- •Published, April, 1939.
- •Introduction
- •Introduction
- •Introduction
- •Introduction
- •Introduction 78-82
- •Introduction 131-135
- •Introduction 297-298
- •Introduction 400-401
- •Introduction 510-511
- •List of maps
- •Introduction to the historical study of the white race
- •Statement of aims and proposals
- •Theory and principles of the concept race
- •Materials and techniques of osteology**
- •Pleistocene white men
- •Pleistocene climate
- •Sapiens men of the middle pleistocene
- •The neanderthaloid hybrids of palestine
- •Upper palaeolithic man in europe,
- •Fig. 2. Neanderthal Man. Fig. 3. Cro-Magnon Man.
- •Aurignacian man in east africa
- •The magdalenians
- •Upper palaeolithic man in china
- •Summary and conclusions
- •Fig. 12. Fjelkinge, Skane, Sweden. Neolithic.
- •Mesolithic man in africa
- •The natufians of palestine
- •The midden-d wellers of the tagus
- •Mesolithic man in france
- •The ofnet head burials
- •Mesolithic man in the crimea
- •Palaeolithic survivals in the northwest
- •Clarke, j. G. D., op. Cit., pp. 133-136.
- •38 Fiirst, Carl m., fkva, vol. 20, 1925, pp. 274-293.
- •Aichel, Otto, Der deutsche Mensch. The specimens referred to are b 5, ks 11032, ks 11254b, b 38, b 34, b 37, b 10.
- •Clarke, j. G. D., op. Citpp. 133-136.
- •Summary and conclusions
- •The neolithic invasions
- •(1) Introduction
- •1 Childe, V. Gordon, The Dawn of European Civilization; The Most Ancient East; The Danube in Prehistory; New Light on the Most Ancient East; Man Makes Himself.
- •And chronology '
- •The neolithic and the mediterranean race
- •Vault medium to thin, muscular relief on vault as a rule slight.
- •Iran and iraq
- •Vallois, h. V., “Notes sur les Tfctes Osseuses,” in Contencau, g., and Ghirsh- man, a., Fouilles de Tepe Giyan.
- •Jordan, j., apaw, Jh. 1932, #2.
- •Keith, Sir Arthur, “Report on the Human Remains, Ur Excavations,” vol. 1: in Hall, h. R. H„ and Woolley, c. L., Al 'Ubaid,
- •10 Frankfort, h., “Oriental Institute Discoveries in Iraq, 1933-34,” Fourth Preliminary Report, coic #19, 1935,
- •Civilized men in egypt
- •11 Morant, g. M., Biometrika, 1925, p. 4.
- •12 This summary of climatic changes in Egypt is based on Childe, V. G., New Light
- •18 Childe, op. Cit.Y p. 35. 14 Leakey, l. S. B., Stone Age Africa, pp. 177-178.
- •Brunton, Guy, Antiquity, vol. 3, #12, Dec., 1929, pp. 456-457.
- •Menghin, o., Lecture at Harvard University, April 6, 1937.
- •Childe, V. G., op. Cit.Y p. 64.
- •Derry, Douglas, sawv, Jahrgang, 1932, #1-4, pp. 60-61. 20 Ibid., p. 306.
- •Morant, g. M., Biometrika, 1927, vol. 27, pp. 293-309.
- •21 Morant, g. M., Biometrika, vol. 17, 1925, pp. 1-52.
- •Morant, op. Cit., 1925.
- •Neolithic north africa
- •(6) The neolithic in spain and portugal
- •The eastern source areas: south, central, and north
- •The danubian culture bearers
- •The corded or battle-axe people
- •The neolithic in the british isles
- •Western europe and the alpine race
- •Schlaginhaufen, o., op. Cit.
- •Schenk, a., reap, vol. 14, 1904, pp. 335-375.
- •Childe, The Danube in Prehistory, pp. 163, 174.
- •Neolithic scandinavia
- •Introduction
- •Bronze age movements and chronology
- •The bronze age in western asia
- •The minoans
- •The greeks
- •Basques, phoenicians, and etruscans
- •The bronze age in britain
- •The bronze age in central europe
- •The bronze age in the north
- •The bronze age on the eastern plains
- •The final bronze age and cremation
- •Summary and conclusions
- •Race, languages, and european peoples
- •The illyrians
- •The kelts
- •Vallois, h. V., Les Ossements Bretons de Kerne, TouUBras, et Port-Bara.
- •We know the stature of Kelts in the British Isles only from a small Irish group, and by inference from comparison with mediaeval English counterparts of Iron Age skeletons.
- •Greenwell, w., Archaeologia, vol. 60, part 1, pp. 251-312.
