1 and Supplementary Tables 1 and 2), and all 36 loci are located in the genomic regions known to be associated with height3. We did not find any locus with multiple associated SNPs for BMI (Supplementary Tables 1 and 3). For most of the height-associated loci, multiple associated variants were detected, mainly because of their very low LD (r2 < 0.01), despite their relatively close physical proximity (Table 1). In this case, the effect sizes from a joint analysis were little different from those from single-SNP analyses. For some loci, SNPs were in modest LD and their increasing alleles were positively correlated. One example of this is the rs17720281 and rs7689420 SNP pair at the HHIP locus on chromosome 4 (Table 1), where effect sizes for these SNPs were therefore overestimated in single-SNP analyses. In the joint analysis, although the joint effects were smaller compared to the marginal effects, these SNPs still reached genome-wide significance, and the variance explained by them collectively was larger than that if we only considered the leading SNP at that locus. For the loci at which the increasing alleles of at least two SNPs were negatively correlated, the SNP effects were underestimated in single-SNP analyses,
