frequency of such variants. This possib Llity brings out clearly the sensitivity to the vagaries of the sampling process of any approach which : , determines the frequency of rare variants from a semi-exhiustive sampling of disecate demes in a minimally disturbed population. Otherwise stated, the estimate must have a large variance. The properties of the samples of Micronesiang and Amerindians might be vendered more comparable to those of Caucastans and Japanese by eliminating the old and the young from the former two samples, but tie basic issue would still remain: the individuals in the former two samples will be more related than those in the latter two. The most practical way to meat this issue seems an extension of the sample, thus reducing the tnpact of any one ‘jackpot’. In addition, it is sugseskad that the definitive treatment of comparative variant frequencios which a larger sample will permit must consider a varlety of dafinitions of rare variant and/or analyze the total heterozygosity of the population. The problem is not a trivial one. The frequency of such variants iu natural populations is maintained by a complex balance between selection, mutation, and population structure. Basic parameters though they be, both the manner of action of selection and the rate and types of mitation tin higher organisms remain poorly understood. Both selection and mutation may be studied directly, i.e., thru surveys followed by detailed family studies of each varfant, or indirectly, i.e., through the manipulation of population parameters (35,36). whe former approach is much more laborious than the latter, and it is tempting to pursue the easier course, but the latter approach is only as sound as the y ( which should be hased on as total a q Gh t, e equenc estimates ‘ 5 ‘ot olan im: a 3