This paper provides some properties of the linear-exponential distribution by recordedvalues. Some recurrence relations for single moments of values are derived. In the meantime, all the higher order moments of recorded values may reduce to the first recorded values and simplify the processes of the higher order moments of recorded values. Simultaneously, the expected values of the product moments of recorded values are derived. From these properties, the paper predicts the future unknown recorded values from the given recorded values. 本文提出由記錄值之性質求線性指數分配之特性，導出一些單一動差值之循環關係；同時，求高次序之記錄值動差可以化成第一記錄值，如此可簡化一些高次序動差記錄值的處理；同時也推導出記錄值之乘積的動差期望值，並利用這些特性，由已知記錄值來預測將來未知的記錄值。
