Determination of Racial Affinity


An important aspect of forensic anthropology is the determination of the racial affinity of an unknown skeleton. Today, the world has become so accessible that there are few places where the populace is totally homogeneous. Even in countries where there is minimal migration from other geographic regions, it is not uncommon to see people of different origins passing through as tourists or in exchange programs. Therefore, the geographic isolation of the past is rare. Currently, the assessment of race is complicated by every kind of admixture as well as the variability that exists within races at the population level.
There have been numerous attempts to explain the origin of populations residing in different geographic regions. There are also text topics that focus on what races are, how they are formed, and how they change. Even the use of the term ‘race’ has been a sensitive issue among many scientists because of the unwarranted social connotations that have been associated with it. Therefore, when dealing with the complex and often controversial subject of race, it is essential for experts to convey exactly what is meant. Recent efforts to deal with the controversy and negative image of this designation have led to the use of incorrect and misleading terminology. Some anthropologists have begun misusing the term ‘ethnicity’ (which refers to cultural issues like nationality, religion, etc.) when discussing genetically based biological characteristics readily observable in the flesh and in the skeleton. A more appropriate term is ‘racial phenotype’ because it accurately describes what forensic anthropologists determine when they examine skeletal remains.
Another problematic area is the terminology used to categorize racial phenotypes. To meet the needs of forensic case work, and avoid confusion, we use the three traditional classifications: Caucasoid (white), Mongoloid (east Asian, Amerindian), Negroid (black). Today, a number of anthropologists substitute the term ‘Asian’ for Mongoloid, but this is inaccurate because not all Asians are Mongoloid (e.g., Asian Indians, Turks, Iranians) and not all Mongoloids are Asian (Amerindians). Most humans can be roughly classified as a member of a population within these designations, or some admixture thereof. Unfortunately, this is not always a simple matter, especially where complex admixtures are concerned. In addition, skeletal features give no clue to soft tissue characteristics such as the precise shade of skin color or eye shape.
A single individual can represent the three major racial phenotypes as well as any number of populations within them. The golfer Tiger Woods is a well-known example. At the population level, a good model of a highly mixed gene pool is found in the eastern nations of South America, such as Brazil, where there is much African, European and indigenous Mongoloid (Amerindian) admixture. Western countries like Peru and Chile have a long history of white and aboriginal intermixture, whereas others, like Argentina and Uruguay, have remained primarily white. The United States and South Africa are also centers of mixed populations from all around the world. With this kind of admixture, genes for original ancestral traits may be transmitted unpredictably and randomly to the next generation. However, there do seem to be some characteristics that remain ‘typical’ of obvious ancestral traits, and these are valuable markers that forensic anthropologists use in their investigations to establish the identity of an individual.
People are only recognizable to others by their outward appearance, and this is often dominated by the characteristics of a particular racial phenotype or sometimes an obvious combination thereof. The medicolegal expert must therefore begin investigations with these population specific facial traits even if they do not reflect someone’s actual genetic constitution. Thus, the purpose of this article is to investigate the ways in which a given individual or victim of a crime can be classified as Caucasoid, Mongoloid or Negroid from skeletal remains.


As in the determination of sex, race assessment from the skeleton can be carried out using both morphological and metric traits. Morphological traits involve shape and size differences that can be assessed with the naked eye. This type of analysis requires consistency in observation and improves with practice. Therefore, the forensic anthropologist who relies on the visible configuration of a particular structure to establish race must have an advanced education and considerable experience in observing and understanding skeletal variation between individuals, populations, and age groups. Metric determination of race is based on the use of selected measurements that show statistically significant population differences. Metrics require both statistical expertise and a thorough knowledge of osteometric techniques and principles. The most commonly used statistical approach for separating two or more groups is uni- or multi-variate discriminant function analyses.