- •Morant, g. M., Biometrika, 1926, vol. 18, pp. 56-98.
- •The romans
- •46 Whatmouffh. J., The Foundations of Roman Italy.
- •The scythians
- •88 Browne, c. R., pria, vol. 2, ser. 3, 1899, pp. 649—654.
- •88 Whatmough is in doubt as to their linguistic affiliation. Whatmough, j., op. Cit., pp. 202-205.
- •Fig. 29. Scythians, from the Kul Oba Vase. Redrawn from Minns, e. H., Scythians and Greeks, p. 201, Fig. 94.
- •Doniti, a., Crania Scythica, mssr, ser. 3, Tomul X, Mem. 9, Bucharest, 1935.
- •The germanic peoples
- •Stoiyhwo, k., Swiatowit, vol. 6, 1905, pp. 73-80.
- •Bunak, V. V., raj, vol. 17, 1929, pp. 64-87.
- •Shetelig, h., Falk, h., and Gordon, e. V., Scandinavian Archaeology, pp. 174-175.
- •70 Hubert, h., The Rise of the Celts, pp. 50-52.
- •71 Nielsen, h. A., anoh, II Rakke, vol. 21, 1906, pp. 237-318; ibid., III Rakke, vol. 5, 1915, pp. 360-365. Reworked.
- •Retzius, g., Crania Suecica, reworked.
- •78 Schliz, a., pz, vol. 5, 1913, pp. 148-157.
- •Barras de Aragon, f. De las, msae, vol. 6, 1927, pp. 141-186.
- •78 Hauschild, m. W., zfma, vol. 25, 1925, pp. 221-242.
- •79 Morant, g. M., Biometrika, vol. 18, 1926, pp. 56-98.
- •8° Reche, o., vur, vol. 4, 1929, pp. 129-158, 193-215.
- •Kendrick, t. D., and Hawkes, c. F. C., Archaeology in England and Wales, 1914-1931.
- •Morant, Biometrika, vol. 18, 1926, pp. 56-98.
- •Lambdoid flattening is a characteristic common to Neanderthal and Upper Palaeolithic man, but rare in the exclusively Mediterranean group.
- •Calculated from a number of series, involving over 120 adult males. Sources:
- •Peake, h., and Hooton, e. A., jrai, vol. 45, 1915, pp. 92-130.
- •Bryce, t. H., psas, vol. 61, 1927, pp. 301-317.
- •Ecker, a., Crania Germanica.
- •Vram, u., rdar, vol. 9, 1903, pp. 151-159.
- •06 Miiller, g., loc. Cit.
- •98 Lebzelter, V., and Thalmann, g., zfrk, vol. 1, 1935, pp. 274-288.
- •97 Hamy, e. T., Anth, vol. 4, 1893, pp. 513-534; vol. 19, 1908, pp. 47-68.
- •The slavs
- •Conclusions
- •The iron age, part II Speakers of Uralic and Altaic
- •The turks and mongols
- •I® Ibid.
- •Introduction to the study of the living
- •Materials and techniques
- •Distribution of bodily characters
- •Distribution of bodily characters
- •Distribution of bodily characters
- •2. Skin of tawny white, nose narrow,
- •Hair Flaxen
- •Gobineau, a. De, Essai sur Vinegaliti des races humaines.
- •Meyer, h., Die Insel Tenerife; Uber die Urbewohner der Canarischen Inseln.
- •46 Eickstedt, e. Von, Rassenkunde und Rassengeschichte der Menschheit.
- •Nordenstreng, r., Europas Mdnniskoraser och Folkslag.
- •Montandon, g., La Race, Les Races.
- •Large-headed palaeolithic survivors
- •Pure and mixed palaeolithic and mesolithic survivors of moderate head size56
- •Pure and mixed unbrachtcephalized mediterranean deriva tives
- •Brachtcephauzed mediterranean derivatives, probably mixed
- •The north
- •Introduction
- •The lapps
- •I Wiklund, k. B., gb, vol. 13, 1923, pp. 223-242.
- •7 Schreiner, a., Die Nord-Norweger; Hellemo (Tysfjord Lappen).
- •8 Gjessing, r., Die Kautokeinolappen.
- •10 Kajava, y., Beitr'dge zur Kenntnis der Rasseneigenschaften der Lappen Finnlands.
- •17 For a complete bibliography of early Lappish series, see the lists of Bryn, the two Schreiners, Geyer, Kajava, and Zolotarev.
- •Schreiner, k. E., Zur Osteologie der Lappen.
- •Gjessing, r., Die Kautokeinolappen, pp. 90-95.
