A fault does not necessarily cause an immediately effective error, the
reby leading to a system failure. Many experiments have shown that mos
t hardware faults do not cause immediately detectable errors. In fact,
it has been found that a significant proportion of the faults injecte
d during the experiments remained latent, i.e. went undetected. In thi
s paper we call the phenomenon the latency problem. The paper describe
s an experimental study of the latency problem, with the aim of valida
ting the analytical model developed by Soh and Dillon [Microelectron.
Reliab. 36, 1223 (1996)]. Copyright (C) 1996 Elsevier Science Ltd.