Morphological assessment

Morphology isshape or configuration. A morphologic trait isone that can be evaluated by observation rather than through measurements. Over the years, race determination hasrelied primarily on the morphologic features of the skull and facial skeleton since these provide consistently reliable results in the majority of cases.
Some race-specific traits may not be present or fully developed in the young until they reach sexual maturity. The only exceptionsto thisseem to be the traditionally race-related morphology of the nasal sills, and alveolar prognathism. The wide nasal form and alveolar prognathism in Blacks can be clear even in infants. However, one must keep in mind that the nasal bridge is not well developed in any infant.
Morphologically, race differencesare most pronounced and observable in the skull, with the facial skeleton being the most diagnostic portion (Fig. 1). Table 1 lists some craniofacial features that have been linked to race in the three major groups. Following the process of dental analysis, a craniofacial race complex can be identified to separate Whites from Blacks. In Whites, this complex includes a high, wide, skull, rounded sagittal contour, low cheek, short, gracile zygomatic arches, orthognathic face, narrow interorbital distance, high nasal bridge, sharp nasal sills, and narrow nasal aperture. The Black complex is characterized by a long, low, narrow skull, flat sagittal contour, long, robust zygomatic arches that project laterally (relative to the narrowness of the head), alveolar prognathism, wide interorbital distance, receding chin, low nasal bridge, smooth, guttered nasal sills, and wide nasal aperture. The Mongoloid complex featuresa rounder skull, anterior and laterally projecting zygomatic bones, flat face with little projection, shallow nasal root, and shovel-shaped incisors. The most distinctive Mongoloid feature is the cheekbones. They exhibit high malar projection, both anteriorly and laterally, aswell asa malar tubercle at the inferior aspect of the zygomaxillary suture. In some cases the bizygomatic breadth (measurement of facial width) may be as wide as the breadth of the skull. Also, shovel-shaped incisors are much more frequent in Mongoloids, as are taur-odont molars. Interestingly, shovel-shaped incisors are also common in Caucasoid residents of the Asian Indian subcontinent (e.g. Indians, Pakistanis).

Frontal and profile views of Caucasoid (A), Negroid (B) and Mongoloid (C) skulls.

Figure 1 Frontal and profile views of Caucasoid (A), Negroid (B) and Mongoloid (C) skulls.

Table 1 Craniofacial traits of the three major human races

Characteristics Caucasoid Negroid Mongoloid
General appearance High, wide Long, low Rounded
Skull length Varied Long Long
Skull breadth Varied (often broad) Narrow Broad
Skull height Medium to high Low Medium
Sagittal contour Round to arched Flat Arched
Face breadth Narrow to wide Narrow Very wide
Face height Medium to high Low High
Cheek No projection Some lateral projection Anteriorly projected
Orbital opening Round to angular Rectangular Rounded
Facial profile Orthognathic Prognathic Medium to flat
Nasal opening Narrow to medium Wide Narrow to wide
Lower nasal margin Sharp Smooth and/or guttered Sharp
Nasal profile Straight Downward slant Straight
Palate shape Narrow to wide Wide Moderately wide

One of the best racial indicators is alveolar prog-nathism. In Blacks the tooth bearing portions of the jaws protrude noticeably in contrast to the straight or orthognathic face of whites and Mongoloids. Nasal configuration is also very telling. A short, wide, low bridged, flat nose with smooth and often guttered sills isan extremely reliable marker of Negroid affinity. A long, narrow, protruding nose with sharp sills points quite reliably to a white person.
Race differencesare also apparent in the mandible and dentition. The tall, thin, gracile ramusnoticeably narrowsin the middle in Whitesasopposed to the robust, rectangular or almost squared ramus in Blacks. This contrast is distinctly visible in the lateral photographsin Fig. 1. A wide, vertical ramus is associated with Mongoloids. Blacks (especially Africans) are also more likely to have fully erupted third molars, supernumerary teeth (e.g. fourth molars), and receding chinsthan other groups. Moreover, Blacks retain their dentition with age much better than do Whitesand are lesslikely to have dental work (fillings, caps etc). Finally, there are significant differ-encesin tooth size, with Blackshaving the largest, and Whites the smallest dentition. Australian aborigines have the largest teeth of any living humans. However,these are only visually apparent to the most experienced observers.
Although there may be considerable variability, some generalizations can be made about head shapes. Whitestend to have a high, wide head, Mongoloids are more rounded, and blacks often exhibit long, low crania. Other morphological featureshave also been described. The nasal bridge has been observed to have a distinct distribution among the races. The Cauca-soids have a ‘church with steeple’ shape (high and pinched), Negroidsa ‘Quonset hut’ (low and rounded), and Mongoloidsa ‘tented’ appearance (low to moderate with relatively straight sides).
Morphologic differencesare lessobviousin the postcranial skeleton. One exception is femoral curvature which differsnoticeably in that American Blacks have a much straighter femur in contrast to the pronounced anterior curvature in Whitesand Mongoloids. The pelvis is also quite disparate between Whites and Blacks. The white pelvishas been described as a broad basin with a lower symphysis than Blacks to compliment a broad torso, and in Blacks as the pedestal of a narrow torso. There are significant shape differences in the pelvic brim that give Whitesa wider, rounder (brachypellic) inlet and a narrower (dolichopellic) inlet in blacks. However, this is not always easy to distinguish visually and the metric approach to race assessment in the postcranial skeleton is preferred.
Determination of race in children hasfocused on the variation of several dental characteristics. Using 10 deciduous dental characteristics, it is sometimes possible to determine racial affinity in children. The Mongoloid dental complex includessix traitswith higher than average frequencies. One of these is the shovel-shaped morphology in the upper central and lateral incisors aswell ascanines(Table 2). The percentage of shovel-shaped incisors in il is about 77 in the Japanese, 10 in American Blacks and zero in Whites. The same trend is also seen in i2. In the Caucasoid complex, Carabelli’s cusp is as high as 35% in Whites and about 12% in Blacks. The canine breadth index (upper canine crown mesiodistal diameter x 100/upper central incisor mesiodistal diameter) is also higher in Caucasoids and Negroids than Mongoloids. Some trace this to the fact that American Blacksand White shave had more opportunity for admixture during the last 250 years or so.
When clear, morphologic traitsare very effective for racial assignment Interestingly, many of these traits have also pointed scientists in the direction of quantification as another approach to the definition of certain formations. In this way, some morphologic traitshave formed the basis of many metric techniques. For example, projection of the alveolar region of the face can be observed as either orthognathic, mesognathic or prognathic by looking at how far this region projects forward, but it can also be assessed by measuring the angle of the face.

Metric differences

The primary focus of metric assessment is size and proportion. Because there are significant size differences between races and the populations within them, metric techniques can be used to separate them. In cases where morphologic variants are not clear, or the facial skeleton is not available, measurement-based methodscan provide good levels of discrimination. This technique necessitates a thorough knowledge of skeletal landmarks, proper equipment, and precise measuring skills. Furthermore, there can be complex combinations of shape and size differences between populations that are not obvious to the eye, and these may be quantified and evaluated by using a set of measurements. The statistical technique most commonly used is discriminant function analysis. This approach assumes that human variation spans a continuum across space and populations, but concentrations of people with similar featurescan be found towards the centers, where as at the peripheries, there may be an overlap with neighboring groups. Because of the overlap, it is rare to be able to divide humans into distinct groups with 100% accuracy. Optimal diagnosis is obtained in areas where there are two major races, with little admixture and no intervening populations. Discriminant function analysis has become very popular, especially during the last 40 yearsand has been applied to many parts of the skeleton for purposes of determining both race and sex.
In order to test variation among three biologically diverse samples, data were obtained from American Whites and Blacks in the Terry collection (United States) and Japanese (Mongoloids) from the Jikei collection (Japan). They were analyzed to determine the extent to which these three groups can be differentiated from each other (Table 3). The first set of discriminant function tests assesses the differences between two samples at a time, e.g. American Whites and Blacks(Table 4). The second set is a multivariate discriminant function statistic that determines if an unknown skull can be assigned to one of the three groups(Table 5).
Table 3 lists descriptive statistics for 12 cranial measurements of both sexes of Japanese and American Whites and Blacks using a sample ranging in size from 29 to 44 specimens per race-sex group. Blacks have the greatest cranial length, Japanese, the largest bizygomatic breadth, and Whites, the largest maximum frontal and biasterionic breadths. Nasal width is the narrowest in Whites and widest in Blacks. Analysis of variance (F ratio) showed statistically significant differences between all groups. Using step-wise discriminant function statistics, only four to seven male and four to five female dimensions were selected to provide the maximum separation among the three samples (Table 4). It must be kept in mind that this type of assessment of an unknown individual is based on pair group comparisons (e.g. White/Black, White/Japanese, Japanese/Black), rather than as one of all three. Racial affinity can be calculated by multiplying each coefficient by the value of that dimension plus the constant and adding them to obtain the discriminant score. The score is then compared with the sectioning point calculated to separate each pairing of the three groups. The same table also provides the percentage classification accuracy. On average, femaleswere more accurately classified than males. In white-black comparisons, for example, accuracy isabout 90% for malesand 98% for females.

Table 2 Population variation in deciduous dental crown characteristics

Crown characteristics Frequency (%)
Whites Blacks Japanese
Mongoloid complex
Shovel shape (upper i1) 0 10 76.6
Shovel shape (upper i2) 0 15 93.3
Deflecting wrinkle (lower m2) 13 19.1 55.6
Protostylid (lower m2) 14.5 17 44.7
Seventh cusp (upper m2) 41.8 46.8 73.1
Metaconule (upper m2) 3.5 9.5 41.8
Caucasoid complex
Carabelli’s cusp (upper m2) 35.1 11.8 11.9
Canine breadth index (upper c) 106.3 107.8 101.5

Table 3 Descriptive statistics, univariate F ratio, and significance of differences among American Blacks and Whites, South African Blacks and Whites, and Chinese and Japanese

Dimensions Whites Blacks Japanese F ratio and statistical significance between
Mean SD Mean SD Mean SD White-Black White-Japanese Black-Japanese
Male 35.0 32.0 44.0
Cranial length 182.2 7.47 186.9 5.65 178.5 7.74 8A5b 4.53a 27.23c
Cranial breadth 142.9 5.06 138.3 4.43 142.2 5.87 15.54c 0.39 9.55c
Max. Frontal b. 123.7 5.81 119.4 4.48 119.2 5.39 11.10° 12.67c 0.04
Min. Frontal b. 95.9 5.50 97.5 4.45 95.4 5.18 1.79 0.16 3.49
Bizygomatic b. 130.7 4.62 131.9 7.30 135.6 5.70 0.65 16.46c 5.93a
Basion-nasion 100.4 4.17 101.4 4.14 101.4 3.92 0.98 1.21 0.00
Basion-bregma 133.8 8.15 131.0 6.88 138.4 5.06 2.29 9.71c 29.76c
Basion-prosthion 93.3 5.16 103.2 8.81 94.7 4.98 32.36c 1.65 28.25c
Mastoid height 30.3 3.04 32.6 3.58 32.9 2.33 8.55c 19.61° 0.20
Biasterionic b 111.8 5.77 109.0 5.85 107.9 3.84 3.97a 13.16° 0.99
Nasal height 51.7 2.63 53.1 4.92 52.0 2.49 2.01 0.13 1.75
Nasal breadth 23.9 1.59 27.2 2.19 26.2 1.90 47.93c 30.49c 4.48″
Female 31.0 30.0 29.0
Cranial length 177.2 8.09 178.0 6.80 173.2 7.33 0.21 3.85a 6.79″
Cranial breadth 139.2 5.50 133.6 5.82 137.3 6.49 14.90c 1.58 5.17a
Max. Frontal b. 118.5 5.71 113.0 4.68 113.1 6.06 17.16c 12.52c 0.01
Min. Frontal b. 93.4 4.47 93.8 3.20 91.2 4.70 0.14 3.29 5.86a
Bizygomatic b. 122.1 3.14 124.4 5.48 128.9 5.40 4.33a 36.86c 10.09c
Basion-nasion 96.0 4.75 96.1 4.08 97.7 4.27 0.00 2.09 2.23
Basion-bregma 129.2 4.74 124.1 6.59 132.4 5.86 12.23c 5.29a 25.96c
Basion-prosthion 87.4 5.18 98.4 6.91 94.1 5.16 49.55c 25.05c 7.28″
Mastoid height 28.0 2.70 28.4 3.10 29.0 2.98 0.24 1.87 0.64
Biasterionic b 109.6 6.76 104.4 5.14 104.9 4.90 11.44c 9.47c 0.14
Nasal height 49.8 3.41 49.2 3.10 50.1 3.06 0.53 0.08 1.08
Nasal breadth 23.2 1.76 26.1 2.41 26.2 2.16 28.35c 33.75c 0.01

Table 4 Population determination from paired combinations of Japanese and American White and Black crania and their classification accuracies for males and females

