There are a number of limitations to our findings that temper our conclusions. Given that we used non-parametric linkage, which requires affected sib-pairs, from the 161 Finnish families for which we had both genotypic and phenotypic data, only a smaller sub-set of families (N= 45 families) had an affected sib-pair contributing to the significant linkage peak. While the two microsatellite marker flanking this linkage peak on 11p were both greater than 2.0, it is possible that this LOD score may be inflated due to interpolation. On the other hand, given that linkage rests on the assumption of genetic effect sizes that are at least modest (Risch, 2000), suggests that our genomewide significant linkage finding is less likely to be a false positive. 1000 simulations found that a LOD score of 3.1 was the threshold necessary for genomewide significance (e.g. genomewide adjusted p value < .05). The lack of consistency between Australian and Finnish samples may also be due to the greater genetic homogeneity of a Finnish sample relative to other Caucasian samples (Peltonen et al., 2000). On 11q, the combined
