The ceiling principle is an intentionally conservative way of estimating the frequency with which individuals who share particular alleles appear in the general population. It establishes frequencies for each allele by taking random samples of 100 individuals from each of 15 to 20 populations and using the largest frequency with which the allele is found in any of these populations or 5 percent, whichever is larger, as an estimate of the allele's frequency in the population of interest. These frequencies are then multiplied to yield an estimate of the likelihood that a randomly selected person would exhibit the same allelic configuration as that which characterizes samples of DNA associated with the defendant and with some crime. The ceiling principle blends scientific and value considerations; the blend of the two is what makes the ceiling principle vulnerable to scientific criticism. It is also what makes it an arguably more attractive forensic approach than current product-rule procedures. The core value that motivated the National Research Council's recommendation of the "ceiling principle" was its sense that the calculation of random match probabilities should be conservative, such that any errors in the random match probability should overestimate rather than underestimate the probability that someone other than the defendant left the DNA evidence.
Lempert, Richard O. "DNA, Science and the Law: Two Cheers for the Ceiling Principle." Jurimetrics Journal 34, no. 1 (1993): 41-57.