- •Hatt, g., Notes on Reindeer Nomadism, maaa, vol. 6, 1919. This is one of the few points regarding the history of reindeer husbandry upon which these two authorities agree.
- •The samoyeds26
- •Scandinavia; norway
- •Iceland
- •Sweden64
- •Denmark62
- •The finno-ugrians, introduction
- •Fig. 31. Linguistic Relationships of Finno-Ugrian Speaking Peoples.
- •Racial characters of the eastern finns
- •The baltic finns: finland
- •The baltic-speaking peoples
- •Conclusions
- •The british isles
- •R£sum£ of skeletal history
- •Ireland
- •Great britain, general survey
- •Fig. 32. Composite Silhouettes of English Men and Women.
- •The british isles, summary
- •Introduction
- •Lapps and samoyeds
- •Mongoloid influences in eastern europe and in turkestan
- •Brunn survivors in scandinavia
- •Borreby survivors in the north
- •East baltics
- •Carpathian and balkan borreby-like types
- •The alpine race in germany
- •The alpine race in western and central europe
- •Aberrant alpine forms in western and central europe
- •Alpines from central, eastern, and southeastern europe
- •Asiatic alpines
- •The mediterranean race in arabia
- •Long-faced mediterraneans of the western asiatic highlands
- •Long-faced mediterraneans of the western asiatic highlands: the irano-afghan race
- •Gypsies, dark-skinned mediterraneans, and south arabian veddoids
- •The negroid periphery of the mediterranean race
- •Mediterraneans from north africa
- •Small mediterraneans of southern europe
- •Atlanto-mediterraneans from southwestern europe
- •Blue-eyed atlanto-mediterraneans
- •The mediterranean reemergence in great britain
- •The pontic mediterraneans
- •The nordic race: examples of corded predominance
- •The nordic race: examples of danubian predominance
- •The nordic race: hallstatt and keltic iron age types
- •Exotic nordics
- •Nordics altered by northwestern european upper palaeolithic mixture: I
- •Nordics altered by northwestern european upper palaeolithic mixture: II
- •Nordics altered by mixture with southwestern borreby and alpine elements
- •The principle of dinaricization
- •European dinarics: I
- •European dinarics: II
- •European dinarics: III
- •European dinarics: IV
- •Dinarics in western asia: I
- •Dinarics in western asia: II
- •Armenoid armenians
- •Dinaricized forms from arabia and central asia
- •The jews: I
- •The jews: II
- •The jews: III
- •The mediterranean world
- •Introduction
- •The mediterranean rage in arabia
- •The mediterranean world
- •7 Lawrence, Col. T. E., The Seven Pillars of Wisdom.
- •The Distribution of Iranian Languages
- •The turks as mediterraneans
- •Fig, 37. Ancient Jew.
- •North africa, introduction
- •Fig. 38. Ancient Libyan. Redrawn from
- •The tuareg
- •Eastern barbary, algeria, and tunisia
- •The iberian peninsula
- •The western mediterranean islands
- •The basques
- •The gypsies
- •Chapter XII
- •The central zone, a study in reemergence
- •Introduction
- •8 Collignon, r., msap, 1894.
- •9 Collignon, r., bsap, 1883; Anth, 1893.
- •Belgium
- •The netherlands and frisia
- •Germany
- •Switzerland and austria
- •The living slavs
- •Languages of East-Central Europe and of the Balkans
- •The magyars
- •The living slavs (Concluded)
- •Albania and the dinaric race
- •The greeks
- •Bulgaria
- •Rumania and the vlachs
- •The osmanli turks
- •Turkestan and the tajiks
- •Conclusions
- •Conclusion
- •Comments and reflections
- •The white race and the new world
- •IflnrlrH
- •Alveon (also prosthion). The most anterior point on the alveolar border of the upper jaw, on the median line between the two upper median incisors.
- •Length of the clavicle (collar bone) and that of the humerus (upper arm bone);
- •Incipiently mongoloid. A racial type which has evolved part way in a mongoloid direction, and which may have other, non-mongoloid specializations of its own, is called incipiently mongoloid.
- •List of books
- •Index of authors
- •54; Language distribution, 561, map; Jews in, 642; Neo-Danubian, ill., Plate 31, Jig. 4.
- •Map; classified, 577; racial characteristics, 578-79; ill., Plate 3, fig. 3.
- •Ill., Plate 6, Jigs. 1-5; survivors in Carpathians and Balkans, ill., Plate 8, figs. 1-6; Nordic blend, ill., Plate 34, figs.
- •61; Associated with large head size, 265, 266. See also Cephalic index, Cranial measurements.
- •Ill., Plate 36, fig. 1. See also Great Britain, Ireland, Scotland.
- •Ill., Plate 30, fig. 2.
- •85; Von Eickstedt’s, 286-88; Gzek- anowski’s system, 288-89; author’s, 289-96; schematic representation, 290, chart; geographic, 294- 95, map.
- •396; Cornishmen in France, 512, 514.
Chapter
VIII
At
this point we have completed the survey in which, with the help of
the combined disciplines of osteology, archaeology, history, and
linguistic science, we have attempted to trace the development of
racial entities in the territory occupied by the white race, from
the earliest human times to the Middle Ages, the threshold of the
modern period. We are now faced with the problem of working with a
different body of material— that furnished by the anthropometry of
living peoples. We must further attempt to fit this material into
the frame furnished us by our study of the dead, so that from the
combination of the two a complete and orderly reconstruction will
result.
While
we were dealing with the data gleaned from the measurement and
observation of bones, the chief difficulty which faced us was the
lack of adequate samples in most of the periods, regions, and
cultural units under consideration. On the other hand, while
metrical accuracy was by no means to be assumed, yet the
measurements on the dry skulls and long bones were for the most part
comparable, and technical difficulty was subordinate to the paucity
of documents. In dealing with the living material, however, we
have vastly larger samples. In some countries, as in Norway, Sweden,
and Poland, these comprise the entire military age group of the
nation, and thus cease to be samples in the strict sense, and assume
the character of total populations. In relatively few regions is it
necessary to use samples of less than one hundred individuals.
Our
authority has, therefore, increased immensely. We may speak with
some confidence of the superficial physical composition of most
European nations. But, at the same time, what we have gained in
volume, we have ' to a certain extent lost in accuracy, for the
present state of anthropometry is partly one of confusion and
mistrust in regard to technical methods. Despite various attempts in
the past and in the present to establish a standard corpus of
technique,1 different schools have arisen in different
countries. What discrepancies may exist between the work of members
of each school
241Introduction to the study of the living
Materials and techniques
Cf.
The Geneva agreement of 1912; the standards established by R.
Martin in his Lehrbuch
der Anthropologie;
the present laudable attempt of Miss Miriam Tildesley to bring
about unification.
242
THE
RACES OF EUROPE
can
usually be determined and allowed for; but this is not the root of
the trouble. The chief difficulty is that much measuring has been
done not by professional anthropometrists but by amateurs, while
some with professional status have not been properly trained.
Therefore we cannot be sure that such men belong to any school, nor
that they follow any standard other than their own. The
accuracy of existing documents on the living is far less than that
of skeletal data, and it is not always possible to know what
techniques have been used. This lack of consistency is often an
obstacle to mathematical comparison, but not enough of an obstacle
to render many series wholly useless. We still have a better tool
for the study of race in the living than we had in the documents of
the dead.
Let
us review the more important measurements in which technical
difficulties most commonly arise. Stature, unfortunately, heads the
list. One would suppose that the maximum height of the body while
standing would be a constant dimension and one easy to measure, but
neither assumption is true. Some investigators allow the subject to
be measured in his shoes, and then attempt to make a standard
subtraction for the heel. This is seldom if ever satisfactory. On
the other hand barefooted negroes with horny soles are raised up
several millimeters by their callouses, when compared to thin-soled
white men standing with their shoes removed. Differences in posture,
and in degree of conscious stretching, may attain the dimensions of
centimeters.
Furthermore,
it has been established 2 that the human body, except in
senility, shrinks as much as 2.5 cm. during a daytime spent either
afoot or in a chair, the amount depending partly on the degree of
and nature of the day’s activity. It makes some differences,
therefore, what time of day the investigator habitually chooses for
his work. At the same time the state of nutrition and of health
makes some difference, and one must beware of series measured
entirely in hospitals.
For
the reasons above outlined, and without doubt for others as well, we
must not, in studying stature as a statistical criterion of racial
value, even if our samples are equivalent in age, expect to find
accuracy down to the millimeter. Therefore the common statistical
devices used to check the validity of the series on the basis of the
sampling process are set at too fine an adjustment in view of the
coarseness of the measurement itself, and in view of the great
variability caused by factors other than sampling or racial
attributes. What applies to stature applies in varying degree to
measurements of its segments and of other bodily dimensions; the
breadths of the shoulder and hips, and the diameters of the chest,
are dependent in some degree on the highly variable amounts of
sinew, muscle, and fat present at the points of measurement.
2
Backman, G., FUL, N. F. vol. 29, 1923-24, pp. 255-282.
INTRODUCTION
TO THE LIVING
243
In
the dimensions of the head and face, most of the difficulties found
in stature and bodily measurements cease to exist. On the whole, a
much greater accuracy is not only possible but has been attained.
There are but two important matters in which serious inaccuracies
arise with any frequency; these are the measurement of auricular
head height and the location of nasion.
The
first of these, the measurement of the height of the cranial vault,
is without doubt the least satisfactory of all common anthropometric
techniques. Although technique #15 of Martin 3 is
considered standard, not all use it, and few do it in the same way.
Some investigators use special metal head-spanners which measure the
height of the vault from the middle of the ear hole, others measure
from the top of the ear hole; still others, following Martin, from
tragion. There is also a dispute as to whether the height taken
should be to the vertex, as stated by Martin, or to a point exactly
above the ear hole when the head is held in an approximation to
the eye-ear plane.
As
a result of these technical difficulties in taking head height on
the living, differences of from ten to fifteen millimeters exist
between the results of different investigators working on
identical populations, and reports embodying these
discrepancies are published without comment. Since the difference
between techniques is as great as the difference between
extremely disparate racial groups of mankind, head height on the
living is a useless criterion when employed uncritically. Unless the
compiler knows the technical peculiarities and personal
equation of each investigator whose work he uses, he should leave
this material alone. In the present work, this ruling immediately
excludes from consideration the majority of published data on head
height.
The
second major difficulty, the location of nasion in the living, while
not quite as inaccurate, is even more serious, since three important
vertical diameters of the face, morphological face height,
morphological upper face height, and nose height, are theoretically
limited, at their upper boundary, by this landmark; and nasion
is an extremely hard point to determine. Ashley-Montagu, however,
has recently devised a method which promises to overcome this
difficulty in most cases.4 On adult male whites, luckily,
there is usually enough ruggedness of facial relief to make this
difficulty less serious than with mongoloids or negroids. Still
technical differences of from five to ten millimeters render the
works of different investigators incomparable, and one must again be
sure of the individual equation of each investigator, or of the
school in which he was trained. Since the facial
8
Martin, R., Lehrbuch
der Anthropologie,
vol. 1, pp. 185-186.«Ashley-Montagu,
M. F., AJPA, vol. 20, 1935, pp. 81-93; vol. 22, 1937, #3, Suppl. p.
6.
244
THE
RACES OF EUROPE
and
nasal indices depend upon vertical as well as lateral diameters, and
hence upon nasion, these important racial criteria must be taken
with great reserve, for the constancy of the lateral diameters
serves only to exaggerate, in the indices, the differences between
the vertical dimensions.
So
much for the most serious metrical difficulties. In measurements on
the living we see a more bountiful but less accurate counterpart of
the criteria already familiar to the craniologist. There is another
large body of data, however, unique in living material; the
observations on the soft parts, including such features as hair
form, hair texture; skin, hair, and eye color; the shape of the
various component segments of the nose, the lips, and the external
eye. These are important diagnostic racial characters and
deserve as careful study as do measurements and indices. But,
unfortunately, accurate comparisons between the work of different
investigators is even less possible here than with metrical
data, since observation is a matter of judgment, and no two
men’s judgments are the same.
The
use of standard pigment scales in determining hair, skin, and eye
color has helped enormously, but has not entirely eliminated the
difficulties in the pigmentation field. There is no really
adequate eye-color scale on the market, although Martin’s series
of sixteen glass eyes is far better than nothing. Von Luschan’s
skin-color scale does not always approximate human shades, and this
is especially true with whites. The Saller-Fischer hair-color scale,
made from actual human hair, is excellent, in most respects,
but has not yet come into common use; the earlier Fischer scale,
made of bleached and dyed vicufia hair, is also good.
Unfortunately,
however, the majority of our observational data has been collected
without reference to scales, and published without accurate
definitions, and it is impossible to tell, in many instances, what
color or what degree of blondism or pigmentation is implied by a
given term. Then too, environment and age make great differences in
pigmentation; the degree of tanning or of uncleanliness in regard to
the skin color is seldom indicated; eyes often grow lighter with
age, and the deposit of fat in the cornea, called arcus
senilis,
which gives a grayish-blue tone to the peripheral zone of the iris,
is often mistaken for eye blondism. Hair color is notoriously
transitory, changing, in all but pure brunets and extreme blonds,
continuously from birth to grayness, baldness, or death.
Most
observations, other than those referring to pigmentation and the
morphology of the pilous system, are divided into the following
categories: absent, sub-medium, medium, pronounced. These are
frequently expressed by the symbols, abs., sm., +> ++•
Often ssm. and +++ are added for greater refinement. In general, the
standard for the + or medium category is a roughly estimated and
ideal mean or intermediate white or European male condition. Thus in
nasal tip thickness almost all
INTRODUCTION
TO THE LIVING
245
negroes
would be ++ or H—f- + ; in beard development almost all Eskimos
would be abs., ssm., or sm. There is a tendency for the observer to
make the mean condition of the people he is studying + or medium, or
to be unconsciously influenced by his own facial form.
Various
attempts have been made to standardize these quantitative
observations, and the most promising is perhaps that of the Moscow
school, where a series of plaster casts has been made to show
standard stages of sm., +, and H—(-in each of the more commonly
studied criteria. Still, whatever standards are used, the location
of the borderline between categories must always be a matter of
individual judgment.
Our
first difficulty with the study of race from existing data on living
populations, whether these data be metrical or observational, is
therefore one of technical inaccuracy and inconsistency. But it is
not the greatest difficulty which will be encountered, and it is not
insuperable. The careful compiler can usually discover what are
the technical idiosyncracies of a given investigator, and if he is
familiar with the material as a whole, he can usually sense
improbable divergences from standard technique. The comparison of
different samples selected from the same population by different
investigators often makes a standard adjustment possible.
Technical
inconsistencies and inaccuracies render the study of race on the
living something less than an exact science, but it remains
something more than a plaything. The manipulation of metrical data
requires experience and judgment, and the uncritical use of
existing materials on a purely statistical basis, no matter how
erudite in the mathematical sense, can never be more than a sterile
exercise. Those who employ experience and judgment, and who make a
discreet use of the simpler statistical methods, may learn much from
the handling of the immense body of anthropometric data.
In
the introduction to the study of the skeletal material, we made only
the briefest mention of the statistical methods to be employed in
that segment of the book.6 This was done because the
numerical size and the nature of the cranial samples employed
limited the treatment, in most cases, to a discussion of individual
crania and to a comparison of simple means. With the living material
however, the use of much larger samples, and of non-metrical soft
part criteria, will necessitate reference to more elaborate methods,
and therefore a brief allusion to the better known statistical
principles and techniques which are commonly employed seems
indicated.
Modern
physical anthropology, in company with other technical and
THE
USE OF STATISTICS IN PHYSICAL ANTHROPOLOGY
ChaDter
I. dd.
14—15.
246
THE
RAGES OF EURpPE
biological
disciplines, has entered a stage of increasing dependence
upon
mathematics, and lengthy formulae which involve the use of
several alpha-
bets are currently employed by most physical
anthropologists. Although
there are several schools each of
which has assembled a favorite collection of
symbols, the
method as a whole is a product of the English biometric
school
founded by Gal ton and Pearson. Aside from the
calculation of means,
the purposes for which these formulae and
numerical techniques are em-
ployed may be reduced to four,
which, expressed in the simplest possible
form, are as follows:
To
determine the degree of homogeneity or heterogeneity of a given
statistical
sample,
in the
various criteria measured or observed,
and to
compare it in these re-
spects with other samples.
To
determine whether or not two statistical samples may be
considered
random selections from a single population.
Having
found that the two samples represent demonstrably different
popula-
tions,
to
determine exactly how different,
in a
metrical sense,
they are.
To
determine whether or not a given sample is racially mixed,
and if it
is,
to
discover
its component elements.
Let
us review these four purposes and the techniques by which they
are
accomplished, in as simple and brief a manner as possible.
To
study the relative variability of samples.
This is done by means of the
two constants, standard deviation
and coefficient of variation.6 The former,
in which
the variability of the extremes is emphasized by the
quadratic
treatment, indicates how many unit points the
average individual in the
sample deviates from that mean. When
used to compare approximately
equivalent means within the same
criterion, it is a simple and useful
constant. The coefficient
of variation is designed to facilitate comparison
between
criteria in which the metrical values of the means are
quite
different, in order to eliminate the size element. By
comparing <r’s and
V’s of a given sample with those of
a general compilation, such as that of
Howells,7
one may gauge the relative variability of the sample, and
may
compare it with other specific samples in this regard.
This technique is
100
• <r
V
fx2
Vn
V2N
—
-
A2.
V (Coefficient of Variation) =P.
E. M. (Probable error of the mean) =M.67450-.6745<rP.
E. <r (Probable error of the standard deviation) =P.
E. V. (Probable error of the coefficient of variation) =V2N.6745VP.
E. Diff. « VP. E. M.i
+ P. E. M.*
Howells,
W. W., HB, vol. 8, 1936, #4, pp. 592-600.
INTRODUCTION
TO THE LIVING
247
not
by its nature limited to living material, but it may be profitably
employed with many more published series of the living than of
crania.
To
test the statistical independence of two samples.
