January 04, 2004
"Discordant" Clines
One of the argument against the existence of race is that of "discordant clines". This argument goes as follows: if you look at feature A, say "skin color", then you'll tend to group people in a certain way, e.g., northern vs. southern. If you look at feature B, say "frequency of blood group B", then you'll tend to group them in another way, e.g., eastern vs. western. If you look at genes for sickle-cell trait, or lactase intolerance, etc. you get an entirely different grouping. Since all these traits don't "go together", then there's no reality in race: it all depends on what you choose to look at, it's all subjective!
There is some truth in this argument. The reason it fails however is because it assumes that race is valid only if all traits are distributed in a concordant manner, i.e., if we had a "light-skinned, low-B, no sickle-cell trait, lactase tolerant" and a "dark-skinned, high-B, sickle-cell trait, lactase intolerant" race.
In reality there are both concordant and discordant clines in trait distribution. This is simply due to the stochastic nature of human variation and the different ecological adaptations of individual populations within broader racial groups. However, it is the fact that populations within the same race tend to match in more traits than they don't that allows us to discover the existence of races in the first place.
As an example, consider three cars: a blue Audi, a red Audi and a red Ferrari. If we say that the Audis belong to one "race" of cars, and the Ferrari belongs to another, then the argument could be raised that the red Audi and the Ferrari are more similar in their color to each other than to the blue Audi. But, if we study a great number of traits we discover that the blue Audi is in fact more similar to the red one, because it agrees with it in more traits than with the Ferrari. Simply put, the existence of multiple concordant clines is unlikely to have occurred by chance or identical selection pressures: it speaks of racial affinity.
To test this, I took a well-known craniometric dataset originally by W. W. Howells and (using ~1400 male crania from 28 populations) calculated one big array CLOSER[i,j,k]. Each entry in this array indicates on how many (out of a total of 45) cranial metric traits population i is closer to population j than to population k.
Each entry can be at most equal to 45: this means that i is closer to j in every trait, when compared to k: perfect concordance. A number ~22-23 means that there are as many concordant as there are discordant clines.
It is difficult to present all these numbers (I might organize these results in a better form later on), but I'll give some examples. For starters, let's take the Europoid populations of Oslo as i and Medieval Zalavar in Hungary as j and look at the number of traits in which these are closer to each other than to the 26 remaining populations: these range between 25 (for Berg in Austria) to 42 for the San Bushmen.
What does this mean? It means that Oslo agrees with Zalavar in 25 traits and with Berg in the remaining 20 traits. But, Oslo agrees with Zalavar in 42 traits and with the San Bushmen in 3 traits. It is clear that even though some discordant traits exist, we can safely classify Oslo/Zalavar and the San as different races.
Or let's take the population of Hainan (China) and Hokkaido (Japan) and repeat this experiment: the numbers now range between 19 (for Kyushu, in Japan) and 41 (for the Andamanese). Hokkaido agrees with Hainan in 19 traits as opposed to 26 for Kyushu. But it agrees with Hainan in 41 traits as opposed to 4 with the Andamanese. Again, we can place Hainan and Hokkaido in one race and the Andamanese in another.
This simple experiment illustrates the fallacy of the discordant clines argument: certainly discordant clines exist, but it is the excess of concordant clines over discordant ones which allows us to determine racial affinity.
PS: This simple analysis could be improved by taking into account the interdependence of traits and the degree of agreement or disagreement. But, it suffices as is to demonstrate the simplistic logic behind the discordant clines argument.
Posted by Dienekes at January 4, 2004 10:38 PM | PermaLinkLet't think about the car analogy a little more. Assume that Ford and Honda both make two vehicles: blue minivans and red subcompacts. What's more similar: the blue minivans or the two Fords of highly different shape and function?
The answer depends on the perspective of the questioner. If I'm in the market for a minivan, then the two minivans are most similar for me. If, however, I'm an investor trying to decide whether to buy stock in Ford or Honda, the sales performance of the two Fords vs. the two Hondas is most relevant.
When we talk about race, we are thinking about ancestry, in terms analogous to the the question: who built these cars? The two Fords would be most similar from this viewpoint because they are made by the same company, just as short African pygmies and tall African Dinkas are more similar racially (i.e., they have more recent ancestors in common) than are short African pygmies and short Andaman negritoes.
Posted by: Steve Sailer at January 7, 2004 05:57 PMThe key question is this: if you had four vehicles (a minivan and a subcompact from each of the two companies) and scraped off all company-identifying labels, would you able to tell that these four vehicles came from two companies?
The vehicles are obviously the result of "selective forces", not natural selection but e.g., "adapted for large families" in the case of minivans. So the two minivans might look similar due to the fact that they were exposed to similar selective forces. But hopefully if you looked under the hood you'd be able to see e.g., the the Ford minivan/subcompact share some components, indeed many of them. So, you'd perhaps be able to tell that they were made by the same company.
Posted by: Dienekes at January 8, 2004 03:13 PM