Abstract
In studying electron tube life, measuring the life of a tube in service in a system whose age is T amounts to lifelength proportional sampling. The problems of estimating mean life from such data are considered for finite T and for “equilibrium” (large T). Several variations on the straight-forward life-biased sampling plan which are avilable to the experimenter are mentioned and conditions are found under which these other sampling schemes would be better than the life-proportional scheme. Particular attention is paid to the gamma and Weibull families of distributions. Other applications of these results are discussed.

This publication has 0 references indexed in Scilit: