A CRITICAL ANALYSIS OF THE BANTU PELVIS,
WITH SPECIAL REFERENCE TO THE FEMALE.

Thesis submitted for the Doctorate in Science
to the Department of Anatomy,

University of the Witwatersrand.

By
O.S. HEYNS

Lecturer, Department of Obstetrics,
University of the Witwatersrand.

November. 19^5

ACKNOWLEDGMENTS

I am indebted to the following members of university
and hospital staffs who have given me much appreciated
help and advice.

Professor R.A. Dart has put his department at my
disposal, including his invaluable Bantu skeletal
material. I wish to express a word of appreciation
concerning this Bantu Museum which has been built up
under great difficulties’, but with a clear vision of
how much it would mean to the many workers like myself.
Professor Dart has animated me with the feeling that he
never tired of wishing to help me accomplish something
of value, and he was prodigal of his time in reading
my writings.

Prflfessor J.P. Dalton, Head of the Dept, of
Mathematics, was kind enough to give advice personally
on several little statistical matters, but more parti
cularly on the method by which use might be made of
pelvic data in the foetus. An incidental result of our
talks was the feeling of inspiration which one had
unobtrusively gained.

- Professor -

T A B L E O F C O N T E N T S
Pag#

PART I.
I. INTRODUCTION......................... 1
II. HISTORICAL SURVEY

1. Anatomical
(a) South Africa................ 3
(b) Elsewhere..... ........ 7

2. Obstetrical.................... 6

III. SUMMARY.............................. 11

PART II. MATERIAL AND METHODS
I. INTRODUCTORY NOTE.................... 13
II. MATERIAL............................. l6
III. METHODS

1. General......................... 17
2. Measurements.................... 203. Analysis of Data................ 26

PART III. CLASSIFICATION
I. INTRODUCTORY NOTE.................... 34

II. VARIOUS CLASSIFICATIONS.............. 35
III. CALDWELL AND MOLOY CLASSIFICATION

OF BANTU PELVES..................... 39
IV. EVOLUTION OF THE HUMAN PELVIS:

CRITICISM OF CALDWELL AND MOLOY1S
VIEW................................ 43

COMMENTARY................. 51

Page
PART IV. THE PELVIC GIRDLE AS A WHOLE
A. CERTAIN EXTERNAL MEASUREMENTS OF THE

PELVIC GIRDLE............................. 54-
B. THE PELVIC BRIM AND CAVITY................ 73
C. MEASUREMENTS AFFECTING THE OUTLET.......... 91
D. RESULTS OF X-RAY MEASUREMENTS............ 102

SUMMARY 110
NOTE ON FURTHER RESULTS OF EXPERIMENTS...
IN X-RAY PELVIMETRY..................... 111a

E. MISCELLANEOUS
I. THE PUBIC ARCH....................... 112
II. SPLAY OF THE SIDE WALLS.............. 115
III. THE ISCHIAL SPINES................... Il6

SUMMARY.................................. IIS

PART V. THE SACRUM
I. DIMENSIONS OF THE SACRUM............... 123
II. NUMERICAL CONSTITUTION................. 14-1
III. SHAPE OF LOWER END OF SACRUM........... 14-5
IV. THE NUMBER OS SACRAL VERTEBRAE IN

DIRECT RELATION TO THE SACRAL AURICULAR
SURFACE............. 1*7

V. SACRAL NOTCH........................... 150

VI. PREAURICULAR SULCUS............. 153

VII. INCLINATION OF SACRUM.................. 157
SUMMARY................................... 158

CONTENTS

VI.
CONTENTS

PART VI. THE FOETAL PELVIS......
Analysis of Foetal Data.............
Results

Pelvis.......................
Saorum and Sacro-sciatio Notoh
Indices......................The Pubic Arch.............. .

Page
l6l
161*.

173
190
207
213

Pelvis ©f the Infant and Child.... ......... 223

SUMMARY............................. 226

PART VII. GENERAL DISCUSSION
A. SEXUAL DIFFERENCES IN THE PELVIS...... 23O
B. THE ADULT FEMALE PELVIS IN GENERAL.... 240
C. THE PELVIC BRIM INDEX................. 2^7

Ontogeny and Phylogeny.............. 2^9
Commentary............. 254

D. COMMENT ON 4- AND 6-PIECE SACRA....... 259
Cartilaginous Lateral Mass.......... 267Phylogenetic Relationships between
different Primates.................. 272

E. ROSENBERG'S THEORY.................... 263

PART VIII. GENERAL SUMMARY
I. OUTLINE OF MATERIAL AND METHODS...... 233
II. GENERAL CONCLUSIONS.................. 239

BIBLIOGRAPHY, 29^

APPENDIX

VII.
APPENDICES

1. reprint In cover.
Studies in X-ray Pelvimetry etc.

2. reprint in cover.
Birth-weight of Urban Bantu etc.

3.
The Size of the Bantu Head at Birth. p.JOS

k.
Histological Preparation of Tissue
comprising the Lateral Wall of the
5th Sacral Foramen in the Foetus. P»3^7

5.
Foetal Pelvic Measurements.

6.
Analysis of the Dimensions of the
Series of Bantu Pelves.

PART I
1.

I INTRODUCTION

The student of pelvic morphology is compelled
sooner or later to consider the two main functions
of the girdle, those of weight-bearing and parturition.
These functions Inevitably guide his Investigations.
Comparatively little work has been done on the pelvic
mechanics of erect posture. The evolution of man's
erect posture, on the other hand, has been considered
by Friedenthal (1910), Weidenreich (1913), Keith (1923),
Morton (1926), Westenhofer (1929), and Reynolds (1931)*
The last author's disappointing paper succeeds only in
demonstrating the difficulty of establishing the prin
ciples underlying the orthopaedist's approach to the
mechanics of man's posture.

Parturition, the second pelvic function, has
throughout the ages stimulated scientific thought with
greater force and much more constantly than has the
weight-bearing function. The reason for this is that
trauma of the vertebral column and lower limb girdle
has always been much less frequent than aberrations
during parturition; any parturient indeed may make

- some -

2
some call on a full knowledge of labour mechanisms,
and the mind of man has been occupied since the dawn
of civilisation - and perhaps before - with the
mysteries of childbirth problems.

In the first place it was the desire to become
well acquainted with the bony passages as one factor
in parturition that stimulated the present exercise.
Since it is an accepted generalisation that a prom
inent cause of difficult labour in the Bantu lies in
differences between the European and Bantu pelvis - a
generalisation unsupported by any study of the bony
pelvis it is essential at the outset in South Africa
to lay down the foundations of such a study.

For the work on the Bantu* about 50 metrical and
non-metrical values were analysed, all measurements
chosen referring to the pelvic girdle rather than to
separate component bones. So far from thinking that
the analysis has been exhaustive, it is submitted that
only 6 functions of the pelvic girdle -including area,
indicial, angular, and linear values - are of essential
importance to obstetrics (Heyns, 1944). Measurements
Itt"Bantu11 is a linguistic name which, strictly speeking,
should be replaced by the more appropriate physical term
South African Negro.

of -

of the pelvis might well he infinite, and scientific
method must seek those that are essential to racial
or sexual differentiation, to other anthropological
purposes, or to obstetrics. The 6 pelvic functions
are the predominant features to be considered in the
analysis of any pelvis or group of pelves; and if a
preliminary survey of a series is being made, these
values should receive consideration first.

II HISTORICAL SURVEY
1. ANATOMICAL
(a) South Africa. The Bantu pelvis has not been
studied on a collection of skeletal material compar
able numerically with the present. G-illman (1929)
and Shore (193°) published anatomical studies on the
Bantu sacrum; Orford and Wells (1936) studied 100
living Bantu women on the basis of body habitus and
of measurements obtained by clinical pelvimetry;
and Orford (193*0 described the Bush pelvis with a
small series of Bantu pelves.

Last century Frltsoh (1&72) recorded some metri
cal data concerning Hottentots, 7 Kaffirs, and a few

- Bushmen -

K
Bushmen. Turner (l£g6, a & b) dealt with similar
material in his papers on sacral and brim indices.
Verneau (1&75)> and t0 a less extent other anatomists,
also recorded measurements derived from pelves of
Negro and related stocks found in South Africa.

The Boskop skull was discovered in 1913 and
subsequent discovery of more complete skeletal material
at the caves of Zitzikama (Fitz Simons, 1921; Dart,
1923) gave some information about the skeletal fea
tures other than the skull of this Boskop type. Thus
arose new concepts concerning the physical composition
of our living South African races. Gear (1925, 1926)
gave the first description of the Boskop sacrum. Later
Weiner (193*4-) in a symposium on the skeleton of the
Bush race described 13 Bush sacra. Since the early
nineteen-twenties there has developed a crystallisation
of the view that the Boskop, Bush, and Bantu races are
three generic groups as distinct in chronology as in
physical type. Galloway (1937) shown that all the
South African Middle Stone Age skulls are within the
normal range of variation of a single physical type.

Dart (1937) approaching the problem from a

- historical -

5.
historical point of view suggests that ten to twenty
thousand years ago Afrioa was inhabited by the Boskop
race. The Bush race migrated south, making contact
with the Boskop people. Miscegenation gave rise to
physical variation of features within both groups.
This anthropological complexity was increased by con
tacts with people alien to Africa, for Mongoloid and
even European skulls have been identified with a pre-
Bantu cultural horizon. The Bantu arrived later and
it is suggested that, "linguistically, the Southern
Bantu can be subdivided into 3 main zones:-

1. Western, spreading north and south of the Okan-
vango River and including the Ambo, Herero, and
Mbundu (as far north as Benguella on the Angola
Coast);

2. South-Central, comprising the Shona tribes of
Southern Rhodesia;

3. South-Eastern, situated principally in the Union
of South Africa and Portuguese East Africa but
also extending into Bechuanaland and the head
waters of the Zambesi. It Includes:-
(a) the’Nguni peoples (such as the Xhosa, Zulu,

- and -

and Swazi) on the south-eastern sea-board,
all of whose dialects are affected by Bush
clicks;

(b) the Shangana-Tonga (Thonga) peoples of
Portuguese East Africa, who have no clicks;

(o) the Sotho peoples, such as the Southern
Sotho of Basutoland, Northern Sotho of the
Transvaal, and Tzwana of Bechuanaland; and

(d) the Venda of North-Eastern Transvaal." (Dart,
1937).

While a great -deal is known about skulls of South
African physical types, living and fossil, little is
known of the rest of the skeleton. The osteology of
the Bush race was dealt with in a published summary
of a symposium on the Bush skeleton (1^3^)• The value
of using the whole skeleton for racial dlagnoUsvwae
shown in the material from Mapungubwe (Galloway, 1937);
but although there is available in the collection of
the Department of Anatomy, University of the Witwaters-
rand, a large collection of Bantu pelvic material,
very little work has been done on it. Elsewhere in the
world also few comprehensive studies on the osteology
of the pelvic girdle have been made.

- (b) -

(b) Elsewhere. Verneau (1675) aad Turner (1665-
1666) studied the pelvis comparatively fully,
Turner presenting no less than 35 separate data and
devoting an entire monograph to the pelves of the
H.M.S. Challenger skeletal material. The paper of
Koganei and Osawa (1900) (not available in South
Africa) is reported as an exhaustive study of the
pelvis which might well serve as a guide (Wilder, 1920),
Derry (1909, 1923) and Straus (1927-1926) studied
specialised aspects of the innominate bones or the pel
vic girdle as a whole. In Hooton’s work (193®) on the
skeletons of the Pecos Pueblo Indians, the pelvis was
treated briefly; and Reynolds, writing a seotion on
the pelvis in the same monograph for the purpose of
assessing the data, confines himself almoet entirely
to a discussion on the mechanism of labour. Hill (194-1)
described 20 Veddah pelves and Pan (1929) 64 Hindu
female pelves, studies which, though not exhaustive, are
valuable. Ssoson-Jaroschewltz, (1925) studied an ex
tensive collection of Russian pelves. As these studies
have followed usual procedures, they have not con
tributed anything to the method by which a series of
pelves should be investigated. Specific aspects of

- the -

3
the pelvis have been explored, but workers have
rarely presented a full range of metrical data.
This statement does not apply to the sacrum which
has been exhaustively studied by several anatomists,
e.g. Radlauer (1903), Paterson (1396), Fischel
(1906).

2. OBSTETRICAL
The majority of studies on the morphology of the

girdle has been made by obstetricians. In the be
ginning, interest in this sphere was confined to
measurements on the parturient or pregnant woman, so
that in time small but useful collections of data
were accumulated. It was on this foundation that
later the physical anthropologist commenced building
in a scientific manner. Before Vesalius (15^3) no
one is known to have described the anatomy of the
pelvis accurately. It was he who demonstrated that
it was impossible for the pelvic bones to separate
adequately in dystocia due to contracted pelvis.

Belief in the Hippocratic theory of pelvic bone
separation during labour resulted in disregard of the
fixed measurements of the pelvis. Thus the rational

- develooment -

9
development of pelvimetry was retarded. The first
known description of contracted pelvis was given only
In 1572 by Arantlus. It is true that Soranus of
Ephesus, perhaps the greatest obstetrician of all
time, in the second century A.D. observed a narrow
pelvis to be a cause of dystocia, but he attributed
the contraction to failure of the bones to separate
during labour.

Van Deventer (1701) erected a milestone.by being
the first to introduce an illustrated description of
pelvic osteology into a work on obstetrics, but it
was Smellie (1752) who first gave definite oelvic
measurements for normal and abnormal pelves. He also
gave a tolerable description of the plane and axis
of the pelvic brim. For many years subsequent to
this clinicians strove to gain a maximum of measure
ments of the pelvis on their patients only. Thus,
nearly a century ago Michaelis took measurements on
500 living women, and Litzmann on 200.

Martin (1866) measured l6 dried pelves and used
20 cadavera for comparison, and also gave brim measure
ments for pelves from different races. Zaayer (1866)
measured 26 Javanese female pelves, and introduced the

- pelvic -

10.
pelvic brim index as a measure of brim shape. Skele
tal study was carried out by Runge (12>69) on 5° Rus
sian female pelves, Rodriguez (12>95) S&nohez
Gomez (1S95) on Mexican pelves, Koganei and Osawa
(1900) In Alno and Japanese, Garson (lgH>l-12>8>2) on
5 Australian and 13 Andamanese women, Emmons (1913)
on 217 female pelves of American Indians, Thoms (1935)
in America, and Caldwell and Moloy (1933) on 26S fe
male pelves (American white and Negro stocks).

Recently radiographic studies on the living have
taken the place of skeletal studies. Dry pelves take
a considerable time to accumulate, whereas the living
subject is available everywhere. Although X-ray pel
vimetry can be very unreliable (see Appendix 1 ), con
tributions from living material will add considerably
to knowledge of the pelvis and must remain the founda
tion of progress in the future. Caldwell et al. (193^>
1939 )> though providing no metrical data, have made
such extensive radiographic pelvic studies on morpholo
gical lines, before and during parturition, that to-day
they are considered the leading authorities on the in
fluence of the bony pelvis on labour, and by inference
the undisputed leaders in the wider field of pelvic

- morphology. -

morphology. In England Nicholson (193&) and Ince
and Young (19^0) published very large series of
pelvic measurements derived from X-ray studies of
pregnant women, and Kenny (I9MJ-) investigated radio
graphically 1,000 pregnant women who had a "clinical
ly suspect pelvis".

Ill SUMMARY

The pelvic girdle of the Bantu has received but
scant attention. On the morphological side work has
usually been done on the girdle either because of an
interest in specific features such as the brim index,
or because of the influence of the bony pelvis on
parturition. Few exhaustive studies on the pelvis
have been made, though the sacrum has been fully in
vestigated by several anatomists.

Obstetricians were first in the field of pelvime
try, only living subjects being used for this purpose.
The physical anthropologist carried the early work in
to the sphere of science; and since the time of Martin
and Zaayer in 1&66 the dried pelves of many races have
been measured. Recent investigation has been based on
radiography which offers greater possibilities than

11.

- does -

12
does the slowly accumulated material of osteological
collections. X-ray pelvimetry, although precarious,
has in the past decade led to a better knowledge of
the female pelvis.

PART II
MATERIAL AND METHODS

I INTRODUCTORY NOTE

For the present survey 100 male and 67 female
pelves have formed the skeletal material, and 125

adult female patients were subjected to X-ray pel
vimetry. These pelves were studied biometrically
In the main, using quantitative values; and where
morphological characters provided the material for
analysis, the statistical method was applied as far
as possible. The purpose of the investigation was
to assess the Bantu pelvic girdle, with special
reference to the female, and to determine its range
of anatomical variation. Neither the innominate bone
nor its 3 component elements were treated individually.
Against crltioism of this omission there must be ad
vanced the reason that the present study does not
claim to be comprehensive. A complete study would
embrace (a) phylogenetic and ontogenetic aspects;
(b) linear, indicial, angular, two-dimensional and
perhaps three-dimensional considerations of the fea
tures of the sacrum, ilium, ischium, pubis, innominate

bone -

bone, and pelvio girdle as a whole - and the ampli
tude of the range of mensuration would be either in
finite or merely arbitrary; (o) morphological stud
ies; (d) pelvic orientation in varying postures;
and finally, (e) the application of the findings to
the pelvic functions of weight bearing and parturi
tion. When the data have been collected there comes
the responsible task of correlation, classification,
and comparison of variables which in itself can be
without end.

Of the above-mentioned branches of research the
ontogenetic development of the human pelvis, for
example, is virtually in its infancy; and there is
no recorded evidence of the application of radio-
graphic methods which alone can trace developmental
changes that occur in given pelves from Infancy to
adolescence.

The present report contains (1) an analysis of *
such dimensions of the pelvic girdle and the sacrum
as were thought from a practical obstetrical point
of view to be of value, (2) a less adequate study of
X-ray findings of the same kind, (3 ) a preliminary
consideration of the foetal and infantile pelvis and

1^.

its

15.

Its sexual differences, and (4-) an estimation of the
weight and head measurements of the Bantu newborn,
of which a knowledge is necessary for comparing fe
male pelves of different racial groups. The last
section is embodied in an appendix which also con
tains the experimental data on which the method of
X-ray pelvimetry is based. The total data presented
comprise some 17,000 determinations of which two-
thirds belong to the pelvis and 7,000 of which were
based on pelvic mensuration in some form. While the
investigation is necessarily incomplete, it is being
followed by further studies. Because of its greater
urgency, ontogenetic development is being investi
gated on 100 Bantu foetal pelves and radiographs of
infants and children. A long-term plan which is pos
sible only with European material has been the col
lection of pelvigraphs of a series of children who
will be X-rayed periodically at the ages of one year,
5 years, 8 years, 11 years, and 15 years. The latter
series of films will give valuable evidence of the
direction of growth of the pelvis, and should remove
the necessity for speculation of the kind made by
Breus and Kolisko (190I+).

II -

II MATERIAL

The adult skeletal material used was obtained
from the Department of Anatomy, University of the
Witwatersrand, where the disarticulated pelves are
preserved. The male pelves, 50 Basuto and 50 Zulu,
as well as the 67 female specimens were taken from
dissected cadavera of known age, race, and sex.

Forty of the Basuto pelves were selected from
the skeletal series of which the skulls were des
cribed by Galloway (1937) for the sake of comparison
with his Mapungubwe material. Of the 5° Zulu skele
tons used, Galloway has studied the skull in 31 cases.
In the case of the female, SO pelves were examined.
Of these, 67 pelves were adult Bantu, 2 were Bantu
aged 13 and 15 years respectively, and 11 were from
individuals of mixed "coloured" Eurafrican race. The
present study concerns the 67 adult Bantu pelves only,
and the measurements refer to this group. (Totals of
6̂ , 65 or 66, however, indicate an occasional im
practicability of measurement owing to damage or gross
malformation.) The races Included in this group of 67

were: Basuto 2k, Xosa Ik , Zulu 11, Flngo "J, Swazi 2,

- Barolong -

17
Barolong 2, Bechuana 2, Tembu 1, Pondo 1, ? Mochopi
1, and 2 unknown.

The women whose pelves were radiographed were
all Bantu, but their specific race was not ascertained.
Dart (1937) discussed the origin of the Bantu people,
and it is conceded that "Bantu" is a generic term which
is unsatisfactory anthropologically and ethnically.
The word is here used to refer to those racial in
digenous stocks in Southern Africa which are clearly
differentiated from Boskop and Bush peoples, the lat
ter including the Hottentot and the Strandloper (beach
comber. )

The foetal material was obtained at the Bridgman
Memorial Maternity Hospital, and was known to be Bantu.
Thirty foetal pelves were measured directly. An ad
ditional 20 foetus were used to determine the number
of sacral vertebrae in the Bantu foetus. The children

6 (1X-ray^were aged 1 to 15 years, and were patients or
visitors to the two Johannesburg hosoitals for Non-
Europeans.

Ill METHODS
1. fl-eneral
Adult Pelvis. It was found that the bones could not be

reliably -

reliably articulated with rubber bands. Where this
was possible, the pelvis had insufficient stability
to remain unchanged during handling. Plasticine
and plaster of Paris soil the bones sufficiently to

#*ruin them. Articulation was effected by "Metallic-X"
which proved to be efficient for the purpose, and is
easily removed either with a knife or a solvent like
acetone. '’Metallic-X'* was applied to the sacro-iliac
articular surfaces, the bones being carefully placed
in wet sand until this adhesive hardened. Satisfac
tory articulation was Judged by the following criteria:
(a) the most accurate coaptation possible for the
sacro-iliac Joint surfaces,
(b) symmetry of the whole pelvic brim,
(c) symmetry at the symphysis pubis and subpublc
angle by viewing the Joint from above and below, and
(d) special care when separation at the symphysis
was greater than normal.

X-ray pelvimetry. Use of X-ray films for the measure
ment of pelvic diameters is unsatisfactory, and many
views are required for a large range of dimensions.
A paper (Heyns, 19̂ 4-5) is included as Appendix 1 to

18.

show -

19
show what method was used, the experimental work
upon which the method was based, and the problems
associated with a study of the living subject.

Foetal pelvis. Exhaustive experimentation demon
strated that there was no easy method of cleaning
the foetal pelvis: the cartilage dries and shrinks
rapidlyy and even preservation in formalin results
in some degree b f distortion. Pelves were accordingly
separated from the foetal trunk and lower extremities,
cleaned with scissors, knife, and forceps, and measur
ed without delay.

A further series of 20 foetus was subjected to
excision of the sacrum and a portion of the innomin
ate bones. In 12 of these cases the vertebral col
umn from immediately above the first lumbar vertebra
was included with the sacrum. The object of this
second foetal investigation was the determination of
the normal number of sacral segments.

Other workers have not cleaned foetal pelves and
measured them on the same day. For precision, how
ever, this is desirable. Moreover, cleaning of this
delicate structure is exacting, and demands 3 to k

- hours -

hours work before proficiency Is gained. Laboratory
assistants not aware of the care required In an in
vestigation of tills sort do not have the patience
to carry out this type of work with thoroughness and
skill; and it was found that cleaning of selected
portions only was incompatible with accurate mensura
tion. The investigator of the foetal pelvis should
personally handle the dissection throughout, arduous
as the task may be.