White-Black White-Japanese Black-Japanese
Dimensions coefficient coefficient coefficient
Cranial length - 0.0403828 - 0.0876780
Cranial breadth -0.0723247 0.1273584
Max. frontal breadth - 0.0955337 -0.1089742
Bizygomatic breadth 0.1231765
Basion-nasion length -0.1097166
Basion-bregma height 0.0575340 0.1139350
Basion-prosthion length 0.1125220
Mastoid height 0.1511110
Biasteronic breadth -0.0623639 -0.1207576
Nasal breadth 0.3694600 0.2081077
Constant 7.6988166 - 2.2330051 - 4.3549907
Sectioning point -0.5850100 - 0.9330200 - 0.7740400
% Accuracy 90.4100000 88.3700000 86.0500000
Cranial length -0.0881222
Cranial breadth -0.1237725
Max. frontal breadth - 0.0702601
Min. frontal breadth -0.1585874
Bizygomatic breadth 0.1706030 0.1416418
Basion-nasion length -0.1180378
Basion-bregma height 0.1269521
Basion-prosthion length 0.1514658 0.0686073
Mastoid height 0.1536866
Biasteronic breadth -0.0883313
Nasal breadth 0.2334410 0.1919761
Constant 4.0805893 -14.7170241 - 4.0549844
Sectioning point - 0.7309400 - 0.6727900 - 0.5958100
% Accuracy 98.48 95.08 89.19

Table 5 Population determination (three group) from Japanese and American White and Black crania and their classification accuracies for males and females

Dimensions Males Females
Function 1 coefficient Function 2 coefficient Function 1 coefficient Function 2 coefficient
Cranial length -0.042814 0.046077 - 0.022793 - 0.074372
Cranial breadth 0.085895 -0.102257
Max. frontal breadth -0.159036 0.076478 - 0.089974 - 0.034754
Bizygomatic breadth 0.087832 - 0.035634 0.154269 0.076963
Basion-bregma height 0.049601 - 0.065688 0.039990 0.125313
Basion-prosthion 0.040147 0.079534 0.098570 - 0.075530
Biasteronic breadth -0.105740 -0.019376 - 0.088455 0.049045
Nasal breadth 0.253434 0.245891
(Constant) -2.358715 - 1.475785 - 9.859351 - 6.808660
Group centroids Whites Blacks Japanese Whites Blacks Japanese
Function 1 - 1.438670 0.117840 1.058690 - 1.641180 0.542170 1.193500
Function 2 - 0.548220 1.591600 -0.721440 0.257780 - 1.159270 0.923690
Accurate N of total N 34/35 24/32 34/44 28/31 27/31 23/29
% accuracy 97.1 75.0 77.3 90.3 87.1 79.3

When three (unpaired) groups are considered, discriminant function statistics can be expected to yield lower separation accuracy because the effectiveness of some variables that are particularly good discrimi-natorsbetween two groupsmay be diminished by the intervening population. Using the same samples (Japanese, American Whites and Blacks), Table 5 presents discriminant function coefficients, and classification accuracy. Of nine variables that differ significantly, eight were selected by the multivariate stepwise discriminant function (all but mastoid height) for males and six (all but mastoid height, cranial breadth, and nasal breadth) for females. Accuracy averaged 82.8 for malesand 85.7 for females. As in the univariate function (Table 3), both sexes of
Whites were more successfully diagnosed than the others. Furthermore, there was no particular tendency in the distribution of misclassified cases. Although nasal breadth was the most distinguishing characteristic between Blacks and Whites, the addition of the Japanese sample (with an intermediate nose width closer to that of blacks) canceled the effectiveness of that variable in females. As expected cranial length showed more population variation than cranial breadth in this three-group analysis. Figure 2 shows the territorial placement of three samples for both sexes. Calculation of discriminant scores for both functions 1 and 2 can be made by multiplying each dimension by its coefficient and adding them up along with the constant. The resulting two scores can be plotted in this figure or compared with the group centroids given in Table 5.
There is an important caveat to the use of discriminant formulason other populations until thorough tests have been conducted. Although the Japanese are, for example, Mongoloid, they may not be representative of other Mongoloid groups such as Amerindians or even the neighboring Chinese. The same applies to using American Black standards for the much smaller southern African Blacks since there are significant mensural differences between them.
Similar studies were also done on the postcranial skeleton such as the pelvis and long bones of American Whitesand Blacks. It was shown that when several standard measurements from the pelvis, femur and tibia were combined, a person could be classified as White or Black with about 95% accuracy. The pelvisalone can differentiate Whites from Blacks by as much as92%. The femur and tibia are less discriminating but, individually and in combination, can still produce up to 88% separation between American Whites and Blacks.
There are well-known proportional differences among the races based on limb to height ratios. In Blacks, limbs are longest relative to height, and the reverse is true for Mongoloids. Whites are intermediate. This variation, which has been associated at least in part with climate, is reflected in the fact that it was necessary to calculate race-specific regression formulas to estimate stature from the long bones. This proportional variation also explains why the sternal end of the rib demonstrates White/Black racial dimorphism with Whites being significantly larger. This is consistent with other torso components such as those of the pelvis where Whites are also larger in all dimensions tested.
Population placement (via 3 group analysis) of American White and Black, and Japanese of both sexes using cranial dimensions.
Figure 2 Population placement (via 3 group analysis) of American White and Black, and Japanese of both sexes using cranial dimensions.
Recent research on the metric analysis of race in sub adults has revealed that the dimensions of the calvarium, especially cranial breadth, resembles the adult pattern of differences between Whites and Blacks after the age of 5 years. Ideally, these results will be confirmed by testing on other, larger samples.


