The present study was designed to examine so-called context effect on performance in number conservation tasks. Twenty-two nonconservers.(mean age 4 years 11 months), in standard number conservation tasks received same kind of tasks in three modified conditions. Main findings were as follows. In the first place, even the subjects who failed a one-to-one correspondence task gave conserving responses in a meaningful context. In the second place, many subjects gave conserving responses even in the condition in which the transformation of elements was accompanied with addition of one element and therefore non-conserving responses were in fact correct. In the third place, conserving responses could be induced even in a condition without context, if only a perceptual contrast of elements after transformation would be enough weakened. These results were interpreted as evidences of degeneration theory, proposed by the author, according to which conserving responses in a meaningful context would not mean the facilitation of conservation competence inherent in young children, but induced by evading cognitive perturbations which were essential in standard conservation tasks.