Historically, IQ tests were designed to test the limits of a child's abilities. All vocabulary items on the Stanford-Binet Intelligence Scale were shown to the child, and credit was given for every correct response, even after most had been missed. In modern testing, rigid discontinued criteria combined with low test ceilings prevent gifted children from demonstrating the full strength of their reasoning. Extended norms do not resolve these issues. The historical context of testing are discussed and data presented on the extent to which gifted children fail to reach discontinue criteria and fail to benefit from extended norms.