The determination of race is a complicated task in part because humans are biologically dynamic, through both adaptation and interaction. Aspects of culture, such as advancing technology and changing attitudes, are also major factors for they bring together genetically and geographically diverse peoples. In many cases culturally and socially assigned race may not reflect the true genetic make-up of the deceased. For example, no matter how much white admixture there is in the offspring of a White-Black union, dark skin color usually dominates the physical appearance, and the general public considers this to be the race of that child. The difficulty increases as cultures change and transportation improves. Although not everyone has access to airlines, few people are more than 24 hours apart. There are, however, certain groups who have maintained their isolation from others much more successfully and their gene pool is less mixed than those of other contemporary populations. Some of these can be found in the Kalahari desert of southern Africa, remote regions of China and Siberia, islands of the Philippines, Indonesia, and New Guinea, northwestern populations of Asia, and even bands of Inuits.
As in all skeletal assessments, both morphological and metric examinations must be carried out meticulously and the results carefully considered. There are always individual differences in the size or shape of a particular structure and equally important is the interobserver variation that forms the basis for interpreting what is seen. It must also be kept in mind that expertise in one population does not always make a person equally expert in analyzing variation in others. Discriminant function statistics have also been problematic in terms of applicability to populations of similar racial origin. It is highly unlikely, for example, that the standards based on American Blacks will be applicable to their African counterparts.
In the medicolegal arena, racial placement is an essential piece of the identification puzzle. Even the most skilled expert could not attempt facial reconstruction without this knowledge as a necessary guide to soft tissue morphology. Moreover, race differences can interfere with the accurate assessment of other demographic characteristics like age, sex and stature. For example, race affectsboth age and sex determination in the rib. Ribsof Blacksare known to age at a different rate and pattern from those of Whites and thus may be assigned to the wrong age phase using White standards. And, if sexing is attempted using metric values for White ribs, the vast majority of Black males would be classified as female. Similar results would be expected using metric sexing standards for American Whiteson Asian Indiansor figuresfor American Blackson southern African Blacks. The proper regression formula must be used for statural estimation from long bones because of variation in body proportions. Height will be underestimated in a White if the formula for Black is used and overestimated in a Black using the formula for Whites.
Asnoted above, the best one can do in determining the racial phenotype of a contemporary person is to assign the individual as Caucasoid, Negroid, Mongoloid or some admixture thereof. Although diagnostic results are often accurate for most adults, further research is needed to improve the identification of race in infants and children.
In conclusion, it is nearly impossible to establish the identity of skeletal remains without determining race. Forensic examiners must be extremely cautious in making a decision on this aspect of the identification of unknown remains. Incorrect racial assignment will thwart identification by eliminating from consideration the population containing the individual in question. Forensic anthropologists and law enforcement agents must be careful not to rule out missing individuals who do not show ‘typical’ features for a given racial phenotype.

Next post:

Previous post: