New and Exciting: The Racial Analysis Calculator

[Dienekes]

For the first time in the history of the Internet, a do-it-yourself racial analysis tool.

Try Version 1.0 now and come back here to tell us what your results are -- or if you've encountered any bugs.

PS: This works for Caucasoid males; future versions will include data for non-Caucasoids and females.

dienekes.angeltowns.net/calc/rac/

[AWAR]

I got the result proto-Europoid The closest plate was an Ugrian Ladogan

I guess I didn't measure things right.

[AWAR]

How do you measure upper facial height?

[Dienekes]

racialreality said:

Your height/length index is: 113.1 (hypsicephalic)
Your height/breadth index is: 145.5 (acrocephalic)
Your estimated endocranial volume is: 2708 cc (aristencephalic)

Your head height is probably inaccurate. It should be measured from the earhole to the top of the head. I hadn't put the instructions for the head height in my figure, so when I do that, I'll add that little graphic to the page.

[Dienekes]

AWAR said:

I got the result proto-Europoid The closest plate was an Ugrian Ladogan

I guess I didn't measure things right.

The Ladogan consistently groups with other Proto-Europoids. It should be remembered that it is physiognomic traits and the degree of facial flatness that indicate Mongoloid admixture, and these aren't measured by the calculator.

[Dienekes]

Here are my results, by the way:

Euclidean Distance Cosine Similarity
Proto-Europoid: 16 0.031
Mediterranoid: 12 0.055
Alpinoid: 4 0.74
Irano-Nordoid: 15 -0.694
Dinaroid: 12 -0.174

Your classification is Alpinoid (using Euclidean distance) and Alpinoid (using Cosine similarity).

The confidence in the accuracy of your input data is 78%

According to Euclidean Distance:

You are metrically furthest from Plate 4, Figure 4 (distance = 50)
You are metrically closest to Plate 14, Figure 1 (distance = 3)

According to Cosine Similarity:

You are metrically closest to Plate 14, Figure 1 (similarity = 0.764)
You are metrically furthest from Plate 16, Figure 3 (similarity = -0.806)

I will be adding more explanatory material in the calculator in the future, describing how to obtain anthropometric measurements, the description of the five metric types and of anthropometric terminology in general.

[Melnorme]

Why are 'closest' and 'furthest' not listed in the same order for both Euclidian and Cosine Distances?
It's kind of confusing.

[skord]

Your cephalic index is: 79.8 (mesocephalic)
Your height/length index is: 90.4 (hypsicephalic)
Your height/breadth index is: 113.3 (acrocephalic)
Your facial index is: 85.7 (mesoprosopic)
Your upper facial index is: 50 (euryene)
Your nasal index is: 54.5 (leptorrhine)
Your estimated endocranial volume is: 2022 cc (aristencephalic)


Euclidean Distance
Cosine Similarity
Proto-Europoid: 30 -0.776
Mediterranoid: 5 0.654
Alpinoid: 8 0.148
Irano-Nordoid: 8 0.243
Dinaroid: 14 -0.419

Your classification is Mediterranoid (using Euclidean distance) and Mediterranoid (using Cosine similarity).

The confidence in the accuracy of your input data is 84%

According to Euclidean Distance:
You are metrically furthest from Plate 5, Figure 1 (distance = 74)
You are metrically closest to Plate 28, Figure 3 (distance = 3)

According to Cosine Similarity:
You are metrically furthest from Plate 6, Figure 4 (similarity = -0.902)
You are metrically closest to Plate 28, Figure 3 (similarity = 0.745)

Closest match,this Danubian:
www.legioneuropa.org/Racediv/CSCoon/Images/p28f3.gif

[Dienekes]

Melnorme said:

Why are 'closest' and 'furthest' not listed in the same order for both Euclidian and Cosine Distances?
It's kind of confusing.

Good observation. I changed it.

[Zetaman]

Greeting Dienekes and good job.

ps: try to analyze my results (It appears I am atlanto-med with borreby admixture, just like I thought, just like many other Montenegrins, and speaking of Montenegrins I say hello to Awar ).
__________________________



My results

Results of Racial Analysis

Terminology is according to Vallois (1965). Endocranial volume is estimated according to the Pearson formula (Comas, 1960).

