the french samples were obtained from multiple regions within france, although I can't find many details about precisely where these regions were. the "italian" samples were obtained from Bergamo, north-east of Milan (could be a population isolate, but seems unlikely). 
 
http://www.cephb.fr/HGDP-CEPH-Panel/HGDP-CEPH_Table1-1.html

have any of these height association studies seriously broken down "height" into endophenotypes, e.g. femur length, tibia length, spine parameters? there's some effort to do that in this study (comparing standing and sitting heights), but it's a weak proxy for actual anthropometry. 
 
it seems plausible that there are genes of large effect for, say, femur length, that don't affect other components of height - so they'd be "small effect" genes in a scan that just looked at total height but they'd stand out when you looked at the endophenotypes. of course there's a trade-off with multiple testing problems, so you'd need big cohorts.