2. Measurements.
Adult pelvis. The following measurements present
no difficulty of assessment: the interspinous, inter-
crlstal, diagonal conjugate, antero-posterlor of the
cavity, height of the symphysis, external oblique,
maximum height of the pelvis, interspinal distance,
antero-posterior of the outlet (pubo-sacral diameter).

The following diameters are more difficult to
standardize: transverse of brim, oblique of brim,
depth of pelvic basin. The following measurements are
often based on subjective impressions:
(a) Oonjugata vera, which may be the anatomical or
the obstetrical conjugate. The former is measured to

20.

- the -

21
the anterior continuation of the ilio-pectlneal
line and is not always well defined; the latter
is shorter and is measured to 'a point on the sum
mit of the inwardly projecting eminence which lies
a little below the upper end of the symphysis pubis1
(Ince and Young, 19^0).
(b) Width of sacro-sciatic notch (ischial spine to
lateral edge of sacrum). This was measured to a
point on the lateral edge of the sacrum, midway
between the lowest two posterior sacral foramina.

/
(c) Transverse of outlet (inter-tuberal diameter).
The tuberosities of the ischium are such large masses
of bone that this measurement is frequently arbitrary,
varying from observer to observer.
(d) Anterior sagittal diameter of outlet: posterior
sagittal diameter of outlet. These depend for their
length on the position of the intertuberal diameter
already considered.

A large sacral promontory projecting far across
the pelvic basin unjustifiably shortens the conjugata
vera and lowers the pelvic index. Reliance on the
basic value of this diameter, still persistent in
obstetrics, can make a pelvis appear small, whereas

- the -

the area of the pelvic brim, discounting this en-
roachment of the promontory, may be larger than the
estimate by as much as 10 sq.cm. Even if the con
jugate is shortened sufficiently to obstruct passage
of the foetal head, the transverse diameter of the
brim may be absolutely increased and allow easy en
gagement of the head. The pelvic cavity below this
plane would then, all things being equal, easily ac
commodate the head.

Some measurements are purely anatomical and are
accurate for purposes of comparison (e.g. the inter-
ischial spine diameter), some are based on obstetrical
considerations (e.g. the obstetrical conjugate and the
transverse of the brim as seen radiographically), and
some are valueless to both anatomist and obstetrician
(conjugate diameters of the brim employing the promon
tory as one extremity).

X-ray pelvimetry. It is explained in Appendix 1
that the measurements possible were the following:
conjugata vera, transverse of brim, interspinal, and
intertuberous diameters, depth of the pelvic basin
(antero-lateral and anterior), subpubic angle, and

- symphyseal -

symphyeeal height. Areas of the pelvic brim, cavity,
and outlet and the brim Index were calculated from
these data. No sacral measurements were taken.

*
Foetal pelvis. In order to measure diameters ac
curately, a travelling microscope was used. The ver
nier soale made it possible to read to an accuracy of
0.05 mm. For each measurement the small pelvis was
oriented so as to bring the diameter into a horizontal
plane. Red ink marks were made for guidance, and a
pointer used to identify the mark to be viewed through
the microscope. As a single observer effected all the
measurements, it is believed that uniformity was ob
tained. This is of considerable importance because
many of the diameters, e.g. the transverse of the brim,
are indeterminate and would vary within small limits
from observer to observer. In assessing sexual dif
ferences, however, the specimens of a particular
series are comparable.

Three measurements were taken with a sliding
compass without vernier attachment - the length of
the sacrum and the diagonal conjugate, because the
termini were not both visible through the microscope;

23.

and -

and the depth of the pelvic basin, because measure
ment with the compass was as precise as that with
the travelling microscope. It must be pointed out
that measurements with callipers or compass on the
*foetal pelvis are as unsatisfactory as those taken

similarly on soft tissues. As far as can be Judged
from the literature, foetal diameters have been ob
tained only by this unsatisfactory method, and one
which does not allow of reliable readings to 0.1 mm.
Most observers have presented data showing tenths of
a millimetre (Thomson, 18>99i Kappers, 193^1 Falk,
1908, etc.).

Infantile pelvis. With 2 exceptions, aged 13 and
15 years respectively, the postnatal material was not
skeletal but derived from living children who were
submitted to pelvic radiography.

In radiographs of the child's brim the y-shaped
epiphysis is well seen when the brim is not parallel
to the film, but after infancy the cartilaginous gap
is not readily identified in horizontal brim shadows.
This of course is due to over-lapping of the dense
shadows of the ischium lying immediately below the

- cartilaginous -

cartilaginous tissue. The most direct and con
venient method of taking the pictures is that recom
mended in Appendix 1 , i.e. with horizontal brim.
Measurements may then be corrected, and further in
spection made of a brim which is undlstorted. Un
fortunately, it was found very difficult to induce
subjects below the age of 5 years to sit in an at
titude of semi-recumbency with moderate lumbar
lordosis, which is the position required for the
horizontal brim. At their best)Juveniles are un
satisfactory subjects for X-ray work of this kind
where precision is essential.

For these reasons recourse was had to Nichol
son's stereometric method (193&) which gave satis
factory results. This thesis contains only a pre
liminary consideration of the relative growth of
pubis, ilium, and sacrum at the level of the ilio-
pectineal line during the age period 1 to 15 years.
Therefore, further development of radiographic tech
nique will not be discussed here. Accurate measure
ment of the length of these bones was made in a
straight line, and the greatest transverse diameter

- of -

of the bria determined. The purpose of this was
simple, viz. to find (a) whether growth ooours mainly
in the pubis or in the ilium and (b) what is the ef
fect of sacral development during childhood. Solution
of the radiographic problem fully Justifies this pre
liminary report, for this is the only method of ap
proach to an important investigation whioh has uni
versally been retarded owing to the impracticability
of securing suitable skeletal material.

3 . Analysis of Data.
It is quite clear that a mere studying of the

facts will not give the truth. There is, however,
the necessity to investigate the capacity of the
Bantu pelvis, in order to discover what morphological
relation it bears to the pelvis considered as average
or normal, e.g. the European pelvis. "The thinker
formulates his question from a given point of view
determined by the context of hie own experience.
This context of experience includes those beliefs
that are widely and firmly held by his contemporaries"
(Stebbing, 1930). From the point of view of obstetrics
the conclusion presents itself that there are no

26.

available -

available criteria for the Bantu pelvis on which
any hypothesis could have been formulated by these
contemporaries. If, for example, it is suspected
that parturition in the Bantu varies considerably
from that in the European, it would be an indis
pensable preliminary to an Investigation into this
matter to establish in a precise quantitative manner
something of the range of anatomical variation of
the female Bantu pelvis. A metrical survey of the
pelvis would, in addition, afford the opportunity
of assessing racial differences suspected on the
basis of a similar "climate of opinion", as A.N.
Whitehead has named it.

The method used was based mainly on measurement,
both linear and two-dimensional. The data were treat
ed statistically, as were non-metrical features such
as the shape and size of the pubic arch. Glassifi
cation of pelves based on the criteria laid down by
Caldwell et al. (1933* 193*0 was made. All results
were compared with those of Europeans and the figures
available for other races.

The biometrical processes used in the present in
vestigation are essentially based on counting or

- enumerating -

enumeration. "It cannot be over-emphasized that
estimates from small samples are of little value
in indicating the true value of the parameter which
is estimated. Some estimates will be better than
others, but no estimate is very reliable. In the
present state of our knowledge this is particularly
true of samples from universes which are suspected
not to be normal.

"Nevertheless, circumstances sometimes drive
us to base Inferences, however tentatively, on scanty
data. In such cases we can rarely, if ever, make any
confident attempt at locating the value of a para
meter within serviceably narrow limits. For this
reason we are usually concerned in the theory of
small samples, not with estimating the actual value
of a parameter, but in ascertaining whether observed
values can have arisen by sampling fluctuations from
some value given in advance. For example, if a
sample of 10 gives a correlation coefficient of +0.1,
we shall Inquire, not the value of the correlation in
the parent universe, but, more generally, whether this
value can have arisen from an uncorrelated universe,
i.e. whether it is significant of correlation in the

- parent -

parent" (Yule and Kendall, 19^0).
It can be asserted that logicians now recog

nise that the foundation of the theory of induction
is to be found in the theory of probability (Steb-
bing, 193°; Cohen and Nagel, 193*0 • Thus an at
tempt is made to obtain a fair sample of all possible
Instances of the class under consideration and to
test the probability of the sample being representa
tive of the instances of this class. In this manner,
for example, a closer approach is possible to the
truth as to whether a certain feature occurring in
the Bantu is an essential racial character or, alter
natively, that it occurs in the Bantu with a signifi
cantly different frequency from that found for other
races.

If a study be initiated on the pelvic girdle
of a particular racial group, the objective is
elementary, if fundamental. The aim, in the first
place, is not a general study of the pelvic girdle as
an entity, but simply an analysis of the characters
of this particular racial sample in order to contri
bute more to the sum-total of knowledge of the pelvis
gained from all sources. A conception of different

- aspects -

aspects of the pelvis is best derived from the
full aggregate of available knowledge, and it is
obvious that the process of construction in that
case is entirely divorced from the study of the
particular racial sample which contributed to the
whole. Certain generalised inferences may be made
during analysis of the sample (e.g. a suggested
morphological classification), but for such prac
tice; the author may be arraigned, though by a few
even acclaimed.

The first purpose in the use of the data ob
tained here is that of comparison with other results.
For this reason the quantitative method seems to be
the most convenient. Where attributes cannot be
expressed quantitatively ab initio and depend on sub
jective analysis, counting of the data followed by
analysis (e.g. presence or absence) will enable the
statistical method to be used. If these observations
are termed statistics of attributes, as opposed to
statistics of variables where the magnitude of the
variable character is at once determined, it can be
seen that essential anatomical analysis, based on
morphological features, need, not "be lmpalre4. Wore

reliable —

reliable conclusions are reached if the biometrical
method is applied throughout the investigation.

Frequency distributions of variables have not
been shown because in the Bantu skeletal material
the series were too small. In order to claim sig
nificance for a departure from the curve of best fi$,
a low P value in the chi-square test must be de
manded - and with the small racial series analysed
such values are too improbable to justify the labour
involved in curve fitting and calculations such as
the chi-square test. Morant (1939) stresses the
fact that a normal curve of distribution is almost
invariably found for samples dealt with by the anthro
pologist. He refers to 3 examples provided by Felix
van Luschan, and states that he knows of no other
published distributions for any anthropometric charac
ter which show a clear departure from normality. It
is unlikely, therefore, that Bantu samples would in
dicate either appreciable skewness or more than one
mode for any of the variables or attributes here con
sidered.

For present purposes, it may be said that the
adult pelvis is in a static, well crystallised phase,

and -

and that its characters should be stated in a con
ventional manner provided that such convention is
progressive and efficient. The foetal pelvis is a
dynamic object of which the characters are elusive
and transient, thus making analysis - even on the %
ontogenetic side alone - more difficult. In the
latter case ingenuity of the individual observer must
be encouraged in order to remedy a serious lack of
knowledge, whereas the study of the adult pelvis must
follow well-trodden paths with, however, constant
removal of the debris of false method and non-valid
conclusion. If the anthropologist labours in the
direction of becoming, in Plato's words, "the spectator
of all time and of all existence", the data to be
presented here can but aid him, and should not simulate
any form of guidance.

To state summarily the value of the present col
lection of facts, the following are now available:
1. Data relating to the capacity and shape of the

Bantu pelvis and sacrum which will make possible
a comparative study in parturition, using a race
which has lived under primitive conditions for

centuries.

centuries.
2. Data which may throw light on racial diagnosis.

A larger series of Bantu pelvic measurements and
attributes is now available to the physical
anthropologist who may be aided when he is con
sidering skeletal material other than the skull,
on which of course his chief reliance is placed.

3. It would be frivolous to disregard the possibility
that the standard of living in the Bantu is in
the process of being altered, or even considerably
improved. There exists, then, the opportunity
to assess the results of improved nutrition. For
this purpose, it is urgent to observe the range
of variation of the Bantu pelvic girdle before any
possible changes set in: comparison, in a few
generations, with the new findings will demonstrate
the nature of the changes that may occur. Apart
from any other considerations, therefore, it is
Imperative that the pelvic girdle, whose morphology
is held to be sensitive to nutritional states,
should be studied without further delay.

PART III
CLASSIFICATION

I INTRODUCTORY NOTE
Most systems Of classification of the pelvis

are obstetrical, and have an aetiological basis.
There appears to be no classification by an anatomist.
Anatomy in itself, as Bolk (1925) pointed out, has no
creative power but is entrusted with the function of
controlling, a control indeed which morphology can
not dispense with. Morphology is dependent on a
consideration of function (physiology) in addition
to anatomy - the latter observes, the former exolains.
The anatomist has pursued his line of thought con
cerning the pelvis into the more profound region of
ontogeny to the exclusion of the simpler considera
tion of the Influence of the bony pelvis on parturi
tion. Therefore, classification of the pelvis has
been left to the accoucheur. The latter has at
tempted quite reasonably to classify the girdle on a
pathological basis, having as his first consideration
the recognition of the types that cause dystocia. It
has been more difficult to identify the morphological

characters

charsetere of the pelvis that are most favourable
to labour, l.e. to discover what Is the Ideal fe
male pelvis.

II VARIOUS CLASSIFICATIONS
1. Van Deventer (1716) In the first classification
of the female pelvis recognised 3 types: too large,
too small, and too flat. Ingenuous as this grouping
may appear, the analytical perspicacity displayed has
not been significantly surpassed during the last two-
and-one-half centuries. It is unlikely that there
will ever be a simple and satisfactory classification
of the pelvis, and the tyoes of a given system will
be useful for special purposes only. Pelves will,
therefore, have to be grouped according to a prescrib
ed purpose so that the basis may be, for example,
obstetrical or anthropological.

2. Obstetrical classifications have been concerned
with peculiarities - usually of a pathological nature -
that give rise to dystocia: the morphology of the
normal pelvis and its variations has'been disregarded,
except for a recognition of the small pelvis with
average proportions (justo-minor or generally con

tracted -

tracted pelvis). Thus Litzmann's (12>6l), Kerr's
(1923), Williams1 (1936), and Zangemelster's (1927)
classifications vary only slightly, and are based
on deformities due to (a) faulty development,
(b) disease of the pelvic bones, (c) disease of
the vertebral column and (d) disease of the lower
extremities. These classifications are of little
interest to the anatomist.
3. It was not until 1933 that a more satisfactory
classification on morphological lines was reported
by Caldwell et al. Following von Stein's (1844)
scheme of grouping pelves into round, longitudinal-
elliptical, and transversely-elllotical, these work
ers recognised 4- parent types: 1) the gynecoid or
average female type, 2) the android tyoe with a
strong male tendency, 3 ) the anthropoid type which
is longitudinal-elliptical, resembles the brim shape
found in anthrooold apes and oresent to a less extent ♦
in all mammals, and 4-) the platypellold or trans-
versely-elliptical type. In addition to the oarent
types there are 2 groups with mixed characters, e.g.
gynecoid with anthropoid tendency or android with
gynecoid tendency (eee Table 1 ).

- Caldwell et al.

37
Caldwell et al. (1933, 193^) drew attention to the

moderate frequency (about 12 percent) of the anthropoid
type pelvis and described its labour mechanism. It is
actually their great contribution to labour mechanisms
which were studied radiographically in parturients
(193^ b, 1935), that has made this team of workers cele
brated. This fame has unfortunately spread to their
classification which has received universal recognition.
The classification is sufficiently complicated to be
poorly understood by the majority of its would-be ad
herents, and its value has only rarely been discussed. In
addition to the statement that a pelvis belongs to one of
the 12 classes seen in Table 1, it is necessary to as
certain (1) whether the forepelvis is narrow, rounded
etc. and (2) whether the pubic arch is wide, moderate,
or narrow. Such a classification is too cumbersome, hav
ing only the merit of describing a pelvis, whereas the
rationale of classifying an object is to reduce descrip
tion by placing the object in a previously defined category.
Because the Caldwell and Moloy system is accepted to-day,
the 167 Bantu dry pelves have been classified accordingly
(Table 1), and reference will be made to the

principles -

principles underlying the scheme.
To avoid a lengthy explanation, the authors'

diagrams of the 12 pelvic types with a short des
cription are given in Fig Cl . Classification is
based entirely on the nature of the pelvic brim,
and more particularly on that part of it lying
posterior to the widest transverse diameter of the
brim, this segment being only a little different
from Derry's posterior segment of the pelvic chllo-
tic line (1923) and Straus1 (1927) lower iliac height.
In the anthropoid type pelvis the distance of the
transverse diameter from the sacrum is great, in the
gynecoid type intermediate in length, and in the
android type the diameter lies close to the sacrum.
The anterior half of the brim is also characteristic
for the different types, and may give rise to mixed
types. This scheme of grouping is followed fairly
faithfully by Turner's brim index, making the types
mentioned respectively dolicho-, mesati-, and platy-
pellic. There is indeed small merit in applying a
new terminology to these pelvic inlets.

In addition to these morphological types there
occur pathological types in regions where rickets is

- common, -

3«.

Fig. 3.— The gynecoid type'with anthropoid tendency ( mixed type).
The posterior segment is gynecoid, the transverse diameter being closer to the prom

ontory. The long, oval appearance is caused by the long, narrowed fore pelvis.
Fig. 4.— The gynecoid type with a narrow fore pelvis ( mixed type).
This pelvis is practically normal. The narrow fore pelvis is evidence of a weak

anthropoid or male influence.

Fig. 5. Fig. 6.
Fig. 5.— The true gynecoid pelvis ( parent type).
In the developmental cycle this inlet is round and typically female in all portions.

The female influence predominates.
Fig. 6.— The gynecoid type with a flat tendency ( mixed type).
Note that the pelvis is wider and flatter, indicating a trend toward the true flat

type. 4

Fig. 7. Fig. 8.
' Fig. 7.— The flat ( platypelloid) type (parent type).
The wide transverse and short anteroposterior diameter are the salient characteristics.
Fig. 8.— The android type with anthropoid tendency ( mixed type).
The posterior segment conforms to the male type. The long oval appearance is

caused by the long narrow fore pelvis. This places it also in the anthropoid class.

Fig. 9. Fig. 10.
Fig. 9.— The android type with gynecoid tendency (mixed type).
The flat masculine posterior segment is characteristic and defines the pelvic type.

The fore pelvis is gynecoid, similar to the fore pelvis seen in Fig. 5.
Fig. 10.— The true android type (parent type).

Note the flat male posterior segment, the narrow anterior segment, the forward
sacrum and converging side walls. This pelvis approaches very closely the shape of
the average male.

Fig. 11. Fig. 12.
Fig. 11.— The android type with flat tendency (mured type).
Although a male posterior pelvis is present, the general form is broad and flat.
Fig. 12.— The asymmetrical type. '
The asymmetry is limited to the right side. The sacrosciatic notch is narrower

on that side and the iliopectineal line is straighter.

common, and which may range from antero-posterior
flattening to the complete distortion found in os
teomalacia.

Comparatively rare pathological and develop
mental anomalies may cause asymmetry due to faulty
posture (disease of spine or lower limbs), or the
contraction seen In the Naegele and Robert pelvis
in which there is faulty development of sacral alae.

Ill Caldwell and Moihoy Classification of Bantu Pelves.
The classification in Table 1 of Bantu pelves

is based on the standards laid down by Caldwell et al.
(1933, 193*0, whose paper (193I1) presented a clas
sification of 215 American white pelves, the figures
relevant to which are here given in brackets after
the Bantu fig\u?es.

It had been observed that the male pelvis ex
hibited several of the Caldwell and Moloy types, and
an attempt was made to classify the male series ac
cording to the scheme for the female. It was sur
prising to f ind examples of nearly all the tyoes, and
even a male pelvic type which was hardly distinguish
able from the essentially female gynecoid type. In
Table 1 male figures are seen opposite gynecoid

and -

and android divisions: this analysis is recorded
for the purpose of comparison, for whilst similarity
to these female types does exist in the male, it is
not intended to extend the connotation of gynecoid
and android to male pelvic types.

Table Bantu Pelves (Caldwell and Moloy classification)
American white figures in parentheses

Per-
No. of c<

Classification
True anthropoid type
Anthropoid with gyne

coid tendenoy
G-ynecoid with anthro
poid tendency

G-ynecoid with narrow
forepelves

True gynecoid type
G-ynecoid with flat

tendency
True flat (platypel-
loid) type

Android with anthro
poid tendency Android with gynecoid
tendency

True android type
Android with flat ten

dencyAsymmetrical forms

No. of centage 100
pelves incidence Hale
9 (25)i 13.14- (11.6) 16
g (l4)' 11.9 ( 6.5) 2

k (10)I 6.0 ( 4.6) 12

g (2*0l 11.9 (11.1 ) 7

1111 2g. k-
1 6.0

(39.5)
< 3-3)

150
1 ( 2) 1.5 ( 0.9) 2

k ( g) 6.0 ( 3-7) 14-
1 (ll) 1.5 ( 5.1) 0

6 ) 9.0 (11.6) 39■ 2 ( K> 3.0 ( l.g) 2

1 ( 4) 1.5 ( l.g) 1

With a few exceptions, the table shows a fairly
close correspondence between the 3 sets of figures.

- The -

The Bantu sex differences qua Incidence are seen
(1) in the android parent type where the difference
in favour of the male is 6.6 times its standard er
ror; (2) in the gynecoid parent type with a dif
ference in favour of the female of 2.33 "times its
standard error. The higher incidence of true an
thropoid types in the male has no statistical sig
nificance. Thus in the male series the highest
incidence was in the android group, and in the fe
male a high incidence of the same order was found
in the gynecoid group. The European female figures
correspond in this respect with the Bantu. Apart
from these 2 fundamental sex types, there is an
appreciable correspondence between the females of
the 2 races on one hand and the male Bantu on the
other.

The true anthropoid type occurs in much the same
proportion in both the Bantu and the American white
series. But if this group is taken together with the
pelves showing anthropoid tendencies, the Bantu fe
male incidence is 37 percent (3^ percent for the
male), and the white 26 percent. The difference for
the female is 1.5 times its standard error; and for

- male

male and female Bantu combined the difference from
the white material is 2.0 times its standard error:
significance for this difference can only be estab
lished by analysing more Bantu pelves.

The true gynecoid type - the nearest approach
to what is considered to be the ideal female pelvis -
occurs in 2S percent only of female Bantu pelves as
compared with 4-0 percent of whites. This might be
regarded as possibly significant, were it not for
the fact that the gynaecoid type with corresponding
tendencies (e.g. anthropoid, flat, narrow forepelvis)
occurs with much the same frequency in both female
series.

While it was surprising that in the female the
true android type actually had a lower incidence
(9 : 12 percent) in the Bantu than in the white, the
difference found is not significant. If all pelves
exhibiting android tendencies are grouped together,
the Bantu female percentage is found to be 20 as
compared with 22 in the American series.

The female tendency to flatness occurring both
in the gynecoid and in the android type shows in each
case nearly twice the percentage in the Bantu series

42.

- as -

as in the American. The gynecold figure at 6.0 per
cent and the android at 3.0 percent are so low as to
lack significance. If this tendency were established
in a larger series, it might be suggested that the
underlying cause is associated with some factor such

■L

as rickets which is so prominent during Bantu child
hood.

The purely asymmetrical forms shown could not
readily be subclassified. Where asymmetry occurred
in pelves of other types, an alternative diagnosis
was made.
IV Evolution of the Human Pelvis: Criticism of

Taldwell and Moloy's view.
Caldwell et al (193*0 have ventured beyond their

classification in an attempt to explain the origin
of the parent types. Fig C2 is reproduced to il
lustrate their ideas diagramatically, and in the
printed paragraph beneath the figure it will be seen
that the authors refer to their belief in an evolu
tionary factor which is responsible for producing
the typically platypellic human forms ranging ortho-
genetically from extreme dolichopelly to an ultra
human platypellold human type. In the same paper
they claim that this hypothesis is the most important

*3.

- contribution -

Again referring to Fig. 13, the android forms may be noted in the
right section of the outer circle. The term “ android” is used to
designate the female pelvis possessing a masculine type of posterior
segment. This may.likewise occur in any pelvis within the develop
mental cycle just described. Hence, we find android pelves ivith an
anthropoid tendency, android pelves ivith a gynecoid tendency, the true
android type, and, below this type as the flat form is approached, the
android pelvis with a flat tendency (Figs. 8 to 11).

Fig. 13.— Diagrammatic illustration of the influences which affect the shape of the
adult pelvis. The evolutionary factors are shown in the vertical series of the inner
circle and the sexual or hormonal factors in the transverse series. The resulting liv
ing forms are illustrated in the outer circle.

Is it not possible that two great influences may be at work, one, the
arrest in evolution from the ape form, the true anthropoid, to the per
fect human form which is characteristically flat, and the other, a sexual
or hormonal factor which may determine certain masculine characters in
the female form?

If this is true, two distinct but radically different trends in pelvic
shape result: evolutionary and sexual (hormonal). The evolutionary
cycle has been described above. In Fig. 13 the evolutionary factor is

contribution of their work.
The Caldwell and Moloy view is that ortho

genesis begins in lower mammals where the pelvic
inlet is always longest in the antero-oosterior
axle (brim index over 100), and ends with platy-
pelly found in man (brim index under 100). The
authors assert that the occurrence of anthropoid
type pelves (10 to 20 percent) is due to an arrest
of evolution (193^)* These assertions will now be
examined.

If the recapitulation theory is being invoked
for the argument concerning the arrest of evolution,
there is a danger that evolution, or this process of
"arrest in evolution", might be by-passing the foetus
in respect to the pelvis. If the Caldwell and Moloy
hypothesis is to mean anything, the picture before
one is a high incidence of primitive "anthropoid type"
pelves in the foetus and embryo; a more human
(possibly flattish) pelvis later during ontogeny;
and at the end of ontogeny some more highly develop
ed pelvis which is well removed from the ape (or
lower mammalian) pelvis morphologically. Our authors
choose to believe that this last, fully evolved, pel
vis is their platypelloid type, l.e. a pelvis with an

average -

*5
average brim index below 30. This presumption is based
on very poor evidence, for not only have Caldwell and
Moloy (193*0 reported an incidence of only 2.0 percent
for this type, but the Bantu incidence does not rise
above this figure, nor does the English (Ince and Young,
19̂ -0; Kenny, 19MO. Moreover, Caldwell and Moloy have
made no reference to any personal experience with foetal
pelves, nor has any reliable foetal pelvic work been
reported by any writer in English. Kappers' monograph
(1933) is the only available study of this kind which
can be considered authoritative.

An ontogenetic survey of the Bantu pelvic brim
demonstrates the following facts.
(1) Seventeen of 100 foetus had an index over lBo
(or }.00 and over).
(2) In 39 children aged 1 to 15 years, 15.^ percent
have indices over 100 and are of well marked anthro
poid type. But 35.9 percent (ik pelves) have an index
of 100 or more.
(3) For adult indices of 100 or over, there were 17
of 100 male dry pelves, 12.0 percent of 67 female dry
pelves, 17 of 100 female X-rayed subjects, and 20.0
percent of another series of 25 X-rayed women. The

- dry -

dry pelves exhibited the Caldwell and Moloy anthro
poid type in 13.̂ - percent of females and l6.0 per
cent of males, both figures being very close to
those given above (12 and 17 respectively).

Concerning the foetal brims mentioned under (l)
above, the index reached 110 on only 2 occasions,
and the brims never resembled the pronounced "anthro
poid type" seen in the adult: they certainly do not
approach the relative proportions of the ape brim.
These high-index foetal brims are in fact only round,
or "more round" than the average brim; and 33 per
cent of all the foetal brims were dolichepellic
(index over 95)•

The infant and child pelvis moves further in the
direction of "roundness" than the foetal with a per
centage difference which is 2.23 times its standard
error. Now for the first time do outspoken "anthro
poid types" appear - but the outstanding observation
concerning the Infant brim is the trend from flatness
to roundness.

In the adult there occurs on the average a flat
ter pelvis than in the child. The "anthropoid type"
incidence is still equal to that in the foetus, though

the

the "anthropoid" appearance reaches greater em
phasis in the specimens so characterised. Some
thing happens at pubertal age to arrest the develop
ment of round or ultra-dolichopellic brims, but the
basis of this is not evolution. The câ use may be
some combination of factors concerned with sex hor
mones, the influence of nutrition on metabolism, and
physical forces based on weight-bearing and progres
sion.

There is no question here of running with the
Lamarckian hare and hunting with the Welsm&nn hounds.
The possibility tinder consideration is whether or
thogenesis is being exhibited in the case of the pel
vis, for on this matter Caldwell and Moloy speak of
an arrest in evolution. A point is, therefore, reach
ed at which it appears that many dolichopelllc forms,
slight and exaggerated, should occur; but coincident
in time there ;emerges 'the sex factor, and the adoles
cent phase is associated with only occasional develop
ment of the sort of pelvic brim which is found through
out the mammalian class. In the anthropoids, on
account of their greater facial angle, the foetal
head is held in flexion during labour and thus re-

- quires -

quires adequate width further forward In the pelvis
than is necessary in, say, ungulates.

It is a point of no mean importance that Kenny*s
(1944) paper has established finally the belief that
the anthropoid type pelvis in women is associated
with a minimum of dystocia, complicated labour being
much rarer than with the "gynecoid" pelvis. If labour
in mammals be considered, it is not unreasonable to
conclude that difficult labour may jeopardise the con
tinuation of the species. Hence only pelves favourable
to easy expulsion of the foetus would evolve. In the
anthropoid apes greater and rounder heads have develop
ed, and the problem becomes more acute. Observations
in Homo sapiens have demonstrated that the flatter
the pelvis, the more difficult does parturition become.
Antero-posterlor shortening in morphological types is
very rare, but common in diseases pausing softening
of bone. Evidence is accumulating that human types
of superior physical status possess on the average
round brims, while showing a high incidence of
'•anthropoid1' types. There is mathematical certainty
in the fact that, with a given perimeter such as is

- represented -

represented by the brim, a circle encloses the
greatest area: as the circumference becomes com
pressed and alters shape, so there Is a loss of
area.

So far, then, from supposing that evolution
strives for platypelly, the evidence is in favour
of dollchopelly being the highest and most efficient
attainment. Bolk (1925-1926), in his hypothesis on
retardation and foetalisatlon, has shown that mor
phologically the apes have advanced much further
than man along the road of evolution. His idea was
that human anatomical features are all tainted with
foetal characteristics; and what represents only a
transitional stage in apes, has in man become the
consummation of ontogeny. All human races have not
made equal progress on the developmental road, and
Bolk mentions such examples as the Mongolian physiog
nomy which is a foetalisatlon effect absent in white

araces, and the persistence of^metopic suture which
is much more frequent in white than in black races.
Bolk emphasizes the necessity for embryonic poten
tialities in human tissues such as epiphyses which
continue to grow over a very long period: phylogene-

**9.

- tic

tic maturity could only be reached very late, while
retardation during ontogeny is inhibiting this evolu
tionary progress all the time. This inhibition is
often essential for man (vide the late obliteration
of cranial sutures).

In the pelvis the same process with side-effects
occurs. During ontogeny (5th to 10th years), attempts
at evolving the dolichopelllc ancestral pelvis are
seen. These attempts appear to be suppressed at
puberty; but a certain number of girdles, neverthe
less, break the bonds of the foetalislng influence
and become "anthropoid". The majority, however, now
resemble the foetus more closely than they did during
infancy.

Although the factors which, in groups such as
those studied by Thoms (1939)# make possible a pel
vic development beyond the foetal stage are unknoiirn,
the available evidence contradicts the Caldwell et al.
belief that the ultra-dollchopellic pelvis represents
an arrest in an evolutionary movement which has platy-
pelly as its purpose. There is no evidence for this
ill-conceived viewpoint; and it is submitted that
in one sense the reverse is true, viz. that the ape

pelvis -

pelvis has developed further than man’s which is
foetal in character, though the latter shows a
tendency to reach towards more efficient dollchopelly.
If the foetus exhibited many vrell developed "anthro
poid" forms, and if during adolescence these forms
diminished in number and essential characters leaving
only a few "evolutionary arrests", Caldwell and Moloy
would have gained their point. Unfortunately, the
facts are inconsistent with their thesis.

Commentary.
The view of Caldwell et al. (193*0 that the

platypelloid pelvis is the end-point of evolution in
man, while dollchopelly (resembling the apes) demon
strates an arrest in evolution, has been refuted.
Although only Bantu data were used for this argument,
it must be stated that the figures of other observ
ers provide equal support. Thus Kappers (193&) had
only 4-. 0 percent of foetus with brim index ovef 100,
and Turquet's series (1334) of 7° infantile pelves
shox̂ ed a brim index of 100 or more in 23.5 percent.
Preservation of these Juvenile specimens has the
effect of raising the index, with the result that the

51.

- childhood -

childhood period of ontogeny has always been credited
with a very high dolichopellic Incidence. Evidence
gained from fresh pelves, however, shows that this
observation is sound, and the Bantu X-ray results
support it strongly.

The view advanced here is that the evolutionary
purpose in man is to produce the "anthropoid type"
pelvis. In accordance with Bolk's hypothesis, the
retardation of the factors leading to pelvio maturity
is sufficiently effective to leave the stamp of foetal-
lsation on the majority of adult pelves: a minority
only excel, become phylogenetically more highly evolved ,
and are more efficient as ultra-dolichopellic pelves.
The latter have relatively larger inlets, presumably
because ilium end pubis grow more, in order to pro
vide the long inlet which vrould have possessed a smal
ler area had it been round because of the earlier
cessation of growth.

To classify pelves until more is known about
their evolution is futile. Further ontogenetic studies
must be directed to X-ray investigation of many pelves
during puberty and adolescence in an attempt to de
termine the factors responsible for development or

- arrest

arrest of the fully evolved human pelvis. Some
knowledge of other primate foetal pelves Is also
necessary. (Bolk in 1926 published a paper on the
comparative anatomy of a gorilla and a chimpanzee
foetus, but this study is not available in South
Africa.) The brim index is a promising basis for
classification, but it may be necessary to consider
other features of which the most significant appears
at present to be the pubic arch. Turner's index is,
therefore, still the most important determination as
a basis for classification. A further observation is
found in the fact that a pelvis may be small, average,
or large in size; and for parturition this character
of absolute size is an essential feature. Pathological
anomalies must be considered in a category which bears
no relation whatever to morphology. 'When the ontogene
tic investigation now in progress in the Department
of Obstetrics has been pursued further, and the rela
tive capacity of dollchopellic girdles has been de
termined accurately, a scientific classification will

53.

be evolved

PART IV
THE PELVIC aiRDLE A3 A WHOLE

A. Certain External Measurements
of the Pelvic Girdle

The measurements considered here are the inter-
cristal, interspinal, and external oblique diameters;
the maximum height of the pelvis, the depth of the
pelvic basin, the height of the symphysis pubis, the
intertuberal diameter, and the vrldth of the sacro-
sciatic notch in the line of the sacro-splnous liga
ment. The external conjugate could not be measured,
though reference to it is made.

In the table for each of these diameters the last
column shows values on which the significance of dif
ferences is based. If the difference between the means
for the two male series is less than 2.0 times its
standard error, the series are grouped and the mean
and other values for the combined male pelves (100) is
shown. A calculation to determine whether the dif
ference between the female series (67) and the com
bined male series (100) is significant is then made,
and also shown in the last column mentioned above.

While for the purpose of the present work only a
difference which is 4- times its standard error can be

5*.

said -

55.

eald to be certainly significant, a value of 3 times has
been taken to show significance, one of 2.5 to 3 times a
probable, and one of 2.0 to 2.5 times a possible signifi
cance. Measurements for the Bush pelvis are those given
by Orford (193*0, and the European figures those of Ver-
neau (l£>75) except inhere otherwise stated. In the fol
lowing tables the mean t the standard error of the mean
is shown, not as in a previous paper on the Bantu female
(Heyns, 19*0-) the mean t twice the standard error of the
mean.

1. Intercrlstal diameter.
Series

67 Female
50 Basuto
50 Zulu
100 Male

Mean (cm.) S.D.(cm.)

23.go t 0.17 1.4-
24-. 00 t 0. 2*4- I.6 7

24-. 57 t 0. 24- 1.69

24-. 29 - 0.16 1.63

Range (cm.) Significance:
Dlff.of msane

20.9 - 26.5

20.1 - 26. g
21.6 - 27-S
20.1 - 27-2

Stand, error
)

i i

•! *-7 ! “
)
)

In the male series the Zulu values were a little high
er than the Basuto, but otherwise there was no difference
of note. The difference between the combined male and the
female, series is possibly significant, and this tendency
for the male figure to be higher is recognised elsewhere.

Thus

Thus the B u s h a n d E u r o p e a n f e m a l e d i a m e t e r s w e r e r e s

p e c t i v e l y 19.0 cm. a n d 26.6 cm., the c o r r e s p o n d i n g m a l e

f i g u r e s b e i n g 20. 2 cm. a n d 2 7.9 cm. The l a r g e s t inter-

c r i s t a l d i a m e t e r s In b o t h B a n t u sexes w e r e a c t u a l l y 0.1 mm.
less t h a n t hose f o r the r e s p e c t i v e E u r o p e a n means.

It h a d b e e n thou g h t that the g r e a t e r d i m e n s i o n in

m a l e s m i g h t be due to m o r e m a s s i v e iliac crests, b u t V e r -

,neau (1S75) m e a s u r e d the d i s t a n c e s in c o n t r o l series

b e t w e e n the m e d i a l aspe c t s of the crests. The p r o p o r t i o n

in f a v o u r of the m a l e w a s 23.1 to 22.2.

2. I n t e r s p i n a l d i a m e t e r .

Series Mean (cm.) S.D,(cm.) Range (cm, ) Significance:
D l f f . o f m e a n s
Stand, e r r o r

67 F e m a l e 2 0 . S 6 + 0.19 1.60 17.0 - 23.8 )
)

) )
) 1.1 ) 3 . 1 7

50 B a s u t o 21.4-9 +
1

0.24- 1.6S 17.3 - 25.0

50 Z u l u 21.2>7 + 0.25 1.77 I S . 6 - 25.7 ) )
)
)100 M a l e 21.6S + 0.17 1.74- 17.3 - 25.7

The Zulu measurements here, as with the intercristal
diameter, showed a tendency to be higher. The difference
between male and female means was significant. Means for
females were Bush 17.0 cm., and European 22.2 cm. Whilst
the smallest Bantu female measurement here was equal to

- the

the average Bush diameter, 2 Bantu oelves, aged 13 and
15 years, not included in the series, measured less than
17.0 cm.

The other male figures were Bush 17.6 cm., European
23.1 cm. The European female mean is considerably higher
than that of the Bantu male, and the Bantu female in turn
shows an even greater preponderance over the Bush male.
This diameter shows a pronounced tendency to be greater
in the male.

3. Inters-olnal/lntercristal index.

57.

Series M e a n S.D. R a n g e S i g n i f i c a n c e :
D i f f . o f m e a n s
Stand. e rror

67 F e m a l e 3 7 . 7 0 3 0.4-5 3.7 2 76.0 - 93. 9 )
)

| 0.76 13.03
50 Ba s u t o 39.74- - O .52 3 . 7 0 so.o - 97.^
50 Zulu 39.16 I 0 .55 3.92 79.0 - O O • Oa ) )

)
)100 Mal e 39.52 - O .3S 3 .S5 79.0 - 100. g

H e r e the male i ndex is the g r e a t e r a n d the d i f f e r e n c e

between the sexes is significant.

4-. External oblique diameter.
The measurement is made from the posterior superior

iliac spine of one side to the anterior superior spine of
the other side. The diameter is bilateral, and the cor
respondence between the two sides follows fairly closely

the

52.

the correspondence between lengths of the oblique diam
eters of the pelvic brim. In some pelves, however,
equal external oblique diameters occurred in the presence
of a variation of ̂ - 1 cm. in the oblique brim diameters,
and vice versa. Both sets of diameters are measurements
of the ilium alone, but may be Influenced by the sacrum.

4. External oblique diameter.
Series

(67 Female
A<u(50 Basuto
(50 Zulu

M e a n (c m . ) S . D . (c m . )

13.33 1 0.11 0.33
Ig.7lj-l0.l4 0.97
19.13 1 0.15 1.09

Range (cm.) Significance:
Dlff.of means
Stand, error

16.2 - 20.6 j
16.9 - 21.1 ) ) 1.6

) 2.1 )
16.9 - 22.5 ) )

(67 Female 12.93 + 0.11 0.39

tARO Basuto IB. 67 + 0.15 .1.04
(50 Zulu 19.12 + 0.14 0.92

16.3 - 20.3
16.5 - 20.3 )

, ) 2+
17.2 - 21.6 )

For both left and right means the Basuto male is less
than the female which lies between the two male grouns. The
mean of the left female diameter is O.R mm. shorter than
the right. Asymmetry at the brim was a striking feature of
the present material, and a shortening of the left oblique
was emphasized. The mean of the right oblique diameter is

- 1.4 mm -

59.
l.lj- mm. greater than the left. On the contrary, the
Basuto male left diameter Is the greater by 0.7 mm.,
with a left oblique of the brim greater than the right
by 0.7 mm. The Zulu male deft oblique of the brim Is
0.2 mm. greater than the right, corresponding with the
Basuto, though in the Zulu the two external obliques are
equal. While there Is no significant difference between
the means of the two sides In these series, the prepond
erance of left over right in the male and the reverse In
the f e m a l e m a y be due to chance. But this trend persists
In other dimensions, such as the ares of the auricular
surface, and must not be entirely disregarded.

5. Maximum height.
Series

(67 Female
^(50 Basuto
3(50 Zulu
(100 Male

(67 Female
£(50 Basuto
§(
£(50 Zulu
(100 Male

Mean ( cm.) S.D.( cm.) Range (! cm. )

19. os Hr 0.11 0.92 17.3 - 21.6

19.62 + 0.16 1 . 1 k 16.9 - 22.3

19.20 + 0.15 1.03 17.3 - 22.5

19.7k + 0.11 1.09 16.9 - 22.5

12.99 t 0.11 0.22 17.0 - 20.9

19.75 Hr 0.16 1 . 1 k 17.1 - 22.5

19.91 H* O.lll- 1.01 17.3 - 22.2

19.23 + 0.11 1.02 17.1 - 22.2

Significance:
Dlff.of means
Stand, error

)
)

)
) O.55 k.2

i 0.07 > 5.*<■1 I
The -

6o.

The differences between individual right and
left sides were found to show no significance what
ever for all four groups: female, Basuto and Zulu
male, and the combined 100 males. The differences
were all less than 1 times their standard errors.
However, the sex differences for maximum height
showed the following:

Female and Zulu male (left): 3*9 X standard error.
Female and 100 combined male (left): k.2 X stan

dard error.
Female and 100 combined male (right): 5*^ X stan

dard error.
Therefore, the smaller diameter for the female was
a significant finding, and on both sides the Zulu
dimension was greater than that of the Basuto.

In both male groups the left was less than the
right maximum height.

In the female, however, left maximum height was
greater than right.

It will be seen below that the depth of the lesser
pelvis is the reverse of the maximum pelvic height
for the txiro sides.

In both male groups left depth was greater than

- right -

right depth.
In the female left depth was less than right.
-No cause for this disparity can be suggested. The

common point for the two measurements is the tuberosity
of the ischium, the second point being in one case on the
crest of the ilium and in the other on the superior ramus
of the pubis. It is obvious that during growth the ilium
may be influenced by factors different from those affect
ing the pubis, but no explanation of their nature can be
offered.

The Bush and European figures for the female were
respectively l6.2 cm. and 19.7 cm., and for the male
17.4 cm., and 22.0 cm.

6. The height-breadth index. (maximum height
intercristal diameter)

Series Mean (ea,) S.D,(e».) Range (era. ) Significance:
Dlff.of means

66 Female 20.4-7 ± 0.4-7 3.32 13.3 - 32.7
Stand, error

)
)50 Basuto 22.44 ± 0.62 4-.4-0 74.3 - 97.3 ) )

. ) 0.26 ) 1.92
50 Zulu 20.26 I 0.60 4.22 74.2 - 29.6 ) )
100 Male 21.70 ±0.44 4.36 74.3 - 97.3 )

The other female figures were Bush 29.4 and Eurooean
74.0; and the corresponding figures for the male were

- 32.2 -

3S.2 and 79.0. The tabulated figures show that the
sex difference is probably not significant, and sug
gest that the female pelvis in this respect has the
same proportions as the male, but is smaller. In
view of the Insignificant sex difference for the
intercrista.1 means, a discrepancy here is apparent.
But' this is explained by the fact that for this par
ticular index, the mean maximum height for each pelvis
is employed: it will be recalled that in the male
the right side was greater and in the female the left.
Herein is found the reason for the highly significant
sex differences for the two sides, these being res
pectively H.2 and times the standard error for
left and right sides.

7. The depth of the pelvic basin - "from the ilio-
pectineal line in the region of the lllo-pectineal
crest, to the lowest point of the sciatic tuber of the
same side" (Wilder, 1920) - is an Important obstetric
pelvic measurement, and has been used in Tabled . In
the present series the superior point was anterior to
the llio-pectineal eminence, and where possible,
at right angles to this part of the ilio-pectineal

62.

line

Ri
gh
t

Le
ft

line: the line f r o m the t u b e r o s i t y of the I s c h i u m always

runs in r e l a t i o n to the p o s t e r o - l a t e r a l b o r d e r of the

o b t u r a t o r f o r a m e n a n d a p p r o x i m a t e l y Is a tangent to the

a n t e r i o r p o r t i o n of the a cetabulum. In the X - r a y m e a s u r e

m ent of this d i a m e t e r the lin e wa s d r a w n as a tangent to

the acetabulum. S e v e r a l w o r kers, I n c l u d i n g V e r n e a u ,

m e a s u r e d f r o m the i l i o - p e c t l n e a l eminence. Oare has,

therefore, to be e x e r c i s e d in m a k i n g c o m p a r i s o n s w i t h f i g

ures w h e r e the m e a s u r e m e n t is not defined.

The sex d i f f e r e n c e has a p r o n o u n c e d s i g n i f i c a n c e a n d

it is seen that

In the mal e left d e p t h of p e l v i c b a s i n is g r e a t e r than
r i g h t .

In the f e m a l e left d e p t h of p e l v i c b a s i n is less than
r i g h t .

63.

Series

(67 F e m a l e
(
(50 Bas u t o

(50 Zulu
(
(50 Mal e

Mean (cm.) S.D.(cm.)

2.75 ± 0.06 0.51
9.26 + 0.07 0A9

9.37 t 0.02 0.52

9.31 ± 0.05 O.55

Range

7.9 -
7.9 -

7.7 -
7.7 -

(cm. )

10.9

10.2 )
1 0 . 5 )

10.5

S i g n i f i c a n c e :
D l f f . o f m eans
Stand, e r r o r1T
1.5 ) 6 -7

( 6 7 Female
(50 B a s u t o

(50 Zulu
((100 Male

2.77 t 0.07
9.20 1 0.07

9.35 1 0.02
9.27 i 0.05

0.56

0.50

0.52

o.55

7.9 - 11.1
7.9 - 10.1 )

7.7 - 10.5 )
7.7 - 10.5

)
)

1.5 ) 5.7 5

)
)

In -

In the female, w h e r e the b a s i n d e p t h was

9.0 cm. or more, the p u b i c a r c h t e n d e d to be narrow,

a n d there w a s a sl i g h t t e n d e n c y to h e a v i n e s s of the

bones: p e r c e n t of the en t i r e series h a d a d e p t h

m e a s u r e m e n t of 9.0 cm. or more, as he.d 9 out of 13

a n d r o i d p e l v e s as a g a i n s t 27 p e r c e n t of the other

types. This m e a s u r e m e n t is g r e a t e r in the male

t han in the female, a n d the f i n d i n g that a n a r r o w

p u b i c a r c h a n d i n c r e a s e d w e i g h t of the b o n e s tends

in the male d i r e c t i o n w i t h i n c r e a s e in the d e p t h of

the p e l v i s e m p h a s i z e s the o b s t e t r i c a l s i g n i f i c a n c e

of this mea s u r e m e n t . The B a n t u female m e a s u r e m e n t

of 2.6> cm. wa s r e m a r kable, s i n c e in 5° A m e r i c a n w h i t e

fe m a l e p e l v e s C a l d w e l l a n d M o l o y (1933) f o u n d the

a v e r a g e p e l v i c b a s i n d e p t h to be 9.0 cm., a n d in 53

A m e r i c a n w h i t e mal e p e l v e s 10.1 cm. The B a n t u female

a v e r a g e is thus l o w e r than that f o r the w h i t e woman.

V e r n e a u ' s (12>75) means, w h i c h must be grea t e r

as p o i n t e d out above, w e r e 9*3 °m . for the female,

a n d 10.7 cm. for the male.

2>. The l e n g t h of s y mphysis p u b i s w a s m e a s u r e d in the
a r t i c u l a t e d p e l v i s to c o r r e s p o n d as c l o s e l y as p o s
sible w i t h the l i m i t s of the r o u g h e n e d conte„ct areas

b e t w e e n -

65.
between the two bones.

Series Mean (cm.) S.D,(cm.) Range (cm.) Significance:
Dlff.of means

66 F e m a l e 3 . ^ 5 + 0.04 0.30 2.7 - ^.5
Stand, e r r o r

50 Basuto 3.5 7 + o . o 4 0.30 2.6 - 4 . 4 )

50 Zulu 3.59 + o . o 4 0.27 3.0 - 4.1
) 0.35 | 2.6

100 M a l e 3.53 + 0.03 0. 26 2.6 - 4 . 4 )

The sex difference in the means Is small but, never
theless, probably significant. The European averages were
3.9 cm. for the female and 4.3 cm. for the male (Verneau).
Most authors show a difference of about 4.0 mm. in the sex
means, and some (Litzmann, 166l) show high values because
their symphysis measurements include the subpubic liga

ments. It has been found that symphyseal height measured
in the living subject is large, no doubt owing to this
last named fact, and in two European cases of dystocia was
well over 6.0 cm.

This measurement is not only extremely important
obstetrlcally, but shows profound changes during ontogenetic
development - a point which is elaborated in the foetal sec
tion of this thesis.

In -

In his extensive investigations into age changes in
the pubic symphysis, Todd (1920-1921) pointed out that the
vertical diameter of the symphysis could not be measured
accurately until the age of 27 to JO years, and that even
at 35 there might still be doubt. Todd's aim was the
assessment of age of a skeleton by observation of the sym
physis pubis in which he was able to show regular age
variations. 10 phases were described, the last of which
had an age factor of JO years and over. Todd's significant
findings were not applied to the Bantu material because
only a very extensive experience with these symphyses would
justify such analysis.

66.

9. The lntertuberal diameter.
The diameter used was the inner lntertuberal breadth

measured from the most medial points on the lower surfaces
of the ischial tuberosities.
Series

67 Female
50 Basuto
50 Zulu
100 Male

Mean (cm,) S.D.(cm,)

10.20 1 0.16 1.30

g.37 i 0.15 1.03

7.29 ± 0.16 1.15

2.13 1 0.11 1.12

Range (cm.) Significance:
Dlff.ofmmeans
Stand, error

6.6

5.5

13.2
10.3

10.2

)
)

) )) 2.2 ) 10.7
)

5.5 - 10.3 )

In view of the fact that the difference between the

- male

male means was 2.2 times the standard error of the
difference, it may not be justifiable to group the
two male series. There is little reason for re
garding the difference as significant and attempt
ing to explain the smaller Zulu diameter, when the
Zulu pelvis tends to give greater measurements for
other diameters. The sex difference, however, is
extremely significant. The greatest male measure
ments are seen only to approach the female mean, as
in Verneau's (12>75) female mean of 13.6 cm. with a
range of 12.0-15.5 cm. and corresponding male mean
of 12.2 cm. Verneau's termini, rarely used, lie on
"le milieu de 1 'espa.ce, compris entre la termlnaison
de la levre profonde de l'ischlon et l'epine sciati-
aue11 (K&ppers, 1936).

Howexrer- the d i a m e t e r be defined, the p o i n t s are

mostly chdsen arbi t r a r i l y , and it is f o r t u n a t e that
this d i a m e t e r is l o s i n g caste in o b s t e t r i c a l c o n s i d e r

ations, a n d g i v i n g way to m ore I m p o r t a n t f e a t u r e s c o n

n e c t e d w i t h the o u t l e t (e.g. i n t e r s p l n a l diameter, size

of the p u b i c arch, a r e a of the p l a n e of l e a s t n e l v l c

d i m e n s i o n s ) . W h i l e the i n t e r t u b e r a l d i a m e t e r is c o m

m o n l y s t a t e d to be k inches in the n o r m a l pelvis, the

67 .

B a n t u -

Bantu female measurement is on the average a fraction
of a millimetre above this. The smallest female
diameter found in the series was 6.6 cm., or 2 (<r inches.
In Ince and Young's series (19*4-0) of *4-7*4- London women,
radiography gave a mean of 10.9 * 0.03 cm., standard
deviation 0.7 cm., C.V. S.K percent, and a range of
2.6 - 13.3 cm. -*-n "the London series the diameter is
7 mm. greater than in the Bantu, yet the mean value
for the Bantu female shows that this oart of the out
let is adequate.

Thoms (1935) refers to "records" of 1,000 women
"presenting what I assumed to be normal external pel
vic measurements", and an average of 9.5 cm. for the
intertuberal diameter. This figure becomes improbable
when the London and the Bantu series are critically
considered. Thoms' measurements were apparently taken
by different observers, and the unusually large post
erior sagittal diameter (averaging 7.3 cm.) implies a
taking of the transverse measurement too far forward.
For statistical purposes clinical measurement of the
intertuberal diameter is useless.

- 10.

Le
ft

69.
10. The width of the«.srcro-sclatlc notch was measured
from tlie tip of the Ischial spine to a. point on the lateral
edge of the sacrum midway between the lowest 2 posterior
sacral foramina.
Series

(65 Female
(
(50 Basuto
($0 Zulu
(100 Male

(65 Female
■£(•g/50 Basuto
*[$0 Zulu

Mean (cm. ) £5.D. (cm. ) Range (cm.

4.92 + 0.11 O.gg 2.g - 6.9
3.75 4* 0.07 0.51 2.5 - 5.0
3. 64- + o.og 0.55 2.5 - 4.5
3.69 4- 0.05 0.53 2.5 - 5.0

4.94 + 0.10 0. g2 3.0 - 7.0
3.75 + 0.07 0.49 2.3 " 4. g

3.65 + O.Og O.56 2.1 - 4.7

Dlff.of m e a n s
Ettand. e rror

)

) )) 1.0 ) 10.1
) )

)
)

The respective notch diameters on right and left
sides are identical when the male groups are considered.
This is due to chance: the sides frequently differ in
individual pelves on account of variations in the Ischial
spines in a given specimen, and the fact that the articu
lation of the pelvic bones is not always perfectly sym
metrical.

The sex difference is pronounced, a fact generally
recognised. No other estimation of the size of the sacro-

sclatic

sciatic notch was made, though in the foetus three
separate measurements were used in an attempt to give
metric values for this variable feature of the pelvis.

11. The external conjugate was not measured be
cause articulation of the lumbo-sacral joint is dif
ficult and apt to be inaccurate, and because this
diameter does not appear to be of value. Ince's
(1940) formula has, however, been used to arrive at
some values for the external conjugate, thus - G.V.
in mm. = O.36 X external conjugate in mm. + 47.47,
and the standard error is 9.° 9®. Taking the mean
value of C.V. (conjugata vera) in the female Bantu,
viz. IO7.6 mm., the external conjugate by calculation
is I67.O mm. or 6.5S inches: even adding the 9.0 mm.
of the standard error, this diameter in the Bantu
remains below 7 inches. Clinically an impression is
gained that the Bantu external c onjugate averages
about 7 inches; but in a large proportion of cases
it is certainly less. In obstetric practice, a con
tracted pelvis is suspected when the measurement is
below 7 inches, e.g. by authorities like Thoms (1935)*

70.

Eut

But in the Bantu at any rate this suspicion has no
valid foundation.

Perin (194-5) has published clinical pelvlmetrlc
findings on 2,033 indigenes from Kimvula, Belgian
Congo. The measurements are much smaller than those
of the present series, though the externa}, conjugate
averages 17.5 cm. Perin believes that these natives
are a degenerating race, and no female had a height
of more than 1 m. 50 cm. The neonatal cephalic
measurements show the head to be only slightly smaller
than that of the Bantu, the biparletal and suboccipito-
bregmatic diameters both being just under 9.0 cm. In
these small people, then, the external conjugate still
averages 17.5 cm.

In Thoms' series the conjugate averaged inches,
and in Ince's series over 2 inches.

Ince's formula is Inaccurate for small conjugata
vera values. His smallest true conjugate of 27 mm.
would give an external conjugate 4 7j inches (110 mm.),
and the commonly accepted figure of 4 inches of the
standard true conjugate would rnahe the external conju-

- gate -

71.

72
gate I50 min. or almost 6 Inches. A superficial
examination of the prepubertal pelvis shows that
an adult pelvis with external conjugate as low as
5 inches is almost inconceivable. Measurement of the
osteomalacic pelvis show how difficult it is to get
any external conjugate figure much below 7 inches,
and how obstetrically valueless this diameter is.

The above argument is necessary because no
figures are available to calculate the co-efficient
of correlation for true and external conjugate. The
co-efficient is almost certain to be too low to
justify the derivation of the true from the external
conjugate by biometricians dealing with the living
subject.

B. The Pelvic Brim and Cavity

The figures in Table 2A call for comment:
these should afford the most reliable data, being
from the skeleton directly, whereas the X-ray
measurements are from living subjects. Caldwell
and Moloy's (1933) figures were puzzling, and com
pelled certain calculations, after which their Table 2
became Intelligible. Their figure of 109 mm. for the
conjugeta vera. in American whites seemed very low, an
impression strengthened by the higher corresponding
figure (110.9 mm.) for Negro stock. Their male
averages of 101 and 92> mm. also seemed relatively high.

These unusual figures became understandable when
it was realized (from their Table 1) that, of 14-7
white and 121 Negro female pelves available, only 73
white and 90 Negro were measured. Further, the
respective numbers of gynecoid, android, $nd anthro
poid types of pelvis for the white series were 26, 25,
and 19; whereas their percentage incidence for the
whole series of 14-7 pelves was 4l, 3 2 , and 23. Similar
considerations affect the figures for the Negro pelves.
Thus, of a series of 263 female pelves, only 163 were
actually measured: hence the figures from this series

- must, -

must, unfortunately, be accepted with reserve.
Pan (1929) studied 6k female pelves from Hin

du cadavera, measuring but three diameters each of
inlet, cavity, and outlet: all pelves exhibiting
any deformity were excluded from his investigation.
However, Table 2 shows that these pelves are ex
tremely small: the brim area by calculation averages
approximately 9̂ -.̂ - sq.cm., as compared with a Bantu
average of 101.2 sq.cm., and with English averages of
121.0 and 126.S sq.cm. The average birth weight of
100 Hindu babies was only 5 lb. 61* oz. , against a
Bantu mean of 6^ lb. (Heyns and Hersch): the Hindu
head measurements were all less than the accepted
averages.

Table 2B show's a fairly close correspondence of
measurements obtained by X-ray pelvimetry, but the
formula used for the London series is not available;
Thoms' antero-posterior view method does not outline
the whole of the sacral promontory, whilst his lateral
view method, though more accurate, must sometimes give
rise to errors in determining the end-point at the
symphysis pubis. Nicholson's stereometric technique
is accurate to the nearest millimetre, and this pre

7^ .

cision

X O rH P d Eh
— TO tZ '- '

•
TO TO SEj 0 0

• P rH • •
<D <D •h to «a rH O 1 1 1 1 KN
X p ,p s • 0 m 1 t 1 1 .TO-
P TO 2= O rH rH

bO >_r P O -r-s.C O •H P TOa "■» O d — 0 0 in 0 on TO p to a
•H C P •d P • • • • • •h a o
p 0 TO P TO rH O N in 1 r -l-d-.ro- x H 0 4 1

0 0 rH •H Ph O rH rH I O TOvO d O > H
TO W — iH rH rH rH TO

O rH
■H TO TO bO
> P ctf • P
f—f O TO 0 0 3 3 OnH ■H
to bo .P P *H • • >
P, a} | -1 b o p < a 0 in 1 1 1 1 0 ■H

TO TO • rH CVJ 1 1 1 1 ON rH
TO T> p x a o rH rH
X 0 4-1
P 0) O

,p p p —
<M P c cti • P 41O TO TO O X O rH P rH TO —

■d a P -H • • TO TO <H •
m c TO •H p «8 CT\in I 1 1 I KN

0 m 1 1 1 i n
a P H S5

P a! P ,P TO . TO 3 bflw
to d ? 6 0 rH rH P 01 Pp •» TO < ! '— d W© > , TO TO
a p TO TO
TO -H a >d TO
•H > 41 P O O O O CVJ a
■d TO p CD O • • • • •

t» 0 d <P m N -m o N ! inro- ►»
a> TO m p rH O rH CVJ 1 O TO
bo TO P 0 rH rH rH rH rH P -*—«s.
TO .P •H 1 jd •
P p O

novo knvo inro- X P m>H
CD d O >H
> Vi p ■d rH <*)
TO O p r—ONtH CVJ N -O 1— P bO •• TO O H CVI CVJ rH ONVO • O RH
4-r P
O O < m i—1 rH rH rH rH m rd W '—

•
p u
O CD
CD P
•H CO ■
p 0 • ^
to f t a p
p 1 a -ri
a 0 >c p • •* cj

O TO a t o o
P a p
P TO P
TO <% • bp O

to a d -h
p a p <0
TO « P TÔ Kt TO
> ■» a o p >

TO 1 O 10 M H
TO CO O TO TO

p P P •‘ rH ft< d f t PTO TO TO TO TO 1 P TOP bO > d P O H I h P
TO d to o4 o p o TO
a n C H bO TO a a
TO P TO rH TO P H • TO•H O P O -H P P O

O O E - t O O « j m S 5

inovr̂ -

No
.

of
p

el
ve

s
50

9
35

O
13

2
10

^
Co

nj
ug

at
a

ve
ra

,
mm

.
11
21
.3

ll6
.k

12
1

Tr
an

sv
er

se
,

mm
.

I3
O.

6
13
2.
3

12
5.
̂

12
1

Br
im

I
nd

ex
,

%
90
.3

S3
.3

91
.2

' 12.
cislon can be relied on where shadows are clearly defined.

Con.1uga.ta vera.
Series Mean (cm.) S.D.(cm.) Range (cm.) Significance:

66 Female IO.76 + 0.10 0.60
50 Basuto 1C.39 + 0.15 1.09
50 Zulu 10.26 + 0.1*]- 1.00

100 Male 10.33 + 0.11 1.06

Dlff.of means
Stand, error

9.1 - 12.6 )
)

6.1 - 12.6 ) )
) 0 . 3 k ) 3.0

7.9 - 12.7 ) - )
7.9 - 12.6 )

These figures show an unimpressive sex difference: the
range for the female lies completely within that for the
male, the difference in means is only 3 times its standard
error, and the male conjugate is above k inches which is
the length usually given for the standard female pelvis.

The conjugata vera in the Bantu female of 107.6 mm.
inches) may be compared with that in English women of

ll6 - 116 mm. (between 44s- and 4-a/j inches). Ince and Young
(19^0) discussed this diameter fully, and suggested that
teaching that the standard length is or inches may be
a legacy from Smellie's "Treatise on Midwifery" in which
the length is given as inches. There has always been
a strong, though unsubstantiated, opinion in South Africa
that the Bantu pelvis is much smaller than that of the

- European

European. If one accepts European conjugate vera averages
of about 4-% Inches, the smaller Bantu corresponding diam
eter is of interest. Urban Bantu removed from the tribal
mode of life are subjected to a low standard of living and
nutrition, whilst those living in the native territories
may have to withstand occasional periods of great privation.
It is possible that with an improvement of nutrition the
female Bantu pelvis may alter considerably both in form
and dimensions.

13. Transverse diameter of brim.
Series

67 Female
50 Basuto
50 Zulu
100 Male

Mean (cm.) 3.D,(cm.) Range (cm.) Significance:
Dlff.of means
Stand, error

11.96 I 0.10 o.go 9.g - 13.g )

11.22 t 0.12 0.g3 9.5 - 13.3 ) )
> 0.77 ) 5.65

11.34- I 0.10 0.73 9.9 - 12.7 ) )
11.26 1 o.og o.go 9.5 - 13.3

The sex difference here is very significant statisti
cally, and is more pronounced than the sex difference for
the conjugate vera.

Table 2 shows that the American Negro me.le conjugate
(Caldwell and Moloy, 1933) is 0.53 cm. lower than the Bantu
male, but that the Negro transverse diameter is 0.34- cm.

greater

77.

greater. The Negro male pelvis is, therefore, smaller
at the brim than is the Bantu.

The female Bantu had a brim transverse mean more
than 1 cm. less than that of English women or of Cald
well and Moloy's American white material (I3.5I cm.),
and was even less than their figure (13.O cm.) for the
male American whites. Thoms' clinic women approached
it more closely with a measurement of 12.5 -̂ cm., and
his series of women of superior physique (12.19 cm.)
gave an average of only 0.23 cm» more. This last com
parison may be accounted for by the fact, emerging from
the experiments described in the appendix, that no
known X-ray technique can show the outline of the pelvic
brim: the skiagram will only show the shadow of a lower
level of the pelvic ring. But in different series, if
the X-ray measurements alone are compared, English hos
pital women appear to have a wider brim than the cor
responding American class, and American women of super
ior physique possess an even narrower pelvic brim.

13a. The pelvic brim index.
So many figures have been published on this im

portant pelvic index, and as in Turner's paper (1SS6)
so many races providing only a few specimens have had

- their

their respective indices put on record, that it is of
value to have available Bantu figures derived from a
reasonable series. For Europeans alone, Martin (1923)
and Turner (1326) collected figures for both sexes from
seven authors, three of whom found a higher figure for
the female, and four a lower figure for the female.

13a. The pelvic brim Index. In the English literature
this is known as the index of Turner (1336) xvho himself,
however, explains how Zaayer (1366) came to be the first
to use this valuable numerical expression.
Series

65 Female 90.
50 Basuto 92.
50 Zulu 91.
100 Male 91.

Mean S.D.

35 t 1.02 2.20

30 ± 1.34- 3.47
03 1 1.17 3.24-
56 t 0.91 9.11

Range

69.6 - 104-.4-
70.2 - 117.0

75.5 - 110.0

70.2 - 117.0

Significance: Dlff.of means
Stand, error

)
)
)

)
)

It can now be said that the Bantu male has an index
comparable with that of the female, and that there is no
significant difference xdiatsoever between the sexes; that
the index is over 90, and that it is, therefore, mesatlpel-
lie in Turner's (1336) sense.

The female Bush pelvis gave a brim index of IO5.2, the
4- pelves studied having an index of 99 > 112.6, 97. J>, and
112 respectively. The male Bush pelvis has the high brim

index -

index of 97.7 ranging in 9 reives between 22.6 and
IO5.5 (Orford, 193*0. Slome (1929) found the aver
age index for 2 female Bush pelves to be 99.7 and
for 5 male pelves 29.2. When Turner wrote in 1226,
2 female Bush pelves and 5 ma}.e had.been measured by
different authors. The females showed an average in
dex of 29.0 which suggests that the true mean is
lower than Orford'e high figure of 105.2; and the
males an index of 99.5. The male indices ranged
from 93 in Fritsch's (1272) specimen to 109 in Turn
er's. Thus the total of 19 male pelves give a doli-
chopellic index, or one over 95* The females appear
to have a lower index. The male pelvis does not
necessarily show a slightly higher index than the
female: the index is probably variable, and a lower
index in the ma}.e is the expected tendency.
14-. Antero-rosterlor diameter of pelvic cavity.

The sex difference here is much more pronounced
than in the case of the conjugate vera, being 9.3

times its standard error, whereas the differentiation
based on the true conjugate was found to be of the
order of J>.0 times the standard error. The male con
jugate vera was only 1.5 mm. lees than this correspond
ing cavity measurement, whereas these W o diameters

in -

go
in the female shoved a difference of 10.0 mm. If a sig
nif