In this article, the feasibility of reconstructing sibship relationships from genetic data without genotypes of possible parents is explored. A Bayesian approach is taken in the sibship estimation, and the sensitivity of the problem to the population allele frequencies and the feasibility of estimation with unknown population allele frequencies is examined. The approach is illustrated on forensic data from falcons, consisting of unlinked mini-satellite markers; but the approach is also applicable to other types of genetic markers. The results indicate that situations in which the problem is relatively insensitive to the population allele frequencies can occur, and that if an appropriate prior distribution over the population allele frequencies can be selected, then analysis can be carried out. The effect of mutations is also examined and indicates that with the relatively high mutation rates possible in mini-satellite markers, the presence of mutations can alter the results significantly.

Key Words
Bayesian; Dirichlet; Forensic data; Genetics; Likelihood; Mutation.

Ian Painter is Postdoctoral Research Associate, Department of Statistics, North Carolina State University, Raleigh, NC 27695-8203.