I’ll add some stuff to tom~'s fine post.
I understand the SCD example. Sure, self-reported blacks in the USA have a greater incidence of this disease than whites. Sure, self-reported blacks have a higher incidence of dark skin and curly hair. Sure, they have a higher incidence of flat noses. There are two problems in saying this, though.
The first is that the definition of race is hazy at best. A large percentage of people self-reported more than one race on the 2000 US Census. So, for those people, you could not make any significant conclusion. We get into lots of problems as well – how about non-US populations? How about putting any types of limits as to who self-reports themselves as what? What percentage of blood is necessary to be considered black? Old-South “one drop” rule or Nazi 1/8 rule? You can see why I am reluctant to go there.
The second much more subtle. What we are seeing here are advantageous mutations in certain populations. All of the traits listed above (dark skin, curly hair, SCD, flat noses) along with a few lesser known traits (beta thalassemia, G6P deficiency and favaism, I known of a few others) have one thing in common: they are advantageous mutations for tropical, malaria-filled conditions. Look at populations around the world at that latitude and what do you see among West Africans, Andaman Islanders, New Guinea lowlanders, Australian Aboriginals, Sri Lankans, and Native Americans? Dark skin, curly hair, flat noses, and in places with malaria, hemoglobinopathies and other disorders giving rise to some protection. Just like you see loss of pigmentation and mutations giving protection to tuberculosis (cystic fibrosis, it is thought) and other diseases among higher latitudes.
What I am getting at is that it is possible to have a small set of shared advantageous mutations with no other genetic similarity whatsoever. This is the nature natural selection and the gene flow of advantageous mutations through populations. Basically, if something is advantageous, it gets not only propogated throughout a population, but other similar mutations are likely to crop up in other separate populations in the same environment.
According to all of the biometric measurement done in properly controlled acquisition sets, the similarity ends there. Penis size I would bet would go along with height, weight, and lung capacity, and show no significant differences between self-described races.
The mantra again: just because two arbitrary groups show a handful of similarities does not mean that you can conclude that they are genetically or scientifically separable. Cultural divisions exist and are important, but have no relation whatsoever to measuring physiologic data.