Your cephalic index is: 74.9 (dolichocephalic)
Your height/length index is: 112.8 (hypsicephalic)
Your height/breadth index is: 150.7 (acrocephalic)
Your facial index is: 84.8 (mesoprosopic)
Your upper facial index is: 53.9 (euryene)
Your nasal index is: 59.8 (leptorrhine)
Your estimated endocranial volume is: 2787 cc (aristencephalic)


------------------------------- Euclidean Distance---- Cosine Similarity
Proto-Europoid: 45 0.814
Mediterranoid: 120 -0.774
Alpinoid: 81 0.068
Irano-Nordoid: 91 -0.294
Dinaroid: 72 0.391


Your classification is Proto-Europoid (using Euclidean distance) and Proto-Europoid (using Cosine similarity).

The confidence in the accuracy of your input data is 62%

According to Euclidean Distance:
---You are metrically closest to Plate 5, Figure 1 (distance = 35)
---You are metrically furthest from Plate 10, Figure 5 (distance = 173)

According to Cosine Similarity:
---You are metrically closest to Plate 24, Figure 1 (similarity = 0.903)
---You are metrically furthest from Plate 43, Figure 1 (similarity = -0.786)

www.legioneuropa.org/Racediv/CSCoon/Images/

The five types listed in the results of your racial analysis represent five clusters discovered in a multivariate analysis of 152 individuals whose anthropometric measurements are given in the photographic supplement of Coon (1939). The names of the five clusters are arbitrary labels corresponding to traditional deignations. They are based on metrical similarity based on nine variables of the head, and should not be interpreted as denoting "races".

References

Coon, C. S. (1939) The Races of Europe, MacMillan, New York.
Vallois, H. V. (1965) "Anthropometric Techniques", Current Anthropology 6(2): 127-143.
Comas, J. (1960) Manual of physical anthropology, Thomas, Springfield, Ill.

[Dienekes]

Zetaman said:

Greeting Dienekes and good job.

Your results are probably inaccurate. You must have entered some wrong or inaccurate values.

In the next version, I will think of a way to give warnings to users for inaccurate measurements. Roughly speaking, if your distance to your closest match is more than 5, something foul is probably at play.

[Racial Reality]

Dienekes said:

Your head height is probably inaccurate. It should be measured from the earhole to the top of the head. I hadn't put the instructions for the head height in my figure, so when I do that, I'll add that little graphic to the page.

Then this site isn't so helpful after all. It says to measure Head Height starting from the chin.

I've remeasured correctly and updated my post with the new data. What happened to the confidence value? I'm not getting that with my results anymore.

[Dienekes]

racialreality said:

Then this site isn't so helpful after all. It says to measure Head Height starting from the chin.

I've remeasured correctly and updated my post with the new data. What happened to the confidence value? I'm not getting that with my results anymore.

I've put a new graphic on the calc page which explains what to measure. Also, now if measurements are too weird it gives warnings. This is more informative than the confidence value.

[Dienekes]

Limesucker said:

Dienekes Pontikos,

3 questions.

One, are you going to make the Coon data available for download?

Two, what clustering algorithm did you use to determine your five main clusters?

Three...
I assume that Coon, for his photographic exhibits, selected heads that best exemplify certain predetermined racial types, instead of using a random sample of heads from random Caucasoid populations. Don't you think that the clusters discovered by your multivariate analysis might be, to some degree, an artifact of Coon's subjective selections?

1. Eventually, yes.
2. K-Means. The value of K=5 was arbitrarily selected as the highest value for which each cluster had at least 15 individuals (~10% of the total).
3. You are right about that potential problem. It'd be nice to find a sample that doesn't have that problem. The only two comments that I can make at this moment is that (i) visual inspection of traits and scatterplots of trait combinations don't show a peculiar distribution which could be thought of as due to selection in favor of "extreme" types, and (ii) Coon's classifications don't always agree with the cluster assignments. Hence, the clustering procedure didn't just rediscover Coon's artifical "types".

[Arawn]

How do you do it?

I could not concievably go about measuring my head in various ways in front of a mirror without cracking up in hysterical laughter.