too strict. We give a more relaxed sufficient condition for treatment connectedness.