One hundred and sixty‐three children were tested for transitivity of length, and also for ability to remember the relative lengths of the three rods taken in successive pairs. It was found that of 107 children who failed to make the transitive inference, only 20 were unable to remember the comparisons. The results are intrepreted as contrary to the hypothesis of Bryant and Trabasso that lack of transitivity is attributable to failure to remember the appropriate comparisons. The Mathematical and Psychological basis of transitivity is considered, and it is concluded that an incorrect definition of the task has been introduced into the literature. Those studies which appear to show that preschool children can make transitive inferences may be interpreted as requiring, not transitivity, but knowledge of the specific stimuli use