The second purpose is, in effect, to tell whether or not two
samples may be considered separate statistical entities. The
technique most commonly employed is to compare the difference
between two means with the probable error * of that difference. If
the difference is three times or more its probable error, then the
two samples are considered distinct in the criterion under study.
If, in a large number of criteria, the two samples are consistently
distinct, then two separate populations are represented. If, on the
other hand, the two samples are not distinct, owing to the relative
smallness of differences compared to their probable errors, then we
may make one of the following deductions: (a) the two groups
represent the same anthropometric population; (b) the two
groups are really different, but owing to the small numerical size
of one or both samples, or to the excessive variability of one or
both, such a difference cannot be established statistically.
In
order to determine which of these two premises is the more likely,
the exercise of judgment must inevitably be interpolated. If both
samples are large and of reasonable variability, the two are
probably, in fact, alike; if both are very small and the probable
errors large, the chances are great that the samples are
statistically worthless. The chief utility of the sampling check,
therefore, is to find out whether or not apparent differences
are really of significance. It is not an automatic proof of
identity.
To
measure the anthropometric difference between samples.
The third purpose, to tell how close or how distant two
samples are in a metrical sense, may be fulfilled in any one of a
number of ways. One is merely to compare the means, and to compute
the differences. Then, for convenience, one may pool the
differences for separate statistical categories. For example, the
difference between sample A and sample B in head length may be 4.35
inm.; in head breadth 7.32 mm.; ip head height 1.09 mm. The
average difference in three vault diameters is therefore 4.19
mm. The average for the same three diameters, between sample A and
sample C, on the other hand, may be 9.73 mm. Therefore we may say
that sample A resembles sample B, in the totality of three vault
diameters, more than it resembles sample C. Similarly one may pool
the vault indices, or the head and face measurements, or the head
and face indices, but one may not average measurements and indices
together. To do so would be to commit the kindergarten fallacy of
adding oranges and apples. But there are anthropologists who
have not only done this, but who have also added centimeters
and millimeters together as equal units, in pooling body and head
measurements.
*
See footnote 6 on preceding page.
248
THE
RACES OF EUROPE
It
has long been the wish of many anthropologists to find some means
whereby they might express the degree of similarity of difference
between two populations by a single figure. Taking population A as
zero, B would be, say, 5.6; C = 7.3; D = 11.9. Thus the
relationships of B, C, and D in respect to A could be determined.
Taking each of the others in turn, it would be possible to
triangulate and to plot the mutual relationships of any number of
populations in a simple, graphic manner. Morant, working with a
formula invented by Pearson, has proposed and employed such a method
in the form of the coefficient of racial likeness.8 Some
have accepted this in principle, others have rejected it.9
Whatever its theoretical validity or error, however, it does
actually give approximately the same results as a simple pooling of
the several categories of differences. Unfortunately neither a
simple pooling nor the coefficient of racial likeness takes into
account correlative influences which compel several characters to
vary in concert, and thus to weight, in a variable degree, the
totality of characters chosen. According to Morant, these
correlative influences could be eliminated, but only by an
unfeasible amount of statistical labor.
Before
proceeding to the fourth purpose, let us pause to make a few
reflections upon the uses to which the three systems already
outlined may be put. Although all are useful, not one automatically
answers any important questions. The first technique, that
which is concerned with variability, tells us how variable samples
are, but not why they are variable. Unusual variability may
indicate an active evolutionary tendency, the recent and as yet not
fully amalgamated mixture between two populations, or any one
of a number of other causes. Unusual homogeneity, on the other hand,
does not necessarily mean racial “purity,” in the historical
sense, but rather a complete amalgamation and a static evolutionary
condition. The second is useful mainly to eliminate from serious
consideration statistically inadequate samples. The third gives
a detailed idea of degrees of metrical similarity and difference.
But neither the second technique nor the third tells the
investigator what is the genetic
relationship between two samples.
To
analyze a racially mixed sample.
Let us now turn to the fourth and last important use which the
physical anthropologist makes of statistics. This is his attempt to
divide a given sample, which he considers to have resulted from a
mixture of races, into its component elements, and to see what
these elements are and how much there is of each in the mixture.
This is a rather complicated process, and many different methods
have been devised to accomplish it.
8 Morant,
G. M., Biometrika, vol. 14, 1923, pp. 193-264; vol. 16, 1924, pp.
1-105.
Pearson,
K., Biometrika, vol. 18, 1926, pp. 105-117.
•
Fisher,
R. A., JRAI, vol. 66, 1936, pp. 57-63.
INTRODUCTION
TO THE LIVING
249
One
is the system employed by Hooton and his school, in which the author
was trained. That is for the anthropometrist, working either with
crania or with the living, to divide his series into what seem to
him natural groupings, and to specify on each measurement blank
which of these types is represented. After the sample has been
seriated as a whole, the sub-samples of the different types are
seriated separately, and statistically compared with each other and
with the total mean. By this means it may be determined whether or
not statistically different elements have actually been
isolated. If so, the next step is to determine, by comparison, what
the larger racial relationships of these elements are.10
Hooton bases his system on the principle that the individual
possesses a racial identity, as well as does the group to which he
belongs.
Another
method which is less subjective but wholly arbitrary is that of
Czekanowski, who plots the mean differences between individuals in a
sample on a chequered field; this is done only with indices of the
head and face, when the original system is followed.11
Two individuals alike in all indices chosen produce a black square
at the point where their lines intersect; two which are less
alike produce a square which is striped, in varying degrees arranged
to show the degree of similarity; then those which are dissimilar in
all indices are represented by white squares. After these squares
have been completely plotted, the graph is rearranged so that those
which are naturally related are placed in contiguous positions. In
this way it is possible to see how many sub-groups of naturally
correlated individuals occur, and how large these sub-groups are.
The next step is to find the racial affinities of each sub-group.
For this purpose the Polish school has designated a formal list of
races, each symbolized by a separate Greek letter, and each equipped
with a list of ideal metrical positions in the more commonly used
measurements and indices, as well as with a characteristic pigmental
position. Each group of correlated black squares in the graph is
assigned to one of these races, or to a combination of two or more,
and the percentages of each race in the sample is thus worked out.
A
third method is that originated by von Eickstedt, the leader of the
Breslau School, and amplified by Schwidetzky.12 This
method is to sort the sample directly into sub-series by splitting
the distribution frequencies of the characters at arbitrary
racial boundaries, and by combining the results of this process as
applied to pairs of characters; to plot the
Hooton,
E. A., The
Ancient Inhabitants of the Canary Islands; Indians of Pecos;
Science,
vol.
63, 1926, p. 75.
Czekanowski,
J., MAGW,
vol.
42, 1912, pp. 17-217; AASF*
ser.
A., vol. 25, #2, Helsinki, 1925; AAnz, vol. 5, 1928, pp. 335-359.Eickstedt,
E. von, ZFRK,
vol.
2, 1935, pp. 1-32.Schwidetzky,
I., ZFRK,
vol.
2, 1935, pp*. 32-40; vol. 3, 1936, pp. 46-55.
250
THE
RAGES OF EUROPE
distribution
curves of the sub-series, so-created, for measurements, indices, and
percentages of observations; and to test the sorting by a comparison
of these curves with others which represent arbitrary racial norms.
Like all such systems, this one operates on the assumption that the
result of
A
+ B
mixing
A + B in any metrical character is —
The
three methods outlined above are all based on the principle of
correlation. Correlation statistics alone are even more commonly
used than any of these. One may correlate metrical characters with
each other; metrical characters with indices; either metrical
characters or indices with observations, and observations with each
other. By means of these correlation statistics one finds which
characters are associated, in the sense that their variations are
not mutually independent. One finds, for example, that light eyes
are usually if not always correlated with light hair. The elements
of blondism are to a certain extent linked. One will also find that
segments of a dimension are positively correlated with that
dimension, but this is of no racial significance. If they are
not correlated, or are negatively correlated, then there is
something to investigate. One must furthermore expect all gross size
diameters to be intercorrelated to some extent in any population,
for obvious reasons.
Correlations
of racial significance are those which are not dependent on gross
size and are not involved in a part-and-whole relationship. Thus, if
tall stature goes with blond hair and short stature with dark hair,
or if a broad nose goes with a low relative sitting height, and vice
versa, then the anthropologist who is analyzing his series assumes
that he has uncovered linkages showing racial variations within
his sample.
There
is no possible objection to the use of correlations, but there are
many objections to the ways in which they are often interpreted. In
the first place, a valid correlation implies some degree of genetic
linkage. But it does not necessarily imply that this linkage
represents with complete fidelity a combination found in one
component element in a hypothetical mixture. There may have
been no mixture at all—the group may be evolving, by mutation, in
a certain direction which involves more than a single character. Or
if there has been mixture, the correlation may represent a
recombination of characters.
Correlation,
in brief, shows linkage, but what does linkage mean? We must not
forget that a population, in the physical as well as in the social
sense, has an existence of its own in addition to and above the
existences of its component units, and we must not, furthermore,
anticipate the findings of the geneticists. All of the methods which
partition a series, or which employ the principle of correlation,
have some justification in their initial steps, and some utility,
but all of them become unscientific as soon